Title Author(s) Citation Issue Date Across the Hindukush of the First Millennium: a collection of the papers Kuwayama, S. Institute for Research in Humanities, Kyoto University, 2002, 297p., (Report of the Research Project on the Historical Archaeology of the Hindukush areas carried out under the auspices of the Asian Archaeology Section of the Institute for Research in Humanities, Kyoto University). 2002 URL http://hdl.handle.net/2433/120966 Right (c) 2002 by the author, c/o Institute for Research in Humanities, Kyoto University. Type Book Textversion publisher Kyoto University 111111111111111111111111111111111111111111111111 2030052767 Across the Hindukush of the First Millenium REPORT OF THE RESEARCH PROJECT ON THE HISTORICAL ARCHAEOLOGY OF THE HINDUKUSH AREAS CARRIED OUT UNDER THE AUSPICES OF THE ASIAN ARCHAEOLOGY SECTION OF THE INSTITUTE FOR RESEARCH IN HUMANITIES, KYOTO UNNERSITY Across the Hindukush of the First Millenium A Collection of the Papers BY S.KUWAYAMA INSTITUTE FOR RESEARCH IN HUMANITIES KYOTO UNIVERSITY 2002 This booklet contains the papers published in various journals and conference proceedings between 1975 and 1998. The collection was made on the occasion of the author's retirement from Kyoto University on the thirty-first of March 2002 in order for those concerned with the archaeology of the region north and south of the Hindu Kush to get easy access to the author's main works which have since 1963 been directed to how the first millenium of the Hindu Kush areas could historically be understood. The author particularly thinks it most encouraging and feels it extremely happy to have shared with Maurizio our time to live for these thirty years and therefore this humble work is dedicated to his dear memory. Copyright 2002 by the author, c/o Institute for Research in Humanities, Kyoto University, Kyoto 606-8501, Japan. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or in any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the author. To the Memory of Maurizio ----~~~~------------~------------------------------------ CONTENTS LIST OF ILLUSTRATIONS ix SOURCES OF THE ORIGINAL PAPERS xi I Buddhist Establishments in Taxila and Gandhara A Chronological Review 1-11 II The Square Podium of Stupa in Taxila: its Introduction 12-19 III The Buddha s Bowl in Gandhara I 20-43 IV The Wheel-shaped Structure inside Stupa: A Hidden from Augustan Rome 44-68 V Shah-ji-ki Dheri before Kanishka 69-89 VI Aspects of Stupa Court at Hadda: Tapa Shotor and Lalma 90-106 VII The Hephthalites in Tokharistan and Gandhara 107-139 VIII Two Itineraries concerning the Emergence of the Colossi in Bamiyan 140-155 IX Bamiyan and its Buddhist Activities 156-161 X A New Date for Begram III 162-172 XI The Horizon of Begram III and Beyond A Chronological Interpretation of the Evidence for Monuments in the Kapisi-Kabul-Ghazni Region 173-199 vii XII Zhun, Aruna and Two Superimposed Shrines at Khair Khana, Kabul 200-207 XIII Identity of the Napki Coins 208-221 XIV The Turki Shahis and Relevant Brahmanical Sculptures in East Afghanistan 222-248 XV The Inscription of Ganesa from Gardez and a Chronology of the Turki Shahis 249-259 XVI Dating Yasovarman of Kanauj on the Evidence of Huichao 260-270 NOTES FOR CHAPTERS 271-280 REFERENCES 281-297 ILLUSTRATIONS viii LIST OF ILLUSTRATIONS 1 - Stupas on the main street at Sirkap, Taxila. 2- Map of Taxila showing the main cities and monasteries. 3 - Map of Gandhara showing the distribution of monasteries. 4 - Stupa IF, Sirkap (above) and Chir Tope C (below). 5 Plans ofJolian showing three successive phases. 6 - Plan of the stupa courts at Takht-i Bahi. 7 - Plan of the stupa court D at Tareli. 8 - Plan of Mekhkha Sanda monastery. 9 - Plan ofJamal Garhi monastery. 10 - Adoration of the patra: Lahore Museum (above) and Kabul Museum (below). 11 - Map showing the distribution of stupas with and without the wheel-shaped pattern inside. 12 - Examples of the wheel-shaped structure: Augustus's mausoleum, Ghanta§ala, Kankali Tila, Dharmarajika (Taxila), Fil Khanah, Sanghol No. 1. the above left clockwise.) 13- Plan of the Shah-ji-ki Dheri main stupa and its schematic cross-section of the stupa mound. 14 - Plans of stupa courts at Hadda: Lalma (above)and Tapa Shotor (below). 15 - Mouldings from stupas at Hadda (above) and cross-sections showing the superimposed floors between the stupas at Lalma. 16 - Routes of Song Yun in the Hindukush (above) and a map showing a route from Gandhara to the Punjab (below). 17 - Maps showing the routes taken by Buddhist monks before (above) and after (below) the mid-sixth century. 18 - The 55 m. and the 38m. Buddhas, Bamiyan, photogrammetrically surveyed in 1970 and 1976. 19 - Distribution of monuments in the Kapi§i-Kabul region. 20 - Monuments with round bastions: 1. Shotorak, 2 and 5. Begram, 3. Khair Khanah, 4. Tepe Maranjan, 6. Saka, 7. Bagh Gai, 8. Kohna Masjid, 9. Tapa Skandar. 21- Stamp-impressions in north Afghanistan. Durman Tepe (1-3, 11, 15, 16, 18-21,23), Balkh (4,5,7,8,10,12,14,17,22), Kara Tepe (9,13), Shahri Bannu (6) and Chaqalaq Tepe (photographs 1-12). 22 - Pottery types in the Kunduz region: Durman III (1-4), Durman IV (5), Chaqalaq Tepe II (6-14, 22-26), Durman III-IV (17-21), Chaqalaq Tepe III (15, 16,27). Scale 1:6. 23 - Plan of the excavated monuments at Tapa Skandar. Black dots and white circles indicate the find spots of stamp-impressed potsherds and pieces of marble Brahmanical sculptures respectively. ix 24 - Medallion stamps. Floral designs. Begram Bazar (9, 32, 34, 39), Begram III (2,23,29, 33), Gul Dara in Logar (12), Kham Zargar (3), Saka (1, 6, 10,21, 24,26,27,50,53), Shotorak (7,14), Tapa Sardar (4,8,22,28,36,51), Tapa Skandar II (5,11,13,15-17,25,30,37,38,41-45,47,49), Tepe Maranjan (18-20,31,35,40,46,48,52). 25 - Medallion stamps. Animal designs. Begram Bazar (1, 8,43,46), Begram III (9, 17, 46), Ghundai Paisa (37), Gul Dara in Logar (3, 4, 10, 45), Kham Zargar (6, 12), Shotorak (31), Tapa Sardar (26-28, 36, 42), Tapa Skandar II (2,7, 16, 19-23,25,29,30, 33-35, 38, 39,44,47-51), Tepe Maranjan (5, 11, 13-15, 18, 24, 32,40,41). 26 - Plan around Stupa 64 and miniature city wall 69, Tapa Sardar and Verardi 1984) (above). Plan and cross-section showing the interrelation between the constructions H, F, G, ] and c, Tapa Skandar (below). 27 - A plan of Tepe Maranjan revised by Fussman (Fussman and Le Berre 1976) and modified by Kuwayama (above) and the layout of the precinct and monastic buildings, Gul Dara in Logar (Fussman and Le Berre 1976, PI. II). 28 - General plan ofShotorak (Meunie 1942: 10) with its detailed plan of Court D (Meunie 1942: 21). 29 - Stupas with deep niches (D2, D3, D4 and D6 at Shotorak) and octagonal tiers (D4 at Shotorak and TK121 at Hadda). Not to scale for TKI21. (Meunie 1942: figs. 4-7 and Barthoux 1933: fig. 89). 30 - Plan and section of main buildings at Khair Khanah. (Drawn Kuwayama after Hackin 1936: Plans I and II). 31 - Line drawings of the Uma-Mahesvara statue, Tapa Skandar. 32 - Marble Brahmanical images (1). Head of Durga from Qal'a Amir muhammad, Tagao (above left), head of Siva from Charsada (above right), Ekamukhalinga from Hund (below left) and head of Siva probably from Gardez. 33 - Marble Brahmanical images (2). Surya from Khair Khanah and Durga Mahishasuramardini from Gardez. 34 - Inscription of the Ganesa statue from Gardez. An ink rubbing. x SOURCES OF THE ORIGINAL PAPERS The following is to show when and where each paper originally appeared. The original title is given in the parenthis in case of any modification. 1. Buddhist Establishments in Taxila and Gandhara: A Chronological Review (Buddhist Temples in Taxila and Gandhara: A Historical Review of Monastic Arrangements) in A Catalogue of the Exhibition ofGandhara Art in Pakistan, Tokyo, 216-220, 1984. II. The Square Podium of Stupa in Taxila: its Introduction (Notes on the Introduction of Square Podium to the Taxilan Stupas) in Orient 14, 23-34, 1979. III. The Buddha's Bowl in Gandhara (The Buddha's Bowl and Relevant Problems) in South Asian Archaeology 1987. ed. Maurizio.Taddei, Roma, 946-978, 1991. IV. The Wheel-shaped Structure inside Stupa: A Hidden Import from Augustan Rome (A Hidden Import from Imperial Rome Manifest in Stupas) in Gandharan Art in Context: East-West Exchanges at the Crossroads of Asia, ed. R. Allchin, B. Allchin, N. Kreitman and E. Errington, Cambridge, 119-171, 1997. v. Shah-ji-ki Dheri before Kanishka in Lahore Museum Bulletin 911,31-58, 1996. VI. Aspects of Stupa Court at Hadda: Tapa Shotor and Lalma (Lalma and Tapa Shotor: Aspects of the Stupa Court at Hadda) in Annali, Istituto Universitario Orientale, Napoli, 4712, 154-176, 1987. VII. The Hephthalites in Tokharistan and Gandhara (The Hephthalites in Tokharistan and Northwest India) in Zinbun: Annals of the Institute for Research in Humanities, Kyoto University 24, 89-134, 1991: Lahore Museum Bulletin 5/1, January-June, 1-24 and 512, July-December, 1-18, 1992. VIII. Two Itineraries Concerning the Emergence of the Colossi in Bamiyan (Literary Evidence for Dating the Colossi in Bamiyan) in Orientalia: Iosephi Tucci Memoriae Dicata, ed. Gnoli, G. and L. Lanciotti. Serie Orientale Roma 5712,703-727, Roma, 1987. IX. Bamiyan and its Buddhist Activities (Bamiyan) in Enciclopedia Italiana dell'Arte Antiqua, Classica e Orientale. Secondo Supplemento 1971-94, 1 (A - Carro), 598-602, Roma, 1994. X. A New Date for Begram III (Kapisl Begram III: Renewing its Dating) in Orient 10, 1975, 57-78. XI. The Horizon of Begram III and Beyond: A Chronological Interpretation of the Evidence for Monuments in the Kapisi-Kabul-Ghazni Region in East and West 4111-4,79-120,1992. XII. Zhun, Aruna and Two Superimposed Shrines at Khair Khana, Kabul (K.hair Khana and its Chinese Evidence) in Orient 11, 1976,93-107. XIII. Identity of the 'Napki' Coins (Not Hephthalite But Kapisian Khingal: Identity of the Napki Coins) in Ex Moneta: Essays on Numismatics, History and Archaeology in honour of Dr. David W MacDowall, ed. Amal Kumar Jha and Sanjay Garg, 331-349, New Delhi, 1998. xi XIV. The Turki Shahis and Relevant Brahmanical Sculptures in East Afghanistan in East and West 26/3-4,375-407,1978. XV. The Inscription of Ganesa from Gardez and a Chronology of the Turki Shahis (L'Inscription du GaI).esa de Gardez et la chronologie des Turki-~ahis) in Journal Asiatique 279/3-4 , 267-287, 1992. XVI. Dating Yasovarman of Kanauj on the Evidence of Huichao in Zinbun: Annals of the Institute for Research in Humanities, Kyoto University, 29, 1- 15, 1994. xii I Buddhist Establishments in Taxila and Gandhara A Chronological Review The doctrines of Buddha were generally planted in the minds and hearts of people by his followers, but another form of motivation for their acceptance must have been present to cause them to spread over vast areas beyond the sphere where the Buddha lived and preached. This spreading was not only because of something inherent to the character of Buddhist allowing it to find wide acceptance by various races, in contrast to the other Indic regions, Brahmanism and Jainism, but because of an external cause, a secular inducement: political and therewith economic power. The cremated remains of the Buddha are said to have been dug up by Asoka (262-232 BC) out of seven stupas and reburied by him in eighty-four thousand places, establishing the same number of stupas in his extensive territories. This legend suggests that Asoka exploited the stupas for political ends, using them to help control his vast domains. It seems that a large number of stupas said to have been erected him, which had repeatedly undergone repairs in later years or had fallen into a dilapidated condition, still remained at the beginning of the seventh century, as Xuanzang referred to them many times in the Da Tang Xiyu ji. In the Northwest of the South Asian Subcontinent Buddhism must have thus made its first appearance through the building of monuments by Asoka. But information about the Buddhist position in the Northwest after Moka is regrettably limited, except that, as the Milindapanha tells, Menander, an Indo-Greek king in the middle of the second century BC, had a debate with Nagasena about Buddhism in Sagala (present Sialkot). Regarding the early aspects of Buddhist architecture in the Northwest we had to wait for archaeological work uncovering the remains of Taxila. T axila (Ill us. 2) Inside the city wall of Sirkap, constructed with rubble-in-mud masonry in the time of Azes I (Ghosh 1947-48; Allchin 1993: 75-77; Fussman 1993: 89-96), four superimposed occupations represent the successive use of the city from that monarch of the Sakas until Parthian Gondophares or slightly later, according to Marshall. Allchin (1968) thinks that Sirkap lasted longer until Kanishka, while for Erdosy (1990: 668-670) the last ruler is Huvishka. Whether Buddhist structures accompanied the first occupation within this walled-in city-Stratum IV according to Marshall and Phase I according to WheelerGhosh-is quite unclear due to the lack of any extensive excavation at this deep level. As far as the excavation reveals in Sirkap, the Buddhist structure thought to be the earliest is a small circular stupa in a small court (A) at the Northwest corner of the house in Block IE. The level from which it springs shows it to be older than the house itself which surrounds the stupa. It is rooted in the third stratum of the Marshall excavation corresponding to Wheeler-Ghosh's Phase II dated by the latter to the first half of the first century AD The location of the stupa within a smaller enclosure in a house shows it was built for private worship only (Marshall 1951: I, 158). Yet interesting is the fact that, faced with kanjur ashlar, it is elaborately furnished with an unusual type of acanthus ornament boldly modelled in lime stucco and painted. In the next period, Phase III (the later half of the first century AD) based on Wheeler-Ghosh or Stratum II according to Marshall, six stupas were raised, more or less facing the main street running north to south (Illus.l). However, only one of them was in the form of traditional Indian Buddhist architecture and the others were of a type that of an had never been met anywhere in the Subcontinent. The former is a sizable apsidal-shaped ground plan in Block ID. It consists of a spacious nave with a porch in front and a circular apse behind, the whole being surrounded by an ambulatory passage, to which access was gained from the front porch. In the middle of the apse there was originally a circular stupa now extinct (Marshall 1951 : I, 150-155). All of the other stupas in Phase III have a square plinth, in place of a circular one, on which the drum and dome stand (Marshall considers one of them-Stupa in Block lA-to be Jain). The square plinths of the stupas were not built to form an architecturally undecorated box-like cube. As seen in the stupas of Blocks IF and they were supported by simple torus-and-scotia mouldings on the foundation and bordered by cornice and bracket decoration on the upper margin (Illus. 4, above). Each lateral face of the plinth is divided into a certain number of bays by Indo-Corinthian pilasters. The elements were carved out of the local kanjur stone on which stucco was applied and painted. Plinths ornamented in this way presumably were derived from a prototype in the form of a square monument decorated with pilasters, rather than from two elements, a cubic plinth and architectural decorations, being separately introduced and associated together here in Taxila at this specific period. The prototype existed among architectures popular in early Imperial Rome rather than in either the north (Greco-Bactria) or the south (Indo-Greek territory) of the Hindu Kush. Taking into consideration that such square plinths were associated with stupas, the Roman prototype must have been sepulchral. The tomb of Caecilia Metella, probably of Augustan date, in a southern suburb on the Via Appia, the tomb at Pozzuoli near San Vito, and the mausoleum at Marano near Naples all are examples having an imposing square plinth which supports a circular body above it. The association of a square with a circle was really revolutionary since the square is basically inconsistent with such a stupa ritual as circumambulation, which, as Irwin (1979: 799ff.) explicitly explains, might have originated in the concept of an ancient cosmic axis. In the northwestern part of the Subcontinent, or the country of the mleccha, where social and cultural traditions were quite different from India proper, or MadhyadeSa, such a combination, quite heterogeneous from India's point of view, might have been taken by the mleccha as something not unnatural for stupas which were a purely Indian type of religious monument. The combination shows that Northwestern Buddhists were already religiously predisposed to take stupas as a sepulchral monument. It is not far from the truth to think that the stupa was apparently conceived in the time of Azes II -Kujula Kadphises as ideologically equal to such Roman counterparts as exemplified above. Roman and Indian ideas met to create quite a novel type of Buddhist monument in this mleccha country. Within the positive evidence for intercourse between Rome and the Subcontinent in the first centuries BC and AD important is the fact that the technique of building Roman mausolea containing a wheel-shaped structure was transplanted into the Northwest through Southeastern Andhra India in the time contemporary with early Julio-Claudian Rome, as discussed later in detail in Chaper IV (Kuwayama 1991B). The importation of the square plinth, first adapted for stupas in Phase III of Sirkap in the second half of the first century AD seems consistent with other various imports from Rome, tangible or intangible, at the same period. Once the shape of stupas had been accepted in this way, the new type became overwhelmingly popular in Taxila and beyond from this time onwards. The stupas containing pieces of Buddha's cremated bones were the object of worship on the laypersons' side as the symbol of their legendary division among eight representatives of the countries close to the location of cremation. Monks, on the other hand, primarily responsible for developing and deepening the creeds had nothing to do with the stupa cult at the outset, and moreover, they had no usual abode except during the three-month rainy season when they gathered together in one place, such as a grove offered by rich merchants. Yet such seasonal settlements gradually acquired permanent buildings of assembly. We hardly know when they were established except for the remains of primitive structures in Rajgir and elsewhere. For some duration the cult of stupa and the life of monks thus seem to have remained separated, as seen in present Burma where monks do not concern themselves with any stupa cult. Even during A§okan times stupas seem to have still been isolated from monks, as Piggott (1943: 1-10) shows at Bairat. In the first half of the first century BC, however, a caitya-stupa and some cells for monks were hewn out side-by-side in one place as groups of caves in rocky hillsides in West India. This fact might indicate that the stupa had generally become an object of worship even for monks by the late second century BC In Taxila, buildings for monks' dwellings appeared in close proximity with a stupa at the same time as the square-based stupa originated in Sirkap, as seen at Dharmarajika and Jandial B. Such buildings in this earliest period simply consist of small cells joined in a single line basically running straight north to south to the west of the stupa. On the other hand, the stupa has special features such as enclosure walls of rubble masonry with low elevation and thin breadth. The main stupa of Dharmarajika was encircled in its earlier phase (Period I of Dharmarajika) by nine small stupas (or pillars) supported by square plinths. This plan at Dharmarajika along with examples of walled enclosures show that in the earlier phase of Taxila Buddhist architecture the traditional Indian style of enclosing the stupa with pillars or a fence was familiar to local architects. If so, both innovation and tradition coexisted side by side in Taxila. In the following century, the second century, the building of Buddhist temples suddenly became more vigorous, taking on a new aspect: namely, large scale square 3 buildings for monks emerged instead of the previous simple cells aligned in a north-south direction. Examples are at Pippala, Kalawan, Giri (C-D-E), and Chir Tope (A, B, C, DI, and D2) (Illus. 4, below). The single narrow entrance now led to an inner court surrounded by monks' cells closely ranged on all four sides. The ground plan of this new type of square monastery is ostensibly similar to vihara caves in West India and, therefore, looks as if derived from it. But an essential difference exists between the single narrow doorway at Taxila and the wide facade occupying one whole side of the square in cave temples. If such a type of monastery as that characterized by a strongly closed character was used for permanent residential purposes, this peculiarity might have promoted various restrictions as part of the monks' monastic discipline. This new architectural type lasted for a long time in Taxila, forming a combination with a single stupa of the square plinth type, and it became the typical monastic building in other parts of India in Gupta and Post-Gupta times. However, taking this building as exclusively used for monks' permanent residence is dangerous. The number of cells is quite limited, ranging between eleven and forty only on the ground level and accommodating the same number of monks. This may mean that in the building could only live monks who had to learn vinaya after completing the precepts. In any case, the purpose must have been closely related to Buddhist institutions. During these centuries, although stupas with low surrounding walls were still being built at Pippala, Giri C-D-E, and Chir Tope DI, stupas without fences and freely accessible from all directions were typical, as seen at Chir Tope B, and C. Such an open character may have developed parallel with the establishment of the closed-in type of square buildings for monks. There were, however, monasteries with a stupa in their inner court at Pippala and Chir Tope D2, and a monastery existed with griha-stupas at Chir Tope B consisting of a square front room and a square rear room in which a circular stupa was enshrined. At Kalawan, a site geographically rather isolated from other temples in Taxila, five griha-stupas were built neighbouring two stupas facing Northwest and located in front of two square monasteries. At Kalawan in particular, there was no main large stupa overwhelming other ones, and griha-stupas were preferred, increasing in number in the following period. An alignment of cult buildings of similar scale to those in the Kalawan stupa court is also met at Nimogram in Swat, yet the latter would not necessarily be in the same span of time with Kalawan. At Kalawan two more cells probably enshrining cult images and one large square monastic building were also added much later, in Period V, the last period but one in the monastic history of Taxila. This successive erection of griha-stupas makes the Kalawan precinct unique in Taxila, and even in the areas to the west of the Indus, Gandhara and Uddiyana, and the stupa cult long retained something of a conservative character there. The characteristic masonry work changed in chronological order in the cities and monasteries in Taxila, as Marshall observed. Comparison of the masonries employed in a monastery or in different parts of a monument with those used in others can lead one to draw a tentative chronological picture of the development of Buddhist temple architectures in Taxila. Thus the evolution of the temples hitherto discussed has become clear by setting the architecture against two kinds of earlier masonries applied to the facing of a monument: the kanjur ashlar masonry for Period I applied to the core masoned with 4- rubble and the diaper masonry for Period II replacing rubble masonry. Whether use of the latter ended with the life of Sirkap or continued into the earlier time of Sirsukh would be guesswork without further soundings inside Sirsukh. But, if one is allowed to take some parts of the Sirsukh fortification as composed of a type of diaper masonry, it may be conjectured that the diaper work did continue into the earlier phases of Sirsukh. Highly suggestive for the duration of this particular masonry is the observation given by Allchin (1968: 13) that "the new diaper masonry of Sirkap coincides with the arrival of the Kushans under Kujula Kadphises rather than Gondophares as Marshall thought." This is consistent with other evidence in the form of some small stupas closely related to the semi-ashlar masonry, the direct follower of the diaper, at Dharmarajika. Stupa K3 on the platform masoned with semi-ashlar contains three coins of Kanishka, while Stupa P6 faced with kanjur ashlar, of the same appearance as semi-ashlar stupas Kl and N4, contains three coins of Huvishka and seven coins of Vasudeva. K3 is standing on the filling of a rubble-masoned tank which produced three coins of Kanishka and one of Soter Megas, now identified by Cribb to Wima Taktu. If one relies on this numismatic framework and if the rubble masonry of Sirkap and of the tank in Dharmarajika is not later than Azes II, as interpreted by Allchin, the introduction of diaper masonry is roughly datable to the time of Kujula Kadphises. In the light of this stratigraphical evidence of coins, the existence of the semi-ashlar masonry in the time of Vasudeva or later than him is possible, but its existence in the times of Kanishka and Huvishka is not indisputable. Therefore a shift from the diaper masonry to the semi-ashlar was supposedly sometime in the time of the Kanishka-Huvishka reigns. After the kanjur ashlar-rubble and the diaper, four other masonries are recognized by Marshall at Buddhist establishments located outside of the cities, but no sound stratigraphical evidence for the development of these four masonries is possible without soundings inside the Sirsukh site. For the time being we have to settle for what we are given by Marshall. If the earlier masonries are divided into the products of two ages, then it is in Period III that a large diaper masonry work heralds the semi-ashlar in the succeeding period which corresponds with a remarkable new aspect in the stupa. Thus Period III is transitional in view of the masonry datable to the time of Kanishka-Huvishka. Many subsidiary stupas were introduced around a larger stupa at Chir Tope B, Giri C-D-E and Kalawan. Small stupas had already appeared in Period I only at Dharmarajika (G8, Q1, D3, D2, D4 and S8, all being closely related to the deposited coins such as Azes I and Maues) and in Period II on a rather small scale at Pippala, Chir Tope C and Dharmarajika, while during Period III they actually increased in number around the main stupa. In contrast to the idea of a single isolated stupa, an idea had strongly developed of a main stupa being built in combination with smaller subsidiary stupas. However, the construction of these additional small stupas was not a phenomenon common to all precincts in Taxila: the stupa of Kalawan H Monastery inaugurated in this period was not in company with any small stupas. Also in Period III two temples, Jolian and Mohra Moradu, were raised, both of them having one small stupa beside the main one. Among temples built in Period III, and also among temples remaining in use from Period II, there arose a clear 5 distinction between those allowed to erect small stupas, i.e., Kalawan, Chir Tope B, Giri C-D-E, Jolian, etc., and those that did not accept the idea of the cult of small stupas, i.e., Kalawan H, Mohra Moradu, Giri A-B, etc. It is these small stupas that were grandiosely decorated with high reliefs mainly of stucco Buddhas. Period IV began with the appearance of the early type of semi-ashlar and was a time of building activities at Jandial Band Jolian. It is hard to say whether the stupa at Jandial B, originally built with late rubble masonry, had decayed before this period or was deleted in this period to allow for a new building. But in either case, covering entirely the previous square stupa, the remaining elevation of which was a reduced one, a round stupa was raised with semi-ashlar masonry, the same device being used to build two small stupas alongside the main one. In the time of the semi-ashlar masonry conspicuous building activity took place at Jolian (Illus. 5). Small stupas, twenty in number, were set closely together in line around the main stupa built in the previous period with a high plinth. In addition, the main stupa and small stupas surrounding it were also enclosed by cells of shrines ranged on all four sides around the stupas with an interspace to allow for circumambulation. With only a part of the design, corresponding to the front of the main stupa, being left open as an entrance, a strongly closed-in character was given to the precinct achieving the appearance of what might properly be called a stupa court. In Jolian the shrines thus built, about five meters high and on a square plan, seem to have permitted the enshrining of a single colossal statue, stucco-made, of the Buddha in most cases. From the outside, also had the stupa court of Jolian might produce the impression of a stronghold. during this period a spacious terrace some steps lower than the enclosed stupa court, where five more small stupas were raised. This type of a closed-in stupa court at Jolian is the one and only example among the many Buddhist temples in Taxila. The extreme paucity of this type throughout the history of Taxila presupposes that elsewhere, outside of Taxila, there was a place where it had firmly been established and that it reached Jolian in Period IV. If this supposition is acceptable, it could afford evidence for a cross-dating to the temples of this kind in the area to the west of the Indus, excluding Swat where no such example has been revealed so far. In Gandhara and Hadda this kind of stupa court was quite popular, as seen at Takht-e Bahi, Jamal Garhi, Tareli, Mekkha Sandah, and Ranigat in the former area and all sites excavated so far except Chahil-e Ghundai, Prates and Lalma in the latter area. Other conspicuous building activity was done in this period, at Dharmarajika in particular. Here a huge square building for monks, called by Marshall Court G, was constructed at the site just to the north and northeast of the Period I Court B which had become desolate by Period IV. To the south of this huge monastic building and to the east of Court B is Stupa No. 1 in Court A. All that is left of this stupa is a huge, high square plinth with a facing of dry masonry work of kanjur ashlar. This type of masonry is demonstrably characteristic of the Period I stupas. On Stupa No. 1 a pair of torus and scotia mouldings massively increase the height and are surmounted by another element, i.e. a square kanjur block with the upper part slightly projecting which supports pilasters. This particular element coupled with the torus-and-scotia below, or a set of these elements, 6 is observed to appear in the earlier period with stupas worked with rubble faced by the kanjur-ashlar, such as stupas in Blocks IF and IA in Sirkap and QI, TI, and MI2 in Dharmarajika, and particularly came in vogue together with the semi-ashlar masoned structures. Marshall believes that both Stupas Nos. 1 and 4 are contemporary with earlier kanjur-ashlar masonry, yet doubts exist why only Stupa No.4 was deleted while No.1 survived several centuries long to be annexed by some walls in the latest stage of Taxila Buddhist history. This gives cause to think that, despite the facing exclusively of kanjur blocks, Stupa No.1 may be a later architecture. In the following period, Period V, the masonry of semi-ashlar type tends to have the boulders laid in more regular horizontal courses and to have the small slabs between the boulders more equal in thickness. With the masonry of this kind the stupa and monastery at Kunala were built on the hill, to the south of Sirkap, where the active life of Sirkap was already an old story, and the square monastic buildings at both Kalawan and Chir Tope B were added to the west side of the monk's cells which had been built with very early fine diaper work in Period II. In Period VI four Buddhist temples, Badalpur, Bhallar and Bhamala, were constructed, while renovations were made at four main sites: namely, Kunala, Jolian, Pippala, and the monastic area of Dharmarajika. This period therefore can in no way be described as one of decline, although this is the last of the long series of masonry styles in Taxila, assigned by Marshall to the fifth century. In the last quarter of that century, according to him, the Hephthalites first invaded the northwestern Subcontinent destroying the Buddhist monasteries there. However, the Hephthalite destruction of Buddhist Marshall, and activities in the Northwest really is a sheer conjecture fancifully created does not bear any historical authenticity, as discussed elsewhere (Kuwayama 1989) and stated here in Chapter VII. As to masonry, one remarkable feature of this period was the employment of huge dressed limestone blocks. At Bhallar Monastery the stupa reached a gigantic size with such blocks in combination with thick ashlar layers. The feature of employing gigantic limestone blocks is similarly seen in seventh and eighth century Kasmir, where there was also a tendency to use massive stones for temples, whether it be Buddhist or Hindu. Also to this period belongs Bhamala Monastery on the northern bank of the Haro River, the main precinct of which is a stupa with a cross-formed plan. Especially to be noticed is the fact that this form of stupa and the use of gigantic stone blocks as well are manifestly shared with Parihasapura in Kasmir and Ushkar on the south side of the Pir Panjal Ranges. The main stupas at Shah-ji-ki Dheri in Peshawar, Adjina Tepa near Dushanbe, and Rawak in Khotan, all share the same cruciform plan and the seventh-eighth century date with those in Kasmir and Taxila. The same may be true for other stupas such as Zar Dheri on the way connecting Taxila with Kashmir, the later phase of Ahin Posh in Hadda, and Top-e Rostam in Balkh. Period VI, the last, is marked by the revival of Buddhist activities in Taxila after a few decades of decline. The decline had been caused not by the Hephthalites but by the West Turks to take no direct concern for the northwestern part of the Subcontinent, unlike the Kushans and the Hephthalites, leading to a temporary disturbance of the long 7 distance trade which had long supported Buddhism in the Northwest. An eye-witness of the revival in the earlier decades of Period VI undoubtedly is Xuanzang. Buddhism then regained its vigor with the final decades of the seventh century under the hegemony of the Turkish king of the Kabul Shahs, who extended their Buddhist sentiments by building a number of temples not only in the Shevaki and Kamari areas to the south of Kabul but even beyond the Kabul-Indus Valleys in Kasmir. Thus Period VI is definitely separated as a new era of Buddhist activity from the previous, long flourishing phases that are characterized by "Gandhara" stucco and schist sculptures, and roughly speaking, the middle of the sixth century is the watershed between these "Gandhara and Post-Gandhara" Periods. Gandhara (Illus. 3) Gandhara geographically covers an area from Peshawar in the west to Swabi in the east, about 100 km east to west, and from the northern hills, where a route crosses the Pir Baba leading to Barikot in the Swat Valley, to the northern bank of the Indus, where an important fortified ford, Hund, connects Gandhara to Taxila and beyond. The area economically supported four cities: Purushapura, Pushkalavati, Varushapura and Udabhandapura. Long distance trade along the routes running from the Subcontinent through Gandhara and Uddiyana, and Gilgit or Chitral to Central Asia, and vice versa, promoted foreign rulers' and local peoples' support for Buddhist activities. Ruins of Buddhist temples are concentrated in the area to the north of the Shahbaz Garhi-Swabi line, most of them located at the foot and on the top of the northern and western hills and some in the plain. The main characteristics of the Gandhara Buddhist temple generally are twofold: the scattered arrangement of monks' abodes and the construction of a main stupa surrounded by cells of shrines with the interspace in between occupied by small stupas. As for monastic architecture, two to four cells are arranged to form a single, block type building. Buildings, each being rather restricted in size, stand scattered on the hilltops and/or the hillsides, forming an area for monks. Such a plan cannot be found in Taxila even at the temples on hills, where builders always obtained a terrace on which spacious monastic buildings of a square, closed-in form were usual. In Gandhara this closed-in square type is seldom met except for examples at Takht-e Bahi (Illus. 6) and Sahri Bahlol. Even in these examples the building is rather smaller in size with less than half the number of cells of comparable structures at Taxila. At Takht-e Bahi the mainstream of monastic building is not of this type but numerous small blocks of buildings arranged in sporadic fashion on the sides of hills; the quadrangular type is located as if to be an integral part of the stupa court, to which the doorway is open. Although the monks' abodes are scattered, the stupa court seems to have been designed to allow laypersons easy access without disturbing monastic life. Jamal Garhi (Illus. 9), situated on the top of an isolated hill, also has the area of monastic buildings located beyond the stupa court to which lay-worshippers would come immediately after 8 entering a main gateway reached from the plain through a long stepped approach. At Tareli, one goes along the bottom of a deep valley first to arrive at the stupa court located on a terrace at the foot of steep hills, on the higher part of the southern slope of which the monastic cells stand. At Mekkha Sandah (IUus. 8), the court of the stupa was on a ridge halfway up a mountain, while the monks' dwellings were on the southern slopes of another hill. Thus, in Gandhara, at those temples excavated so far, the monastic life seems to have been strictly separated from lay-worshippers. As mentioned earlier, the stupa court at Jolian of Period IV is isolated in style from others in Taxila, but the same style is quite popular in Gandhara and Hadda. This is a court with a main stupa surrounded by cells for image shrines, the intermediate area filled with small stupas. Strictly speaking, there were two such types. At Jamal Garhi and Takht-e Bahi (Illus. 6) the main stupa is in the center of a court enclosed only shrines with a broad interspace between them, and the small stupas stand on another terrace at a level lower than the court of the main stupa also surrounded by shrines. At Jamal Garhi, there are two such terraces on different levels consisting of only small stupas and shrines. The basic plan at these two temples was thus the following: (1) the court of a main, large stupa enclosed by shrines on the highest location without any small edifices; (2) a court of small stupas enclosed by shrines on a terrace lower than No. 1. In this style of stupa court, therefore, Nos. 1 and 2 are clearly separated from each other. At Takht-e Bahi, the court of the main stupa and a quadrangular monastic structure were located vis-a-vis on a median line running north and south with a space between them full of small stupas. Another style of stupa court, as seen at Tareli (Illus. 7) and Mekkha Sandah, is the same as that of Jolian. A difference between the configuration of structures at Tareli and Mekkha Sandah and that at Jolian might be a line of shrines placed on both sides of the passage leading to the stupa court and small stupas raised on the passage itself in the case of the former. But the concept behind that arrangement might be the same as that behind the group of small stupas plus a line of shrines on a lower terrace at J olian. Since there remained no room for constructing such a terrace at Tareli and Mekkha Sandah, the passage might have been used in place of an extensive terrace. The Tareli stupa court does not seem to have had any enlargement, but at Mekkha Sandah the stupa court was enlarged at least three times, as the number of small stupas increased. So these Gandhara temples and J olian of Periods III to IV share common features as to the stupa court. Another aspect in which Gandhara differs from Taxila is the location of the spacious quadrangular building, which scholars have hitherto interpreted as an assembly hall for monks. In this court they might have assembled for either uposatha or other ceremonies. In Taxila the hall and the cells for monks are separated from each other only by a wall, and entry could easily be made through a doorway. This aspect has promoted the supposition that the hall functioned as a court for monks' assembly. In Gandhara, on the other hand, it is located neither in the area of monks' cells nor close to it. The "hall" is attached to the stupa court or closely connected to the passage leading to the main stupa. In Takht-e Bahi it actually is next to and just to the west of the quadrangular monastic building, but this building has no doorway to the hall. Since the "hall" is only open to the court of small stupas several steps lower than it and the monastery, monks would have had to cross the court of stupas to enter it. The "hall" in Taxila and that in Gandhara thus look similar to each other in ground plan and in function as well, yet if more stress is placed on the fact that in Taxila it is kept close to the monks' life and in Gandhara closer to the stupas, namely, closer to lay-worshippers, it might be conjectured that such a spacious hall was not always used as an assembly court for monks only. Also, still open to discussion is the usage of a square depression in the center of the quadrangular monastic buildings. In Swat and Dir, like in Taxila, there are temples where small stupas are allowed, such as Butkara I, Butkara III, Panr, Saidu Sharif I, Shnaisha, Gumbatuna, Chatpat, Andan Dheri, and Nimogram, while Arnluk Darah, Loebanr, Tokar Gumbad, Darah, and Damkot have no subsidiary stupas. The basic form of building for monks is also the same with that in Taxila: a closed-in quadrangle. Therefore, Buddhist temples in Gandhara must have looked anomalous to the eyes of Buddhists in Taxila and as well as Dir. One thing common to all these regions was that in each there was one location where a gigantic stupa dominated other stupas or temples around it, as did the Great Stupa of Dharmarajika in Taxila, Kanishka Stupa in Gandhara, and the Sacred Precinct of Butkara in Swat. Chinese Buddhist sources, mainly of monk-pilgrims, make no mention of such a central structure in Taxila, but Kanishka Stupa is repeatedly described as Qiaoli Stupa or Bai-zhang Stupa, which Biruni also referred to as Kanik Caitya in the eleventh century. The great temple of Butkara is mentioned by Song Yun in the beginning of the sixth century as Monastery Tuola, as rightly assigned by Tucci. Temples in Gandhara and Taxila were built up with stones, the facing of the walls being made of large stones with the interstices between them being filled with small flat stones placed one on top of the other. Yet while possessing tl1ese common characteristics, each region developed its own original style of masonry work. The sequences of such masonry in Gandhara and Swat have not been clearly established, and any relative chronology of Buddhist temples is still to be discussed. It would be highly hazardous to apply an roughly established, chronological succession of masonries in Taxila to the evolution in other regions. Boldly speaking, while the temples in Taxila followed Indian tradition in some aspects, various original devices were developed. Gandhara and Swat, however, do not seem to have followed anything of Indian architectural tradition, such as griha-stupas, except for the technique of constructing a corbelled ceiling, but laid much stress on what originated in Gandhara or in Taxila. It may also be dangerous to say, even based on the discussion hitherto given, that the temples in Gandhara were on the whole one step behind those in Taxila in a chronological context, when it might be admitted that whether or not they adapted the old styles and systems and retained them depended on regional differences. Lastly, we are to set an approximate chronological framework to the evolution of the Buddhist activities in the Gandhara-Taxila areas. During the period from the latter half of the first century AD to the early second century, Buddhism became vigorously active mainly under the Kushans who tended by character to visualize things rather than engage in deep speculation and for whom, therefore, it was natural to develop the Buddha's figure as carved or as drawn. Then, according to Chinese literary sources, 10 Gandhara of the fourth and fifth centuries was the most flourishing center of Indian Buddhism as reflected in the various descriptions of Chinese monk-pilgrims who crossed the passes between Wa'khan and the mountainous area in the south-down to Gilgit and Darel along the Sindh (or Indus) River-and eventually reached the Swat Valley and farther south. The situation of Buddhism in the northwestern part of the Subcontinent was still unchanged in the late fifties of the sixth century, or throughout the time of the Hephthalite rule, as Narendrayasas and even Song Yun inform. But just after that time Buddhist institutions lost their fame due to the political attitude of the Hephthalite successor, i. e., the West Turks who did not cross the Hindukush southwards. Monastic life in the northwestern Subcontinent inevitably underwent prompt deterioration of support from commercial communities there which had long relied on the long distance trade connecting Gandhara-Taxila to Central Asia under the supports of nomadic empires. The revival of Buddhism in these regions however came under the hegemonies of the local Khingal Dynasty at Kapisi from the mid-sixth to the mid-seventh century and the succeeding Turki Shahis at Kabul down to the second quarter of the ninth century. Since the Turks in the north of the Hindu Kush had friendly ties with Kapisians, this time inaugurated were the caravan routes that started India without going northward fron Gandhara running westward to Kapisi and then into the western of the Hindu Kush to reach Tokharistan. Thus Buddhism in Gandhara-Taxila experienced three main waves: the first to the early third centuries, the fourth to the mid-sixth centuries, and the late sixth to the ninth. 11 II The Square Podium of Stupa in Taxila its Introduction The earlier Buddhist stupa known so far at Sanchi appeared with a hemispherical anda supported by a low circular plinth. This type was introduced somewhere in Northwest like Dharmarajika, Manikyala, and Butkara. Such stupas have successively been enlarged, covering the original core structure which is, round in plan, believed to date back to the Ashokan time. Insofar as the core was made round, the enlarged stupa also had the round plan. In the first century AD in Taxila, however, the stupas took quite a new style, supported by the square podium. This type of stupa was used in Taxila side by side together with the stupas with round podium and later on became the most popular type of the Buddhist stupas in the Northwest, being widely accepted in time and space. The square podium is not a simple box-like cube but architecturally decorated on all four sides, from the very beginning, with a certain number of pilasters flanked the cornice and the bracket on the upper margin and by the torus-and-scoti mouldings on the lower. The root of these architectural elements is apparently not Indian. Each architectural element does not seem independently introduced to be juxtaposed on the square podium. An architectural proto-type, having the square plan and the decorative elements on each lateral face, found itself as a base of the stupa in Taxila. Association of the round stupa with the sqaure architecture of Western character was not taken as alien at all in Taxila. The discussion follows in favour of Roman religious architecture. The "Bactrian Hellenism" had nothing to do here. The first introduction of the stupa with a square podium was in Period II of Sirkap. All of the stupas excavated by Marshall at Sirkap have the square podium except one enshrined in the apsidal temp]e in Block ID and another from Period III in Block IE. The stupas excavated by Marshall are in Blocks lA, 3A, ID, IF, IG, 1C', and IE' as well as in the palace. A large stupa found near the northern city gate is situated in Block IA. All that is left of this stupa is the podium measuring about 10m square at the base around which runs a torus-and-scotia moulding surmounted by a series of square pilasters, seven on each side, above which is a frieze-and-dentil cornice. The core of the structure is of rubble and the facing of the squared kanjur stone is masoned without joint. Three subsidiary stupas, one on the back side and two smaller on the north-west corner, are built of rubble faced with plaster. The largest stupa on the back side is adorned with the torus-and-scotia mouldings along the foot of the base surmounted by three Corinthian pilasters on each side. In the small stupas, blocks of the kanjur stone are let into the rubble walls for mouldings, pilasters, etc., which are blocked out of the soft stone and finished off with plaster (Marshall 1951: 142-46). 12 The stupa in Block 3A, about 1.5 m square at the foot of the base, is built of limestone rubble without any kanjur blocks and finished off with plaster directly on the face of the rubble masonry (Marshall 1951: 146). The stupa standing in the court of Block IF, measuring about 6.5 m north and south by about 8 m east and west, is faced with the squared kanjur stone and its core is made of rubble. Round the base runs a well-cut moulding consisting of a torus and scotia divided by a fillet, above which is a row of pilasters surmounted by brackets with a frieze and dentil cornice at the top. The pilasters on the two and back sides have the square shaft surmo.unted by flat capital except for the central one on the two sides which has a round shaft with Corinthian capital. On the western or frontal side all the pilasters are Corinthian with square shafts, but two pilasters have round shafts. The interstices between the pilasters on this face are relieved by niches of three different patterns. The two nearest to the steps resemble the pedimental Greco-Roman buildings; those in the center are surmounted by ogival arches; those at the corners resemble the form of early Indian toranas. Except for the central niches, the tops of the four niches are surmounted by birds. The double-headed eagle on the southern ogival arch makes this stupa unique (Marshall 1951: 163-64). The stupa enshrined in the small court of Block 1G and facing west to the main street is rather small in size, measuring about 4 m square with a projection of about 2 m on the facade. The pilasters, five on three of its sides, are square, with capitals made up of horizontal mouldings. The moulding round the foot of the base is the usual torus and scotia, on which the bases of the pilasters rest directly without the intervening course which is present in other stupas. The cornice is relieved by a bead-and-reel moulding, but is devoid of dentils. The core of this stupa podium is of limestone rubble, faced with squared kan jur stone (Marshall 1951: 167.) At the south-east corner of Block 1C' is a small stupa surrounded by walls and facing east or to the main street. Of this stupa only the high podium is left. It is constructed of diaper masonry with corners of dressed kanjur. The presence of the fragments of kanjur pilasters leads us to suppose that all four sides were protected by them (Marshall 1951: 191-92). At the south-east corner of Block IE' remains a stupa podium constructed partly by rubble and partly of diaper masonry, protected on the outside by a thick coating of lime stucco, and the core of which is strengthened internally by thick foundation walls arranged both crosswise and diagonally, interstices between them being filled with stone debris. Marshall asserted without hesitation that the plan which is somewhat unusual for the foundation walls was designed to have provided for the weight of a circular stupa drum, the outer edge of which would be supported by the four diagonal walls (Marshall 1951: 73-74). On the main street there is another Buddhist chapel of rectangular plan, measuring 68 m east and west by 40 m north and south, in the center of which is the apsidal temple consisting of a round apse and a square nave. On both sides of the fa<;ade of the apsidal temple were two small stupas of square plan in the ancient time, but only the southern counterpart remains. All that is left of this stupa is a lowest course of the square podium 13 faced with squared limestone which is jointed by small pebbles (Marshall 1951: 154). In the palace there was a small stupa-shrine presumably erected for a private use, but nothing is left except for its square podium (Marshall 1951: 173). The stupas, or actually the stupa podium, in Sirkap are divided into two types from the architectural point of view. One is built of rubble and faced with squared kanjur stone. The other is constructed either of diaper masonry or of both diaper and rubble masonries. Especially the former is also found at other Buddhist chapels in Taxila, among which Dharmarajika is a subject of discussion. The main stupa of Dharmarajika consists of hemispherical anda and round medhi which is supported by the round podium. The core of anda is strengthened internally by thick walls, sixteen in number, radiating irregularly from the central circular core. The central core and the interstices between the radiating walls are built and filled with rough rubble of limestone. Important is the fact that the radiating walls stop short above the berm of stupa instead of being taken down to its foundation (Marshall 1951: 236f). This implies that these walls were erected after the original anda had been reduced to ruin. Furthermore, the rebuilt anda was faced with diaper masonry, a small patch of which is seen on the west side of the drum immediately above the berm. These facts lead one to presume that the diaper masonry and the principle of radiating walls are parallel to the foundation walls of the small stupa in Block 1E', however the latter may be arranged crosswise and diagonally. Around the main stupa remain a lot of constructions which were assigned by Marshall to stupas and shrines. Some of them are masoned of rough rubble limestone, faced with squared kanjur stone. This masonry is anterior to that of diaper since the diaper-masoned wall partly covers the construction of kanjur stone, as seen between stupa D 1 and Chapel D6. Although we have not known how the original substance of the Dharmarajika main stupa was, the diaper-masoned structure covering the kanjur-faced constructions can rightly be assigned to the same horizon as a small patch of remains on the west side of the stupa drum above the berm. The kanjur construction is thus supposed to be the earliest among the buildings at Dharmarajika and the stupas in Stratum II of Sirkap. This relative chronology allows us to reconstruct an earlier plan of the stupa court at Dharmarajika; the main stupa was surrounded by nine stambhas, arranged at almost equal intervals, such as DI, D3, S8, S9, B20, D10, and three stambhas between S9 and B20, two of them being lost and one of which not designated by Marshall. The stupas built in the same technique as the stambhas were built one step later than the stambhas since they disturb the regularity of the arrangement of stambhas (Marshall 1951: Pl. 45). This successive sequence at Dharmarajika gives the stupas in Sirkap a position closely related with Dharmarajika: the stupas built of rubble and faced with the kanjur stone belong to the same building activity as the earlier phase of Dhamarajika. The pilasters decorating the sides of podium are exclusively carved out of a block of kanjur stone and surmounted by flat capitals in most cases and by the degenerated Corinthian ones supported by round shaft in some cases. The Corinthian capitals can only be found at the stupa IF. Marshall supposed that the earlier capitals, pilasters and 14 mouldings in Taxila had appeared as an easternmost offshoot of Hellenistic cultures which had overcome the Indian tradition in Taxila. In this connection a discussion follows how far Bactria was responsible for the activities of building such monuments as decorated with the 'Hellenistic' capitals now in question. 2 Archaeological information on Balkh-Bactra, the Bactrian phase of Balkh, have still remained limited, although Foucher and Schlumberger insistently endeavoured to shed light on it. Against all expectations of those who are concerned with the Bactrian problems, a vast Hellenistic city Ai Khanum suddenly appeared before us some 300 km east of Balkh and has become the object of annual digs since 1965 of DAFA . In the Amu Valleys two types of the pillar foundations, made of limestone, had been known even before the discovery of Ai Khanum, but curious was the fact that the bases for pillars had never been found together with any stone columns. The columns, made of the same material with capitals and bases, found at Ai Khanum, therefore, enlarged the scope of architectural aspects in that area, not only for the period of Bactria but also for the later times. Importance lies in the palace or the administrative quarter of Ai Khanum which comprises the building complex interrelated by corridors in the north and the extensive rectangular courtyard in the south. The courtyard is surrounded on four sides by the porticoes to which one may enter from north through a propylaeum consisting of a narrow passage between two esplanades. The inner esplanade boasts two Corinthian columns. On the south this peristyle opens to a hypostyle court or the eighteen-pillared hall surrounded on three sides by massive walls built of mud bricks. In this peristyle all columns, made of white limestone, are supported by the pillar bases of two types, one being the Attic base and the other Persian or Achaemenian base, and surmounted by the two kinds of capitals, one being more properly Corinthian and the other 'pseudoCorinthian' as called by Bernard (Bernard 1968: 111-51). At the south side or the inner front of the propylaeum the Persian bases were found in situ and they are properly supposed to have supported the pseudo-Corinthian capitals which had been fallen down to the floor. The colonnades of the peristyle had been plundered and revealed, when unearthed, in huge piles of the fallen drums and capitals. Each of the east and the west colonnades originally had twenty-four columns and that of the north and the south thirty-four. Thirty-one pillar bases on the south portico found almost in situ are of tl1e type called the Attic base which has the high and shallow scotia, while those of the east portico have the higher scotia. The capitals of these columns are same in type as those of the propylaeum. The pillar bases of the hypostyle court are more close to the Attic base and the capitals found in that court are all the more close to the proper Corinthian, sharing some common characteristics with those of the propylaeum of Breuterion at Miletus, of Olympieion at Athens, and found at Nisa, as Bernard (1968) properly pointed out. 15 In addition, the Attic base and the Persian bases were found in the temple with redans. The inner cella of this sanctuary used two Attic bases when it was reconstructed in the last but one stage. The pillar-capital in this stage was made of timber and Ionian type, and no stone drums were found. The colonnade of the south portico of this sanctuary, which limits the sanctuary in the south, was supported by the Persian bases (CRA! 1969: 327-353; 1970: 317-339; 1971: 414-431). In the later stages of Ai Khanum the capitals and the drums do not seem to have been made of stone despite that the bases carved of stone were usual. This trend was succeeded to Surkh Kotal. Temple A of the fortified sacred precincts of Surkh Kotal consists of a square cella surrounded by the corridor on three sides and supported by the plinth measuring 27m by 34m. In the center of the cella is a square podium, at each corner of which was found the huge limestone pillar base of the Attic type, about 90 cm in diameter and 60 cm high. But any stone capitals and drums were found in this cella. The bases of pilasters decorated with the torus-and-scotia mouldings carved out of the same material as the pillar bases were let into the mud brick walls of the cella without any capitals and shafts blocked out of limestone. The outer face of the podium, however, was faced with the limestone blocks intervened by the square pilasters of limestone supported by the Attic type bases and surmounted by the rectangular capitals which are carved with shallow relief depicting the acanthus leaves that recall the Corinthian ancestor. On this podium were the colonnades which surrounded the outer faces of the cella. Of the colonnades only their bases made of limestone were found. Same is the case with the pillar bases of the colonnades standing along the inner sides of the fortification walls, only sixteen bases of the degenerated Attic type now being left (PBA 77-95). After the conflagration the cella was filled up with mud to a level, 2 m higher than the original floor, on which was made a new floor supporting the four smaller Attic bases. A fortified square city, Dilberjin, in the desolate plain about 40 km northwest of Balkh and about 20 km north-east of Aq-chah, has been a main aim of excavations of the Soviet-Afghan archaeologists headed by Kruglikova since 1970. Along the inner side of the city wall was excavated a temple, which in the first stage had essentially the same plan as Temple A of Surkh Kotal. A pillar base found in situ in a deep sounding implies that the temple was surrounded by colonnades at the first stage. The building is believed to be later changed in character to a Buddhist sanctuary which was attached with the walls that surrounded the earlier building to make a room at each end of the facade and the corridors on the other three sides. In this stage some stone pillar bases of the Attic type were set on the fa~ade between the rooms. After the destruction of the long passage built throughout the city wall, many small rooms were constructed instead of the corridor along the north wall. The easternmost seven rooms Nos. 9-15 had a portico in front of them. From here neither capital nor drum was found, but thirteen bases for pillars were uncovered, five of them being of the Attic type and the others being simple blocks of limestone and of baked bricks (Kruglikova 1974: 16f. and 49f.). The Japanese team excavated two sites in Kunduz each representing two types of the mounds in this region. The very low mound, Durman Tepe, revealed eight pillar bases of limestone. All of them were found as settled in the rooms and some of them 16 were accompanied with rotary quem. Three bigger bases were found with regular intervals between them in a room. They seem to have been used as the bases for columns, but the size and the shape including thetorus-and-scotia mouldings are different and do not show uniformity. Moreover four of the eight bases do not take any regular mouldings; the deep horizontal lines are incised in place of mouldings. The other site Chaqalaq Tepe is a fortified habitation, oval in plan, some 2 km south of Durman Tepe. There were found two bases of pillars on a floor of Room i. This floor had been made by filling the burnt original floor with charcoal and soft mud. Though found in situ along the central long axis of the room, they represent two types, the Attic and the Persian types (Mizuno 1968: 45-47; Mizuno 1970: 97-101). In the north of the Amu Darya, Khalchayan and Kara Tepe also had only pillar bases, without capitals and drums. Those of Khalchayan are mainly the Persian type and the Attic bases are few, while in Kara Tepe the bases are similar to the degenerated types of Durman Tepe (Stavisky 1972: PI. I; Pugachenkova 1966: 132). Only Ai Khanum employed the pillars carved out of limestone from the capital through the column to the base, although the designs of capital had already taken a local style. Presumably the usage of such pillars made of stone were confined to the main imposing buildings of the earlier stage. The Ionic capital with the Attic base of stone was carved out of wood in the temple with redans. After the construction of the peristyle courtyard, technique to make and build stone pillars seem to have come to a standstill. But in the periods later than Ai Khanum in the Amu valleys the Hellenistic flavour remained impressive only in making pillar bases as still seen at Surkh Kotal. After the Kushan period even the stone base of pillar ceased to be made but seems to have been reused by bringing it from some ruins of earlier periods as Durman Tepe, Chaqalaq Tepe and Dilberjin may indicate. 3 In contrast with the north of the Hindukush, the Northwest did not share the common trend with the Amu Darya valleys. Two buildings at Taxila, Jandial C and Mohra Marialan, are characterized by their Western aspects which however were not succeeded to the later architecture in that region. Especially the pillar bases which were very popular in the Amu valleys cannot be found in the Taxila-Gandhara regions except for the above two temples. The recent excavations at Tareli revealed that only schist slabs had been used for the bases of pillars which might be of timber (Mizuno and Higuchi 1978: 152). The Hellenistic architecture could not plant the roots in the soil of that regions despite the Greek advent and hegemony. Jandial C and Mohra Marialan were exceptional and heterogeneous to the architectural history of the Taxila-Gandhara regions: the capitals employed in both temples are of the Ionic type and the pilasters used on the stupas were surmounted by the Corinthian capital from the very beginning. These two monuments do not seem relevant to the later develpoment of Gandharan stupas supported by square podium. 17 The temple unearthed by Cunningham, Mohra Marialan, has an oblong plan consisting of naos and pronaos which opens on the longer side. Two pillars were found in the pronaos and four in the naos, the bases being of the Attic type made of sandstone, and the drums supporting the Ionic capital being made of the kanjur stone (ASIR V, p. 70f). Jandial C, oblong in plan and opening to the shorter side, consists of four parts such as pronaos, naos, opisthodomos and a solid structure between naos and opisthodomos, all of which are surrounded by a wall. At the fa~ade of pronaos two round pillars were standing free between the square pilasters which were placed at the ends of the side walls (Marshall 1951: 222 f). As Bernard pointed out, the decorative structure of the Corinthian capitals at Ai Khanum are of transformed types like that of Olympieion at Athens. The style of the square capitals of pilasters at Surkh Kotal in the Kushan period, where the capital of free-standing pillar did not existed, is supposedly intended to be Corinthian since they are composed of the acanthus leaves as the main decorative element. When the Corinthian capitals were later adopted at the Buddhist stupas where the acanthus leaves were usual, the core structure before applying the plaster was masoned with thin slabs of schist to give it a rough shape of acanthus in order to make it easier to carve acanthus leaves when the plaster covered the core. On the contrary, the core of the capitals of pilasters let into the sides of the square podiums in Sirkap takes a flatter trapezoidal shape of the dressed kanjur stone. This shape might not be expected to be Corinthian even when the lime plaster was thickly applied and carved. The existence of any Corinthian capital could not be expected when the stupas with square podium first appeared in The above discussion therefore leads us to suggest that the Bactrian Hellenism in architectural term had nothing to do with the appearance of Buddhist architecture in Sirkap. The core of anda of the main stupa at Dharmarajika is internally strengthened with the thick walls which radiate from a circular solid structure in the center core. This device principally corresponds to that of the podium of stupa in Block IE', not only confined to stupas in Taxila, but also found at Shahji-ki-Dheri (ARASI 1908-9, p. 48) and at Fil Khana, J alalabad, on the foot of which a cave monastery are hewn out (Mizuno 1967: 76). In each case the spaces between each radiating wall are filled with rubble. Generally the stupas in the Northwest are built with rubble as its core with or without mud joint except for the main stupa in Butkara which is constructed exclusively with smashed schist slabs. The radiating walls therefore were not always necessary and usual for the massive imposing Buddhist stupas. The radiating walls was not unusual for large-scale tombs or mausoleum in the Roman world. There the monumental tombs built of stone or baked bricks had the cylindrical drum supported internally by the radiating walls. Typical is one of the graves on the Via Appia with the walls radiating from the central circular cella which took the role preventing the collapse of rooms inside from the pressure of soil. The imposing kind on this line is the mausoleum of August, built in BC 28, measuring 84 m in diameter and consisting of the many concentric walls divided into many rooms by twelve radiating walls (Robertson 1943: 265-66). This style and technique of these monuments was brought westwards to Britania as 18 the Roman Empire expanded. The tomb at West Marsea in Essex, 20 m in diameter, was built of the thin baked bricks jointed by mortar on the paved basement of the chipped stone and was internally strengthened by six radiating walls from the central hexagonal cella, the outer circular wall being supported by the twelve projections (Collingwood and Richmond 1969:170-71). The stupa drum strengthened by the radiating walls in the Northwest is a remote echo of the Roman ancestor. The similar structures also found in the Buddhist sanctuaries at Nagarjunakonda are closer to those of the homeland as I think that a cella was placed in the center of construction at Nagarjunakonda and that the Northwest architects do not seem to have paid attention to the primary meaning of this structure without making any central cella and the concentric walls. The same examples as those in the Northwest have also been found in Mathura and Sanghol. This device appeared with the diaper masonry in Taxila since the stupa of this device in Block IE' was built with diaper masonry and the outer face of the drum strengthened internally with the radiating walls at the Dharmarajika main stupa was repaired with diaper masonry. Marshall maintained that the diaper masonry had first been introduced to Taxila after the earthquake in 25 as a technique which allowed to make the building much more stable than that of random rubble masonry. If Marshall be accepted, the square podiums, the cores of which being filled with rubble and the wall being faced with the dressed kanjur stone decorated with pilasters were supposedly introduced before 25 and the technique of building the radiating walls were used in Taxila sometime after 25. However, nobody knows that the "earthquake" really happened in 25. Readers may refeer to Chapter IV for further discussion on the wheel-shaped structure within stupa. Presumably something like architectural ideas had already extended to the Northwest from the Roman world in the early days of the first century AD. The square podium of stupa is characterized by the aforesaid ornaments, not a simple box-like square shape. No tradition in Taxila of brewing the Western decorative ideas and no clue of taking the Bactrian Hellenism as responsible for the birth of square podium. We have no positive evidence to suppose that the idea of square podium and that of pilaster were independently introduced to Taxila and that each elements were associated together in Taxila, but that an original type of the square monuments decorated with pilasters was introduced as religiously equivalent with stupa. A prototype, square in plan and decorated with pilasters, must have been supposed at that time to be justifiable even if it were associated with stupa. 19 III The Buddha s Bowl in Gandhara I Evidence to date otherwise undatable trends of Gandhara Buddhism can be drawn out from more cautious search of the Chinese sources, Buddhist or secular, which often date the year of publication. Based on such documents, this paper exclusively devotes itself to the Buddha's alms bowl which Siddhartha was given by the four guarding deities on the first meal after the great Enlightenment. The Chinese sources unanimously record that such patra existed in Gandhara. Why was such sacred relics in Gandhara? The paper aims to depict it as it was, as clearly as possible, with special reference to its role in the context of the popular Buddhism in Gandhara while emphasizing dating certain Buddhist activities. At the beginning, allow me to repeat a point that I have touched on in Chapter VIII: a drastic change occurred around the middle of the sixth century of the routes connecting the Northwest with Central Asia and China. Before the middle of the sixth century, the routes had passed through the mountainous regions north of Gandhara to Wa'khan, and then led to Chinese Central Asia in the east and to Tokharistan through Badakhshan in the west (Kuwayama 1987:703-727). The routes, which one may call the Karakorum routes, seem to have been secured under the successive rules of the nomadic kingdoms such as the Kushans and the Hephthalites. The chapters for the Western Barbarians in the dynastic histories of China describe these people as having subjugated certain of the countries lying to the north of Gandhara.(l) This opportunity had long since caused Gandhara to prosper as a commercial station, which had supported the Buddhist activities. Around the middle of the sixth century, however, the discarding of the routes occurred at the same time as the gradual decline of the Hephthalites in both Tokharistan and the Northwest. The trading routes drastically shifted westwards, and led Kapisi as the entrepot at the southern foot of the Hindukush and Bamiyan as a transit point in the mountains to flourish. The kingdom of Kapisi almost suddenly took the role of Gandhara, as did Takkadesa in the Punjab. The new highway ran to Tokharistan and farther on to Sogdiana. To get to Chinese Central Asia and China, however, one had to proceed from Tokharistan through Badakhshan and Wa'khan. Before the change of the routes, Gandhara had been a centre of Buddhism in India, to which not a few Chinese monks had made pilgrimage and even Indian devotees had proceeded from various parts of the Subcontinent. During the fourth and fifth centuries in particular, Chinese pilgrimage to India blossomed forth in response to the unprecedented arrival in China of Indian monks: eighteen Indian monks came to China, or twice the number that had come in the third century. As for Chinese pilgrimage to India, ninety-six monks are named in the Gaoseng zhuan, which was edited by Huijiao 20 until 519, and in the catalogues of Buddhist scriptures housed in the library of each court which were compiled in each dynasty (Kuwayama 1982B: l33-l39). More than half of those leaving for India and for China had close connections to a country of the Northwest, which is exclusively called in the Gaoseng zhuan Jibin, a focus of scholarly discussion since the last century. Some Indians were born in Jibin, while others came from other parts of India in order to pay hommage to the so-called relics of Buddha and to learn Buddhist philosophy. It is quite usual to see that Chinese monks proceeded to Madhyadda after they stayed in Jibin(2) with the pious purpose to see the sacred places of Sakymuni. Yet they do not seem to have studied Buddhism in any other parts of the Subcontinent other than Jibin. The Jibin country may therefore be emphasized as a centre of Buddhism in the Subcontinent throughout the fourth and fifth centuries. 2 This centre of Indian Buddhism was famous for the alms bowl, which various Chinese monks worshipped as stated in their biographies in the Gaoseng zhuan. The following are some of the quotations from it. (l) The biography ofZhimeng says (T. 50: 343b-c; 55: 113c-114a): He arrived in Jibin (leaving China in 404), and worshipped the Buddha's bowl there. He saw the Buddha's spitting vase of decorated stone in Jisha on the Onion Range (Tashkurghan on the Pamir), and in addition, in this country of Jibin he saw the Buddha's bowl. The Buddha's bowl was lustrous and purple-blue in colour and had the clear mark of the four rims. He worshipped it with incense and flowers, wishing that the bowl might become either heavier or lighter in response to his pious soul. Then the bowl he was carrying with his hands suddenly became too heavy to endure, so that he could not help putting it down on the table. When he did so, however, the weight of the bowl went to nothing. Thus the bowl showed sympathy with his soul of devotion to Buddha's faith (3). (2) The biography of Dharmavikrama says (T. 50: 338b-339a): He also worshipped the Buddha's bowl in Jibin, and stayed there a year to study the Indian language and books. (He is a Chinese monk who left China immediately after 420). (3) The biography of Huiguang says (T. 50: 337a-b): Earnestly searching for the last one-third of the Parinirvanasutra, he wrote to the first emperor of the Liu Song dynasty, saying that he needed financial 21 support for a mission to the Western Regions consisting of ten officials headed by a sramana Daopu. Daopu, who was a Turfanese in origin, had been in the Western Regions, travelling extensively and worshipping the shadow of the Buddha (in Nagarahara), the Buddha's bowl (in Jibin), the Four Stupas (in Buner, Charsada, Taxila and somewhere between Swat and Taxila), the Tree of the Enlightenment (in Magadha), the footprints of the Buddha and the carved image of Buddha. (4) The biography of Huilan (T. 50: 339a) says that Huilan also was one of the worshippers ofdle Buddha's bowl in Jibin (before 439). (5) The most detailed account of the Buddha's bowl is given by Faxian himself, who is supposed to have been in Purushapura in 402. The account is divided into two parts: the first half is a legend of the origin of the temple which housed the Buddha's bowl and the latter is a description of his observance of the bowl: The Buddha's bowl is in this country. Formerly a king of Yuezhi raised a large force and came to invade this country, wishing to carry the bowl away. When he had subdued the country, the king ofYuezhi and others had sincere faith in the Buddha's law and wished to take it away. Accordingly held religious services to Triratna on a large scale. When they had done so, they had a large elephant grandiosely decorated, on which they placed the bowl. But the elephant knelt down on the ground, and was unable to go forward. Again they made a four-wheeled cart, in which the bowl was placed, and made tight elephants draw it, but neither were they able to go forward. Then the king knew that the karma-relation between himself and the bowl had not arrived yet, and was deeply ashamed of himself. Forthwith he built a stupa and a monastery at this place, and left guards to watch the bowl, making all sorts of dedications. The monastery is capable of housing about seven hundred monks. When the sun is about to be in the middle of the sky, many monks bring out the bowl, and hold various kinds of religious services to it along with the laity, then take their midday meal. When they offer incense at the usual rite of sunset, they again do it in the same manner. The bowl can contain about two dou (four litres), and is of various colour, black predominating, with the four rims clearly marked. Its thickness is two fen (about half a centimetre), and it has a bright and glossy lustre. When poor people throw a few flowers into it, it becomes immediately full, while some very rich people, wishing to make offerings of many flowers, might not stop till they have thrown in hundreds, thousands, and even myriads of bushels, and still will not be able to fill it. (A pardy modified version of Legge 1886: 34-35). 22 When comparing this version with the bowl seen by Zhimeng, one can get an impression that the bowl they saw in Jibin and Purushapura respectively was one and the same object in that the lustrous colour is rather dark, purple:"blue or blackish, and the four rims are clearly discernible. Exclusively based on the routes taken by both Indian and Chinese monks, I have elsewhere proved that the place name Jibin as used by Huijiao in the Gaoseng zhuan is not Kasmir but Gandhara (Kuwayama 1987: 708-712). Now this solution is reasonably clear, even apart from the examples of Jibin in the Gaoseng zhuan.(4) As discussed later, the Buddha's bowl as the object of worship had to be the relics that could exist nowhere but Purushapura in Gandhara. More information about the Buddha's bowl corne from other monks, whose complete narratives are missing but remain in part quoted in later books, i.e. the Shuijing zhu (Commentary on the Water Classic by Li Daoyuan, d. 527) and the (an encyclopedia in the early Tang edited between 622 and 624) as well as the Taipingyulan edited in 983. (6) The Shuijing zhu, ch.2 (on Purushapura): Futu says, "The Buddha's bowl is made of blue jade. It contains about three dou (six litres). The people of the country take it to be a precious object. At the time of holding religious services, if one wishes, offering incense and flowers all day long, that the bowl may not become full of flowers, the bowl will respond as one has wished. If one wishes to make it fill with only a bunch of flowers, it will also respond as one has wished." (7) The Yiwenleiju, ch. 73 (Section of bowl), quotes the paragraph from the Xiyu zhuan (Memoir of the Western Regions). The cited paragraph actually originated in the Zhuguo zhi (Record of Various Country). The Zhuguo zhi says: The Buddha's bowl is in the country of Gantuowei and made of blue jade. It contains about three dou. The people of the country take it to be a precious object. At the time of holding religious services, if one wishes, offering flowers and incense all day long, that the bowl may not become full of flowers, the bowl will not become full as one has wished. If one wishes, offering only a bunch of flowers, that the bowl may become full, the bowl will also follow one's wish. (8) The Taipingyulan, ch. 759 (the bowl), quotes the following paragraph from the Xiyu Zhuguozhi (Record of Various Western Regions): The Buddha's bowl is in the country of Gantuowei and made of blue jade. It contains about three dou. The people of the country take it to be a precious object. At the time of holding religious services, if one wishes, offering flowers and incense all day long, that the bowl may not become full of flowers, the 23 bowl will not become full as one has wished. If one wishes, offering only a bunch of flowers, that the bowl may become full, the bowl will also follow one's wish. The similarity of this text to No.7 is unquetionable. Comparing the name of the book Xiyu Zhuguo zhi with the two different books cited in the Yiwenleiju as the Xiyu zhuan and the Zhuguo zhi, it is probable that the title Xiyu Zhuguo zhi may have imprudently been confused at the time of writing or editing the Taipingyulan or in the text which the editors of the Taipingyulan consulted. A personal name seen in No.6, Futu, is unclear. In some editions the name is given as Fututiao and Futucheng. The identifications of the names with the famous monk (died in 348), whose name was restored by Pelliot (1903:100,1912:419) should be left aside, because we know many names of this kind liable to be confused such as Futu, Futucheng, Fututiao, and Zhu Futiao. Also unidentifiable is the book Zhuguozhi of unknown authorship (Petech 1950:7, fn. 2). The sentences common to all these quotations might give a plausible explanation that Futu is the author of the Zhuguozhi, but whether they come from one and the same source is beyond our scope here. The only difference one can easily pick up from Nos. 6, 7 and 8 is an added phrase seen on No.7 which says that the bowl is in the country of Gantuowei. Gantuowei is a transliteration of a Prakrit form *Gandhavai, derived from an earlier *Gandhavadi which corresponds to Sanskrit Gandhavati (Petech 1950:58. Pelliot 1914:393, fn.). Faxian uses it in his narratives on the travels as indicating Pushkalavati, but on the other hand, he clearly saw the Buddha's bowl in Purushapura. As the below-mentioned text show, it would be better to suppose that Gantuowei or Gandhavati in the above quotations points to not a city but to a more extensive area including Purushapura, which we call Gandhara. More clues about the Buddha's bowl derives from the Shuijing zhu, which also quotes other passage from two different source books. One is Zhu Fawei's writing which, according to Petech (1950:6), is called the Fuguo ji (Record of Buddhistic Kingdoms), and the other is the Shishi Xiyu ji (Record of Buddhistic Western Regions) compiled by Dao-an (314-385). (9) The Shuijing zhu, ch. 2 (on Purushapura): According to what Zhu Fawei says, the Buddha's bowl is in the country of Da Yuezhi. A futu is standing there. The height is thirty zhang (about 72 m.) and it has seven storeys. The bowl is on the second stage. The bowl which is entwined with golden threads is suspended by a chain (The original Chinese sentence may otherwise be interpreted as that the bowl is suspended by a chain with gold threads entwining). The bowl is made of blue stone. It is said that Subhuti has put it on the golden table. The Buddha's footprint is at the same place together with the Buddha's bowl. Also in this case, another similar version is found in the Yiwenleiju, ch. 73: (10) The Yi71len1eiju quotes the source book entitled the Waiguoshi (Matters Concerning the Foreign Countries) written by Zhi Sengzai: The Buddha's bowl is in the country ofDa Yuezhi, otherwise called Fulusuoyue (Puru~aval) that means the royal city of devaputra. A futu is standing there. The futu is four zhang (about 9.6 m) high and has seven storeys. On the walls of all four sides are enshrined golden and silver images of the Buddha, the size of which is as large as a man. The bowl is in the centre of the second stage. The bowl is entwined with golden threads, being suspended by a chain. The bowl is made of stone and the colour of it is blue. We do not know who is earlier, Zhu Fawei or Zhi Sengzai, and whether they are based on one and the same source or not. The only clue for dating the above records Nos. 9 and 10 is the order of the biography of Zhu Fawei, where his biography is located, not the content of the biography itself which only mentioned the fact that he was one of the travellers to the Buddhistic countries. This means that his biography is attached to the end of those Huiguang and Daopu (Chavannes 1903 :437 and see tl1e text No.3). So it would be reasonable to expect that Zhu Fawei was their contemporary in the latter half of the fourth century and in the early fifth century. The futu deserves note. This term of architecture used in China in the Six Dynasties period had always been found to be a stupa. But in this context it cannot be that since a stupa in Gandhara, solid from top to bottom, was not used for taking out the objects which have been deposited in it. As Faxian tells in the text No.5, the bowl used to be brought out in order to be paid hommage. Besides, any stupa cannot have a room spacious enough to enshrine the images. We know that multi-storeyed tower-like buildings were prevalent in Buddhist temples throughout China in the fifth century, while the archaeological records from Gandhara do not support the existence of tower-like buildings within holy precincts. The claim that the futu had seven storeys cannot be taken literally, but this wording was given by Chinese Buddhists who well knew multi-storeyed architecture in China. In this connection important is a reference to Xuanzang's and Daoxuan's descriptions of the building that housed the Buddha's bowl. (11) In Chapter 2 of the Da Tang Xiyu ji Xuanzang explains what he saw in Purushapura. In the northeast inside the royal city there remains a foundation of the building which in the old days housed the Buddha's bowL Xuanzang says nothing of stupa. Daoxuan also quotes the above sentence in order to account for the building for the bowl, when he edited a book on Buddhistic Geography entitled the Shijian Fanzhi. The chapters in this book, called the Buddhist Remains, exclusively depend on the Da Tang Xiyu ji, and seem to be an abridged and re-arranged 25 edition of the Da Tang Xiyu ji. The section about Gandhara says that the remain of a building in the northeastern part of Purushapura is a foundation of the building for the bowl. The same author also refers to this building in the biography of Xuanzang with which he supplemented the first edition of the Tang Gaoseng zhuan in 648-649. Daoxuan says that in the city of Purushapura there was formerly a bomiao decorated with various things. The bo is a bowl and the miao is a shrine. The bomiao means a shrine for the bowl. We do not know why he used such a term particularly in the biography and whether he could consult the Yiwenleiju or not, yet the bomiao apparently tells that the building for the Buddha's bowl in Purushapura was not a stupa. No evidence remains for the multi-storeyed building. Rosenfield (1967:223) proposed a view, in relation to Faxian's record on the bowl, that the bowl had been enshrined in the Kanishka stupa in Peshawar, although admitting that the tower built for the bowl was inside the royal city and that the Kanishka stupa was located to the southeast of the city itself. The location of that famous stupa had constantly been described by Chinese pilgrims to east of the city, but only Xuanzang placed it to the southeast. Even Daoxuan placed it to the east of the city in the Tang Gaoseng zhuan. No confusion of its location is therefore found among the sources. Faxian says not that a king of Da Yuezhi enshrined the bowl in a stupa, but rather that he built a temple for the bowl (See the text No. 50.). Hence no evidence lends support for Rosenfield. The Shuijing zhu quotes a passage from the Shishi Xiyu, which Petech (1950:60) interpreted as that "the Shishi Xiyu ji says that to the Northwest of the royal capital of Jiantuoyue (Gandhara) there is the city of Botuluoyue which is the royal of the robe of the Buddha". It seems to me that some words might be left out from between the words botuluo and yue and that the word botuluo should be equivalent with Sanskrit patra. Hence another possible interpretation that to the Northwest of the royal capital of Gandhara there is the botuluo and that the city of [.... ]yue is the royal city of the robe of the Buddha. Other interpretation is also possible that to the Northwest of the royal capital of Gandhara there is the bo which literally means the bowl and that the city of Tuluoyue is the royal city of the robe of the Buddha. In this case the tu of Tuluoyue makes no sense, and should have been replaced by an original Chinese character xi, since Xiluoyue is very close to the town called Xiluo (hi<;l<;la>Hadda) in the Da Tang Xiyu ji that was well known for a shrine housing the robe of the Buddha. Daoan, the editor of the Shishi Xiyu ji, is not a traveller to the Northwest but is known as an eminent disciple of Futucheng. The Gaoseng zhuan (Vol. 9) says that Futucheng has been twice in Jibin before his arrival in Luoyang in 310. If Futucheng were an informant to Daoan, there would be possibility that Daoan learned about the bowl in Gandhara from him and therefore that the Buddha's bowl must have already been in Purushapura in the third century. If the above interpretation is right, it is hard to explain why the location of the bowl differs between Daoan and Xuanzang. To answer this problem a little more sensibly, a reference should be made to the description by Xuanzang of the location of the capital of Bamiyan: Xuanzang describes that the western grand Buddha statue is located to the 26 northeast of the royal city of Bamiyan. In fact, no important remain exists in the area southwest of the western Buddha insofar as Xuanzang describes that the royal city of Bamiyan occupies over precipitous cliffs and across valleys, covering six or seven Ii in length and backed by high precipices on the north. However, in fact, no remains have ever been found to the southwest of this Buddha image. T uming our eyes from the main cliff of the Bamiyan valleys to the southeast, one could find a huge mass of ruins on the top of a hill, difficult of access, which might be a city of the Bamiyan kingdom, that is, Shahr-e Za'hak. To this site in particular, the Arab geographers seem to have referred, saying that the town of Bamiyan has not a hisar (fortification) but is on the mountain in front of which a river flows (Istakhri, Tehran 1340: 220; Hawqal: 328; Ya'qubi: 288-289; al-Fida: 466-467). If it is right to identify the site of the royal city of Bamiyan with Shahr-e Za'hak, the possibility that Xuanzang took the Northwest for the northeast at the time of editing would turn out true, when one examines how Xuanzang used the northeast in the Da Tang Xiyu ji. The above fragmentary passages fortunately quoted in two or three different texts give a picture of the Buddha's bowl in Purushapura. In the northern part inside the city there was a temple which was said to have been built for the bowl by a king of Da Yuezhi. The temple consisted of a stupa and a monastery, capable of housing a great number of monks, as well as a shrine of more than two storeys. In the centre of the second stage was the Buddha's bowl. The bowl, probably covered with golden threads, had been suspended by a chain, but seems to have later been put on the pedestal. On the four walls of this room were carved images of Buddha or Bodhisattva decorated with golden and silver thin leaves. The bowl itself was made of stone, the colour of which was dark in blue but lustrous. Judging from a few texts, which say that it was of jade, it seems to have been made of precious stone such as lapis lazuli. But none of the texts uses the Chinese term for lapis lazuli, which had long since been known to China. The thickness of the bowl was half a centimetre, rather thinner than one expects, given that the capacity ranged from four to six litres. 3 The disproportionate thinness is presumed to have been derived from and closely related to the very origin of the bowl itself. As Zhimeng and Faxian clearly say, the bowl had the four rims which had made it unique. The legend that the Buddha received the bowls from the four lokapalas at the first meal after the Enlightenment and fused them into one is variously told in both Indian and Chinese versions of the sutra praising the legendary life of the Buddha, such as the Lalitavistara, Buddhacarita, xv, 64, the Mahavagga, i, 4, 4, and so on. Among the Chinese versions still accessible are the Taizi Ruiying Benqi jing (T. 3: 479a-b), the Puyao jing (T. 4:28c), the Guoqu xianzai Yinquo jing (T. 3:643b), the Fusuoxingzan (T. 4:28c), the Wufen lu (T. 22:103a) the Fubensuoji jing (T. 3:80la-802a) and the Fangguang Da zhuangyan jing (T. 3:601c-602b). 27 (12) The earliest text is the first of them, which is believed to have been translated into Chinese by Zhiqian between 222 and 253. In this the four rims of the bowl are clearly explained: The Buddha had been in quite a purified state for seven days after the Enlightenment, when the tree god thought to have someone offer the meal to Buddha who had remained seated since the Enlightenment without eating. Five hundred merchants came out of the hill and were going to pass by, when the tree god thought about this. But the yoked bulls stumbled on to their knees and could not go forward. Among the merchants there were two men of virtue whose names were Trapusa and Bhallika. Both of them feared this accident and played together with others to the tree god for their safe departure. The god, emerging with his bright figure, said to them, " The Buddha now came to this world, but has remained seated without a meal on the Nairanjana river in Uruvilva. Fortunately you are good enough to offer the meal. If you do so, you will be given much happiness in the future." Hearing the name of the Buddha, all of them were delighted and said to each other, "The Buddha must be an unrivalled man of holiness, because even the tree god pays respect." Then they mixed up rice flour with honey, proceeded to the tree under which the Buddha was seated, and offered the meal in veneration. The Buddha thought that the Buddhas in the past had always received the offerings from men in a bowl and it would be against the custom to receive them with hands as did the heretics. Perceiving that the Buddha was going to use a bowl, the four lokapalas hurried to the top of the Vinataka hill within an interval in which one could stretch and bend his arms, and made the four bowls come out of a rock with their highly spiritual will. Each of them brought a bowl with him, and offered it with the intention that the merchants might get happy. Considering that taking only one of the bowls from only one lokapala might offend the others, the Buddha received all of the four bowls, put them on the left palm, passed over the rims, and fused them into one. And as a token of the fourfold composition he left the seams of the four rims. The Buddha's bowl as an object of worship in Purushapura was the sacred relics that the Four lokapalas had dedicated, and in addition to the holiness, it had to look like one bowl despite the fact it was really four bowls. The thickness that Faxian tells of seems to be derived from emphasizing the only one despite the four. The Wuftn lu and the Lalitavistara, however, do not refer to the four rims, while the Puyao jing clearly explains why the Buddha showed the trace of the fourfold composition. In this, which is believed to have been transposed before 290 by Dharmaraksha, the personal names of the guardians are cited as Dhritarashtraa, Virudhaka, Viruraksha and Vaisramana, who offered in this order the bowls of blue stone found by them on the 28 Vinataka hill at the time of the midday meal. The dedication of the first meal by two merchants and the offering of the four bowls by the lokapalas are also related in the chapter 9 of the Da Tang Xiyu ji. (13) The English version of Beal (1884: 129-130) is given with some modifications as follows: By the side of the place where Bodhisattva received the rice milk is a stupa where two wealthy men offered the roasted rice flour mixed with honey. The Buddha was seated with his legs crossed beneath a tree, lost in contemplation, experiencing in silence the joy of Enlightenment. After seven days he aroused himself from his ecstasy. Then two merchants passing by out of the woods were addressed by the god of the woods thus: "The royal prince of the Skya family dwells in this wood, having just reached the fruit of a Buddha. His mind fixed in contemplation, he has for forty-nine days eaten nothing. By offering to him whatsoever you have you will reap great and excellent merit." Then the two merchants offered some roasted rice flour and honey from their travelling store. The Bhagvat, or the World-honoured, accepted and received it. By the side of the place where the merchants offered the rice flour and honey is a stupa. This is the place where the four lokapalas presented the bowls. The merchants having already offered the rice flour and honey, the Bhagavat thought over to himself in what vessel he should receive it. Then the four lokapalas coming from four quarters, each brought with him a golden bowl and offered it. The Bhagavat was in silence and did not accept it, thinking that this vessel became not the character of a hermit. The four lokapalas, casting away the golden bowls, offered silver ones; afterwards they offered vessels of sphatika, vairurya, asmagarbha, musaragalva and rohitamukta. The Bhagavat would not accept any of them. Each lokapalas returned to his palace and brought with him a stone bowl, of deep blue in colour and translucent, to offer again. The Bhagavat, rejecting this one and that one, accepted all of them, put them one with the other, and made one bowl of the four. Therefore one may see the four borders on the outside of the rim. Such variety of the materials of the bowls is also found in the other two sutras: the Fuben suoji jing (T. 3 :801c) and the Fangguang Da zhuanyuan jing (T. 3 :602a). The former, compiled between AD 587 and 592 by a Gandharan monk Jinagupta who left Gandhara for China in 555 (T. 49:103b, Kuwayama 1987:717), tells that each lokapala brought with him the bowls of suvarna, rupya, sphatika, vaidurya, rohitamukta, asmagarbha and musaragalva. The latter, transposed by Divakara on the 30th of September in 683 (Forte 1974: 149, No.8), also refers to the seven precious materials without their actual names of materials. The Fuben suoji jing is the first to mention the seven precious materials. The earlier 29 sutras refer only to the one stone bowl. The story of dedicating the bowls of seven precious materials in the Da Tang Xiyu ji therefore seems to have been selected from the Fuben suoji jing at the time of writing. Xuanzang probably did not learn about the story in India. Even in the Fuben suoji jing the last bowls that the Buddha accepted are described to have been the stone bowls of deep blue colour, which were offered filled with the flowers and coated with all kinds of the powder-incense. It is also very interesting to know from these later translated sutras that only these sutras give a decisive role to one of the lokapalas, Vaisramana, the Northern guardian. It is he that recommended the stone bowl to the other lokapalas, saying that he remembered that the devas of blue bodies had come to offer the four stone bowls to the lokapalas. As Foucher (1905 :420) pointed out, such bowls was a favorite object for Gandharan schist carvers, who, I surmise, depicted it as it possibly was in Purushapura. We can find it on the front of a pedestal supporting a Buddha or a Bodhisattva image (lIlus. 10), and in the centre of the uppermost part of a false gable adorning the anda of a stupa, as well as, very rarely, under one of a number of arches depicted in a frieze. In addition, we have some examples detached from still unidentified original reliefs. (See the lists at the end of the article). The examples collected so far, forty-eight in number, indicate that the most favorite depiction is the bowl on the front of a pedestal, twenty-one in number. The pictures of the bowl in gables number thirteen, which is almost equal in number to the examples depicted under arches and in other places. The bowl in relief generally takes a shape of a hemispherical body, resting on a round cushion laid on a pedestal supported by carved legs. Between the legs, in some cases, a pleated cloth is hanging, while at both right and left ends of the pedestal are standing two poles, to which is tied a curtain that looks as if it is surrounding the bowl in the background, and on which is a folded cloth that looks like a canopy with a wavy hem. The pedestal with such decorations is a type common to all that support the other relics like a relic casket, a turban, and so forth. In some instances only the bowl and its supporting base is depicted without any decorative elements (Marshall 1960: PI. 102). It is covered with a single canopy, or catra, supported by a stem which seems standing in the centre of the bowl itself. The seams that represent the fourfold composition of the bowls are depicted incised in parallel horizontal lines and the three lines should show the four rims. Of the forty-eight examples, however, only four of them have three lines, and thirteen of them bear two lines, while twenty of them have only one line. Four of the others are without any mark and the other seven are so defaced and so small that the rims cannot be counted. The variation of the number of lines seems to have depended on whether sculptors demanded such exactitude or not, and does not seem to suggest the multiplicity of bowls at various precincts in various countries, although Kumarajiva is said in the Gaoseng zhuan to have worshipped the bowl in Kashghar on the way back from Gandhara to Kucha. Also doubtful is the remark by Foucher (1905: 420) that partout sa lourde forme d'ecuelle reste sensiblement la meme, et il n'est guere contestable que l'ecole du Gandhara ne nous ait ici transmis des documents authentiques sur l'apparence exterieure 30 de la relique conservee dans sa capital. Let me stress no Buddhist art except the Gandharan art throughout a vast span of time and space has found any need to depict this particular relics of the Buddha. 4 In Singhala Faxian listen to a recited on a pulpit by a monk who had learned it by heart. It said that the bowl went around the world not only as the symbol of flourishing Buddhism but also as the base on which the Buddha's law was to be transmitted to the future. (14) Legge (1886:109-110) translated the passage as follows: Buddha's alms bowl was at first in Vaisali, and now it is in Jiantuowei (Gandhavati). After so many hundred years-he gave, when Faxian heard him, the actual number of years, but he has forgotten it-it will go westward to the state of Yuezhi; after so many hundred years, to Khotan; after so many hundred years, to Kucha; after so many hundred years, to the land of after so many hundred years, to Singhala; after so many hundred years, it will return to Middle India. Mter that, it will ascend to the Tushita heaven: and when the Bodhisattva Maitreya sees it, he will say with sigh, " The patra of Sakyamuni is come." And with all devas he will prersent to it flowers and incense for seven days. When these have expired, it will return to Jambudvipa, where it will be received by the king of the sea nagas, and taken into his naga's palace. When Maitreya shall be about to attain to perfect wisdom and become Buddha, it will again seperate into four bowls; which will return to the top of the Vinataka hill, whence they came. Mter Maitreya has become Buddha, the four lokapalas will again think of the Buddha with their bowls as they did in the case of the previous Buddha. The thousand Buddhas of this Bhadra-kalpa indeed will all use the same patra: and when the bowl has disappeared, the law of Buddha will go on gradually to be extinguished. After that extinction has taken place, the life of man will be shortened till it is only a period of five years .... " However stimulating an actual circulation of the one and the same patra may be, a reference to the then local centres of Buddhism, such as Yuezhi, Khotan, Kucha, and China, naturally raises a question about the authenticity of this sutra. Particularly notable is the phrase which Legge did miss to translate accurately: "Mter so many hundred years, to the land of Han". The Chinese original uses the term laidao which means "come and arrive". Legge's translation does not convey the exact nuance. The original text means that the bowl will come and arrive at the land of Han, but in the case of the other countries it only says "arrive" without accompanying the lai (come). Such usage as "come 31 and arrive" suggests that the Chinese people was responsible to wrote this passage. The local place names in this story cannot also be accepted at face value, since a monk in Singhala would not have known them. Rather it is natural to think that the story was written in China after the pilgrims had heard about the bowl. This also suggests that the original location of the bowl, Vaisali, may be spurious. The stories that tell how the bowl was brought into Gandhara come from two source books: the Maming pusa zhuan (Biography of Bodhisattva Asvaghosha), said to have been translated by Kumarajiva who had been in Gandhara between 3S8 and 361, and the FuJazang yinyuan zhuan, edited in Northern Wei in the early sixth century but in part supplemented later. (IS) The Maming pusa zhuan (T. SO:183c-184a) says: Mter then, when a king of Little Yuezhi in North India invaded and lay siege to a city in Middle India the king of Middle India sent a letter to him and said, "I am ready to pay you if you want anything. Why is it necessary to distress people by staying longer?" The king ofYuezhi replied to the Indian king, "If you agree to surrender, send me nine hundred million pieces of gold, and you will find pardon." The Indian king said, "There cannot be found even one hundred million pieces of gold in every corner country. How can I collect such a huge amount of gold?" The Northern Indian king pointed out that the country had two great treasures: the Buddha's bowl and a bhik~u (monk) whose talent of preaching the law of Buddha was very skillful. He said, "Give me these as substitutes and they will suffice for two hundred million pieces of gold" Thinking that he could not discard both of these, the king of Middle India asked the bhik~u what to do. But the bhik~u persuaded the king, saying that bhikshus were destined to bring salvation to all beings and therefore that the king would get a high reputation for allowing him to leave for North India. The king, respecting his words as usual, gave both of the treasures to the Yuezhi king. The Yuezhi king returned to his homeland with these precious things .... After arriving there the bhikshu, ASvaghosha became famous for his skillful propagation of the Buddha's law. (16) The FuJazangyinyuan zhuan (T. SO:31Sb) says: In Pataliputra there lived nine hundred million people. At that time a king of Yuezhi was very influential and was called Candra Kanishka. His fighting spirit was so ferocious and his bravery was so unrivalled throughout the world that, whenever he made expeditions, he did not fail to subjugate the attacked. So he came to this country at the head of four kinds of forces. After fighting he made the country surrender and promptly asked for nine hundred million pieces of gold. Then the king of Pataliputra offered the king Kanishka three things, Asvaghosha and the Buddha's bowl as well as kind-hearted 32 cock. Each was equivalent to three hundred million pieces of gold. He took them and presented them to the king Kanishka. ASvaghosha was especially excellent in his wisdom, and the Buddha's bowl possessed the merit of Tathagata, while the cock was kind-hearted, did not drink tainted water and could destroy its enemies. Because of these characteristics, they were worth nine hundred million pieces of gold. The king ofYuezhi was much delighted to receive these, withdrew his troops, and returned to his homeland. The legends tell that the Buddha's bowl was brought into North India as ransom together with ASvaghosha who is generally said to have contributed much to the expansion of Buddhist teachings. The king, who is responsible for this event, rules Little Yuezhi in the first source and Yuezhi in the latter, which specifies the name of the king. The sources also differ over where the Buddha's bowl originally was. Taking into consideration Faxian's narratives, original places of the bowl vary among all of the sources, such as Vaisali, Pataliputra, and Middle India. Whoever the king was and wherever an original place of the bowl was, what was of vital importance for Gandharan Buddhists presumably was that the bowl had been transferred to Gandhara from somewhere in Middle India, the place where Sakyamuni had actually been living and preaching. It seems to have been necessary for them to have a plausible reason why an object used by the Buddha himself was in Gandhara. Really the bowl was not an object brought with a king of the nomads but a simple forgery which was created in Gandhara, so an explanation of its originality was needed in order to justify its authenticity. Nevertheless the connection in the story of the bowl with a nomadic king seems to go back to something historical and suggests that he might have participated in Buddhist activities in Gandhara in the earlier phase of its history. We can be sure that Gandhara, or more broadly speaking, the northwestern part of the Subcontinent, was never a place where Sakyamuni had lived or preached. In this sense Gandhara was a putative site of Buddhism. When in such a locality Buddhism flourished, and gained an influential position in a short while, Gandhara must have disguised itself as an originally Buddhist homeland, or originally sacred in Buddhism. Really this part of the Subcontinent is not India in its proper sense, as Xuanzang clearly said of 'Bei (North) Yindu (India)' which extends from modern Laghman in Mghanistan till northwestern Punjab: (17) The Da Tang Xiyu ji, ch.3 (BealI884: 164) says: From the country of Lanbo (Laghman) till this country (Rajapura), the men are of coarse appearance, their disposition fierce and passionate, their language vulgar and uncultivated, with scarcely any manner or refinement. So they do not properly belong to India, but are frontier people, with barbarous habits. (18) The Biography of Xuanzang by Huili and Yancong also shows the same trends (Julien 1853:96-97): 33 Depuis Lanbo (Laghman) jusqu'a ce pays (Rajapura), comme Ie habitants vivent sur une frontiere rude et inculte, ils different sensiblement des Indiens par leurs coutumes, leurs vetements, et leurs langage; leurs moeurs sont vicieuses et grossieres. Gandhara needed something around which Buddhists could gather and upon which Buddhism could find firm roots and ties with Middle India. To this purpose, an object which could exist in places other than Gandhara had to be ruled out, and besides, it had to be something that could last. Hence only the bowl of the Buddha offered by four lokapalas met these requirement. Archaeological evidence in Taxila shows that a sudden development of activities of building stupas and monasteries took place sometime between the middle of the first century and the middle of the second century (Ghosh 1947-1948:44-45; Kuwayama 1984:216-219). On the other hand, the Kushan occupation of the Northwest can generally be supposed to fall into this time scale, whichever date of Kanishka one may select. This simultaneity roughly supports a view that the Kushans were responsible for building activities in Taxila. In this connexion Rosenfield pointed out that the patra must have played a major role in the religious awareness of the Kushans, being connected to the coming Buddha, Maitreya (Rosenfield 1967: 223). But at the outset, contrary to what he says, the bowl must not have been connected with the Maitreya. It must have provided the raison d'etre of the Buddhism in Gandhara, and it was one and the only unperishable relics that supported this view. No sooner was the Buddha's bowl established by the Kushans in Gandhara than it became the symbol of Buddhism there because of its unperishability. Only later did it become a means of transmitting the Buddha's law to the future. The connexion of the bowl with Maitreya can be clearly seen in Gandharan sculptures. An image which appears as the last of or along with seven Buddhas carved out in a row has long been interpreted to be Maitreya. Now we can better understand that an image, whether it may be Buddha or Bodhisattva, which is on the pedestal depicting the bowl of Buddha on its front side, is Maitreya. Insofar as the collected examples are concerned, it should be stressed that it does not always hold a flask in his left hand in the case of Bodhisattva images. From the above descriptions on Magadha in the Da Tang Xiyu ji we know that Xuanzang did actually see a stupa commemorative of offering the bowls. It cannot be stressed too much that he did not see the bowl itself. This fact clearly marks the point that the cultural and spiritual climate of India tended much more toward the speculation, or abstraction, while the Northwest has a tendency to depend upon what is more concrete, visible and tangible. The advent of the Kushans might have fostered, or even initiated this tendency. It is not a matter open to conjecture that a strict difference between Gandhara and India in the nature of cognition existed. The Buddha's bowl in Purushapura and depicted on reliefs is not same in character with those relics that were the objects of worship in the days before the creation of the Buddha image. 34 Judging from Chinese pilgrims' accounts on their travels, Gandhara and its environs are full of objects such as the hairs, the teeth, the eyeballs, the ushnisha bone of Sakyamuni. Although they were made in Gandhara, they had safely escaped from being branded as fakes. Together with such grotesque objects for worship, the land of Gandhara was very famous for a number of sacred places established after the Jataka stories, particularly of Bodhisattva who is a scapegoat destined to rebirth in this world because of his strict self-sacrifice, or killing himself to make others live. Among the places, the Four Great Stupas, thus called by Chinese pilgrims, were dominating Buner, Charsada, Taxila and somewhere between Swat and Taxila. What should be noted in this context is that all of such trends in Gandhara are quite unrelated to the actual deeds ofSakyamuni during his lifetime. As a matter of fact, not a single location of the events in the life of the historical Buddha could have been said to have taken place in Gandhara. If the Gandharans had said so, Gandharan Buddhism would have failed to live long. But if we think that Buddhists in Gandhara were not concerned with the actual life of Buddha, we overlook the essentials of Gandharan Buddhism. The well-known fact of Gandharan sculpture is that there are more narrative scenes from the life of Buddha than from the Jataka stories. Only on the reliefs could the Gandharans express their view of the life of the Buddha. 5 The last but not least is the epilogue of the Buddha's bowl. After telling that only a foundation of the building that had housed the bowl desolately remained in a corner of the city Purushapura, Xuanzang further informs us that the bowl is in the royal palace of Bolasi. Evidently Bolasi is a Chinese transliteration of *parasi, Persia, which is unmistakably Sassanian Persia in the last days under the rule of either Ardashir III (628-630) or Bulan (630/631). It is very curious to find the bowl in Sassanian Persia, since no Sassanian plundering of the Northwest between the fifth and the early seventh centuries happened and no Sassanian king has been known to have been so devout a Buddhist as to enshrine it in his palace. Nevertheless we have to accept the fact that the bowl disappeared in Purushapura sometime after the fifth century and before Xuanzang arrived there. The disappearance of the bowl is obviously told in the other source book, which is called in Chinese the Lianhua mian jing (Sutra of the Visage of Lotus) which has no Sanskrit text at all. According to a catalogue of the Chinese scriptures housed in the Sui court and edited in the Sui period, this is said to have been translated by Narendrayasas in the fourth year of the Kaihuang era, or 584.(6) The sutra consists of six parts, as I see it. 1) The Buddha, who forsees that nirvana is coming near, recommends Ananda that one should contemplate with deep meditation the physical body of the Buddha in case that the Buddha departs from this world. 2) In order to prevail the Buddha's law everlastingly over this world, the Buddha scatters his sarira, broken to pieces by himself, over various worlds of the N agas, Yakshas and Jambudvipa. 3) The Buddha transmits the True Law to the 'Thirty-three' Heaven, then 35 to the palace of the king N aga Sagara, then to the palace of the king N aga Kala, then to the Yaksha world, and to Jambudvipa. 4) The Buddha shows Ananda how this world becomes worse after his departure, giving some concrete examples. 5) The divinities pay hommage to the physical body of the Buddha. 6) Lastly the Buddha gives a prediction to the country of Jibin. The following is an English adaptation of the original Chinese text of the sixth part of the Lianhua mian jing (T. 12:1075b-1077b): (19) The Buddha preaches to Ananda, "In old days I gave a prediction to the country of Jibin at the place of the king of Naga Apalala. "The kingdom of Jibin will continue to flourish and be peaceful and rich like the country of Uttarakuru, after my nirvana. The Buddha's law prospers there. A number of arhats will come to this kingdom, and many disciples of Tathagata will also do so. All of the arhats who live in this Jambudvipa will come to this kingdom, which is like a dwelling of the gods in the Tushita heaven. These arhats will collect twelve classes of the sacred books of Tathagata (such as sutra, geya, vyakarana, gatha, udana, nidana, apadana, itivrittaka, jataka, vaipulya, adbhutadharma and upadesa), and will also make long sastras to them. This kingdom of Jibin is like the garden of Nandana of Indra and the cool lake of Anavatapta. Bharadvaja, Pindola, and others, therefore, wish to come to this kingdom .... There will also appear such five devaputra as Kumbira in Jibin where they make my law prevail and hold religious services on a large scale. "Oh, my disciple! Such a great assembly for religious services will have never been held before in Jambudvipa." The Buddha tells Ananda that in the country of Jibin in the future such a great assembly for the Law will be expected to be held. "Oh, Ananda! After five devaputras disappear, there will be a disciple of the heretics Purana whose name is Lianhua mian (Visage of Lotus). He is intelligent, wise and well acquainted with astronomy, the twenty-eight nakshatras and the five celestial bodies. He has a body like gold. But this blooming fool ever held religious services for the four arhats. When he was doing so, he pronounced the wish that he would destroy the Buddha's law in the future. Because he held religious services for the arhats, he will receive a well-constructed physical body in each birth of his, and will be born at the last birth in a royal family in Jibin, and he himself will be a king named Meizhihelajula (*Mihirakula), who will destroy my law. This blooming fool will break my bowl to pieces. After he does so, he will be born in the grand hell called Avici. After the death of the blooming fool, there will be seven devaputras, who successively renounce their bodies to be born in the kingdom of Jibin, where they re-establish the law of Tathagata and make offerings on a large scale." "Oh, Ananda! Because he destroys the bowl, various disciples of mine gradually dirty the pure defences (fila). In the first stage after my bowl is 36 broken, the bhikshus, despite sullying the pure defences, can still overthrow the heretics as a bull king does. But in the second stage the bhik~us in this Jambudvipa destroy the pure defences and desire doing bad deeds: they themselves take to thieving, tilling, and planting, favorably saving money, being particular about their robes and begging bowls, and do not wish to recite the sutra, the vinayas and the abhidharmas." "Oh, Ananda! In this way the people who wish to read and recite and are wise, all disappear. Then there will be many bhikshus, who are flattering, green-eyed, and often practice ten akusalas. Because they do so, the land of this country produces many thorny poisonous plants, and becomes sandy and barren." "Oh, Ananda! In this time in this Jambudvipa five kinds of condiments turn their flavour and lose their force. As five kinds of condiments like butter, oil, salt, sugar and honey lose their force and flavour, so people practice the evil deeds. Because they much practice the evil deeds, the destroyed bowl of the Buddha ought to go direct to the Northern land. At this time the people of the Northern land see the destroyed bowl of the Buddha and grandiosely hold religious services for it, with various flowers, perfumes, lamps, garlands, and various musics. The bowl being paid homage, there will be those who make up their minds for the Anuttarasamyaksanbodhih, those who do so for the Sravakas, and those who do so for the Pratyeka-buddhas. Then the broken bowl ought to go direct to Boluoboduo (*parvata, *pravada, or even *paravata ? See Levi 1905:300-304). The people of this country see the broken bowl of the Buddha, and make various offerings such as flowers, perfumes, lamps, garlands, and musics. There will be those who make up their minds for the Anuttarasamyaksanbodhih .... " " Oh, Ananda! With the power of the Buddha and with the force of good deeds of people, my broken bowl will by itself be restored as it was, and it will be not different from what it originally was." The sutra continues to say that the restored bowl of the Buddha then goes around the heavens and the worlds, being received with adoration by the divinities and the Nagas. After circulation the bowl farther goes to the end of the world, which is undestructable like diamond, where eighty-four thousand yojanas far from here it waits for the coming Buddha together with the sarira. When Maitreya reaches the fruit of Buddha, he will build the Four Ratnanaya Stupas in honour of the bowl and the sarira. With this event, the light of the Buddha's law again will begin to shine. The last part of the sutra exclusively tells that the bowl destroyed by Mihirakula, the king of Jibin, is a symbol of the extinction of Buddhism in Jibin, and seems to reflect something historical that might have happened in Jibin. Judging from both content and composition of this, this last part occupies one-third of the whole sutra, but does not show any specific relation with the other parts Nos. 1-5. It suddenly tells a very concrete story in which the historical place name such as Jibin is used. Therefore one may well be 37 allowed to doubt whether this was from the outset composed as one and the same text, and even whether it was edited from at least two original texts: the first two-thirds is the one and the last one-third is another. Levi (1905:297-300) and Ryujo Yamada (1955: 110-123) have discussed with this as one of the sutra which expound the extinction of the Buddha's law and supported a view widespread in the latter half of the sixth century in China that the time of the extinction of the True law, or the final age of dharma was come. Believing that the Buddhism in Gandhara was actually destructed by the Hephthalites, whom they identify with the Hunas, they claim that this reflects the historical event that Mihirakula destroyed the Buddha's bowl in Gandhara and that Narendrayasas, the translator of the sutra, has introduced it into the sutra as a witness of the desolation of Gandharan Buddhism. This hypothesis is clearly based on a widely accepted view that the decline of Gandharan Buddhism was first and foremost related to the first advent of the Hephthalites to Gandhara. However, the first Hephthalite invasion of the Northwest in the middle of the fifth century, or more exactly in 460, does not seem to find any support for the decline of Gandharan Buddhism. The date was very mathematically calculated on the basis of a simple schematic reckoning, founded on the description made by a Chinese official of the sixth century who compiled a memoir of the travels done by Song Yun and Huisheng. The description on Gandhara in the chapter 5 of the Luoyang Qielan ji (the Records of Buddhist Temples in Luoyang) says: (20) In the middle of the fourth month in the first year of the Zhengguang era (520), they entered the country of Gandhara. The land is also similar to Uddiyana and formerly called Yeboluo. It was conquered by the Heda (Hephthalites), who eventually installed a chiqin (tegin) as the king (of Gandhara). Now two shis (of the Hephthalite kings) have passed since their occupation. The meaning of the word shi is twofold: thirty and a generation or a duration of the reign of a king. Those who like to date the first invasion to 460 simply took the former meaning without paying attention to the other, so an answer was led to by a very simple calculation: 520 - (30 x 2) = 460! Certainly the shi originally means thirty, as the structure of the Chinese letter implies, but because the author used this character instead of the actual number, I prefer to take a less rigid view. Hence a solution is that the king of Gandhara at the time of Song Yun was either the second king or the third. This solution also suggests that the exact date of the Hephthalite invasion cannot be fixed. Another serious problem on the Hephthalite king, whom Song Yun met, comes out of the passage immediately following the above: (21) The nature (of the present king) is violent and cruel, and he often conducts reckless killing. He does not believe in Buddhist faith but devotes himself to non-Buddhist creeds. As all inhabitants are Brahmans who respect Buddhist teaching and enjoy reading the sutras so it is deeply against their 38 wishes that they suddenly have such a king. This is the famous passage that has always been discussed as a hard evidence by those who have given credence to the Hephthalite destruction of Gandharan Buddhism. But it gives no positive evidence for the event which they have believed. What can be explained from this passage is nothing other than the fact that either the second or third king, of violently cruel nature, often conducted killing and gave little support to Buddhism. Nothing is written about killing Buddhist monks and destroying Buddhist temples. Even if rulers gave no support to Buddhism in Indian subcontinent, it could flourish anyway as it did under the Gupta rules. So the Luoyang Qielan ji describes Gandhara as follows: (22) The land (of Boshafu, *Varushapura, or modern Shahbaz Garhi) on the rivers is fertile and the fortified city is well ordered, the population large and flourishing, the woods and the springs lush and numerous. The land is rich with precious articles, and the customs are refined and good. In and out of the city there are old temples where virtuous priests and monks reside and their devout conducts are highly excellent. One Chinese mile to the north of the city is a temple called the White Elephant Palace (a palace of the father of prince Visvantara), where all of the Buddhist statues are of stone, extrerneJlv fine in decoration and very many in number. The entire body of each statue is covered with thin leaves of gold, producing a dazzling effect on the viewers. No reference to desolate temples and the decline of Buddhist activities appears here. In fact, the opposite is true. This is the written document by an actual pilgrim known to us. It clearly refers to both stone sculptures and to images covered with thin leaves of gold in Gandharan art. Now we turn our eyes to the translator himself of the Lianhua mian jingo The Tang Gaoseng zhuan and other sources like the catalogues of the Chinese scriptures in the court of Sui dynasty say as follows: Narendrayasas was a native of Uddiyana. At the age of twenty-one he received the complete precepts and heard various old devotees praise the shadow of the Buddha, tell that there was the bowl in a certain country and the robe in a certain country, and say that the miracles of the ushnisha bone and teeth were countless. Learning about these, he eventually made up his mind and desired to see these things. But one had to study the vinaya when one received the precepts. So after the five summer retreats (or five years including the first summer retreat), he departed for the pilgrimage (See T. 50: 432a) This passage shows that the Buddha's bowl was still in Purushapura when Narendrayasas was twenty-one years old. His biography seen in the Tang Gaoseng zhuan says that he was forty when he arrived at Ye, the capital of Northern Qi dynasty (T. 50: 432c; T. 49:102c), in the seventh year of the Tianbao era, or 556. Based on this, one may get the absolute date of his age of receiving the precepts, or 537, and in this way we can know that he left for India in order to pay homage to the relics of Buddha in 54l. However it may be, the same biography gives another date, which is unfortunately 39 inconsistent with the above. It says that Narendrayasas died at the age of one hundred on the 27th of the eighth month in the ninth year of the Kaihuang era, in 589 (T. 50:433a). If this is accepted, the year of his receiving the precepts may be 510 and the date of departure for India 514. The former dating tells that the Buddha's bowl must have still existed in the thirties and forties of the sixth century, while the latter informs that it must have existed in the second decade of the sixth century. Turning back to his biography, we are informed of the fact as follows: he started from his homeland for an extensive pilgrimage throughout India from the Snow Mountains till the kingdom of Singhala, staying ten years in the monastery Kalandaka in Magadha. When he returned to Uddiyana, the monastery where he resided was burnt down by a conflagration of natural cause. Therefore he made up his mind to preach the Law in the regions north of the Hindukush-Karakorum. Mter many difficulties in the mountains he farther proceeded to the east and eventually arrived at the headquarters of Ruirui in central Mongolia. As soon as he arrived there, he encountered the Turkish attacks of the Ruiruis. Since the Turks blocked the ways to the west, he could not help giving up hope of ever returning to the homeland. Then he had to wander northward and reached the nihai (muddy sea) which was located about seven thousand Chinese miles to the north of the Turkish territory. But he could not settle down there, and again wandered southward where he to the region under the rule of Northern Qi, the capital of which was arrived in the seventh year of the Tianbao era, or in 556 as I mention~d above (T. 50: 432b-c). The Turkish invasions of the Ruirui's headquarters evidently corresponds to the fact in the Zhoushu, ch. 50, and the Tongdian, ch. 196, which respectively say that Tumen Illiq Khaqan attacked the Ruiruis in 552 and that Mugan Khaqan occupied their territory in 555, with the result that the chieftain Dengshuzi fled to Western Wei. This suggests that some years before 552 Narendrayasas had left Uddiyana for the northern regions. Besides, Mugan Khaqan advanced westward immediately after his victory over the Ruiruis to Tokharistan where the headquarters of the Hephthalites were, in order to cut off the Ruiruis' ties with the Hephthalites which had made them quite free from menace from the north. The Mugan Khaqan's attack in 555 of the Hephthalites in Tokharistan described in the Zhoushu, ch. 50 and the Suishu, ch .84, as well as the Tongdian, ch.196, is referred to in the biography ofJinagupta (Tang Gaoseng zhuan in T. 50: 433c) as ashijian. The term implies a current emergency which Jinagupta encountered in the Hephthalite court in Tokharistan. No mention of the date of the shijian is made in the biography. Nevertheless, the fact that he left Gandhara, his homeland, in 555 for Kapisi, Bamiyan, and the Hephthalite court in Tokharistan and arrived at Shanzhou in 557 gives a high possibility to a supposition that the shijian could imply the Turkish invasion by Mugan Khaqan. Since their defeat, the Hephthalites in Tokharistan often encountered disaster. For instance, Shinjibu Khaqan or Silziblos of the West Turks attacked them in 558 in alliance with Khosraw Anosharwan (Altheim 1969: 260). This date also appears very clearly in the Zhoushu, ch. 50, which says that the Hephthalite tributes to the Chinese 40 court ceased after they sent the last mission in 558. Ten years after then, the West Turks gave the last blow to the Hephthalites in Tokharistan as shown by Menander (Altheim 1969: 261), and this is also corroborated by the passage in the Suishu, ch.83, which states that the Hephthalites had formerly been in a chaotic state and were occupied by the Ton -shad Zi jie of the West Turks. It is not a matter of conjecture that the Hephthalites in the Northwest had closely been connected with their headquarters in Tokharistan. As a decisive result of their dispersal in the northern slopes of the Hindukush, their counterparts in the south also became helpless and disintegrated. As we have seen above, Narendrayasas even worshipped the Buddha's bowl before he arrived at the Ruiruis' headquarters in 552, and no evidence for the desolation of Gandharan Buddhism is found before the middle of the sixth century. Even if Narendrayasas died at his age of one hundred in 589, he must have seen the Buddha's bowl in the second decade of the sixth century, because Song Yun tells of the prosperity of Buddhism in Gandhara in and around 520. If we accept the later date for his birth, Narendrayasas must have seen the bowl in the forties of the sixth century. This suggests that Buddhism in Gandhara was still in good order in the forties of the century. The Tang Gaoseng zhuan tells that in the latter half of the sixth century only three Indian monks reached China. The first and earliest is Narendrayasas who carne to Ye in 556. The second is Jinagupta who arrived at Shanzhou in Qinghai province in 557, and the last is Dharmagupta, who carne to the capital of Sui dynasty in 590. Narendrayasas left Uddiyana just before the Hephthalite decline in both Tokharistan and the Northwest, and Jinagupta met with the first difficulties for the Hephthalites in Tokharistan in their court. After their dispersal in both sides of the Hindukush, Dharmagupta could pass through the regions from Takkadesa to Wa'khan via Kapisi and Bamiyan as well as Tokharistan (Kuwayama 1987: 719f). His biography (T. 50:434c-436b) does not say when he passed there. Nevertheless the total duration of his stays in each country on his route, amounting to ten years, suggests that he left Takkadesa in the end of the eighth decade of the sixth century. Therefore Narendrayasas cannot have known the Hephthalite decline in the Northwest, and Jinagupta could only know their first defeat solely in Tokharistan, not in the Northwest. Dharmagupta was the only one who had ample knowledge of the complete dispersal of the Hephthalites. It is of vital importance and of profound interest in this connexion to know that the biography of Dharmagupta is quite silent about the Northwest in striking contrast to Takkade§a and Kapisi. It says that Dharmagupta, accompanying his master to corne to Takkade§a from Kanyakubja, was informed of the Maha-cina kingdom by merchants who had corne from the 'northern routes', and that the kingdom of Kapisi was at the junction of the routes from the north, where merchants and merchandise from the northern countries had therefore been gathering. In no text dated before his biography do we have a single line of evidence showing that these countries were flourishing, nor have we any earlier written evidence in which these place names appear. The fact that Dharmagupta went from Takkadesa to Kapisi without regard to the Northwest proves that the Buddhism in the Northwest had already been devastated in 4-1 the eighth decade of the sixth century. On the other hand, TakkadeSa and Kapisi had improved their economics after the Hephthalite destruction of the commercial wealth in the Northwest, which had long since supported the Buddhist life and activities that had centered on the Buddha's bowl. It is clear now that the Buddhism in the Northwest had still been active during the Hephthalite occupation and that its decline should have happened in close relation with the disappearance of the Hephthalites, not with their first invasion.(7) Hence no reason to credit Narendrayasas with translating or even editing at least the last one-third of the Lianhuamian jing, even though the first two-thirds were compiled by him, since the event reflected in the last part of this happened after Narendrayasas left Uddiyana, or in the same period as the declining process of the Hephthalites in the sixth decade of the sixth century. The possibility that Dharmagupta would have been the last compiler of the Lianhuamian jing and would have given it the final title after adding the story of destruction of the Buddha's bowl is, to my mind, highly plausible.(8) Table - Lists of the Reliefs depicting the Buddha's Bowl 1. The bowl on the front side of the pedestal of statue: 1. Foucher 1905: Frontispiece (Pl.I). 2. Foucher 1905: p. 419, fig. 211. 3. Foucher 1918: Frontispiece (PI. II). 4. Foucher 1918: p. 223, fig.417. 5. Foucher 1918: p. 225, fig.418. 6. Marshall 1960: PI. 102. 7. Mizuno 1969: PI. 35,2 . From Mekhasandah. 8. Ackerman 1975: PI. LXIII(LM.189-1913). 9. Mizuno 1978: PI. 91,1. From Thareli. 10. Mizuno 1978: PI. 95,1. From Thareli. 11. Mizuno 1978: Pl. 110, 42. From Thareli. 12. From Thareli in 1964 (Neg. No. IAP64-2469). 13. From Thareli in 1964 (Neg. No. IAP64-2650). 14. From Thareli in 1964 (Neg. No. IAP64-2801). 15. Ingholt 1957: PI. XV, I (From the Minkenhof collection). 16. Joshi and Sharma 1969: fig. 25. 17. Bussagli 1984: Figure on p.195. 18. A pedestal in the Swat Museum. 19. A pedestal in the Peshawa Museum. 20. A Bodhisattva statue in the Matsuoka Art Gallery, Tokyo. 21. A seated Buddha statue in the Kabul Museum. II. The bowl in the false gable: 1. Foucher 1905: p. 127, fig. 47. 2. Foucher 1905: p. 129, fig. 48. 3. Ingholt 1957: PI. 168. 42 4. Ackerman 1975: PI. XLVI (3299-1883.1. S.). 5. Ackerman 1975: PI. XLVIIA (3301-1883.1. S.). 6. Mizuno 1978: PI. 112, 1. 7. Mizuno 1978: PI. 112,2. 8. Mizuno 1978: PI. 115,7. 9. Mizuno 1978: PI. 139,24. 10. In the Peshawar Museum. 11. In the Peshawar Museum. 12. In the National Museum, Tokyo. 13. In the Museo Nationale d'Arte Orientale. II\TV. MAl 1550-MNAOR 1186. III. The bowl in the arch: 1. Foucher 1905: p. 353, fig. 180. 2. Ackerman 1975: PI. LXXXV c (1.M. 212-1913). 3. Mizuno 1978: PI. 119, 13. From Thareli. 4. Mizuno 1978: PI. 120, 10. From Thareli. 5. Mizuno 1978: PI. 121, 10. From Thareli. 6. From Thareli in 1964 (Neg. No. IAP64-2350). 7. From Thareli in 1967 (Neg. No. IAP67-1348 a). 8. From Thareli in 1967 (Neg. No. IAP67-1348 b). 9. From Thareli in 1967 (Neg. No. IAP67-1515). IV. Others: 1. Meunie 1942: PI. XVII (54). 2. Ingholt 1957: PI. 230. 3. Faccenna 1962: PI. DXI, o. (Inv. No.812). 4. Joshi and Sharma 1969: fig. 24. 5. An unfinished work in the Lahore Museum. 43 IV The Wheel-shaped Structure inside Stupa A Hidden Import from Augustan Rome More than two dozen stupas in the South Asian subcontinent are known to contain a wheel-shaped pattern. From extant examples, the pattern appears to be a building device, in most instances, of the lower part of such stupas including the drum, or, in a few cases, of the lower part plus the dome. The geographical and chronological distribution of these stupas is peculiar to a specific period and to certain areas (Illus. 11; table 1): density is largest around the mouths of the Krishna and Godavali Rivers on the northernmost Coromandel coast; several exist in the Northwest (Gandhara) and sporadic examples survive in the upper Yamuna. The frequent usage of the pattern in stupas had quite a limited time span, emerging initially in the first century AD. The opinions of the excavators of these stupas and other scholars almost invariably converge on the idea that it sa ymbolises the dharmacakra and aided the economic use of materials. If so, why was the pattern not used in earlier stupas and only introduced in the first century AD? In this paper I reexamine the examples of this mode of architecture and attempt to locate them in their likeliest historical setting. Our survey begins in south India with special reference to the date of the construction of patterns in question. (I) Table 1 - Monuments with wheel-shaped structures and their elements Diameters(m) Souteast India 1. GhantaSala 2. Peddaganjam, No.2 3. Alluru 4. Adurru SA. Kodavali 5B. Salihundam 6. Amaravati 7. Nagarjunakonda No. I No.6 No. 21 No.5 No.9 No.3 No. 30 No.32A No. 14 No. 52 No. 54 No. 105 33.3 11.65 23 20.4 Hub • •• solid 41. 4 27.3 15 15 14.7 12.5 10 8.4 8.1 8.1 6.6 6.6 6.6 Ring walls 2? 1 2 2 1+ ? •• •• •• 0 0 © © © • H 2 2 Width of outer wall 6.15 1. 15 Radial walls (4 + 12) + 16 8? + 12 16 16 + 24 2.4 8 + 16 + 16 8 8 8 + 12 8 + 16 8 6 6 4 4 4 4 No. 27 No. 108 8. Ter 4.5 4.5 26 The Northwest 9. Kankali Tila lOA. Sanghol, No. I lOB. Sanghol, No.2 18.85 16.8 3.7 11. Tahkal Bala A 12. Dharmarajika 13. Block IE', Sirkap 14. Shah-ji-ki Dheri 15. Fil Khana 31. 5 34.5 12 x 10.8 56-72 15.2 Early Imperial Roman Mausolea 16. Augustus 87 17. West Mersea 19.5 20 18. Via Appia 19. Via Tiburtina X X X 4 4 8 2 1. 07 © © X • X X X © © © 8+8 12+24+32 8 2 0 1 0 0 16 + 16 16 4+4 8 5 0+ 0 + (12) + 12 + 24 6 X X: Without hub, . : Solid hub, ©:Hollow hub, II:Square solid hub, D: Square hollow hub. 1. The Southern Examples Ghantafala In a report of the excavations of a Buddhist site near an ancient south Indian port, Ptolemy's Kontakossyla (Sk. Kantakasaila), in the region of Maisolia in modern Andhra Pradesh (Vogel 1947-48: I), Rea (1894: 2-3) gives a description of the structure of a huge stupa known to us as Gha1).tasala. According to him, its plan differs from all other known in the Madras Presidency apart from the second stupa at Peddaganjam. The Ghamasalastupa has an outer ring of brickwork 6. 15 m thick, excluding the projecting basement which is 1. 675 m broad (Illus. 12,2). Inside is a concentric circle 16. 75 min exterior diameter and one meter thick. In the centre is a square cube of solid brickwork surrounded by a hollow brick square. Cross and radiating walls connect these walls; the smaller cells or chambers thus formed are firmly packed with black mud; the bricks used in constructing the walls are of very inferior manufacture. Concerning the superstructure, Rea says the interior walls suggest the existence of several storeys. These walls may, of course, have been simply used to strengthen the outer wall if it rose in the form of a dome, but judging from examples where a ring dome was undoubtedly employed, its thickness of 6. 15 m is proportionately much greater than is customary for a dome wall. Rea says that the wall of the inner circle is of the usual proportions, and the dome seems to have risen over it. According to the plan and section, Rea seems to have reconstructed the whole structure as follows: the hollow dome which he suggests rose over the inner round wall was intersected and supported internally by a square around the central, square, solid brickwork. Sixteen cross and radial walls firmly connected this square structure to the 45 inner circular wall. The outer, thick ring wall and the sixteen radiating walls between it and the inner circular wall, he seems to have interpreted as a high drum with a retaining wall that supported the lower external part of the dome and drum respectively. Peddaganjam Rea further notes the remains of the second stupa at Peddaganjam, which has a similar structure to that of Ghantasala (Rea 1894: 3). In plan it has two concentric brick walls. The exterior diameter of the outer wall is 11. 65 m, with a thickness of 1. 15 m, while the outer diameter of the inner ring wall is 6. 45 m, with a thickness of O. 9 m. The outer and inner structures are separated from each other by a distance of 1. 45 m and connected by twelve walls radiating from the centre; four of these walls also project inwards from the inner ring wall to the centre. Although Rea does not give any more information about the plan inside of the inner ring wall, the walls projecting inwards might have been originally numbered eight walis that radiated from the centre, since there are no other examples of any such a plan comprising only four walls that project from an inner wall without reaching the centre. The examples collected (table 1) clearly show a pattern of eight walls in the centre, not four, associated with twelve walls between two circles. Alluru As Mitra (1971: 28) rightly says, in addition to the above, other examples collected so far are stupas at Alluru, Adurru, Salihundam, Kodavali, and twelve stupas located in the open air at Nagarjunakonda. According to Mitra, the brick stupa at Alluru consists of sixteen spokes radiating from a circular solid hub, 9. 8 m in diameter, reaching a outer tyre which forms the face of the dome. The drum is 23 m in diameter. A limestone pillar, found lying in the compound of the Siva temple of the village, is decorated with a relief of half-lotus medallions and inscribed in characters of the second century AD, on the basis of which, Mitra thinks the antiquity of this Buddhist establishment is at least as old as the second century AD, if not earlier. Adurru Recent excavations at Adurru in East Godavali revealed a Buddhist precinct. It has a brick stupa built on a plan of double rings. The inner ring is intersected by sixteen spokes radiating from a solid hub, 5. 3 m in diameter, while twenty-four walls connect the inner ring to the outer one (IAR 1985 -86A: Iff., pI. I). According to the excavators, "The dome raised over the outer concentric circle left a limestone plastered space, 1 m wide, on the drum as an offset". The circumference of the stupa is 64 m with a diameter of 20.4 m and an extant height of 3.45 m from the plinth level. Of the rough total of five excavation layers, Layer 5 consisted of natural riverine sand, while Layer 4 comprised a rammed floor of brownish earth with brick-bats and the foundation level of the stupa. The report says that the structures are datable to the second- third centuries AD on the basis of the pottery types and stupa architecture. +6 Kodavali and Salihundam Nothing precise is published concerning the number of spokes or the size of a rubble masonry, stone stupa at Kodavali, East Godavali District, except for the mention made by Sarkar (1960: 78) of a wheel-shaped base built on a terraced platform with a solid hub and two concentric circles. The same is true with regard to a small stupa at Salihundam which has a central hub, eight spokes, and a rim (Mitra 1971:. 222; Margabandhu 1985: 36). Nagarjunakonda Sarkar provides measuremnets and other details for only thirteen of the twenty-one stupas that have wheel patterns at Nagarjunakonda, a site usually dated to the third century. The largest and earliest of all is the Mahacaitya, or stupa no. 1, constructed most probably in the sixth regnal year ofVirapurushadatta, the son and successor of Chantamula, the founder of the Ikshvaku dynasty (Sarkar 1960: 68, n. 1; Vogel 1929-30: 3). This stupa has three concentric walls. The four square projections of the cruciform platform are oriented to the cardinal points. The exterior circular wall has a diameter of 27 m. The centre is intersected by eight walls radiating from a solid hub, while the two intervening spaces between the inner, middle and outer rings are each divided by sixteen walls into the cells of the same number. Stupa no. 5 was constructed in the second regnal year of Ehuvala Chantamula, the son of Virapuru~adatta, while stupa no. 9 was renovated in the eighth regnal year of the same ruler (Sarkar 1960: 69; IAR 1957-58: 8-9). Both have two concentric walls supported by eight radiating walls around a central, solid brickwork. The outer ring space of no. 5 is intersected by twelve radiating walls and of no. 9 by sixteen radiating walls. The diameters of their exterior circular walls measure 14. 7 m and 12. 5 m respectively. As Sarkar pointed out, stupas having a diameter of less than about 8 m are generally given fewer spokes than those having a diameter of more than 10m. This is illustrated by Stupas nos 30 and 32 A (six spokes) and Stupas nos 14,27, 52 and 54 (four spokes). Stupas with a diameter of more than 10 m have eight spokes in the centre or radiating from the central brickwork (table 1). Stupas of smaller size have only a single ring wall, which is rather thick, and no inner concentric walls. However, among the larger stupas with a diameter of more than 10m, there are also some examples with only one ring wall, e. g. nos 6 and 21, which measure 15 m in diameter. Most of the stupas with wheel-shaped bases at Nagarjunakonda have a central circular structure of solid brickwork (table 1). A few examples have a hollow, central structure of brickwork in the form of a ring (nos 14, 52, 54) or a box-like square (nos 3 and 9). In the latter case a spoke radiates from each corner of the square and from a centre of each side. Smaller stupas like nos 27 and 108 have no central brickworks; instead, four spokes radiate directly from the centre. Amaravati In the description of the Main stupa at Amaravati Burgess draws attention to a solid mass of brickwork found by Mackenzie about 3. 6 m inside the perimeter railing 17 (Burgess 1887: 20). Barrett also noticed the importance of this brickwork in his reconstruction of the stupa. This brickwork was found in the south-west quadrant of the stupa, within, and concentric with, the circle of the drum and comprised a band of masonry, 8 ft (2.4 m) wide and 40 ft (12 m) in length, with an extant height of 12 ft (4. 3 m)(Barrett 1954: 35). Mackenzie observed: "On the south side, within the circles, a stony work of masonry is discernible, which may probably be the remains of an interior wall, as the people of the village informed me that a similar work had been observed all round, which has since been cleared away in removing earth "(Barrett 1954: para. 2). Burgess thinks that this, in all probability, marked the basement which supported the terrace of the stupa, but steers clear of a definitive commitment as to "whether the dome rose directly from this or whether it was constructed by other terraces above it"(Burgess 1887: 20). However, Barrett has little doub that "it was this eight foot [2. 4 m] wall which supported the dome itself' and quotes similar features at other sites in the Andhra Pradesh, such as Nagarjunakonda, Alluru and Ghantasala, where the "dome wall" is frequently strengthened by radial walls often laid in an elaborate wheel pattern. There is, of course, no means of knowing, Barrett continues, whether this was the case at Amaravati, but he believes that it is a possibility. Mitra accepted Barrett's point that the dome was not of solid brickwork, but she is quite careful about whether the 8 ft (2.4 m) circumference wall was strengthened by radial walls (Mitra 1971: 203, n. 11). However, she fails to provide an alternative for the wheel-shaped pattern when she says that the dome was not of solid brickwork. What then was the role of this thick circumference wall? Is the existence of such a thick circular wall known elsewhere in stupas of solid construction in any material? To answer this question, it will suffice to examine all the stupas of wheel-shaped pattern in Andhra Pradesh: the difference in breadth between that of the exterior circular wall and that of the other concentric walls is apparently quite large in stupas which have concentric circles bisected by spokes and, even in stupas with a single circular wall, the breadth of the ring wall is still disproportionately thicker. At Nagarjunakonda, in the case of stupas built with solid masonry (nos. 7,20, 22 and 38), there is no ring wall at all, or, even where one seems to exist, as in no. 20, it is much thinner. There can be little doubt, therefore, that this 2. 4 m thick ring wall found by Mackenzie at Amaravati was internally strengthened with spokes radiating from the centre, while its approximate diameter can be estimated as about 41m. Thick circumference walls are found almost without exception in all stupas with a wheel-shaped pattern. The significance of this peculiarity is discussed in detail below. The question still remains as to whether this ring wall was an integral part of the original stupa or part of a subsequent reconstruction. A large number of inscriptions found at the site attest to a flourishing Buddhist centre at Amaravati from the third century BC onwards. Mitra suggests that the nucleus of the Main stupa, which formed the principal focus of the Buddhist establishment, possibly dated back to the third century BC, citing as evidence a pillar-fragment with an inscription, most probably of Asoka (Sircar 1963-64: 40-43). An alternative proposal by Chandra (1919-20: 261), also based on the interpretation of inscriptions, is that the Amaravati Main stupa originated in the 48 second century BC. Yet even if the nucleus of the stupa does date back to the third or second century BC, the size of the stupa at that time is likely to have been on a much reduced scale. The ring wall in question, with a diameter of about 40 m, is much too large for a stupa of the third century BC and cannot have been part of the original Mahacaitya. Mitra reveals "recent clearance has shown that the existing ayakas are definitely much later than the original stupa and may even be as late as the second century; relics found in the sockets of two stones, one a re-used railing-post, in the core of the southern ayaka are presumably referred to by Inscription 47 of PI. LX of Burgess, ascribable to the second century". Furthermore, an inscription given by Ltiders (1912: 147, no. 1248) shows that the dharmacakra at the western gate of the Mahacaitya was a gift of a certain gahapati and his family in the reign of the Satavahana king Purumavi (c. 130-159). We can believe from the above that the Mahacaitya at Amaravati was being furnished with ayakas and gates in the second quarter of the second century. Therefore, the ring wall, the exterior face of which was found separated 3. 6 m from the circle of rails (Burgess 1887: 20), must have already been there when these outer decorations were under construction. Though the thick ring wall is not assignable to the earliest stupa, we may believe that it is a reconstruction of the time after the third century BC and before the second quarter of the second century AD, most probably of the first to early second century AD. Ter Farther to the Northwest, at Ter in the Osmanabad District in southern Maharashtra, the base of a large brick stupa with wheel-shaped plan, 26 m in diameter, was also excavated (IAR 1967-68: 25). It reportedly had circular ribs of bricks with eight spokes, surrounded by a circumambulatory path with four ayaka projections at the cardinal directions. The brief description lacks any plan, photograph, or detailed measurements to help us to understand the actual nature of the circular ribs, but the confirmed existence of eight spokes proves that the base of the stupa was built on a wheel-shaped pattern, most probably with a circular hub. According to the excavation report, "the stupa can be dated to the first half of the second century on the basis of an inscription recording the names of masons and a coin of Purumavi. An apsidal brick temple, fronted by wooden mandapa and housing a stupa, had been thrice repaired and seems to be assignable to the same period, since another coin of Purumavi was also found here. Although we must wait for a final report to say anything about the exact date of this precinct, it appears significant that finds from excavation included Roman clay bullae and a Roman glass bottle of the Mediterranean type. This site was originally discovered by M. N. Deshpande some fifteen years prior to excavtion. He brought to light five sculptured limestone slabs from the fields near the stupa mound and from the debris in the vicinity of a temple at a small village on the other side of the town ofTer where other three such limestone slabs had previously been discovered by Fleet (IAR 1961-62: 102, pIs CXLIX, CL). That "the finds suggest the influence of the Amaravati school" is accepted by Mitra.(2) 4-9 2. The Northern Examples Stupas with a wheel-shaped pattern are also known much further north of Maharashtra beyond the Vindhya Range. They are all found Northwest of Mathura: one at Kankali Tila, two at Sanghol in the Punjab, two at Taxila, two near Peshawar and another near Jalalabad. Kankali Tila The only available information on the Jain stupa known as Kankali Tila is an explanatory note by Vincent Smith and a ground plan (Illus. 12, 4) (Smith 1901: pIs II-III). Eight radiating walls apparently project outwards beyond the limit of the ring wall. But close observation of the plan reveals that one of the walls extends northwards to meet the small extant part of another ring wall which is probably identifiable as an outer concentric wall. The structure of this stupa most probably thus consisted ·of an internal and external circular wall, with the intervening space intersected by eight walls radiating from the centre. The eight walls converge from the interior ring towards the centre which contains neither round nor square brickwork. This is particularly curious since the diameter of the outer ring wall at Kankali Tila is 18. 85 m according to my measurement based on the scale given by Smith (1901: PI. III) and such a design is always found in all the stupas with the wheel-shaped pattern at Nagarjunakonda, even in the smallest stupas, like nos 27 and 108 (diam. 4. 5 m). Unfortunately, nineteenth- and early twentieth-century diggings, primarily in search of sculptures, destroyed any direct structural evidence at Kankali Tila for dating the stupa itself. Material for dating therefore only comprises several sculptures of mottled red sandstone found around the stupa: a standing Karttikeya inscribed in the year 11, a pillar with standing Tirthankara inscribed in the year 35 and a crouching Surya, all of which are datable to the Kushan period, i.e. the second to third centuries AD (Huntington 1985: 160-2). Close affinities between the sculptures from Kankali Tila and those found deposited near stupa no. 1 at Sanghol are claimed by S. P. Gupta (Gupta 1985: 50). Sanghol At Sanghol is located south of Rupar and midway between Ludiana and Chandigarh. At the site, two Buddhist stupas (nos 1 and 2) were found in the area SGL-5 and SGL-ll respectively. Sanghol gained its fame for the surprising discovery near stupa no. 1 of sculptured railings deposited just below the surface soil, but of concerns here are the stupas themselves because of their wheel-shaped patterns. The first digging at stupa no. 1 revealed a cylindrical structure with the wheel-shaped pattern surrounded by a number of other retaining structures (Illus. 12, 3). The spokes of the wheel comprise brick walls radiating from a acentral, hollow column or hub and are intersected by three circular walls (IAR 1971-72: 39). Succeeding excavations revealed other facts about the wheel-shaped structure (IAR 1984-85: 62, 66). During vertical excavation of the central, hollow, hub (diam. 3.24 m) there appeared ashes, charcoal, a 50 tooth, some fragmentary bones and a casket base of grey soapstone, the lid of which was also recovered. The Kharoshti inscription on the lid reads Upasakasa Ayabhadrasa. The radial walls were intersected by three successive circular walls concentric to the thick ring-shaped wall of the hub (according to the excavation report, there are four concentric walls, but this number appears to include the wall of the hub). The exterior diameters of the three concentric walls are 5. 04 m (7 m according to the plan given by Gupta 1985: 50), 13 m and 16. 3 tol6. 8 m respectively. The circular wall of 13 m diameter is 1. 07 m wide and possibly supported the superstructures. The outer circular wall, o. 3 m wide, is enclosed by square structure, 17 x 17m, which has been identified as an ipper terrace that supported the hemispherical dome of the stupa. Each side of the terrace is provided with a projection which is internally divided into many rectangular cells of irregular shape and is provided at each cardinal point with steps leading to the stupa. The three intervening ring spaces between the hub and the outer concentric wall are divided at regular intervals by radiating spokes, also of brick masonry (extant height 1. 18 m). There are twelve spokes in the first space, twenty-four in the second and thirty-two in the third. The interspaces between all of these walls are filled up with kankar and yellow earth. According to the second excavation report (IAR 1972-73: 28), the entire structure appears to have been built on the natural soil, this seems to contradict the first report of a pre-structural deposit that yielded sherds of crude black and grey wares, normally associated with the NBP Ware (IAR 1971-72: 39). The report of the monastic complex incorporating stupa no. 2 at Sanghol only briefly describes a small stupa comprising two circles (internal diam. 1.445 m and 3.70 m respectively) each with eight spokes (IAR 1985-86B: 69). However, the photograph included with the report reveals that all the spokes extend from a large, central, circular brickwork to the outer wall, which has a concentric inner face and an exterior facade in the form of a sqaure (IAR 1985-86B: pI. 23). If the photograph records the final stage of excavation, it appears that the stupa was raised on a square terrace. The device of a circle within a square is common to stupa nos 1 and 2 and comparable to the square plinth of a stupa in Block IE' of Sirkap II (Illus. 1). According to the report cited above, among the finds providing clues for dating are numerous seals and sealings with inscriptions (mainly in Brahmi, but also in Kharoshthi), coin moulds of schist, abundant copper coins of Indo-Parthian, Kushan and Huna rulers, and an ivory comb that notably recalls another example from Taxila. The most important find is considered to be an inscription written in ink on the interior of a late Kushan bowl, which, on palaeographical grounds has been dated c. second to third centuries AD. Situated is situated on a route leading from Mathura to Gandhara. The entire form of stupa no. 1 at Sanghol, raised as it is on a square platform, recalls usual Gandharan stupas, while, on the other hand, the 117 pieces of sculptured railings, mostly found as a hoard concealed in the earth, are of the Mathura school. Not only did the major Jain stupa at Kankali Tila in Mathura have a similar plan, but also, as S. P. Gupta notes, the dimension and workmanship of the Sanghol railing pillars most closely resemble those of Kankali Tila, which are shorter than the Bhutesvara pillars and have sculptures in comparatively low relief. Although noprecisely datable material related to the Sanghol 51 stupa itself survive, by comparing it with Kankal Tila, one cannot exclude the possibility that the stupa belongs to the earlier Kushan period of the first to second century. Gupta (1985: 50) clearly manifests his belief in a first century date on the title page of the booklet he edited on the subject.(3) Shah-ji-ki Dheri Further to the Northwest, we have much more evidence for attributing stupas with spoke walls to the first-second century cultural context. First of all, consider Shah-ji-ki Dheri (Illus. 12, 6). Although the results of the excavations at the site described by Spooner and Hargreaves in the early twentieth century are incomplete, the existence of a stupa with a wheel-shaped pattern within the criciform plan is beyond doubt. The ground plans given by Spooner and Hargreaves show at least four spokes in fragmentary condition: two contiguous spokes radiating from the centre toward the north-west and west and the other two extending from the centre toward the south-east and south-west. An excavation trench, 7. 2 m square, around the exact centre of the stupa mound, revealed traces of massive radiating walls of rubble masonry 0.3-0.6 m (i.e. a few feet according to Spooner) below the surface of the mound (Spooner 1908-9: 48-49). Deepening the same trench and avoiding the walls, he found the relic chamber at a point two ft (0. 6 m) below the level of the brick pavement surrounding the stupa. The relic chamber was built off centre, against "the end of that one of the radiating central walls which ran due east from the centre of the stupa". Though not shown on the plan, a radiating wall running due east clearly must have existed. Spooner says only that "after several days" digging we got down to what seemed to be free earth and had lost hope of finding any relics at all, when suddenly, and without warning, the remains of it were reached. From this it appears that the relic chamber was built on "what seemed to be free earth". Spooner continues: "A long, smooth slab of slate had been laid down extending in its length from north to south, and across the southern end of this was laid a heavy slab meeting at right angles with another heavy slab along the western edge. These two thus formed two sides of a possible square, with the corner intact at the southwest". However, no trace whatever could be found of any corresponding slabs on the east and north, and from the general position of the whole, he conceived that the chamber had been enclosed on these sides by the massive rubble masonry of the walls radiating to the east and the north-east. Whether the wall running north-east was really extant, or Spooner merely imagined it, we are not sure. His description raises a number of questions as to actual location of the relic chamber. First of all, he says that it was found against the end of the radiating wall running due east, but no spoke running due east from the centre is shown on the published plan. Was the draftman not able to include the eastward spoke on the plan because of its rather deeper location in the trench? Secondly, Spooner supposed that the spokes running east and north-east were substitutes for the north and east sides of the relic chamber, but it is impossible to imagine such a structure where two walls running from the centre toward the east and north-east formed the north and east walls of a square room or something similar? If the situation of the two sides described by Spooner 52 was indeed right, then the other sides must have comprised walls radiating from the centre toward the west and southwest. But no such walls appear on the plans given by Spooner, or even Hargreaves. Hargreaves' description increases the ambiguity: "An attempt to fix the circumference of the drum of the stupa dome by following one of the radiating walls running from the spot where the relics were discovered yielded as little result, for the wall was broken at a distance of 24 ft [7. 2 m] from the centre of the mound (Hargreaves 1911: 25). The radiating wall he followed is most probably tl1e one running to the north-west, since it is clearly drawn much longer on the plan given by Hargreaves than on that by Spooner. Although this longest wall is said to run from the spot where the relics were found, we do not know the exact location of the relic chamber, nor how many radiating walls were still extant. However, if we take as valid tl1e directions of the two spokes described as east and north-east and include them with those shown on the plan, the total number of original spokes must have been eight, radiating at regular intervals from the centre in the directions of north, north-east, east, south-east, south, south-west, west, north-west respectively. The next question is whether or not the spokes and the cruciform 'plinth are contemporaneous. Reasoning that the latest main wall of the stupa proper was undoubtedly of the same material and technique as the circular tower bases, Hargreaves rightly pointed out the improbability that it had formed part of the oldest structure on this site (Hargreaves 1911: 26). According to Spooner, as we have seen, the spokes were first revealed a few feet below the surface of the mound and seem to have continued downwards to reach "free earth". "Free earth" was also found in the eastern part of the south projection, where Spooner first dug one of the two deep pits. This produced three layers: a top layer, 1 ft thick, of soft "free earth" or surface soil; a middle layer of tightly packed debris some 2 ft thick; and a layer of "free earth" that continued down to a point 15 ft from the existing surface soil. Therefore the tem "free earth" in the latter sense means virgin soil, devoid of any archaeological context. Spooner also confirms that the relic chamber was found 2 ft below the level of the brick pavement surrounding the stupa (Spooner 1908-09: 48).It seems clear from the excavation reports that the four projections were contemporary with the main walls of the stupa. Since the stucco decoration of seated Buddhas and pilasters on the north-eastern corner of the western projection continued in part onto the main wall, the brick pavement is conceivably also contemporary with the main walls of the stupa. Although we do not know the depth to which the radiating walls descended, we can surmise that their excavated depth, to the level of the relic chamber, was at least 2 ft (0.6 m) below the base moulding of the exterior plinth wall. Thus the internal radiating walls are earlier than the outer cruciform base. This tentative conclusion as to when the radiating walls were constructed can be reached in another way. A circle drawn to touch the sides of the square plinth on a plan of tl1e stupa represents the maximum circumference of the supposed drum. However, in the south-western quadrant, part of a remaining radiating wall extends beyond this circle. Only the very small stupas in Block IE' at Sirkap and stupa no. 2 at Sanghol have a square base internally strengthened with spokes. At Shah-ji-ki Dheri, in contrast, the radiating walls are too long for the supposed drum or dome to have been supported by a 53 square plinth of such size. The former would have been confined within our hypothetical circle if the radiating walls and the square main walls were at least contemporaneous. Therefore we have another good reason for believing that the structure with eight radiating walls is earlier than the outer cruciform one. Moreover, the extant length of the southwestern radial wall supports the contention that the diameter of an original round stupa might have been more than 55. 2 m. Near the north-west corner of the square plinth the remains of three small stupas were found. The alignment of these votive stupas is not parallel to the existing wall of the Main stupa but suggests an earlier circular construction with which the small stupas were associated. Furthermore, Hargreave notes that one of these stupas on the north-west appears to have been partly demolished as if to allow for the building of the tower, while a small stupa to the north-east lies directly in the way of the foundations of a pathway that seems originally to have surrounded the monument, and of which traces can still be seen on the south and west (Hargreaves 1911: 26).(4) The distance to the inner margin of these small stupas from the supposed centre from which the radial walls would have started measures about 36 m. Therefore a circular stupa built with eight radiating walls might have had a diameter between 55.2 m and 72 m. To sum up, from the excavation reports of Spooner and Hagreaves, all tl1e available information on the earlier stupa with a wheel-shaped pattern at Shah-ji-ki Dheri is as follows: 1. The masonry was of massive rubble. 2. The radiating walls survived to a considerable height (from 2 ft below the surface of the stupa mound down to the virgin soil, which was 2 ft below the brick pavement that surrounded the entire stupa. 3. There were no remains of a circular or square central structure of masonry from which the walls might have radiated. The walls simply converged at the centre. 4. Eight spokes originally radiated from the centre, but there remained at least five spokes in a fragmentary condition oriented east, south-east, south-west, west, and north-west respectively. 5. The estimated exterior diamater of the inferred circular structure which the eight radial walls internally supported was between 55. 2 m and 72 m at its base. 6. Since Hargreaves says that an attempt to fix the circumference of the drum of the stupa "yielded as little result", the exterior ring wall to which the radiating walls originally extended was presumably not found. However, it is also possible, particularly if the evidence from the Kankali Tila stupa at Mathura is taken into consideration (see below), that a thick ring wall such as those known from examples in Andhra Pradesh never existed at Shah-ji-ki Dheri. Dating the later cruciform plinth is difficult, but the earlier stupa with the wheelshaped pattern was most probably built at the same time, or even much before the relic chamber, since the latter could not have existed without the radiating wall: as Spooner records, it was found against "the end of that one of the radiating central walls which ran due east from the centre of the stupa". So far so good. However, in this case the "end" is quite insufficient as a the description: Was the radiating wall found running to a broken 54 end? Were both ends found broken? These questions are really important for fixing the real chronological relationship between the Shah-ji-ki Dheri stupa with the wheel-shaped pattern and the relic chamber containing the "Kanishka" casket. If the radiating wall was truncated and the relic chamber was located against the broken end, it implies that the stupa might have already once met with a disaster prior to the construction of the relic chamber. In this case, the damaged stupa would then definitely predate the reign of Kanishka I, if the casket is really a dedicatory object of Kanishka55 l Taxila The core of the Dharmarajika stupa in Taxila (Illus. 12, 6) definitely predates the reign of Kanishka. As Marshall writes, it is "of rough rubble masonry strengthened by walls, three feet two inches [0. 95 m] to four feet nine inches [1. 43 m] in thickness, radiating from the centre. These construction walls "stop short above the berm of the stupa, instead of being carried down to its foundations", and it is obvious that they belong to the subsequent reconstruction of the fabric which took place probably during the Kushan epoch (Marshall 1912-13: 10). Marshall gives detailed measurements of the stupa elsewhere: the overall diameter, including the staircase and the raised terrace around its base, is 150 ft (50 m) from east to west by 146 ft 6 in (44 m) from north to south, the body of the stupa having an average diameter of 115 ft (34. 5 m), the berm a projection of 8 ft (2. 4 m) and the step a further projection of 8 ft 6 in (2. 55 m). The present height of the ruin is about 45 ft (13. 5 m). It appears from Marshall's description and plan of the Dharmarajika stupa that the centre of the core is a solid, round structure (diam. c. 50 ftl15 m), which is built of the same masonry as that of the radiating walls. Thirteen of the original sixteen spokes remain extant. Of the existing walls, only the six in the northern half of the stupa seem to converge with any regularity towards the supposed central point; the others are splayed like a vortex whirling anti-clockwise from the outer face of the central circle. Marshall (1951: 236) says further: As to when the original fabric fell to ruin and was rebuilt the stupa itself does afford some evidence: for, although the masonry of the reconstructed drum (of wheelshaped pattern) is limestone rubble, there is a certain measure of regularity in the way it is laid that recalls the masonry of many of the Saka-Parthian buildings in Sirkap (Sirkap II), rather than the earlier masonry of the Bhir Mound or of the lower settlements in Sirkap. Moreover, the principle of the radiating construction walls is paralleled to some extent in the foundation walls of the little stupa in Block C' in Sirkap, which was er ected after the earthquake in the early part of the first century AD But another and stronger reason for ascribing the rebuilding to the first cenhlry AD is to be found in a small patch of diaper facing masonry on the west side of the stupa drum immediately above the berm, which apparently belongs to the reconstruction period and is identical with other diaper work of the first century. In the above, the mention of Block C' is somewhat strange if one refers to Marshall's description of the little stupa there: "The foundations of the Block C' stupa descend to a depth of about 2 ft 8 in. below the ground level. Some 4 ft below this, i. e. at a depth of 7 55 ft from the surface, was a wall running east and west, which evidently belonged to a building of the third stratum (Sirkap III)" (Marshall 1951: 191). Block C' should be read as Block IE', where he found the high plinth of a stupa, measuring 40 ft (12 m) north to south by 36 ft (10. 8 m) east to west, constructed partly of diaper masonry and partly of rubble, and strengthened internally by thick foundation walls arranged both cross-wise and diagonally, interspaces between them being filled with stone debris (Marshall 1951: 183). This high plinth, although lacking a internal circular facing, shows structurally the same device as the foundation of stupa no. 2 at Sanghol. The "small patch of diaper facing masonry on the west side of the stupa drum immediately above the berm" is no longer visible. If it really existed there at the time of Marshall's excavation, the evidence from Sirkap IE' evidence for the coexistence of both rough rubble and diaper masonries within the same stupa can be extended even to the Dharmarajika stupa. If we believe Marshall, the radiating walls are indistinguishable from, and thus part of, the diaper work belonging to a reconstruction that was built over the earlier low plinth. In this connection with the question as to when diaper masonry was introduced at Taxila, Allchin has very carefully and correctly interpreted the coin finds in relation with the strata in Sirkap and given quite a convincing chronological picture of Sirkap II: it was constructed during the last years of Azes II, perhaps during a period of overlap with Kujula Kadphises, suggesting that Gondophares was a somewhat senior contemporary of the former and was at best remotely concerned with the government of Taxila. With this conclusion he implies that the new diaper masonry of Sirkap II, a new city constructed after the earthquake, coincides with the arrival of the Kushans under Kujula Kadphises rather than with Gondophares as Marshall thought (Allchin 1968: 13). If this conclusion can be extended to Dharmarajika, the dome, which is internally strengthened by walls radiating from the central solid circle of rough rubble limestone masonry and externally faced with diaper masonry, is datable to the time of Kujula, if not later. However, what is visible now, and also in photographs and drawings by Marshall (1951: PIs. 46-48), of the outer facing of the stupa, is semi-ashlar masonry. As Allchin says, this is absent from Sirkap and only appears in the houses inside of the Sirsukh fortifications and in a series of stupas (K1, K2, and K3) to the north of the Dharmarajika stupa. stupa K3 overlies a rubble masonry tank filled in with debris. In the filling were discovered two coins of Kujula and one of Soter Megas, while within the stupa was a deposit that included three coins of Kanishka. In the relic chambers of P6, another stupas faced with semi-ashlar masonry, located a little south of K3 and between two other semi-ashlar stupas (P3 and P4), three coins of Huvishka and seven of Vasudeva were discovered. Semi-ashlar masonry therefore, as Allchin (1968: 22) claimed, continued to be in vogue for several centuries after the time of Kanishka. Therefore a tentative conclusion, based on the diaper masonry, is that the construction of the radiating walls is datable to the time of Kujula, if not later, and, based on the semi-ashlar masonry to which they run, earlier than the time of Kanishka. Fil Khana 56 Another stupa with internal spokes is located on top of a conglomerate hill several kilometers to the north-west of Jalalabad, on the opposite bank of the Kabul River. There are two mounds with stupa remains that were most probably associated with the caves hewn out of the side of the hill and known locally as Fil Khana (the House of an Elephant). One of the stupas had been reduced to a mere pile of rubble, but a vestige of the other huge structure still survived when the site was surveyed in 1962 (Mizuno 1967: 76). The latter, 13 m in height, measured about 15 m in diameter at the base of the drum, which was encircled by a low plinth with a diameter of 18 m. The structure rested on two rectangular podiums, the upper one being of greater height. The extant northern facades of both podiums had projections for flights of steps leading up to the circular drum. The drum was constructed with eight walls regularly radiating from a centre which had neither a core nor cell, like Shah-ji-ki Dheri and Kankali Tila. The radial walls were built of river boulders jointed with clay and river boulders. A sketch drawn by Simpson clearly shows that the upper part of the drum was decorated with a band of stucco-plastered arches supported by "Indian" pilasters with Corinthian capitals. The exact date of the Fil Khana stupa is not clear, but its rectangular podiums are later than the examples of the Azes II - Kujula Kadphises period: the square prototypes first introduced at Sirkap II, the small stupas encircling the Dharmarajika Main stupa and stupa no. 4 with kanjur facing located north of the precinct at the latter site.(7) Tahkal Bala Foucher (1905: 88) gives the following account of the excavations at Takkal (or Tahkal Bala) carried out by Lieutenant P. Haslett: "En dedans des murailles interieures, ecrit-il, il y a des murailles transversales, generalement baties comme des rayons partant du centre; on s'est servi, pour combler les intervalles entre ces murailles, de terre et de pierres rondes empilees par couches horizontales qui sont aisees a distinguer". According to Errington (1987: 320-323), who has given easy access to Haslett's report ( Haslett 1876: 238-9), the 1876 excavation of Tahkal Bala revealed two stupas. One of these (Mound A in Area A) was of the pattern discussed here and is described by Haslett as follows: The mound A is about 70 ft (21 m) above the ground level, and built solid. The two concentric walls have been uncovered. These are of 16 sides each, and are interior walls, built to give solidity to the structure. They have projecting stones to dovetail into the earth, and stones filled in between. Inside the inner wall are cross walls, generally built as radii from the centre, and earth and round stones were filled in between these walls in horizontal layers, which are easily distinguishable. When excavating into the mound lower down, a wall was found built in steps, about 8 inches rise and tread. This I do not think is a portion of the outer wall of the tope, nor could any trace of it be found. Errington rightly interpreted the Haslett report as a probable description of the internal core of a stupa, apparently about 105 ft (31. 5 m) in diameter (Errington 1987: 314 and 321). Two concentric polygonal walls of sixteen sides are extraordinary. A plan of Mound A (Haslett sketch no. 3; Errington 1987: 308-309, fig. 4) shows the two 57 concentric, sixteen-sided polygons without indicating the breadth of the walls or the system of ribs. Yet Haslett's description and Errington's interpretation seem to imply that the centre from which the walls splayed outward was not of solid stonework as in the case of Fil Khana and Kankali Tila. The enormous diameter of this stupa can only be compared with that of Dharmarajika Main stupa (diam. 34. 5 m) in the north-west and with those of Ghantasala(3 3. 3 m) and Amaravati (41. 4 m) in Andhra Pradesh. The Tahkal stupa was enclosed by a rectangular wall (or walls), reinforced with round and semi-circular bastions, but it is not known whether the stupa and the enclosure were contemporaneous structures. The enclosure of a stupa court with bastioned walls recalls the Bagh Gai Buddhist temple at Hadda, where the stupa court is separated from the monastic court only by the southern row of monastic cells. Round or semi-circular bastions attached to Buddhist precincts are found elsewhere in Mghanistan. Shotorak has several: one from the last stage of the site datable to the eighth century at the extreme east of the stupa court, and two in the outskirts of the monastery area (Kuwayama 1991A: 103-107, fig. 10). The square monastic building at Gul Dara in the Logar valley is also reinforced by three round bastions on the north-eastern side and is datable to the seventh century (Kuwayama 1991A: 103-107, fig. 9). Also at Tepe Maranjan the square monastic building was renovated with at least three round bastions. At each corner of the square walls of the stupa at Shah-ji-ki Dheri Period II datable to the eighth century (Kuwayama 1991A: 100-103, fig. 8). The exterior wall of the Shah-ji-ki Dheri stupa has four round bastions, one at each corner, which do not necessarily all postdate the square structure that they reinforce. Semi-circular bastions, of, as yet, undetermined date, are also known from the city wall at Sirsukh (Schlumberger and Le Berre 1964: 88, n. 6; Fussman 1974: 91-91). Date proposed by Le Berre, Schlumberger and Fussman for the bastions of all these sites cover a wide time span from the second to the eighth century and do not afford any clear guidelines for dating those of the Tahkal Bala Area A stupa enclosure. 3. TheFunction of the Thick Ring Wall and Spokes The northern stupas with the wheel-shaped pattern have no particular masonry structure in the centre, with the exception of examples at Sanghol in the Punjab and at Dharmarajika, Taxila (table 1). The walls merely radiate from the supposed central of the stupa dome. In contrast, the southern examples have either a circular or square central structure of brick, regardless of the size of the stupa (the two exceptions at Nagarjunakonda are the smallest of the kind). In most cases the central structure is solid, but in some instances, it is in the form of a cylinder or a box-like shaft. Regardless of whether or not a masonry structure existed in the centre, the wheelshaped design was essentially applied to stupas of enormous size throughout the southern and northern regions.(8) We had better exclude the examples at Nagarjunakonda in this instance: construction of Buddhist stupas and monasteries supposedly flourished there not before the Ikshvaku period in the third century AD, as Sarkar has demonstrated, 58 while the building of stupas with a wheel-shaped pattern merely followed a tradition that was established in the area of the Krishna and Godavali river mouths during the Satavahana period in the earlier half of the second century AD. Whether the spoke structure was contained only within the drum or extend up into the dome can only be conjectured at Kankali Tila, Sanghol, Shah-ji-ki Dheri and in the south. Yet the extraordinarily thick ring walls at Amaravarti, Ghantasala and Nagarjunakonda imply great pressure from the superstructure. Such broad thicknesses would not have been required if the drums were as low as Barrett's reconstruction of Amaravati Main stupa (Barrett 1954: fig. 2). This device might have been essential, particularly in such a high drum as that depicted in the Amaravati reliefs (Barrett 1954: PI. Ia). Presumably the wide ring wall was therefore the exterior wall of the drum on which the dome stood, being neither of the dome itself nor of the dome plus its surrounding path. It cannot be far from truth, moreover, that all stupas with thick ring walls had a rather high drum. No thick wall is necessary for the outer, lower casing of a semi-spherical dome. The pressure of the dome is only exerted downward to the drum and eventually causes cracks in the drum's outer wall. As a direct consequence of this, it may well be that the thicker ring wall was necessary in the construction of stupas with large dimensions and high drums. Earlier larger stupas have usually been considered to have had a low drum and dome, as Marshall advocated in his reconstruction of the stupa at Sanchi. But the existence of a thicker ring wall in stupas leads to the hypothesis that a high drum must have been incorporated even into stupas out in the open air, not only into those of rather small scale that were hewn out in cave temples in west India. In the north-western examples the spokes were contained within the dome. At Dharmarajika, spokes are observable at the top of the extant height of the stupa, which reaches 13. 5 m above the ground. The remaining height of the spokes and the central stone work is too tall for the drum, even if the height of the part below the berm is extracted. Moreover, there is no trace of an outer ring wall anywhere. The same is true for Fil Khana, where no ring wall is traceable and the spokes are seen on top of the mound. All of these facts suggest that the stupas of this kind in the north-west contained spokes as a structural device for the dome, not for the drum, and that the tightly packed fill of stone and earth filled in between tl1e spokes did not require the construction of a thick exterior wall. That the wheel-shaped structure at Dharmarajika stops short above the berm of the stupa, instead of being carried down to its foundations, is proof of this assumption, despite the fact that it is a reconstruction. 4. Delusion of the Dharmacakra and the Rathacakra The wheel-shaped pattern contained in stupas seems to remind everyone of the dharmacakra, as stated by writers such as G. B. Sharma and S. P. Gupta (Gupta 1985: 50). D. Faccenna also is of the opinion that Gandharan monuments showing these characteristics (Dharmarajika and Sirkap Block IE' stupas at Taxila and the Shah-ji-ki Dheri stupa near Peshawar) all date from a later period, and that the expression of such a 59 visible pattern of construction may perhaps embody a symbolic-religious significance inspired by the dharmacakra symbol (Faccenna 1980: 29, n. 1). Sarkar (1960: 70) interprets to the introduction of a wheel-shaped plan for the larger stupas at Nagarjunakonda as being motivated by considerations of structural stability and economy of material. However, taking into consideration the instances of small stupas which do not require so much stability and economy, he also does not forget the possibility of symbolism. Accordingly, he says that this developed constructional feature might have drawn its inspiration from the dharmacakra symbol and that, hence, one may notice in wheel-shaped stupas one may not only notice an improvement over an earlier building tradition, not only an improvement over an earlier building tradition, but also a successful attempt at transforming an idea, a symbol, into an architectural entity. Along the same lines, S. L. Huntington (1985: 179) thinks that the wheel clearly has Buddhist significance, representing the Buddhist dharma, and that the method of construction possibly served both a philosophical and structural purpose. Snodgrass (1988: 78) and Kottkamp (1992: 73-78) find only a symbolical implication in the wheel-shaped pattern of the stupa. In view of the above, it is noteworthy to find V. A. Smith voicing the following dissenting view "the arrangement of radiating walls in the basement is merely a constructive device to save expensive masonry"(Smith 1901: PI. III). The most serious fallacy in conceiving such stupas as symbolizing the dharmacakra lies in the fact that the design of the wheel-shape stupa is only evaluated one-dimensionally, in horizontal section from above, while the importance of its elevation, viewed in vertical section, is totally ignored. The restriction in the south of the wheel-shape to the basement, as Smith noted above, encouraged this point of view. If one only considers the horizontal section or plan, the wheel-shaped pattern ostensibly looks like a wheel. But when both horizontal and vertical sections are assessed together, the pattern cannot be a wheel: the location of the wheelshaped pattern in the Northwest is too high, as stated before, to represent the dharmacakra, while the existence of spokes without an exterior wall or tyre could never have embodied a wheel. Even in the south, the thick outer walls there must have been those of high drums, too high to represent the Wheel of the Law. Table 2 - Number of sporks of the Dharmacakra Amaravati Sanchi, Stupa 1 28 : 31 : 63 : Pillar supporting the dharmacakra (Barrett, 1954: PI. 28). Dharmacakra (Barrett, 1954: PI. 16). Pillar supporting the dharmacakra (Sivaramamurti, 1956: PI. 63, 3). 64: (Sivaramamurti, 1956: PI. 20, 2). 81, 81 : Dharmacakra on the Buddhapada (Barrett, 1954: PI. 47). Uncountable: Pillar supporting the dharmacakra (Barrett, 1954: PI. 8, b). (PI. 9, a, b). Uncountable: Uncountable: (PI. 20). 16: (Marshall/Foucher/Majumdar, 1940,2: PI. 40). (Marshall/Foucher/Majumdar, 1940, 2: PI. 22). 20 : 22: (Marshall/Foucher/Majumdar, 1940,2: PI. 69, c). 25 : (Marshall/Foucher/Majumdar, 1940, 2: PI. 26). 60 Sanchi, Stupa 2 Sanchi, Stupa 3 Mathura Sanghol Gandhara 26 : 28 : 32 : 34 : 38 : 12 : 16 : 16 : 16 : 16 : 17 (?): 20: 24 : 27 : 15 : 22 : (Marshall/Foucher/Majumdar, 1940,2: Pl. 27). (Marshall/Foucher/Majumdar, 1940,2: PI. 37, a). Wheel of a cart (Marshall/Foucher/Majumdar, 1940,2: PI. 23). (Marshall/Foucher/Majumdar, 1940,2: PI. 41). (Marshall/Foucher/Majumdar, 1940: PI. 60). (Marshall/Foucher/Majumdar, 1940,3: PI. 91, h). (Marshall/Foucher/Majumdar, 1940,3: PI. 91, g). (Marshall/Foucher/Majumdar, 1940,3: PI. 74, lc). (Marshall/Foucher/Majumdar, 1940,3: PI. 74, 3a). (Marshall/Foucher/Majumdar, 1940, 3:PI. 86, 66b). (Marshall/Foucher/Majumdar, 1940,3: PI. 103, b). (Marshall/Foucher/Majumdar, 1940,3: PI. 101). (Marshall/Foucher/Majumdar, 1940,3: PI. 97, 24). (Marshall/Foucher/Majumdar, 1940,3: PI. 103, e). Dharmacakra on the seat (Joshi, 1966: PI. 9) Dharmacakra (Gupta, 1985: P. 76). 8: Pillar supporting the dharmacakra (Kurita, 1988,2: Fig. 791). 10: Dharmacakra (Kurita, 1988,2: Fig. 852). II: Dharmacakra under the arch (Kurita, 1988, 1: Fig. 299). 15 : Dharmacakra on the Buddhapada (Kurita, 1988, I: Fig. 283). 24 : Dharmacakra on the Buddhapada (Kurita, 1988,2: Fig. 783). 24,24 : (Kurita, 1988, 2:Fig. 786). 35,20: (Kurita, 1988, 2:Fig. 784). 45,42: (Kurita, 1988, 2:Fig. 755). 57 : (Kurita, 1988, 2:Fig. 782). 7: Dharmacakra (Ackermann, 1975: PI. H, 15). 9: (Kurita, 1988, I: Fig. 276). 13 : (Ingholt, 1957, 75-Taxila). 13: (lngholt, 1957, 77-Peshawar). 16 : (Kurita, 1988, 1: Fig. 280). 24: (Kurita, 1988, 1: Fig. p3-II). 4, 4, 4 : Triratna (Kurita, 1988, 1: Fig. 288). 6, 6, 6 : (Kurita, 1988, 1: Fig. 295). 8,9,8 : (Kurita, 1988,2: Fiug. 769). 9,9,9 : (Kurita, 1988, 1: Fig. 282). 10,10,10: (Kurita, 1988, 1: Fig. 277). 11, II, 11 : - (Faccenna, 1962: PI. 230, a). 12,12,12: (Kurita, 1988, 1: Fig. 274). 12,13,11 : (Kurita, 1988, 1: Fig. 285). 14, 14. 13 : (Kurita, 1988, 1:Fig. 298) 14,14,14:- (Kurita, 1988, I:Fig. p3-III). (Faccenna, 1962: PI. 119). 8(?), 18, 18 : 24,24,24: (Kurita, 1988, 1: Fig. 294). 9, 12,9 + 10 : Triratna and dharmacakra (Kurita, 1988, I: Fig. 296). 11,11,13 + 14:- (Kurita, 1988, 1: Fig. 291). (Kurita, 1988, 1: Fig. 290). 12, 13, 11 + II 12,13,13+ 15:- (Kurita, 1988, I: Fig. 292). 11, 10, 8 + Lotus: Triratna and lotus flower (Kurita, 1988, I :Fig. 293). 18, 18, 18 + Lotus (Ackermann, 1977: PI. 24, a). Moreover, certain basic differences certainly emerge from a comparison of the wheel-shaped patterns with relief depictions of the actual dharmacakra symbol, which 61 support this view. First of all, the circular hub of the dharmacakra is always realistically portrayed in the reliefs, while the central structure within the stupas is sometimes square, or may, in stupas of the north-west, even be lacking. Secondly, many stupas of this kind have double, even triple concentric walls. These are not depicted at all in the dharmacakara of the reliefs, while multiple rings would be anomalous in the actual wheel of a cart or chariot. Thirdly, among the depicted wheels, the number of spokes varies too widely to show any evidence of standardization (table 2). For instance, at Sanchi, the spokes of the dharmacakra number 16, 20, 22, 25, 26, 28, 32 and 34 respectively in reliefs from stupa no. 1; 12 and 16 in those from stupa no. 2; andl7, 20, 24 and 27 in those from stupa no. 3. At Amaravati, most of the dharmacakra have innumerable spokes, and even where countable, they number as many as 28, 31, 63, 64 and 81 spokes. In Gandhara, the spokes are depicted in rather reduced numbers: 4, 6, 7, 8,9,10,11,12,13,14,15,16,18, 20 and 24 for depictions oftriratna and wheels supported by a pillar; and 24,35,42, and 57 for those of Buddha's footprint. A comparatively large number of spokes, seen in particular at Amaravati, might be supposed to have been influenced by the Buddhist canons, which say the Wheel of the Law essentially has thousands of spokes. But I rather think that the depiction of numerous spokes on the Amaravati sculptures only followed the elaborate way of expression of the Amaravati school. Although odd numbers are also included in the depicted spokes, the architectural stupas have basically numbers of spokes representable by the following considerations: P (x = 2,3,4,5) or 2X (x = 1,2,3) x 3; e. g. 4 (= 22), 6 (= 2 x 3), 8 (= 2\ 12 (=2 2 x 3), 16 (=2 4), 24 (= 23 x 3) and 32 (= 25). These numbers clearly made the construction of a divided circle much easier at the sites. All of the above discussions lend support to the belief that the wheel-shaped pattern within stupas, regardless of location (north-west or south), was simply a construction device, with no symbolic implication. Apart from symbolism, a brief reference should be made here to a resemblance of the shape contained in these stupas to Vedic altars in the form of a rathacakra or a chariot wheel (fig. 13). According to the Manava Srauta sutra and the Baudhayana Sulba sutra (Satyaprakash and P. R. S. Sharma 1968: 162 ff; van Gelder 1963: ff), an altar in a form of the rathacakra is built for the agnicayana rituals exclusively of baked bricks laid in five layers. The odd-numbered and even-numbered layers of bricks are arranged according to strict regulations. The bricks themselves are also all made under strict regulations as to size, form, as well as number. The same is true for the altar itself: the regulations elaborate the position of the spokes, the hub and the wheel and the number of bricks to be used for them. The sutra also give each odd-numbered layer a meaning: the first layer symbolises the earth; the third the sky; the fifth heaven, the paradise svarga to which the master of the ritual is led. As regards any relationship with stupas, the importance of this altar is that it is built only for a specific rituals and is never re-used. The altar is only considered sacred for the duration of the ritual and it never becomes an object of worship. All of these features of the agnicayana rituals thus are quite inconsistent wih those of stupas with the wheel-shaped pattern. Any similarity is only superficial. Nothing that derives from the Vedic rituals is incorporated in the stupas under discussion here. Moreover, any influence from Vedic traditions of this kind would have been 62 incorporated into the stupa structure at a much earlier period than the date of any of the surviving examples with a wheel-shaped pattern. 5. Origin of the Wheel-shaped Pattern In the earlier stupas there is no trace of the wheel-shaped pattern. At Vaisali, layer 5, which covers the entire area of the mound, is a compact but sandy rath with plenty of white kankars, including sherds of NBP Ware, reddish grey ware and one inscribed coin. According to the excavators, the layer may be assignable to the early NBP Ware Culture, (c. 550-450 BC). On top of the layer was constructed the first mud stupa, about 8. 16 m in diameter, which was then enlarged three times. The northwestern quarter of the last enlargement was encased (or buttressed) in the first century AD (Sinha, B. P. and Sita Ram Roy 1969: 18). Of the successive structures that comprise the Jagannatha Tekdi stupa at Pauni, Maharashtra, the earliest stupa, (diam. 38.2 m) appears to have been a massive construction of large-size bricks set in mortar on a lateritic horizon (Deo, S. B. and J. P. Joshi 1972). A single trench exposed a section of the structure consisting of sixteen courses (h. 1 m) of bricks laid in brown mud mortar. This earliest stupa was encased within the remains of a subsequent enlargement which was built in a box-like technique. The floor surrounding the earliest stupa was covered with soil containing sherds of NBP Ware and Black and Red Ware. A different mode of construction from the earlier stupas cited above is seen at Bhattiprolu stupa. Inscriptions on three stone receptacles found here have been variously assigned to c. 200 BC or even earlier and alternatively, to 100 BC or slightly earlier (Buhler 1894: 323-9; Sircar 1963-4: 224 and 225, n. 5). These dates probably also provide an indication of the antiquity of the stupa itself. According to Rea, the dome measures 40. 26 m in diameter at its base and had a podium 45. 14 m in diameter. The stupa was built of solid brickwork throughout and had a maximum extant height of c. 4. 2 m (Rea 1894;1874: 124): On the top of the mound and in the centre of it is a circular hole 9 in (22. 5 cm) in diameter, which reaches from top to bottom .... The whole structure is one solid mass of brickwork built up in regular courses 6 in (15 cm) in depth .... In the first horizontal ring surrounding the hole there are eight large bricks set in a radiating way. Around the ring thus consisted of eight bricks, sixteen wedge-shaped bricks were placed with the apex pointing to the centre. The triangular interspaces between the bricks are filled in with clay. Surrounding this course of bricks, sixteen bricks are placed under the clay thus forming a sort of band, and so on with each alternate course. The entire circular brick work measures about 2 m in diameter. This mode of construction is limited to the very centre of the stupa and is, therefore, quite different 63 from the wheel-shaped pattern. A similar mode seems to have been found at Piprahwa Main stupa by W. C. Peppe, in so far as he says: "I excavated a trench across the summit of this mound, 10 ft [3 m] broad and 8 ft [2.4 m] deep, and found that the structure was built of bricks measuring 16" x 10.5" x 3" [40 x 27 x 8 cm], or 15" x 10" x 3" [38 x 25 x 8 cm], laid in clay mortar, layer upon layer, in concentric circles"(Peppe 1898: 573). Fabregues (1987: 34) has claimed that techniques used in the construction of main stupas at Dharmarajika and Butkara, in which the hemispheres of the domes are cut by walls radiating from the central axes, are consistent with construction techniques of the oldest period, i. e. contemporary with Shapur Tope near Shah-Dheri (Taxila) (Foucher, 1905: 87) and the Jain stupa at Kankali Tila (Smith 1901: 4). Fabregues's claim is quite strange and untenable. Firstly, from the photograph provided by Foucher, it seems that the Shahpur Tope is most probably identifiable as the Dharmarajika Main stupa at Taxila. Secondly, the "walls radiating from the central axis" of the Butkara Main stupa cannot be regarded in the same context as structural techniques of other sites like Dharmarajika and the Jain stupa at Kankali Tila. At Butkara, "the central axis", as it was called by Fabregues, is not an axis but the core of the stupa. The "radial walls" cannot really be walls, but are solid stone-work horizontally laid in the manner of paved floors surrounding the lowest part of the core of the stupa without any interspaces between them. Faccenna rightly marked the difference between this pattern at Butkara and the radial walls at other sites, saying that the plan recalls the stupas with a wheel-shaped plan, but the building technique is quite different from that involving radial walls (Faccenna 1980: 29, n. 1). The Butkara type of stone structure is in the same category as the central structure of brick at Bhattiprolu and Piprahwa. As Sarkar stated, the stupa with wheel-shaped pattern really appears to belong to an evolved stage of stupa architecture. If so, how is it possible that this constructional form could develop from earlier stupas, where any evidence of such features is entirely lacking? The antiquity of Bhattiprolu, Piprahwa and Butkara predates the earliest possible phase of stupas with a wheel-shaped pattern. The same is true for other stupas like Vaisali and Pauni. Thus, based on the chronological development of stupa construction in India, the wheel-shaped pattern does not seem to be contmpoarary with earlier stupas. Nor did the modes of construction of these earlier stupas serve originally as models for the wheel-shaped pattern. As discussed above, the pattern also did not originate in the rathacakra and never symbolised the dharmacakra. What then inspired this mode of constructional? As far as we can see from the dates, construction of all wheel-shaped stupas was of short duration (i. e. the first-the second century), regardless of their location in the north or the south. The only exception is Nagarjunakonda, the artistic successor of an earlier tradition in the area at the mouths of the Krishna and Godavali Rivers, during the decline of the Satavahana dynasty in the early third century. If one may be allowed to find a historical context for the birth of the stupa with the wheel-shaped pattern within this specified period, the earlier half, or first century, appears to be most significant, coinciding as it does with a peak period for sea trade, which has been recently emphasised by invaluable interpretations of the archaeological materials (Begley and de Puma 1991: 6). In connecton with this evidence, 64 it can also be argued that the construction technique used for stupas of large scale in the South Asian subcontinent mut have been one of the numerous divergent import from Rome. 6. Roman Mausolea and Relevant Sepulchral Architecture. There are striking structural similarities between the circular tombs of stone or brick built in Imperial Rome and its provinces, and the stupas of large scale in the South Asian subcontinent: concentric circles of walls; a thick, outermost wall (hence, a high drum); and a wheel-like plan comprising a hub with the spokes. As Toynbee (1971: 143ff.) notes, one of the most imposing features of Roman sepulchral architecture is the great series of circular masonry and brick tombs, based on Etruscan models. None of the series is pre-Augustan, the proto-typeis an accepted view, and it was Augustus's mausoleum built by the emperor in 28 BC in the Campus Martius that inaugurated the whole series. None of the series is pre-Augustan. The mausoleum of Augustus (fig. 18) is about 87 m in diameter at the base (Dlus. 12,1; table 1) (Eisner 1979: 319 ff.; Virgili, 1984: 205 ff.). In the centre of the sepulchral chamber, a pillar rising to the total estimated height of the monument (c. 44.65 m), supported a statue of Augustus that crowned the apex of the tomb. Around this medial point are five concentric circles of walls that probably supported the superstructures. These decrease in height from the centre to the perimeter and are strengthened by twelve walls radiating most probably from the second circular wall to the outermost one. Each of the twelve interspaces divided by these walls is strengthened with a wall of recumbent arches built against the outer face of the fourth circular wall and also with further radiating walls; the outermost ring space has twenty-four radiating walls in all. Thus the mode of construction of the interspaces between the walls of the outermost ring and the fourth ring helps withstand the pressure from the inside and increases stability. Stability and prevention against cracking under pressure were also taken into consideration in the construction of the outermost circular wall by building it far thicker than that of the other four. The technical similarity should be stressed between the Mausoleum of Augustus and stupas in the South Asian subcontinent, in Andhra Pradesh in particular, where a thick outermost wall was also required when building a high drum. Circular burial mounds of varying size, with spokes and with or without a hub, were constructed in the northern and norht-western Roman provinces mainly in the course of the first two centuries. Toynbee rightly pointed out that the stimulus can hardly have come from elsewhere than Augustan Italy, where the long series of circular Roman tombs and mausolea took its rise and with which Britain as well as northern Gaul and the central European provinces and borderlands were in close cultural contact; it is undoubtedly a far cry from the Mausoleum of Augustus to even the largest and most impressive provincial tumuli; but all the latter shared with their distant Italian relatives their circular shape, earthen mounds, spokes, hub, recumbent arches, and curving walls ... ; these features obviously recall the concentric circles of walls and radiating spokes below 65 Augustus' Mausoleum in Rome-possibly also its crowning monument. It will suffice here to cite several examples of probable Augustan date in Britain and Belgium, listed by Toynbee (1971: 182f.), Collingwood and Richmond (1969: 170-171): 1. The round tomb at Keston in Kent is very likely to have been a small drum-and-cone tumulus. It has an overall diameter of 8. 7 m, a flint ring-wall o. 9 m thick, no entrance and six large radiating external buttresses. 2. The much larger tumulus at West Mersea in Essex has twelve exterior buttresses ones, an overall diameter of 19. 5 m and a ring-wall O. 9 m thick. The tomb is the so-called "cartwheel" type, with a small central hub from which six spokes radiate to join the ring-wall, in corresponding line with alternate buttresses. The spokes have their counterparts in tumuli, one with eight spokes, the other with only four, near the fifth milestone of the Via Appia in Rome. A tumulus beside the Via Tiburtina has eight radiating spokes, but no hub, while a tomb on the Via Salaria has six spokes joining a series of curved walls.(9) 3. Toynbee further says that the Belgium tumuli are structurally the same as those in Britain, but are on the whole larger. The tumuli range from 12 m to 43. 5 m in diameter and mostly date from the 2nd century AD. There can indeed be little doubt, Toynbee says, that the pprovincial builders 4. of the Antoing tumulus, the three Treviran tumuli, as well as the Keston and West Mersea tumuli in Britain, must have had Roman architectural sketch-book or manuals at thier disposal. This must also apply to those who erected the homogeneous group of four additional Treviran tumuli: two near Trier, one at Nennig and one at Fremersdorf (now vanished) (Toynbee 1971: 184). 5. The tumulus at Fremersdorfhad a hub from which four spokes radiated to join the circle of recumbent arches. As at the Keston and West Mersea tumuli, in all the above four Treviran tumuli, the ring-wall is basically a "fence" designed to support and retain the earthen cone (unlike the high, monumental masonry drum, topped by a mound, of the great circular mausolea of Augustus, Caecilia Metella, Hadrian in Italy). These stone "fences" recall the palisades round the "beaker-culture" tumuli and the encircling dry stone walls of the Bronze Age and Hallstatt period tumuli in these regions. Regarding the above four tumuli, Toynbee evocatively points out that the nature of the ring-wall is basically a "fence" retaining the cone. In all of them the breadth of the ring wall is really not thick, unlike the high, monumental masonry drum of the great Italian mausolea. This may well imply that, even in case of stupas, the thick outer wall was a device for building a high drum. Also significant for the emergence of stupas with the wheel-shaped pattern in the South Asain subcontinent is Toynbee's doubtless claim that the provincial builders had at their disposal architectural sketch-books or manuals compiled in Roman heartlands.<IO) Wheeler says that from the woods of Hertfordshire to the palm-groves of the Coromandel, red-glazed cups and dishes are symbols of the adventures of routine traders whose story may be set only a little below that of Alexander the Great himself (Wheeler 1955: 178-179). Really the same is true for the architecture of the wheel-shaped pattern. 66 7. A Provisional Conclusion Prototypes for stupas with a wheel-shaped pattern cannot be found in the South Asian subcontinent itself. There can be little doubt that neither the dharmacakra nor rathacakra provided inspiration for the emergence of the pattern. If they had been the models for stupas, the question why the first emergence of the pattern dates from the first century AD would have been answered. Any similarity is only superficial. The real purpose of the design was to provide stability to the lower part or to the drum of the stupa, particularly in those cases of stupas with a ring wall of thicker breadth. Though some scholars also consider its purpose was to economise on materials, this function may only be considered valid for brick structures, since making large numbers of bricks was laborious. I cannot think it likely that the stupa-builders of Dharmarajika at Taxila and Fil Khana near Jalalabad economised on stone materials for the wheel-pattern, even if quarrying and dressing processes are considered. When all the above points are borne in mind, the most significant fact appears to be that the first century AD is the earliest date for the pattern. That century saw the apex of sea trade between India and Imperial Rome, where the first structure with a wheelpattern, the Mausoleum of Augustus, was built in 28 Be. The same basic pattern, albeit with occasional modifications, became prevalent in other circular tombs, not only in Rome and nearby, but also in the north and the north-western Roman provinces during the first and second centuries. The contemporaneity of wheel-shaped patterns in Romanized, secular, circular tombs and in Buddhist (or, in one instance, Jain) religious monuments is no casual coincidence either. Whether or not any occidental examples are located in the regions between the South Asian subcontinent and Rome is not important. What is significant is that the builders of the subcontinent required such a mode of construction, and the wheel-pattern in stupas thus became the eastern counterpart of the Roman provincial structures. What was the real purpose of introducing the Roman mode of sepulchral architecture? Was it used only as a means of internal rinforcement for the stupas, without changing the traditional exterior appearance of a low podium and dome? Particularly in Andhra Pradesh, this Roman import enabled the height of the drum to be increased, so that stupas became much more loftier in appearance. The existence of a thicker breadth for one of the ring walls supports this contention. On the other hand, in Gandhara, the spokes sometimes were clearly used not in constructing the drum, but the dome, so thicker ring walls are lacking. The stupa builders in Gandhara were not inclined to elevate the circular drum, but, instead, must have introduced another Roman design in sepulchral architecture: a high square podium. An example of a Roman mausoleuth a square podium is the tomb of Caecilia Metella beside the Via Appia (22. 3 m square base, with a round drum 29. 5 m in diameter). Other examples include the tomb at Pozzuoli near SanVito and the mausoleum at Marano near Naples, both of which date from the late first or early second century. The former originally had panels flanked by stuccoed pilasters, that supported pediments 67 or half-pediments and a cornice above. The latter had a quite elaborate terracotta cornice along the top and brick pilasters alternating with panels of opus reticulatum masonry that were surmounted by pediments and half-pediments in brick (Toynbee 1971: 155-156). With decorative elements based on architectural designs, the square podium was first introduced in Gandhara at Sirkap II before the earthquake. The early examples are the stupas built of rubble with a facing of squared kanjur in Sirkap Blocks IF and 1G (fig. 11) and in Court B in the northern part of the Dharmarajika complex. The base of the podium of the Block IF stupa has a well-cut moulding consisting of a torus and scotia divided by a fillet, above which is a row of pilasters surmounted by brackets with a frieze and dentil cornice at the top. The whole facing was originally finished with a thin coat of fine stucco, but several other, and coarser coats were later added, some of which showed traces of red, crimson and yellow paint (Marshall 1951: 163). If diaper masonry was, as Marshall thinks, a new masonry style adopted after the earthquake and really did form the facing of the Dharmarajika Main stupa dome with its internal reinforcing spokes, then the introduction of the square podium in Taxila predates the spoke structure. In wheel-shaped patterns there is a striking difference, apart from the materials, between Andhra Pradesh and Gandhara. Gandharan stupas lack thick ring walls, and in most cases, also hubs. In this respect, the examples at Sanghol are quite significant, for they have square podiums, hubs and thick ring walls, thus sharing common characteistics with both Gandhara and with Andhra Pradesh. Yet the following question remains open: was the Roman wheel-shaped design first used in Gandhara or Andhra Pradesh? Geographical and archaeological considerations suggest the latter: its proximity for sea trade, the extensive finds of Roman exports in south India and the much greater number of known stupas with the pattern located there. Moreover, the Ghal).tasala stupa seems closer to the Roman original, although its exact date of construction is not known. Indeed, a problem in answering the above questions is the lack of precise dates for all the Andhra Pradesh stupas with the wheel-shaped pattern. Weare at present obliged, rather disappointedly, to admit such a broad time-scale as the first century AD for the first introduction of that pattern in this region. On the other hand, in Gandhara materials for re-examining the dates are available to the extent that construction of that pattern at Dharmarajika in Taxila and at Shah-ji-ki Dheri does not appear tp predate the reign of Kujula Kadphises. Based on the absence in Gandhara of such elements of Roman origin as the hub and thick ring wall, I suggest that the Gandharan stupas with a wheel-shaped pattern were generally a little later than those in Andhra Pradesh. If the Dharmarajika and Shah-ji-ki Dheri examples are datable to the time of Kujula Kadphises, then the first date of stupas with the wheel-shaped pattern in Andhra Pradesh falls in between 28 BC and Kujula Kadphises. 68 v Shah-ji-ki Dheri before Kanishka The Buddhist travellers from China to Gandhara in the fifth-eighth centuries describe a huge stupa as founded by Kanishka locating it several Chinese miles to the east or the southeast of Purushapura, while a site comparable in both size and location with Kanishka Stupa had been known in the nineteenth century as Shah-ji-ki Dheri lying outside the Lahore Gate of Peshawar. With the "poor and scanty nature of the discoveries" from his digging at the site, Crompton (1875: 719) corroborated Cunningham's supposition (1871: 80) that "no remains of this great stupa existed," and stopped further work saying that "it certainly is not worth while continuing the explorations here." Sticking to Foucher's identification (1901: 328ff.) of Kanishka stupa with Shah-ji-ki Dheri, however, Spooner and Hargreaves performed afresh a series of excavations from 1908 to 1911 at the circular, easternmost mound surrounded by cultivated fields, to the west of which lay the 6 m high extensive mound where Crompton opened most of his trenches and which Hargreaves later designated the "Monastery Mound." The results were circulated in the following six publications: Spooner, D. B., Excavation at Shah-ji-ki Dheri. Annual Report of the Archaeological Survey of India, Frontier Circle, 1907-08. Peshawar, 1908. pp. 17-22. Spooner, D. B., Excavations at Shah-ji-ki Dheri. Annual Report of Archaeological Survey of India 1908-9. Calcutta, 1912. pp. 38-59. Hargreaves, H., Excavation at Shah-ji-ki-Dheri. Annual Report of the Archaeological Survey of India, Frontier Circle, 1910-11. Peshawar, 1911. pp.12-16. Stein, A., Excavations at Shah-ji-ki-Dheri. Annual Report of the Archaeological Survey of India , Frontier Circle, 1911-12. Peshawar, 1912. pp. 1-2. Marshall, J., Shah-ji-ki-Dheri. Archaeological Survey of India, Annual Report of the Director-General ofArchaeology, Part 1,1911-12. Calcutta, 1914. p.11. Hargreaves, H., Excavations at Shah-ji-ki Dheri. Annual Report of Archaeological Survey of India 1910-11. Calcutta, 1914. pp. 25-32. According to Spooner (1912: 40-48), he first dug up a shrine-like small building on the brick pavement in front of the square main body of the stupa and also found the south face of the stupa to the west of the southern projection. Also uncovered was the rough masonry of the west side of the southern projection. This was decorated with a stucco frieze of a series of seated Buddha images and pilasters supported by mouldings. The frieze on the face of the projection was not directly applied to it but to a mud coating of about 52. 5 em in width that separated the frieze from the rough masonry of the projection. The frieze had been broken at both north and south ends, and only the 69 lowest parts, such as the mouldings, continued further around the base of the projection. The masonry of the south side of the square body also partly remained intact in the area near the southeastern corner. From here to the northeastern corner of the stupa nothing remained except for the east side of the eastern projection. On the north side of the stupa was found the east face of the northern projection which was also faced with the same rough masonry as on the south projection. Excavation was continued on to the west side of the stupa where Spooner found everything surviving better than elsewhere: At the two angles, north and south, where the western projection met the square body, the stucco friezes were revealed decorating not only the projection but also the stupa itself. Spooner clearly found that on the north side of the western projection the coating of mud daubing on the rough masonry wall measured 52. 5 cm in width while on the side of the stupa itself the similar coating measured 30 cm in width. At the northwestern corner of the square the existence of a circular base of a tower was confirmed. Spooner presupposed the existence of the same sort of base at the three other corners, but found nothing except for a stone possibly attributable .to the southeastern tower. Thus the excavations revealed a 54 m square main stupa with a circular extension at the northwestern angle (most probably at each angle) and a 15 m projection on each side making a cross-form, surrounded by other smaller stupas and a shrine-like structure which stood on the brick pavement (lIlus. 13). The stupa also contained four broken walls, all originally radiating from the center to the northwest, west, southwest and southeast. A little to the west of the stupa three superimposed constructions were exposed, the lowest two being of brick and the latest of brick and stone, comprising together square pillar bases and walls faced with circular or semicircular towers identified by the excavators as part of monastic buildings. The extraordinarily extensive size of the stupa complex thus well matched the stupa founded, according to legend, by Kanishka which the Chinese pilgrims unanimously talked about as by far the greatest among all stupas in the Western Regions. Furthermore, a crude bronze casket recovered from off center on the natural soil seemed to have strengthened the identity when Spooner (1912: 51 ff.) and Konow (1929) found the name of Kanishka and a date, Year 1 of the Kanishka era, in the dotted Kharoshshi inscriptions. The reading was so influential as to lead scholars to produce a number of debates in relation to the first visualization of the Buddha. However, technical treatments in the British Museum (Organ and Werner 1964) made possible Mukherjee's reading (1964) that revealed no evidence of a date despite his reaffirmation of Kanishka's association with the casket. A new interpretation of the inscription provided by Fussman (1987: 77 ff.) has further separated the site from Kanishka, as Kreitman (1992: 195-197) recently summarizes: Fussman "acknowledges Kanishka neither as the founder of the stupa nor as the donor of the casket, the donation being assigned instead to Mahasena and Sangharakshita," low-ranked monks in charge of a fire-room, and the apparently illegible date "does not necessarily fall within the period of Kanishka's reign," as Rosenfield (1967: 259-262) pointed out. Only Dobbins (1971) prefers Kanishka I as the donor. 70 Contrary to much attention in the inscription and casket, little concern has hitherto been given to how and where this important casket was found in close relation to the structure of the stupa itself. Some questions, as to the find spot of the casket in particular, arose during my examination of the radial walls for another topic, namely, stupas containing a wheel-shaped structure inside. As soon as that work was tentatively completed-part of the summary is necessarily included in the chapter below, I looked into the ways to better understand Shah-ji-ki-Dheri stupa and the casket. Unfortunately prevented by village houses and graveyards today covering the mound any further reexamination of the stupa at the site itself is hardly possible. The only way left to us for reevaluating Shah-ji-ki Dheri stupa, therefore, is a fresh excavation of the aforesaid reports. Yet the excavators have left only 44 pages of descriptions, no more than three sheets of line-drawn plans and several unclear photographic illustrations. Besides such absolute paucity of information, the descriptions contain ambiguous explanations and often lack details we really need, some of the illustrations hardly providing any information. Despite such difficulties, everybody can easily realize that the site produced only a few of the pieces of schist sculpture which are normally recovered in great number at Gandhara Buddhist sites and also that baked bricks were familiar material for the builders of Shah-ji-ki Dheri Main stupa and other buildings surrounding it. The use of baked bricks as a building material in Buddhist temples throughout Gandhara, Taxila and Swat is highly unusual: After digging a number of Buddhist sactuaries at Taxila Marshall refers only at Bhamala to the baked bricks which however were used as a material for pavement around the main stupa, not for the facing of wall as at Shah-ji-ki Dheri. If only for these peculiarities the Shah-ji-ki Dheri buildings are quite exceptional and anomalous among comparable monuments. The plentiful use of bricks may suggest a fairly late date for the site-but not because of the geographically difficult access to stone as the excavators supposed. Yet the discovery of the "Kanishka Casket," even if it was produced later than that monarch, resists assigning the entire site to a late date. The coexistence of such earlier and later factors does not support a single building activity. If so, how many rebuildings did the site undergo? Where should the earlier stupa be found? What is the later phase? To which phase of rebuildings is the "Kanishka Casket" attributable? If those questions go unanswered, the casket cannot correctly be interpreted in the history either of the site itself or of Gandharan Buddhist monasteries. Thus questions arising from ambiguous descriptions in the excavation reports, primarily of the find-spot of the casket, have led me not only to carry out an overall inspection of the Spooner and Hargreaves reports but also to check the ASI Frontier Circle's photographs in the British Library. Among the photographs taken during the excavations some that escaped from the excavation reports show stucco images of tlle seated Buddha in front view. All that was published by the excavators about the stucco images at Shah-ji-ki Dheri, however, is the one taken from the side that gives no clue to solve the date from stylistic points. The images in front views manifestly show the characteristics ascribable to a style posterior to the usual Gandharan stucco Buddhas. This fact has basically led me to a new understanding of Shah-ji-ki Dheri. 71 According to the plans and descriptions given by Spooner and Hargreaves the Shah-ji-ki Dheri Main stupa had several radial walls incorporated within a cross-form framework. Though destroyed and disconnected, they clearly indicate that the stupa had once contained a wheel-shaped structure. Is this structure actually framed by the cross-formed structure or only concealed by it? The answer is the latter and the main stupa had been circular with this structure inside as a foundation before a square stupa which later transformed itself into the cross-form. This paper only deals with how a circular stupa can be identified within a cross-form. To reach this conclusion it is first necessary to locate Shah-ji-ki Dheri's place within the broader context of stupas with the wheel-shaped structure which have wrongly been interpreted as symbolizing the dharmacakra. 1. The Wheel-shaped Structure As Imported From Augustan Rome Possibly datable to the first three centuries, stupas with that structure are geographically concentrated in the areas around the mouths of the Krishna and Godavari Rivers, such as Adurru, Alluru, Ghantasala (Fig. 2: 2), Kodavali, Nagarjunakonda, Peddaganjam, Salihundam, and possibly Amaravati. Further, to the northwest, in Maharashtra, there is a large stupa with this pattern inside at Ter, known as Tagara in the Periplus Maris Erythraei (Casson 1989: 212-213). The structure within these stupas serves as the lowest part of a stupa which basically consists of a significantly thicker tyre-wall and several spoke-walls radiating from the central hub, circular in most cases but square in some. Also some are internally strengthened by a few intermediate concentric walls. Strangely, the wheel-shaped cylindrical structure contained in stupas has reminded everyone of the dharmacakra, as stated by writers such as G. B. Sharma and S. P. Gupta (Gupta 1985: 50), Sarkar (1962: 78), Huntington (1985: 179), Snodgrass (1988: 78 ff.), and Kottkamp (1992: 73-78). However, the most serious fallacy in conceiving such stupas as symbolizing the dharmacakra lies in the fact that the design of the wheel-shape is thereby only evaluated one-dimensionally, in horizontal section from above, while the importance of its elevation, viewed in vertical section, is totally ignored. If one only considers the plan in horizontal section, the wheel-shaped pattern ostensibly looks like a wheel. But when both horizontal and vertical sections are assessed together, it cannot be a wheel: the elevation of the wheel-shaped pattern in the aforesaid examples is too high to represent the dharmacakra. Moreover, certain basic differences clearly emerge from a comparison of the wheelshaped structures with relief depiction of the actual dharmacakra symbol. First, the circular hub of the dharmacakra is always realistically portrayed in the reliefs, while the central structure within the stupas is sometimes square. Secondly, many stupas of this kind have double, even triple concentric walls. These are not depicted at all in the dharmacakra of the reliefs, while multiple rings would not make any sense in the actual wheel of a cart or chariot. Thirdly, among the depicted wheels, the number of spokes 72 varies too widely to show any evidence of standardization, while the architectural stupas have spokes within the limit of numbers such as 4, 6, 8, 12, 16, 24 and 32. This clearly made the construction of a divided circle much easier at the sites. All of these factors lend support to the belief that the wheel-shaped pattern within stupas was simply a construction device, with no symbolic implication. Symbolists should explain the reasons why not a single sign of the dharmacakra appeared along with the earlier stupas, e.g., those at Vaisali (Sinha and Roy 1969), Piprahwa (Peppe 1898), Pauni (Deo and Joshi 1972), Bhattiprolu (IA 1874), and Sanchi (Marshall et al. 1940) and why stupas suddenly appeared in the first century with that pattern inside. Actually, there is no clear evidence for various stages of development from the stupas without the wheelshaped pattern to those with it inside. One cannot deny the chronological gap between the stupas at Vaisali, Piprahwa, Pauni, and Bhattiprolu and those with the wheel-shaped structure at Adurru, Alluru, Ghantasala, Kodavali, Nagarjunakonda, Peddaganjam, Salihundam, and Amaravati. Apart from symbolism, a brief reference should be made here to a resemblance of the shape contained in these stupas to Vedic altars in the form of a rathacakra (the wheel of a chariot). According to the Manava Srauta Sutra and the Baudhayana Sulba Sutra (Satyaprakash and Sharma 1968: 162 ff.; Gelder 1963: 311 ff.), an altar in the form of the rathacakra is built for the agnicayana rituals exclusively of baked bricks laid in five layers of which the arrangement of bricks in the odd-numbered and even-numbered layers differs according to strict regulations. The baked bricks to be used are also all made under strict regulations as to size, form, as well as number. The same is true for the altar itself. The regulations also intricately refer to the placement of the spokes, the hub and the tyre and the number of bricks to be used for them. The sutras also give each odd-numbered layer a meaning: the first layer symbolizes the earth, the third the sky and the fifth heaven, the paradise svarga to which the master of the ritual is led. As regards any relationship with stupas, this altar is built only at the time of the rituals, and the remains of the altar are never used again. The altar ceases to be sacred after carrying out the rituals and never comes to be an object of worship. All of these features of the agnicayana rituals are quite inconsistent wih those of the stupas with wheel-shaped pattern. Only the outward appearance of the shape is similar to that of the stupas, and nothing derived from the Vedic rituals is incorporated in the stupas now in question. Further, any influence from Vedic traditions of this kind would not have found its way into stupa construction earlier than the time of construction of any of our surviving examples with a wheel-shaped pattern. The above evidence shows that technique of building a wheel-shaped structure does not seem to have been fostered in the Subcontinent. Evidence exists which favours the introduction as a new idea from outside of the Subcontinent. If one searches for a historical context conducive to the birth of the stupa with that pattern, profound significance can be attached to the first century: this is the peak period for sea trade between Rome and peninsular India, as recent interpretations of the archaeological materials have invariably and invaluably pointed out (Begley and de Puma, 1991: 6 in particular). Builders must have found the technique of construction for stupas, of larger 73 scale in particular, among the divergent imports, tangible or intangible, from Augustan Rome (BC 28-AD 14). In fact, there is a striking similarity of structure between circular tombs built of stone or brick in very early Imperial Rome, and its provinces as well, and stupas of large scale in the Subcontinent: the concentric circles of walls, the thick outermost wall (hence, a high drum or a high plinth), and the spokes and hub like a wheel in plan. As Toynbee (1971: 143 ff.) says, "one of the most imposing features of Roman sepulchral architecture is the great series of circular masonry and brick tombs. That their models were in Etruscan cemeteries is an accepted view, and it was Augustus's mausoleum built by the emperor in 28 BC in the Campus Martius that inaugurated the whole series (Illus. 12, 1). None of the series is pre-Augustan". The mausoleum of Augustus is about 84 m in diameter at the base. In the centre of the sepulchral chamber is a pillar rising to the total height of the monument, calculated as about 44.65 m, which originally supported a statue of Augustus that crowned the apex of the tomb. Around this medial point are five concentric circles of walls probably for supporting the superstructures, decreasing in height from the centre to the perimeter and being strengthened by twelve walls radiating most probably from the second circular wall to join the outermost one. Each of the twelve interspaces divided by these walls is strengthened with a wall of recumbent arches built against the outer face of the fourth circular wall and also with the radiating walls; the outermost ring space has twenty-four radiating walls in all. Thus the mode of construction of the interspaces between the walls of the outermost ring and the fourth ring helps to withstand the pressure from the inside and increases stability. Stability and prevention against cracking under pressure were also considered in the construction of the outermost circular wall: it was built far thicker than that of the other four. The technical similarity between the Mausoleum of Augustus and stupas in Andhradesa should be stressed in the case of this thick outermost wall that was required in the case of building a high circular plinth. Circular burial mounds of varying sizes with spokes, with or without a hub, were also constructed mainly in the course of the first two centuries by the populations of the northern and northwestern Roman provinces. Toynbee (1971: 180) rightly pointed out: ... the stimulus can hardly have come from elsewhere than Augustan Italy, where the long series of circular Roman tombs and mausolea took its rise and with which Britain as well as northern Gaul and the central European provinces and borderlands were in close cultural contact; it is undoubtedly a far cry from the mausoleum of Augustus to even the largest and most impressive provincial tumuli; but all the latter shared with their distant Italian relatives their circular shape, earthen mounds, spokes, hub, recumbent arches, and curving walls; these features obviously recall the concentric circles of walls and radiating spokes below Augustus' mausoleum in Rome, possibly also its crowning monument. Really the same is true for our stupas with this pattern, which might be taken as 74 eastern, or Asiatic counterparts of the northern and northwestern Roman provincial tumuli. The wheel-shaped patterns in the Subcontinent and the Roman provinces must have stemmed from the Italian prototype, namely, Augustus's mausoleum, despite quite different cultural contexts. Wheeler (1954: 150) says that "from the woods of Hertfordshire to the palm-groves of the Coromandel, red-glazed cups and dishes symbolize the routine adventures of tradesmen whose story may be set only a little below that of King Alexander himself." In this statement the wheel-shaped structure may well be included even if no tradesman was directly involved. Significant for the emergence of stupas with the wheel-shaped pattern in the Subcontinent is Toynbee's (1971: 184) doubtless claim that the provincial builders had at their disposal architectural sketch-books or manuals compiled in southern lands. Even apart from sea-borne trade between Rome and peninsular India, it is well documented by Strabo, Dio Cassius, and others that "Indian embassies" reached the court of Augustus first in 26 BC, then successively in the years 25, 21, and 14-11 BC (Warmington 1974: 35-38). As the Res Gestae mentions, "embassies from kings in India were frequently sent to me (Augustus) ; never before had they been seen with any Roman commander". It is undoubtedly believable that the embassy from the Subcontinent in Rome around 26 BC actually saw the Mausoleum of Augustus, the most colossal building in the world of that century, informed about how such a huge, circular building had architecturally been possible. It is very significant to recall that a fairly low, circular plinth had been usual with earlier stupas of greater size, as explained above, while a higher circular plinth was already popular among stupas of lesser size in or before the first century such as those hewn out in the cave temples in West India or depicted in relief at Sanchi and elsewhere. Such a difference in the outward appearance of stupas between greater and lesser sizes may have been derived from a technical problem: when desired in the case of smaller stupas a higher plinth might have easily been attained, while, in the case of large-scale architecture in the open air, building techniques for high circular plinths which would prevent cracking under the pressure may not have been available, even if the problem of building with huge numbers of bricks could be solved. Briefly speaking, they could not build in the open air a higher circular plinth which was indeed desired at that time. The situation changed as early as the first century for builders of stupas in Andradesa, and Maharashtra, too, when they suddenly became familiar with the wheelshaped structure introduced from Augustan Rome. Built inside the plinth the wheelshaped pattern helped to withstand the pressure from the inside, prevent cracking, increase stability and economize materials and labour when coupled with an outermost ring wall of great breadth. Thus, by introducing the architectural technique of the cylindrical wheel-shaped structure they first succeeded in elevating greatly the circular plinth of stupas of monumental size in the open air. 2. Northern Types of the Wheel-shaped Structure as Learned from the South 75 Stupas with the wheel-shaped pattern also found their way to northern districts: a Jaina stupa at Kankali Tila (Smith 1901), Mathura, and the three recently excavated Buddhist stupas at Sanghol (IAR 1971-72; 1984-85; 1985-86; Gupta 1985), near Ludiana, are the only examples to the east of the Punjab. Dharmarajika of Taxila (Marshall 1912-13; 1951) is in the extreme north of the Punjab. Tahkal Bala A (Errington 1987), Fil Khanah of Jalalabad (Mizuno 1962) and Shah-ji-ki Dheri are all to the west of the Indus. These northern examples include different types as follows: (1) Dharmarajika Main Stupa (Illus. 12, 6): It has only radial walls converging to the center without any outer ring wall being applied to the dome structure. The interior of the dome and drum is built of rough rubble masonry and strengthened by sixteen walls radiating from the central core with a thickness of 3 ft 2 in (0. 95 m) to 4 ft 9 in (1. 43 m). According to Marshall (1912-1913: 10) these construction walls stop short above the berm of the stupa, instead of being carried down to its foundations and that obviously belong to a subsequent reconstruction. Based on the description and plan given by Marshall, the central core of the stupa is a solid, round structure with the same kind of masonry as that of the walls which radiate from it. It measures about fifty ft (15 m) in diameter, which is quite enormous when compared with the total diameter of the stupa. Thirteen of the original sixteen spokes are extant. Of the existing walls, only the six in the northern half of the stupa seem to converge toward the supposed central point rather regularly; the others are splayed like a vortex whirling anti-clockwise from the outer face of the central circle. The wheel-shaped pattern applied to such an upper element of the stupa as the dome is never met in southern examples, not to mention the whirling radial walls and the enormous size of the central circular structure. That the Dharmarajika radiating walls are definitely datable before Kanishka I is discussed later in detail. (2) Fil Khanah Stupa (lllus. 12, 5): The stupa, with spokes inside, is on the top of a conglomerate hill several kilometers to the northwest of Jalalabad on the opposite side of the Kabul River. There are two mounds with stupa remains that were most probably associated with the caves locally called Fil Khanah hewn out of the side of the hill. One of the stupas had been demolished to a mere pile of stones. A vestige of the other, of huge scale, remained when the site was surveyed in 1962 (Mizuno 1962: 76) measuring 13 m in height and about 15 m in diameter at the base of the drum which was supported by a low, circular plinth with a diameter of 18 m. These were built on two rectangular podiums, the upper one much taller. The front of both survived with projections for flights of steps leading up to the round drum to the north. The drum with a decayed dome above it was constructed with eight walls regularly radiating from the centre where there was no hub structure, as at Kankali Tila. The core of the radial walls was built of river boulders jointed with clay; the facing was of small, thin, schist slabs fitting tightly together. The interspaces between the radial walls were filled in with clay and river boulders. A sketch drawn by Simpson, which is quoted in Mizuno's report, clearly shows a part of the band of 76 stucco-plastered arches and pilasters with Corinthian capitals supported by "Indian" pilasters which decorated the upper part of the drum. Thus the Fil Khanah wheelshaped pattern shares a common feature with that at Shah-ji-ki Dheri as to the location of the pattern inside the stupa, while the non-existence of the hub is common to both theShah-ji-ki Dheri and Kankali Tila as explained below. The exact date of the Fil Khanah stupa is not clear, but the rectangular podiums should be later than the square ones which were first introduced in Sirkap in the time of Azes II - Kujula Kadphises, were arranged in the form of small stupas in a circuit around Dharmarajika Main stupa, and faced with dressed kafijur blocks in the case of stupa No.4 in Court A to the north of the precinct of the main stupa. (3) Stupa in Sirkap with Radiating Walls: According to Marshall the wheel-shaped structure at Dharmarajika is to some extent comparable with the foundation walls of the little stupa in Block 1C' in Sirkap. Comparison of the radiating walls with those in Block 1C' is incorrect, and 1C' should be read as 1E'. All that is left of this stupa in Block IE' is the high plinth, measuring 40 ft (12 m) north to south by 36 ft (10. 8 m) east to west, constructed partly of diaper masonry and partly of rubble and strengthened internally by thick foundation walls arranged both cross-wise and diagonally, interspaces between them being filled with stone debris (Marshall, 1951: 183). This high plinth, though not having a circular facing inside, structurally shows the same device as the foundation of stupa No.2 at Sanghol. (4) Sanghol Stupas: Examples of another type with the wheel-shaped pattern settled within the square plinth. This pattern within the square plinth is quite anomalous if compared with the southern examples which are structurally close to the Roman examples. At Sanghol three stupas have been excavated so far, two of them having structures of this type and another being of a normal type without the square plinth. Stupa No.1 (lllus. 12, 3): The spokes of the wheel are represented by brick walls radiating from a central, hollow column or hub (IAR 1971-72: 39), intersected by three successive circular walls concentric to the hub. The exterior diameters of the three concentric walls are 5. 04 m (7 m based on the plan given in Gupta 1985), 13 m and 16. 3 to 16.8 m. The circular wall of 13 m diameter is 1. 07 m thick and possibly supported the superstructures. The outer circular wall, O. 3 m wide, touches at each cardinal point the straight wall which forms a square of 17 x 17m. The three intervening ring spaces are divided at regular intervals by the radiating spokes of similar brick masonry, the extant highest point of which is 1. 18 m. Twelve spokes are placed in the first space, twenty-four in the second, and thirty-two splay outward to end at the third concentric wall. The interspaces between all of these walls are filled up with kankar and yellow earth. The entire structure seems to be built on the natural soil (IAR 1972-73: 28), although the first report (IAR 1971-72: 39) concludes that the pre-structural deposit yielded sherds of crude black and grey wares, normally associated with the sherds of Northern Black Polished Ware. 77 Stupa No.2: The earlier report merely says that it is a structural complex comprising a small stupa having two circles, 1. 45 m and 3. 70 m in interior diameter, with eight spokes and a monastery on all three sides (IAR 1985-86B: 69). Yet the photograph attached to the report (IAR 1985-86B: PI. 23) reveals that all the spokes run from a large, central, circular brickwork to the circular, inner face of the outer wall, and the outer face of that wall forms a square. If the photograph was taken at the final stage of digging, the stupa supposedly was raised on a square terrace strengthened by the wheel-shaped pattern. In fact, the latest brief report (IAR 1989-90: 91, 93) says that the stupa is 20 m away from stupa No.1 on its southwest side, and the platform, measuring 8. 50 x 8. 50 m square, was erected to strengthen the entire stupa from the outside with a flight of steps on the southern side. According to the excavation report, the stupa has three ring walls intersected by 24 and 12 spokes in the outer ones, while the third inner one could not be excavated. It is not clear how the earlier photograph and this later report can be reconciled. The bricks measuring 36 x 22 x 6 cm were laid in English bond in mud-mortar (IAR 1989-90: 93). Stupa No.3: The latest excavation report (IAR 1989-90: PI. XXVII) very briefly explains the discovery of a third stupa on the southeast side of the monastery in the same excavation area with stupa No.2. This stupa had almost been deleted down to the ground and only a small portion was excavated. However, it appears to have had at least two ring walls intersected by spokes radiating from a solid hub structure. The device of making a circle within a square is common to stupas Nos .. 1 and 2 and probably comparable to the square plinth of the stupa in Block IE' of Sirkap, although the latter has no circular inner face (see No.2). According to the report cited above, among the finds at Sanghol giving clues for dating are numerous seals and sealings bearing inscriptions in Brahmi mainly and Kharoshthi, coin moulds of schist slate, copper coins found in abundance belonging to Indo-Parthian, Kushan and Huna rulers, and an ivory comb notably recalling one from Taxila. The report further says that the most important find is an inscription written in ink in a concentric fashion on the interior of a late Kushan bowl, which, according to the report, is datable to circa second to third centuries AD based on palaeographical considerations. Situated midway on a route leading from Mathura to Gandhara, the entire outward appearance of stupa No. 1 at Sanghol raised on a square platform recalls usual Gandharan stupas, on the one hand, while, on the other hand, the 117 pieces of sculptured railing pillars, mostly found concealed beneath the earth as a hoard, are of the Mathura school of carvers. Not only did the major Jain stupa at Kankali Tila in Mathura share the wheel-shaped pattern with the Sanghol ones, but, as S. P. Gupta quite significantly notes, the Sanghol railing pillars are nearest in dimensions as well as workmanship to the Kankali Tila sculptures which are shorter than the Bhutesvara pillars and decorated with sculptures in comparatively low relief. Here at Sanghol, too, although no precisely datable materials related to the stupas themselves survive, 78 one cannot exclude the possibility that they belong to the earlier Kushan period of the first to second century AD by comparison with Kankali Tila. S. P. Gupta (1985) clearly manifests the first century date on the title page of the booklet edited by him. (5) Kankali Tila (Illus. 12. 4): Lastly comes the type of wheel-shaped structure at Kankali Tila reported by V. Smith (1901). The only information available on this stupa is the ground plan shown on his Plates II and III and the explanatory note attached to them. Eight radiating walls apparently project outwards beyond the limit of the ring wall. A close observation of the plan reveals that one of the walls radiates to the north to meet the small remaining part of another ring wall which is probably the outermost wall concentric with the inner. The structure of this stupa thus most probably consisted of two concentric circular walls, interior and exterior, intersected by eight walls radiating from the centre. The eight walls converge toward the centre where there is neither round nor square brickwork. This is particularly curious because among the stupas with wheel-shaped pattern at Nagarjunakonda such a design is always found, even in the stupas of the smallest diameter, 4. 5 m, such as Nos 27 and 108, while the diameter of the outer ring wall here at Kankali Tila is 18. 85 m according to my measurement based on the scale given by Smith on Plate III. Such a device of radiating walls converging to a center with no hub is also seen at Fil Khanah stupa (see No.3), but never met among the other examples in Andra Padesh. Also curious here at Kankali Tila is an almost equal thickness throughout all of the walls, including the outermost wall and radial walls. The nineteenth and early twentieth century excavators of the site, only looking for sculptures of better quality, took little care of the stupa and regrettably spoiled any direct evidence for its dating. Remaining materials for dating the stupa at Kankali Tila, therefore, are only several sculptures of mottled red sandstone found around the stupa: e. g., a standing Karttikeya dated to a year 11, a crouching Surya, and a pillar with standing tirthankara inscribed with a year 35, all of which are datable to the Kushan period in the second to third centuries AD according to Huntington (1985: 160-162). The close affinities between the sculptures from Kankali Tila and those found deposited near stupa No.1 at Sanghol claimed by S. P. Gupta (1985: 50) were already cited above. Builders in the north do not seem to have understood how the wheel-shaped structure should be used and why it was necessary to the stupa. If they had been well aware of the real purpose for which southern builders adopted the pattern, they would not have settled it in the dome plus drum (e.g., Dharmarajika and Fil Khanah) or have built ring walls of equal thickness (e.g., Kankali Tila) or within a square plinth (e.g., Sanghol). Really such constructions do not make any practical sense. Ostensibly the northern and southern examples look similar to each other, but the northern examples lack features structurally essential to the wheel-shaped pattern. Granted that there existed a geographical difference in the plan of the plinth, round in the south and square in the northwest, the examples at Sanghol are still quite strange in the point that the wheelshaped pattern settled inside of the square plinth makes the existence of the outer ring 79 wall useless. It clearly exposes the builder's ignorance about the primary meaning of the pattern. All these anomalies and divergence of types of that pattern among the northern examples suggest, therefore, that the northern stupa-builders must have transplanted that pattern from the south at a stage when it had already become quite popular with the southern builders. In no way would such northern patterns have originated immediately from Roman prototypes. The technical significance of the wheel-shaped structure in the Mausoleum of Augustus survives most vividly inside the stupas in peninsular India which have the thick outermost ring wall and the radial walls running from the central hub structure, all being the basic factors that enable the round plinth to be built with a high elevation. In this sense the northern examples should be taken as one step later than the southern examples. 3. Relation of Radial Walls to the Relic Chamber The plan eventually given by Hargreaves shows four of the radial walls, all being broken halfway (Illus. 13; Hargreaves 1914: PI. XIII). The neighbouring two running from the northwest and the west meet each other near the centre. The remaining length measures 7. 2 m long for the former and about 3 m for the latter. The angle formed by these two walls clearly shows that the wheel-shaped pattern here does not contain any hub structure as at Kankali Tila. Another radial wall running to the southeast starts from the midway point, not from the center, and runs only about 3 m. Trying to find the south face of the stupa to the west of the southern projection, Spooner (1908: 19-20) admitted the presence yet further to the north inside the mound of a roughly built wall running towards the south wall of the stupa itself at an acute angle to it. He identified this as one of the radial walls saying that it is not uncommon to find the central portion of a large stupa intersected by a series of radiating walls intended to strengthen the structure and distribute the pressure of the massive core (Spooner 1908: 19-20). Spooner and Hargreaves do not mention the breadth of these radial walls at all nor show it in their plan except for the short wall: it is about 120 cm if a measurement based on their plan and scale can be considered valid. At Shah-ji-ki Dheri these radial walls are closely related to the relic chamber as the latter was found built against one of the radial walls. Spooner describes the discovery of the relic chamber as follows (Spooner (1912: 48f): A large pit, 24 feet square, was outlined covering the exact centre of the monument, and then taken downwards. A few feet below the present surface of the mound, traces were found of the very massive radiating walls in the heart of the stupa [sic], and these greatly delayed the progress of the work, for we were anxious not to remove any portion of these walls unnecessarily. Avoiding these, therefore, as much as possible, the pit was taken down by slow degrees to a very low level without result. Indeed, after several days' digging we had got down to what seemed to be free earth, and had almost lost hope of finding any relics at all, when suddenly, and without warning, the remains of the relic chamber were reached at a point which proved to 80 be two feet below the level of the brick pavement surrounding the stupa as a whole. Making a 7. 2 m square sounding in the center of the stupa mound Spooner reached the remaining tops of "the very massive radiating walls," about 30 cm below the surface (Illus. 13). Then after several days work, the chamber was found standing on "what seems to be free earth" (probably natural soil) 60 cm below the brick pavement surrounding the stupa (Spooner 1912: 48-49). Here again Spooner does not mention the remaining height of the walls, but it is not impossible for us to calculate the height based on Hargreaves' description of the structure on the north face of the stupa. The stucco frieze on the west face of the northern projection which possibly starts from the brick pavement remains up to the modillion cornice, measuring 135 cm in height, and the natural soil on which the radiating walls stand is 60 cm below that pavement. The highest level of the modillion cornice seems to be more or less equal to the top of the central stupa mound as the photograph at PI. XIV (a) of the Hargreaves report published in 1914 shows. If so, the remaining height of the radiating walls presumably is at least 195 cm, or about 2 m, and actually more than that. This measurement holds good if the three contour lines showing the whole stupa mound in the plan attached to the Crompton report represent a height of 10ft, or 3 m (Crompton 1875). So far so good. Yet Spooner's description of the find spot of the chamber is not easy to read since the plan of the stupa does not include any wall running due east beside which the casket was discovered. Spooner continues (1912: 48-49): What the exact structure of this chamber had been originally, it was unfortunately quite impossible to say. It was not in the exact centre, but a little more to the east, and appeared to have been built against the end of that one of the radiating central walls which ran due east from the centre of the stiipa. The chamber itself must originally have been roofed in some way, but this roofing, whatever it was, had completely disappeared, and the original open space was packed with earth. The chamber itself was of rudest construction. A long, smooth slab of slate had been laid down extending in its length from north to south, and across the southern end of this was laid a heavy slab meeting at right angles with another heavy slab along the western edge. These two thus formed two sides of a possible square, with the corner intact at the south-west. But no trace whatever could be found of any corresponding slabs on the east and north, and from the general position of the whole it is my opinion that the chamber was not enclosed on these sides save by the massive rubble masonry of the radiating walls to the east and north-east. This is, of course, a surprising fact and one which it is difficult to explain. But the fact as such is unquestionable. The eastern wall was penetrated to a considerable distance, but there was no trace of any continuation of the chamber, not even so much as a definite pavement of any kind. The whole thing seems to have been almost primitive in its rough simplicity. Even the 81 definite "floor" of the chamber was not decorated or dressed anywhere except in the very corner where the relic casket stood. Here a little daub of chuna had been laid on, on which the casket had rested, and wherein its outline was found clearly impressed when the casket itself was removed, but the rest of the floor was the plain unadorned slab. (Italics were added by Kuwayama. ) Apart from one slab used for flooring, possibly with a longer side to the west and a shorter to the south, the chamber seems to have consisted of only two heavy wall slabs which met at a right angle at the southwestern corner forming an L-shape, without anything remaining of the corresponding walls on the north and east sides. Therefore Spooner considered that the massive masonry of the radiating walls running to the east and northeast had taken the place of the slabs for both the north and east sides. Although not shown in the plan, the radiating wall running to the east must have actually existed since the chamber "appeared to have been built against the end of that one of the radiating walls which ran due east from the centre of the stupa". The radiating wall running to the northeast also suddenly appears only in the above description without anything of that wall showing in the plan. This radial wall which runs northeastward also must have been discovered. We are not informed about how these two neighbouring walls stood. In any case, Spooner's mention of "the end of that one of the radiating walls" is instructive. The "end" suggests the fact that the portion of this radial wall closer to the center of the stupa was found destroyed because otherwise the "end" would not make any sense. Despite Spooner's supposition, the massive rubble masonry along the radiating walls running to the east and northeast cannot be the opposite sides of the L-shaped walls. If Spooner's description of the two slabs composing the two walls of the L-shaped chamber is correct, the corresponding radial walls should have been either to the west and the southwest or to the south and the southwest. But whichever the case might be, the locations of the uncovered radial walls would resist the claim that the chamber was found a little east of the center. More attention should be given to the "end" of that radiating central wall which ran due east and against which, as Spooner describes, the chamber was built. As said above, the "end" suggests that the radiating wall in question was found broken at a distance close to the center of the stupa. Considering Spooner's belief that the radiating walls to the northest and the east formed the north and east sides of the chamber, these sides should be enclosed by the broken ends, or the breadths of the radiating walls, not by the long faces of the radiating walls. According to this supposition, the wall radiating to the northeast must also have been found broken to its foundation, although nothing about how it was found is described by Spooner. Hargreaves' description given below (Hargreaves 1914: 24) amplifies the advantages of this argument: An attempt to fix the circumference of the drum of stupa dome by following 82 one of the radiating walls running from the spot where the relics were discovered yielded as little result, for the wall was broken at a distance of 24 ft from the centre of the mound. Which one of the radiating walls is that running from the spot where the relic casket was discovered? On the plan of Shah-ji-ki Dheri given by Hargreaves (1914: PI. XIII), the one that runs to the northwest is drawn much longer than that drawn in the same direction by Spooner (1912: PI. X). The difference of length suggests that Hargreaves followed the wall to the northwest. If so, the place where the radial wall starts to the northwest is approximately the spot where the relics were found. In light of the fact that Spooner thought that the radiating walls to the east and the northeast took the place of the slab walls, the location of the chamber supposedly was a spot a little to the east from the center, where the radial walls converge from the east, the northeast, the west and the northwest. Thus we have reached a conclusion as to how and where the chamber was located, although it remains quite hard for us to add to the Spooner plan the exact location. The discovery of the radiating walls as broken to the foundation at the central part of the wheel-shaped pattern particularly conveys an important fact as to when the relic chamber was built with the "Kanishka Casket" inside. The plans given by the excavators show that the radiating walls running to the three directions-the northwest, the west and the southeast-had already been broken when excavated. The plans and the descriptions as well show that all of the radiating walls were found broken, some of them lacking the central part and others being damaged after running for some distance from the center. Since the wheel-shaped pattern on natural soil is a component of the lowest foundation, Shah-ji-ki Dheri Main stupa must have suffered complete destruction which reached even the lowest of all the structures. The fact that the relic chamber was built against the broken ends of the radial walls laid on the natural soil shows that the relic chamber was made after the disaster, not together with the construction of the wheel-shaped structure. The "Kanishka Casket" found standing on the plastered floor inside the southwest corner of the chamber therefore is not attributable to the initial phase of Shah-ji-ki Dheri Main stupa but to a later phase after the disaster. Among the two dozen examples of stupas containing the wheel-shaped pattern known so far, Shah-ji-ki Dheri is in the most destroyed condition. No example has ever been found destroyed down to its lowest portion, except for the stupas at SanghoI. Even the stupas without that pattern have hardly ever been found destroyed to such an extent. If such extreme destruction had been caused by any natural agency, the stupa would have been exposed without any care quite a long time. A more likely alternative for the cause of such severe damage would certainly be vandalism. 4. The First Main Stupa As a Circular Plan The Shah-ji-ki Dheri Main stupa ostensibly belongs to the same category as the 83 Sanghol stupas for the reason that the radial walls are inside a square and also comparable with Kankali Tila (Illus. 12, 4) for the reason that they are settled in the foundation without a hub. However, a further examination does not provide any evidence for rendering the Shah-ji-ki Dheri wheel-shaped structure within the Sanghol category. At Sanghol stupa No.1 the wheel-shaped pattern is not exposed but framed by square walls (Illus. 12, 3). This evidently shows that both round and square structures were made at the same time. The outward appearance of the stupa might have taken the form of a square plinth supporting the circular superstructure, including the drum, above a platform. All of the upper circular elements above the platforms, however, were not built directly on the square plinth but more deeply rooted on the ground. In other words, the cylindrical construction with the wheel-shaped pattern inside stands on the ground, the lower part being framed on all four sides by the retaining walls which support the square platform and the upper part being exposed above the platform. Nowhere at Shah-ji-ki Dheri can this device be found. The Shah-ji-ki Dheri Main stupa is chronologically interpreted by Errington so as to relate it to the cross-formed stupa at the site of Tahkal Bala (Errington 1987: Fig. 4). According to the plan of Shah-ji-ki Dheri redrawn by her the cross-formed stupa and the radial walls belong to Phase 1, or the first of three building phases, and the circular towers at the comers to Phase 3. However, from our structural and chronological viewpoints the cross-formed stupa and the radial walls are clearly independent from each other for three reasons. The first important point is the clear difference between the level of the natural soil on which the radial walls stand and the ground level on which the cross-formed stupa stands (Illus. 13). According to Spooner, the radial walls first appeared about 30 cm, or a few feet, below the surface in the center of the mound and went down to the natural soil with the aforesaid remaining height, more than 200 cm. If the surface of the stupa mound was more or less flat despite the existence of Crompton's trench as shown in his plan (Crompton 1875) and the presumed height is allowed to be approximately correct, the level of the top of the natural soil would be at least about 230 cm below the surface of the central part of the mound. Natural soil was also discovered elsewhere. A trench to the east of the south projection revealed three superimposed layers. Spooner (1912: 40-41) describes them as follows (Illus. 13): For the first foot or so (30 cm) free earth of the wheat field was passed; then came a stratum some two feet (60 cm) thick of tightly packed debris among which one copper coin-too badly corroded to be recognized-and a few small and very badly damaged sculptural fragments were found. But below this the pit passed again into perfectly free earth to the depth .... Spooner seems to use the term free earth as a layer containing nothing artificial. The second level of free earth mentioned here therefore might be taken as natural soil. If so, in this sounding the natural soil was reached only 90 cm below the surface. Natural soil was also found to the west of the south projection below 60 cm of debris on which the floor was paved with baked bricks. The base of the cross-formed stupa was found faced with three successive coatings. The final coating includes the stucco frieze, and it corresponds to the period of brick pavement around the cross-formed stupa. As mentioned in the context of the relic chamber, the brick pavement is 60 cm higher than the natural soil on which the chamber is built. This 60 cm layer could be the same in character as the debris found between the surface and the natural soil to the west of the south projection explained just above. Thus a conclusion is reached that a strict difference exists between the levels on which the radial walls were standing on the one hand and on which the cross-formed stupa was built on the other. If the radial walls were a structural element settled inside the cross-formed stupa, the process would have to have been such a forced one as to have had in the first stage of construction a circular body with the wheel-shaped pattern built on the natural soil. This was followed by a second stage in which the ground level would have been raised 60 cm above the natural soil with earth around the circular body. In a last stage the cross-formed stupa would have been built on the raised ground on top of which the brick-paved floor was laid. In this hypothetical case the 60 cm-deep layer would have to be quite free from artificial objects since no time lapse would have existed for anything artificial to be left on the first construction of the stupa at the site. Yet, on the contrary, the 60 cm-deep layer was discovered including debris; it was not without artificial objects. Even if it had not contained anything artificial, raising the ground level around the wheel-shaped structure and building the cross-formed stupa on it would have been quite a strange way of constructing a stupa. For this reason, the existence of the 60 em-deep layer of debris implies that the radial walls and the cross-formed stupa chronologically differ from each other: the circular stupa with the radial walls inside predates the cross-formed. Secondly, the center of the wheel-shape does not meet at the point of intersection of two diagonal lines connecting the corners of the cross-formed stupa-or the square body with four projections, nor does it meet the centre of a circle touching the insides of the square (Illus. 13). If Shah-ji-ki Dheri stupa were a construction like Sanghol stupa No.1 which has the wheel-shape built in one body with the square, the above two different elements would share one and the same central point. In fact, tlle stupa at Sanghol not only has only one central point for both the wheel-shape and the square, but also has radial walls which correspond to diagonal lines between the corners of the square. However, at Shah-ji-ki Dheri the directions of the radial walls running from the center to the northwest and to the southeast somewhat differ from that of a diagonal line running from the northwest angle to the southeast. Furthermore, the intersecting point of the two facing radial walls running to the northwest and to the west is fairly close to the point where the two diagonal lines meet. This proximity of the two points shows that the real center of the wheel-shape is elsewhere, a little to the east of the center of the square. The third important point is an interpretation of the building sequence at the northwest corner of the square. As Hargreaves' plan shows, a number of constructions 85 with smaller sizes, round and square, are laid out parallel to the main stupa. Among them, however, a group of four square and one round structure, all probably being remains of small stupas, near the northwest corner significantly differs in direction from the others. According to Hargreaves (1914: 26), the northeastern section (actually the eastern corner) of the square base of a stupa in particular was found partly destroyed so as to allow for the building of the northwestern tower. As Hargreaves also describes that the round tower is faced with masonry similar to the square body of the stupa, the tower was presumably built as part of the same building program. This simultaneity gives cause to think that the group of small structures near the northwestern corner predate the square body with the projecting tower at each corner. The disposition of this group characteristically appears to conform to something circular, not square. Therefore, it suggests the possible existence of a circular stupa near which tllis group of small stupas was settled. As to this circular stupa, nothing other than the construction with the radial walls inside can be supposed. From the above interpretations it is apparent that the circular stupa containing the wheel-shaped pattern had existed as the main stupa at Shah-ji-ki Dheri before either a square or a cross-formed stupa was built. This first stupa had the following features (Illus. 13): (1) A huge circular plinth: Internally strengthened by the radiating walls, it must have had a diameter shorter than 72 m and longer than 56 m, the former being of a circle touching the inner sides of the bases of the three stupas at the northwest corner and the latter of a circle touching the outer broken end of the radiating wall running to the southwest, the outermost end of the extant radial walls. (2) Eight radial walls: Massively masoned with rubble, six of them were excavated, four being shown on the plan and two only described in the excavation report. These remaining spokes show there must originally have been eight in all. The neighboring two radiating walls shown on the plan of Shah-ji-ki Dheri stupa head to the northwest and the west with an angle between them of 42 degrees which is very close to 45 degrees that make eight spokes at regular intervals. Despite no description of the breadth of the radial walls in the excavation reports, it most probably is about 120 cm, if we are allowed to take the wall running southwestward in the plan given by Hargreaves as proper for measuring. (3) Absence of hub and ring walls: The intersecting point formed by the same neighboring radial walls shown in the plan does not suggest the existence of a hub structure at all. Although no ring wall was found here, the pattern seems quite similar to that at Kankali Tila, not to the Sanghol examples. In addition, at Kankali Tila no thicker outermost wall exists and the breadths of the walls are common to all regardless of whether they are ring or radial walls. Such a circular plinth cannot be given structural stability and thus lacks the original technical significance of both the stupas in Andhra Pradesh and the Roman ancestor. The existence of a high plinth presumably was 86 impossible for the stupa at Kankali Tila and that at Shah-ji-ki Dheri, neither of which had a hub in the center. (4) The "Kanishka Casket" is not the one for this first circular stupa. A supposition that vandalism rooted the stupa down to its foundation laid on the natural soil is due to the fact that all of the massive radial walls had disappeared except four of them which had also been broken in many places. The "Kanishka Casket" would not have survived such a calamity, if it had stood as it was found. This is another reason for ascribing the casket to the following period, not to the time of the first circular stupa. At the Shah-ji-ki Dheri the only subsidiary stupas of the first period to survive subsequent destruction remained near the northwestern angle of the main stupa. 5. Inauguration Before Kanishka Use of the wheel-shaped pattern for the dome at the Dharmarajika Main stupa and for a presumably low plinth at Shah-ji-ki Dheri apparently is of no originally technical significance. But the Shah-ji-ki Dheri and Kankali Tila builders seem to have been aware that that pattern ought to be placed in a lower portion of the stupa. In this sense Shah-ji-ki Dheri and Kankali Tila stupas are less far from the Andhra Pradesh examples than Dharmarajika Main stupa, even though their circular plinths were rather lower Dheri than the Andhra examples. If a time gap bears this difference between and Dharmarajika, the circular Shah-ji-ki Dheri stupa may predate the rebuildings of the dome and drum at Dharmarajika. The initial construction of Shah-ji-ki Dheri stupa may thus be dated within this framework. As Marshall (1912-13: 10) writes, the dome and drum of Dharmarajika is "of rough rubble masonry strengthened by the walls, 3' 2" (95 cm) to 4' 9" (143 em) in thickness, radiating from the center. ... These construction walls stop short above the berm of the stupa, instead of being carried down to its foundations, and it is obvious that they belong to the subsequent reconstruction of the fabric which took place probably during the Kushan epoch". Marshall (1951: 236) further describes as follows: As to when the original fabric fell to ruin and was rebuilt the stupa itself does afford some evidence: for, although the masonry of the reconstructed drum is limestone rubble, there is a certain measure of regularity in the way it is laid that recalls the masonry of many of the Saka-Parthian buildings in Sirkap .... Moreover, the principle of the radiating construction walls is paralleled to some extent in the foundation walls of the little stupa in Block 1C' in Sirkap, which was erected after the earthquake in the early part of the first century AD. But another and stronger reason for ascribing the rebuilding to the first century AD is to be found in a small patch of diaper facing masonry on the west side of the stupa drum immediately above the berm, which apparently belongs to the reconstruction period and is identical 87 with other diaper work of the first century. (Block 1C' should be read as Block IE'.) As to when diaper masonry was introduced to Taxila, Allchin (1968) gives quite a convincing chronological picture very carefully and correctly interpreting coin finds in relation with the strata in Sirkap. According to Allchin (1968: 19-23), Sirkap at the level of Stratum II was constructed during the last years of Azes II, perhaps during a period of overlap with Kujula Kadphises, suggesting that Gondophares was a somewhat senior contemporary of the former and was at best remotely concerned with the government of Taxila. With this conclusion he implies that the new diaper masonry of Sirkap coincides with the arrival of the Kushans under Kujula Kadphises rather than Gondophares as Marshall thought (Allchin 1968: 13). If this hypothesis can be extended to Dharmarajika, the dome and the drum of the main stupa, internally strengthened originally by sixteen walls radiating from the central solid circle of rough rubble limestone masonry and externally faced with diaper masonry, is datable to the time of Kujula Kadphises. On the other hand, what is visible now at the site and on several photographs and drawings provided by Marshall (1951: PIs. 46-48) of the outer facing internally strengthened by radiating walls is semi-ashlar masonry. When the semi-ashlar masory was introduced in Taxila has not been clear in the stratigraphical sequence within the cities. Only clear is that the semi-ashlar masonry is absent from Sirkap and appears in the houses inside of the Sirsukh fortification. This does not afford any secure date to the introduction of the semi-ashlar. Allchin therefore draws attention to a series of stupas, including K1, K2, and K3, to the north of Dharmarajika Main stupa, all being faced with the semi-ashlar masonry. stupa K3 overlies a rubble masonry tank filled in with debris. In the filling was discovered a deposit containing three coins of Kanishka and one of Soter Megas, and in stupa K3 itself was discovered a deposit including three coins of Kanishka. In the relic chamber of P6, another stupa faced with semi-ashlar masonry located a little south of K3 and between two other semi-ashlar stupas, P3 and P4, Allchin (1968: 22) brings up the fact that three coins of Huvishka and seven of Vasudeva were discovered (Marshall 1951: 263). This coin sequence roughly shows that the semi-ashlar might have first appeared no earlier than Kanishka and continued to be used much longer than the diaper. The wheel-shaped pattern of Dharmarajika stupa is datable to the time of Kanishka at the earliest based on the semi-ashlar masonry. However, as to the rough limestone rubble masonry of the reconstructed drum including the radial walls of Dharmarajika Main stupa, Marshall sees that "there is a certain measure of regularity in the way it is laid that recalls the masonry of many of the Saka-Parthian buildings in Sirkap, rather than the earlier masonry of the Bhir Mound or of the lower settlements in Sirkap". Furthermore, the rubble masonry does not seem to be adopted at the same time with the semi-ashlar masonry elsewhere in Taxila monastic buildings. The semi-ashlar masonry now visible on the outer facing of the dome presumably is not original but of a later restoration. Comparing the northern examples of such structures with those in the southeast of 88 the Subcontinent, which are closer to the Roman ancestor, the wheel-shaped structure laid in the' foundation here at Shah-ji-ki Dheri might be one step earlier than that at Dharmarajika set in the dome and drum with a facing of diaper masonry supported by a low circular plinth apparently ascribable to a previous construction work. This previous work is broadly contemporaneous with the building of the earliest small stupas around it and several stupas on the main street in Sirkap, all being masoned with rubble and faced with kanjur ashlar and some producing as reliquary deposits coins of Azes I and Maues together or Azes only. No coin is found later than these monarchs. Therefore the diaper facing and the rubble radial walls are most probably placed chronologically sometime between these monarchs and Kujula Kadphises. The dating of the Dharmarajika wheel-shape thus gives a chronological framework to the inauguration of the main stupa at Shah-ji-ki Dheri: it is no later than the time of Kujula Kadphises. Last but not least about the legend as to how Kanishka built the stupa near Purushapura. A small boy was making a stupa on the road when Kanishka asked him about his deed. According to Xuanzang, the king, being awakened to right belief by what the boy told him and deeply respecting the Buddha's teaching, tried to make a stone stupa covering the small one which the boy was making. However high King Kanishka might try to build over the small stupa, the original stupa itself increased in height and size and always rose above the king's stupa by three Chinese feet. Kanishka stupa, therefore, became higher and higher and eventually the king could completely cover the original only when it measured more than four hundred Chinese feet in height. Scarcely had he finished his work when he saw the little stupa project about half-way up at the lower southeast corner of the great foundation. The king was quite unsettled at seeing this and immediately gave up doing further construction work. Consequently the small stupa remained projecting from the inside of the stone foundation below the second storey of stupa and, moreover, a smaller stupa emerged at the original location. Essentially the story should be taken as concealing the fact that Kanishka built a gigantic stupa covering the one that had previously remained there. Archaeologically, Shah-ji-ki Dheri Main stupa was inaugurated in Period I as quite a large circular stupa which was later destroyed down to the foundation. In the following period a square stupa was newly raised covering the damaged earlier one. This assumption may lead to a hypothesis that the legend might be based on a historical fact that Kanishka constructed a square stupa over a previous, circular one which had been standing, even though damaged, occupying quite a huge space and therefore hard to cover. In this connection the remains of the Period I small stupas in the northwestern corner of the main stupa are very significant. 89 VI Aspects of Stupa Court at Hadda Tapa Shotor and Lalma The extensive excavations by Jules Barthoux at Hadda, Jalalabad, follows the preliminary work of Foucher and various amateur digs in Jalalabad of the nineteenth century. While Barthoux concealed the fact that schist sculptures figured among the finds, strong impression was given to only stucco sculptures and let Marshall conceive the idea of the Indo-Afghan school. Gradually scholars have come to realize just how far his published excavation reports fall short of modern standards. Lacking any description whatsoever of excavation techniques, they cause anyone intending to re-examine the sanctuaries afresh to encounter many difficulties including desolate ruins, which have not been protected in any way since Barthoux left the area. The prolonged excavations and restorations at Tapa Shotor, Hadda, therefore, not only help to make up for deficiencies resulting from past digs at other sites in the area but also, because well-executed specimens of clay images are still intact in every niche and chapel, to enlarge our knowledge concerning sculpture at Hadda. It is the clay figures at Tapa Shotor that shed light on Buddhist plastic art. The sculptures found here should be compared with those found at the Basawal cave monasteries, Tepe Maranjan, Gul Dara in Logar and Tapa Sardar in the south of the Hindu Kush, and Fonduqistan in the mountains, or even with those at Surkh Kotal and known from the sites beyond the Oxus. In Jalalabad too, we have evidence for the import of sculptures from Mathura. Two fragments of Mathura sandstone depicting drapery were collected by me both in 1964 and 1972 at Shah Nasr Ghundai of Chahar Bagh, where the capital of ancient Nagarahara may have been located, some 10 kilometers to the west of modern Jalalabad bazaar. The importance of Jalalabad therefore is that Buddhists there decorated their sacred areas with images made of various kinds of materials such as schist, stucco, clay, limestone, and even sandstone from Mathura. No Buddhist sanctuaries elsewhere provide us with examples of Buddhist art in these different media. It should be particularly noted, moreover, that Tapa Shotor was the first site excavated by a team of all Mghan archaeologists. Shahibai Mustamandi began excavating the site during the period 1965-1973, and Zemaryalai Tarzi conducted the last six campaigns in 1974-1978 with exemplary thoroughness. Thus this sanctuary, so unusual in many ways, had been put by 1976 into the context of modern Afghan archaeology, or more generally, of Gandharan archaeology. Detailed site reports, apart from a few summaries published by both the then Directors General of Archaeology, Afghan Institute of Archaeology, Kabul (Mustamandi 1969a: 15f., 1969b: 119f., 1971: 43f.; Tarzi 1976: 381f.), might well have been expected. Yet all those knowledgeable about Afghan archaeology realize that, since 1980, the site has been destroyed. 90 In 1967, when the stupas in the courtyard had almost been entirely uncovered the present writer, then completing the third and last campaign of the excavations at Chaqalaq Tepe near Kunduz, was given generous permission by Mustamindi to examine the site freely with the help of Mr. Ehsan Aram, a member of the above-mentioned Institute. Observations of the stupa court at this extraordinary site suggested comparisons with that of Lalma. The Lalma site was excavated in 1965 by the Kyoto University Archaeological Mission to Mghanistan under the direction of the late Prof. Seiichi Mizuno of the Kyoto University Institute for Research in Humanities, to which I also belong (Mizuno 1968). With the permission of Mustamindi who kindly supplied with necessary photographs, my views and observations about Tapa Shotor were published in Japanese in No. 45 of the Toho Gakuho Gournal of Oriental Studies), Kyoto, 1973, under the title 'Hadda Saikin-no Hakkutsu-ni kansuru Mondai' (Problems Concerning the Recent Excavations at Hadda), with intention of introducing this Buddhist temple of vital importance to the Japanese scholarly world. This short note, published by an archaeologist who affectionately thinks of Afghanistan as his second home, both archaeologically and personally, is an English adaptation of the original article with slight modifications mainly resulting from the lapse of time. This version, moreover, would not come about without the encouragement of Prof. Maurizio Taddei who was deeply concerned, too, about current situation in Afghan archaeology where any information collected in the past is now vitally important and virtually irreplaceable. Hence to him many thanks are due. I should also like to express my gratitude to Dr. Zemaryalai Tarzi for his earlier suggestion to do this translation and, in particular, for his friendliness given to me during my work in Afghanistan up to 1978 and beyond. 1. Stupa Court The main stupa at Tapa Shotor, square in plan and facing southeast, is surrounded on all four sides by rows of small subsidiary stupas that are separated from the main stupa by a procession path. Around the rows of small stupas there is another procession path which is also surrounded by rows of chapels limiting the area of the stupa. All of the small stupas had been uncovered by 1967, but only seven of the chapels were visible. The whole courtyard is only accessible from outside through a narrow and short passage at the east corner (lIlus. 14, below; Mustamindi 1969a: fig. 2). The cores of all stupas including the main stupa are constructed usually of the squared conglomerates, but sometimes of boulders from riverbeds laid in mud. The facings of the plinths and drums are composed of schist slabs, carved sandstone, or squared limestone. In most cases either two or three kinds of stone are associated in one stupa: for example, in stupa No.3 the base mouldings and pilasters are formed of schist slabs, and the niches between pilasters are carved out of a block of sandstone. The faces of stupas are coated, as usual, with stucco, while the chapels and the images enshrined in them are made of mud bricks and clay respectively. So, the stupas and the chapels 91 contrast strongly with each other. In Taxila and the Peshawar plain as well as Swat valleys, Buddhist temples are without exception masoned. Even at Basawal the chapels built of schist slabs seem to have housed both stucco and clay images (Mizuno 1971: 109-111, PIs. 17-22). In the district to the west of Jalalabad and south of the Hindu Kush the main stupas are always built of stone, even if the temple has chapels and monastic buildings constructed of mud bricks. The example of Top-e Rustam mud brick stupa at Balkh makes it clear that stone is not the only suitable material for building stupas. Whether stupas were constructed of stone or mud bricks may have depended on local doctrinal tradition or regional variation in the accessibility of materials. Along the east wall, which limits the stupa court and measures 14.8 m. long, were excavated three low bench-like projections of similar dimension, square or trapezoidal in plan, laid on the floor at regular intervals of about 3 m. Similar but larger projections, three in number, are also set along the southwest wall. These might be thought to be pedestals for images, but they are flush with the wall, and no trace of anything on top of them remains. Similar pedestals are found at Tapa Kalan, where Barthoux reported finding only the feet of the images (Barthoux 1933: 115-116. See the plan of Tapa Kalan.). The use of these objects at Tapa Shotor is still open to further inquiry. 2. Chapels First of all, chapel D deserves particular mention because of various important features. Its width measures 1.8 m. at the front and 1.9 m. at the back with a depth of 1.6 m. On the floor, which is a little higher than the level of the procession path, is a 40 cm. wide low bench set along the side and back walls. Each wall is divided into upper and lower parts by a horizontal band projected from the wall. On the band are standing the square pilasters carved out of mud, the presumably Corinthian capitals having already been demolished. The extant shafts of pilasters, which are supported by a socle consisting of coarse layers of mouldings, are simply decorated with a usual pair of vertical incised lines terminating at both ends with a concave incision. For the lower portion of the wall still indispensable are the pilasters that support the pointed arches reaching the band. A standing Buddha image was enshrined under each arched space. Under one of the arches his halo and bare feet only are extant and under another his upper body is missing. The ornamental flutings on the shaft look as if they were cast, the topmost of them turning over in form of flower petals. The base is trapezoidal in elevation form, horizontally incised with the two rush lines that recall torus and scotia mouldings. The appearance of the walls of this chapel thus leads us to suppose that the composition is more similar to that of some of the caves in Bamiyan rather than to the outer decoration of still standing huge stupas at Top Dara, Shewaki, Shankardar, and so on. Also in this chapel, the triangular space formed by the band and two neighbouring arches is filled with an eagle-like bird with his wings half extended and with feathers 92 depicted in great detail. Birds in stone reliefs are usually carved on both sides of a gable (Ingholt 1957: No. 168), and there are few examples where they fill such a space as found on the Bimaran casket (Wilson 1841: PI. IV). The specimens close to Tapa Shotor are found in Gul Dara monastery in Logar (Fussman and Le Berre 1976: 19-20, PIs. XXXVIIIXL, designate the birds as garuda), and, even closer, on the front of the square plinth of fire altar in the shrine B at Surkh Kotal (Schlumberger et al. 1983: PI. 69, Nos. 235 and 236). The Surkh Kotal birds, also in clay, are cut off from the breast up and they almost reach the top of the pilasters between which the birds stand. Their current condition as published by the above authors suggests that the pilasters had originally supported two arches, i.e. that the arches and the upper halves of the birds have since been cut off. The chapels tentatively called J and K and located at the southwest corner of the courtyard also deserve mention. Both measure 2.2 m. deep by 2.0 m. wide and are covered with the barrel-vaulted roofs, which are mostly decayed but traceable along the top of the remaining side walls. As far as we know, the real barrel-vault does not seem to have been known to the builders who designed Buddhist monuments in the Peshawar plain and farther east. In every site where the ceiling is structured as a semi-circle in cross-section, schist slabs are vertically laid one over another to form a tapering top (Foucher 1095: 109 f; Mizuno 1978: PIs. 56-57). In addition, only the end that is destined to face outward when the slabs are piled together is sharply chiselled to make a shape of a single-edged blade, and so achieves an effect of gentle curving. In short, architects were familiar with corbelling technique, but not that of the real barrel-vault. This is also the case with the construction of domes in the Peshawar plain and beyond as depicted on the ceilings of tlle chapels and the cells as well as the passages in Takht-e Bahi, Thareli, Ranighat, and other sites in Swat. No squinch-arches are found in these districts. The excavations at Tapa Shotor done subsequent to my first visit have shown that some of the corridors in the monastic quarter are covered with barrel-vaulted roofs, and this type of ceiling is known from another site in Jalalabad (Mizuno 1970: PI. 32, Nos. 2 and 3). In the districts west of Jaialabad the dome supported with squinch-arches and the vault are usual device for ceilings. Hence Jalalabad can be thought of as the easternmost boundary of this architecture. 3. The Buddha Dissuading the Naga King (Chapel H) This chapel deserves special mention because all the images are carved in round as those in other niches on the walls of monastic halls are and because their composition is not a random collection of different figures but represents one of the legends from the Buddha's life. The chapel located near the westernmost corner of the courtyard is a small oblong in plan, measuring 2.9 m. deep by 2.4 m. wide, and extant walls reach a height of about 2 m. Regrettably a fire caused the beams of ceiling to fall to the floor and the damage to the clay images extended even to their cores made of timber. 93 The restored images and the cavities that once contained images suggest a scene from a story to which a fountain, or a pond, or even a lake is closely related. The sculptors succeeded in effectively utilizing both walls and floor to express a water scene; lotus leaves are trembling in whirling flow of water in which a kind of fish, among which a two-headed one is particularly noticeable, swimming from the back wall to the floor. Against such background about fourteen figures were performing a legendary scene. The seated Buddha had completely decayed. On the back wall only he leaves a trefoil-shaped aureole of his, within which some cavities having supported the image by means of tenons remain and around which are carved in mud heaps of flames that look identical to the whirls and flows of water depicted below. In front of the Buddha, two characters seem to have originally been laid face-to-face in the center of the floor. But only one figure to the left of the viewer remains, and he is easily identified as a naga, as a serpent is creeping up on his back from the floor. Another crucial character in this scene is a figure whose upper half is decayed. He is standing facing to and along the left wall with his legs straddled, and with his short tunic tucked up at his waist, and with his left· arm vigorously stretched out. His gesture, his clothes, and his location suggest that he may be identified with Vajrapani. Whoever the other figure opposite to the naga may be (possibly a nagini), the subject shown by at least two restored images, naga and Vajrapani, and by the flow of water is definitely one of the legends where a naga (or a naga king) converted himself to the Buddha's faith. Among various legends in which the naga figures as a chief actor, the most common in Gandhara is the naga Kalika who worshipped the Bodhisattva Siddhartha realizing that the Enlightenment was coming soon. The scene in this chapel, however, does not give any specific evidence suggesting this interpretation. In the North West one of the most popular and localized stories was the legend of malicious nagas who were in turn to be persuaded by the Tathagata to cease their evil ways. For example, two legends from the Da Tang Xiyu ji (the Great Tang Records of Western Regions by Xuanzang) can be singled out as closely related to this in geographical context. The following is one of the stories for which Nagarahara was famous. About 20 Ii to the southwest of the city Dipamkara we come to a small ridge, where there is a sangharama .... To the southwest of this sangharama a deep torrent rushes from a high point on the hill and scatters its waters in cascading falls. The mountain faces stand like walls. On the eastern side of one is a great cavern, where the naga Gopala resides .... In old times, when the Tathagata was in the world, this naga was a shepherd who used to provide the king with milk and cream. On one occasion he unexpectedly failed to serve him politely. Already blamed for this fault of his, he bore anger in his heart, purchased an offering of flowers with gold coins, and proceeded to the predicative stupa with the prayer that he might become a destructive naga so as to afflict the country and injure the king. Then going to a cliff, he threw himself down and died. Forthwith he became a great naga king, occupied this cavern, and wanted to go out from there in 9+ order to accomplish his original wicked purpose. When this intention had risen within the mind of the naga, the Tathagata just observed the exact object of the naga and was moved with pity that the people in this country were about to be destroyed. He came from middle India via his spiritual power to where the naga was. When the naga saw the Tathagata, his murderous purpose was stayed, and he accepted the precept not to kill and vowed to defend the true law .... (Original translation). In the section regarding U ddiyana in the same book is a story of the conversion of the naga Apalala. To the northeast of the city Mengjieli (Mingora) we enter a great mountain and arrive at the fountain of naga Apalala .... This naga, in the time of Kasyapa Buddha, was born as a man and was called Kingki (Gangi). He was deeply learned in the ways of magic and used magic to restrain the power of the wicked nagas so that they could not afflict the country with violent storm of rain. Thanks to him, the people were thus able to get harvest and, both having a sense of gratitude and being influenced by his virtue, made an of grain collected one dou per house as a yearly tribute. After a lapse of some years there were some who failed to pay their tribute. Gangi was so angry with them that he prayed that he might become a poisonous naga and injure them with storms of rain and wind which would destroy their crops. After the end of his life he became the naga of this country. As the result a fountain emitted a white stream which damaged the fertility of the soil. SakyamuniTathagata, his great pity guiding the world, was moved with compassion for the people of this country who were singularly afflicted with this calamity and intended to descend to this country in order to convert the violent naga. When Vajrapani beat with his weapon against the mountain cliffs, the naga king, trembling with terror, came out of the fountain and converted himself to the Buddha's faith. Listening to the Buddha who preached the dharma, his heart became pure and his faith was awakened. The Tathagata forthwith forbade him to injure the crops .... (Original translation) These stories in the Da Tang Xiyu ji seem to be the closest to our theme. Mustamindi opined that the conversion of the naga Gopala is the theme in this chapel, admitting that the scene at Tapa Shotor is not a faithful version of it. His identification solely depends on the facts that the cavern actually existed in Jalalabad at the time ofXuanzang, and that Tapa Shotor is nearby. In the Gopala story told in this Chinese document, Vajrapani does not appear. Another version of the story is recorded in the named the Guanbo sanmei hai jingo This, usually believed to have been translated from the original Indian text, is supposed to have been edited in China in early fifth century by Buddhabhadra who was born at 95 Nuoheli (possibly Nagarahara) in North India, according to the Biographies of Eminent Monks edited by Huijiao, and learned Buddhism in Gandhara and its environs. So, it is highly possible that Buddhabhadra was so well informed of the Gopala legend as to describe it very long and in great detail. Vajrapani in this vigorously sways his weapon in order to awake the wicked naga and his family as well as the raksasas to convert themselves to the Buddha's faith. In the Gopala legends of both sources, however, Gopala is linked with a natural background of steep cliffs and falls as well as caverns. The scene of this chapel at Tapa Shotor does not show such circumstances but rather a lotus pond or a fountain. The tells that the Tathagata entered into the wall of the cavern and was seated in meditation at the request of the naga who would otherwise become wicked even after his conversion. The seated image on the back wall at this chapel might reflect such a legend, but we cannot be sure for this interpretation. Mustamindi rejected the Apalala legend as a possibility for the scene in question since it had been told in the Swat valley, not in Jalalabad. We have to keep in mind that even though not a single specimen of sculpture showing the Dipankara Jataka has ever been found in J alalabad, the action of this J ataka took place there. It is necessary, too, to remember the fact that the reliefs or stelae depicting this Jataka have mainly been produced (or at least found) in the neighbouring districts like Kapisi and Gandhara. With this in mind the Apalala legend should not be excluded as a possibility for the scene in the chapel H. In this connection we refer to a paragraph on Apalala's conversion translated from the Vinaya of the Mulasarvastivadins by J. Przyluski (1941: 511), as cited by Giuseppe Tucci (1958: 227). The Vinaya says that since Apalala did not show any willingness to be subdued by the Buddha, Vajrapani smashed with his vajra the mountain overlooking the lake where stood the palace of the naga; the Tathagata sat in meditation, then entered the meditation of fire so that from every side there was nothing but a heap of flames. This version also seems to be closer to our Buddha and scene, especially in the illustration of the Buddha sitting among flames. 4. Classification of Stupas Small stupas, thirty-one in number, are supported by square plinths classifiable into two types determined by process of composition: a single lofty plinth and a two-layered one with a high plinth on a low base. Mustamindi divided the stupas into two chronological groups, depending on their location, structural chracteristics, and stylistic details. According to him, those in the earlier group (Nos. 1,2,3,5,9-11,13,15,16,18,21,22,25-27,31 and 32. See the Illustration 14, below) are laid more closely to the main stupa. The cores of these stupas are of stones joined together with mud, and those of pilasters are of schist slabs covered with stucco. The top of the central .i.acanthus; leaf of the capital is turned downwards with a central vein that has many horizontal nodes notched in it. Of the later group (Nos. 4, 7, 7 bis 8,12,14,17,19,23,24,28,29 and 30) the pilasters and base mouldings are chiselled out of sandstone blocks and then finished off with a coat of stucco. The schist slabs are not included in this type of stupas. The leaves of the capital 96 are arranged in double layers, the upper reaching the abacus without turning downwards, and the shaft is usually incised with two vertical parallel lines closed at each end with a semi-circular line. Mustamindi maintained that the acanthus leaves with a notched vein and the incised lines of the shaft could provide clues for dating each group: the former originated from those found at Ai Khanum (Bernard 1968: PI. XIII) and the latter were stylistically similar to those excavated at Surkh Kotal (Schlumberger et ai. 1983: PI. 7). According to him, the techniques characteristic of different sites in different periods must have been transmitted to Tapa Shotor in its earlier and later periods respectively. This is a forced explanation of the chronology of small stupas. So the following is a list of the details of small stupas as I observed them. (I) The group identified as earlier by Mustamindi: No.1: The drum is faced with small squared limestone, but all details including brackets, cornices, pilasters and mouldings are of schist slabs. The facing of the plinth cannot be visible, as it is covered with stucco. No.2: Same as No. 1. No.3: The lower plinth is completely covered with a coat of stucco, so the facing and details cannot be detected. The pilasters and mouldings of the second plinth are made of schist slabs, but the niche between them is carved out of sandstone. No.9: As for the lower plinth, the sides are faced with slightly finer rubble of limestone and finished off with stucco, while the pilasters, the trefoil-shapcd arches between them, and the mouldings as well as cornices are built of schist slabs, then finished off with stucco. The shaft of the pilaster is decorated with vertical incised parallel lines with an incised semi-circle at both ends. No. 10: The walls of the lower plinth are neatly faced with small blocks of limestone. The techniques used in the other details are the same as in No.9, but witnout decoration on the pilasters. No. 11: All details are schist slabs, while the facings cannot be observed because the seated Buddhas and Bodhisattvas are still intact. No. 13: Same as No. 11. No. 16: Same as No. 10. No. 18: The sides and all details are built of schist slabs, while stupa No.5 to the opposite of No. 18 is faced with limestone blocks, and the details are of schist slabs. Nos. 26 and 32: The pilasters are carved out of sandstone blocks, and the mouldings and cornices are of schist slabs. (2) The group identified as later by Mustamindi: No.4: The four sides of the lower plinth are faced with dressecl blocks of limestone, roughly squared, the interstices between them being filled with neat piles of small pieces of limestone. The cornice and the base mouldings are composed of the sandstone blocks 97 worked out in great detail. Some of the pilasters are carved out of one block, the others being fixed together on the base mouldings after each part, such as the capital, the shaft and the base, were separately carved. No.7: The core of the cornice and pilaster is of schist slabs. The other details cannot be detected. No. 12: Certainly this stupa is later than both stupas Nos. 11 and 13, as it is inserted into between them. The only characteristic detail I noticed is that the cornice of both plinths are made of schist slabs. No. 14: This lies between Nos. 13 and 15 and is later than them. The core is of rubble and river boulders. The facing is untraceable for the second plinth. Nevertheless, the base mouldings are laid with schist slabs, and the trapezoidal niches on them are carvings of sandstone blocks. The schist slabs are used for the cornice of the lower plinth. No. 19: The facing and the pilasters are cut from different sandstone blocks. No. 28: Same as above. No. 29: The facing is of schist slabs, while only the pilasters are carved out of sandstone. The examination of small stupas reveals that the classification given Mustamindi is quite unacceptable. The associations of many kinds of stone and their usage do not seem to point to such a simple conclusion, albeit a tentative one. To make his scheme to him, an even less valid, the ornamentation of incised lines on pilasters is, indication of the later group, but it occurs on stupa No.9, which he included in the earlier group. 5. Composition of the Main Stupas All that is left of the main stupa at Tapa Shotor in its composition is a high square-shaped plinth, 4.7 m. square in plan, approached from the southeast by a long flight of steps. The superstructure including the dome and the drums has now disappeared, but the lofty plinth is still standing at its almost full height with no stucco coating over the top portion that has been damaged. The core of the structure is of conglomerate with its facing of the same material covered with coarse stucco. The figural decoration on the northeast (right), Northwest (back) and southwest sides of the plinth consists of a standing Buddha, cast in stucco without attendants, occupying the bays between each pair of pilasters. Six pilasters divide the sides but the facade into five bays. Turning our eyes to the Lalma site, we notice that the main stupa there has a high plinth of the same appearance as that of Tapa Shotor (Illus. 14, above). It measures 8.5 m. square in plan. The core of the structure and the facing are of blocks of conglomerate. The remnants of the sides are observed to have been coated with lime plaster, which is more whitish and finer than that used at Tapa Shotor. On some bays, however, coarser plaster mixed with sand and similar in quality and colour of material to that of Tapa 98 Shotor was used, after the original fine face had been cut off along ihe pilasters with a sharp knife. A marked contrast between the repaired bays and the original ones on each side can be seen even from a long way off (Mizuno 1968: PIs. 18, No.6 and 19, No.1). The figural decoration between each pair of pilasters dividing each side except the front one into seven bays is originally a Buddha, standing on a low socle supported with base mouldings of the plinth, with smaller Buddhas who also stand as attendants on each side of the central Buddha and are fixed a little higher than the base mouldings (Mizuno 1968: PIs. 20-21). Though the type of coated plaster at Tapa Shotor and Lalma differs, the style and expression of the Buddha images on both stupas show close similarities. Attention should be paid to the bays repaired with coarse plaster, where the figures are neither carved nor cast but drawn in brush work of coarser vermillion lines that have almost faded away. Thus at Lalma following its renovation the statues in stucco on the main stupa were replaced by this type of drawing. The coarser plaster similar in quality used at both sites suggests that Lalma and Tapa Shotor co-existed for some time. Seven other statues of the .i.Buddha;, seated or standing, are laid along the base of the plinth at Lalma (Mizuno 1968: PI. 18), while such arrangement of images is not visible at Tapa Shotor. However, the interstices between the step and the pilaster at the corner on the front facade are punctuated by trefoil-shaped niches which house the seated Buddhas. In one niche to the right of the step is the dhyana Buddha attended by a worshipper who is kneeling in adoration outside of the niche. In the left niche the Buddha preaching is carved without followers. Both niches look as if they are similar in their shape, but differ in minor points as follows. In the right niche, the ratio of height of the central arch to that of the side foils is slightly greater than that of the left niche: and the whole left niche is supported with dwarf pilasters set on the base mouldings of the plinth, but the right niche is directly based on the base mouldings. Along with such ornaments, the seated on a high throne is fixed to the right and left sides of a projection formed at the corners of both step and main square body of the stupa. The left facade of the projection is adorned with a standing Buddha image whose style is the same as those punctuating other bays. On both sides of a flight of steps the Buddha and his attending images, both of them standing on socles, are fixed. The main details such as the pilasters, the mouldings and the niches are formed of schist slabs carefully selected in order to serve as the basis for the desired forms and then finished off with a coat of coarse lime plaster. The device is also same as in the cases of the small stupas Nos. I and 2 as well as of the main stupa at Lalma. Also common to each main stupa is the feature that the cores of the plinth and pilasters were not built at the same time: namely, first the cores of the plinths were made, leaving spaces for the pilasters to be loaded, then the schist slabs were carefully inserted one on another into the moulds for the pilasters. Elaborately cut slabs for this technique seem to have been necessary. 6. Elements of the Mouldings 99 The plinths of the main stupas at Lalma and Tapa Shotor are supported by double sets of mouldings of the usual torus-and-scotia pattern (lllus. 15, 3 and 4). But it is worthy of note the upper scotia of Tapa Shotor is finished off in beve]led facets (Illus. 15,4). Though the lowest scotia and the lower part of the upper base are also of beve] led facets at both main stupas, the only difference is in the facetted scotia seen at Tapa Shotor. As shown in the illustration No. 15, if we view the pattern of the mouldings at stupa No. 141 of Tapa Kalan, Hadda, as more rounded and more faithful to the original style, then such angular details may be regarded as the product of the decision to bypass the extensive labour needed to carve such rounded mouldings (Illus. 15, 1). If indeed such a stylistic difference can be ascribed to the passage of time, stupas may be arranged into chronological order. In other words, such stylistic difference should be taken into consideration as a way to reconstructing a history of monastic architectures. (1) Styles of Plinths: An Enigma of the Stupa Court at Tapa Shotor Excavation revealed that the stupa court at Lalma underwent two successive stages. During the first stage the aforesaid high plinth of the Lalma Main stupa, which resembles that at Tapa Shotor, was actually the second (upper) plinth which was supported by a low basic plinth: that is, in the later stage the second plinth of the first stage still survived, but the lower plinth of the first stage had been filled up with debris (Illus. 15, center). In the first stage the main stupa at Lalma had two-storeyed plinths. A lofty upper plinth, the extant height being 4.3 m. above the floor, is supported by a low basic plinth of bigger dimension measuring 9.32 m.long on the back side and 0.82 m. high above the floor. The small subsidiary stupas, fifteen of them along the eastern half of the main stupa having been excavated so far (Nos. 2, 3, 5-17), are laid out so closely to the main stupa and also to one another that one cannot make a procession through the space lying between them (lllus. 14, above). Accordingly the procession around the main stupa might only have been made on the lower plinth. Another possibility is that a procession was made along outside of the whole small stupas. The cores of almost all the small stupas at Lalma are of boulders laid in mud with the facings of sandstone blocks (Mizuno 1968: PI. 26). Each block is well-shaped in order to fit with its required position. Lime plaster covers the facings as usual. The pilasters and the niches between them are also carved out of sandstone blocks (Mizuno 1968: PI. 27, Nos. 1 and 2). At the last phase of the earlier stage the small stupas, Nos. 5-17 in particular, had been cut offleaving their square plinths (Mizuno 1968: PI. 27, No.6). The height of the remaining plinths corresponds to that of the lower plinth of the main stupa. The cutting off resulted in the paving of a new floor at this level, which was filled with mud, rubble, Buddha images, and dismembered blocks of sandstone (Illus. 15). As a result, the main stupa took a new appearance, which retained only a lofty plinth that had previously served as the upper or second plinth. Close to the cornice of the southwest side of the previous lower plinth of the staircase, a small stupa, No.1, was built of sandstone blocks, 100 and another stupa, southwest of this, was also made of the same kind of stone blocks. All that is left of these small stupas of the later stage are the lowest courses of the plinths, namely the lowest courses of sandstone blocks, but notable is the fact that these stupas were made in the same fashion and material as those of the earlier stage. So without data concerning stratification noted above, one can not dispute which stupa is earlier or later. Solely typologically, we cannot solve any problems of chronological order of stupas. Such a clear stratification in the courtyard of stupas does not occur elsewhere. But Tapa Shotor seems, to my mind, to be an exception. The thickly plastered floor of the stupa court at Tapa Shotor shows the sharp lines resulting from the difference of levels on the existing floor, which regularly surround the main stupa (Illus. 14, below). I would argue that one of the lines seems to be no other than the margin of the cornice of the lower plinth: the main stupa, and accordingly the stupa court, at Tapa Shotor might have undergone two stages, earlier and later, as that at Lalma did. The floor is divided by the lines into different levels: the floor closer to the main stupa is slightly higher than the floor outside of the lines. Beginning from the west corner of stupa No.4 in front of the main stupa, a line runs clockwise to the east angle of stupa No. 18, and reappears between No. 22 and the annex of No. 2. It is observable that this line keeps running parallel with the main stupa and that all stupas except No. 18 are laid outside of the line. When we keep in mind that the later small stupas at Lalma were constructed on a new level, and that they abutted what had been the cornice of the lower plinth, this line at Tapa Shotor becomes very significant in that it might indicate the cornice of the lower plinth of an earlier stage of construction. The narrow regular space between the main stupa and small ones at Tapa Shotor can only be explained if it is the upper surface of the lower plinth from a previous phase of construction. Certainly and actually this space had been and was for circumambulation in both stages, and it had to remain without small stupas even after a new floor was laid. Because the builders knew well the dangers of building even small stupas on earth surfaces of varying hardnesses that resulted from the filling-up of the area outside of the original body of the stupa and the original body itself, they did not build stupas here. Outside the line 'a' separating the earth of the original stupa from the area that was filled in later is another line (Line 'B'). It surrounds rows of stupas on all four sides except on the southwest, where several stupas (Nos. 19,28-30,32,26,21, and 33) lay beyond this line. These two lines mark off different levels. The area between the two lines is slightly lower than the center area and also slightly higher than the surrounding area on which the southwestern row of small stupas lies. The area between the two lines is supposed to have been filled up after the initial phase of construction. In it, most probably, lie the remains of the flight of steps which led up the facade of the original lower plinth. All of this suggests that the existing floor, where the edifices are intact, belongs to a later stage of the stupa court at Tapa Shotor while the floor or floors of earlier stupa or stupas are sealed beneath it. Whether the outer line of the two may delimit the boundaries of an earlier courtyard or not is only a matter of guess. (2) Relative Sequence of Stupas 101 During the earlier stage at Lalma the floor of the stupa court underwent three successive coatings with coarse lime plaster (IHus. 15, bottom), which we designate the floor IA (earliest), IB and IC (latest). On the earliest floor lA are laid out the three small stupas Nos. 2 4. Special mention should be made of stupa No.4: the lower plinth adorned with five pilasters on the sides is an extension of the lower plinth of the main stupa which is not decorated with any pilasters. This extension is built between the square body and the staircase leading to it. The second plinth of stupa No.4 on this base has facings, extant only on both sides facing the staircase and the right facade of the main stupa, which are punctuated by five pairs of pilasters standing on the base mouldings built of schist slabs. The bays are adorned with the dhyana and dharmacakra Buddhas respectively, and on the side facing to the staircase, the dhyana Buddha is enshrined in a semi-circular arch built of very thin slabs of schist and the dharmacakra Buddha in the trapezoidal one, while on the other side the arches are only semi-circular. All arches are supported by the dwarf pilasters with Corinthian capitals on which other more slender 'Indian' pilasters rest. The core of this stupa consists of bigger round boulders from river-beds, and the details such as pilasters are coated with stucco which is very weathered. Although stupas Nos. 4 and 8-10 are all on level IB, the fact that the Northwest borders of stupas Nos. 8-10 run flush with stupa No.4 suggests that stupa No.4 was built earlier. On the floor IC, or the latest floor of the earlier stage, are stupas Nos. 5-7, 11-17. The cores of these stupas are made of boulders from river-beds with facings of sandstone blocks, from which the pilasters, cornices and mouldings as well as the niches on the bays have been carved out. Such facings are the same as those of stupas Nos. 8-10 on the floor IB and of stupas Nos. 2 and 3 on the floor lA. So, even on the earlier stage of Lalma there were two kinds of construction which are also observable at the later stage of the courtyard at Tapa Shotor, though boulders were rarely used for the cores at Tapa Shotor. The development of stupas at Hadda was once discussed by Barthoux, who maintained that schist-faced stupas were earlier than the others (Barthoux 1933: 157f.). A forced chronology of the stupas at Tapa Shotor was also proposed by Mustamindi (1969a: 17-18). The excavation at Lalma, however, cannot support any of their views. The composition of the main stupas at both sites is very similar, but they differ in details such as the scotia of the mouldings and the quality of applied plaster. And if we take the later stages of both sites as contemporaneous, then we cannot agree with those regarding the schist-made stupas as predating others. Even in a restricted area such as Hadda, putting stupas into more reasonable chronological order is not a simple matter. At this point what should be re-examined is the prevailing view that holds that schist predates stucco not only as a building material for stupas but also as a sculptural one. Nothing stratigraphical from the Peshawar plain, Swat, Jalalabad, or Taxila has proved Marshall's hypothesis (Marshall 1951: 75 f.). In Taxila the reserve may even be true. Among the earliest known stupas in Taxila (Sirkap) some were neither constructed of nor decorated with schist, but were built of the kanjur stone which made stucco easier to adhere. 102 7. For Absolute Dating Lastly our review touches on a system of absolute dating as proposed by Mustamindi (1969 a: 23-24). In stupa No. 24, which, according to him, belongs to a later group, was found a pottery vessel in which a lump of copper coins had been deposited. The coins include one copper coin of Menander and small copper coins attributable to the Sassanian Shapur III (383-388). Based on the latter, he opined that the later group of the small stupas might date from the end of the fourth century. And he supposed that the earliest possible date was given by many Kushan coins found in the courtyard. the Sassanian coins of Shapur III only give clue to the terminus a quo for the construction of stupa No. 24 and they cannot be used as evidence to date the site itself. Mustamindi also defined the end of the site as the Hephthalite trampling of Buddhist temples that has generally been maintained as a plausible explanation since Marshall adopted it in his excavation report on Taxila (1951: 76). To my mind, the Hephthalite destruction of Buddhism and Buddhist temples in Gandhara remains sheer conjecture. Song Yun's record edited in the Luoyang Qielan ji (Record of Buddhist Monasteries in Luoyang), which Marshall cited as providing conclusive evidence for the Hephthalite sacking of Buddhist temples in Gandhara, gives no such account: In the middle of the fourth month in the first year of the Zhengguang era (520) they entered Gandhara previously known as the state ofYeboluo. It was conquered by the Hephthalites who installed a tegin as the king (of Gandhara). Now two generations of Hephthalite kings have reigned (since the occupation). The nature (of the present king) is violent and cruel, and he often conducts massacres. He does not believe in Buddhist faith but devotes himself to non-Buddhist creeds. As all the inhabitants are Brahmans who respect Buddhist teaching and enjoy reading s, so it is deeply against their wishes that they suddenly have such a king. Relying on his bravery, the king has been fighting against Kasmir for control of territories for three years .... The king stays on the border without returning at the end of the day. The soldiers become weary and people grow tired. All the inhabitants sigh with resentment.. .. After five days journey to the west of the Buddhist temple (where Song Yun and Huisheng were sent by the king to lodge), (Song Yun and Huisheng) arrived at the place where the Tathagata cut his head and gave it to an old Brahman. There was a stupa and a monastery where about twenty monks were residing. Again travelling westwards for three days, they reached the great river Xintou (the Indus) .... Travelling again westwards for three days, they arrived in the city Boshafu (Shahbaz Garhi). The land on the rivers arc fertile and the fortified city is rightly straight, the population large and 103 flourishing, the woods and the springs lush and numerous. The land is rich with precious articles, and the customs are refined and good. In and out of the city there are old temples where virtuous priests and monks reside and their devout conducts are highly excellent. One Ii to the north of the city is a temple called the White Elephant Palace, where all of the Buddhist statues are of stone, extremely fine in decoration and very many in number. The entire body of each statue is covered with thin leaves of gold, producing a dazzling effect on the viewers .... (Original translation). Who can demonstrate the Hephthalite sacking on the basis of the above passages? The Hephthalites occupied Gandhara and beyond, as the nomadic peoples from the north of the Hindu Kush had always done so, and claimed the territories extending from the west bank of the Indus to the Salt Range in the south and to the regions around modern Jhelum in the southeast. Their claim brought them into frequent conflict with the king of Kasmir, who also wanted to be politically influential in the regions. We have no positive evidence for the point that the Hephthalites destroyed Buddhist temples and killed monks when they first came into the Northwest. The above passages only inform us of the fact that Song Yun met either the second or third tegin (king of Gandhara), not the first, and moreover, that the king conducted massacres and did not believe in Buddhism, not that he massacred Buddhist monks! Not accepting Buddhism and massacring are two different matters that should not be regarded as one and the same. It would be natural to suppose that the Hephthalites had their own religion and hence gave no support to the thriving Gandharan Buddhism and temples. Their main interest was presumably to tap the richness and fertility of the Gandharan lands. In Tokharistan the Hephthalite headquarters were first attacked in 555 by the Turkish chieftain, Mugan Khaqan, then in 558 by the joint expedition of Sinjibu, the Khaqan of the West Turks, and Khusraw 1, the Sassanian king who was the son-in-law of Sinjibu. Also in 568 Sinjibu again struck a blow against the Hephthalites and extended his sway over Tokharistan (Altheim 1969: 260 261; Haussig 1956: 23). The Chinese Annals (the sections of the Hephthalite state in the Zhoushu and the Suishu) also inform us that the Hephthalite tributes to the Chinese court ceased after they had sent the last mission in 558 and that they ware defeated and occupied by the West Turks. The Hephthalites in the Northwest, who are naturally supposed to have had close connections to their headquarters in the north of the mountains, in consequence, became helpless and disintegrated with the loss of their homelands. The Northwest gradually declined after Song Yun's visit in 520, and certainly by the seventies of the sixth century had completely lost its former fame and fortune. 8. Conclusion Here I have touched on several different points concerning specifically Tapa Shotor and Lalma, and more generally Gandharan archaeology. I have described different stylistic 10+ features of the stupa such as the presence of corbelled or barrel-vaulted roofs, suggested which naga legend may be depicted in chapel H at Tapa Shotor, and challenged the historicity of the Hephthalite destruction of Buddhism in Gandhara. The importance of, as well as the difficulties in, establishing reliable chronologies for the Gandharan region should be clear to the reader by now. I have reviewed Mustamandi's classification of stupas at Tapa Shotor partially in order to demonstrate where his scheme breaks down and partially to suggest how such an analysis might be done. Different stylistic features as well as varying composition of the stupas makes dating them far from straightforward. The similarity of certain stupas at Lalma to those at Tapa Shotor suggests that the two sites were contemporaneous for some duration, and, accordingly, some of our findings about the Lalma site may apply to Tapa Shotor as well. Lalma is an unusual site in that the different layers of occupation are so clearly distinguishable. There are slight variations in depth which accord with different periods of construction. Most importantly, a floor of mud, rubble, Buddha images, and dismantled blocks of sandstone lie over the earlier level and clearly mark a second stage of occupation. This pattern at Lalma enables us to make sense of the unusual configuration of stupas at Tapa Shotor. The sharp differences in step surrounding the main stupa mark off the cornice of the original lower plinth, which, as in Lalma, became the new floor level in the later phase of construction. The varying hardnesses of the earth-a solid original stupa and a softer filled-in area-supports my contention; as does the fact that the builders, fully cognizant of the danger of building on surfaces of different hardnesses, did not construct any stupas there. Careful comparison of the composition of the stupas reveals that some of the building techniques that characterize the earlier Lalma stupas also characterize the later Tapa Shotor stupas. Marshall's argument that schist predates stucco as a building and sculptural material simply does not square with the data from Tapa Shotor and Lalma. Finally, I would like to raise a point of particular concern to me. Given the enormous difficulty involved in establishing a chronological sequence for Gandharan sites, and so Gandharan art objects, I think we ignore the potential of pottery as a means of dating at our own peril. Even today excavators often tend to slight, if not neglect, pottery vessels found in temples, which are, to my mind, indispensable for placing Buddhist temples in relative chronological sequence. Pottery was always a necessity in the daily lives of monks, without whom one cannot discuss the history of Buddhist temples. Closely related to temples, moreover, are lay-people who lived in towns and in cities. Archaeological examination of both types of sites, religious and secular, is vitally important for establishing a comparative sequence of pottery that will illuminate the relative chronology of Gandharan temples. Even though this method may not produce immediate results, it may be the only way to determine a chronological sequence. If only we could establish guidelines for recording pottery. As it is now, with the classification of pottery left to the whim of individual archaeologists, it is impossible to use published reports to analyse pottery. Determination and designation of different pottery types vary as do even the words used to describe the colour of clay. I suggest that 105 we form a committee that will help to coordinate our future excavations and set up strict guidelines for all aspects of archaeological investigation, even including the recording of pottery types. The chronology of Gandhara is simply too important. Pottery analysis may be the only way to get at that chronology. *Numbering of stupas at Tapa Shotor in this article is used as given by Mustamindi himself (1969a) and the designation of niches and chapels such as A, B, D, .... is given by me as done in the originalJapanese text. 106 VII The Hephthalites in Tokharistan and Gandhara The sixth century was the turning point in the history of the regions in the north and the south of the Hindukush. In the middle of the fifth century Tokharistan the Hephthalites had already been so active as to oppose the Sassanians . Overwhelming Sassanian Persia under Peroz and Kawad I they debouched into the Northwest where the son of Kidara of the Kushans had been ruling. By the early years of the sixth century Hephthalite power had culminated in both Tokharistan and the Northwest with the Hephthalite king receiving tributaries from some forty neighbouring countries. The decline of the Hephthalites seems to have begun in the mid-fifties of the sixth century when the newborn nomadic state of the Turks in the Northern Asiatic steppe made a sudden a.ppearance across the Amu Darya. A decline in the Hephthalite branch in the Northwest must have inevitably followed after ties with their headquarters in Tokharistan were cut. As a reflexion of the Hephthalite confusion and their following decay a drastic change occurred around the middle of the sixth century regarding the routes that had long since connected the Northwest with Tokharistan or, broadly speaking, with Central Asia. Quite a new highway was opened crossing the western passes of the Hindukush via Kapi§i and Bamiyan to Tokharistan. The old routes that had connected Gandhara with T okharistan through the valleys between the Karakorum and the Hindukush were to be ignored. This change is mutually related to the history of the regions on two sides of the Hindukush: the rise of Kapi§i and Bamiyan in and at the southwestern foot of the ranges and the fall of Gandhara at the southeastern foot (Kuwayama 1987A). With this background in mind it is time to challenge the long accepted view of Hephthalite vandalism against Buddhist activities in Gandhara and to readdress the Hephthalites' presence in Tokharistan before and after the Turkish advent to include problems related to their qishlaq (winter residence) and yailaq (summer residence) in Tokharistan. 1. The Hephthalites in Gandhara and the Authenticity of their Destruction of Buddhism I have elsewhere stated that the Hephthalite destruction of Buddhism and Buddhist temples was sheer conjecture and briefly discussed how such a superficial view had been posed from inaccurate interpretations of the Chinese literary documents, actually quoting the paragraphs on which such scholars had based their theories (Kuwayama 1987B: 173; 1990: fn.8). Xuanzang's witness of the desolate Buddhist monasteries in Gandhara and its environs in the late twenties and early thirties of the seventh century shows that the Gandhara Buddhism was decayed before Xuanzang. Yet I cannot agree with those believing that it happened with the Hephthalite attacks of Gandhara. If so, when and how did the 107 Buddhism run down here? I will again discuss in more detail that the Hephthalites had nothing to do with this decay and that really historical significance existed with a series of events which happened after the political withdrawal of the Hephthalites. The hypothesis that Buddhism in the Northwest was destroyed by the Hephthalites or the Hunas was proposed by Fleet (1885: 65-68;1886: 245-252;1888: Nos. 13-15, 18-20 and 33-37;1889: 225-230) and Stein (1900, 1: 43. fn. 289;1905: 73-87) who put together literary documents edited in China, inscriptional sources discovered in India and numismatic evidence. Marshall (1912-1913: 8-9) quite fantastically found in the two different kinds of skeltons dug up at the Dharmarajika monastery his own clear evidence of fighting between the Hephthalites, represented by one type of skeleton, and the monks who were attacked by them, represented by the other type. Marshall who had already believed in the Hephthalite (White Hunic) invasion very superficially combined his interpretation with the hypothesis which Fleet and Stein had proposed. Even Foucher (1918: 586f) accepted Marshall's view. Marshall (1936: 52£) again strengthened his own idea in 1936. Since then scholars have been choosing their own flavours of reality (Rowland 1967: 77; Ghirshman 1948: 108; Wheeler 1949: 10-11; Wheeler 1963. 557; Ingholt 1957: 16-17; Hallade 1968. 71). Although Dani (1986: 5-6 and 75-78; Bussagli 1962: 24; Biswas 1973: 108-109) do not seem to accept the view of Hephthalite destruction of Buddhism, they are not yet able to show how the hypothesis does not match well with historicity. According to those who have promoted the view, Gandharan Buddhism seems to have been devastated in the latter half of the fifth century, as Marshall and Wheeler say, during the first advent of the Hephthalites into North India. Some have laid much importance on Mihirakula whom they suppose to have ruled over Gandhara in the early sixth century. This group includes a number of scholars in the field of the history of Chinese Buddhism (Yamada, R. 1955: 110-123; Kawakatsu 1981: 101-102; 1987: 501-537; Fujiyoshi 1987: 29-56; Yamada, M. 1990). Scholars of Chinese Buddhist history believe that the fall of Buddhism in Gandhara can be closely linked with the idea current among Chinese Buddhists in the Northern Qi and Sui dynasties that there had come so corrupt an age that Buddhist practice and attainment were no longer possible (Yamada R. 1955:110-123; Fujiyoshi 1987.29-56). According to them, the Indian monk Narendrayasas conceived the idea that the Final Age of the Dharma had come, while still in Uddiyana where he was born and grew up in the midst of the Hephthalite vandalism. In Northern Qi Narendrayasas translated into Chinese some Buddhist scriptures which preached the fear of the extinction of the True Law. As the most important among these scriptures such scholars put great importance on a rather short called the Lianhua mian jing (the Sutra of the Lotus-Masked: T. 12, no. 386) which includes a story that Gandharan Buddhism was destroyed by the king Mihirakula. Mihirakula is recorded as an infamous king for such deeds in the Da Tang Xiyu ji (the Great Tang Record of Western Countries:T. 51, no. 2087) and the Fu Fazang Yinyuan zhuan (the Account of the Causes and Conditions for the Transmission of the Treasury of the Dharma: T. 50, no. 2058). Scholars who believe the hypothesis to be historical fact have identified Mihirakula in these sources with a king who appears 108 with the same name in numismatical and epigraphical sources such as the Mandasor stone pillar and the Gwalior stone (Fleet 1888: 142f. and 161£.). However, we should be very cautious about Mihirakula in the Chinese sources, all of which were edited with views of Chinese Buddhist monks formed when the conception of the extinction of the True Law was current. In this social circumstance, they possibly laid all the blame for the decay of Gandharan Buddhism simply on a recognizable name associated with destruction, Mihirakula, particularly in the case of editing a new scripture giving authority to the current ideas of Buddhist theologians. In support of this warning Xuanzang does not say anything of Mihirakura or the Hephthalites when describing the dissolution of Buddhism in the Northwest, while devoting many words to Mihirakula's cruelty when describing the city Sagala (modern Sialkot). Furthermore Mihirakula on the stone inscriptions only appears as a king of the Huna, not a Hephthalite one. In this context it is noteworthy that M. Yamada[1990] find no reasonable ground for identifying the Hephthalites with the Hunas that exclusively appear on some stone inscriptions in India and therefore identifying Mihirakula on the inscriptions with the Hephthalites king whom Song Yun met. For a reevaluation of the Hephthalite invasion of Gandhara, the fifth volume of the Luoyang Qielan ji (A Record of Buddhist Monasteries in Luoyang: Zhou 1963) remains the single basic document. In this volume Yang Xuanzhi edited the narrative of the travels of Song Yun who met a Hephthalite king at his headquarters in Tokharistan and a Hephthalite tegin who was the king of Gandhara in 520. An original translation follows: In the middle of the fourth month in the first year of the Zhengguang era (520) [Son Yun] entered the state of Gandhara. The land is similar to Uddiyana, previously known as the state of Yeboluo. It was conquered by the Heda (Hephthalites) who eventually installed a chiqin (tegin) as the king [of Gandhara]. Now two shi's [of the Hephthalite king] have passed since their [first] occupation. (Zhou 1963: 209-210; Chavannes 1903A: 416-417). The Chinese word shi should not be read as signifying an actual number, i.e. thirty years, but interpreted as either a generation or a reign. This paragraph only says that at the time of Song Yun either the second or the third king of the Hephthalites was ruling Gandhara. This does not support those who date their invasion of Gandhara in the year 460 from the simple computation: 520 - (30 x 2) = 460. It is impossible to expect from the above passage alone the date when the Hephthalites first overran Gandhara (Enoki 1955: Mitchner 1975B:168). The next paragraph following the above has been thought to directly concern the massacre of Buddhist monks in Gandhara by a Hephthalite king, or more specifically, by the king Mihirakula. The Luoyang Qielan ji says:The nature of the king is violent and cruel, very often conducting massacres. He does not believe in Buddhist faith but well worships their own heathen 109 gods. As all the inhabitants in the country are Brahmans who respect Buddhism by much reading the s, so it is deeply against their wishes that they suddenly had such a king. Relying on his bravery the king has been fighting against Jibin (Kasmira) for the control of the territory for three years .... The king stays on the border all day long without returning to his residence. The soldiers become weary and people overburdened. All the inhabitants sigh with resentment. (Zhou 1963: 210: Wang 1984: 235; Chavannes 1903A: 417). This paragraph only describes the usual situation of the land at the time of nomadic intrusion. It was not unusual for the nomads to worship their own deities. It was also quite natural for the Buddhist population in Gandhara to have a king who believed in non-Buddhist creeds and therefore to face difficulties. But no statement is made in any paragraph proving that the Hephthalite king killed Buddhist monks or destroyed Buddhism in Gandhara, only that either the second or the third king of the Hephthalites in Gandhara had such a violent nature as to conduct killings and that he did not tend to believe in Buddhism but worshipped his own (or their own) deities. Indeed, the description of Shahbaz garhi, or the central area of Gandhara, in the Luoyang Qielan ji is far from a picture of decay either in regard to general well-being or devotion to Buddhism. The land on the rivers is fertile and the city wall is exactly straight. The people and houses are flourishing and many. The woods and the springs are lush and brimful. The land is rich with precious articles and the customs are refined and good. In and out of the city there are old temples. There are learned priests and men of virtue whose devout conduct is highly excellent. One Chinese mile to the north of the city there is a temple called the White Elephant Palace. All of the Buddhist statues in this temple are carved of stone, extremely fine in decoration and very many in number. The entire body of each statue is covered with thin leaves of gold producing a dazzling effec on the eyes of people. (Zhou 1963.213; Wang 1984: 238; Chavannes 1903A: 419). The description of Uddiyana in the Luoyang Qielan ji is consistent with that of Shahbaz garhi: In the early twelfth month of the second year of the Shengui era ( January to February AD 520 ) [Song Yun and Huisheng] entered Uddiyana which bordered on the Pamir to the north and India to the south. The climate is mild. The territory covers several thousand square Chinese miles(li). The products are as abundant as in the holy land of Linzi, and the beautiful fields are equal to the best land of Xianyang .... The king [of Uddiyana] is diligent as a Buddhist and purifies himself, taking vegetarian foods. He worships the Buddha in the morning and evening, beating the drums, blowing the 110 conches, together with mandarins, lyres, pipe-flutes and pan-flutes, all of which are played. After midday [the king] takes the state affairs .... North of the city is the Tuoluo monastery where the greatest number of the Buddha statues are and the stupa is lofty and the monks' cells are crowded side by side. There are six thousand gilt Buddhist statues surrounding. The pancavarsika ceremony of the king is held here every year. [On this occasion] the monks in this country all come here like clouds .... (Zhou 1963. 199; Wang 1984: 228-232; Chavannes 1903A: 407-411) Scholars who believe in the Hephthalite destruction of Gandharan Buddhism without any regard for nomadic modes of invasion have linked two quite independent matters: conducting massacres and not believing in the Buddhist creed. This oversight has promoted the false image of the Hephthalite king as killer of Buddhist monks. However sticking strictly to the language of the passages, all that the Hephthalite king in Gandhara may have done was to show no interest in Buddhism and to fight against Kasmira. Rather important than no interest in Gandhara is the fact that the king had been fighting against Kasmira for the control of certain territory. As we can well understand from the text itself, the Hephthalite fighting against Kasmira might have exhausted the inhabitants of Gandhara through increased exactions of taxes above the usual burden imposed by such nomads. Table 1 - Distance between Military Camp of the Hephthalite King and Charsada. Distance Faxian SongYun Xuanzang (days) o Military camp of the Hephthlite king Site of the VyaghriJataka 4 Place where the Tathagata gave an old brahman his head Town of cutting head Takshasila 6 8 Udabhandapura The Sindhu River 9 10 11 12 Boshafu (Shahbaz garhi) Varushapura Place where the Tathagata gave an old brahman his eyes Pushkalavati The real reason for fighting against Kasmira seems not to have been the claiming . III of territory, but controlling Kasmiri trade with the region of the Salt Range. This can be discussed from the whereabouts of the military camp of the Hephthalite king. There is no firm evidence that he was staying in Gandhara proper (Table 1 and Illus. 16, below). According to the Luoyang Qielan ji (Zhou 1963: 211-212), Song Yun met the Hephthalite king in his military camp located an eight day journey to the east of the Indus or the usual ford of the Indus called Udabhandapura in the Rajatarangini (Stein 1900, I: v. 153, 232, vii. 1081; II. 337, 338) or Waihind in Biruni's India (Sachau 1888, I: 206). The original text says that after a five day journey to the west of the monastery where Song Yun was sent by the Hephthalite king to lodge, he arrived at a place where the Tathagata cut and gave his head to an old brahman (Chavannes 1903: 418; Wang 1984: 237). Faxian also refers to a town whe~e Bodhisattva gave his head to a man (T. 51, no. 2085. 858b; Legge 1886: 32). Xuanzang mentions the Jataka story as taking place at Takshasila (T. 51, no. 2087.884 c). Travelling again westward from Takshasila for three days, Song Yun arrived on the river called Xintou (Sindhu), the Indus. Travelling westward from the Indus for three more days he reached the town Boshafu (Varushapura), modern Shahbaz garhi, which is identical with Jiantuowei, or *Gandhavati in Faxian's narrative. Both pilgrims, Song Yun in the sixth century and Faxian in the fifth century, took six days in all from Taxila to Shahbaz garhi. This fact implies that six days represent the official distance between the two towns in those times, not the time required for specific individuals to pass between them. Therefore, comparing the distance between Taxila and Shahbaz garhi with that between Taxila and the military camp of the Hephthalite king, I presume that the king was camping somewhere around modern Jhelum, a town strategically important as a gateway to India and a gateway connecting Punjab with Gandhara. More importantly in the context of the Hephthalites' fighting against Kasmir, Jhelum is also situated between the eastern end of the Salt Range and the southern foot of the Pir Panjal Range. It is through the latter that the Kasmiris gained access to markets of salt in the Punjab or salt mines in the Salt Range, where, as Xuanzang stated in Chapter 3 of the Da Tang Xiyu ji, Singhapura was located. The Kasmirian author in the eleventh century, Kshemendra, gave a glimpse of the ancient salt trade saying that his heroine Kankali in the Samyamatrika proceeds to Surapura as the wife of a load-carrier engaged on the 'salt-road' (Samayamatrika 2,90; Meyer 1903: 24, no. 91., f.n. 4). Stein also refers to the route of salt trade connecting Kasmir and the Punjab through the Pir Panjal pass. According to him, it had remained to his day the chief route by which the produce of the Punjab salt mines coming via Jhelum and Bhimbhar entered Kasmir, as he used to meet daily the long strings of salt-laden bullocks when marching into Kasmir by the Pir Panjal route (Stein 1900: 2, 395, no. 44. f.n. 47). With this fact in mind we may be allowed to think that the Hepththalite king in 'Gandhara' at the time of Song Yun's visit was checking the salt trade of the Kasmiris encamping in and around Jhelum. To summarise, the interpretation of Song Yun's narrative edited by Yan Xuanzhi in the Luoyang Qielan ji gives no positive evidence for the direct destruction of Gandharan Buddhism by the Hephthalites. Instead the narrative itself proves the prosperity of Buddhism in Gandhara and Uddiyana. Dissatisfaction from the general populace with 1I2 the Hephthalite king was likely to be a reaction to an unpopular and costly war. Now we turn our eyes to further evidence that the Hephthalite destruction of Buddhism is an illusion. It is provided by the biography of Narendrayasas to which scholars have never drawn attention in this context. Since this source is crucial for giving a clear picture of Gandhara and Uddiyana in the first half of the sixth century, an English version of the biography is given first and is followed by a detailed discussion. 2. Significance of the Biography of N arendrayasas First, it is necessary to mention the various editions of the biographies of Narendrayasas. The earliest known is included in the twelfth volume of the Lidai Sanbao ji (Records of the Triratna through the Ages) edited by Fei Zhangfang and completed in 597 (T. 49: no. 2034). The second and the third ones are found in the fifth volume of the Da Tang Neidian lu (Great Tang Record of the Buddhist Scriptures, completed in 664; T. 55: no. 2149) and the Tang Gaoseng zhuan (T. 50: no. 2060), both of them edited by Daoxuan (596-667). The biography in the Da Tang Neidian lu is a copy of that in the Lidai Sanbao ji and shorter than the more detailed version in the Tang Gaoseng zhuan. The biography of Narendrayasas is also included in the sixth and the seventh volume of the Kaiyuan Shijiao lu (The Kaiyuan Record of Sakya's Teaching) completed by Zhisheng in 730 (T. 55: no. 2154). This is the same copy of his biography as in the Tang Gaoseng zhuan. At the end of the biography of Narendrayasas in the Tang Gaoseng zhuan Daoxuan points out that Yancong edited another biography of Narendrayasas (T. 50: 433b). In the thirty-third volume of the Suishu, which is a catalogue of the books in the Sui imperial library, there is included a five volume publication entitled the Da Sui Fangjing-poluomenfashi Waiguo zhuan (Great Sui Record of Foreign Countries Informed by the Indian Monk-translator) which is now lost. The eminent Indian translators in the Sui period were Narendrayasas, Jinagupta (Chavannes 1905) and Dharmagupta. Yancong is known to have compiled the biography of Dharmagupta in four volumes as described in a catalogue ofYancong's writings in the fifth volume of the Da Tang Neidian lu (T. 55: 278c). Yet there is no evidence in any accessible documents that a biography was compiled about Jinagupta except for that included in the Tang Gaoseng zhuan. So, if the Da Sui Fangjing-poluomen-fashi Waiguo zhuan were not edited with information drawn from all three Indian monks, this book might have been Yancong's biography of Narendrayasas which Daoxuan mentions in the Tang Gaoseng zhuan. If so, it is possible that Daoxuan used it as a basic source for editing his larger biography of Narendrayasas in the Tang Gaoseng zhuan. Thus two versions of his biography are at our disposal. The following translation of the most detailed biography of his, to which we now are able to have access, is quoted from the Tang Gaoseng zhuan: Narendrayasas. Zhuncheng in Sui Chinese. He came from Wuchang (Uddiyana) in North India .... In his seventeenth year (at the age of sixteen) 113 he made up his mind to associate himself with the monkhood. He was fortunate enough to have virtuous priests as his teachers, from whom he learned the right doctrine of Buddhism. In his twenty-:first year (at the age of twenty) he received the complete precepts. [When he did so] he heard about the Shadow of the Buddha which various senior monks were admiring. Some told that there was the alms bowl of the Buddha in a certain country, the robe of the Buddha in another, the skull bone, and the teeth, the miracles of which were various. He eventually made up his mind to see them with veneration. Receiving the complete precepts, however, one ought to become well versed in the vinaya. [Therefore] he accomplished this study in five summer retreats (varsa) and then departed for other countries since such places as the stone foundation of the heaven's ladder and the precious stupa of the Nagas had still remained over many countries of India. He himself worshipped all of them with his forehead touching to the earth. No place to visit was left. He resided at the Kalandaka monastery [in Magadha] for ten years. Visiting monks' cells he met many eminent monks among whom was a priest of virtue good at fortune-telling. He said to Narendrayasas, "If you are serenely trained, you will have attained perfect results. But extensively travelling you will probably succeed in nothing. You do not understand me from the bottom of your heart when you hear me today. It is no use crying over spilt milk even though you later call to mind what I am talking to you". Narendrayasas successively visited the sacred remains from the snowy mountains (the Himalayas) in the north to the country of Lion (Sri Lanka) and returned to his homeland (Uddiyana) where, seeing the king of the country, he realised that the king was a real devotee of Buddhism. [He could find] few who shared the acquaintance of such a personage, even from his experience [of long journey]. The following is a tentative outline of him: With justice the king puts the mind of people at rest, so the people love him as they love their parents. Around five o'clock in the morning, first of all, he worships the Three Treasures (the Buddha, the Dharma and the monks) from the bottom of his heart with incense and flowers as well as music. When the sun rises he sets up the court and attends to the affairs of the state. Then around 8 o'clock he sprinkles scented water over the statue of the Buddha. In the palace he daily serves the mid-day meal to one hundred monks. The king and queen serve it with their own hands. For digestion after the meal he practices martial arts. When the sun is getting low he copies the ten lines of the scripture, discusses the meaning of the Dharma with various virtuous monks, and inquires into administration together with the whole body of officials. Mter the sunset he enters the Buddhist chapel to make prayer to the Image of the Buddha and to recite the s, holding the lamp with his own hands. Nothing unusual has ever happened in any of these matters. [After] carrying through all of the 114 daily routines he retreats to his private room. He has never failed for these thirty years to complete these meritorious deeds. The king has hundreds of sons and daughters, all of them being faithful and adoring their father. The good custom of the Sakyas has survived in this country. However, his monastery was so closely located to the hills that it was burnt down by wildfire. Each monk therefore was sent to distant countries where they were to reside. Making a group consisting of six monks [Narendrayasas] went to the region north of the Snow Mountains (the Karakorum) where they were to convert people to Buddhism. Reaching the steep top of the mountain, they found the two roads, one for men and the other for demons. The former was desolated and difficult to go, while the latter easily traversable. Therefore travellers waver in their judgement and many take the demon road. As soon as entering the territory of demon, one may be killed. Once was a wise king, raising at the entrance of the roads a stone image of Vaisravana whose hand indicated the road good for men. A monk in company got distracted and entered demon's road. N arendrayasas, sensing the danger, hurried after him for the distance of one hundred paces, chanting the sacred mantra of the Avalokitasvara. But the monk had already been killed. Narendrayasas himelf could narrowly escape the difficulty due to that mantra. Accordingly continuing to proceed, they encountered the thieves. Again chanting the same mantra, Narendrayasas was protected by divine power: thieves came to stab, yet they could not see him despite they were just in front of him. Following the routes to the east Narendrayasas arrived at the country of the Ruirui, where he met with the Turkish invasion that blocked up the routes to the west. Narendrayasas cut his wish to return to the homeland. Then he wandered as fate would take them to reach the coast of muddy sea in the north which was located seven thousand Chinese miles to the north of the Turkish territory. They never felt easy there and came farther down to the land of the Chinese Qi dynasty. In the seventh year of the Tianbao era (556) they arrived at the capital Yeo The emperor Wenxuan received Narendrayasas with marks of distinction which were exceptionally different from the case of others. Narendrayasas was in his fortieth year (thirty-nine years old). He was upright and excellent as well as elegant in personality, so no one was critical of him. The emperor's honors were therefore quite cordial putting him in the Monastery Tianping and asking him to be the Master of Translating the Tripitaka. About one thousand sets of Sanskrit texts of the scriptures housed in the court were sent under the Imperial order to the Monastery Tianping where they were placed in the superior chamber. Moreover, a seminary for the priesthood was established there with a good supply of treasures. Separately building the treasury and the kitchen the emperor thus expressed his profound 115 respect to him. The emperor also ordered the sramana Fashang, chief controller of the priests, and others, more than twenty in number, to supervise the translation work.... In the beginning some fifty volumes of the scriptures were translated .... Whenever Narendrayasas was free from his duty of translation he often muttered magic words to save one's life and help others, with much success. He was, after a short while, appointed as the assistant controller of the priests and suddenly became the chief. He did not spend for his own use anything that he received as his own salary.... In the end of the Jiande era (557) the emperor Wu of the Northern Zhou dynasty [which was governing the western half of the late Northern Wei dynasty] subjugated Northern Qi. [As the emperor had interdicted Buddhism in his state since 554], both the state of Northern Qi and its Buddhism were at the same time destroyed. Narendrayasas fled for safety here and there wearing the monk's robe inside and putting a lay costume over it without taking a bit of peaceful rest .... Such difficulties continued for four years. When the Sui dynasty came to control over the country the dynastic authority again led the Three Treasures to prosper. In the beginning of the Kaihuang era the Sanskrit texts of the scriptures came from far in response to this fortune. Thus the Imperial rescript requested Narendrayasas to come to translate them. In the seventh month of the second year his disciple Daomi and other monks accompanied him to the capital where he was to dwell in the Monastery Daxing Shan. In the last of the three winter months (the twelfth month) of that year the translation work was thus inaugurated. About thirty members headed by sramana Tanyan, the controller of the priests, were ordered by the Imperial rescript to take charge of translation .... Mterwards he moved to reside in the Monastery Guangji and became the head of monks from abroad .... When he suddenly died, his age was just one hundred. That was on the twenty-ninth day of the eighth month in the ninth year of the Kaihuang era. In his early age Narendrayasas met a man who well told one's fortune by physiognomy and this man told Narendrayasas that he would live to be a hundred and ascend to the world of legendary wizards. He ended his life fulfilling this longevity. The saying of that man was thus embodied. Although the reason for ascending to that world was difficult to anticipate, his physical shape of face and head was great and unusual: the top of his head was swollen like a ushnisha looking like a mountain above the clouds; his eyes were situated in the middle of the face, dividing it into equal halves; his ears were lofty and long, provided with a perforated lobe. Among fortunate physiognomies his mask had never comparison. That is why he was from the outset a man of virtue succeeding the Law (dharma) .... According to Narendrayasas, his wandering covered more than forty 116 years and more than fifty countries as well as one hundred fifty thousand Ii .... The things were really so tremendous and extensive that there had not yet been any time to describe it. Therefore sramana Yancong compiled his biography and widely distributed it. (T. 50: 432a-433b) As I discussed elsewhere (Kuwayama 1990), the biography itself gives two different dates for his life. On the one hand, if we base ourselves solely on the passage which says that he was in his fortieth year (at the age of thirty-nine) on his arrival at Ye in the seventh year of the Tianbao era (556), his date )f birth would be 517 which would give 537 for the year he received the precepts and 541 for his departure for India to pay hommage to the relics of the Buddha and others. On the other hand, the date of his birth would be 490 if we rely on the passage that he was in his one hundredth year (at the age of ninety-nine) at his death in the ninth year of the Kaihuang era (589). In this case his receipt of the complete precepts would be in 510 when he was in his twenty-first year (at the age of twenty) and his departure for India after five summer retreats would be in 514. The second dating is supported by his biography in the Lidai Sanbao ji which says that he was more than ninety in the tenth month of the fifth year of the Kaihuang era (585) when the compilation of all his translated works in the Sui period was completed. This age matches well with the dating calculated from the date of his death in 589, since he would be ninety-six in the fifth year of that era. In addition, the biography edited by Daoxuan says that he had wandered for more than forty years. If we take it right that the forty years were calculated from his date of arrival at the capital of Northern Qi (556), the year he left Uddiyana after five summer retreats would be 516. This date almost corresponds to the year calculated for his first departure for India, 514, based on 490 as his date of birth. Thus a date of birth of 490 and age of one hundred at death in the ninth year of the Kaihuang era (589) are consistent with all of the issues except the passage stating that he was forty on his arrival at Ye in 556. The age of forty in 556 which is described first in the Lidai Sanbao ji and copied by Daoxuan in his Tang Gaoseng zhuan could be an original misunderstanding: forty should have been written as the duration of his wandering. Thus the date of birth 490 places the receipt of the complete precepts at the age of twenty-one in 510. Likewise his date of departure for India to see the sites of the Buddha must have been 514 because he departed after the five summer retreats, i.e. four years after receiving the precepts. Herewith we obtain clear and important dates that give a clue to solving whether or not Buddhism in Gandhara remained unchanged despite Hephthalite rule. To benefit from this clue it is first necessary to examine the biography's references to the Shadow of the Buddha, the Buddha's bowl and other relics like the Buddha's robe, skull bone and teeth, each of them still existing in the early decades of the 6th century in Peshawar, Hadda and Jalalabad. (1) The Shadow of the Buddha and Other Sacred Sites in Nagarahara The Shadow of the Buddha was in a cave in Nagarahara (modern Jalalabad). In the early fifth century Faxian actually worshipped the figure of the Buddha at the spot in the 117 cave. James Legge (1886: 39) gives the English version of the original Chinese text describing the Shadow in the biography of Faxian as follows: South of the city (Nagarahara), half a yojana, there is a rock-cavern, in a great hill fronting the south-west; and here it was that Buddha left his shadow. Looking at it from a distance of more than ten paces, you seem to see Buddha's real form, with his complexion of gold, and his characteristic marks in their nicety clearly and brightly displayed. The nearer you approach, however, the fainter it becomes, as if it were only in your fancy. When the kings from the regions all around have sent skillful artists to take a copy, none of them have been able to do so. Among the people of the country there is a saying current that 'the thousand Buddhas must all leave their shadows here. After listening to a monk who had come from Nagarahara and had been well acquainted with the shadow, even a Chinese monk in China who had never been there, Huiyuan, made a copy of the Shadow of the Buddha in his hermitage. The text of the inscription written by Huiyuan himself in 413, reproduced in Volume 15 of the Guang Hongmingji (compiled in 664 by Daoxuan; T. 53: no. 2103. 198a), says that Huiyuan had been informed of the Shadow of the Buddha by monks coming from the Western Countries but that he only knew about it clearly when he met a monk from Jibin (Gandhara) who was well versed in medidation (dhyana) and a learned framana from the south (India) who had deep knowledge about the vinaya. The historical context allows us to believe that the former is Buddhabhadra and the latter is Faxian. According to the Gaoseng zhuan (the Biographies of Eminent Monks, edited by Huijiao in 519; T. 50. no. 2059; Wright 1954; Shih 1968) Buddhabhadra, born in Northern India, is known to have visited Huiyuan's hermitage at Lushan and left there in 412 (T. 50. 334c-335c: Shih 1968. 90f.: Kuwayama 1985: 136-137). Most important is the fact that Buddhabhadra had a very good knowledge of what the shadow of the Buddha actually was since he quoted the story of the shadow in detail in the scripture, the Fushuo Guanfu-sanmei hai jing (T. 15: no. 643) which explains how one should practise contemplation of the Buddha. The scripture explains that, in response to the earnest request of the king of Nagarahara, Buddha came from Central India with a number of his attendants to Nagarahara and converted the king of wicked Nagas to Buddhism. He left the shadow of his seated figure in the wall of a cavern in order to prevent the wicked Naga king from resuming his evil deeds. Even through the story of the shadow of the Buddha given by Buddhabhadra is not necessary for the construction of this scripture a large number of pages are devoted to it. In addition, no Sanskrit text is left of this scripture. It is quite conceivable that the whole of the original scripture was never translated but that Buddhabhadra compiled the Chinese version at least partly without the original texts. In such a case a vivid description of the shadow would never have been made were it not for a writer, such as Buddhabhadra, whose native land was Nagarahara and who actually saw the shadow himself. It is very likely that Buddhabhadra was the very monk who inspired lIS Huiyuan. No written or archaeological source is known which might shed light on the actual origin of the shadow of the Buddha in N agarahara and there is no way to prove how and why the story was made. Yet it is well documented in Chinese biographies of both Chinese and Indian monks that the Buddha's relics in Nagarahara, including the shadow, were very famous in the first half of the fifth century at the latest. For instance, Daofu, Baoyaun, and Zhimeng saw the shadow (T.50: 336a, 339c, 343b; Shih 1968: 106, 123, 145). Daorong also saw it as cited in the Song Yun's narrative in the Luoyang Qielan ji. He is not documented in any Chinese Buddhist sources. Nevertheless, as Chavannes (1903A: 415) rightly considered, Daorong should be read as Daoyao who was listed by Daoxuan in the Shijia Fangzhi, or the Buddhist Geography, (T. 51: no. 2088. 969c) as a framana who had travelled through Kashgar and the upper Indus valleys to Sankasya in the later years of the Northern Wei Emperor Taiwu (424-451). Daoyao saw the shadow in the forties of the fifth century. It is quite important for us to find in the biography of Narendrayasas that monks in U ddiyana were still paying hommage to the shadow of the Buddha when he was there in and around 510 for by the early seventh century the shadow was rarely visited by travellers or by local people and hardly visible aulien 1853: 79f.) as described in a biography ofXuanzang, the Cien zhuian. Before that time, i.e. between the fourth century and the second decade of the sixth century, monks travelling from China to India tended to stop off at Gandhara and most of them continued on to Nagarahara. For the Chinese in particular a visit to Nagarahara was a most popular kind of pilgrimage since many sacred sites and objects of the Sakyamuni in addition to the shadow of the Buddha attracted them (Kuwayama 1987A: 714f.). For example, the parietal bone of the Buddha in Hadda and his teeth in the city of Nagarahara were among the famous objects of worship. Faxian says that proceeding sixteen yojana to the west from Gandhara (or Purushapura) he reached the town Hadda which was otherwise called the Place of the Buddha's Parietal Bone. Eye-witnesses at about the same time as Faxian were Zhimeng who departed Changan in 404 and Tanwujie, or Dharmakara according to Chavannes (1903A: 403-436) or Dharmodgata according to Shih (1968: 115) who left Changan in 420. Faxian recorded how the local nobility performed the rite of the parietal bone (T. 50: no. 2059. 338c, 343b; Legge 1886: 36-37) and that in the midst of the city Nagarahara there was the stupa for the Buddha's teeth where offerings were made in the same way as the parietal bone. Daoyao says, as quoted in the Song Yun's narrative of his journey in the Luoyang Qielan ji, that the Buddha's teeth were kept in the monastery Qihelan si, the location of which is still to be identified. The parietal bone and the teeth of the Buddha there were worshipped as objects of faith even long after the fifth century. Xuanzang gives a clear picture of eccentric beliefs in these objects in seventh century Nagarahara aulien 1853: 78; Bea11884: 91f.). In Nagarahara the Buddha's robe and staff were also sacred remains. As in the case of the teeth, the robe and the staff at the time of Faxian were not in Hadda but to the northeast of the city Nagarahara. Ayojana to the north-east of the city brought Faxian to 119 the mouth of a valley, where there is Buddha's pewter staff; and a vihara also had been built at which offerings are made. The staff is made of Gosirsha Candana, and is quite sixteen or seventeen cubits long. It is contained in a wooden tube, and though a hundred or a thousand men were to try to lift it, they could not move it. Entering the mouth of the valley, and going west, Faxian found Buddha's sanghati-kafaya, where also there is reared a vihara, and offerings are made. It is a custom of the country when there is a great drought, for the people to collect in crowds, bring out the robe, pay worship to it, and make offerings, on which there is immediately a great rain from the sky (Legge 1886: 38-39). Daoyao's account indicates that these objects were still absent from Hadda in the middle of the fifth century: After arriving in the country of Nagarahara we found a piece of Buddha's bone from the top of his skull, square outside and round inside, four inches long and beige in colour. At the lower end of the bone was a hole big enough to insert a finger; lining the inside were tiny pockets as numerous as the cells of an inverted hornet's nest. Visiting the Qihelan si monastery, we found a thirteen-patch kiifaya of the Buddha. Measured with a ruler, some of the patches were long; others short. There was also a pewter staff of Buddha's. Seventeen Chinese feet long and gold-plated, it was deposited in a water pail that was also gold-plated. Its weight varied from time to time: when it was heavy, one hundred men could not lift it, but when it was light, two persons could handle it (See also Wang 1984: 243-244). Xuanzang also notes the existence of the robe and staff. He specifically mentions a multi-storeyed building in Hadda as the location of the Tathagata's sanghati-kafaya, pewter staff, eyeball, skull and aforesaid parietal bone, each sealed in a precious box. Since Faxian locates the robe and staff northeast of the city of Nagarahara and Daoyao locates them in the monastery of Qihelan si the site of which is still unknown and nowhere specifically associated with Hadda, it appears that they had been relocated for some reason by the time of Xuanzang' s visit in the fourth decade of the seventh century. (2) The Buddha's Bowl in Purushapura The Buddha's bowl in the biography of Narendrayasas is the real witness that resists the hypothesis of the Hephthalite destruction of Gandharan Buddhism. As I specifically discussed before (Kuwayama 1990), the key object of Buddhism in Gandhara was this bowl enshrined in a specific building located either in the northeastern (Da Tang Xiyu ji, II; T. 51: 879; Beal 1884: 98) or in the nothwestern (Shuijing zhu, Commentary on the Water Classic,by Li Daoyuan-d. 527, II; Petech 1950: 60, no. 65) ofPurushapura. Various Chinese documents, Buddhist and secular, give a clear picture of it. There was a temple said to have been built by a king of the Great Yuezhi in order to enshrine the bowl (Legge 1886: 34-35). The temple consisted of three buildings such as a stupa, a monastery capable of housing a great number of monks and a shrine of more than 120 two-storeys which housed the bowl. On a four-legged pedestal in the centre of the first floor was the famous bowl made of blackish blue stone, covered with golden threads (Petech 1950: 60, no. 64) and horizontally incised with no less than three lines on the rim. The incised lines well show that the bowl was a composite of four bowls, all of them being the same in size, dedicated to Buddha by the four heavenly gurdians at the first meal after the Enlightenment. According to the Shishi Xiyu ji (Record of Buddhist Western Regions, compiled by Daoan) quoted in the Shuijing zhu, the bowl already existed in Peshawar in the early fourth century. Buddhist literatures say that the king Kanishka or a king of the Yuezhi brought it back from central India to Peshawar as a ransom. The story is simply a ruse designed to obscure the fact that the bowl was created in the land of mleccha itself, i.e. in Gandhara. After all, a place such as Gandhara had no historical connection with the Sakyamuni. When Buddhist activities later became vigorous in Gadhara, the country presumably needed raisons d'etre to support the authenticity of Buddhism in the area. The Buddha's bowl, created and enshrined in the main city of Gandhara, thus became a central object of worship without which Buddhism might not have been able to consolidate its position in Gandhara. In the early fifth century even a monk as far away as Sri Lankawell knew the importance of the bowl in Gandhara, as Faxian heard him recite the following from the pulpit: Buddha's alms bowl was at first in Vaisali and now it is in Gandhavati. After so many hundred years-he gave the actual number of years when Faxian heard him, but Faxian forgot it-it will go west to the state ofYuezhi; after so many hundred years to Khotan; after so many hundred years to Kucha; after so many hundred years to the land of Han (China); after so many hundred years to Singhala; after so many hundred years it will return to central India. Then it will ascend to the Tushita heaven. When Bodhisattva Maitreya sees it he will say with a sigh that the patra of Sakyamuni has come. And with all the devas he will present to it flowers and incense for seven days. When these have expired the bowl will return to Jambudvipa where it will be received by the king of the sea nagas and taken into his palace. When Maitreya shall be about to attain to perfect wisdom and become Buddha the bowl will again separate into four bowls which will return to the top of the Vinataka hill whence they came. After Maitreya has become Buddha the four heavenly gardians will again think of the Buddha with their bowls as they did in the case of the previous Buddha. The thousand Buddhas of this Bhadra-kalpa indeed will all use the same bowl. And when the bowl has disappeared the Law of the Buddha will go on gradually declining to extinction. After that extinction has taken place the life of man will be shortened till it is only a period of five days. As mentioned above, Narendrayasas learned about the sacred objects like the Buddha's bowl when he received the precepts in 510. He departed Swat after fulfilling the five summer retreats in 514. These dates clearly show that Buddha's bowl in Gandhara 121 was still an object of worship in Peshawar in the first decades of the sixth century. The above discussion on the Shadow of the Buddha, Buddha's bowl, and other relics such as the robe, the staff, the skull, and the teeth iIi Hadda and Jalalabad shows how important they were as destinations for Buddhist pilgrims and as instruments for legitimising Buddhism in Gandhara. The discussion has also shown how important the relics can be as clues to proving the flourishing Buddhism in Gandhara in the first part of the sixth century. But just as important as the biography's notations on the relics is its description of the king ofUddiyana. N arendrayasas returned to U ddiyana long after setting out for India in 514, after living in the monastery of Kalandaka ten years and extensively travelling in India. It is quite unlikely that he would have returned if the Hephthalites had destroyed Buddhism in the Northwest. In fact, he found the king of Uddiyana practicing Buddhism with methodical devotion. As quoted above, the biography states: N arendrayasas successively visited the sacred remains from the snowy mountains (the Himalayas) in the north to the country of Lion (Sri Lanka) and returned to his home land (U ddiyana) where, seeing the king of the country, he eventually realised that the king was a real devotee of Buddhism. [He could find] few who shared the aquaintance of such a personage, even from his experience [of long journey]. The following refutes even more strongly those scholars who maintain religious vandalism by the Hephthalites: The following is a tentative outline of his life: with justice the king puts the mind of people at rest, so the people love him as they love their parents. Around five o'clock in the morning, first of all, he worships the Three Treasures (the Buddha, the dharma and the monks) from the bottom of his heart with incense and flowers as well as music. When the sun rises he sets up the court and attends to the affairs of the state. Then around eight o'clock he sprinkles scented water on the statue of the Buddha. In the palace he daily serves the midday meal to one hundred monks. The king and queen serve it with their own hands. For digestion after the meal he practices martial arts. When the sun is getting low he copies the ten lines of the scripture, discusses the meaning of the Buddhist dharma with various virtuous monks, and inquires into administration together with the whole body of officials. Mter the sunset he enters the Buddhist chapel to make prayer to the image of the Buddha and to recite the s, holding the lamp with his own hands. Nothing unusual has ever happened in any of these matters. After carrying through all of the daily routines he retreats to his private room. He has never failed for these thirty years to complete these meritorious deeds. The king has a great number of sons and daughters, all of them being faithful and adoring their father. The good custom of the Sakyas has survived in this country. 122 Very significant is the fact that the king had been practicing this program for thirty years. For a better understanding of the starting point of this thirty years it is once again fruitful to return to his biography. It tells that Narendrayasas encountered the Turkish invasion of the Ruiruis at the court of the latter. The conflicts between the Ruirui nomads and the newborn Turkish forces are well documented in the accounts of the Turks in the Zhoushu, the Suishu and the Beishi. They respectively record that Tumen Illiq Khaqan of the Turks attacked the Ruiruis in 552 and that Mugan Khaqan then occupied the territory of the Ruiruis in 555 with the result that the chieftain of the Ruiruis, called Dengshuzi, fled to the court of 1,"lTestern Wei in China. The above events prove that Narendrayasas reached the Ruirui's headquarters in the central part of Mongolia no earlier than 552. This date provides an important key to solving the question of when Narendrayasas left Uddiyana if combined with an additional clue available in the itinerary of Jinagupta. According to Jinagupta's biography (Chavannes 1905) which is put next to that of Narendrayasas in the Tang Gaoseng zhuan, he left his homeland Gandhara in 554/555 for Northern Zhou and reached Shanzhou (modern Xining in Qinghai) in 557/558 after passing through Kapisi, Bamiyan, Tokharistan, Wa'khan, and Khotan (Kuwayama 1987A: 718). Thus, in the case of Jinagupta, it took only two or three years to travel from Purushapura to Shanzhou. Although it IS not known which route Narendrayasas took, the duration of Jinagupta's travel provides a realistic estimate of the duration of Narendrayasas' own trip. We may therefore be allowed to surmise that he started for the northern region in the very early fifties of the sixth century. If the date of departure be thus placed, the king ofUddiyana would have been such a devout Buddhist as told in the biography of Narendrayasas since around 520. In support of this date, Song Yun similarly describes the king of Uddiyana based on his visit there in 520: The king [of Uddiyana] is diligent as a Buddhist and purifies himself, taking vegetarian foods. He worships the Buddha in the morning and evening, beating the drums, blowing the conches, together with mandarins, lyres, pipe-flutes and pan-flutes, all of which are played. After midday [the king] takes the state affairs .... (Zhou 1963: 200; Wang 1984: 228-229). Thus Buddhism in the first half of the sixth century in Uddiyana looks quite stable despite the hegemony of the Hephthalites. The descriptions of two sources, the biography of Narendrayasas and the Luoyang Qielan ji, provide clear evidence that the king of Uddiyana had for at least thirty years been one and the same person who devoted himself to the Three Treasures. However, one problem remains: no written evidence positively affirms that the Hephthalites were ruling Uddiyana at that time. To attempt to a solution to this problem it is necessary to examine the evidence for the arrival of Hephthalites in the Northwest. Ghirshman (1948), GobI (1967), and Mitchner (1975A;1975B) believe that the Hephthalites came to the Northwest from the west beyond Nagarahara: the Hephthalites first occupied Zabulistan, Kapisi and even Bamiyan 123 and from there invaded Gandhara. But, in opposition to this view, we have a source for the Kidara Kushans in volume 102 (the Western Region) of the Weishu, which states that they came south to subjugate the five countries located to the north of Gandhara:The king of Da Yuezhi called Jiduoluo (*kidara), brave and fierce, eventually dispatched his troops southward and invaded North India (modern the Northwest), crossing the great mountains to subjugate the five kingdoms which were located to the north of Gandhara. This account at least allows us to think that the Kidara Kushans did not come from the west, but does not go so far as to specify the five kingdoms north of Gandhara under Kidara rule. Nevertheless, other passages of the Weishu clearly show several kingdoms as subjugated by the Hephthalites. Such passages are usually believed to have been reused for the original edition of the Weishu quoted from the Huisheng Xingzhuan, the Account of Travels of Huisheng, now lost. Huisheng was a monk who accompanied Song Yun. Zhujubo, or Yarkand on the southwestern margin of the Tarim basin, is described to belong to the Hephthalites; Kepanduo, Tashkurghan on the Pamir, follows the Hephthalites; Buhuo, Wa'khan to the west of Tashkurghan, is governed by the Hephthalites; Shemi, Chittal in the upper valley of the Kunar, also follows the Hephthalites; and Ganduo, Gandhara, has been defeated by the Hephthalites. Highly suggestive, in the connection with the Kidarites, is the fact that the countries are five in all, the number being comparable with the number of the kingdoms which were located to the north of Gandhara. Since the other countries between Tokharistan and Gandhara are under the Hephthalites, it is unrealistic to exempt only Uddiyana. This tentative conclusion also suggests that the Hephthalites subjugated these countries in the same nomadic way as the Kidaras and as nomads have always intruded the Northwest whoever they may be: the Hephthalites, the Kidara Kushans, or the Kushans. Further it well strengthens the viewpoint that the Hephthalites came to the Northwest from Tokharistan via the valleys between the Hindukush and the Karakorum without passing Bamiyan, Kapisi, or Zabulistan. The passes in the western part of the Hindukush were only opened in the mid-fifties of the sixth century when the Western Turks replaced the hegemony of the Hephthalites who thereafter lost their vigor confined to the region associated with their traditional headquarters in Tokharistan (Kuwayama 1987 A). The nomadic inroad of the Northwest before the sixth century seems to have always been through these valleys, not through Bamiyan and Kapisi. To summerise my points, the invasion of the Hephthalites gave no cause for the fall of Buddhism in the Northwest. The narrative of Song Yun and the biography of Narendrayasas prove that it had prospered at least until the Hephthalite rule, or the middle of the sixth century. Yet in the fourth decade of the seventh century Xuanzang saw the desolation of Buddhist monuments. The decay of Buddhism therefore came in the Northwest in the latter half of the sixth century after the political withdrawal of the Hephthalites whose homelands in Tokharistan were mostly occupied by the Turks in the sixth and seventh decades of the sixth century. 124 3. The Hephthalites in Tokharistan before 558 As the nomads usually do when they politically overcome sedentary populations, the chieftain of the Hephthalites, supposedly titled the tarkhan which may be synonymous with "dahan" or "taihan" in Chinese sources, dispatched an army headed by a tegin to occupy Gandhara. There is no doubt that such a chieftain of the Hephthalites had his headquarters in Tokharistan. But exactly where he had his base within Tokharistan is still an open question despite much debate between scholars who have easy access to the Chinese documents on the Hephthalites and the town names relevant to their lives (Specht 1883; Marquart 1901; Chavannes 1903B; Shigematsu 1919; Hermann 1925; Fujita 1931; Markwart 1938; Ghirshman 1948; Enoki 1951; Enoki 1955; Funaki 1954; Matsuda 1970; Uchida 1972; Yu 1986). The sinologists tend to be confined by the limits of traditional ways of dealing with Chinese sources with little regard to the fact that the Hephthalites are nomads shifting with their daily utensils between qishlaq (winter quarters) and yailaq (summer pasture lands), and in fact take no notice of either the natural topography or aspects of human ecology which are so closely linked with each other in T okharistan. As a matter of fact, there is a great difference in Tokharistan between the area which lies to the west of the river courses running from the south via Rui and Samangan to Khulm and the area which extends to the eastern mountains from the course of the Surkhab. The division of Tokharistan into two distinct areas, eastern and western, is due to natural topography and human ecology however it may be that the Muslim geographers distinguish the Upper (east) and the Lower (west) Regions (Minorsky 1970: 337). The geographical line dividing these two areas lies in the low ranges of hills between the Surkhab and the Khulm. The routes connecting them are only two: one is in the north, running in the dasht which extends along the southern bank of the Amuand the other is in the south running over the Robatak pass which directly connects Baghlan in the eastern area with Samangan in the western and is only impassable for snowfall. Eastern Tokharistan is surrounded by mountains in the east and the south, but is made up of vast alluvial plains in the west and the Northwest. From the southern mountains the Surkhab runs northwards forming an alluvial plain in the Baghlan and Gori districts after meeting the Andarab at Doshi. It then flows farther northwards, collects much more water from the Talaqan-Kunduz river originating in the eastern mountains and eventually reaches the Amuat a point north of Qal'a-ye Zal. The rivers such as the Talaqan-Kunduz gradually descend from the east to make alluvial plains especially in their lower courses. These lands, extensively cultivated, support major towns like Kunduz, Khanabad, Talaqan and Khwaja Imam Saheb all busy collecting and distributing products in their bazaars. In the district to the west of the dividing mountains, on the contrary, the land is clearly split into two distinct regions, a mountainous area in the south and a plain in the north, and there is no gradual slope on the northern side of the mountains but a sharp 125 cliff separating the two areas. Rivers such as the Balkh and the Khulm do not make any open valleys in the mountainous area of the south but form alluvial fans suddenly meeting the plain, as the Khulm River does at Khulm. There they either form channels underground or dry up in the plain; there is no river that reaches the Amu. The extensive plain in the north can be made fertile only if well irrigated and a systematic irrigation system would only be possible under a strong political power. Lacking these such a town as Balkh could never have been born. The geography of eastern Tokharistan, more amenable to cultivation, must have better suited the natural development of numerous communities and the accumulation of wealth which would attract a powerful nomadic force, like the Kushans and the Hephthalites, to consolidate power there. As is the case with the lands of Sogd and the Tarim basin, there has never been any local sedentary chieftain who politically unified the lands lying between the Hindukush and the Amu. The oldest literary document ever known on these lands is the Shiji by Sima Qian who describes them after the account given by Zhang Qian who came there in 129 BC. It tells that there is no big political power governing Daxia, a Chinese transcription of Tokhara, but only minor chiefs each ruling a limited territory. The Hanshu supports this saying that originally Daxia had no major overlord or chief and minor chiefs were frequently established in the towns (Hulsewe and Loewe 1979: 121). More than seven hundred and fifty years later Xuanzang found that the royal race had been extinct for several hundred years, that the chieftains had by force contended for their possessions and that once again each held their own territory independently only relying on the natural divisions of the country. Xuanzang also describes that they thus constituted twenty-seven minor states divided by natural boundaries yet as a whole dependent on the Turks (T. 51: 872a; Bea11884: 37-38; Yule 1873). One may be allowed to replace the Turks with the Kushans or the Hephthalites for earlier periods of nomadic domination. We can see that Tokharistan had been populated with two kinds of inhabitants: (1) the sedentary population living in towns and villages; (2) the people of a nomadic way of life camping around the towns in the winter months and on the highlands like Badakhshan during periods of summer pasturing. Usually they lived side by side sharing their daily necessities but in most cases the nomads could lord over the sedentary people by advantage of their mobility on horseback. When Song Yun arrived at the headquarters of the Hephthalites in Tokharistan in the very early twenties of the sixth century he observed that the cultivated lands were extensive and the mountains and marshes were visible for a long distance (Zhou 1963: 195; Wang 1984: 225). The account of the Hephthalite country in the Western Region chapter of the Weishu tells that the king did not reside in the fortified towns but moved from place to place every month taking with him camping tents and military force and leaving his wives at intervals of two or three hundred Chinese miles. It further says that during the three winter months the king stayed at a place without going around his country. Although he did not reside in the town he must have had a winter residence (qishlaq) close to a leading town. The season in which Song Yun visited as an ambassador of the Empress Dowager Hu of Northern Wei was in the beginning of the tenth month 126 which is almost identical with the middle of November on our calendar. Song Yun therefore met the king of the Hephthalites at his winter residence. Since the qishlaq near a town is always in the lower courses of rivers in Tokharistan, the Hephthalite king received Song Yun somewhere on the Surkhab or the Talaqan-Kunduz but nowhere in Badakhshan, a site of summer pasturing. Sources mentioning the rise and fall of the Hephthalites are scarce and fragmentary but the Hephthalite section in the chapter of the Western Region of the Weishu describes that they subjugated Samarkand, Khotan, Kashgar, Margiana (Anxi) and other thirty smaller countries in the Western Region and that they called their state a big country having marital ties with the Ruiruis. Song Yun's narrative also describes that their power extended to the Turks in the north, to Khotan in the east, to Sassanian Persia in the west and to the still unidentified region Dieluo at the time when Song Yun visited the Hephthalite king and the tegin of Gandhara. It further says that some forty countries sent their envoys to the Hephthalite headquarters. The same source also informs us that they controlled the important places to the east of the Pamir, saying that Tashkurghan on the Pamir and Yarkand on the southwestern margin of the Tarim Basin were under their rule. The Hephthalites coped with the menace from the north by securing marital relations with the Ruiruis in central Mongolia, then the most powerful nomadic state in Northern Asia. The accounts of the Ruiruis in both the Weishu and the Beishi tell that the Hephthalite chieftain married three sisters of the Ruirui chief Poluomen and that in about 523 Poluomen tried to flee to the Hephthalites in quest of protection on his failure in rebellion against the local Chinese government at Dunhuang. This event is also mentioned in the biography of Feimu in the Weishu. This strong relation of the Hephthalites with the north also enabled them to ensure fearless control over oasis cities like Samarkand in the lands of Sogd and like Yarkand and Kashgar on the margin of the Tarim basin. Relations with the western power, Sassanian Persia, however, were not so congenial. The Hephthalites had conflicts in Khorasan with Sassanian Yazdegard II (438-457) between 441 and 452 and several times put him to rout. Yazdegard II paid them as the condition of armistice a huge amount of Sassanian silver coins which they struck with countermarks and used as their own currency. Among the Sassanian coins with Hephthalite countermarks those ofShapur II (309-379) are the earliest (Mitchner 1975A: 157-165). A series of countermarks on the coins of such Sassanian kings as Bahram V (420-438), Peroz (459-484), Kawad I (484, 488-497 and 499-531), Balash (484-488) and Khosraw I (531-579) comes from the heiday of Hephthalite power. The most numerous of the Sassanian coins with the Hephthalite countermarks are those of Peroz who several times denounced the Sassanian-Hephthalite border treaty and finally had to send his son Kawad as a hostage to the Hephthalite court. Peroz again proclaimed war against them in 484 but was killed in Mervrud. Ascending the throne after Peroz, Kawad I faced rebellion from the Karens in Armenia on the one hand and by Balash on the other, a brother of Peroz who claimed the throne. Kawad fled to the Hephthalite court and stayed there for four years. In 488 Kawad expelled Balash with the great aid of the Hephthalites but again had to ask for their protection as a result of 127 religious antagonism from his nobility. In 499, after a three year stay in the Hephthalite court, Kawad I returned but completely as puppet king in the hands of the Hephthalites (Christensen 1944: 289f. and 335f.). The king of the Hephthalites in the time of Song Yun had great influence over both the Sassanian king Kawad I and the tegin/king of Gandhara who had been fighting for three years against Kasmir. The first known official envoy of the Hephthalites to the Chinese court of Northern Wei was sent in 456, that is four years after fighting against the Sassanian king Peroz had ceased (See Table 2 below). But the next mission reportedly arrived only in 507 in the court of Northern Wei. Therefore one can surmise that the Hephthalites were busy with their western boundary, Khorasan, that had several times been claimed by the Sassanians. The movement of the Hephthalites toward the Northwest also took place during 456 and 507. The nomadic occupant of the Northwest preceding the Hephthalites was the Kidara Kushans who originated in Tokharistan. The chapter of the Western Region in the Weishu says as follows:The state of Little Yuezhi: the capital is Purushapura. The king of Little Yuezhi is a son of the king of Da Yuezhi. Kidara had been driven away by Xiongnu and fled westwards, and later made his son assume the defensive. That is why the son was called Little Yuezhi. Kidara cited in this paragraph was a descendant of the Da Yuezhi who resided in Tokharistan but was driven away by the newborn power of the Hephthalites (Xiongnu). According to Chapters 5 and 6 of the Weishu, a mission of Jichang arrived at the capital of Northern Wei two times in 459 and 460 and one of Juduoluo arrived in 477. As it records, in 459 the country of Jichang sent a mission to present gifts and in 460 the king ofJichang presented a trained elephant. In 477, it reports, countries such as Juduoluo, West India, Sravasti, and Zabul(?) sent a mission to bring a tribute. If Jichang and Juduoluo are phonetically identifiable with Kidara, Kidara would have been expelled after 477 by the Hephthalites (Xiongnu). The Hephthalite invasion into the Northwest may be datable to the time between 477 and 507. Table 2 shows the tributaries of three countries to the Chinese courts. It clearly points out the Hephthalites' political relations with both Sassanian Persia and Samarkand. Most notable is the blank space in the Hephthalite column between 456 and 502 during which they were busy first with the Sassanians and later with the Kidara Kushans as mentioned above. In contrast, the Samarkand column is full during that period. Apparently during the time that the Hephthalites and the Sassanians were fighting against each other Samarkand could be free from the yoke of the Hephthalites and send their own missions to China. After the beginning of the sixth century Samarkand disappeared from the scene and the Hephthalites reappeared after a long silence of nearly half a century, apparently exercising great influence over Sogdiana and Persia. We may surmise that the real heydays of the Hephthalites were in the first four decades of the sixth century. However according to the Zhoushu the Hephthalites stopped bringing tributaries to China after the second year of the Emperor Ming of Northern Zhou which corresponds 128 to 558. The reason we are told was the Turks. Table 2 - Tributaries to the Chinese Courts. AD Persia (Sassanians) 455 0 456 Hephthalites Samarkand 0 461 0 468 0 00 473 474 0 476 0 0 479 0 00 480 0 487 0 491 0 502 507 0 0 0 509 0 0 511 0 512 0 513 0 0 516 0 517 0 0 518 0 0 519 0 520 0 521 0 522 0 524 00 525 0 526 530 0 0 532 00 00 533 0 535 0 0 546 0 0 553 0 0 555 0 537 558 (Twice a year) 00 0 0 564 0 In 555, the second year of the Emperor Gong of Western Wei, as soon as he defeated the Ruiruis and occupied their territory, the Turkish chieftain Mugan Khaqan 129 extended his power to the west and defeated the Hephthalites (the Accounts of the Ruirui in volume 196 of the Tongdian and volume 98 of the Beishi; the Accounts of the Turks in the Zhoushu, the Suishu and the Beishz). The Turkish invasion of the Hephthalite country is reflected in the life of Jinagupta who stayed there in 555 (Kuwayama 1987A: 718;1988: 14). According to the biography of Jinagupta as included by Daoxuan in the Tang Gaoseng zhuan (Chavannes 1905), when Jinagupta arrived in the country of the Hephthalites he observed that the land was extensive but barren and the people did not produce anything to eat or drink. One need not go further to see the clear difference between the conditions which Song Yun saw in 520 and those which Jinagupta observed in 555. The difference may mark the period when the Hephthalites lost their vigor. The political situation of the Hephthalites in Tokharistan is also well shown in Jinagupta's biography. The specific Chinese phrase, shi (the political situation)-jian (hardships) which Jinagupta often suffered at the Hephthalite court literally means "current emergency" and well suggests the Turkish attacks against the Hephthalites in 555. In 558 the Turkish chieftain Sinzibu Khaqan, a younger brother of TumenIlliq Khaqan, attacked the Hephthalites from the north in alliance with Sassanian Khosrow I (531-579) who had replaced the pro-Hephthalite king Kawad I and had close ties with the newly emerged Turks as the son-in-law of Sinzibu. They released from Hephthalite control the regions around modern Tashkent area, the Ferghana basin, and Samarkand. In between 562 and 568 Sinzibu Khaqan finally conquered the Hephthalites and the Turkish power crossed the Amu to settle in Tokharistan (Marquart 1903: 64; Altheim 1969: II, 260-261; Haussig 1956: 23). These events are well corroborated the description in the Hephthalite account of the Zhoushu, noted above, which says that the Hephthalites, dispersed by the Turks, at last stopped sending embassies [to China] after the second year of the Emperor Ming of Northern Zhou, i.e. 558. The Suishu also records in the account of the Hephthalites that their country had been in a state of disorder and that the Turks had dispatched the Ton-shad Zijie to occupy it by force. This information about the invasion of the Turks against the land of the Hephthalites shows that the chieftain of the Turks tried to control the territory of the Hephthalites by sending his men to a place in T okharistan. The Chinese sources do not give an exact name of the place, but there are no lack of clues with which to solve it. There is clear written evidence for the later history of the Hephthalites in Tokharistan, with their headquarters around Baghlan, and for the history of the Turkish invaders themselves, based in the district of Qal'a-ye Zal, i.e. Huo by Xuanzang or Warwaliz by Muslim writers. 4. The Hephthalites and the Turks in Tokharistanafter 558 In the preface of the Xiyu Tuji (the Illustrated Accounts of Western Region) two names, Tuhuoluo (the Tokhara) and Yida (the Hephthalites), appear side by side in reference to Tokharistan. This book was edited by Pei Ju, a Sui officer, from the hearsay of long distance traders who had come and gathered at Zhangye in Gansu province 130 (Kuwayama 1975: 100-102; Uchida 1973: 115-128). He dedicated this book to the Emperor Yangti in the first half of 606. The whole of the Xiyu Tuji is lost, yet the preface written by Pei Ju himself is still accessible in his biography in the Suishu. The preface briefly refers to the three main routes connecting Dunhuang at the gateway of China with the western civilized worlds in the early seventh century, and the southern route among them was the only one leading to the subcontinent, the northern gateway to which was the Northwest. The preface mentions the place names on this route in the following order starting from Charkhliq, Khotan, Qaghaliq, Tashkurghan, the Pamir, Badakhshan, Tuhuoluo, Yida, Bamiyan, Cao and North Brahman country (Kuwayama 1975: 100). While the Xiyu Tuji refers only to the names, the detailed accounts of Tuhuoluo and Yida are found in the Suishu's chapter of the Western Region for which the Xiyu Tuji most probably served as a basic source. In relation to the country Cao identifiable as Kapisi (Kuwayama 1975: 99-101) each location is given: Tuhuolo is 1700 Chinese miles and Yida 1500 Chinese miles to the north of Kapisi. In addition, Bamiyan is described as 700 Chinese miles to the north of Kapisi. The information on the geographical interrelation of these four countries increases our understanding of their actual locations. A simple calculation locates Tuhuoluo 200 Chinese miles to the north of Yida. The two sources are available to locate Tuhuoluo and Yida in relation to the Amu. As the Suishu says (Table 3), Table 3. Interrelations between Tuhuoluo and Vida. The Wuhu River (the Amu Darya) 1700 The Tuhuoluo Country (=Huo[*Awan as capitall=Warwaliz=Qal'a-ye Zal) I 1500 The Yida Country (=BaghlaniGhorI. Huoluo as capital) I 700 The Fanyin Country (=Bamiyan) I o The Cao Country (=Kapisi) Yida is 200 Chinese miles south of the Wuhu River (the Amu). Another Chinese source, the Tongdian (chapter 193), mentions that Tuhuoluo was known to the Chinese court for the first time during the Sui period and that it was located on the south bank of the Wuhu. The problem we have here is the 'first time' in this sentence because the Tuhuoluo 131 ------------------------------------------------------------------------------------------~~ country had already been known to Northern Wei China as described in the Western Region chapter of the Weishu. If so, the Tuhuoluo which Sui China knew about for the first time is a country quite different from the Tuhuoluo country in the Weishu whch reportedly is so extensive a land as to cover the area between the Hindukush and Samarkand. By piecing together the complementary facts available in the Xiyu Tuji, the Suishu, and the Tongdian it is clear that Tuhuoluo in the Suishu must be a very limited district on the south bank of the Amu 200 Chinese miles to the north ofYida, not an extensive country described as Tokharistan (the Tuhuoluo country) in the other Chinese historical sources and by Arab geographers, too. All accessible information therefore points out that Tuhuoluo was on the south bank of the Amuand that Yida was two hundred Chinese miles to the south of it and that both of them were between the Hindukush and the Amu. According to the Cien zhuan-the most extensive biography of Xuanzang, a Turkish ruler resides in the country Huo which has generally been identified with Warwaliz in the Islamic sources (Yule 1873). When Xuanzang first arrived there, the ruler was the Tardu-shad by title and he was the eldest son of the Ton-Yabghu Khaqan of the Western Turks (Doerfer 1975: 124). During his stay, the Tardu-shad was poisoned by a man called by Xuanzang the tegin, a son of the Tardu-shad, who became the new shad and eventually called himself the Tokhara Yabghu. The same biography tells us that the Tokhara Yabghu was ruling the country Huo on the southern bank of the Fuchu River (the Amu). Therefore one may be allowed to believe that Xuanzang's Huo is the Tuhuoluo of Sui times. Another biography of Xuanzang included in the Tang Gaoseng zhuan says that the Turks residing at the southern headquarters in Huo (with the northern headquarters to the north of modern Tashkent [theJiu Tangshu, chap. 194b]) governed all of the smaller countries south of the 'Iron Gate' and extended their hegemony to Balkh. According to a detailed examination of the countries in Tokharistan described by Xuanzang in the Da Tang Xiyu ji, this statement seems a little exaggerated although Balkh was definitely under the Turkish occupation. This geographical book says that only such districts as Mengjian, Kuoxiduo and Andanluofu were subjugated to the Turks. Mengjian is modern Talaqan west of Kunduz, while Kuoxiduo is identifiable with Khost-Nahrin districts and Andanluofu is definitely phonetically equal to Andarab. These attributions suggest the Turkish influence may have been confined to the zone extending from Balkh through Huo to the highland districts east of the Surkhab and the zone extending from Khanabad/Talaqan southwards to Andarab. Significant is that not under the Turks was the middle course of the Surkhab, or the Baghlan-Gori area, where the Hephthalites must have had the original homeland. This means that their winter home in the Baghlan/Gori plain could be kept with them despite the loss of their most useful pastures (most probably in Badakhshan) at the advent of the Turks into the area south of the Amu. Furthermore, it is likely that the Hephthalites shifted their summer quarters to the Himatala district which was located at the gateway to Badakhshan. The account of the country Himatala in the Da Tang Xiyu ji suggests their fate in the face of and after the Turkish invasion:- 132 The first king of this country was a Sakya, fearless and bold. To the west of the Pamir most of the people were subdued to his power. But since the frontiers were close to the Turks, they adopted their low customs. When they suffered from the attack of the Turks, they protected their lands. However, the people of the Himatala were dispersed into different districts, where were many tens of fortified cities over each of which a separate chief was placed. The people live in tents made of felt, and lead the life of the nomads (BeaI1884: 290-91; With modification of the present writer). On the way back to China, Xuanzang reached the upper course of the Andarab, crossed the Hindukush at the pass of Khawak with his big elephant that had been given to him by Harshavardhana Siladitya on his farewell day and then proceeded 300 Chinese miles northwards to the Khost-Nahrin districts. He eventually arrived at Huo which is described as being about 600 Chinese miles north of the district of Andarab (Cien zhuan, chapter 5;T. 50: 249c-250a). According to the Da Tang Xiyu ji (T. 51: 872b, 939c-940a), Huo is northeast of Baghlan which is north of Ru-Siminjan. Ru-Siminjan is southeast of Hulin which is situated between Huo and Balkh. In the commentary of the Hudud al- 'Alam, Minorsky (1970: 338) properly identifies Ru-Siminjan with modern Rui and Samangan located in the upper and the middle courses of the Khulm respectively. Hulin in the Da Tang Xiyu ji therefore is no other than modern Khulm (Tashkurghan). The above geographical reconstruction affirms that the state of Huo in the Da Tang Xiyu ji and that of Tuhuoluo were one and the same district that covered an extensive alluvial plain at the confluence of the Talaqan/Khanabad and the Surkhab. Le Strange (1905: 408) quotes the description of Warwaliz by Hauqal and Muqaddasi. It is a town between Khulm and Talaqan reachable within a two day march from either of them. Xuanzang also notes that he reached Mengjan from the town of Huo in two days. Huo could thus be Warwaliz since all of the above sources, both Chinese and Muslim, agree in distance and direction. Whatever the towns of Huo and Warwaliz may be, I do not think it correct that scholars usually assign Huo to Kunduz, even if Kunduz means Bala Hisar in the northeastern suburb of modern Kunduz (Ball 1982: 222, no.931). The Chinese sources discussed above tell that the main town of Huo is on the south bank of the river but Kunduz is situated in the midst of the alluvial plain formed by a number of the branches of the Talaqan-Khanabad. The only candidate for the town of the Huo district which is compatible with this context may be Bala Hisar not at Kunduz but at Qal'a-ye Zal. It is located to the south of the Kunduz and just east of its confluence with the Amu (Barger and Wright 1941: 44;Le Berre 1965: 87-88; Bernard and Francfort 1978: 57; Ball 1982: 214, no. 892). Th Ciensi zhuan informs us that Huo is the southern headquarters, i. e., their qishlaq, of the Turks. If so, the summer pasture must have been in Badakhshan. Huichao visited the king of Tokhara at his residence in 726 and recorded that the king, or the Tokhara Yabghu in other sources, had been expelled to Badakhshan which was a month's march to the east of Huo when the Arabic forces occupied Huo (Fucks 1939: 449). 133 Huichao usually uses a specific term, 'wang(king)-zhu(residing)-cheng(town), , for the capital of a region. But in this case he does not use that term despite the fact that Tokhara Yabghu was forced to live there. This suggests that in Badakhshan there was no major town serving as the king's residence. The towns in Tokharistan shown in various Chinese literary sources therefore must be only in the lower course of the rivers and they must represent qishlaqs where the nomadic populations gathered in the winter season. Tokhara Yabghu of Tokharistan became an important figure in the history of China as the Tang Dynasty expanded westwards. From at least the middle of the seventh century on the Turks sent envoys and tributes to the Tang court: they were received at the New Year's day ceremony in the nineteenth year of the Zhengguan era (645). In 653 envoys reported to the Tang court that one of the Turkish chieftains, Wushibo (Wujingbo), had ascended to the position of the Tokhara Yabgu (Tang Huiyao, Vol. 99). From 658 on, Tang China was giving the official titles of China to the Turkish leaders and even attaching Chinese administrative names to each capital town. According to the chapter regarding Tuhuoluo (Tokhara) in the Tangshu, the Tang court in 658, on the occasion of establishing hegemony over the Western Regions as far as Persia, appointed the Tokhara Yabghu to be the Yuezhi Dudu (Governor of the Yuezhi) in charge of the Yuezhi Fu (Chinese Local Government of the Yuezhi) at his town Ahuan (*Awan. [Sogd.] "w'n(h): river(?)1,1. Sims-Williams 1983: 46). By 661 the Tang presence in the Western Regions was firmly installed. The local names of the sixteen towns between Khotan and Persia were replaced by Chinese names and local Chinese governments were established under the direct care of local chieftains who were given Chinese official titles. On the commemoration of Chinese supremacy the Tang court sent as ambassador Wang Mingyuan who was specifically appointed in charge of settling the local governments and commissioned to raise at Huo a stone monument inscribed in honour of the Emperor's benevolence (the chapter on Tokhara in the Tongdian Vol. 193). The Jiu Tangshu records the following: 'The Yuezhi Dudu-fu is settled at the town Ahuan, the capital of the Tokhara country; The Tang emperor has the king Yabghu (the Tokhara Yabghu) govern over his country which is divided into twenty-four sub-regions (twenty-five, according to the chapter on Geography in the Tangshu, which may include the capital itself), all of which are under the Yuezhi Dudu's rule'. As the Zizhi Tongjian informs, one hundred twenty-seven towns and cities west of Kashgar came under the Tang rule in 659. Included among such states is the Yida, or the Hephthalites. In the Jiu Tangshu the Hephthalites also clearly appear among the sixteen major regions: The Dahan government-general (otherwise called Taihan in some editions, the first character tai has one more stroke than da) is placed at the town Huoluo, the capital of the Hephthalite ulaq; [The Tang emperor] has their chieftain Dahan govern over his territory consisting of fifteen ulaqs which are under the rule of the [Hephthalite] Dahan. 134 Among the sixteen local governments fifteen ulaqs are the largest except for that of Tokhara. The number of ulaqs which this information refers to must have been close to the truth, since another source supports the number: the Xiuxian government-general established in Kapi§i at the same time as the Dahan covers eleven principalities, while Xuanzang mentions in the Da Tang Xiyu ji that Kapi§i rules over about ten principalities. The Hephthalites were still keeping power over the extensive area represented by these fifteen ulaqs even a century after the Turkish advent into Tokharistan. As demonstrated earlier, the Hephthalites with their headquarters at Huoluo must have shared the territory of Tokharistan in the middle of the seventh century with the Turks who were ruling at Ahuan in the Huo district at the confluence of the Kunduz (Surkhab) and the Amu. In view of this it is hardly possible to identify Huoluo with Huo despite the claim raised by Enoki (1951: 13 3-13 9) who regarded them as one and the same and assigned them only to Kunduz. In addition to the geographical evidence, to the contrary, the district Huo is the only place name in the Da Tang Xiyu ji that is constituted by just one Chinese character despite the usual transliteration of local place names with plural characters. For this reason Huo does not seem to be an abridged form of Huoluo. Really Xuanzang is quite right when he speaks of Huo in place of Tuhuoluo (Tokhara). At the time of Xuanzang Tuhuoluo was known as equivalent to Tokhara, not Tokharistan of the later Arab geographers but a place name within broader Tokharistan. Actually it must have been during his trip to the northern side of the Hindukush that Xuanzang first realized the truth: Tuhuoluo in the Sui period should have been called but was not equivalent to the entire expanse of Tokharistan. Therefore the town Huoluo must be found in another district of Tokharistan, most probably in the same district as the Yida country recorded along with Tuhuoluo in the Suishu. It is reasonable therefore to identify Huoluo with a town in the Baghlan-Gori plain which is located some 200 Chinese miles to the south of the Amu. The Baghlan-Gori plain is in the middle course of the Surkhab where some hundreds of years before the Hephthalites the Kushans had likewise dominated with a great fort-like temple at Surkh Kotal overlooking the extensive plain (Schlumberger, Le Berre and Fussman 1983) and from where in the later history of the Hephthalites a prince called tarkhan nezak rallied his force to move against Qutaiba b. Muslim (Minorsky 1970: 338; Frye 1979: xviii). 5. Towns in Tokharistan and their Whereabouts In connection with Huoluo, we have to discuss several place names which Chinese written sources gave to towns in Tokharistan, such as fudiye, boti, bodiyan, badiyan. Huichao gives the name fudiye to the town from which the king of Tuhuoluo (Tokhara) was expelled by the Arabs. The town fudiye is therefore the same as Ahuan in the chapter of Geography in the Tangshu and the capital town of the district Huo in the Da Tang Xiyu ji. The name fudiye also appears in the account of another country in the chapter of the Western Region in the Tangshu. This source says that from fudiye southwards one enters the Snow Mountains (the Hindukush) and reaches Bamiyan within a distance of four 135 hundred Chinese miles. But this account of the country in which the fudiye appears is on Chaghanian which is in fact located to the north of Tokharistan in the upper valley of the Surkhan, a tributary of the Amuflowing from the north. Despite the very obscure description of fudiye in Chaghanian, making one doubt whether it has later been properly inserted, Fuchs (1939: 449) regards fudiye in the account of Chaghanian as the same as 'the fudiye noted by Huichao and, moreover, identifies both of them with a town called boti in the account of Tokharistan in the Western Region chapter of the Weishu. Fuchs even more imprudently maintains that all of them are assignable to Balkh. According to Huichao, the distance between fudiye and Bamiyan is a twenty day march, while that between Bamiyan and Ghazna is seven days. Istakhri and Huaqal agree closely with the Bamiyan-Ghazna distance (eight days) but give the distance between Bamiyan and Balkh as only ten days. If we follow Fuchs' claim that fudiye is identical with Balkh, we cannot fill the gap between the above two sources as regards the distance between Bamiyan and BalkhijUdiye. Clearly Balkh is not a likely candidate for fudiye. The identification of boti is also not as simple an issue as claimed by Fuchs but certainly Balkh is not a candidate for boti as the account of Tokharistan given by the Weishu states as follows: In the center of Tokharistan is the town boti. It measures sixty Chinese miles in circuit. To the south of the town is a large river flowing westward, which is called the Hanlou. Firstly, no large river that well matches the above passage flows in the area south of modern Balkh. Secondly, according to Witzel (1980: 98), boti can be restored as *baxdi which means Balkh in the Avesta, and Balkh in the fourth-fifth centuries was normally called *baxl. Since boti is a transliteration in the time of Northern Wei (386-534), it cannot be Balkh. Volume 186 of the Taiping Huanyu ji, edited in Northern Song, describes the Tokhara country quoting some known and unknown documents that are now lost. It says: The town boti is now called bodiyan which is located in the northern part of the country. This implies that boti was an abridged form of bodiyan but it is still unclear if either boti or bodiyan is assignable to Balkh. Following the sentence cited above, the Taiping Huanyu ji gives the measurement of boti, saying; In the Tuhuluo country in the chapter of the Western Region in the Weishu there is the town called boti, the perimetre of which is sixty Chinese miles. This size is a little different from that of the town boti in the Xifan ji which measures fourteen or fifteen Chinese miles [square]. The Xifan ji, though lost too, is a record of the Western Barbarians edited by a Sui 136 official, Wei Jie, who travelled to the barbarian countries in the west, like Jibin and the wang(king)-she(residence)-cheng(town), as a diplomat for the Emperor Yangti. Presumably the accounts of the Western Region in this book were one of the basic sources for those in the Suishu. But, unfortunately, there is no clue as to the authenticity of the information given in the Taiping Huanyu ji which quotes the Xifan ji. The Weishu says that the town boti is located in the 'guo (country)-zhong' (in or in the centre of the country). Therefore the large river flowing westwards and located to the south of the town is no other than Darya-ye Arnu, while the town itself may be on its northern bank (Bernard 1978: 56-61, 99-100). If so, the town boti in Tokharistan in the Weishu is neither Balkh in modern topography nor fudiye in the accounts of Huichao and the Tangshu. The evidence therefore suggests that there were a number of towns such as fudiye, boti, and bodiyan which do not seem to represent the proper names of specific towns. A question also arises as to the whereabouts of another town called badiyan. This also looks similar to all of the above (Funaki 1954; Matsuda 1970), but it only appears in the Zhoushu (Vol. 50) as the capital town of the Hephthalites and is originally explained as that it means the 'wang(king)-she(residence)-cheng(town).' The size of that town is also alluded to as '10 Chinese miles on each side.' The Zhoushu contains the account of the Hephthalites instead ofTokharistan. In other words, Tokharistan and the Hephthalite country was synonymous in the time of Northern Zhou. In this sense the town badiyan can be taken as a leading town in Tokharistan. Such a big town was in the plain of the middle or lower valley of a river, not in mountainous regions where no wang-she-cheng existed. The claim by Funaki (1954) that badiyan was a town in Himatala is thus ruled out. The Da Tang Xiyu ji does not refer to any town there. This conclusion also resists all of the views proposed by scholars such as Specht (1883) who identifies badiyan with Badghis, Hermann (1925) who maintains that the Hephthalite summer quarters were around badiyan and Ghirshman (1946) who identifies badiyan with Boharak near Faizabad. Markwart (1938: 36-37) restores badiyan as *padiyan in Middle Persian and believes that the original word meant a king's city. His restoration however is not based on the phonetical system of Middle Chinese, nor is it supported by any evidence other than the above annotation in the Zhoushu itself that badiyan means the 'wang(king)she(residence)-cheng(town).' So we are still left without a reasonably restored original word for badiyan. The word would not seem to be a proper noun indicating a specific town. A more reliable clue to identifying the town badiyan is given in the Suishu. The Suishu tells us that the capital town of the Hephthalites measured about ten Chinese miles square. Therefore the town boti in the aforesaid Xifan ji that measured fourteen or fifteen Chinese miles square is apparently different from the town badiyan. In view of the size provided by the Zhoushu, the town badiyan corresponds to the capital town of the Hephthalites given in the Suishu and probably the town Huoluo in the Tangshu as well. Further, if one accepts that the Suishu and the Xifan ji are confined to the issues of the Sui time (581-618), the town boti in the Xifan ji cannot be the same as the Hephthalite capital in the Suishu. In order to show the size of towns the editor of the Da Tang Xiyu ji 137 uses only four measurements: "about ten Ii", "fourteen or fifteen Ii", "sixteen or seventeen li" and "about twenty li." It is not likely that any confusion would have existed between two towns ten and fourteen/fifteen Chinese miles square. One must also refrain from identifying the town badiyan with the capital town of the Tuhuoluo country in the Suishu which is now evidently assignable to Xuanzang's Huo, Huichao's fodiye and the Tangshu's Ahuan. The size of the Tuhuoluo capital is given in the Suishu to be 2 Chinese miles square, while the Da Tang Xiyu ji says that the capital of Huo is about 5 Chinese miles square (about 20 Chinese miles in circuit). The length of one side of a town given in the Suishu is often shorter and no town is given a size larger than 10 Chinese miles square except for the examples of the Hephthalites and the Persians. The size of fourteen or fifteen Chinese miles square associated with the town boti in the Xifan ji is clearly out of line with the size of the Tuhuoluo capital as known from all available sources and should not at this point be identified with the Tuhuoluo capital in the Suishu. The town badiyan was possibly the Hephthalite capital town described in the Suishu. The discussion of the names of town in Tokharistan such as bodiyan, boti, badiyan and fodiye may converge to a probable conclusion that each of them only indicates the town of king' residence as the author of the Zhoushu properly annotates. In this connection it is well reminded that people around Kabul do usually call it "shahr" instead of Kabul. The Hephthalites had their winter quarters around the town Huoluo in the BaghlanGori plain, as did the Kushans. Their summer pasture was first in Badakhshan and then in Himatala after the advent of the Turks into the Huo (Tuhuoluo) district. There is no written source to show that the Hephthalites had occupied Badakhshan and Huo before the Turkish invasion. Yet the Suishu suggests that the Hephthalites were living together with the people of Tuhuoluo. This statement may support their occupation of Huo. Also in view of the fact that the Hephthalite hegemony extended through Wa'khan, Chitral and Gilgit to the Northwest, no hindrance exists for a supposition that Badakhshan was theirs before the Turkish occupation. When the Turks came across the Amu, they apparently first occupied the mouth of the Surkhab (the Kunduz), the Huo district, as their qishlaq. Inevitably Badakhshan, the superb pasture land in Tokharistan, came under their control. The Hephthalites had to give up Huo and were probably able to retain only their homelands in Baghlan-Gori, ceding their pasture land in broader Badakhshan. Possibly the Hephthalites kept the western half, Himatala, while the powerful invaders took the better eastern half, Badakhshan. The head of the Turks was officially called Tokhara Yabghu and the Hephthalites were ruled by Dahan or Taihan. It would seem assignable to Tarkhan in Muslim sources. The Hephthalites and the Turks divided the area between tlle Amuand the Hindukush into their own territories. The Hephthalites thus seem to have been independent, even during Turkish hegemony, until the first decade of the eighth century. Their situations began to change in the second decade of the eighth century. According to Tabari, Tarkhan Nezak, a king of the Hephthalites, was also the vassal of the Tokhara Yabghu of the Western Turks and seized a Turkish lady of his sovereign in 710 but was captured and killed by Qutaiba b. Muslim (Chavannes 1903B: 49. fn. 2). 138 The Cefu Yuanggui (Vol. 999) includes a memorial presented in 718 to the Tang emperor by a younger brother Puluo of the Tokhara Yabghu Nuo(Pan)dunili. According to the memorial the Hephthalite chieftain had fifty thousand soldiers and horsemen at his command. In the same memorial an equal military power was also at the command of the kings of such neighbouring states as Khuttal, Chaghanian, Akharun-Shuman, Shughnan, Wa'khan; Guzganan, Bamiyan and Badakhshan. The Turkish Tokhara Yabghu himself had four time larger forces than the above chieftains supervising the kings to the south of the Hindukush such as Kapisi and Zabulistan who were holding the same amount of troops as the Tokhara Yabghu. In 729 the Tang emperor appointed Kutluk Ton Tardu, a chieftain of Tokharistan to the Tokhara Yabghu according to the Tang Huiyao (Vol. 99) and the Cefu Yuanggui (Vol. 996), and also to the king of the Hephthalites according to the Tangshu (Vol. 221b) and another information in the Cefu Yuangui (Vol. 964) . This is the last and latest Chinese historical record which touches on the Hephthalites. 139 VIII Two Itineraries concerning the Emergence of the Colossi in Bamiyan* Dating the colossal standing Buddha images in Bamiyan is faced with problems such as the lack of source materials comparable in archaeological terms and documentarily referable clues. The prospect of dating them may not be dark in light of the excavations at Tapa Sardar, Ghazni and at Adjina Tepe, Dushanbe (Taddei 1974: 379-381; Litvinsky abd Zeyrnal 1971). The clay figures of the gigantic Buddha found in situ lying in the sleeping position as when he attained nirvana, fifteen m. and twelve m. long respectively in the original measurement, have roughly been dated to the eighth century. So far as the Bamiyan colossi are comparable with those images in the archaeological contexts, there is no reason to place the standing colossi in the particular situations of Central Asian activities of art and architecture. However, as regards more exact dating, as to when such huge images were popular, even a reliable reference of Xuanzang to them is no more than the terminus ante quem. More exactly, his date of departure from Changan has been much disputed, but in the light of the fact that Xuanzang met the Ton-Yabghu Khaqan of the West Turks just before his assassination in between February and September, 628, Xuanzang possibly started before January 628. This date leads us to a more accurate chronology of his extensive travels: he visited Bamiyan in the beginning of 629(Kuwayama 1983). Because of unavailability of clues, schlolars, Foucher in particular, called to mind rather colossal seated figures of the Buddha at the Yungang caves in northern China as either the eastern counterparts or the eastern successors of the Bamiyan images. If the Yungang Buddhas were most probably hewn out in or just after 460 (Mizuno and Nagahiro 1956: 13-17), the western 'seniors' can be dated as before the middle of the fifth century (Godard, A., Y. Godard and]. Hackin,]. 1928: 12; Rowland 1942: 230-231). The appearance of the colossi should be discussed in its socio-religious or politico-religious circumstances of a certain age. The problem is more the Buddha, not as a speculated one in mind, but as a visible and furthermore, gigantic appearance that surpasses beyond the scope of human imaginativeness, and does not seem to be so simple as to be explained solely in the expansion of ideas. 1. Jibin as a Terminal of the Route taken by Indian and Chinese Monks The itineraries of Buddhist missionaries from both India and China may help lead us to a conclusion despite more indirect approaches. In the first three centuries, the Buddhist monks with the Central Asian surnames (Parthia, Samarkand, Yuezhi, and 140 Qizil),(l) eighteen in number, are twice as many as those of Indian origin. To the contrary, such Central Asian surnames suddenly decreased in number from the fourth century onward, in striking contrast to the increase of Indian monks. In the same time a great number of Chinese pilgrims reached India. Before the fourth century there was no one on the way to India at all. This trends imply that the Buddhist interchanges between China and India had burst upon the scene during the fourth-fifth centuries. Such interchanges could not exist without the kingdom of Jibin which had a more crucial situation than any other parts of the Indian subcontinent. How prominently Jibin played its role is described as follows. The earliest pilgrim who came from Jibin is Futucheng, or Buddhadana as restored by Pelliot (1903: 100; 1912: 419; Wright 1948: 321-371; The Jinshu 95 and T 50: 383b-387a). According to the biographer he had twice been to Jibin as a monk from the Western Regions in order to study the Buddhism and eventually became well known throughout the Western Regions. Called the enlightened he reached Luoyang in the 4th year of the Yunchia era or in 310. Chronologically speaking, Futucheng is followed by Kumarajiva, a central figure in the Chinese history of translation of the scriptures, born in Qizil as a son of an Indian immigrant who married a sister of the king of Qizil (T 50: 330-333a; T 55: 100a-102a; The Jinshu 95). At the age of nine, he went to Jibin with the mother, crossing the Indus river. There he was an eminent disciple of Bandhudatta, a younger cousin of the Jibin king. Kumarajiva was famous for the high intelligence in Buddhist Jibin, entertained by the king with special hospitality given only to the most honorable priests. Several years later he returned to Qizil crossing the mountains to the north ofYuezhi (Yuezhi Beishan) and proceeding to modern Kashgar and Aqsu areas. On the advent of Lugang he was arrested in Qizil in the eighteenth year of the Jianyuan era (382), and then brought to Changan under Yaoxing' rule of Later Qin. Buddhayasas (T 50: 333c-334b; T 55: 102a-c), of a Brahman family of Jibin, arrived in Kashgar where he was paid reverence by a prince and met Kumarajiva. Ten years later he proceeded to Qizil where he heard of the seizure of Kumarajiva. Buddhayasas hurried to Liangzhou in search of Kumarajiva. Kumarajiva had succeeded in inviting Buddhayasas to Changan to make more complete translations of Buddhist texts. Some kinds of the vinaya texts were translated by Buddhayasas in between 410 and 413. Buddhabhadra (T 50: 334b-335c: T: 55: 103b-104a), a native of Nagarahara which appears in the biography as Naheli, had studied together with his friend Sanghadatta many years in Jibin, where he met a monk from Langzhou named Zhiyan, who had three years resided there to learn the dharma under Buddhasena. With pious intension to make the true laws of the Buddha prevail in China, Zhiyan requested Buddhabhadra to accompany him to Changan. The biographer records that they proceeded to China beyond the Onion Range (Conglin), six countries, deserts, and precipices. Buddhabhadra also met Kumarajiva in Changan (T 50: 339a-c; T 55: 112b-113a). The Chu Sanzangjiji says that Zhiyan is not known of his homeland. Dharmaraksha, a native of Madhyadesa, visited Jibin carrying with him the Mahaparinirvanasutra which had been given to him by his master elsewhere in India (T 141 50: 335c-337a; T 55: 102c-l03b). Among the pragraphs of his biography it is very interesting to find that most of the monks in Jibin did not like to study the Mahayana, since it was a country of learning the Hinayana Buddhism. The unpopularity in Jibin of the Mahayanic thoughts even in such later years as the fouth century is very stimulative. Anyway, for this reason, Dharmaraksha was forced to travel further to Qizil and then to Liangzhou. In Liangzhou he was received with honor by Juqu Mengsun, a nomadicNorthern Liang king. By the king's order he started to make a Chinese version of the first one-third of the Mahaparinirvana sutra. However, he returned to his homeland in the course of the work, and on the way back to China, he found the middle one-third of this sutra in Khotan. Later an envoy despatched by the king in search of the Buddhist texts could get the final part of the sutra in Khotan. Dharmaraksha was able to complete the edition of the translated Mahaparinirvana sutra with the assistance of local monks in 421. Bhimaraksha (T 50: 333b-c), a native ofJibin, had long resided in Qizil, famous for his deep understanding of the vinaya which Kumarajiva had learned from him. As Buddhayasas did, he also went after Kumarajiva into Changan in 406. Another of the highest priests of Jibin was Dharmamitra. Making his mind in his youth to work as a propagandist (T 50: 342c-343a; T 55: 104a-b), he eventually reached Liangzhou, travelling through many countries including Qizil and Dunhuang. Among those who were of Jibin and came to China by land were Sanghabhatta (T, 50, 328a-b.), Sanghadeva (T, 50, 328c-329a.), and Punyatara (.T, 50, 333a.). Sanghabhatta came to China in 381. Sanghadeva arrived at Changan during the Jianyuan era or in between 365 and 384. At the request of Huiyuan he made the Chinese version of the Abhidharma texts during the Taiyuan era (between 376 and 396), and later translated one of the Agama texts with assistance of Sangharaksha who was also of Jibin origin. A brief survey of the first three books of the Gaoseng zhuan informs that most of the eminent translators of Buddhist texts in China were closely related to Jibin: they studied the Buddhist philosophy there, whether they may have been natives of Jibin or not. It is also likely that the route leading to Kashgar and farther to Qizil was usually taken by the monks who intended to proceed from Jibin to China. It is also reasonable to suppose that the route between Jibin and Kashgar was through one of the most precipitous mountain areas lying at the junction connecting both ranges, the Hindukush to the west and the Karakorum to the east. 2. Jibin, i.e., Gandhara The focus of many disputations since the last century is the whereabouts of Jibin, which is usually believed to have roamed about from place to place as geographical information might change in China, however it may confine to the south of the Onion Range (the Pamirs). Shiratori, agreeing with Levi who identified Jibin as Kapisi in the Tang time, further claimed that it had been the denominations of the Kabul valleys including Gandhara in the Han times, of Kapisi in both Sui and Tang periods, and of 142 Kasmir in between the Han and the Sui dynasties (Levi 1895: 371-384; 1986, 161-162; 1897,521, fn. 2; Shiratori 1917). In more recent time Petech devoted an exhaustive study to the subject (Petech 1950: 63-80), claiming that in the dynastic histories from the first century BC to the end of the fifth century AD it indicated the Indian territories of the great political power of the Northwest, whatever it was at the time of the writings (the Sakas, the Kushanas, the Hephthalites). In the Buddhist tradition, he also says, Jibin is Kasmir from the beginning (second century AD) till the time of Xuanzang. His solution was, however, sharply attacked from philological points of view by Pulleyblank, who maintains the traditional identification with Kasmira (Pulleyblank 1963: 58-144, 206-265) We cannot but hesitate to approve the above views, especially when Petech says 'in Buddhist tradition from the beginning till the time of Xuanzang' and Shiratori fixes Jibin to Kasmir throughout the periods of the Six Dynasties. The location of the Buddha's alms bowl leads to a better understanding that Jibin at least in the Gaoseng zhuan is assignable to Gandhara and that the fourth-fifth century Jibin is Gandhara. First of all, attention should be drawn to what the Da Tang Xiyu ji tells about. When two merchants intended to offer to the Buddha a meal such as roasted barley dough mixed with honey, Four Guardian Deities came to present identical bowls. At first each deity offered the bowl made of precious metal such as gold, silver, crystal, lapis, and so on, but the Buddha refused to receive all of them. When they finally offered brilliant ultramarine-coloured stone bowls, he received and fused four bowls into one. As a result the rim of the bowl clearly showed the trace of the fourfold composition (Beal 1884: 129-130; Rosenfield 1967: 222-223). Following the story, Xuanzang tells about a stupa in the vicinity of Buddhagaya commemorating this miracle. What is interesting here is the fact that Xuanzang never recorded anything about edifices enshrining the bowl itself. In the Fufazang Yinyuan zhuan, which Maspero rightly suggests to be not the translation of an original version but a work edited on various sources from a dynasty to another in China (l\1aspero 1911;130f.), Kanishka took away the Buddha's alms bowl together with Asvaghosha as well as a merciful cock from Pataliputra when he besieged the city (T 50: 314b). In the Biop-aphy of the Bodhisattva Afvaghosha is found a slightly modified version of the similar story which Kumarajiva transposed into Chinese in the fourth century. The Biography says that a king of the Little Yuezhi of North India had carried away both the bowl and ASvaghosha with him as the booty substituting for two hundred millions of gold (f 50: 183c). Whoever Kanishka or a king of the Little Yuezhi may have been, the main point of this story is that the bowl was taken away from the religiously proper location to the Northwest, or a frontier. The bowl was actually seen by Faxian at Purushapura in Gandhara. As the Faxian zhuan describes (f 51; 858b): The Buddha's alms bowl is in this country. Formerly a king ofYuezhi raised a large force and invaded this country, wishing to carry the bowl away. Having subdued the kingdom, as he and his generals were sincere believers in the Law of the Buddha, and wished to carry off the bowl, they proceeded to present their offerings on a great scale. When they had done so to the 1+3 Three Treasures, he made a large elephant grandly caparisoned, and placed the bowl upon it. But the elephant knelt down on the ground, and was unable to go forward. The king knew that the time for an association between himself and the bowl had not yet arrived, and was sad and deeply ashamed of himself. Forthwith he built a stupa at the place and a monastery, and left a guard to watch the bowl, making all sorts of contributions. There may be there more than seven hundred monks. When it is near mid day, they bring out the bowl, and along with the common people, make their various offerings to it, after which they take their midday meal. In the evening, at the time of incense, they bring out the bowl again. It may contain rather more than two pecks and is of various colours, black predominating, with the seams that show its fourfold composition distinctively marked. Its thickness is about the fifth of an inch, and it has a bright and glossy lustre. (Legge 1886: 34-35). The bowl which Faxian saw had been formed exactly as in the story. He further says that Huiying, one of his attendants, died in this monastery.(2) Faxian saw it in Purushapura in the 2nd year of the Yuanxing era of the EastJin dynasty (403). Just after him, Zhimeng also worshipped in prayer the bowl in Jibin. Leaving Changan in the sixth year of the Hongshi era of the Later Qin Dynasty (404), the year in which Kumarajiva entered into Changan, after climbing the Onion Ranges to the southwest of Khotan (where he arrived presumably from Qizil, crossing the Taklamakan desert). Zhimeng proceeded to Bolun through Tashkurghan where he saw the Buddha's spitting vase of decorated stone (T 50: 343b-c; T 55: 113c-114a; Uchida 1967: 19-34). From Bolun he crossed the Snow Mountains and passed over the Sindhunadi to arrive in Jibin. The bowl he adored there had a bright lustre of purplish dark blue color with the distinctive fourfold composition. The color and clear mark of the rim of the bowl is decisively identical with those of the object that Faxian saw in Purushapura. I do not agree, therefore, with those still sticking to the traditional solution that Jibin is Kasmir. Could Faxian and Zhimeng respectively see the same object in Purushapura and in Kasmir at about the same time? (3). Moreover, both itineraries of Faxian and Zhimeng seem to be almost identical. Zhimeng left Changan, via Liangzhou, Yangguan, Miran, Qizil, and Khotan to Tashkurghan, and reached Jibin, crossing the Snow Mountains and the Indus after departing from Boluluo. He extended his itinerary westward up to Nagarahara. Faxian, departing from Jiecha which is equivalent with Jisha on the route of Zhimeng, identical with Tashkurghan, crossed over the Pamirs to enter Tuoli (Darel), the first country he reached in the northern lands of India as his biographer says. Then he went onward to the southwest for fifteen days along the foot of the mountains as he is described: The way was difficult and rugged, running along a bank exceedingly precipitous, which rose up there, a hill-like wall of rock, ten-thousand cubits from the base. When one approached the edge of it, his eyes became unsteady; and if he wished to go forward in the same direction, there was no place on which Iff he could place his foot; and beneath were the waters of the river called the Indus .... After crossing the river, the travellers came to the kingdom of Wuchang (Uddiyana) .... Faxian and others performing the summer retreat, they descended south, and arrived in the country of Suheduo (tentatively identifiable with modern Buner, although this place name is phonetically close to Swat) .... and going downwards from here towards the east, they came to the country of Gandhara in five days (Legge 1886: 26-31) On the other hand, Xuanzang visited Darel, following the reverse course. He went on from the town Mengjieli (tentatively identified with modern Mingora) to the northeast direction, crossing mountains and valleys and then again ascending the Indus. He also saw a wooden figure of Maitreya standing beside a large sangharama in the Darel valleys. Faxian explains the image as connected to the northward expansion of the Buddhism. Going eastward from Darel, he went up the course of the Indus. Then, after going 500 Chinese miles or so by the help of flying bridges and footways made of wood laid across the chasms and precipices, he arrived in the country of Boluluo, which is equivalent with Bolun on the route of Zhimeng, or either Bolor or Palur (BeaI1884: 133f.). Faxian and Xuanzang took the same route. In modern geography the route runs along the Swat river from Mingora to Khwaja Khela from which turning to the east and crossing the mountains Shang is reached in the Indus valley. The routes of both priests coincide by ascending along the Indus from Shang. From natural topography that the Indus changes its course from the westward to the southwestward, Tuoli (Darel) seems to have been located just east of this winding, or around Chilas where a number of petrographies and inscriptions were discovered by the Pak-German team headed by K. Jettmar. Dharmavikrama also took a look of the bowl in Jibin 50: 338b-339a). He left China in 420. Huijiao, the biographer of Dharmavikrama, records his experience between the Snow Mountains and Jibin as follows: cr Beneath the foot there is a large river flowing rapidly like arrows. Between both sides of the east and west mountains, there is a bridge made of ropes of twisted twigs which connects each side of the mountains. When crossing, one must make a group of ten persons. Then, after crossing, they light a signal fire, and the followers, seeing it, know of the predecessors' safe arrival to eventually cross the bridge. In case that no fire is seen for long time, they understand that a strong wind blew them down to the river, or what happened to those who have tried to cross. Going thus for three days, D harmavikrama passed over the Great Snow Mountains. The cliffs stand like a wall, and so there is no place to step on. On the wall are hewn out pairs of old cavities for inserting pegs, using four of which one can climb up. Dharmavikrama passed there in several days of difficulties and at last reached Jibin. He worshipped in prayer the Buddha's alms bowl there and learned spoken and written local Indian language during a year's stay. Then he went westwards to the Sindhu River and along the west of it he entered into the kingdom ofYuezhi, where 145 he made prayer to the u~ni~a bone of the Buddha. Dharmavikrama's way to Jibin and the kingdom ofYuezhi is same with Faxian's. In this case the kingdom of Yuezhi is Nagarahara, since the ushnisha bone of the Buddha existed in a temple of Nagarahara at the time of Faxian. Jibin of Dharmavikrama therefore is also identical with that of Zhimeng, that is, Gandhara. Hui-Ian (T 50: 339a) also was in Jibin where he adored the bowl, too. On his way back to China he was esteemed by the nomadic chieftain of the Tuyuhun kingdom, and died in Changan at his age of about sixty in the Daiming era (457-464). All those such as Faxian, Zhimeng, Dharmavikrama, and Huilan saw the bowl during the limited times pan of the earlier half of the fifth century; Faxian in about 403 in Purushapura, Zhimeng a little later than 404 in Jibin, and Dharmavikrama as well as Huilan several years later than 420 in Jibin. Furthermore, the way thither from north seems to have been shared by all of the pilgrims. Insofar as the Buddhist sources in the fourth-fifth centuries are concerned, Jibin is no other than Gandhara. Now the location of Jibin, thus interpreted, makes it clear that pilgrims took the route through the western fringe of the Karakorum, or through Bolor, Darel, and Uddiyana. When they came from either Kashgar or Yarkand, they firstly ascended to Tashkurghan, then proceeded to the passes such as Naizatash and Ishkoman or Darkot, and reached Bolor. Chinese pilgrims in Gandhara seem to have usually extended their steps farther westwards to Nagarahara, as Faxian and Dharmavikrama did. Nagarahara was well known not only for the sacred relics and the remains of the Buddha but also the places where the Dipankara Buddha appeared. The remains of the Buddha had been established on the base of the legends concerning the unbelievable visit of Buddha to the Nortllwest. Of various remains in Nagarahara, one of the religious attractions was the sacred image in a cavern located to the southwest of the capital called Denggang, or Dipankara. The detailed story is found in the Fushuo Guanfu Sanmeihai jing which is generally believed to be one of the s translated by Buddhabhadra, a native of Nagarahara.(4) In China, moreover, even a priest who had never been to Nagarahara admired the Buddha's shadow in Nagarahara: Huiyuan had that image of Buddha painted on the wall of his abode in the mountain call Lushan in 412 and even wrote a text to be inscribed on the stone stele so that the pious reproduction of the image was commemorated.(5) Nagarahara, just to the west of Gandhara, was, in fact, spiritually very close to Buddhist monks in China, whether or not one may have actually visited (Illus. 17, above). It is very natural for Chinese Buddhists to proceed southeastwrd to the Gangetic plain after visiting Gandhara. Yet especially important is that they went all the way in order to see Nagarahara and then return to the east, or the Indian subcontinent. Faxian crossed the Spin Ghar range to reach Bannu and then Mathura; Song Yun also visited N agarahara coming from Gandhara, not from the west. For Chinese Buddhists, N agarahara was the western ternimal of Buddhist India. Chinese Buddhists had not known about the countries farther west of it until the middle of the sixth century. This geographical and chronological frameworks seem suggestive of dating the Bamiyan Buddhas. 146 3. The Route of Song Yun and its Historical Significance According to the fifth volume of the Luoyang Qielan ji Song Yun entered into the territory of Kharbandan, or modern Tashkurghan area in the beginning of the eighth month of the second year of the Shengui era (519). Then the Luoyang Qielan ji records that he visited Bohuo (Wa'khan) before reaching the Hephthalite kingdom. In the beginning of the tenth month or in the first decade of the tenth month (probably in 519), he arrived in the kingdom of Heda (the Hephthalites). Following the description of the Hephthalites, Yan Xuanzhi, the editor of the Luoyang Qielan ji, tells that Song Yun entered into the kingdom of Bozhi in the beginning of the eleventh month and reached Shemi (Chitral according to Chavannes or Kafiristan in the upper Kunar according to Stein(6) in the middle of the eleventh month. Then he entered into Uddiyana in the beginning of the twelveth month. Which way Song Yun took from Shemi to Uddiyana is not described, but the follwing paragraph does not seem to indicate that Song Yun visited Bolue on the way (Illus. 16, above). The steep and precipitous path is dangerous for passing, and only a man and horse can barely pass; a path penetrates from Bolule (Bolor) to Uddiyana; the bridge is made of iron chains and, suspended in the air, it forms a passage; one cannot see the bottom of the valley beneath his foot and there is nothing to catch beside him ... In the beginning of the twelfth month he entered into the kingdom ofUddiyana. The Account of the Western Regions in the Weishu is believed to have been in part compiled after the information given by Song Yun. The Account says that Shemi is situated in the mountains to the south of Bozhi, and the Buddhist religion is not believed there but devotion is solely given to deities of paganism. The Account also notes that the kingdom of Shemi belongs to the Hephthalite rule, to the east of this kingdom being Bolule, and that the path between Shemi and Bolule is steep and passable only with the help of iron chains, beneath which the deep bottom of the valley is invisible. This paragraph has no difference from the description of the Luoyang Qiela ji, but only the Account includes what the Luoyang Qiela ji does not contain: During the Xiping era (516-17) Song Yun and others could not arrive in Bolule. It is furthermore important that the similar sentence is found in Vol. 186 of the Taiping Huanyunji. If the Taiping Huanyun ji quotes it from the original narrative of Song Yun, Song Yun would not have visited Bolule. Song Yun proceeded directly to Uddiyana. His route from Shemi must have passed through modern Dir, whether Shemi may be Chitral or the upper reaches of Kunar. From Uddiyana he entered into Gandhara in the second decade of the fourth month of the first year of the Zhengguang era (520). Following the description of Gandhara and its related matters, the editor of the Luoyang Qielan ji then includes Nagarahara quoting from a record of travel of Daorong, not of Song Yun. Chavannes 147 therefore doubts about whether Song Yun had actually visited it(7). In an edition of the Luoyang Qielan ji, however, the character rong of Daorong is replaced by yao. In fact, Daoyao appears in the fifth section of the second book of the Shijia Fangzhi as a sramana who "penetrated from the Sule (Kashgar) route through the Hanging Passage to the kingdom of Sengqieshi (Sankasya) in the last year of the Later Wei emperor Taiwu's reign, or in 451, and, on returning to China, again took the same itinerary as he had used before"(Uchida 1961: 113-116). The same source book also explains that the record of his travel, once circulated, comprises one book. The editor of the Luoyang Qielan ji alludes in his postscript to the information he was able to employ as follows: The narrative of the travel of Huisheng is incomplete in many points. So the narrative of Daoyao and the personal memoirs of Song Yun had to be consulted. Accordingly, all the sources used in order to fill up the blanks of texts are clearly mentioned here. The sources he employed to make his edition more accurate seem to have first of all been based on the narrative Huisheng, who had accompanied Song Yun searching for the Buddhist scriptures and making prayer of the sacred places by the imperial order of Empress dowager Hu. The editor supplemented the incompleteness of Huisheng's memoirs with those of both Song Yun and Daoyao. Depending on such sources, he intended to cover all of their travels as completely as he could. Presumably Song Yun really visited Nagarahara. The Hephthalite headquarters was in the middle reaches of the Surkhab, not in Badakhshan. When they entered into that country, it was in the beginning or in the first decade of the tenth month, that is in early winter or in the end of autumn of our calendar. In such seasons the nomadic people was accustomed to return from the summer pastures in the mountainous regions down to their winter settlements in the plain. In Tokharistan, the summer pasture is/was located in the areas extending beyond Faizabad in Badakhshan. The winter residences were in the middle and lower courses of the rivers: the Surkhab and the Kunduzab. The Hephthalite headquarters that Song Yun visited is in the plain, since the account of Song Yun clearly informs us that "The cultivable lands are extensively numerous; mountains and marshes are as far as one can see". The scenery is no longer in the mountains. The views that the Hephthalite headquarters was in Badakhshan is untenable (Enoki 1951: 137-139; 1955: 231-236). Chavannes identifies the kingdom of Bozhi, which was on the way between the Hephthalites and Shemi, with the valley of Darya-e Sanglich around Zebak (Illus. 16, above). This is quite agreeable since Bozhi is situated to the north of Shemi according to the editor of Song Yun's itinerary lying to the south of Buhuo or Wakh'an according to the Weishu. The areas covered by Song Yun in company of Huisheng in the northern and southeastern slopes of the Hindukush. They were obviously charged in both diplomatic and Buddhist mission to the Hephthalites and Uddiyana respectively. Thinking of their 148 audience with the Hephthalite king in Tokharistan and its tegin in Gandhara, Song Yun and Huisheng possibly travelled under the Hephthalite protection through the countries in Tokharistan down to Gandhara. The areas covered by their travelling indicate the Hephthalite territories. If so, significant is that they did not cross the Hndukush to Bamiyan and Kapisi from the middle course of the Surkhab, where the Hephthalite headquarters was. Bamiyan had never been controlled by the Hephthalites (8). Even in the Account of the Western Regions in the Weishu, there cannot be found any evidence for the routes leading to Bamiyan and Kapisi from the northern slope of the Hindukush. However, Pelliot believes that the name Fanyang in the Beishi could be a transcription of Bamiyan, noting that "au lieu de Fa-yang de Pei-che (compose de fan, nom de famille, et de yang, principe lumineux), on a ici Fan-yen ecrit avec fan, voile de navire, et yen, etendre; la valeur de transcription est bam + yan". In the Beishi or in the existing Weishu, however, the name Fanyang is used as indicating the eastern limit ofTokharistan. Bamiyan is situated outside of the territory ofTokharistan, as Xuanzang entered into the territory of Bamiyan after leaving the Tokhara countries (Beal 1911: 52. Julien 1853: 68). The name Fanyang-zhou appears in Volume 73 of the Tang Huiyao as one of the twenty-four Divisions of Tokharistan under the Tang hegemony. The headquarters of this Division is the town Bateshan, and Bateshan also appears in the narrative of Huichao's travels in India as Puteshan (Fuchs 1939: 449 and 466), which is located about one month march to the east of the capital town of Tokharistan (Qal'a-e Zal). It is very clear that Bateshan or Puteshan is identical with Badakhshan. Hence the idea of Pelliot is right in his restitution, but Fanyang in the Beishi is no other than a denomination during the Tang period of some town in Badakhshan. Therefore, as regards Bamiyan, no positive evidence exists for the real information in the Northern Wei dynasty. 4. Change of the Route Crossing the Hindukush From the fourth century to the second decade of the sixth century, Buddhist monks, travelling from China or from India (actually Gandhara), shared a common Karakorum route which starts Gandhara for Tashkurghan through Swat, Dir, Gilgit and passes such as Mintaka, Kirik, Ishkoman, and Borogir (Illus. 17, above). After Song Yun, however, both pilgrimage and its routes saw a drastic change. Chinese pilgrims disappeared from sight, while Buddhist Indian propagandists suddenly became fewer: Indian monks having arrived in China during the sixth century were only fifteen in number. A new highway replaced the Karakorum route, running, beyond Nagarahara, through Kapisi and Bamiyan to Tokharistan. Jinagupta is the earliest traveller known so far on this road in between 554 and 557 and Dharmagupta also took it a little later in the seventies of the sixth century, as their itineraries are shown in each biography edited by Daoxuan in the Tang Gaoseng zhuan. Jinagupta (Shenajueduo) was born in 528 in PurushapuraChavannes 1905: 332, fn. 1). He entered the monkhood there under the guidance of Jiianayasas in a vihara called Great Forest (Mahavana). Then he went around India for making prayer to them in 149 company with his master, since the divine vestiges still remained in the sacred lands of India. In 554, at the age of twenty seven, he could observe the complete precepts and stayed in summer retreat for three months. His biographer says about his travels as follows: The master and his disciples then contracted the resolution of voyage for propagating the Law. The mission, consisting of ten monks at their departure from Gandhara, left for Kapisi. Their stay covers about one year. The king of Kapisi very earnestly requested his master to be the head of a state ritual with very extensive advantages. Regarding the itinerary they took after Kapisi, the editor of his biography, Daoxuan, gave the following accounts: Pursuing their voyage furthermore, they went over the western foot of the Great Snow Mountains (Daxueshan xizu) which certainly is the most precipitous in the world. They arrived in the kingdom of the Hephthalites where they stopped [longer] for the first time. The land is extensive but inhabitants are scarce. There are none who prepare drinking and eating, so he dared to take his robes off in order to do his best for the sake of his master. Although he suffered troubles caused by current political affairs, he luckily kept out of harm's way by the protection of supernatural power. He then traversed the kingdoms such as T ashkurghan and Khotan. Since he often endured chilly weather such as rains in summer and icy snowfalls, he had to stay in such kingdoms a little while. Yet he did not stay long because of little propagation of Buddhism. He then reached the kingdom of Tuyuhun (the capital was fifteen Ii west of Kuku Nor) and Shanzhou (Xining) in the first year that succeeds the fall of the W estern Wei dynasty (557) .(9) He had suffered much trouble and peril throughout his travels. Nevertheless his heart was still more energetic. When he arrived there, three years had passed since he left Gandhara. More than half of the members had passed away on the road. Still alive are the four who were able to arrive. In the years of the W udu era under the rule of Emperor Ming of the Zhou Dynasty (559- 560), Jinagupta first arrived in Changan. He stayed there in the temple Zaotang.... As stated above, Jinagupta arrived in Xining in 557 after three years' journey from Purushapura. Since we know he observed the precepts in 554, his departure seems to have been just after his summer retreat. Certainly he was in Kapisi and Bamiyan between 554 and 557, or more possibly in 555. According to the same biographies of the eminent monks, Dharmagupta, born in Lata (10) as an eldest son of the Kshatriya family, went to Kanyakubja (Kanauj) at his age of twenty three in order to be associated with monks' community. After two years' hard works he received the complete precepts under Samantadeva, a upadhyaya, in the Yellow ISO Flower sangharama. Then Dharmagupta accompanied his master on the latter's leaving for Takkadesa at the request of the king and stayed there for one year. After his master returned to Kanyakubja, he remained in Deva-vihara four more years. Having stayed in various countries and viharas, Mahayanic and Hinayanic, he gained twice more knowledge than he had in Kanyakubja. Merchants frequently came to Takkadesa from the northern routes and informed him about the great kingdom of China in the far east. He could not believe them from the bottom of his heart when he first heard it. However, since no specific place to go for the propagation was conceived despite his earnest will, he eventually proceeded to the kingdom of Kapisi together with six attendants, and stayed in a temple built by the king (wangsi).(ll) Then he moved into the city (modern Begram) to reside two years and more with four of the attendants. In Kapisi he learned Buddhism in various temples he visited. Nevertheless, he was still not well satisfied holding the will of further propagating the dharma. As the biographer says that the kingdom of Kapisi was at the junction of roads running from north, and therefore merchants and merchandise had flowed into the territory of Kapisi from the north of the Hindukush. At the places where such merchants stayed, Dharmagupta again heard about the great kingdom of China where the Three Treasures were flourishing. So he and his followers wanted to see China with all their hearts. They not only wished to see sights and people, but also intended to propagate Buddhism. Hereby they crossed the "western foot" of the Snow Mountains, Bochuluo (Bakhla), Boduochana (Badakhshan), and Damoxibinduo (Wa'khan) (121, and reached Tashkurghan where they stayed one year. Dharmagupta eventually arrived in the city Daxing, the capital of the Sui dynasty, in the 10th month in winter of the Kaihuang era or in 590. When was Dharmagupta on the road from Takkadesa to China? The biographer does not refer to how old Dharmagupta passed away in the 2nd year of the Wude era of the Tang dynasty, in 619. Instead he notes how long D harmagupta stayed in each main kingdom: two years in Kapisi, one year in Tashkurghan, two years in Qizil, two years in Kara Shahr, two years in Turfan and one year in Hami. The total duration amounts to ten years. Taking into consideration the time he spent on the way, Dharmagupta left Takkadesa in the late seventies of the sixth century. This chronology well explains the reason why no mention is made of the Hephthalites in his travels. By the second invasion of the Turkish chieftain, Sinjibu Khaqan or Siljibulos they utterly became unresisted in 567 (Zhoushu 50; Uchida 1975: 436-437). Dharmagupta saw the Hephthalites politically weak when he arrived in Tokharistan in the eighties of the sixth century. Dharmagupta and Jinagupta made way from Kapisi to Tokharistan tcrossing the "western foot" of the snow mountains. Th snow mountains clearly is the Hindukush. A clue to locate these words more accurately is found in the Xiyu Tuji (Illustrated Account of the Western Regions) which is edited by Pei Ju, a Sui diplomat. According to his biography in the Suishu (VoL 67), he was sent to Zhangye in Kansu in order to controll the long-distance traders from the west. The Xiyu Tuji is a record of informaion of the western countries thus collected by him at first hand. The book was dedicated by him to the emperor Yangti in the first half of 606 (Uchida 1973), but lost in the early Tang period except for author's original preface which was fortunately quoted in his biography 151 of the Suishu. The three routes, such as the northern, middle and southern, led from Dunhuang to the western sea are especially explained in the preface. The southern route is a way to India which is entered through the Northwest. The western half of this route, i.e., from the Pamirs westward, perfectly corresponds to the roads taken by Dharmagupta and Jinagupta. It runs to the Northern Brahman country through Tokhara, the Hephthalites, Bamiyan and Kapi§i (Kuwayama 1975: 99-101). A comparison of this itinearary with that taken by Dharmagupta and Jinagupta leads us to an understanding that the "western foot" of the snow mountains is Bamiyan. The route through the western Karakorum seems unpassable sometime after Song Yun returned to China, and the highway connecting India to Central Asia drastically shifted from the Karakorum area to the west. On this new road emerged Kapi§i and Bamiyan. The use of this route, which may have been a detour for pilgrims, was not accidental (Illus. 17, below). First of all, no mention is made of the Gandhara-Bolor-Tashkurghan route in Pei Ju's preface dated to 606. The Northern Route referred to by Pei Ju is from Dunhuang to the western sea via Hami, Barkul Nor, the steppe areas north of the Tian-shan where the headquarters of the West Turks were, the extensive areas with various rivers running northwards, and the Byzantine empire. The Middle Route runs through Turfan, Kara Shahr, Qizil, Kashgar, the Pamirs, Ferghana, and various fortified cities in Sogdiana, to the Sassanian Persia. Secondly, the Bolor-Tashkurghan route was usual in the fifth century when one entered India from the north.The general preface of the Account of the Western Barbarians in the Tongdian (Vol. 191) refers to the four routes which lead to the Western Region: (1) 2200 Chinese miles to the west of Yumen Oade Gate) and the Flowing Sands, one reaches Shangshang; (2) 2200 Chinese miles to the north of Yumen, crossing over the Flowing Sands, one reaches Cheshi (the areas lying across the Bogd-ola); (3) 1000 Chinese miles to the west of Yarkand, one reaches the Pamirs, and more 1300 Chinese miles to the west of it, one reaches Wa'khan; (4) 500 Chinese miles to the south-west ofYarkand, one reaches the Pamirs, and more 1300 Chinese miles to the southwest of it, one reaches Bolor. The passages from Yarkand though the Pamirs to Bolor and Wa 'khan are same as those actually taken by Song Yun. The editor of the To ngdian, Du Yu, quoted these routes as reported by Dongyan, who had been sent in 437 by the Later Wei Emperor as a diplomat to the Western Region (13). Such routes were used in the fifth century. Accordingly no evidence supports the existence of the route running in the western Karakorum after the mid-sixth century. Thirdly, Pei Ju's account of the three routes shows that the route through the western foot of the Hindukush was one and the only highway connecting Central Asia to India around 606. About twenty years after Pei Ju's edition of the Xiyu Tuji, Xuanzang also took the same way as the western half of the Southern Route from Tokharistan to India through Bamiyan, Kapi§i Laghman, and Nagarahara. Lastly, Xuanzhao, a monk-pilgrim, does not seem to have taken a different route from Xuanzang in the middle of the seventh century. As Yijing stated in the first volume 152 of the Da Tang Xiyu Qiufa Gaoseng zhuan (Great Tang Biographies of Eminent Monks Being in Search of the Law in India), Xuanzhao could not find any way back to China on his second travel to India. His biography writes the reason why he could not, saying that it happened that the Tibetans blocked up a way through Nepal, while the "Kapisi road" was disturbed by the Arabian invasions. The "Kapisi road" is a way to Central Asia through Kapisi from India. This statement implies that there were only two ways en route which could connect India with China in the mid-seventh century (T 51: 1c-2a). 5. The Rise of Bamiyan and the Decline of Gandhara Located at the southern foot of the western fringe of the Karakorum, Gandhara was a place, where long distance traders from the north unloaded goods and for those heading for Central Asia to prepare for crossing the mountainous area. From here the route run for Central Asia through Swat, Gilgit or Chitral, not through the regions to the west of Gandhara. From the middle of the sixth century onward Kapisi took the place of Gandhara located just at the starting point for the ascent of the Hindukush. The route changed from that leading to the north but to the one that proceeds westward. The cause of this change reflects the historical background that happened around the Hindukush. According to Song Yun and Huijiao, Gandhara was overthrown by a chieftain of the Hephthalites, who appointed a tegin from among them to the king of Gandhara; by the time of Song Yun two or three reigns of tegins had passed; but by his brutal character the present tegin pursuesd massacre not believing in the Buddhism but devoting himself to heathen gods. He had contested the territory against Kasmira for three years, and this had caused trouble to the land and people of Gandhara.o 4) Attention should be drawn to Shahbaz garhi (Chavannes 1903A: 419, fn. 5). There Song Yun was only admiring the magnificent Buddhist temples and images, never referring to any destruction of them by the hands of the Hephthalite tegin. About a century later, however, in Gandhara at the time of Xuanzang, the royal family had already been extinct, the Kapisi deputies governed few of the inhabitants in deserted towns and villages, most of them belonging to heretical schools and few believing in the Right Law. There were one thousand Buddhist monasteries, which had been deserted in ruins, filled with wild shrubs, and solitary to the last degree. The stupas had mostly decayed, but the heretical temples, numbering about one hundred, were occupied without order by heretics. The desolation of Buddhism in Gandhara therefore chronologically falls in a time between visits of Song Yun and Xuanzang. The decline of the Hephthalites started in their homeland around 554 and suddenly accelarated with the first Turkish attack headed by Mugan Khaqan. The successive attacks resulted into Turkish hegemony over oases in Sogd and Ferghana such as Samarkand, Shash, and Kish (Marquart 1901: 64; Altheim 1969: 260; Haussig 1956: 23). The Turkish Sinjibu Khaqan and his ally Khusraw Anushirwan gave a last decisive blow in 558. The Suishu records that in the same year a Turkish Tunshad was dispatched to take over the Hephthalites and that the latter's defeat was too bitter to send their envoys to China which the Northern dynasties had received almost every year since 507.(15) It is just before their defeat, in 554, thatJinagupta 153 stayed in the Hephthalite country. The decline of Gandharan Buddhism was due partly to the Hephthalite disintegration. Without the political power no traffic security in the mountainous region was ensured. Contrary to Gandhara, Kapisi and Bamiyan entered the historical field on a new highway. Dharmagupta's biography says that Kapisi was a junction of the routes coming from the north. Xuanzang states that these people (in Bamiyan) are remarkable, among all their neighbours, for a heart of pure faith and that from the highest form of worship to the triratna, down to the worship of various deities there was not the least absence of earnestness and the utmost devotion of heart. As he further describes, the merchants, in arranging their prices as they come and go, fall in with the signs afforded by the deva deities. If good, they act accordingly; if evil, they seek to propitiate the belong to the powers. There are ten convents and about one thousand monks. Hinayanic school, and learn the Lokottaravadin's discipline (Bea11884: 50f). Thus the Indian wealth, no longer saved in Gandhara, pushed the way toward Kapisi and Bamiyan, or farther through the Hindukush to Tokharistan and even beyond the Amu. Together with merchants, Indian gods of popular beliefs travelled to such countries and cities. In remote countries beyond the original frontiers of Indian cultures, the images of Indian gods and goddesses occupied their seats among various deities and cults of Central Asian origins, as found in the recent decades in the excavations at Dilberdjin near Balkh, at Piandjikent near Samarkand, and so forth. All that is left of the deities on and beyond the Amu is types of the Sivaite images. At Dilberdjin is a composite Siva and Parvati image fig. 30). This is seated on the bull Nandin and painted on the wall (Kruglikova 1974: closely relevant in its composition to the marble Uma MaheSvara image found at Tapa Skandar to the north of Kabul (Kuwayama 1978A: 381-383; 1972B). In Piandjikent are found several fragments of painted deities that unavoidably presuppose the essential feature of Siva, as Belenitsky and Marshak rightly pointed out (Belenitsky and Marshak 1971: Sf. and figs. 3-5). According to Xuanzang, two standing and one lying images of Buddha between the two were in Bamiyan. On the declivity of a hill to the northeast of the royal city is a standing image of the Buddha made of stone one hundred fifty feet high. To the east of the image is a Buddhist monastery. To the east of the monastery is a standing image of the Buddha Sakyamuni made of brass which is one hundred feet high. In the monastery is a lying image of the Buddha in nirvana which is one thousand feet long. (BeaI1884: 50ff.) The location of each standing figure has been very clearly fixed. Between two standing colossi is a monastery which housed a huge lying Buddha. The Da Tang Xiyu ji also pointed out that to the east of the one hundred fifty feet Buddha figure is a monastery which was built by a xianwang. The xianwang has variously been interpreted. According to Beal and Watters it means a former king (Beal 1884: 51; Watters 1904-1905: 118). In the historical context, the word should be interpreted as the king before the present one 154 by whom the lying figure of huge dimension housed in this monastery must have been dedicated. Xuanzang came to Bamiyan not so long after such a cult image was completed. We have discussed the routes which connected India, or Gandhara, to China or Central Asia from the fourth century onward to find out a drastic change in the middle of the sixth century. From that time onward the earlier route was completely discarded, a new route was opened as a highway running beyond Nagarahara toward the west and crossing the Hindukush at Bamiyan. Supported by more and more frequent exchanges of wealth from both sides of the Hindukush, Bamiyan, and Kapisi, appeared from an earlier, rather isolated position in both geographical and hence economical contexts. In such circumstance the constructive activities such as building gigantic Buddha figures could have been embodied. It should be further discussed that in Bamiyan the nirvana Buddha images are closely related to those of Maitreya in the compositions of painting in some caves. I may suggest that the lying Buddha and Maitreya have heralded the real extinction of Buddhism in that area. The Bamiyan colossi emerged when Bamiyan became quite prosperous on the highway, that is, from the middle of the sixth century on.(16) 155 IX Bamiyan and its Buddhist Activities Bamiyan is a small town and the modern capital of the Bamiyan province in Afghanistan, located in a mountain valley, 2500 m. above the sea level, of the western Hindu Kush ranges and 236 km. to the west of Kabul by road. The Bamiyan valley is secluded by the Aq Robat pass on the north from Mghan Turkistan or mediaeval Tokharistan and by the Hajigak and Unai passes on the south from Ghazni and the Upper Argandab Valleys or ancient Zabulistan as well as the Shibar pass on the east from the Parawan-Kabul districts or Kapisi in old time. The geographical situation had historically given it a political independence at least before a Korean pilgrim Huichao in the second or third decade of the eighth century described that other countries were not brave enough to invade it. Nevertheless, the early history of Bamiyan is quite unclear, especially of the times before the sixth century, although some of the Indo-Greek and Kushan coins have reportedly been collected. The earliest information of Bamiyan is given in the Tang Gaoseng zhuan (the Tang Biographies of Eminent Monks, edited in the first half of the seventh century by Daoxuan) which says that the two Indian monks such as Jinagupta, of Gandhara origin, and Dharmagupta, native of West India, visited the 'western foot of snow mountains' (daxueshan xizu), e.g. Bamiyan, in 555 and 580s respectively on the way to Tokharistan from Kapisi. Bamiyan also appears in the preface of the Xiyu Tuji (edited by Pei Ju in 606) which says that it lies between the Hephthalite headquarters, e.g. Baghlan, and Kapisi, and in the Suishu which says that it is located 700 Chinese miles to the north of Kapisi and that a Bamiyan emissary was received by the second Sui emperor Yangti on the New Year's day in 615. The most detailed account of Bamiyan in the first volume of the Da Tang Xiyu ji, completed in 646 by Bianji who had been given by Xuanzang his whole records of the extensive travels). It tells that Bamiyan, with the capital resting on a steep hill, covers a district extending about 2000 Chinese miles from east to west and about 300 Chinese miles from north to south and that the letters, customary rules and monetary system are same as those of Tokharistan, the language being a little different. According to the Tangshu and the Jiu Tangshu, Bamiyan became a state politically controlled by the Chinese general headquarters in Qucha from the middle of the seventh century on, consisting of the five divisions with each central town such as Jianna, Silin, Fushifu and Weilasadanna which are governed from the state capital of Bamiyan Fuli (according to the Tangshu and the Taiping Huanyujt) or Luolan (according to the Jiu Tangshu and the Tang Huiyao), although whether such Tang rule really worked musht be checked. About seventy years later, Huichao described that in Bamiyan was a king of the Iranian stock who did not belong to any other countries with the aid of strong military forces, while 156 devoting himself to Buddhism. Contrary to the Da Tang Xiyu ji, which only tells the existence of two of the Hinayanic schools, the Lokottaravadins and the Mahasanghikas, Huichao admits both Mahayanic and Hinayanic Buddhists, and the customary rules similar to those in Kapisi, despite the language being isolated. The Muslim sources are rather eloquent for the events after the mid-eighth century. In the time of aI-Mansur (754-775), the second Abbasid Caliph, Bamiyan was subjugated to the Muslim power by Muzahim b. Bistam, the governer of Warwaliz, whose son, Abu Harb Muhammad, was ordered by his father to marry a daughter of Shir-e Bamiyan. In 7921793 his son aI-Hasan took part in a military campaign to Kabul under the leadership of al-Fadl b.Yahya b. al-Khalid b. Barmak, an Abbasid governer of Khorasan, and marched to Ghurwand. Al-Hasan was invested by al-Fadl to the king of Bamiyan and called the Shir-e Bamiyan after his grandfather. After probable hegemonies of the Tahirids and Samanids, Alptegin, according to the Siyasat Nama, fighted on the way to Ghazna against a Bamiyan amir called Shir Barik(?) and defeated him. During the Ghaznavid rule Bamiyan is politically illiterate. In 1150/1151 'Allah aI-Din Husayn of the Ghurids conquered Bamiyan after putting fire over Ghazna, and made his brother Fakhr aI-Din Mas'ud rule over the countries including Tokharistan and Juzjan. Bamiyan was transcribed in China as Fanyang in various characters and the Shir-e Bamiyan in the Muslim sources as Shi-Fanyang. The srur is believed in all probability to be derived from ser (MP) which possibly originated in sar, which means a king in the East Iranian dialect, of xsa8riya(AP). The existing chapter on the Western Regions in the Weishu however curiously tells that Fanyang is the eastern boundary of Tukharistan. Fanyang in this source seems to have been wrongly confused with Yuantang which was a Chinese designation of Badakhshan in the time of the Tang presence in the Western Regions. Some years later than the Tang presence was established, the Beishi was edited in the Tang court and Yuantang was used in the chapter on the Western Regions of that History as the eastern boundary of Tokharistan. Since the chapter on the Western Regions of the Weishu had long since been lost the newly edited chapter of the Western Regions in the Beishi supplemented the lost chapter on the Western Regions of the Weishu. The Beishi and the Weishu have shared one and the same chapter. One may therefore be allowed to take Fanyang in the Weishu as confused with Yuantang since the Chinese characters fan and yuan are very similar to each other and those such as yang and tang are also liable to be confused. Such strict attention drawn to Fanyang in the Weishu as noted here also suggests that the Wei China had not been informed of Bamiyan. The Chinese sources on Bamiyan call the capital of Bamiyan either Fuli or Luolan. Luolan was identified by Le Strange with al-Lahum which he read with uncertainty in a text of Muqaddasi. In another text of Muqaddasi in the BGA III (Leiden 1906) al-Lahum is clearly read as al-Iuhum, meaning meat, of which Ghaznin is plenty. The identification of Luolan with al-Lahum is therefore disputable. Xuanzang, regardless of the actual name of the capital, tells that it is located to the southwest of the western colossal Buddhas. The fact that no site in this direction is located has raised an insoluble question as to the whereabouts of the royal capital of the Bamiyan kingdom. On the other hand 157 the existence of the capital on a steep hill described by Xuanzang might be close to, the location given by the Arab geographers who say that the town of Bamiyan, half as large as Balkh in size, is on a hill with a river in front of it and does not form a hisar which should be on the plain (Istakhri, Hauqal and Ya'qubi), if one did not take the direction given by Xuanzang as serious, whichever one may choose Shahr-e Gholghola or Shahr-e Za'hak. The Chinese and Islamic sources also give several towns in the territory of Bamiyan, Baz(s)ghurfand of them in the latter sources being regarded as Fushifu in the former by some scholars. According to Xuanzang who stayed in Bamiyan in the spring 629, the affectionate belief in Buddhism expressed by the royal family, local people, and mechants of long distant trade was particularly deeper than those in neighbouring countries. They cordially pay respect to the religion not only from the triratna but also down to various minor spirits. Ten (or several dozens of) Buddhist temples and several thousand monks were supported by such people. Xuanzang also notices three of the colossal Buddha statues: (1) A standing Buddha made of stone, 140-150 Tang-chi (43 -47 m) high, to the northeast of the royal town was sparkling the golden hues with its precious ornaments and dazzling one's eyes. The brightness and decorations of this Buddha recently led Klimburg-Salter (1989) to hypothesize that it was so-called the Buddha pare; (2) A lying Buddha in the pose of nirvana, about 1000 Tang-chi (about 300 m) long, housed in a temple to the east of No.1, was dedicated by the latest king. The king now holds every five years a great assembly of uninterrupted donation, giving all of his possessions from his wife and children down to his country's treasures, and giving even himself when they were out of stock. On the occasion of this ceremony his vassals always took a responsibility of buying all of the dedicated treasures back from monks. A certain Chinese source is significant telling that the lying Buddha is made of clay. The contexts of this image in its material, size and chronology connnect it with the known huge lying Buddhas found at Tapa Sardar (Ghazni) and Adjina Tepa (Dushanbe); (3) Farther east of this temple was the other standing Buddha, or the Sakyamuni image, cast in the toushi (pure copper or brass, evidently not bronze), about 100 Tang-chi (about 30 m) high and made in different parts to be jointed together in a complete form. Two of the existing colossal Buddhas are carved out of conglomerate as they are standing in the imposing rock-cut niches covered with barrel-vault, 55 m. and 38m. high (IIlus. 18). They have drawn attention of both Arab geographers and the nineteenth century travellers. Ya'qut informs that in Bamiyan is rising a high-pillared temple (bayt) in the sky, in which are drawn all kinds of birds created by the Supreme God and are standing two colossal idols carved from the top of mountain down to the foot, called surkh (red) but and khink but (white) which have no equals in the world. In the late ninth century AD Bamiyan seems to have been overrun by the Saffarid Ya'qub b. al-Lais who on the way to Hindustan sent an idol of Bamiyan together with war trophies to the Caliph in Baghdad. Even in the tenth century al-Kindi, as quoted in Fihrist, tells that Indians make pilgrimage with their sacrifices and incense to both statues called jun bukt 158 and zun bukt which are carved out on the side of the valley. Still to be discussed is a question whether the two standing colossi described by Xuanzang and the two existing gigantic images are quite identical. A question as to whether the existing 38m. colossus is identical with the toushi Sakyamuni has recently been raised by Carter(1985) who claims based on a false identification of the capital town with Shahr-e Gholghola that the latter had been standing free from the cliff to the east of the 38m. Buddha and that the 55 m. Buddha was built later in the late seventh century after Xuanzang. Carter's view was strictly criticized by Klimburg-Salter (1989). There is no trace of metal casting on the 38m Buddha and also no evidence for proving whether Xuanzang told the truth about the toushi. Other than the two rock-cut colossi, still intact is a standing Buddha, 7.7 m high, in a deep niche in the Kakrak valley, and remain on the cliff of the main valley six larger niches, similar in size to that of the Kakrak valley, which ever enshrined huge statues. Two of them are to the east of the 38 m. Buddha and another to the west of it. Three others are on the western half of the cliff. The largest of them, about 13 m high, has a shape similar to that of the 55 m. Buddha. The device of coating with clay the two colossi differs from each other. The conglomerate core of the 55 m. Buddha are regularly driven with the wooden pegs along each line of drapery, which are tied with each other by ropes and covered with clay. Thus a cross-section of draperies gives an appearance that a number of semi-circles are lined up on the surface of the body. On the other hand, the surface of the core of the 38 m. Buddha is irregularly pierced with small cavities to facilitate thick clay to adhere well in order to get reaaly wavy draperies, as seen on a seated image in a niche west of tl1e 38 m. Buddha. Different methods may suggest that the 55 m. Buddha was built based on either a model or a detailed original planning. The exact date of the existing colossi in Bamiyan is obscure despite the parallels found at Tapa Sardar in the south and at Adjina Tepa in the north. Significant for dating them is that such colossi in a small secluded mountain valley could have come into reality solely in the time of economic prosperity which was supported in Bamiyan by long distant traders who connected India with Central Asia. Commercial importance of Bamiyan seems to have appeared around the middle of the sixth century when caravan routes began changing from the one crossing the eastern parts of the Hindukush to the other crossing the western foot of the Hindukush to Tokharistan through Kapisi and Bamiyan. (See Chapter VIII.) The cave monasteries themselves in Bamiyan had not been referred to by any Chinese and Arab sources. They have since 1920s successively been described in detail with modern discipline by the French, Italian, Afghan and Japanese archaeologists who seem to have surveyed only about one hundred of no less than nine hundred caves in all. On the northern cliff of the main valley there are more than seven hundred fifty caves including eight notable niches noted above, and about eighty caves in the Kakrak valley as well as about thirty in the Foladeh valley. Most usual is the cave with a rectangular plan covered with either flat or barrel-vaulted ceiling. The sixty-two caves so far surveyed are square in plan and twenty-four are octagonal. They are covered with a dome in most 159 cases and with a corbelled roof in some. The space between the sided walls and the dome is usually filled with either a simply projected band or a drum on such a band. In some of the octagonal rooms a drum is often replaced by sixteen-sided panelling. A rather decorative squinch arch resting on the band is a usual device for supporting the dome at each corner, but in some caves the squinch arch is more decorative carved on the drum. The band jutting out from the top of the walls in square room is given a quadrantal curvature at each corner. Such device is not necessary for supporting a dome or squinch, but peculiar to a corbelled dome which is quite usual in 'Gandharan' architecture, or more widely speaking, native in Indian archi tecture. The octagonal caves, which are densely located on the main cliff and around the feet of the both colossi in particular, seem to give a clue to dating them, when compared with octagonal architecture, like stupas, which is supposedly specific feature of later, if not latest, Buddhist precincts in other regions (i.e. Shotorak, Tapa Skandar, Tapa Sardar, Hadda, and Takht-e Bahi), where it seems to be associated with constructing shrines or enlarging the previous building in a larger scale. The distribution of most of the outstanding caves covered with a dome or a corbelled roof is quite dense in the area around the 38m. Buddha in the eastern half of the main cliff and they consist of twenty-four groups, each having at least one cave covered with a dome or a corbelled roof. In the western half, on the other hand, only five groups consisting of such caves are placed among a great number of usual rectangular caves. The fresco painting associated with caves and niches is still existing in sixty-four of the caves. It also remains in four of the rectangular caves. Since most of rectangular caves must have been used for monk's cell or even warehouse, it is dubious if all of them were painted. The caves much preserving the painting are either domed or covered with a corbelled roof. Some of the niches have well been known of the painted composition of Buddhist universe. It should be particularly pointed out that all extant paintings are drawn on whitish plaster with vermillion lines and painted with thick pigments of various colours, among which the bright green is exceptional, on the light or dark blue background. The subjects of pictures are quite unrivalled throughout the Buddhist world. No narrative scene from the stories of the Jatakas and the life of Buddha are present except the nirvana scenes in some five caves. The nirvana scenes however are quite independent from the other episodes. Of striking importance is the fact that the Buddha in nirvana painted mainly on the wall above the doorway is closely related with the Bodhisattva Maitreya holding a flask in his right hand and occupying the central zone at the zenith of a dome or in the centre of a vaulted ceiling. The Buddha in nirvana in Bamiyan is always attended by many characters crying out of their deep sorrow with their arms raised or injuring themselves with knives, among whom MahaMaya and Maha KaSyapa are important figures which never appear on the Gandharan reliefs. That Buddha therefore seems to have been conceived not as the Buddha experiencing spiritual awakening but as the dead, or the demise of the True Law, who will live again in the paradise of the Tushita Heaven where Maitreya will be waiting for him in order to advocate the True Law in the future. On either each side of or above the lying Buddha, the solar and lunar deities are depicted 160 respectively in a medaillon. Whatever the ceilings in square and octagonal caves may be, the Bodhisattva Maitreya, usually seated, is a popular main deity throughout the subjects of paintings in Bamiyan, surrounded by thousand smaller Buddhas within a multi-circled aureole whether seated or standing in various mudras. The smaller Buddhas are tightly arranged in several courses of concentric bands radiating from a central Bodhisattva at the zenith, and even on each corbelled arch of a squinch. Only in several caves, a wall of the room has a large niche housing a clay statue which is now missing. Clay statues which has disappeared were often housed in each arch of archades carved out on the drum and/or on the lower part of the dome. Such composition as painted or carved in the concentric bands on a dome suggests that the dome in Bamiyan has deep roots in an architectural soil of the Indian world since the painted concentric patterns on the ceiling do not seem to have existed without the corbelled concentric structure which was characteristic of Indian dome. The solar deity holding a long spear and wearing a long sword, painted within the bright white aureole, or a sun disk, occupies the whole barrel-vaulted ceiling of the eastern colossal niche. He is standing in a chariot driven by four horses and attended by two guardians on the lower and by two wind-deities on the above. On each side of the upper part of the niche below this huge composition, a row of royal personages seated behind the balustrade together with one seated Buddha between two 'bejewelled' ones is depicted as performing something of religious ceremony. On the other hand, all that is left of more huge composition on the ceiling of the western colossal niche is only three of the four personages seated on a chair covered with decorative cloth with their legs crossed and with their hands in various madras. On each side of them at the foot of the vault is a panel depicting an archade consisting of semi-circular and trapezoidal arches, and in each of them is a personage similar to the above ones. Such figures are not always identified with Bodhisattvas. Quite a spacious portion of the ceiling leaving nothing of painting has raised an enigma as to which deity is painted. The Maitreya image might be expected, if the same deity in most of the significant caves was taken into consideration. The fact, however, that the lunar deity appears in association with the solar one in the nirvana scene may give another possibility to interpreting it as the lunar one, particularly when one may refer to Xuanzang's record that a clay statue of the nirvana Buddha housed in a temple is situated between the two colossal Buddhas. 161 x A New Date for Begram III 1. Ghirshman's Dating Begram III Crossing the black ranges and probably descending along the Ghorband gorge, Xuanzang reached the royal city of Kapisi in 629. That city at the time of Xuanzang was first identified by Foucher with Begram, which is called by him the "Nouvelle Ville Royale"(Foucher 1942-47: 140). His identification disagrees with the result of excavation by Ghirshman (1946: 41) whose theory is that the end of Begram III came with the Chionites (in the end of the fifth century according to him) who supplanted the Kidara Kushans (IHus. 19). This chronology led Ghirshman to a new identification ofXuanzang's royal city of Kapisi with another extensive flat mound called Ghunde Peisa, about five kilometers to the southeast of Begram. There was found a fragment of the earthenware stamped with a circle within which two horses are facing each other (Ghirshman 1946: 41). The distance and direction of each monument in Kapisi is usually measured by Xuanzang with the royal city as the starting point. Exceptions exists with the ones that are too close to the city to measure the distance: to the northwest of the city on the south bank of a large river is a monastery built by the former king; To the southeast of this monastery is another built by that king; To the southeast of the latter is a monastery built by the former queen. The first monastery is on the south bank of a large river, apparently assignable to the Panjshir. The proximity of these monasteries to the river suggests that the royal city was on the Panjshir. Ghunde Peisa is about five kilometers to the southeast of Begram, which is almost equal to ten Chinese miles of Tang measures. Ghirshman's identification of the royal city with Ghunde Peisa does not seem supportable. Ghirshman's date of Begram III should be checked. 2. Two Criteria for Renovation. A rush report on the excavation of the central part of the Nouvelle Ville Royale called "Bazar" by Meunie only includes a general plan of the excavated area, a list of collected objects and no more than nine lines of description (Carl 1959B: 84-102). Mainly unearthed are pottery with the stamp decorations and the bronze and iron objects. In certain rooms are the two superimposed layers of occupation. The lower layer produced the coins identifiable as of Hermaeus, Soter Megas, and Wima Kadphises, while the coins of Kanishka, Huvishka and Vasudeva (Carl 1959B: 85) were from the upper layer. These layers must be stratigraphically identical with Niveau I and II of Ghirshman's excavations at the west half of that city. 162 However, Ghirshman recognizes that only Niveau III, called Begram III, the uppermost period of the area, produced the pottery with the stamped decorations, which, according to Ghirshman, first appears in that period, and that the pottery tradition became extinct in that same period. The numismatic evidence used by both excavators raises the difficulty that the upper layer of Meunie, which produced the stamp decorations, is identical with the Niveau II, Begram II, of Ghirshman which does not produce any stamped decorations. Ghirshman (1946: 99) states that the latest coins attested in Niveau II are those of Vasudeva, the last ruler of the second Kushan dynasty. He pointed out elsewhere that the coins discovered in Niveau III are attributable to the Kushan III and IV dynasties (Ghirshman 1946: 97). He does not show any exact distribution of coins in each Niveau. The description given by Ghirshman (1946: 85-86) is a classification of the coins themselves and their quantities of each issue, not a list of the coins unearthed from each layer. As to the coins, the exact data of his excavations at Begram is not available. If so, it is much safer to use the stratigraphical evidence offered by Ghirshman: Begram Bazar is chronologically equal to Begram III sharing common features such as the stamp-decorated pottery with the latter. The other two excavation quarters at Begram should be referred to here: two oblong buildings are strengthened with bastions or tower at each angle, one being in the city and the other about four hundred meters to the south of the southern city wall. The former, found by Meunie, is strengthened with a bastion of circular type at each angle. The northeast bastion and the east part of the north wall were built on the wall of Room T ofNiveau II, one of the rooms connected to Room No.1 0. This superimposition duly led Ghirshman to an understanding that the bastioned building belong to Begram III (Menie! 959: 108). Material evidence for his interpretation is the stamp-decorated potsherds collected in one of the rooms of this building (Meunie 1959B: 104). The building with bastions outside the city wal (Illus. 20, 5) has the same ground plan as the above with a different composition of rooms (Illus. 20, 2). The coins of Vasudeva from the lower layer and those of Eucratides from the upper cannot give any clue to solving the chronology of the monuments Meunie 1959: 106). Despite less informative report Meunie's excavation at the south gate of the rampart is very important (Meunie 1959: 107-113). As Meunie mentions, during the last phase marked by potsherds with stamp impression, the habitation area extended across the gate to the south front of the city wall and the entrance was eventually flanked on both sides by the stupa-like square momuments. The city wall and its gateway no longer served any defensive role. Instead the fortress reinforced by four round bastions was built for a new purpose to defend the expanded habitation area. This phase of Begram III is clearly the last one at the site. The bastioned building within the city wall would also be curious if the city wall was still active. The two bastioned buildings in and out of the city wall and all phases in Meunie's 'Bazar' should be attributed to a broader period which Ghirshman called Begram III. Such common features as the stamp decoration and the round bastion seem firm clues, instead of dubious coins, to date more properly in this paper for renewing the dating of Begram III. 163 3. Circular Bastions in the Kapisi-Kabul Reion. The similar bastioned buildings exist in the same region as Begram to the south of the Hindukush at Shotorak, Tapa Skandar, Khair Khana, Tepe Maranjan and the Saka fort. The plan of bastion is basically circular, except for the the city wall at Begram II which is faced with square bastions (Ghirshman 1946: 16, fig. 5). The fortress at Khair Khana called 'H' by J. Hackin is located at the westernmost and highest part of this religious complex. According to Hackin the masonry of bastions is later than that of the principal edifices (Hackin and Carl 1936: 5) (IUus. 20, 3). The westernmost and highest location of the bastioned building, however, suggests a defensive role for the eastern holy building complex. The ground plan quite similar to that of Begram III should particularly be recalled since the latter played defensive role after the city wall became useless. The marble seated Surya and other images were once enshrined in the "Trois Sanctuaires" at Khair Khana. Schlumberger dates them to the seventh century following Ghirshman who interprets two Nezak-type billon coins unearthed at 'F' (Schlumberger 1955: 110-119; Ghirshman 1948: 51-54; G6b11967: 134-135) to belong to that century (Kuwayama 1972: 28-49). If the "Trois Sanctuaires" were built for enshrining such marble carvings dated to the seventh century, the fortress protecting the holy edifices is also dated to the same time span. The fortress may have been introduced in the later phase of the "Trois Sanctuaires", if the masonry is later than that of the sanctuaries (I). The chronological proximity of the Saka fort (IIlus. 20, 6) to the "Ancien Sanctuaire" at Khair Khana is suggested by a particular device of the facing of wall shared by both sites. Meunie explains it as follows: l'aspect particulier de ce crepi reside en ce qu'il imite les murs de planches placees horizontalement: chaque planche recouvrant de son arere inferieure la planche placee immediatement audessous d'elle forme de la sorte jet d'eau, evitant les infiltrations dans les joints. (Carl 1959A: 14) The "Trois Sanctuaires" are in fact a later construction on the flat roof of the "Ancien Sanctuaire" which was systematically filled up with stone and mud-bricks when the "Trois Sanctuaires" were built (Illus. 30). A billon coin of the Nezak type struck only on the obverse was unearthed in the higher part of the bastion 'G' at the Saka fort (Carl 1959A: 18). The Saka fort is also contemporaneous with the later complex of Khair Khana, which includes the westernmost fortress. Another building strengthened with more than four bastions is the 'Castle' of Tapa Skandar (Illus. 20, 9; Kuwayama 1974A: fig. 31). The monument is roughly square in plan with two huge bastions being at the south angle, one on the north face and the other smaller four on the west face. The western bastions are of original structure of the monument, while the others belong to a later phase of construction (Kuwayama 1974A: 7). The site has two occupations of different age. The later period, to which the 'Castle' 164 belongs, saw the flourishing Sivaite cult mainly in the seventh century with several sanctuaries, one of which was dedicated to the cult of the marble image of U ma MaheSvara seated on the bull Nandin together with a child deity (Illus. 31; Kuwayama 1972B: illus. 14, 7-8). The later phases of Tapa Skandar and of Khair Khana share the common feature, the Brahmanism, with each other. Architecturally the circular bastions are also common to these sites. Tapa Skandar is identifiable with a town called by Xuanzang Xibiduofalaci, as pointed out in the first report (Kuwayama 1972B). Tapa Skandar had already existed when Xuanzang visited Kapisi in 629. The attribution of a Hephthalite billon coin found in the 'Castle' in 1972 (Kuwayama 1974A: 12, fig. 14) to GobI's Emission 150 would roughly date such monuments to the time in and after the reign of Narana-Narendra (570-580) (Gobi 1967: 115). Tepe Maranjan should be divided into two successive periods if one follows the description by]. Carl and Hackin (1959: 7-12). The bastions and the rectangular stupa enclosure belong to the later period, since the staircase built on the inner face of a wall of "Qal'a" is an annex which is contemporaneous with the bastions (Illus. 20,4; 27, above). The coin hoard concealed with the staircase suggests an epoch when the reinforcement of a monument with four bastions was needed. The terminus post quem for this concealment relies on the date of a coin of this hoard which contains the Kushano-Sassanian coins and Shapur III's. According to GobI, one of the KushanoSassanian coins may be attributed to the later degenerate series of Emission 84 datable to the time of Bahram IV (388-399) (GobI 1967: I, 31). The earliest date of concealment is thus inferable. At Shotorak three bastions were uncovered (Illus. 20, 1; 28). Apart from the two circular, the other is similar to that of Tepe Maranjan with a square plan with each angle rounded off. Meunie thinks it to be one of the earliest constructions of the Shotorak monastery. However, it seems that the round bastion at Shotorak was built later than the other buildings in the stupa courts F and D. All constructions such as the main stupa and the enclosure wall are arranged symmetrically on a long median line from the court F to the stupas D3 and D4. Only Room D 11 abruptly disturbs this regularity by changing its direction (Meunie 1942: 20-22) (Illus. 28). It is at the outer southeastern angle of this room D 11 that the bastion is constructed. At the same time as building bastions at Shotorak and Tepe Maranjan the blockade of doorways in proximity of bastions happened. Architectural evidence for a relative date of such defensive phase at Shotorak comes from Qol-i Nader, a Buddhist monastery near Shotorak, which has no bastion. As the ground plan shows, the stupa and a rectangular monastic building of Qol-i Nader are symmetrically laid out on the same median line with a doorway to the monastic court which faces the rear side of the stupa. In Taxila (Kuwayama 1974C), a popular Buddhist temple takes a plan consisting of a single square stupa which faces the rectanglar monastic building. This layout appears alongwith the introduction of diaper masonry which is no later than the deposit of the copper coins of Kanishka and the rectangular court of monks' cells is established in the following Period III (Kuwayama 1974C: 327-354).(2) This established pattern is given a position to Qol-i Nader. In comparison with Shotorak, 165 the plan of Qol-i Nader is strikingly different in many points. Qol-i Nader represents an earlier type compared with the patterns in Taxila. But Shotorak does not show any affinity with Qol-i Nader although they are geographically close to each other on the northern side of the Koh-i Pahlavan. Therefore the difference of plans should indicate a chronological separation Bagh Gai Buddhist temple located on a flat, conglomerate hill of Hadda is characterized by a recangular enclosure strengthened with circular bastions. (Illus. 20, 7). Bagh Gai is similar to Qol-i Nader in the regularity of plan, but the Bagh Gai main stupa is enclosed at three sides by small shrines with many small stupas in the processional path. This type of the stupa court appears in Taxila of Period IV and in Gandhara at the last phase of Mekh Sanda (Kuwayama 1974C: 342; Marshall 1951: 368-387, PI. 101). In light of architectural history at Taxila, Bagh Gai may be later than the first construction of Qol-i Nader, if the Taxila trend of monastic layout is applicable elsewhere outside it. Bagh Gai and Shotorak can be assigned to a period later than Period IV in Taxila rather than contemporaneous with it. The Giri fort at Taxila is strengthened with many circular bastions built by the same masonry as Period IV monasteries (Marshall 1951: 342-343, PI. 81). All of the bastioned buildings discussed here suggests a date later than Begram III which is according to Ghirshman before the fourth century. Period IV saw a depression of building activity in Taxila (Kuwayama 1974C: 341-343). The appearance of the Giri fort with circular bastions in the same horizon is relevant with the emergence of bastioned buildings in the Kapisi-Kabul region(3). 4. Circular Bastions in the North Clues to a chronology of the circular bastions in the Kapisi-Kabul region come from the north of the Hindukush. Here circular bastions suddenly appear among the square bastions. In the south of the Hindukush round bastions are popular, while square bastions are prevalent in the north of the Hindukush at such places as Ai Khanum, Surkh Kotal, Balkh, Dalberjin, and Kohna Masjid. At Kohna Masjid the round bastions were juxtaposed among the square ones (Illus. 20, 8). A part of the rampart at Ai Khanum revealed a massive square bastion of considerable dimension (CRAI 1970, 316-317). Despite excavators' silence, the latest possible date of the circular bastions may be assigned to 50 BC, according to the stratigraphical evidence of pottery cleared up in a house of the South Quarter (CRAI, 1971,447-452). At Surkh Kotal the square bastions with a small room inside are attached to the outer face of the rectangular enclosure of fortress-like Temple. Cutting through the south wall of Balkh Schlumberger and Le Berre established the sequence of the city Bactra: I, lA, II, and III. Bactra I was recognized on Bala Hissar, while Bactra IA was identified as the outskirts which was developing at the foot of Bala Hissar covering the vast extension of the city and its fortification towards the south. Bactra II was an extension of Bactra IA with its fortification towards the west. The precedent extension was abandoned 166 during Bactra II. Schlumberger assigned Bactra I to the period before the Kushans, while Gardin providing pottery evidence from the soundings at Bala Hissar. But his chronology of Bactra IA must have been divided into either pre-Kushan or Kushan. Bactra II is difficult to be dated since the sizes of the employed mud-bricks give no clue. Schlumberger concludes that if Bactra I and IA are of the Great Kushans Bactra II may be around the beginning of the Hephthalite rule invasion (Schlumberger and Le Berre 1964: 87-89). The city wall of Dilberdjin Kazan is reinforced at the four side by square bastions, while a circular fortification in the center of the city is strengthened with circular bastions. Along the inner face of the city wall the galleries were recovered built of mud-bricks, the sizes of which are 38 x 28 x 12.5, 35 x 30 x 10, and 35 x 34 x 13 (cm). Based on the coins collected in and around the site Krouglikova believes that the city had been inhabited before the earlier half of the fifth century. However, the stylistic affinity of the mural painting discovered in one of the galleries to those found at Balalyk Tepe and pottery evidence comparable to that from other sites converge to a much later date (Fitzsimmons 1994 and 1996). Kohna Masjid, situated a short way to the southeast of Surkh Kotal, is a citadel of the modest dimension built on an isolated limestone hillock (Bernard 1964: 212-221; Veuve 1974; See also fig. 2. 1). Bernard suggests that Kohna Masjid is later than the Kushans and contemporary with the Sassanians. According to him, the Period the earliest of the site, was constructed on natural rock and ended with the conflagration after five successive building phases. Throughout these phases dominant was the pottery with burnished lines and with stamped decoration in the medallion types. Bernard further notes that the dwellers of the following Period II partially used the remnant walls of the Period III alongwith their own robust constructions and that they used pottery similar to that of the previous period. Bernard particularly emphasizes the juxtaposition of both types of bastions and regards the round bastion attached to the southwest angle of the citadel as one of the rebuildings which were likewise seen at other places of the site, such as the vaulted ceiling on the ground floor at the rectangular tower called 'C'. Bernard does not directly point out which period these rebuildings belong to. Presumably the round bastion had been built in the Period III, since the domestic houses were built outside of the rampart during the Period II and made the round bastion senseless. The bastions in the areas between the Amu and the Hindukush are square from the Greek period on down to the time of Sassanian hegemony. Therefore exceptional are the round bastions at Kohna Masjid which seem to have been introduced from outside to Tokharistan. The most possible homeland of this type of bastion is the region to the south of the Hindukush where it was quite often used, as exemplified earlier. However, the use of round bastion in the south of the Hindukush is confined to a time after Begram I-II of Ghirshman's and Sirkap IV of Marshall's, where square bastions were popular. 5. Stamp Impressions 167 There are clearly two categories of stamp impressions in the northern and southern sides of the Hindukush. Popular in the north is a palmette pattern within small oval, circular, or square frame, the diameter of which is about 2 centimeters: e.g., from Ai Khanum, Surkh Kotal, Dilberdjin Kazan, and Durman Tepe to the south of the Amu and from Kara Tepe on the right bank of it (Illus. 21). Usual in the south of the Hindukush is a large circular medallion-shaped stamp impression with a diameter of c. four centimetes, filled with floral or animal motifs such as horse, antelope, birds and their composite scenes. In some cases a spiral is impressed together with other motifs. Several types were used on one and the same vessel. The geographical distribution of these two categories is very clear. Th finds of the smaller stamp impression is confined to Tokharistan, while the medallion types are peculiar to the south. Here again Kohna Masjid is anomalous for the finds such as potsherds decorated with a large circular medallion-shaped stamp impression of horse motifs. Some of the stamp impressions found at Chaqalaq are in the same line as the finds at Kohna Masjid. The sequence of pottery vessels in the Kunuz regions from Durman III to Chaqalaq III gives the two categories a tentative, comparative chronology (Illus. 27): (1) Dominant from Durman III to Chaqalak III is the red ware with a decoration of burnished multiple lines on thickly red-slipped surface. A type of dish with the vertical rim is very popular. Kohna Masjid II - III and Balkh I - III also produced such dishes. (2) In Durman III in Kunduz a type of bowl which carries two opposing, looped handles on the flared rim made the first appearance. A sharp carination at a short way from the mouth characterizes the shape of a cross section (fig. 5, 1, 2). Carination later lost its sharpness. The final stage saw the multiple incised lines instead of carination. (3) In Balkh III and Kobadian V a code of clay is twisted and attached to the rim as handle. Later in Durman III, a clay code was applied with decorative notches instead of being twisted. In Chaqalaq II, the handles themselves were replaced by clay bosses. In Chaqalaq III this type of bowl itself disappeared. (4) Contemporary to the clay boss replacing the handle in Chaqalaq II, refined pottery ewers suddenly appeared for a short life. The overall shape recalls the "Sassanian" metal ewer. The slender neck with a beak-like mouth surmounts an oval body which is supported by a tiny bottom similar in shape to that with a goblet. An elegant handle connects the mouth to the shoulder. However, throughout Chaqalaq III, the crudeness prevails in shape and fabric, and a good tradition of potter's technique ceased to work. (5) The stem of goblet had usually been hollow until it became solid in Chaqalaq II. The period in Durman IV-Chaqalaq II is a transitional stage of pottery in Kunduz, shown in the following table. The clear discontinuity of the pottery tradition existed in 168 the north of the Hindukush. The phases such as Durman IV and Chaqalaq II chronologically separate the medallion stamp impression from such tiny impressions as found at Surkh Kotal, Kobadian II (MIA, 37, pI. XII) and Ai Khanum (BCH, 625-30, fig. 12-14). The atrophied types lived long into Kobadian V despite an interruption of Kushan Kobadian IV. Some similar motifs appeared during Balkh III (MDAFA, XV, 21, fig. 5), Durman III, Durman IV and Chaqalaq II as well. But in Durman IV and Chaqalaq II only a small cavity impressed with a finger tip appears as a simple decorative pattern in place of much more elaborate, palmette designs. Dish Carinated Bowl typw Carination / Handle Ewer Stem of goblets Stamps Balkh III 19C sharp / twisted - ? oval palmettes Kobadian V ? sharp / twisted hollow oval palmettes Durman III ? sharp / incised - hollow oval palmettes DurmanIV ? dull / - - hollow small round dots Chaqalaq II A dull / boss fine ware solid Kohna Masjid III-II Chaqalaq III 0 A ? / ? I popular -/- coarse ? solid coarse small ropund dots medallion types medallion types At Chaqalaq Tepe a kind of stamp impression appear exclusively on the rim of storage jar. Except for the incised marks, they are divided into two types. One is marked by its circular outlines like a medallion, inside of which a swastika, kirtimukha-like mask, wild goat, winged animal, and floral pattern are depicted (See the photographic part of Illus. 21). The other looks like the impression of the gems of the Sassanian types and of an animal under a crude, stepped merlon (Illus. 21, above). The chronological peculiarity about these impressions is that the former type exclusively comes from the last stage of the pottery tradition known so far in the pre-Muslim Kunduz, in Chaqalaq III. The degenerated palmette motifs impressed on the above bowls and the amphora-shaped vases were popular in Durman III, while such motifs became simple impressed cavities in Durman IV and its following stages. On the other hand, Chaqalaq III does not produce bowls with flared rims nor such impressed decorations. All these factors lead us to an understanding that in the Kunduz the stamp decoration of the smaller size ceased to work in or before the intermediate stage, from which time on the medallion impressions suddenly began to be popular (4). According to Bernard, two main streams of pottery tradition exist in Periods II (later) and III (earlier) at Kohna Masjid: (1) Vne cerami que lissee, generalement a engobe rouge, au l'outil a lisser laisses des zebrures d'un rouge plus fonce. Les formes les plus notables sont la gargoulette a panse piriforme, la cruche a bee, et l'assiette. (2) Vne ceramique estampee avec motifs flo raux jou animaliers. La piece la plus remarquable est un fragment de grande cruche avec sur l'epaule une frise de medaillons representant des cervides allonges a ample ramure. Ce type de 169 ceramique est abondamment repandu en Mghanistan .... (Bernard 1964: 217). Such phases of Kohna Masjid is comparable with those of Chaqalaq II and III, possibly closer to Chaqalaq III. Accordingly, the important objects closely relevant to the date of Chaqalaq III are as follows: (1) A circular bronze coin with a hollow square inside, but without chinese letters, found in the uppermost layer (Higuchi and Kuwayama 1970: 21-22, PI. 63, 5) differs in thickness from non-Chinese coins as well, despite the weight and size indicating it to be of the Sogdian coinage (MIA, 66, 216-280, PIs. I-VI). This cannot be also assigned to any of a series of the Wuzhu coins in size and style. Most possible attribution is to a local coin imitated after the Kaitong Yuanbao which is one of the most widely distributed Chinese coins found in Central Asia. In this case, the earliest possible appearance in Kunduz is not earlier than the first issue in 621. (2) An Arab-Sassanian silver coin of the Khosraw II type issued in the twenty-first year of the Yezdigird era, AD 652, at the earliest (Higuchi and Kuwayama 1970: 22, PI. 63,2). Found from the disturbed floor (Walker 1941: xxxvii and 237). (3) A three-winged eddy-like mark impressed on the fragment of buff ware, identical with the countermark No. 118 by GobI (1967: II, 148). According to GobI, it appears on the second quarter of the Sassanian coins of King Kawad issued in his regnal year 18, 516/517. He also notes that the clan-tamghas of the West Turks could be the nearest. The same mark was also reported from Kei-Kobad-Shakh on the inner surface of one of the bowls (MIA, 66, 297-310, fig. 11 ) . (4) Chaqalaq III and Pianjikent share common features in the medallion-type impressions on pottery (MIA, 66, 125, fig. 20), the type of the scale bar (MIA, 66, 136) and, a thickset ewer (MIA, 124, fig. 19) which differs from that of Chaqalaq III in the clay used and the surface finishing. These factors support dating Chaqalaq III to the seventh century at the earliest. Kohna Masjid can properly be settled in the late sixth or seventh century, since it is located between Chaqalaq II and III. 6. Medallion Types Ghirshman thinks the stamp impression of medallion type as a device introduced to Begram III, while the painted decoration on pottery goblets chracterized Begram II (Ghirshman 1946: 69). Meunie also collected the same kinds of medallion-impressed pottery from Begram Bazar, which is, according to his brief remarks, one of the main objects in that excavation quarter. This was also the case with two square buildings 170 reinforced at four angles by round bastions. At the Saka fort the pottery with such decorative motifs was found in the square rooms inside the round bastions D and F. At Tepe Maranjan they were found outside the northeastern walls which were added at the latest stage of the site together with strong round bastions. At Tapa Skandar several round bastions strengthening the inner castle shared the same time with six hundred and forty-four potsherds bearing the medallion-stamp decorations that were collected in the two seasons' works (lllus. 23; Kuwayama 1974A: 10-12, figs. 29, 30). As mentioned in the interim report, the motifs are classifiable to three categories which is valid for all medallion impressions from the monuments in the Kapisi-Kabul region (lllus. 23 and 24). The affinities of the motifs from Tapa Skandar and Shotorak are especially strong: a horse galloping with ribbons waving from the neck and with tiny three circles between the legs; rosette motifs are also comparable with those from Koh-i Mori, the Saka fort, Tepe Maranjan, and Begram III. Almond-shaped patterns are shared by Begram III, Tepe Maranjan, and Tapa Skandar, giving those sites a simultaneity which is also stressed by the affinities of motifs, such as double spirals and eddies, found from those sites. Based on such motifs shared by potsherds from various sites in the Kapisi-Kabul region, some potters--not to say one and the only-seem to have distributed the products to many sites throughout the Kapisi-Kabul region. This assumption may be opposed to various datings hitherto given to the monuments. Ghirshman suggested Begram III to be datable to a time from the end of the third to the fourth century with a superficial hypothesis that these medallion-stamped motifs appeared in this region as a Sassanian impact resulted from the conquest of Shapur I (241-272). Yet a clue to dating Begram III exists in the cultural phenomena in the north of the Hindukush: the simultaneous existence of the round bastion and the medallion-stamped pottery attested at Kohna Masjid, near Baghlan. Such simultaneity reflects the phenomena in the south of the Hindukush just before the influx of the Muslims in Tokharistan. Existence of the two factors of southern culture gives Kohna Masjid a position that can make Begram III much later than Ghirshman postulates. 7. Begram in the Seventh Cetury The first chapter of the Da Tang Xiyu ji says that about forty Chinese miles to the south of the capital is a town called Xibiduofalaci which is quite undisturbed even though the rest of the region is visited by earthquakes and landslides. Julien reads this town as Sphitavaras, while de Saint-Martin and Beal interpret it as Sveta:varas. On the other hand, Svetavat is suggested by Watters. According to him, this is one of the epithets of Indra .riding a white elephant. His translation follows as the town where was Svetavat-alaya, the shrine of Indra. Xibiduofaluo may be equivalent to Svetasvatara. Svetasvatara originated his Saivite school. The Svetafvatara Upanishad describes that believing Siva as a sole deity he devoted himself to this deity wearing a white yajfiopavita and kaupina on his body smeared with ash. Svetasvatara himself worshipped Mahesvara, following the old style of pasupata. 171 In the Da Tang Xiyu ji, the pasupata school first appears in Kapisi. Of deva temples there were about ten; and that there were about one thousand professed sectarians, such as diganbaras, pasupatas, and those who wear wreaths of skulls as head-ornaments like the kapalikas. Existence of such Brahmanism as the above suggests that among the pasupatas in Kapisi were devotees belonging to a sect originated by Svetasvatara. Moreover, if Xibiduofalao possibly is Svetasvatara, an identification of this town in the Da Tang Xiyu ji with Tapa Skandar would be possible, since Tapa Skandar was a place of Brahmanism where the marble Uma Mahesvara was enshrined. Tapa Skandar had already existed on Xuanzang's visit to Kapisi. As he calculated the distance to Tapa Skandar from the capital, Begram must have been inhabited at the topmost phases when he stayed there. Begram III existed in the second quarter of the seventh century. Begram survived long after the Kushan IV dynasty of Ghirshman's. 172 XI The Horizon of Begram III and Beyond A Chronological Interpretation of the Evidence for Monuments in the Kapisi-Kabul-Ghazni Region 1. Standpoint A new date for Kapisi-Begram III in the previous chapter which was first published in 1974 (Kuwayama 1974: 57-78) is based on two of its cultural characteristics: medallionstamped pottery and round bastions. In the same year Fussman has independently given both of these elements quite a different date by interpreting material from Wardak (Fussman 1974: 65-95) and Gul Dara in Logar (Fussman and Le Berre 1976: 51-51, 95-99). Taddei has pointed out that evidence from Wardak and Tapa Sardar did not seem to lend any support to my hypothesis (Taddei 1978: 266-267). A reinterpretation of their evidence therefore leads me again to stressing my thesis. Further, the scope of my analysis is widened to envelop all sites in the Kapisi-Kabul-Ghazni region which coexisted on the same horizon as Begram III, the latest of three superimposed occupations. First, the points of the previous article is given for convenience' sake. Ghirshman (1946: 41) dates Begram III to sometime before the fourth century at which time the people of Begram III supposedly abandoned the city in the face of the Hephthalites who were believed by him to have invaded from north of the Hindukush.(l) With the geographical situation solely in mind, on the other hand, Foucher (1947: 140) identified Begram, before the site was opened, with the royal city of the Kapisi country described in the Da Tang Xiyu ji by Xuanzang, who visited there in 629.(2) Thus Begram III has been given two opposing datings without any proper reexamination since 1946. The excavations by Ghirshman between 1941 and 1942 revealed that quite a new ceramic device characterized Begram III: stamp decoration executed within a round frame, which I call the "medallion" type. This kind of decoration on pottery vessels has been widely found not only in Begram Bazar (Carl, 1959b), attributable to the same phase as Begram III, but also at other sites in the Kapisi-Kabul-Ghazni region such as Kham Zargar (Kuh-e Mori) (Mustamandi 1967/68, 67-79), Shotorak (Meunie 1942), Tapa Skandar (Kuwayama, 1972, 1974a, 1976, 1978a and 1980), Khair Khana (Hackin 1936 and unpublished find by the Kyoto University Archaeological Mission in 1964), Tepe Maranjan (Carl and Hackin 1959), Saka fort (Carl 1959a), Gul Dara in Logar (Fussman and Le Berre 1976), Tapa Sardar (Taddei 1968 and Silvi Antonini 1979), and Jagatu (Scerrato 1967)-all of them variously dated (For the locations of the sites, see fig. 1). Important, however, is the fact that the medallion-stamp decoration found at these sites clearly shares common features in its designs. Some of the designs, from several sites, were executed with one and the 173 same stamp that presumably derived from a specific workshop of potters (Illus. 24 'and 25). In view of this fact, dating Begram III is vitally important not simply for defining the history of Begram itself but also for drawing a chronological layout for sites in the Kapisi-Kabul-Ghazni region as a whole. The previous attempt to clarify the problem of the date of Begram III stesses the great importance of the simultaneous existence of both cultural elements, pottery with medallion-stamp decoration and round bastions, at Kohna Masjid (Bernard 1964: 212-221) in the Baghlan-Gori plain. Their coexistence at Kohna Masjid is contrary to the tradition of the region north of the Hindukush where square bastions had been prevalent since the time of the Greeks in Bactria and stamp decoration on pottery had been executed but without medallion frames, in quite limited size, in different designs, and for a different duration of usage than the stamps with medallion frames found to the south. However, these two cultural elements found at Kohna Masjid are very common in the regions south of the Hindukush. The round bastions have been found at two fortress-like buildings in Begram, Shotorak, Tapa Skandar, Khair Khana, Tepe Maranjan, Saka Fort, Gul Dara in Logar, and Tapa Sardar, all sites where medallion-stamped pottery was also found. Their coexistence at Kohna Masjid should therefore be regarded as a reflection of the current fashion on the south side of the Hindukush. Kohna Masjid can be placed in between the Periods II and III at Chaqalaq Tepe in view of the comparative stratigraphy and the chronological sequence of pottery in the Kunduz-Baghlan region or the lower and middle valley of the Surkhab, and dated to sometime in the late sixth-seventh centuries (Veuve 1974, Kuwayama 1970 and Kuwayama 1974b: 69-74). Taken as a counterpart of the monuments south of the Hindukush, Kohna Masjid can, therefore, give Begram III a date much later than the one suggested by Ghirshman. 2. Reexamination of the Wardak Evidence. In a detailed review of archaeology in Mghanistan of the mid-seventies, Taddei, however, remarks, as I stated at the beginning, that 'this strict cultural relationship between the round bastions and the medallion-stamped pottery does not seem to find further support elsewhere,' particularly at Wardak and Tapa Sardar (Taddei 1978: 266267). Taddei refers to Fussman (1974) whose important report on Wardak appeared in the same year as my paper: His [Fussman's] surface sherding in the ancient town ofWardak valley where the round bastions are extant has given practically no specimen of that kind of pottery. He thinks that the medallion stamped pottery makes its appearance sometime after the reign of Vasudeva, i.e., towards the end of the second century AD .... The same date is probably to be attributed to the miniature fortress with square, round, and semi-octagonal bastions ... The miniature fortress was found in Early Period II at Tapa Sardar, the period connected 174 with the first appearance of medallion-stamped pottery there. This pottery continued to be used in the Late Period of the same site; i.e., the seventh - ninth centuries AD In his interpretation of the chronology of sites in the Wardak valley, Fussman (1974: 90-91) accepts Ghirshman's dating of Begram, about which doubts now exist. He dates all the surface potsherds collected by him in Wardak to the 'Kushan' period since their types are parallel to those of Begram II and III but, more precisely, to those of Begram II because of the almost total absence of medallion-stamp decoration. Furthermore, he recollects that this kind of decoration does not appear at various Kushan sites such as Surkh Kotal, Begram II, Sirsukh, and Shaikhan Dheri, all of them ending during or before the reign of Vasudeva. Therefore, the round bastions in Wardak, Fussman suggests, are not later than Vasudeva. This conclusion places the simultaneous existence of both cultural elements, round bastions and medallion-stamped pottery, to sometime after Vasudeva. The evidence from Wardak Valley is capable of a different interpretation. Fussman chooses to discount any significance to the discovery of a fragment of a pottery vessel bearing circular stamped decorations among the sampling of ninety-one potsherds collected from the surfaces of the ancient sites there (Fussman 1974: 103). In view of the doubts now raised about Ghirshman's dating of Begram III, it is hasty to ignore the existence of this example before the evidence of further archaeological excavation at Wardak is available. At Tapa Skandar only four sherds depicting medallion-stamp decoration were collected on the surface of the mound in 1967 before I began excavate there in 1970, but seven hundred seventy specimens of potsherds bearing that decoration were unearthed throughout the five season dig which was provisionally closed in 1978. Moreover, most of them came from Inner Fort A (Illus. 23). When the surface of a mound is disrupted by modern villagers in need of soil for agriculture and architecture, as at Inner Fort of Tapa Skandar, one should not neglect even a single specimen of pottery. In the case of medallion-stamped pottery, this warning applies not only to Wardak but also to Khair Khana where no excavated specimen was reported in the descriptions given by the excavator despite an example collected by the Kyoto University Archaeological Mission in 1964. In fact, the specimen from Wardak is described by Fussman: Fragment de panse de jarre; pate grise-rose grossiere, avec points noirs et blanc de degraissant. Engobe exterieur beige avec traces d'engobe noir pardessus. Traces du bord d'un grand estampage circulaire avec semis de cercles en relief. He actually compared this type of medallion-stamp with those found at Begram III (lIlus. 25: no.46; Ghirshman 1946: PI. L, BG506) and Tepe Maranjan (lIlus. 25: no. 40; Carl and Hackin 1959: PI. II, 8). The former are further comparable with examples found at Begram Bazar (Carl 1959B: PI. VII on p. 89, nos.5 and 6) and from Gul Dara in Logar (Illus. 25: no. 45; Fussman and Le Berre 1976: PI. LI, P60 and P61) and the latter with example from Tapa Skandar (Inus. 25: no. 39; Kuwayama 1972: fig. 18). 175 As for pottery types, some of the potsherds collected by Fussman in Wardak (Fussman 1974: fig. 29, nos. 7-10 and 17) are clearly of the same types as those unearthed by Ghirshman in Begram II (Ghirshman 1946: PI. XLIII, nos. 38,40 and 44). However, some of the vessels (Fussman 1974: figs. 3,4, 11-13 and 18) can also be identified with types used only in Begram III (Ghirshman 1946: PI. LIII, No. 29; PI. LIV, No. 660) and, most importantly, to types common at Tapa Skandar of Period II during which phase the marble, seated image of Vma Mahesvara was dedicated at a shrine at the centre of the site. A comparison of pottery vessels thus leads the site of Wardak to the same horizon as Begram III and Tapa Skandar II. Also, the lack of medallion-stamped pottery at the Kushan sites of Surkh Kotal, Shaikhan Dheri and Sirsukh may not represent the chronological turning point which Fussman claims for it. At Surkh Kotal, stamped decoration on pottery seems to have actually been used to a limited degree.(3) But, it is quite different from that executed within a circular border in typological, geographical and chronological points, as I previously demonstrated (Kuwayama 1974b: 69-74). In Gandhara, no medallion-stamped pottery was used during any period, pre-Kushan, Kushan or post-Kushan. The lack of stamped medallion decoration on pottery vessels in the region east of Jalalabad rather seems due to a difference in geo-historical and cultural traditions between the areas east of Jalalabad and the Kapisi-Kabul-Ghazni region.(4) As I showed in 1974 and will show again now, other evidence indigenous to the regions south of the Hindukush indicates that production of medallion-stamped pottery did not originate soon after the disappearance of Vasudeva, as claimed by Fussman and Taddei who accepted Fussman, but considerably later. 3. Tapa Skandar Evidence from the 1976 and 1978 Excavations (5) To reexamine further the date of Begram III, I will first give new stratigraphical evidence for the use of medallion-stamped pottery at Tapa Skandar: the fourth excavation at Tapa Skandar in 1976 revealed that the pottery in question first appeared in Period II (the later period of the site) and that the round bastions had already existed in Period I (the earliest period).(6) Both cultural elements at Tapa Skandar therefore coexisted for the first time in Period II. Taddei is right in so far as he thinks that the two cultural elements did not appear simultaneously in the regions south of the Hindukush. But my stress is put on their coexistence on the horizon of Period II at Tapa Skandar, a fact which can give Begram III the same dating as suggested in my previous paper. The relevant stratigraphical evidence at Tapa Skandar appeared in the area around Buildings F and H. All that is left of these buildings are the masoned foundations and the lower structures, which are built of layers of mud bricks (38 x 38 x 9 cm) and faced with masonry consisting of both river boulders and quarried rubble used together in one and the same wall. Building H stands on a compact filling of three superimposed layers which conceal sections of the walls (c) of an older structure built directly on the bedrock. The walls had become broken, worn out and isolated from each other before they were 176 filled in. Despite their dilapidation, the walls were elaborately faced with quarried rubble tightly jointed without clay. It is clear that this structure clearly predates Building H. Building F likewise realed similar walls which had been built directly on the bedrock (Kuwayama 1975a: Illus. 2 and 3). All that is left of Building F is only mud brick layers making up its foundation within a square masoned frame, which conceals the dilapidated walls of earlier period. Two other walls of this kind, also built on the bedrock, were exposed in 1976 to the east of Building F, running parallel with each other and evenly reduced to a very low elevation. The western counterpart has on its western face two semicircular projections that served as bastions (Kuwayama 1976: Illus. 3). Walls G and J provided an embankment for this building complex preventing the fill retaining Building H from flowing away onto the bedrock that forms a slope westwards. It was because of this sloping bedrock that the older structure was filled up and transformed into a stable floor for Building H. As for Building F, the older structure had to be incorporated into the mud brick layers of the new foundation since the bedrock of that site was more even. The structures thus incorporated or buried clearly belong to the period preceding Buildings Hand F themselves; i.e., Period I. Since Embankment J was attached to the southwest angle of the Vma Mahesvara Shrine, Buildings F and Hare one step later than the shrine, but attributable to Period II. Stratigraphical and other relevant evidence used for setting the above buildings in their proper chronological order is also pertinent to the pottery found at Tapa Skandar with the medallion-stamp decoration. No pottery of this kind was found on the bedrock on which the isolated walls of Period I stand, but examples have been unearthed in the filling and on the mud floor of Building H. The potsherds which include the medallionstamped pottery are homogeneous in the quality of clay, slip and firing. The sherds are hard, brownish red ware, well oxidized in the kiln and mixed with mica in most cases, and with fine sand in other cases. They are thinly covered with dark red or dark brown slip. Important is the fact that the main types of vessels, of great variety, include some which are similar to the pottery of Begram III (Ghirshman 1946: PI. XLIX, BG516, PI. LIII, nos. 29 and 31, PI. LIV, nos. 51-55,60,63-66 and 72). As mentioned earlier, they are also familiar at Wardak (Fussman 1974: Pottery nos. 3,4, 11-13 and IS). These types of potsherds mark a sharp contrast to those of Period I found on the bedrock which include both red and grey wares. The pottery vessels of red ware of Period I were found only as small fragments, in very limited quantity, and give no clues for reconstructing their whole shapes. They are made of well levigated clay with no admixture, are generally less hard after firing than Period II pottery, and the core of fired clay is greyish brown in most cases. The outer surface is so thickly covered with dark red slip that one can strip it off with the tip of one's nail. The grey ware is of the same clay as the red ware, but rather harder than the red one, and confines itself to smaller types of vessels. Some of the sherds of goblets among the red ware closely resemble a specific type peculiar to Period II at Begram (Ghirshman 1946: PI. XL, BG34S, BG117, BG150 and BG207) and thus support a chronological parallelism between Tapa Skandar of Period I and Begram of Period II. In addition, only a few types of the vases and carinated dishes of red ware-unearthed in restricted number-suggest that the 177 occupants of Period I at Tapa Skandar had any connection with regions to the east. These examples were imports from Shah Naser Ghundai at Chahar Bagh in Jalalabad or either Bala Hisar or Shaikhan Dheri in Charsada, areas which, as previously mentioned, seem to have had a different cultural tradition. Two facts suggest that Period II at Tapa Skandar was inaugurated during quite a new era, long after the preceding occupation: the walls of Period I are broken and worn out and the masonries differ significantly between the two periods. A parallel succession of events seems to have also occurred at Begram. Ghirshman states that the plans of the houses at Begram III were laid out in a different direction from those of Period II, neglecting the preceding layout (Ghirshman 1946: 30-32). Ghirshman also mentions that the walls of the preceding period were not used for the foundations of new houses built in the beginning of Begram III. This implies that the walls of Begram II were seldom exposed on the surface of the ground by the beginning of the later period. The initiation of Tapa Skandar II and Begram III thus shares two trends: a quite new occupation begun after a time vacuum following the end of the preceding occupation and the introduction of medallion-stamped pottery. Even though it is quite sure that the medallionstamped pottery was inaugurated in Begram III and Tapa Skandar II, it is unclear how the pottery was introduced and whether it was used from the very beginning of these phases.(7) During Period II, Tapa Skandar acquired its most extensive and large scale constructions covering the whole of the mound (fig. 4): for example, the imposing Inner Fort at the site's highest location (A), the Vma Mahesvara shrine at the top of the westward slope (C), the two big foundations (D and E) of the same solid masonry as the Vma Mahesvara shrine and attached to it, the shrine of modest size just to the east of the Inner Fort (B), a barrel-vaulted shrine (S2) built on a highly raised terrace built of solid masonry (S), Buildings F and H, a building complex including one construction roofed with a dome (M, N, 0, P and Q), and a big structure with a rounded projection (R). The seventh-eighth century date for the Vma Mahesvara shrine and connected buildings can be derived from a detailed stylistic comparison of the excavated marble image of Vma Mahesvara with other Brahmanical and even Buddhist devotional objects in eastern Afghanistan (Kuwayama 1978b). The date also finds support in a palaeographical interpretation of the inscribed hymn on the pedestal of the Vma Mahesvara image (Kuwayama 1972 and Yamada 1972). 4. Tapa Skandar II and Khair Khana II as described by Xuanzang. I do not think it irrelevant or inappropriate to repeat here my previous hypothesis (Kuwayama 1975: 93-107) that Tapa Skandar of Period II and Khair Khana can be dated from the written documents of Sui and Tang China. (1) Tapa Skandar II. The first and foremost point is the identification of Tapa Skandar with the town 178 Xibiduofaluo-ci mentioned in relation to Kapisi in the Da Tang Xiyuji (Vol. 2) edited by Xuanzang.(8) This town was located about 40 Chinese miles to the south of the royal city of Kapisi. The royal city of Kapisi is apparently on the south bank of a large river and must be identical to the modern site Begram on the south bank of the Panjshir.(9) Various restorations of Xibiduofaluo-ci have been suggested. Julian (1853: 393) associated it with Sphitavaras which, I am afraid, has no specific meaning. Watters (1904-1905: 126) thought it referred to a shrine of Indra by dividing the word into two elements: Xibiduofaluo, which, according to him, was an epithet of Indra; and ci, which he correctly interpreted as 'shrine'. Despite his proper division of the town's name into two elements and its acceptance by]. N. Banerjea (1956: 9) who more radically defined the city as Svetavatalaya or Indrapura, the interpretation of Watters does not sit well with references to Brahmanism made by Xuanzang who never refers to Indra but does give much evidence for the existence of Siva cults represented by the Pasupata and Kapalika sects. Svetambara, which is one of the two main sects of J ainism, seems to be phonetically close to a supposed original Indic name for Xibiduofaluo, but Xuanzang only refers to the existence of the other Jain sect, Digambara. In view of the above inadequacies and the discovery of an Uma Mahesvara image with inscribed hymn for Mahesvara at Tapa Skandar, the town's name and its attribution to a specific archaeological monument can better be interpreted in another wayYO) Geographically and culturally located on the outskirts of India, it is possible that knowledge of Sanskrit was not accurately maintained at Kapisi and that Xibiduofaluo is a transcription of Svetavara, a corrupt form of Svetasvatara. According to the Svetasvatara Upanishad, Svetasvatara was a religious leader-clad in rags, wearing a white yajnopavita, his whole body smeared with ash-who originated a school in continuation of the ancient pasupata sect which believed in Siva as the one and only god. Under this interpretation Xibiduofala-ci could be restored as 'the town where the shrine for Svetasvatara was.' The only candidate for such a town is Tapa Skandar where a shrine was found housing a statue of Uma Mahesvara. Significantly, the pedestal of this statue contains an inscribed hymn in Acute-Angled Brahmi the theme of which is clearly Mahesvara (Yamada 1972: 15-22). If the above identification with Xibiduofalaci is sound, the Uma MaheSvara shrine of Period II at Tapa Skandar must have existed at the time of Xuanzang's visit to Kapisi, i.e., in 629. (2 ) Khair Khana II This date for Period II at Tapa Skandar finds further support in another description of Xuanzang: 'About 30 Chinese miles to the south of the town Xibiduofaluo-ci was Mount Aruna.' I have identified Mount Aruna as the site of the Trois Sanctuaires, or the Three Shrines, of Khair Khana where Hackin found a marble Surya seated on a chariot driven by Aruna and a marble pedestal on which a bare foot of a central deity and a smaller standing statue of a warrior attendant remained intact (Kuwayama 1975). In 1980 another Surya image, one of the most elaborately carved, was found by chance (Bernard and Grenet 1981). At Khair Khana there was also found a still bigger shrine, called by Hackin the Ancien Sanctuaire, or the Early Shrine, systematically filled up with 179 mud and debris in order to support the floor on which the later Three Shrines stood (Illus. 30). To date the Early Shrine we have two Chinese written accounts: The first is the Xiyu tuji (Illustrated Accounts of the Western Region) edited in the earlier half of AD 606 by Pei Ju, a Sui official in charge of relations with the Western Region, based on information collected from western traders on his official visit to Zhangye in Gansu Province. The accounts are now lost except for Pei Ju's preface of his biography written by the Suishu editors. Next comes the Western Region chapter of the Suishu, edited in 629 and 630, to which does theXiyutuji apparently offer the basic sources together with the Da Sui Xiguo zhuan (Great Sui Record of the Western Countries). The chapter on the Western Region in the Suishu speaks of a shrine on Mount Cong(onion)-ling(mountain) (11) in the Cao (Kapisi) country (12) as follows: Worshipping the licentious god is a custom of this country; there is a shrine for the god Zhun whose ceremonial institution is extremely magnificent, roofed with gold and silver plates and paved with silver. Xuanzang later came to eastern Afghanistan and described in the section on J aukuta (presumably identifiable with Zabulistan in Islamic sources) in the Da Tang Xiyu ji a shrine on Mount Zhuna-hira at the southern border of that country. It was very rich with many precious objects dedicated to the god who had come there from Mount Aruna. Xuanzang also states in the section on the Kapisi country in the same book that the god of Mount Aruna enviously looks afar at the richness of Mount Zhuna-hira since he refused a previous proposal of the god Zhun to live together on Mount Aruna at the time when Zhun had just arrived from far away and was seeking a place to rest. The two tales cited above well suggest that in the early seventh century there occurred in Kapisi something like a religious conflict between two groups worshipping Zhun and Surya respectively, a conflict which might also reflect contemporary political issues. If the tale Xuanzang heard was simply propaganda maintained by the Aruna (or Surya) sect and we are allowed to regard Mount Congling in the Suishu as an earlier name for Mount Aruna in the Da Tang Xiyu ji, it might be true that Khair Khana is the very site where Zhun had been worshipped-in the Early Shrine-until the arrival of the worshippers of Surya who forced the Zhun sect south, changed the name of the site from Mount Congling to Mount Aruna, built their own shrine-the later Three Sanctuaries-and who invented the story about the god of Aruna rejecting the newcomer, Zhun, so as to legitimatize their hold on the site. If this hypothesis is accepted, the date of this change seems attributable to the period between 606 and 629 since Mount Congling in the Suishu was first recorded by Pei Ju in 606, when he edited the Xiyu tuji, and Mount Aruna had already existed there on Xuanzang's visit to Kapisi in 629. The later Three Sanctuaries at Khair Khana-call it Khair Khana II- were installed after sometime between these dates. Since there are a number of links between Khair Khana II and Tapa Skandar II-marble Brahmanical sculptures, architectural designs, and a variety of material finds-the above evidence for dating the construction of the Three Sanctuaries serves to verify the (13) ISO attribution of the initiation of Tapa Skandar II itself to the very early seventh century at the latest. 5. Tapa Sardar Evidence. In addition to Tapa Skandar, the Italian excavations at Tapa Sardar also provide recently established stratigraphical references for the appearance of medallion-stamped pottery. This pottery made its appearance at Tapa Sardar in the Early Period II which has two phases, A(earlier) and B(later), and continued to be used in the Late Period (Taddei and Verardi 1984: 43). For the chronology of the Early Period II the most relevant area is around stupa 64 and the miniature city wall (69), both of them located midway on the northwestern slope of the mound. stupa 64 was built on top of the remains of Early Period I structure 65 (Illus. 26, above). According to Taddei, medallionstamped pottery was found in the filling collected to support a mud floor around stupa 64. The mud floor connects the stupa with the miniature city wall (69), strengthened with the square, round and semi-octagonal bastions, and partially covers the foot of 69. This pottery was also found in Stratigraphical Layers 5 and 6 representing debris from the destruction of Early Period II which covers the area of stupa 64. No specimen was found in Layer 7 which precedes the monuments of Early Period II in this area. Taddei first provisionally dated Early Period II in the third century on the grounds of the palaeographical interpretation of a Brahmi inscription executed on the above mentioned mud floor between stupa 64 and the miniature city wall 69 (Taddei 1972: 383). Parlato later suggested that it showed archaic features without actually being so old and placed it in the fourth century (Parlato 1979: 265). In a more recent personal communication replying to my scepticism about an early date for stupa 64, Taddei suggests, with several factors in mind, that the inscription does seem to be later than the third century. He thinks that 'if the ancient sanctuary - including the area of stupa 64 was destroyed sometime in the seventh century by an Arab invasion (Table 1), the inscription cannot be four hundred years older unless the area of stupa 64 was kept safely closed for some particular reason.' He further opines that, 'although a seventh century date for the site's destruction is only a hypothesis, it matches well with the style of the clay sculptures found in the debris (Layers 5 and 6),' leftovers attributable to Early Period II. He suggests the sixth century for most of the sculptural fragments found in the filling that covered the area of 64,69, and the mud floor with Brahmi inscriptionP4) Table 1 - The Arabic Intrusions into the Kabul-Ghazni Region in the Seventh Century According to Bosworth (1968). AHlAD 311651-52 33/653-54 Al-Rabi' b. Ziyad made campaign to Sistan: Arrival of the Arabic troops in Sistan via Kerman. 'Abd ar-Rahman b. Samura became the Governor of Sistan and attacked the local chieftain Zhunbil in both Zamin Dawar and Zabulistan. 36/657 The troops of Ibn Samura fought against Zhunbil and Kabul Shah. 181 46/666 Al-Rabi' b. Ziyad fought against Zhunbil in al-Rukhkhaj and Zamin Dawar. 511671 'Ubaid Allah subdued the lands of Zhunbil and Kabul. 53/673 Appointed as Governor of Sistan, 'Abbad b. Ziyad invaded Qandahar and Kabul. Yazid b. Ziyad made an extensive campaign toward Zamin Dawar and Kabul, but was defeated by 611681 Zhunbil. 'Abd al-Aziz b. 'Abd Allah drove backZhunbil who had invaded Sistan. 67/686 74/693-94 'Abd Allah b. Umayya fought against Zhunbil and made peace with him. 78/697 'Ubaid Allah b. Abi Bakra fought against Zhunbil but was bitterly defeated. 79/698-99 'Ubaid Allah b. Abi Bakra made peace with Zhunbil. 80/699 Ibn al-Ash'ath who rebelled in Sistan leagued with Zhunbil, but was defeated by the Arabs troops and fled to Zhunbil at Bust. Discussing the inscription, Parlato makes an important point when she states that the region around Tapa Sardar is not purely Indian. The inscriptions at both Tapa Sardar and Tapa Skandar actually reflect an environment on the outskirts of India where old-fashioned or transitional styles of script possibly lived much longer than in their homeland. For example, the Vma Mahesvara inscription from Tapa Skandar bears a distinctive mark of the sixth century style of Acute-Angled Brahmi in India proper. Nevertheless, some of the characters show characteristics of the succeeding Nagari script (Yamada 1972). Thus, relying on a strict palaeographical interpretation for dating is only valid within India itself. The peculiarities encountered in both inscriptions might allow for the possibility that the Tapa Sardar Brahmi script was used sometime before or even at the same time as the strangely mixed characters of the Vma Mahesvara inscription. The hypothesis that the area around stupa 64 at Tapa Sardar was destroyed at the advent of the Arabs into Zabulistan in the sixth and seventh decades of the seventh century also finds support in the sizes of the site's mud bricks (See the Table 2). Mud bricks of dimensions 40 x 40, 38 x 38, and 38 x 19 cm should be regarded as the sizes most common to the monuments south of the Hindukush: for example, Begram III, Shotorak, Tapa Sardar (Late Period), Tapa Skandar II, Tepe Maranjan (later phase), and Saka-all sites which revealed medallion-stamped pottery and round bastions constructed of bricks of the above three sizes. Exceptions to this pattern occur at Tepe Maranjan (earlier phase) where bricks of unusually large dimension (SOx SOx 15 cm) were found in the earlier phase of the qal'a, Fonduqistan (t.p.q. 689) where 38 x 38 cm bricks were exclusively used but no medallion-stamped pottery was reported, and Gul Dara in Logar where medallion-stamped pottery was discovered but no bricks of the usual dimensions were used in construction. The largest of the three common dimensions, 40 x 40 cm, made its first appearance at Begram I, the earliest among the sites in the Kapisi-KabulGhazni region and datable to sometime before Kanishka I if one believes Ghirshman, and continued to be used at Begram II with no bricks of the other sizes suggesting that this size of brick had a much longer life than the 38 x 38 and 38 x 19 cm bricks. Table 2 - Mud Bricks at the Monuments in Kapisi-Kabul-Ghazni Region. 182 Site Size of Brick Begram I 40x40x 10. Begram I (City Wall) 40 x 40 x 10, 12, 14. Begram II 40x40xl0 Begram III (House) 40x40x 10. Begram (Qal'a in the city area) Pakhsa. 38x38x9. 38 x 38 x 9. Begram (Fortress outside of the city) Pakhsa. F onduqistan Gul Dara (Chapel '0') 38 x 38 x 9. 41 x41 x 9.S 40x40x9,10. Gul Dara (D2) Gul Dara (B7) 42 x42 x 10. Gul Dara (20) 42.S x 42.S x 11. Khair Khana (Early) 40 x 40 xII. Khair Khana (Late) Khair Khana (south of 'E') Pakhsa. 38x38x9. 42 x 42 xII. 38x38x9. Saka Fort (Bastion I) 40 x 40 x 10. 38x38x9. Saka Fort (Bastion C) 40 x 40 x 10. 38x38x9. Saka Fort (Court T) 40x40x 10. Shotorak (Between D and F) 40 x 40 x 10. Shotorak (Bastion at D12) 40x40x 10. 38x38x9. 38x38x9. 38 x 38 x 9. Shotorak (Shrine D9) Shotorak (Wall between D3 and Dll) 38x38x9. Tapa Sardar (Early II-A) 42 x 42 x (?). Tapa Sardar (Early II-B) 47 x47 x(?). 46 x 46 x (?). Tapa Sardar (Late) 40 x 40 x (?). 38 x 38 x 9,38 x 19 x 9. Tapa Skandar (Fort A) 40x40x 10. 38x38x9,38xI9x9. Tapa Skandar (Shrine C) 38x38x9. Tapa Skandar (Shrine B) 38 x 38 x 9. Tapa Skandar (Building N) 38 x 38 x 9,38 x 19 x 9. Tapa Skandar (Shrine S) 38 x 38 x 9,38 x 19 x 9. Tapa Skandar (Fortification) SO x SO x IS, 46 x 46 x 10, 12, 40x40x 10. 38 x 38 x 9,38 x 19 x 9. Tapa Skandar (Vaulted corridor) 46x46x 10. 38x38x9. Tapa Skandar (Building F) 38 x 38 x 9,38 x 19 x 9. Tapa Skandar (Building H) 38 x 38 x 9, 38 x 19 x 9. Tepe Maranjan (Qal'a) Tepe Maranjan (Bastion) SO x SO x IS. 40x40x 10. 38x38x9. According to Taddei and Verardi, mud bricks measuring 42 x 42 cm belong to Phase A (earlier) of Early Period II at Tapa Sardar and those of 46/47 x 46/47 cm belong to Phase B (later). As mentioned before, medallion-stamped pottery first occurred at Tapa Sardar during this period which Taddei now suggests existed up to the sixth or 183 even seventh century. In the Late Period, which they date to the seventh-ninth centuries, the three smaller, common sizes were used. This evidence of Tapa Sardar relating differences in brick size to specific stratigraphic levels seems valid only within the site of Tapa Sardar, particularly when viewed against additional evidence available to us at Tapa Skandar and Khair Khana. At Khair Khana bricks measuring 42 x 42 and 38 x 38 cm were both found still intact in 1974 in the walls of the building to the south of Building E.(15) Excavations at the northwestern margin of Tapa Skandar revealed a vaulted corridor, attributable to Period II, which runs through the fortification wall. It contains walls built up of layers of bricks measuring 46 x 46 cm alternating with layers of pakhsa which support the barrelvaulted ceiling made of bricks measuring 38 x 38 cm (Kuwayama 1976: 11-12 and Kuwayama 1978a: 11-12). As an aside, layers of bricks alternating with layers of pakhsa are not only found at Tapa Skandar II but are also used at the qal'a in Begram. Presumably this device further represents a chronological parallelism between the two sites. The use of both the larger and smaller dimension bricks side by side in one and the same walls at the same time at sites like Tapa Skandar and Khair Khana which, as previously shown, must have existed in the very early seventh century, suggests that bricks used in the Early Period at Tapa Sardar, 42 x 42 and 46/47 x 46/47 cm, cannot always be assumed to be earlier than those measuring 38 x 38 cm in the Late Period. This fact indicates the possibility that both of the phases, A and B, in the Early Period II at Tapa Sardar overlapped Tapa Skandar II and Khair Khana II and reinforces the other evidence suggesting that the date for the appearance of medallion-stamped pottery in the area around stupa 64 at Tapa Sardar is much later than one might expect from the palaeographical evidence. It is consistent, as well, with the hypothesis that the Early Period ended with an Arab invasion from Zamin Dawar in the south (Table 1). Further, the evidence may even indicate that the Early Period should be dated quite close to the Late Period which also overlaps Period II at Tapa Skandar-and Begram III, as well. At least, there is no evidence necessitating a long chronological vacuum between the end of the Early Period and the beginning of the Late Period at Tapa Sardar. 6. Two Periods at Tepe Maranjan and Shotorak. Now our scope should be expanded from Begram III, Tapa Sardar (Early and Late Periods) and Tapa Skandar II to two Buddhist monasteries, Tepe Maranjan in Kabul and Shotorak in Kapi§i, which may not be as early as previously supposed. (1) Tepe Maranjan Drawing proper attention to the published layout of Tepe Maranjan which differs at several points from descriptions and photographs given by the excavator, Fussman (Fussman and Le Berre 1976: 95-99) claimed that Tepe Maranjan underwent three successive phases (Illus. 27). In the first and earliest phase it was only a secular building of square plan, called by the excavators a qal'a, externally reinforced by round bastions at 184- each angle and internally comprising one oblong and two square rooms. Fussman maintained that the site later changed into a Buddhist precinct with the building of the stupa to the west of the qal'a and two arched niches in the western face of the western wall of the qal'a for enshrining Buddha and Bodhisattva images. At the last stage of Tepe Maranjan these niches were blocked up with a 90 cm wide wall (Wall A; lIlus. 27) running from the northwestern bastion to the southwestern one and closing off the doorway to the qal'a. The stupa was also enclosed on all four sides by the construction of other new walls of the same width (Wall C and D; Illus. 27). In an attempt to date these three phases, Fussman gave importance to the coin hoard hidden in the wall at the level of third step of a staircase which he claims must have been attached to the northern wall of the oblong room of the qal'a. Fussman, recalling that the vinaya did not allow monks to possess money, attributed the Sassanian silver and Kushano-Sassanian gold coins to the time of secular usage of the site. Accordingly he dates the qal'a, which has round bastions, to the fourth century. He dates the Buddhist phases of Tepe Maranjan to the sixth and seventh centuries for the reason that the seated clay images in the niches predate in style those from Fonduqistan which are not earlier than 689. Despite his proper doubt about the word 'creuser' which Carl used (Carl and Hackin 1959: 7), Fussman eventually followed the excavator who had observed the niche as follows: 'sur la face ouest, une niche de 1 m. 25 de largeur, creusee dans l'epaisseur du mur de briques a la chaux.' Yet I am quite sceptical of the hypothesis that a wall built of mud bricks of such a large size as 50 x 50 x 15 cm was dug out simply to make niches. Moreover, one can hardly imagine the construction of a rough arch with mud bricks on top of a niche cut out of such a strong wall (Refer to the figs. 5 and 6 in Carl and Hackin 1959). Therefore I do not hesitate to propose that the two arched niches E and F-only one of which, however, is shown on the ground plan given in (Carl and Hackin 1959: fig. C)-were made at the same time as the qal'a. Further, existence of the niches on the outer side of the western wall of the qal'a means that the qal'a and the stupa to the west of it were contemporaneous: the niches enshrining the images would not make any sense without the stupa located to the west of them. The round bastions may not have existed with the qal'a in this stage. According to Carl, the qal'a itself was built solely with the 50 x 50 x 15 cm bricks and probably not supported by a masoned foundation, while the round bastions "simply added at at least three angles" were built on a retaining wall masoned with very large river boulders and with 'mud bricks' of much more restricted size than the above sizable bricks. The size of brick employed in the round bastion was not mentioned by the excavator, yet 'bricks of much more restricted size' suggest the usage of either the 38 x 38 x 9 cm square brick or the 40 x 40 x 10 cm bricks, or both, when compared with the known sizes of bricks at other monuments in the Kapisi-Kabul-Ghazni region. From Tapa Sardar and Tapa Skandar those measuring 46/47 cm are known, but these types are too large to match well with what is termed 'much more restricted.' If the qal'a and the bastions were built at the same time, as Fussman thought, bricks of the same size would have been used for both of the constructions and Carl would not have given a description that the bastions 185 was simply jointed to the mud bricks of the previous wall without any direct connection (Carl and Hackin 1959: 7). In the case of at least the southeastern bastion, it was built together in one and the same body with a completely new wall, encasing the previous one. Technologically the addition of such a new wall provides stability: it enlarges the building at that part. A high round tower directly attached to a previous wall would have by no means been stable. Archaeologically the addition testifies to at least two periods not only at Tepe Maranjan but at any other monuments that have such bastions (See Section 7, b). The enclosure walls of the stupa must have been built simultaneously with the round bastions. Although the excavator's ground plan shows a 90 cm wide wall stopping a short way north of the southwestern bastion (Carl and Hackin 1959: fig. C), the photograph and description given by them (Carl and Hackin 1959: figs. 1-6) clearly demonstrate that this wall originally connected the northwestern and southwestern bastions. The excavators apparently intentionally destroyed it in order to reveal the Bodhisattva niche-or perhaps to open up the western entrance to tlle qal'a. This fact was already referred to by Fussman, but he thought that this wall had been added later to the bastions which he considered built at the same time as the qal'a. It is also possible that the wall was partially destroyed by Ghulam Muhieddin Khan who opened up some trenches at the site before the arrival of the French team. My interpretation of the Tepe Maranjan excavations therefore allows for the monument to have undergone the two periods: Period I From the beginning it was established as a Buddhist precinct which comprised the stupa on the west and a square building (qal'a) on the east. To this first period belong the seated images of Buddha and Bodhisattva enshrined in the niches (E and F) on the west face of the square building. Period II - The square building was reinforced with the round bastions and the b. stupa was enclosed with the 90 cm wide walls (A, C and D), the eastern wall (A) of which encased the western wall of the square building and blocked up the niches and the doorway connecting the stupa to the square building. It was necessary to construct this wall connecting the southwestern and the northwestern bastions for stability. Enclosed by walls on all four sides, the stupa found itself situated about 130 cm east of the real center of the enclosed court. The unsymmetrical location of the stupa on the new plan helps to confirm that the stupa had already been there before the enclosure was made. Medallion-stamped potsherds were unearthed in the area to the north and the east of the square building at Tepe Maranjan. Also, Hackin explains the discovery of the top part of a medallion-stamped jar immediately after he describes Carl's discovery of the coin hoard while pulling down the staircase leading to the roof-terrace (Carl and Hackin 1959: 12). Hackin does not give any accurate information about where Carl actually found the jar and how closely the jar had been placed to the hoard. However, judging from the order of his description, it was possibly found during the work of demolishing 186 the staircase. If so, the medallion-stamped pottery may have belonged to Period II, the phase of building the bastions to which the staircase gave access. As for the coin hoard, it is virtually impossible and certainly impracticable to attribute it to a well-defined period. According to Hackin, it comprised twelve gold coins of the Kushano-Sassanian series and three hundred seventy-three silver coins issued by Shapur II, Ardashir II and Shapur III. A vague description, a probable false printing of the photograph of the staircase (Carl 1959: fig. 4) and the absence of the staircase on the plan of the site combine to raise a handful of unsolvable questions as to when the hoard was concealed: Was it concealed at the same time as the building of the wall? Was the wall given its final coating after concealing the hoard and before attaching the staircase? Was it concealed immediately at the same time as the building of the staircase, without coating? The numerous uncertainties about the circumstances surrounding the hiding of the hoard compel us to rely on the clues available to us for establishing a relative chronology not only between different constructions within Tepe Maranjan itself but also between different monuments throughout the Kapisi-Kabul-Ghazni region. A summary of such clues will be provided after the following discussion of Shotorak. (2) Shotorak Surface examinations of the Buddhist monasteries of Top Dara near Charikar, Borji Kafir near Tapa Skandar and several others in Shevaki and Kamari, all unexcavated, tentatively allow us to suppose that they consist of an imposing, lofty stupa lying on the same long median line as a rectangular monastic building as in the case of the excavated site of Gul Dara monastery in Logar (Illus. 27; Fussman and Le Berre 1976: PIs. I and II). Another category of monastic complex as exemplified by Shotorak (Illus. 28), Kham Zargar, Tepe Maranjan (Illus. 28) and Tapa Sardar, on the other hand, has quite a different layout for a Buddhist sanctuary from that of the above group. Weare still not in a position to say whether the difference between the two categories was derived from differing Buddhist sects or whether it represents a chronological difference. In any case, among the latter category Shotorak underwent at least two periods, an earlier and a later one, and the Late Period can be further divided into two phases (Illus. 28). Insofar as the DAFA excavation at Shotorak is concerned (Meunie 1942), all that was made accessible through Meunie's dig is presumably the later Buddhist temple that was completely renovated after the decay of an original precinct, that of the Early Period. Nothing of the Early Period, however, seems to have remained except for some of the schist images well recalling sculptural idioms found in Swat and Gandhara to the east (Meunie, 1942: PIs. XIII, no. 45; PI. XVII, nos. 56 and 57; XIX, no. 62; XXII, nos. 69 and 70; XXVII, no. 81; XXVIII, no. 87) and elements indigenous to faraway Mathura (Meunie 1942: PI.XXXIX, no. 130). A clue for assigning the original construction of the Main stupa found by Meunie, which was later encased with many deep recesses, to the Late Period is its facing masoned with thin slabs of schist which, as I will suggest in the next section of this paper, appears to be a later feature of stupas in east Afghanistan. In any case, the date of this structure built of thin slabs of schist can by no means be 187 consistent with any date potentially attributable to the above schist images. Sometime after the first construction, the main stupa F1 completely changed its outward appearance, decorated on the south, west and north sides with three deep niches, each nich being flanked on each side by a column which was characteristically built of square dressed stones and schist slabs of regularized thickness. Evidently the monuments of the Later Period belong to at least two sub-phases; A (earlier) and B (later), and both of them can be attributed to the same period as Begram III, based on Meunie's report (1942: 67) that the potsherds found in great number at Shotorak did not differ from those found in Begram Bazar (Begram III). Particularly characteristic of Shotorak in Phase B of the Late Period are the deep niches, which were also applied to other smaller stupas, D2, D3, D4 and D6, in Court D (Illus. 28, below). Most of the schist sculptures excavated by Meunie-except for those definitely assignable to the Early Period- may, in reality, have originated in Phase A of the Late Period and have been reused as materials not only for building stupas and small shrines, but also as images for decorating the niches of stupas and shrines of Phase B of the Late Period (Meunie 1942: 27). The type of sculpture most peculiar to Shotorak and used, for example, in the deep niches is the Buddha statue carved together with an aureole rounded or arched at the top (e.g., Meunie 1942: PUII, no. 10 and Pl.X, no.36). If it is admitted that such arched aureoles were carved to fit well with the shape of niches, the group of sculptures with such an aureole may be attributable to the later phase of the Late Period. Also attributable to Phase B are clay sculptures found in the recess made in the eastern wall of Room D4 (Meunie 1942: PI.VII, no. 23), in the recess later encased by the north wall of construction D9 (Meunie, 1942: PI. nos. 17-20) and the clay standing Buddha image attached to the southern side of stupa D2, which looks too large for the size of stupa itself (Meunie 1942: PI. V, no. 16). To the period of construction of the deep niches on the Main stupa the last stage of stupa F6 clearly corresponds (lllus. 28). In the first stage, stupa F6 was one of two stupas of square ground plan flanking the first step of the stairway to the Main stupa of Phase A. The redressing work which took place twice on only the east side of F6 changed this simple stupa into a shrine-like edifice facing to the east, composed of bas-reliefs and sculptures which were possibly brought from other locations within the site of Shotorak (Meunie 1942: 14-15). Especially in the last phase of construction on F6, structures quite similar, in a technical sense, to the columns surrounding the main stupa F1 of Phase B were built on each side of the central socle which supported either a statue or a stele already lost at the time of excavation. Meunie further noted that a small reliquary pot was found 'inside of the socle' of F6 (Meunie 1942: 67). It is a burnished pottery jar of fine black ware which, according to Meunie, shares a common characteristic with pottery goblets and dishes unearthed at Saka Fort which I have already cited as a monument which produced medallion-stamped pottery and was reinforced by round bastions. This kind of column was also raised on the south side of F6, bordering a deep niche which enshrined a seated schist statue of Buddha with flames on both shoulders and with its face mutilated (Meunie 1946: PI. III, no. 10), apparently by the hands of iconoclasts. 188 Four stupas, Dl, D2, D3 and D6 in Court D at Shotorak were most probably all raised at the same time, since the first three are horizontally linked by the same slab pavement (lllus. 28) and the last three have some clear typological similarities in the construction of the niches recessed on the lower bases of the square plinths (lllus. 29). First, all these niches are characteristically deeper than those usually designed under the archades which surround the different stages of the bases of stupas. Second, some of these niches are trapezoidal in elevation except for the destroyed example of D6 and flanked by pilasters which differ from all the other pilasters supporting the stages of the base of the same stupa (Illus. 29: D2, D3 and D6). As will be shown, such deep niches not only help establish the relative chronology of constructions within Shotorak but also provide a clue as to the chronological relationship between Shotorak and other monuments in the Kapisi-Kabul-Ghazni region. Stupa D4 also has this kind of deep niche (Illus. 29: D4). In this case the niches are recessed in each bay of the octagonal stage. The device of placing a deep recess on all sides is similar to the final construction of the central stupa Fl during Phase B. stupa D4 is in the same chronological position as Stupas D2 and D3 since they all are surrounded by similar pavement and have the deep niches. Clay images were enshrined in another kind of niche: Niches D3 and D4 which were recessed into the walls facing Stupas D3 and D4 (Illus. 28). The logical spatial relationship between these two niches and Stupas D3 and D4 indicates a close chronological relationship between them and consequently with Stupas D2 and D6 as well. Wall niches enshrining clay images are not only found at Shotorak but also at Tepe Maranjan of Period I and at Tapa Shotor in Hadda (Tarzi 1976 and Kuwayama 1987b), not to mention Fonduqistan. To a phase post Late Period belong the constructions D9, Dll and D12 (Illus. 28). The latter opened onto the Room Dll. It blocked up Niche D3 and disturbed the pradaksina-patha on the south side of stupa D3. The eastern face of the wall containing Niche D3 is not parallel with the western face, but exactly parallels the other long side of Room D 11 which is horizontally aligned a little to the east, reinforced by circular Bastion D12. Since Dll and D12 are clearly askew in relation to the structures to the west and since D9 is structurally associated with D 11, they must all be later additions belonging to a more recent phase. Belonging to the same phase may be Bastion K located in the very southwestern part of the site (lIlus. 28). Like D 12, it is also a round bastion and, although not mentioned in the excavation report, from personal observation it was clearly added to a previous structure in conjunction with a new encasing wall, just as can be seen at the southeastern bastion at Tepe Maranjan (Illus. 27) and at the southeastern bastion at Inner Fort (A) of Tapa Skandar (lIlus. 23). The above evidence indicates that a stupa recessed with a deep niche may have been a later device in the Kapisi-Kabul region. At Tepe Maranjan we have seen that the stupa with a deep niche on the north side of the plinth existed from the earliest phase of the site. If we are right in thinking that a stupa with deep niche may represent a chronological parallelism between the sites, the earlier period of Tepe Maranjan could possibly be dated to the same time as Phase B (later) of the Later Period of Shotorak. 189 The structures reinforced by circular bastions at both Tepe Maranjan (southeastern bastion) and Shotorak (Bastion K) would then be one step later than this phase (Section 7, h). In this connection, the building with bastions at four angles found within the fortification walls at Begram should be reexamined. Ghirshman regarded this building as belonging to Begram III simply for the reason that it was founded directly above Room T of Begram II. However, as Hackin already suspected, the existence of a bastioned qal'a within a city area already protected by fortification walls is redundant. This redundancy may suggest that the qal'a is later than the occupation of Begram III. If this hypothesis is correct, round bastions appeared in the Kapisi-Kabul-Ghazni region as early as the period between Begram II (overlaps Tapa Skandar I) and Begram III (overlaps Tapa Skandar II) and continued to be used after Begram III, while the medallion-stamp decoration was first introduced in Begram III and also continued to exist after it, since a medallion-stamped potsherd was found in a room of the bastioned qal'a at Begram (Meunie 1959: 104). The defacement of a seated Buddha settled into the south side of F6 (Meunie 1942: PI. III, no. 10) suggests that the image was intact at the time of the destruction of the sanctuary by the hands of iconoclasts. Usually iconoclasts have been assigned to no other than the Muslims. It is, however, not clear how aggressively the early Arabs in the Kapisi-Kabul-Ghazni region took action against the icons and whether they really destroyed the Buddhist or the Hindu images in the early period of their invasions as part of their initial attempt to settle their religion in the pagan lands. Highly suggestive as to this question is the recent analysis of the religion and society in early Arab Sind done by Maclean (1989: 22-82). Accepting the hitherto disputed views on Buddhist conversion to Islam, he persuasively concludes that not Hindus who were based on the agriculture but only Buddhists converted into Islam since Buddhist commercial activities, which had supported Buddhist establishments in Sind, were supplanted by the Muslim traders who were also based on the long trade. If so, there could be the possibility that Buddhist images were defaced by the converted Buddhists themselves as their confession of faith, although it remains only an imagination whether they did so at their own will or by compulsion. This hypothesis also suggests an answer to the question why the Hindus survived longer than Buddhists in the Kapisi-Kabul-Ghazni region, as politically shown by the rise of the Hindu Shahi Dynasty in the ninth century (MacDowall 1968). 7. Chronological Criteria for Monuments in the Kapisi-Kabul-Ghazni Region. It is my thesis that archaeological features shared in common between monuments in the Kapisi-Kabul-Ghazni region can be used as a basis to address their chronology. I have stressed the importance of medallion-stamped pottery and round bastions on numerous occasions. In this paper I have also stressed the importance of those round bastions which were clearly added to previous structures, stupas with deep niches, and the dimensions of mud bricks used in construction. Other criteria are also available. A 190 consolidated list and a tentative comparative chronology of the monuments follow: a. Medallion-stamped pottery: The occurrence of medallion-stamped pottery within the context of excavations at Begram III, Tapa Skandar II, Tapa Sardar of Early Period II and Late Period has already been discussed in this paper with a special reference to Wardak. Its discovery at other sites in the Kapisi-Kabul-Ghazni region has also been briefly mentioned: Shotorak (Meunie 1942) (See Illus. 25: no. 31 and Illus. 24: no.14), Kham Zargar (Illus. 25: nos. 6, 12 and filIus. 24: no. 3), Tepe Kalan (Carl 1959C: 132, no.22), Tepe Maranjan (Carl, 1959c) (Illus. 25: nos. 5,7, 11, 13, 14, 15, 18,24,32,40,42 and Illus. 24: nos. 18,20,31,35,40,46,48,52), Saka Fort (Carl 1959a) (Illus. 25: nos. 1, 6, 10,21,24,26,27,53), Gul Dara in Logar (Fussman 1976) (Illus. 25: nos. 3,4, 10,45, and Illus. 24: no. 12), and Jagatu in Wardak (Scerrato 1967: 1 If. and figs. 47-48). b. Round Bastions and Specific Device of Annex to Existing Building: All sites mentioned above with medallion-stamped pottery also had constructions with round bastions-with the exceptions of Kham Zargar and Jagatu. Further, examples of such bastions having been added to a previous building have been found at Tepe Maranjan (southeastern angle of the rectangular building), Tapa Skandar of Period II (southeastern corner of the Inner Fort), Shotorak of the Later Period (in the extreme southwestern area of the site), Saka Fort (northeastern portion of the northern tumulus), and Gul Dara in Logar (eastern corner of the monastery). In each of these cases the bastions were attached to a right-angled corner of the previous building, but not directly to the older building itself. Instead, they were built together in one and the same body with completely new walls which encased the previous ones. Thus each site must have at least two phases. The examples at Saka Fort and Shotorak are not described as such in the excavation reports but were visible at the time of my visits to the sites. Peculiar to all these examples is the outer line of the new wall: it connects to the round bastion via a short interruption where it turns inwards at an obtuse angle (Illus. 27: B). c. Stupas Constructed of Schist Slabs of Regularized Thickness: This particular masonry, broadly attibutable to the seventh-the early eighth centuries, is common to the Main Stupas at Tepe Maranjan, Shotorak of Phase A (earlier) of the Later Period and Fonduqistan and to some of the smaller stupas at Tapa Sardar of Early Period II around the Main stupa and on the terrace of the northwestern slope (Taddei 1968: figs. 34-37;1972: fig. 16; 1978: fig. 5.51). Some of the small stupas in the Courts F and D at Shotorak are also in this category. One may regard Stupas 11, 20, 21 and 64 at Tapa Sardar as the same type as Stupas D1, D2, D3, D4 and D6 of Phase B of the Later Period at Shotorak, the stupas at Tepe Maranjan and Fonduqistan. d. Stupas with Deep Niches: Four examples of stupas with deep niches have been described at Shotorak: D2, D3, D6 and D4. They are found at other sites as well. The deep niche on the lowest square plinth at Tepe Maranjan may be related to this particular category. But an undoubtedly much better example is stupa 20 at Tapa Sardar of the 191 Early Period II, located on the southwestern side of the Main stupa. These two examples outside Shotorak were not covered with a flat ceiling but with a barrel-vault masoned with very thin slabs and they were not flanked by pilasters. As an aside, particular attention should be drawn to the barrel-vault of the niche in stupa 20 which has keystones at its apex. This kind of vault with a true arch is never found elsewhere in the south of the Hindukush, as far as I know. In Gandharan monasteries barrel-vaulted ceilings were built using traditional corbelling techniques, as mentioned long ago by Foucher (1905). In the regions west ofJalalabad they were built in successive courses of mud bricks-without keystones-inclined at a 45 degree angle in relation to the side walls against which they rested (Kuwayama 1987b: 158). e. Octagonal Constructions: Among the above stupas at Tapa Sardar built of schist slabs of regularized thickness, stupa 64 (on the lower terrace) and stupa 11 (at the east corner of the Main stupa) have octagonal stages laid on their lowest square plinths. The octagonal design is very important in the context of later Buddhist monuments in· east Afghanistan and Northwest Pakistan. It was also used on the second plinth of small stupas at Shotorak (D4) and at Tepe Kalan (TKI21) in Hadda (Barthoux 1933: 101-102) and in the design of some of the Bamiyan caves. As I elsewhere stated (Kuwayama 1987 and 1990), the prosperity of Buddhist activities, including the making of cave temples, in the kingdom of Bamiyan burst on the scene after Gandharan Buddhism had decayed parallel with the withdrawal of the Hephthalites from the Northwest which cannot have been earlier than the middle of the sixth century. The octagonal caves in Bamiyan are thus attributable to sometime after this date (See also Klimburg-Salter 1988: 305-312). In addition, Tapa Sardar, Shotorak and Bamiyan-as well as Tepe Maranjan which has, however, no octagonal plan-share sculptural idioms including clay as raw material and certain stylistic peculiarities. Stupas built with octagonal plans in the region south of the Hindukush, therefore, should not be given so early a date as the fourth or the fifth century. f. Cinerary Urns: Another strong common feature between these sites is the custom of depositing cinerary urns. Besides the urn found at Fonduqistan underneath the clay sculpture of a princely couple in Niche E and containing the Arab-Sassanian coins that give the terminus post quem at least to the above sculpture, a pot containing calcified human bones was found under the low bench for clay statues placed along the outer face of the southern enclosure wall of the stupa at Tepe Maranjan of Late Period (Carl and Hackin 1959: 8). It was also reported to have been found at Tepe Kalan of Koh-e Pahlawan, although we are not informed about where it was found (Carl 1959c: 129, No.9). At Shotorak, a cinerary urn containing calcified bones was also unearthed in Chamber D8, placed 110 cm below the floor. According to Meunie, this pottery jar is similar in its fabric to a huge number of potsherds unearthed at Shotorak of the same category as the one from Begram 'Bazar' (Begram III) to which the medallion-stamped pottery belonged. 192 g. Brick Dimensions: Sizes of bricks have been discussed in the context of the chronology of Tapa Sardar (Section 5) to include a list of sizes for monuments in the Kapisi-Kabul-Ghazni region (Table 2). In particular, a comparison of sizes proved useful for proposing a relative chronology between Tapa Skandar II, Khair Khana II and Tapa Sardar of Early Period II (Tables 1 and 3). h. A Comparative Chronology of the Monuments in the Kapisi-Kabul-Ghazni Region: In the below chart I placed the beginning of Begram III in the middle of the 6th century. The kingdom of Kapisi was first mentioned in the biography of Gandharan monkJinagupta in the second volume of the Tang Gaoseng zhuan. Jinagupta as a member of Buddhist mission left Gandhara for Kapisi after his summer retreat in 554 and stayed there one year. The biography says that the Kapisian king very eagerly wanted his master-monkJinayasas to give a series of lectures supported by the king himself and that the advantages which Jinayasas and Jinagupta received from the king were extensive. The Biography also refers to their itinerary that further extended from Kapisi through Bamiyan to the Hephthalite headquarters where Jinagupta encountered with the first West Turkish attack of the Hephthalites in 555. Taking into considerations the rise of the local political power, the Xingnie dynasty (Section 8), and the decline of the Hephthalites in both Tokharistan and the Northwest, it may not be so far from truth that Begram III was initiated in the middle of the 6th century (Kuwayama 1987a and 1990). Table 3 - A Tentative Chronology for the Monuments in the Kapisi-Kabul-Ghazni Region. (AD) 550 600 650 700 750 800 850 (?)III (Bazar) ....................... Qal'a/Fortress Begrarn/ Kabul! Shotorakl Early(?) (?)Late (A) ..... Late (B)........ Tapa Skandar I(?) (?)II ........................................................................... ? Khair Khana/ (?)I. ..... .III. ................................................................ ? Tepe Maranjanl Tapa Sardar/ .Post Late (D9, DII, D12) Early ............................. Late ........................... EarlyI(?) (?)EarlyII .............. .ILate(A) .................. Late(B) .................. ? Fonduqistanl 689(t.p.q.) ................... ? 8. From Kapisi to Kabul Last but not least, the historical context should be discussed. After the victory over the Hephthalites in Tokharistan and the Hephthalites' consequent decay in the Northwest in the middle of the sixth century, the West Turks only occupied the lands most useful to the nomads in Tukharistan. They never passed through the Hindukush either southwards or southeastwards, but stayed within Tokharistan with their main qishlaq at Huo (phonetically restituted as *War) in the Da Tang Xiyu ji or Warwaliz in Islamic sources, 193 which is evidently attributable to Bala Hisar near Qal'a-e Zal on the south bank of the AmuDarya. By 628/629, the time of Xuanzang, they had already subjugated the local principalities covering Balkh to the west of *War, Talaqan to the east of Khanabad, and the Khost and N ahrin districts on the highland which extends to the east of the middle valleys of the Surkhab and to the north of the Andarab at the foot of the Hindukush. It is through the latter area that they seem to have established friendly relations with Kapisi on the southern skirts of the Hindukush. Turkish relations with Kapisi are suggested by a description in the Da Tang Da Ciensi sanzangfashi zhuan that the Tardu Sad, a Turkish ruler residing at *War, promised Xuanzang an official escort to accompany him to KapisiY6) Particularly important for the history of the Kabul valleys was tllis policy of the West Turks never to penetrate southeast of the Hindukush, unlike their predecessors the Hephthalites, the Kidara Kushans and the Kushans. The withdrawal of Hephthalite power from the Northwest and the unconcern of the West Turks toward the Northwest naturally stimulated local powers to rise. The vast regions to the west of the Indus fell into the hands of a local dynasty having originated from a ruler called Xingnie. The historical sources regarding the foreign relations of the Tang Dynasty such as the Jiu Tangshu (Vol. 198), the Tang huiyao (Vol. 99) and the Xin Tangshu (Vol. 222a) describe that in Kapisi in 658 there was King Hejiezhi, the twelfth king on the line originated by Xingnie, and the Cefu yuangui (Vol. 970) also tells that Hejiezhi ascended the throne on his father's retirement in 658. According to other information in the above sources, this Kapisian dynasty governed over about eleven principalities-this number matches well with that given by Xuanzang-between Kapisi and Gandhara along the Kabul River Witll Kapisi as its summer residence and Gandhara as its winter one, and the dynastic capital city was called Ehe, which can possibly be identified with Kapisi-Begram of Period III which came into existence after a long vacuum following the Period II of Begram which is probably contemporary with the Kushan rule in India. This local kingdom to the west of the Indus appears in Chinese sources as either Jibin or, more properly, Kapisi, but I prefer the Pan-Kapisi Kingdom to either of the Chinese denominations. In the regions to the east of the Indus, on the other hand, the new born Karkota Dynasty of Kasmir became powerful extending its political influence to the Chenab in the southeast and to the Salt Range in the south. This political environment from the middle of the sixth century onward caused a drastic change in the trade routes connecting the north with the south through the Hindukush, or the Great Snowy Mountain as it is called by Xuanzang. Before that time the Karakorum highway had connected Gandhara with Tokharistan to the Northwest and with the Tarim Basin to the northeast. This route had flourished throughout several centuries since at least the Kushans and brought continuous prosperity to Gandhara and its Buddhist establishments. It was eventually replaced by a new route through the western Hindukush which became secure under the West Turkish hegemony. This change was really an epoch-making event by virtue of its atrophying effect on Gandhara and its promotion of Bamiyan and Kapisi as trade centres stimulating their sudden prosperity in and at the south foot of the Hindukush respectively (Kuwayama 1987a). It 194 was the first time in history that the plain of Kapisi was firmly connected with the north of the Hindukush. On this newly opened highway, therefore, the fashion current in the region south of the Hindukush for pottery with medallion-stamp decoration and for round bastions could suddenly be conveyed to the north. Relevant to this point are the strongholds reinforced by round bastions in the Hindukush recently published by the French team and their discovery of a sherd bearing a medallion stamp found at Ruin 6 near Gumbad-i Kalu between Bamiyan and the Hajigak Pass (Le Berre 1987: 78,101), which was on one of the routes toward Tokharistan, the location of Kohna Masjid. The Kapisian missions came to the Tang court in the years 619, 629, 637, 640, 642, 647, 648, 651, 652, 653, 654, 658, 670, 671, 692, 710, 719 and so on (Table 4). We do not know how long the Kapisian king Hujiezhi ruled after his accession in 658 and therefore whether he sent such missions as those in the years 670, 671, 692 and 71 0. Yet the year 719 is apparently the year of the first tribute of the Turkish king of Kapisi, to whom the Tang court gave on this occasion a new Turkish title, Qaradachi Tegin. In the years between 658 and 719 something politically important certainly happened in Kapisi. The event was most probably the Turkish usurpation of the local Khingal Dynasty. Twenty years later, in 738, a Kapisian king with the title Khorasan Tegin Shad sent an envoy to the Tang court and reported that his son, FROM KESARO/HROM KESARO (Roma Kaisar), had succeeded him since he had become too old. After that year all of the envoys from Kapisi came to Changan with Turkish official titles. The aged Khorasan Tegin Shad in 738 must have been the first Kapisian Turkish ruler and the usurper. Table 4 - The Kingdom of Pan-Kapisi in Chinese Sources. (AD) 619 The First Mission from Kapisi to the Tang Court (Jiu Tangs/nt, Vol. 198. In Chinese sources such as ]iu Tangshu, Xin Tangshu, Tang huiyao, Ton dian, Cefu Yuangui, Kapisi appears as Jibin). 629 637 Xuanzang's visit to Kapisi. The Second Kapisian Mission to the Tang Court (Jiu Tangshu, Vol. 198). The Third Kapisian Mission to the Tang Court (Ton dian, Vol. 192, 8.). The Fourth Kapisian Mission to the Tang Court (Cefu yuangui, Vol. 970). 640 642 The Fifth Kapisian Mission to the Tang Court (Cefu yuanglli, Vol. 970). The Sixth Kapisian Mission to the Tang Court ('jiu Tangshu, Vol. 198, Tangshu, Vol. 221a and Ceflt yuangui, Vol. 970). 647 The Seventh Kapisian Mission to the Tang Court (Cefu yuanglli, Vol.970). 648 The Eighth Kapisian Mission to the Tang Court (Cefu yuangui, Vol. 970). 651 The Nineth Kapisian Mission to the Tang Court (Cefu yuanglli, Vol.970). 195 652 The Tenth Kapisian Mission to the Tang Court (Cefa yuangui, Vol. 970). 653 The Eleventh Kapisian Mission reported the accession of a new king to the Tang Court (Cefa yuangui, Vol. 970). 654 The Twelfth Kapisian Mission to the Tang Court (Cefo yuangui, VoI.970). 658 The Thirteenth Kapisian Mission reported that the Kapisian king called Hejiezhi was the twelfth king on the line originated by Xingnie. On this occasion the Tang Court appointed this king to a GovernorGeneral Xiuxian and set up the Government-General Xiuxian at the capital city of Kapisi (Jiu Tangshu, Vol. 198). 661 The Tang Court established the Protectorate-General Anxi at Kucha which governed over the sixteen states in the Western Region including Kapisi at which capital locally called EM was again set up the Government-General Xiuxian (Jiu Tangshu, Vol. 198, Tang huiyao, Vol. 99 and Zizbi tongjian, Vol. 200). 665 Xuanzhao's Visit to Kapisi (Da Tang Xiyu quifa gaoseng zhuan by Yijing, Vol. 1). 670 The Fourteenth Kapisian Mission to the Tang Court (Cefu yuangui, Vol. 970). 671 The Fifteenthe Kapisian Mission to the Tang Court (Cefu yuangui, Vol. 970). 692 The Sixteenth Kapisian Mission to the Tang Court (Cefu yuangui, Vol. 970). 710 Xieyue (Zabulistan) and Jibin together sent their missions to the Tang Court (Cefo yuangui, Vol. 970). aibin indicates Kabul after the Turki Shah's usurpation, but Kapisi is used here for Kabul only for convenience). 719 The Seventeenth Kapisian Mission to the Tang Court. On this occasion the Tang Court admitted the king of Kapisi as the Qaradachi Tegin (Jiu Tangsbu, Vol. 198). 720 The Eighteenth Kapisian Mission to the Tang Court (Cefu yuangui, Vol. 971). A mission also came from Zabulistan, too. In the ninth month, the Tang Court sent an envoy to Kapisi and Zabulistan and admitted Qaradachi Ilt1ib1ir Zabul as the king of Zabulistan and Qaradachi Tegin as the king of Jibin (Kapisi or Kabul) (Cefu yuangui, Vol. 964). 727 Huichao's visit to Jibin and Zabulistan before 727, or probably in the third decade of the sixth century. 738 The aged king of Jibin, Khorasan Tegin Shad, sent a memorial to the Tang Court on the occasion of the Nineteenth Mission and asked the Tang Emperor to admit to abdicate the throne in favour of his heir Fulin Jipo (FROM KESARO/HROM KESARO). The Emperor accepted it and appointed Fulin Jipo as the king of Jib in (Cefo yuangui, Vol. 964). 745 On the occasion of the Twentieth Mission the Tang Emperor admitted that Bofuzhun, son of Fulin Jipo, had succeeded to the king ofJibin and Uddiyana (Jiu Tangshu, Vol. 198 and Tang buiyao, Vol. 99). 196 746 The Twenty-first Kapisian Mission to the Tang Court (Cefa Yllanglli, Vol. 971). 748 The Hephthalite and the Twenty-second Kapisian Missions to the Tang Court (Cefu Yllanglli, Vol. 971). 750 The Twenty-third Kapisian Mission arrived at Changan and asked for Tang official mission to Kapisi (Narratives ofWukong). 751 Emperor Xuanzong sent an envoy of returning courtesies to Kapisi headed by Zhang Taoguang who was accompanied by about forty officials including Che Fengzhao (Wukong) (do.). 753 Zhang Taoguang arrived in Gandhara and returned to China within this year, but Che Fengzhao remained in Gandhara because of his illness and later in Gandhara became a monk named Dharmadhatu. Dharmadhatu again received the complete precepts when back to China and was renamed Wukong (do.). The Zabulistan and the Twenty-fourth Kapisian Missions to the Tang Court (Cefu yllangui, Vol, 971). 758 The Gandharan king sent an envoy headed by Tafu Tegin and Chebishi Tarkhan to the Tang Court (Cejityuanglli, Vol. 976). This man is possibly identifiable with Barha Tegin in the Ta'rikh ai-Hind by Biruni and Aye in the Wang Wu- Tianzhuguo zhuan (Record of Travels to India of the Five Districts) by Huichao. Based on his visit in the third decade of the eighth century to the regions between Jibin and Gandhara, which were ruled by the Turkish king with his Turkish troops in hand, Huichao explained how these Turks had succeeded to political power there as follows: This country had originally been ruled by a king of Jibin (Kapisi) under whom a Turkish king Aye was subject with his tribesmen and his troops; when the Turkish forces became more active afterwards, he killed the king of Jibin and became chief of the kingdom; accordingly this kingdom abutted on the frontier with the supreme king of the Turks [of the northern land], but this usurper was not in the same line as the northern Turks. A story about the rise of the Turks in the south of the Hindukush described by Biruni sits well with that by Huichao: The Hindus had kings residing in Kabul, Turks who were said to be of Tibetan origin; the first of them, Barhatakin (Barha Tegin), came into the country and entered a cave in Kabul, which none could enter except by creeping on hands and knees .... Some days after he had entered the cave, he began to creep out of it in the presence of the people, who looked on him as a newborn baby; he wore Turkish dress, a short tunic open in front, a high 197 hat, boots and arms; now people honoured him as a being of miraculous origin, who had been destined to be king, and in fact he brought those countries under his sway and ruled them under the title of a shahiya of Kabul. (Sachau 1888: II, 13-18) In view of the fact that no written evidence exists indicating that the West Turks in Tokharistan invaded Kapisi and the fact that Huichao clearly says that the Turks to which the usurper of Kapisi belonged were not the same as those in the north, we have excellent reasons to suggest that these Turks were quite different from those of the north-i.e., not in the lineage of the West Turks-and that they had long resided in a region south of the Hindukush, even earlier than Xuanzang who noted that a Turk was the ruler of Fulishi-sthana. In the time of Xuanzang, a Turkish ruler residing south of the Hindukush was exceptional. The district called by Xuanzang Fulishi-sthana was, according to the Da Tang Xiyu ji, situated about five hundred Chinese miles to the north of Jaukuta (Jawudasthana or Zabulistan) and between Zabulistan and Kapisi, measuring about two thousand Chinese miles from east to west and about one thousand from north to south. It is also identifiable, from the viewpoint of geographical interrelationships, with Hushijian in the Western Region chapter of the Tangshu which is located to the north of Zabulistan and to the southeast of Bamiyan. Notwithstanding that both Fulishisthana and Hushijian are quite unknown in other documents, it is reasonable to identify them with the district around Kabul, which may have included the areas from Maidan, or even Wardak, in the west to the Lataband Pass in the east. This tentative conclusion about the situation of Fulishi-sthana would suggest that the Turkish minority under the Wusan (Khworasan) -Tegin Shad had been subject to the local king of the Xingnie Dynasty but after increasing their power in and around Kabul they finally replaced the Xingnie Dynasty itself, long before Huichao's visit to the region. In this connexion something about Kabul should be mentioned. The strange fact is that the name Kabul appears only in Islamic sources without any mention of Kapisi in relation to the Arabic intrusion from Zamin Dawar northwards. The Chinese sources never mention the area, although it is sometimes suggested that the Chinese denomination Jibin might cover not only Kapisi but also Kabul. A solution to this enigma is perhaps available within the context of the political changes just described: the political centre in the western Pan-Kapisi kingdom was shifting from Kapisi-Begram III southward to Kabul as the Turks in the regions around Kabul became increasingly powerful in the last years of the seventh century. This would not be only a hypothesis if the qal'as in and outside of Begram III could be proven to be one step later than the closing years of the city Begram III. In the qal'a within the city area there were found potsherds with medallion-stamp decorations, while Fonduqistan, datable to around the end of the seventh century, produced no such pottery. Therefore I very tentatively and modestly date the disappearance of this kind of pottery in the Kapisi-Kabul-Ghazni region to the early eighth century. This conclusion resists the seventh-ninth century dating given by Taddei to the Late Period of Tapa Sardar in which medallion-stamped pottery reportedly was still used. However, 198 some features of the Late Period clearly appear to belong to a phase after the early eighth century. How to reconcile this conflicting evidence? The contradiction would be resolved if the period should prove to have two sub-phases: an earlier phase with medallionstamped pottery which may architecturally be represented by the row of chapels-i.e., Chapels 23, 17,37 and Vihara; 54-located to the southwest of the Main stupa and a later phase after the disappearance of the pottery, to which may belong the rows of stupas and thrones-i.e., 1-11,22,24, 19,38, and 42-regularly placed on the southeast and southwest sides of the Main stupa (Taddei and Verardi 1985: fig. 1 and 117-120). Investigation of this hypothesis might shed further light on the chronology not only of Tapa Sardar itself but also of the other monuments in the Kapisi-Kabul-Ghazni region which reveal the shared archaeological features this paper has attempted to document (Section 7, h). 199 XII Zhun, Aruna and Two Superimposed Shrines at Khair Khana, Kabul 1. Buddhists in the North and Non-Buddhists in the South The density of remains of Buddhist sanctuaries in the area around the capital of the kingdom of Kapisi, or in the northern part of Kapisi, is consistent with the distribution of the sites given by Xuanzang. According to Xuanzang's biography by Huili and Xuancong, the Shaluojia monastery built in honour of the Kashgarian hostage is at the northern foot of a hill about three or four Chinese miles to the east of the capital of the kingdom of Kapisi. On the top of the hill to the north of this monastery are caves where the hostage practiced samadhi, and about two or three Chinese miles to the west of the caves, there is an AvalokiteSvara image on the high mountain. Further northward on the top of the Great Snow Mountains, about two hundred Chinese miles to the northwest of the royal city, there is a lake of the naga king, by which stands a stupa containing the bone-flesh sariras of the Tathagata. To the northwest of the royal city is a large river, which may reasonably be identified with the Panjshir River after the confluence with the rivers Ghorband and Salang. On the southern bank are three Buddhist monasteries: two built by the previous king and the other by his queen. To the southwest of the capital is Mount Filsara, on the great solid rock of the summit of which is a stupa called Filsara. To the north of the stupa is also a sangharama. Then Xuanzang says that about thirty Chinese miles to the southeast of the capital is the Rahula sangharama and a stupa with a height of about a hundred Chinese feet. Such a distribution of Buddhist monuments in the early seventh century corresponds well to the density of Buddhist ruins and stray finds in the vicinity of the third and last occupation of Begram site such as Shotorak, Qol-i Nader, Top Darah, Kham Zargar, Jabal as-Saraj, and Paitava (Illus. 19). In fact, no Buddhist sanctuary is mentioned by Xuanzang to the south of the capital except the Rahula sangharama isolated from the northern group. Such isolation also corresponds well to a still standing stupa called Borj-e Kafir that is the one and only site which still exists in the south far away from Begram. Contrary to this distribution of Buddhist monuments in northern Kapisi, as found both in Xuanzang's description and in archaeological reality, are two clearly Brahmanical monuments referred to by Xuanzang in southern Kapisi: a town called Xibiduofalaci, about forty Chinese miles to the south of the royal capital, and Alunuoshan, about thirty Chinese miles to the south of Xibiduofalaci. In this context special mention should be made of the two non-Buddhist monuments, Tapa Skandar and Khair Khana. At Tapa Skandar the Kyoto University Archaeological Mission unearthed in the center of the site 200 in 1970 the marble Vma Mahesvara statue seated on the bull Nandin together with Skanda, the pedestal of which contains an inscribed hymn in Acute-Angled Brahmi dedicated to Mahesvara (Kuwayama 1972). On the other hand, the discoveries of the marble Surya images, one by Hackin in 1936 and another by chance in 1980 (Bernard and Grenet 1981), have been well known at Khair Khana. In modern times not a few marble Brahmanical images have been discovered especially in the area to the south of Mir Bach Kot, or the southern part of ancient Kapisi. Xuanzang usually points out, when describing the heretics in the nearby countries, that non-Brahmanical people live mixed with the rest of the population without isolating themselves from the others. In the case of Kapisi, however, no mention is made of how and where they are living. From the viewpoint of the above distribution, it seems therefore that in Kapisi the Buddhist and Brahmanical populace did not live in the same area, but Buddhists were mainly in the north and the non-Buddhists in the south. In addition, such a coincidence of distributions known from both Xuanzang's description and modern archaeological surveys may reveal that the dates of most of the remains in the ancient Kapisi area could converge within a limited time span which includes the time ofXuanzang's visit to Kapisi, 629. 2. Aruna and Zhuna On Xuanzang's route from Changan to Nalanda, Kapisi is the last country before entering India, the record of which is placed at the end of the first volume of the Da Tang Xiyuji. In Kapisi Xuanzang first mentions non-Buddhist deities, the shrines and supporters being represented by the 'deva' shrines and heretics that include ascetics such as the Pasupatas with naked bodies smeared with ashes, the Kapalikas, and the Digambaras. Thus Brahmanism had expanded to Kapisi, east Mghanistan in the early seventh century. The pasupatas and the supporters of this Sivaite sect most probably based themselves in the town Xuanzang calls Xibiduofalaci, which should be interpreted as the town with a shrine dedicated to Svetavara, probably a corrupt form of Svetasvatara, and identifiable with an actual site, Tapa Skandar or Bala Hisar near Sarai Khwaja, Mir Bacha Kot, about thirty kilometers to the north of Kabul and some fifty kilometers south of Begram. The marble Vma Mahesvara statue found in a shrine of this site has correctly been associated with the seventh-eighth century Brahmanical statues discovered at other sites in the Kabul Valley and its vicinity (Kuwayama 1976). Another site of non-Buddhist character referred to by Xuanzang is Alunuoshan, which can be taken as Mount Arona. Xuanzang locates it about thirty Chinese miles to the south of Xibiduofalaci. About this mountain there was told a legend which suggests the conflict between two heretic sects. The following is a translation based on Beal's and modified by the present author: Thirty Chinese miles or so to the south of the town with a Sveta[s]va[ta]ra shrine inside is a mountain called Alunuo (Aruna). The ridges of this mountain 201 are precipitous, its rocky valleys being dark and deep. Each year the peak increases in height several hundred Chinese feet. As soon as it dimly faces Mount Zhunahira «*dz'iunahila) in the kingdom of *dzaukiuta (>dzaul[i]/dzaul/=Zabul in the Islamic geographic sources), it falls down. According to local tradition, the deity Zhuna first came from far to this mountain desiring to dwell on it, but the original deity of the mountain trembled with anger and shook the valleys. Zhuna said, "As having no wish to live together, you might be thus trembling. If you only entertained me as a guest, I would confer on you great riches and treasure. Now I go to Mount Zhunahira in Zabul. Whenever the king and his ministers may offer me their tributes every year, then you shall stand face to face with me". Therefore Mount Aruna increases its height, and as soon as it stops doing so, it crumbles down at the top. Zhuna thus could not stay on the mountain in Kapisi and went to the kingdom of Zabul. This country recorded by Xuanzang in the twelfth volume of the Da Tang Xiyuji is one of the largest countries to the south of of Kapisi. According to his account, although the people in this country worship various gods, Buddhism is the most honoured. Among the heretics the most powerful are those worshipping Zhuna. Xuanzang says as follows: There are several dozen 'deva' temples, and the sectarians of various denominations dwell together. Among those counted, the Tirthakas are many in number and very powerful. They follow Zhuna «*dz'iuna). In the past this deva came from Mount Aruna of Kapisi to live on Mount Zhunahira on the southern border of this country. He showed dignity and gave the people happiness, or perpetrated violence and evil. Those who believed in this deity attained their wishes, whereas those who looked down on him received misfortune. Therefore all people, both from far and near, worshipped him; and all people, both from the upper and lower classes, held him in reverence. People from neighbouring countries and of different manners and customs, kings and courtiers assemble together every year on an auspicious day of that year, and sincerely devote themselves by presenting gold and silver as well as rare treasures, or by competitively dedicating cattle and horses as well as domestic animals. Accordingly, the floors were full of gold and silver, while the valleys were full of sheep and horses. Nobody has any intention to steal them, but solely tries to offer such objects. The heretics who intently serve the deva practice asceticism, and then the deva gives them magical power in return. The heretics effectively perform magic to treat illness, by which many recover completely. 3. Cao, i.e., Kapisi 202 It is not hard to take what was enshrined on Mount Aruna as the image of Surya, a deity in close connection with Aruna, the personification of dawn in the form of a driver of the chariot of the sun, although Zhuna is of an uncertain character (Simms-Williams 1997: 17-19). The story is simple, but suggests a historical background: there happened a conflict between the two religious groups, the Surya followers and the Zhuna worshippers. The whole story, with a special stress on the relationship between a deity of Mount Aruna in Kapisi and Zhuna in Zabul, allows for two interpretations. The first standpoint is that what Xuanzang tells about may be accepted as historical fact: when the Zhuna group came to Kapisi the mountain had already been occupied by the group of the Surya followers. Moul!l.t Aruna had been a center of the sect who honored such a god, when the other group worshipping Zhuna came there to enshrine the god. But the former group did not accept the new one. Following a conflict between them, the latter had to shift its base to Zabulistan. The other interpretation is based on a view that the reverse is the case. The incident which the story talks about happened before Xuanzang's visit. When he first reached Kapisi, Zhuna had already gone to reside in Zabul and Mount Aruna was occupied by the opposing group. Hardly believable is that the Surya followers opposing themselves to the Zhuna supporters told Xuanzang a story sympathetic to the other. So, the Surya group told Xuanzang that it had long resided on Mount Aruna insisting on its priority of occupying that mountain contrary to the truth that the Zhuna group had actually dwelled there before the Surya group came. According to this second interpretation, the earlier occupant of Mount Aruna was the Zhuna group, and later the Surya worshippers arrived there and dispelled the Zhuna to Zabulistan. In fact, we have good documentation for taking the second interpretation as more probable. A paragraph from the kingdom of Cao in the Western Region chapter of the Suishu says as follows: As a custom of this country, people worship a deity of dubious character. On Mount Congling is its statue called Zhun. The ceremonial institution is extremely gorgeous. The shrine is roofed with both gold and silver plates and paved with silver on the floor. More than one thousand people visit this shrine everyday. In front of the shrine is a back bone of fish. At the center is a hole through which a mounted horseman can pass freely. There is no room to doubt Zhun as identical to Zhuna in the Da Tang Xiyuji. However, which Zhun does the Suishu refer to, the one in Kapisi or in Zabulistan? Levi (1895: 374-375) solved the location of both Jibin and Cao with the conclusion that Cao of the Suishu is identical to either Caojuzha or Caoli in the Da Tang Xiyuji. This interpretation was later supplemented by Marquart (1901: 285; 1915: 249) and Shiratori (1917: 33-102) and followed by Fujita (1931: 49a) and Uchida (1972: 69). Using such a traditional solution of Cao, we have to take the above paragraph as explaining what was going on in Zabul. However, the discussion which follows favors the identification of Cao with Kapisi. 203 Untenable is the hypothesis that Cao is an abridged form of Caojuzha. Xuanzang annotates Caojuzha (*dzaukiuta) as also called Caoli (*dzauli), while the Tangshu says that Zabulistan is called either Caojuzha or Caoju. Daoseng, an eminent monk-scholar and a biographer of Xuanzang, gives another form, Zaoli (*dzauli), to Caoli. The early Tang sources thus show Caoju, Caoli and Zaoli instead of Caojuzha, but no evidence supports a view that a single character Cao represents Caojuzha. The account of Zabulistan in the Tangshu says that Empress Wu (her reign: 690-705) officially used *Ziajiwat in place of *Dzaukiuta or *Dzaukiu which had been used prior to her. This chronologically fits well with what Xuanzang and Daoxuan used. In early Tang before 690, therefore, Caojuzha was called Caoju and Caoli as well as Zaoli. If Cao in the Sui period had been the same as Caojuzha, why was Caojuzha not called or not abridged as Cao? Cao is different from Caojuzha. The kingdom of Cao again appears in another Sui source which indicates that it is located at the southern foot of the Hindukush on a long distance trade route connecting Central Asia with the Northwest. TheXryu Tuji, the Illustrated Accounts of the Western Region, is a book of information about the Western Region bearing the same character as the Western Region chapter of any official dynastic history, such as the Suishu. The book was personally compiled by Pei Ju (557-627), a Sui official of low rank in charge of relations with long distance traders at Zhangye. He succeeded in editing information about various western countries he got from such traders at Zhangye and dedicated it to Emperor Yangdi, knowing well his interest in such curiosities. This book itself is now lost except for Pei Ju's preface which is found in his biography in Chapter 67 of the Suishu. The preface includes an important reference to three main routes, the Northern, Middle and the Southern, which lead from China, or actually from Dunhuang, to the Western Sea (Xihai). While China is located in the extreme east, Xihai means the western end which represents the three major civilizations at that time, i.e., the Byzantine Empire, Sassanian Persia and India. On the Southern Route is the Cao kingdom which is the last country of Central Asia located just before entering the Bei Boluomen guo which literally means the North Brahman countries. Evidently Bei Boluomen guo is identical to Bei Yindu guo which appears in the Da Tang Xryuji as indicating one of the five Indian districts, or the northwestern part of the Subcontinent, which extends from Laghman and Nagarahara in the west to the Punjab in the east. The Southern Route runs from Dunhuang westward to the Western Sea through the countries listed in the following order: the Charkhlik-Miran area, Khotan, Kaghaliq, Tashkurghan, the Pamirs, Wa'khan, Tokharistan, the land of the Hephthalites, Bamiyan, Cao and Bei Boluomen guo, or the Northwest Subcontinent. The route after the Pamirs corresponds well to that taken by Xuanzang on his way to Nalanda, i.e., the route running from Bakhl to Laghman through Tokharistan, Bamiyan and Kapisi. He notices Laghman as belonging to the Bei Yindu guo, or the countries included in the Northwest Subcontinent. Comparing Xuanzang's itinerary from Tokharistan to the Northwest with the route recorded by Pei Ju one might be led to an understanding that Cao in the time of Pei Ju (in or before 606) is identical to Kapisi in the time of Xuanzang (629). Really the Jibin section of the Tongdian (Vol. 192) says that 204 in the reigns of the Sui Emperors Jibin was called Cao which was located southwest of the Pamirs. The Tangshu also states that the country called Jibin in the Tang period is the same as that called Cao in the Sui period. Jibin in this context is equal to Xuanzang's Kapisi, as rightly predicated by Sylvain Levi in the article cited earlier. Xuanzang indicates that between Kapisi and Caojuzha is Fulishi-sthana. Kapisi is different from Caojuzha. Cao, therefore, is Kapisi, not an abridged form of Caojuzha. Although Caojuzha is not found in any source in and before the Sui, Cao undoubtedly differs from Caojuzha which is Zabulistan. The above identification of Cao with Kapisi enables us to think that Kapisi was called Cao by the Chinese court about ten years before Xuanzang's visit. The Cao kingdom in the Suishu therefore should be read as Kapisi ten years before Xuanzang. If so, the second interpretation rather hits the mark: Zhun in the Suishu must reflect a previous situation on Mount Aruna, as the place was called on Xuanzang's visit, and this mountain must have originally been called Congling (Mount Pamirs) in the Sui Period. 4. Dating the Appearance of the Surya Group The next question is when the cult of Zhun in Kapisi was known to China. The Cao envoy arrived in the capital city during the Daye era (605-617), according to the Suishu. The exact year is not known. Only after his dedication of the Xiyu Tuji did Pei Ju give the emperor advice to invite western countries in order to fulfil the emperor's wish to get curious objects from the western world. So, any information about Cao that Pei Ju got in Zhangye is presumably the earliest in the Daye era. If so, the date of the edition of the Xiyu Tuji may solve the problem, yet no direct evidence for it is anywhere recorded. In addition, it differs among different sources. In the Zizhi Tongjian, the whole record about the editorial work is placed in the very end of Vol. 180, i.e., after a series of events in the tenth month of 607, but does not intrude into the new volume which begins with the records concerning the first month of 608 (Zizhi Tongjian: Vol. 180). The author of the Zizhi Tongjian therefore seems to have regarded the edition as completed before 608, possibly in 607. On the other hand, in the Jiu Tangshu (Vol. 63) we find Pei Ju's work in between 605 and 607. However, Uchida (1973:115-128) convincingly succeeded in dating the edition to some time between the middle of 605 and the middle of 606. In comparison with Pei Ju's career in the court, the earlier half of 606 fits well with a period when he could compile it. The Zhun had believably left Mount Conglin sometime in between 606 and 629 (Xuanzang's visit to Kapisi) for a mountain on the southern border of Zabulistan which they were to name Mount Zhunahira, while the new Surya occupants of Mount Conglin changed the name to Mount Aruna. 5. Dissimilarity of the Two Superimposed Sanctuaries 205 The excavation by Hackin revealed two superimposed shrines, earlier and later, of different styles on a hill of Khair Khana (Illus. 30). The later shrine complex consists of several buildings extending on a ridge going downward to the east and ending with a flat square terrace supported by stone work at a lower point. The holy precinct consists of three independent, rather small cells for images, of similar plan and size, facing east and placed on the same north-to-south line within a space enclosed by walls on three sides. This is called "Trois Sanctuaires" in the report of the excavations. Access to the sanctuary is only given by tiny doorways on both north and south walls, while the east side of the sanctuary is open to the air supported by a high retaining wall based on the floor of the already mentioned terrace built below on the limestone bedrock of the hill. The retaining wall actually is the front wall of the earlier shrine abandoned and systematically filled with mud bricks and rubble up to the roof level, on which a new floor was built for the "Trois Sanctuaire". The later shrines were built on top of the previous "Ancien Sanctuaire", but set back to the west with a space on the east or in front of them. The earlier "Ancien Sanctuaire" consists of three rooms under one and the same flat roof. The large and square central room is open to the east through a narrow doorway, and connected with the two side rooms by the barrel-vaulted passages. The walls measure about 4 meters high with a breadth of about 2.4 meters. The method of construction, not only the plan, of the two superimposed sanctuaries quite differs from each other. The retaining walls of the terrace in the western half of which the earlier shrine is built is of coarse stone work recalling the diaper masonry in Taxila, while the walls of the earlier building are made of pakhsa layers thrown on a foundation of finely packed slab masonry. The core of the walls seems to have been filled with mud and tiny chipped stones. The face of the walls is very thickly coated with four em-thick white plaster (chunah), which remained still intact at the time of my observations in 1964 and 1974. Each later shrine has a low foundation worked with large dressed limestone, the interstices being filled with small slabs. On the three sides of these square foundations are the enclosure walls built with square mud bricks, the size of which measures 38 cm. x 38 cm. x 9 cm. A building of the later period, called H by Hackin, located at the topmost part of the complex is strengthened with a circular tower at each angle. Nothing is left of the architecture of this fortress-like building except for the foundation of coarse, stone work, the height of which is about one meter above the ground. Since the direction differs from that of the main buildings, Hackin takes the bastioned building as much later than the main group of three shrines (Hackin et Carl 1936: 5). It is most unlikely that the bastioned building was built after them. Building H is located on the highest part of the Khair Khana complex in order to look over the whole area and even to the eastern, extensive plain. Comparable in plan and function with this are the two fortresses at Begram, one located inside the city walls and the other at the southern border of the dwelling area which had come into existence as a result of expansion beyond the then useless city walls. Therefore, such fortresses are evidently of the last stages of the town (Hackin, Carl & Meunie 1959: 104 and 106). These buildings appeared in a period when people needed something to defend the habitation area in place of the city walls. The 206 existence of the one inside the city suggests that the role of city walls had indeed come to a standstill. The later sanctuary of Khair Khana strengthened by such a fortress must have still been alive with the latest stages of Kapisi Begram in the later half of the seventh century (see also Chapter II). The later shrines enshrined at least the three marble Surya statues. The Surya statue seated on a chariot with the driver Aruna in front was from the debris in between two of the shrines. A base with a standing warrior and without any main deity was found in situ on the plinth set up against the back wall of one of the shrines. The third statue, a standing Surya of most elaborate workmanship, was found by chance in 1980 at the foot of the Khair Khana hill and promptly reported by Paul Bernard and Franz Grenet (1981: 127-146). The seated Surya is worn out on the surface, whereas the standing sun god appears as if it has just been finished. In addition to the three main sculptures unearthed so far, sockets were found on the plinths, actually three in each shrine. Undoubtedly they originally received nine Brahmanical images. The shrines are similar in size to each other and regularly placed on the same line, allowing for a supposition that the nine images were regarded as of equal power and value. On the other hand, for the earlier shrine, the square central room is markedly big in both horizontal dimensions and vertical size. The remaining height of the walls measuring no less than 4 meters suggests a colossus. Such dissimilarities in design and location between the two superimposed religious architectures at Khair Khana lead us to the following identification: the 'Ancien Sanctuaire' was dedicated to the deity Zhun on Mount Conglin, as the Suishu describes, and the 'Trois Sanctuaires' were the Surya temple on Mount Aruna, as Xuanzang records. Such a change at Khair Khana is datable to some time in between 606 and 629. The later shrines at Khair Khana which contained at least three marble Surya images were installed sometime between these dates. Therefore, the marble Surya statues, seated or standing, once displayed together with other pieces of Brahmanical sculpture on the second level of the Kabul Museum but since 1992 unfortunately missing, are roughly datable to the seventh century. 207 XIII Identity of the N apki Coins The golden crown decorated with a bull head worn by Kapisian kings in the time contemporary to the Chinese Sui Dynasty was first known to 'orientalists' in Europe when Remusat (1829, 211) published a selected translation of the account on Central Asian kingdoms in the last volumes of the Wenxian Tongkao edited by Ma Duanlin in the fourteenth century. Gobi later drew attention to a similar bull-headed crown in relation to a series of coins with the Pehlevi legend Npki, Nspk, or degenarate versions thereof, but concentrated on the bull exclusively from the zoological point of view. However, neglect of the bull crown, at least in the light of the evidence given by Remusat, has unfortunately prevented the placement of the coins themselves in their proper historical context since they have been assumed to be of the Hephthalites. Such a misconception about the coins has been promoted by a belief in an illusory Hephthalite movement toward the south of the Hindukush which has influenced scholars to take a firm hold of a historical unreality: i. e., the Hephthalite presence in Kapisi and Zabulistan. The Hephthalites really had nothing to do with these kingdoms, but directed their concerns only toward the Northwest Subcontinent. Also historians and numismatists have overlooked the existence of a king called Khingal and his successors in Kapisi despite the earlier scrutiny by by Petech (1988, 187-194). A kingdom with capitals at Begram and Hund was inaugurated by Khingal-henceforth I call it the Khingal Dynasty-in parallel with the political weakening of the Hephthalites in the middle of the sixth century and lasted until the rise of the Turks in Kabul in the latter half of the seventh century. One of the earliest mentions of that dynasty is the Cao kingdom in the Suishu which is identifiable with Kapisi in the time of the Sui (Kuwayama 1975). The original Suishu's reference to the bull crown was later quoted in the Beishi and the Tondian, the latter of which was eventually copied in the aforesaid Wenxian Tongkao which was translated into French by Remusat. An allusion in the Suishu to the crown worn by a king of Cao lends strong support for the identification of the Napki coins with those issued by the kings of the Khingalline of Kapisi. 1. Historically 1. The Absence of the Hephthalites in the Kapisi-Kabul-Zabul Region No written document contemporary to or soon after the Hephthalites supports the extention of their power beyond the western Hindukush to Kapisi and Zabulistan. Sources 208 mentioning the rise and fall of the Hephthalites are scarce and fragmentary, but the Hephthalite section in the Western Region chapter of the Weishu (6. 2279) describes that they subjugated Samarkand, Khotan, Kashgar, Margiana (Anxi) and thirty other small countries in the Western Region and that they called their state a big country (Da Guo). The account of Song Yun's travel to the Hephthalites as edited in Volume 5 of the Luoyang Qielanji (Record of the Buddhist Monasteries in Luoyang) also tells that their power extended to the Turks in the north, to Khotan in the east, to Sassanian Persia in the west and to the still unidentified region Dieluo at the time of Song Yun's official visit to the Hephthalite king at his headquarters in Tokharistan and to the tegin of Gandhara in 520 (Zhou 1963, 195, 197; Wang 1984, 225). It further says that some forty countries sent their envoys to the headquarters. The same source also informs that they controlled the important places to the east of the Pamirs, saying that Tashkurgan in the Pamirs and Yarkand on the southwestern margin of the Taklamakan Desert were under their rule (Weishu 6. 2279f.). It is very important that the Weishu and Song Yun's account, which are of great value as sources of contemporary information about the Hephthalites, make no mention of Bamiyan and Kapisi among the vassal states of the Hephthalites.(l) The Great Yuezhi section of the Weishu (6. 2275) says that the king of the Great Yuezhi called Jiduoluo (>*kidara), brave and fierce, eventually sent his troops southward and invaded the Bei Tianzhu (the Northwest including parts of the Punjab) crossing the great mountains (Da Shan) to subjugate the five kingdoms located to the north of Gandhara. The Kidara Kushans do not seem to have come from the west to the Northwest, particularly considering the importance of the statement that they 'subjugated the five kingdoms located to the north of Gandhara.' The very same number of kingdoms is recorded as subjugated by the Hephthalites in the Western Region chapter of the same source book (Weishu 6. 2279f.): Yarkand (Zhujubo), Tashkurghan (Kepanduo), Wakhkhan (Buhuo), Chitral (Shemi) and Gandhara (Ganduo), all of them being governed by the Hephthalites in their heyday in the first thirty years of the 6th century. Since the course and method of nomadic invasion into India is not likely to be manifold, the Hephthalites may have also followed their predecessors like the Kidara Kushans and the Great Kushans. The existence of the five kingdoms to the north of Gandhara and no mention of Bamiyan, Kapisi, or Zabulistan may lead us to think that the Hephthalites invaded India through the area between the eastern Hindukush and the western Karakorum. Song Yun's narrative and its relevant paragraphs in the Weishu significantly keep silent about all of the kingdoms in and south of the western Hindukush. Useful for proving the absence of the Hephtl1alites in Kapisi and Zabulistan is Song Yun's itinerary to the residence of the Hephthalite tegin in Gandhara from their headquarters in T okharistan. As the Luoyang Qielanji tells (Wang 1963: 192-209; Yang 1984: 224f.), he proceeded from Tashkurghan in the Pamirs through Wa'khan to the Hephthalite headquarters in Baghlan(2) and left there for Gandhara through Zebak (Bozhi), Chitral (Shemi) and Swat (Wuzhang). There is no mention of crossing the Hindukush to Kapisi. Especially since Song Yun was the head of an official mission sent by the Northern Wei emperor to the Hephthalites he must have passed specifically through Hephthalite territory, particularly on his way to the Gandharan tegin. 209 The five kingdoms to the north of Gandhara and Song Yun's itinerary lend strong support to the view that the Hephthalites never occupied the areas of Bamiyan, Kapisi, Laghman and Nagarahara, all south of the Hindukush, not to speak of countries further to the south of Kapisi, e. g., Zabulistan. To the contrary, the kingdoms such as Bamiyan, Kapisi, and Zabulistan first appear in Chinese records exclusively posterior to all the documents cited above, such as the Suishu, as well as the Xiyu tuji (An Illustrated Account of the Western Regions) and the biographies of Jinagupta and Dharmagupta in the Tang Gaoseng zhuan (Tang Biographies of Eminent Monks edited by Daoxuan in the middle of the seventh century) that offered the basic materials for the former two. In the earlier half of the sixth century the Hephthalites and the Ruiruis were two strong nomadic states dividing Central Asia into the east and the west. After rising in the area between these two nomadic states and defeating the Ruiruis in Mongolia in the middle of the century, the Turks next had to rid themselves of the menace of the Hephthalites. The Hephthalites, in fact, had had strong ties with the Ruiruis: the Hephthalite chieftain had married three sisters of the Ruirui chief Poluomen and in about 523 Poluomen had tried to flee to the Hephthalites in quest of protection on his failure in rebellion against the Chinese government at Dunhuang (Weishu 6. 2302;Zhoushu 10.3262). The decay of the Hephthalites at their main headquarters in Tokharistan had already begun when the Turkish chieftain, Mugan Khaqan, attacked them immediately after defeating the Ruiruis in between 552-555 (Tongdian, Vol. 196;Beishi 10.3287). A clear witness of this event is Jinagupta, a Buddhist monk of Gandharan origin, who left there in 554 for Chinese Central Asia via Kapisi, Bamiyan and Tokharistan. His biographer Daoxuan makes a special allusion to the 'current political emergency' which Jinagupta often suffered from on his stay in 555 at the Hephthalite headquarters where he saw the land extensive but barren without producing anything to eat (Tang Gaoseng zhuan, Vol. 2; T. 50. 433c). Three years later Sinzibu Khaqan, another Turkish chieftain, attacked the Hephthalites from the north in alliance with Khusrow I who had replaced the pro-Hephthalite king Kawad I and had close ties with Sinzibu as his son-in-law (Marquart 1901: 64; Altheim 1969: II, 260-261; Haussig 1956: 23). This attack released the Tashkent, Ferghana and Samarkand regions from the yoke of the Hephthalites, and they became unable to send missions to the Chinese Northern Zhou after that date as recorded in the Hephthalite sections of the Zhoushu (3. 918). In between 562 and 568 Ton-shad Zijie eventually crossed the Amu under the leadership of Sinzibu to occupy the most favorable lands for pasture (Suishu 6. 1854). As the Hephthalites in Tokharistan were weakened at the advent of the Turks, their ties with the Gandharan Hephthalite tegin naturally were broken. Into this political vacuum a local kingdom in the Kabul Valley rose to political power covering all the valleys extending some 300 km east to west between their summer headquarters at Begram at the confluence of the Ghorband and the Panjshir Rivers and their winter residence at UdabhandapuraIW aihind on the right bank of the Indus. The kingdom is called Kapisi by Xuanzang in the Da Tang Xiyu ji and Jibin in the Chinese historical sources of the Tang Dynasty. The identification of Jibin with Kapisi was rightly made by Levi (1895;1896). The rise of Kapisi was not a direct result of the Hephthalite weakening, 210 but depended much more on the Turks' policy never to invade India, unlike the Hephthalites, the Kidara Kushans, the Great Kushans, and the Bactrian Greeks before them. The successive invasions of these people from their headquarters in Tokharistan into the northwestern part of the Subcontinent had long since connected the latter with the former, and more extensively with Central Asia, and had led Gandhara to economic wealth that enabled the long survival of large numbers of Buddhist monks and monasteries. The Turkish attitude to leave Gandhara untouched caused the traditional trunk road running through the eastern fringe of the Hindukush to become deserted. Eventually Gandhara and environs inevitably fell into economic declinewere; hence the desolate Buddhist communities as witnessed by Xuanzang in Jalalabad, Gandhara, Swat, Taxila, and other minor states in the Punjab. As shown by the fact recorded in a biography of Xuanzang-the Cieni zhuan-that the Turkish chief, Tardu-shad, in Tokharistan offered to escort him to Kapisi, the Turks politically had good relations with countries to the south of the Hindukush, most probably through Bamiyan. The first appearance of the actual routes connecting Tokharistan to Kapisi through Bamiyan occurs in the biography of the aforesaid Buddhist monkJinagupta who was at the Hephthalite court in Tokharistan in 555 (Tang Gaoseng zhuan, Vol. 2; Chavannes 190: 333f.; Kuwayama 1987: 718f.). Based on this evidence, it is presumably correct to say that Gandhara lost its economical superiority to Kapisi some years before 554 - 555. 2. The Khingal Dynasty of Kapisi Insofar as the accessible written sources are concerned, the kingdom of Kapisi sent a dozen tributary missions to the Tang court from A. D. 619 onward.(3) Speaking of 658, the time of the official setting up of the Xiuxian Dudu Fu (Protectorate General Xiuxian) atJibin (Kapisi), the Jiu Tangshu (16. 5309), the Tangshu (20. 6241) and the Tang Huiyao (Vol. 99) record that the first king of Kapisi was Xingnie-*x[i]aTj-Tjar in Middle Chinese restorable to *henger >Khingal (Khingar)-and that the kingship had been inherited from father to son for the twelve generations prior to the present king called Hejiezhi-*yaryartsie in Middle Chinese restorable to *ghargharci>*ghar-ilci. Association of this statement with the evidence for the first mention of Kapisi in the biography of Jinagupta makes it clear that Khingal founded his dynasty in the middle of the sixth century. Further, according to the Cefu Yuangui (Vol. 970), on the occasion of their tribute on the eleventh month of the fourth year of the Yonghui era (653), Kapisi reported to the Tang court that the princely heir had succeeded to the throne in their country. This account most probably indicates that in 653 Ghar-ilchi had already ascended the throne in Kapisi. Later, in 661, when Tang China officially established his presence in Central Asia, the Xiuxiang Dudu Fu of Kapisi was again confirmed in his position at the capital town, archaeologically known to us as Begram, which Chinese historical sources call Ehe. The name of the king is not shown herewith, but the very silence about it in the sources rather suggests that Ghar-ilchi must have still been the king even in 661. In any case, we are allowed to admit that there was in Kapisi a line of kings which was inaugurated with Khingal in the middle of the 6th century and continued at least until 661. 211 In the middle of the dynastic history of Kapisi a politico-religious event happened in between 606 and 630 at Khair Khana, a hill in the southern district of Kapisi. The following is a summary of the discussion which I fully gave elsewhere (Kuwayama 1975). This event happened before 630, the time of Xuanzang's visit, since Xuanzang recorded it in the Da Tang Xiyu ji. It refers to a tale about a hill called Aruna in Kapisi. The mountain deity Aruna did not receive Zhun when the latter, having come from afar, asked for lodging so that Zhun had to move on farther to the Zhunahira hill in southern Zabulistan. The tale reflects, in reverse, a religious conflict between two sects of Brahmanism: the Surya group gained a victory over Zhuna, drove them away and occupied the hill which had been the Zhuna group's home. This interpretation is supported by the still earlier written evidence that the Aruna hill had been called the Conglin hill, an abode of Zhun deva, as recorded in the Kapisi section of the Western Region chapter of the Suishu. Since information about Kapisi in the Suishu is believed to be based on what was told to Pei Ju in the early months of 606 by long-distance traders gathering in Zhangye, the existence of an abode of Zhuna on the Conglin hill must have been current information at least in and before 606 and the conflict is datable to sometime after 606. The place of the conflict was in fact the site of Khair Khana in northern Kabul, or southern Kapisi, which has revealed two superimposed shrines: the upper shrine, where marble Brahmanical images such as a seated Surya were discovered, is identifiable with the Aruna hill and the lower huge shrine, found underneath the floor of the upper one, is related with the former dweller, Zhun, the remains of which still survive as massive walls coated with very thick mud plaster suggesting a hasty removal of the images. The great difference between the two shrines as to their plans, elevations, building materials, and techniques of construction belies an interpretation that one and the same religious community ceased to use the shrine for some reason and then renovated it with quite a different style at the same place after filling it up with debris. People who paid hommage to Surya drove those worshipping Zhun away from Khair Khana, occupied it and built instead quite a new shrine there. The image of Zhun removed by his worshippers to Zabulistan is possibly identical with a colossus often referred to as Zun in the early Muslim sources. Ya'kut (II, 906) tells that a gigantic statue of Zun on the hill Jabal az-Zun was a trophy taken by the Arabs, while Murgotten (1924,143-144) clearly translates passages from Baladhuri saying that Ibn Sammurah brought Zamin Dawar under his control in the latter half of the seventh century attacking his opponents holding a shrine of Zun and ravaging the golden idol of Zun with its eyes inlaid with ruby. Baladhuri's record suggests that the Zhun/Zun cult and the rulers in Zamin Dawar politically had close relations. The Da Tang Xiyu ji also describes Zabulistan as follows: The king of Zabulistan, of a long line of kings and sincere in his faith, applies himself assiduously to religious work (merit) and is well instructed and fond of learning ... ; non-Buddhist heretics are very numerous principally worshipping the deity Zhun who formerly came from the Aruna hill in Kapisi, and took up his abode right here on the Zhuna-hira hill in the 212 southern district of this kingdom; the deity is severe or good causing misfortune or exercising violence; people who invoke him with faith obtain their wishes and those who despise him reap misfortune; therefore people both far off and near show for him deep reverence; high and low alike are filled with religious awe of him; the princes, nobles, and people of this as well as of foreign countries assemble every year at a season of rejoicing which is not fixed, and offer gold and silver and precious objects of rare value, with sheep, horses, and domestic animals; all which they present in simple and confiding trust, so that though the earth is covered with silver and gold, and the sheep and horses fill the valleys, yet no one would dare to covet them; they consider them as things set apart for sacred purposes; the heretics, by subduing their minds and mortifying their flesh, get from the deity sacred formulae; by the use of these they are frequently able to control diseases and recover the sick. (A partly modified version ofBea11884: 284-285) A close involvement of the rulers in the cult of Zhun, thus mentioned by both Baladhuri and Xuanzang, lead us to a supposition that the conflict at Khair Khana was not simply religious but disclose a political implication. The ruling class of Zabulistan defending Zhun is most probably those who had moved from Kapisi as a result of the conflict which took place in between 606 and 630. "A long line of kings of Zabulista" in the paragraph above quoted of the Da Tang Xiyu ji therefore suggests that the ruling class was not indigenous to Zabulistan but a branch of the Khingal line of kings in Kapisi. 3. Identity of Khingal First, careful attention should be drawn to identification of the Khingal Dynasty with a dynasty of the Cao kingdom in the Chinese sources of the Sui period (581-618). The kingdom Cao is not an abridged form of either Cao-juzha or Cao-li which are assignable to Zabulistan, but synonymous with Kapisi which the Tang Chinese called Jibin. I proved this in detail elsewhere (Kuwayama 1975, 93-107), including the fact that the identification of Cao in the Sui time with Jibin in the Tang had already been recognized by Du You in the middle of the 8th century. He reproduced all of the Suishu's record on the Cao kingdom in the section on Jibin in the Tongdian when he edited the latter in 766 - 768. Xuanzang tells that on his visit in 630 to Kapisi the king belongs to a family of Chali (Da Tang Xiyu ji, Vol. 1), which is a Chinese synonym for the Kshatriya. This clearly shows that the Khingalline of kings belongs to the Kshatriya. In the manuscripts of the Da Tang Xiyu ji datable to the Tang period found in Dunhuang and in others preserved in existing Buddhist monasteries in Japan such as Chusonji and Ishiyamadera, Chali is replaced by Suli which usually means Sogdians. Yet the existence of a Sogdian king of Kapisi does not seem of any historical reality. This word might presumably be derived from a copyist's fallacy, at least if one does not take seriously the statement of 213 1 the Suishu that the king of Cao (Kapisi) had the family name Zhaowu which was usually held by local Sogdian dihqans. This statement of the Suishu may have been confused with the account of a Sogdian kingdom also called Cao. The Chinese character applied to the Sogdian kingdom only lacks the first three strokes of the character for Kapisian Cao and shares the same phonetical value. These characters therefore are liable to be confused. Insofar as the Indian terminology is concerned, the Khingal king of Kapisi was not a descendant of the nomadic Hephthalites but indigenous, belonging to the warrior class. Since the king in Xuanzang's time was not Hephthalite, all kings of the Khingal Dynasty in Kapisi, succeeding from father to son from Khingal to Ghar-ilchi, were not Hephthalite. There is other evidence for separating Khingal from the Hephthalites: different coinage. If the dynasty inaugurated by Khingal was truly local and issued coins, they must have differed from the last series of Hephthalite coins depicting the peculiar busts of Hephthalite kings. In fact there exists an enigmatic series of coins bearing on the obverse the Pehlevi legend Npki MLK-a, Nspk MLD-sh, or degenerate versions thereof as read by GobI (1967: E. 198-205,217,221,222,262,264-269 and 271).(4) This specific coinage has always been attributed to the line of Hephthalite kings and dated using many different methods by different scholars; e. g., de Morgan (1923-1936), Ghirshman (1948), GobI (1967), and Mitchner (1972), who have all held fast to an interpretation of history by which the Hephthalites governed over Zabulistan and Kapisi even after the Turkish occupation of Tokharistan. We have already argued against this illusory interpretation. Actually a substantial difference between the above-noted Hephthalite coin series and that bearing the Npki legend is beyond question, and the Napki coins depicting a king who wears a crown with a bull's head either on top of it or at the front (GobI 1967, I, 132f. and II, 71£.) are attributable to the Khingal Dynasty.5 In the historical context of the Northwest and eastern Mghanistan, however, several names of kings quite similar to the founder of the Kapisi kingdom, Khingal, are known: (1) Deva $ahi Khingila on the Hephthalite coins, (2) Khinkhila in the Rajatarangini and (3) Khinkhil (Khinjil? or Khinjal?), an Arabic transcribed form of Khingila or Khingala, in al-Ya'qui~s Ta'rikh (Houtsma 1969,479). We also add to the above another Khingala in the two line inscription of the marble Ganesa statue from Gardez, who is mentioned to be Sri $ahi / Uc.lc.liyaJ)a $ahi (Kuwayama 1991B: 283). Petech (1988, 187-194) thinks that Khingal is not a personal name but an eponym based on the name of a dynasty which passed down from generation to generation in east Mghanistan and the Northwest. However, it does not seem reasonable to regard Khingal as an eponym and to attribute these names to one and the same ethnic line. Khinkhil or Khinjil was one of the Kabul Shahs of Turkish origin in the time of al-Mahdi (775 - 785) as al-Ya'qubi mentions that al-Mahdi sent to kings of various countries messengers who asked for their submission and that many of them did submit among whom were the Kabul Shah called Khinkhil and others. Khingala in the Ganesa inscription is also one of the Kabul Shahs, identifiable with Bofuzhun who ascended the throne in 745 as stated in theJiu Tangshu, Vol. 198, and the Tang Huiyao, Vol. 99 (Kuwayama 1991A: 283). We do not know whether Arabic Khinkhil is identical with Chinese Bofuzhun, but it is sure that neither is a descendant of 214 the Hephthalites. Based on the Tang sources mentioned earlier, the king Ghar-ilchi was the twelfth king of the Khingal Dynasty who ascended the throne in or before 653. The twelve generations are so long that Xingnie (Khingal) seems to have founded the dynasty in the remote past. If so, Khingal of Kapisi cannot be identical with either of the above names associated with the Kabul Shahs. Kalhana's Khinkhila and Khingila on Hephthalite coins are chronologically closer to Khingal of Kapisi but, nevertheless, are also historically distinct. The Rajatarangini (Stein 1900, I, 52, Taranga 347) says that "his [Gokar1).a's] son Narendraditya ... bore the second name of Khinkhila .... " In this case Khinkhila clearly indicates a personal (individual) name. In the same way, Deva Shahi Khingila on coins also is a distinct personality one among the many specific Hephthalite kings known from their coins such as Toramana and Mihirakula. Khinkhila Narendraditya is attributed by Kalhana to a dynasty called Gonandiya along with the names of other rulers, some of whom, e. g., Toramana and Mihirakula, are clearly Hephthalite. Indeed, Stein's assumption seems acceptable that Khinkhila Narendraditya "must be recognized in the Ephthalite ruler who calls himself on his coin Deva Shahi Khingila" (Stein 1900, I, 80). The Gonandiya Dynasty is said to be much earlier than the Karkota Dynasty. The Karkota Dynasty was founded by Durlabhavardhana Prajfiaditya who was, according to Kalhana, on the throne between the years 3677 and 3713 of the Laukika era (601/602-637/638) and contemporaneous with Xuanzang who passed through Kapisi and Kasmir some sixty years after the Hephthalite breakup. So, the date of Khinkhila Narendraditya is assignable to sometime before or even long before 600, which is consistent chronologically with Khingal of the Chiense sources, however he was a Hephthalite. Since Khingal of Kapisi, as previously shown, was not a Hephthalite, he cannot, therefore, be associated with either Deva ~ahi Khingila on the Hephthalite coins or Khinkhila in the Kasmir chronicle. But an additional question arises: Could Khinkhila Narendraditya really be a Hephthalite ruler over Kasmir? The account of Song Yun's visit in 520 clearly refers to the Hephthalite tegin as a king of Gandhara who "has been fighting against Kasmir for the control of the territory for three years" (Zhou 1963,210; Yang 1984, 235). As I interpreted this passage (Kuwayama 1989, 95f.), the king, in fact, resided in the vicinity of Jhelum to the east of the Indus rather than in Gandhara to the west of the Indus, fighting for control of the salt trade to Kasmir which had caused both of them to claim the territory. It is really incredible that there would have been two different Hephthalite kings, one in Gandhara and one in Kasmir, fighting against each other. The names and events of the Hephthalite kings that are described before the verses regarding the Karkota dynasty in the RajataTangini may have intentionally or unintentionally been edited into the dynastic lists by Kalhana for whom any vacancy in history was unacceptable. Even if he was a Hephthalite, Khinkhil in the Kasmir chronicle presumably was not a king who resided in and actually ruled Kasmir, but some area outside it. 4. Authenticity of the Bull Crown in the Suishu The bull crown of the Cao (Kapisi) king is referred to in the Western Region 215 chapter of the Beishi, the Northern Dynastic History (10. 3238f.), edited by Li Yanshou in 659. That chapter includes information gained from the then existing Weishu, the Zhoushu. The bull crown is also seen in the Jibin section in the Tongdian, Vol. 192, edited by Du You between 766 and 768. It is doubtless that Du You used the Suishu as one of the main source books. However, the 'bull head' is not found in any accessible edition of the Western Region chapter of the Suishu, but is replaced by a "fish head". The same is true with the Cefu Yuangui (VoL 960) which was compiled in the eleventh century. Considering the chronological order of these references to the crown, differences possibly existed between the existing editions of the Suishu and the ones that Li Yanshou used for the Beishi in 659 and for the Tongdian by Du You in the middle of the eighth century. Were it not for the bull crown in the Suishu available to Li Yanshou and Du You, they could not have described it in their accounts of the Cao kingdom. The original Suishu certainly alluded to the bull crown. This supposition gives cause to think that the king of Cao (Kapisi) actually wore the bull crown by the early seventh century at the latest. Yet, before accepting the above solution, we should furthermore give a brief explanation as to the source materials used for the Western Region chapter of the original Suishu to prove how that chapter is authentic. In the end of the Wenti regime and in the early half of the Yangti, or in 602-610, information about Central Asia and India was variously collected by several officials to utilize for Yangti's policy toward the Western Region and edited in such publications as the Xifan ji (Record of the Western Barbarians), the Da Sui Xiguo zhuan (Great Sui Record of the Western Countries), the Xiyu tuji (Illustrated Accounts of the Western Region), the Tianzhu ji (Record of India), and the Damojiduo zhuan (Biography of Dharmagupta). All of them must have served as basic sources for the Western Region chapter of the Suishu when later compiled by the early Tang officers in 629-630. The Xifan ji was edited in an unknown year between 602-610 by a Sui official, Wijie, who travelled as the head of an official mission to the western countries in the company of Du Xingman. Although it is lost, some paragraphs from it explaining some Sogdian city-states are quoted in the Tongdian. According to the preface of the Western Region chapter of the Suishu (6. 1841), only three kingdoms of many countries actually visited by Wijie and Du Xingman are cited since they gained an agate goblet in Jibin (Kapisi), brought ten dancing girls and other strange things in the kingdom of Shi (Shahr-e Sabz) and found Buddhist scriptures in the "town of a king's residence (wangshe cheng)",<6) In c. 602 the Da Sui Xiguo zhuan was edited by Yancong by the emperor's order. According to the biography of Yancong in the Tang Gaoseng zhuan (VoL 2), being well versed in both Buddhist and non-Buddhist literature and famous for his profound knowledge in Chinese and Sanskrit languages, Yancong assiduously devoted himself to participation in translating Buddhist scriptures brought either by Indian monks themselves with them or'by Chinese monks who had been in Central Asia in search of scriptures. He thereby became well acquainted with Dharmagupta whom he knew well enough to publish his biography. Dharmagupta arrived at Sui Changan (properly called in the Sui time Da Xing) in 590 leaving Takkadesa in the Punjab and travelling in the 580s, after 216 the Hephthalite decay, through Kapisi, Bamiyan, Tokharistan, Badakhshan, Wa'khan, Kucha, Turfan and Dunhuang. Yancong's biography of Dharmagupta, though lost, therefore may have included the extensive descriptions of the above Central Asiatic kingdoms which made up the main parts of the Da Sui Xiguo zhuan. Daoxuan's biography of Dharmagupta in Volume 2 of the Tang Gaoseng zhuan clearly states how this record was compiled: Realizing that Dharmagupta's knowledge about these countries surpassed the information given by other Indian monks who had reached Changan before him, Yancong edited on behalf of the emperor a record of the western kingdoms. It consisted of ten chapters on products, climates, towns, politics, education, codes of conduct, foods, dress, treasures and natural topographies. (T. 50, 435c) The Da Sui Xiguo zhuan unfortunately ceased to circulate and was lost after Xuanzang's Da Tang Xiyu ji became popular enough to take the former's place. TheXiyu tuji edited in the earlier half of 606, of which only the author's preface is extant, was a description of Central Asian kingdoms given by long-distance traders from Central Asian towns at Zhangye in 606 as edited, with illustrations, by Pei Ju who was in charge of foreign affairs for the Yangti regime (Kuwayama 1975: 102). The preface by Pei Ju, quoted in his biography in the Suishu (6. 157ff.), remarkably elucidates the three major routes leading from Dunhuang to the three contemporary civilized worlds: the Byzantine Empire, Sassanian Persia and India (Suishu 6. 1579f.; Kuwayama 1975: 100). On the southern route, to India, one can reach Northern India by passing in due order several kingdoms on the southern fringe of the Taklamakan Desert, Tashkurghan in the Pamirs, the Tokhara (Tuhuoluo) and Hephthalite (Yida) regions, Bamiyan (Fanyin) in the Hindukush and Kapisi (Cao) south of it. The above discussion proves that the source materials used for the Western Region chapter of the Suishu are authentic. Hence the description in the Suishu that a Kapisi king in the late sixth or the very early seventh century wore a golden crown decorated with bull head is historically reliable and additional proof that he was quite distinct from the Hephthalites. The late sixth or the early seventh century covers the period of the Khingal Dynasty which lasted from the time of Khingal through Xuanzang's time until 661 at the earliest. If so, we are allowed to think that the crown decorated with bull head is the dynastic one of the Khingalline of kings in Kapisi, not that of the Hephthalites. II. Numismatically 1. Collapsed and Uncollapsed Types Beside the reverse depicting within a single beaded circle a fire-altar in the center and an attendant on each side of it, on the obverse, also, certain Sassanian elements are 217 manifest on the more or less degenerate types of the bull-crown coins classified by GobI to E. 262-E. 271. The latter require discussion first in order to obtain an approximate dating of the whole series (GobI's E. 198-205,221,222,262-271). E. 262 and E. 264 have on both sides a single circle and on the obverse in particular the star-on-crescent mark at three corners outside the circle. These Sassanian elements are confined to the coins of Hormizd IV (579-590) and of Kawad I from his 13th regnal year (500) to the 19th (506). E. 263 has crescents without stars placed at three corners outside a single circle as found only on the coins of Khusrow I (531-579) and Vahram IV (590-591). On the other hand, E. 265 and E. 266 bear a specific mark, a 'wineglass,' replacing the star-on-crescent seen on E. 262 and E. 264. Coins with the wineglass-like mark therefore represent a further modified or more remote version of the Sassanian prototypes. These three Sassanian coin types are chronologically confined to the sixth century, specifically the years 500-506 and 531-591. If we exclude Kawad 1's coins which are exceptionally earlier, the others mostly converge toward the second half of the 6th century, but significantly have nothing to do with any coin of Khusrow II (590/591-628) and his successors. Although it is hard to determine which is the exact prototype of E. 262 and E. 264, Kawad 1's coins or Hormizd IV's, the two emissions might presumably depend on Hormizd IV's coins rather than Kawad I's. The reason for this presumption is that Kawad I was the Sassanian king closest to the Hephthalites: he had been a hostage in the Hephthalite headquarters in the time of Peroz and later was a puppet king of the Hephthalites during their most powerful period. For the Khingal Dynasty, coming to power after the Hephthalite decline in the 55 Os, the coinage of Kawad I must have been historically more distant than the others. Therefore the bull-crown coins bearing the above Sassanian coin elements are roughly datable to sometime in or after the second half of the sixth century. The other degenerate types of coins with bull crown are assignable to sometime after 615-627. The U-shaped beard of the attendants beside the fire-altar on the reverse of E. 266 only appears on the Khusrow II coins issued in his eleventh regnal year (602). Throughout E. 267-E. 271, the following points are observable: (1) The collapse of the Pehlevi script is clearly discernible, e. g., the illegibly degenerated 'p on the obverse and bg' on the reverse. (2) A star-on-crescent motif on both sides of the king's head is shared with the specific coins of Khusrow II issued only in his regnal years 26 (615/616), 27 (616/617), 36 (625/626) and 37 (626/627). (3) The reverse of E. 267 depicts a markhord-like deer, while E. 268 and E. 269 bear a standing king-like figure, in addition to a wineglass-like motif on the reverse. The reuse with countermarks and the entire imitatation of the coins of Khusrow II, as in the case ofYazdegard III's, began with Arab-Sassanian coins. Of the five groups of Arab-Sassanian coins classified by Walker, only those issued by the Umaiyad Caliphs, i.e., those from AH 31(651) to AH 83 (702), are closely related to the coins under 218 discussion. It was parallel with and in response to the Umaiyad way of minting ArabSassanian coins that coins imitating those of Khusrow II so often appeared in the Kabul region from the second half of the seventh century to the eightth century. However, quite a basic difference exists between these Arab-Sassanian coins and the coins with bull crown: the latter did not use the dies of Sassanian coins but only took some of the elements such as the star-on-crescent; for the coins with bull crown only minor elements of the Sassanian coins were needed. The difference chronologically suggests that the bull-crown coins are anterior to the Arab-Sassanian coins or roughly before the last days of the Sassanians. Of the bull-crown coins which bear Sassanian elements, therefore, some are datable to after the middle of the sixth century and the others after the second decade of the seventh century, while none of them is later than the third quarter of the seventh century. The bull-crown coins without Sassanian elements, or uncollapsed types (GobI's E. 198-205,217,221,222), began to be minted with the inauguration of the Khingal Dynasty approximately in the middle of the 6th century or a little earlier. The coins have the king's bust facing right with a crown decorated with a bull head either on top of it or at the front and flanked at the right and left sides by extended wings. Another usual element is a crescent placed at the lower front, which looks like an element supporting the bull head. In the case where the bull head is placed on top of tlle crown, the crescent supports another element such as a disk or a star-like object. To the right of the king's bust is the Pehlevi legend Npki or Nspk and to the left of it a Pehlevi script clearly the legend on readable as a in the case of Npki and sh in the other. According to some coins, such as E. 198,200,201,202 and 205, can be read as Npki MLK plus a and that on others, such as E. 217, 221 and 222, as Nspk MLD plus sh. GobI thinks that Nspk MLD-sh is chronologically prior to Npki MLK-a. On the other hand, Humbach (1966, I, 59), without any regard to a and sh , reads Npki MLK as npky MLK', which is restorable to 'nafak shah, and Nspk MLD as nycky MLDH or ssfky MLDH, which is a collapsed version of the former. The script n is hard to read in some cases or actually not written in other cases (E. 217-1,217-5, 222I-222VI). In addition, the sp of Nspk is often replaced by a strange script similar to the Arabic numeral 8 which does not make any sense. These facts mean that Nspk MLD-sh was not a basic or the earliest legend but a type which was transformed from Npki MLK-a and that Npki MLK-a is earlier than Nspk MLD-sh. Therefore the coins of the Npki MLK-a series presumably appeared earlier than those of the Nspk MLD-sh series later. Among the Npki MLK-a coin series some legends lack Nand L to make 'pki Mk-a' (E. 203 and E. 204) ,md even 'pki MK-a' cannot be read on coins such as E. 267-E. 271, on which GobI read the script written to the left of the king's bust as p which should have been written as a or sh. GobI does not misread it, but the inscriber himself could not understand what should be written at this location. On the reverses of these coins something like a deity (E. 268 and E. 269), a kind of symbol (E. 271), and a deer (E. 267) replace the usual design, a fire altar with attendants. These coins with the 'pki MK' legend are also much lighter in weight than the others and characteristically 219 bear a wineglass-like mark on the reverse. In view of the degenerate legends and other extraordinary characteristics, the coins with the 'pki MK' legend are still later than those with NspkMLD or even spkMLD. 2. Typology The degradation of the legends suggests a chronological sequence of the bull-crown coins as follows: (A) the Npki MLK-a series; (B) the Nspk MLD-sh series; (C) the pki MK series. Association of these three legends with the crown types tentatively gives us the following typological groups: [Type IA] Bull head on top of the crown and crescent at the front. Npki MLK-a legend (E. 198). [Type IB] Bull head on top of the crown and crescent at the front. Nspk MLD-sh legend (E. 222). [Type IIA] Bull head on a crescent at the front. Npki MLK-a legend (E. 200 - E. 203). [Type IIB] Bull head on a crescent at the front. Nspk MLD-sh legend (E. 217 and E.221). [Type IIIA] Bull head on top of the crown and crescent decorated with a small dot at both ends placed at the front and supporting a large disk. Npki MLK-a legend (E. 205). Of the above five groups, Type IA is earlier than Type IB and Type IIA is earlier than Type IIB; viz., Types IA, IIA and IIIA (with Npki MLK-a) are the oldest of all and are followed by Types IB and IIB (Nspk MLD-sh) primary if we attach importance to the chronology of the legends. From this point of view, the location of the bull head on the crown is chronologically meaningless. In addition to this sequence, the coins on which Sassanian elements are discernible can be classified as follows: [Type IV] Bull head on top of crown and two neighbouring crescents on the diadem, the right crescent supporting a star-like object. pki MK legend and star-on-crescent motif on either side of the king's bust (E. 267 - E. 270). They are not earlier than 615-627 since a star-on-crescent on either side of the king's bust only appears on coins issued by Khusrow II in his 26th, 27th, 36th and 37th regnal years. [Type V] Bull head on top of crown and two neighbouring crescents on the diadem. 'pki MK' legend (E. 265 and E. 266). E. 265 is later than the 5th regnal year of Khusrow I, or 535. E. 266 is later than 601, or the 11th regnal year of Khusrow II, since the U-shaped beard of the guardians of the fire altar on the reverse only appears on the coins issued in the 11 th regnal year of Khusrow II. [Type VI] Bull head on top of crown and element of reversed C at the front of the diadem. Legend unidentified. The examples are E. 262-E. 264. A star-on-crescent placed outside of a single circle is an imitation of Khusrow 1's coins. The appearance of this type of a star to the right and a crescent to the left of the fire altar on the reverse, which is characteristic of the coins of Hormizd IV (579 - 590) and Bahram VI (590), suggests a date later than 580. 220 The above examination of the coins themselves and the dynastic trends in the Kapisi-Kabul-Zabul region allows us to attribute the coins with bull-head crown to three chronological groups as follows: (1) Types lA, lIA and IlIA-GobI's Emissions 198, 199, 200, 201 and 202-are the earliest of the coin series that bears a crown with bull head and attributable to the Khingal lines of kings in Kapisi, one of them being the king described in the Kapisi section of the Western Region chapter of the Suishu. (2) The coins succeeding the above are Types IB and lIB, the legend Nspk MLD-sh losing the original meaning represented by Npki MLK-a despite its intention to express the same legend. They might be attributable to the rulers of Zabulistan having split off from the Khingal line of rulers at Kapisi after the Khair Khana conflict and datable to sometime after the event's occurence between 606 and 630. (3) Coin Type IV and followings are the latest groups of the bull-head series and hard to assign to specific rulers. However, taking into consideration that the Turkish rulers of Kabul replaced the Khingal dynasty in the third quarter of the 7th century and that their coins are quite different from the bull-head series-depicting a king wearing a crown with a trisula on top which I shall discuss in a separate paper-these groups may not be later than the third quarter of the seventh century. 221 XIV The Turki Shahis and Relevant Brahmanical Sculptures in East Mghanistan Brahmanical marble images known to us since 1934 in East Afghanistan and Northwest Pakistan have failed to find their proper locations in history, because of the lack of accurate information about the levels in which they were unearthed. Therefore, simply based on their stylistic comparisons to the post-Gupta sculptures, many scholars have proposed various chronological views which converge to a period from the seventh century to the ninth century, despite much earlier chronology by Joseph Hackinwho first found the pieces at Khair Khanah, 10 kilometers to the northwest of Kabul. Such a broad timespan, however, covers the reigns of the two different dynasties, the Turki Shahis and the Hindu Shahis. Some have interpreted the pieces as belonging to the time of the latter Shahis with the reason that they are Brahmanical. The confusion seems to have derived from the least careful examination of the objects themselves. Therefore, more detailed discussion of the individual pieces would find the factors, which can characterize the pieces, to lead to a more reasonable framework which the marble pieces may deserve to be given. 1. Images Known so far by 1976 1. Surya, h. 42 cm. (IHus. 33, left) This sculpture was discovered in 1934 by J. Hackin and J. Carl in the debris accumulated between the chapels Band C of the later phase at Khair Khanah (Hackin 1936: 4). It is assumed that it was originally placed in one of these two chapels. The piece can be divided into upper, middle and lower parts. In the centre of the upper part is Surya, flanked by Danda and Pingala. In the middle part is the driver Aruna, holding the reins of two horses, whose backs are shown as they veer upwards to the right and left. The lower part consists of a roughly hewn pedestal. Surya's hair lies in small loops, starting from the centre and flowing towards the back in a symmetrical movement, almost indistinguishable from the crown. The hair is braided at the back and falls over the shoulders. The crown has a tripartite leaf design, with the ribbons attached to its both sides and a single knot at the back. The upper part of the ribbons is incised vertically, while the lower part has multiple nail-like incisions. Conspicuous is a tenon on the ribbon, indicating that a halo had originally been affixed to the head of Surya. A small and a large necklace encircle Surya's neck. The drapery of Surya's robe is indicated by lines, knots and folds. The folds hanging between the knees are not part of the drapery but a kind of scarf. Both elbows 222 rest on the thighs and the arms seem to be lifting an object. The wrists are adorned by bracelets, but the hands are missing. The pearl design on the lower legs could indicate the decoration of the boots. The expression of the face is closely related to that of Vishnu excavated at Taxila (Banerjea 1956: 401, PI. XXXI; Siddiqui 1935-36: 35, PI. XLa). The eyebrows, however, are similar to those of Siva from Gardez discussed later. The flanking Danda and Pingala are on a much smaller scale. They lean sideways, looking up at Surya. Both wear similar mantles, closed in front and covering the shoulders, and similar tunics, draped in front, with pearl decoration along the hem. The spear held by Danda has a peculiar leaf or herring-bone design. It is interesting to note that the covers of the horses' saddles bear the same design. The back of the sculpture is polished without depicting the back of the images. So the clear contrast is observable between the depicted back of the main deity and the polished back of these smaller images. 2. Pedestal with standing warrior, h. 21 cm. From Khair Khanah. The warrior and the pedestal were reported to have been excavated separately from the chapel A (Hackin 1936: 4). The warrior was found in the debris of the Northwest corner of the inside of the chapel A, while the pedestal was in its original position in the chapel. On the pedestal are a pair of feet of a missing standing figure and to the right, leaning against a column, a warrior is looking up at the missing deity. The left arm and right leg of the warrior are also missing. His long hair is parted in the centre and tied with ribbons. The long belted tunic, with a pearl design along the hem, is the same as that worn by Surya. A sword and a dagger hang from a belt. The left hand rests on the pommel of the sword. He wears two necklaces: the one of pearls is similar to Surya's. Boots are indicated by vertical and horizontal lines under the knees. Several curved lines between the tunic and the boots indicate the bulkiness of the pantaloons worn under the tunic. This type of costume has already been seen on Kushan coins and sculptures, such as the male figure in the relief of the pedestal from Paitava and Shotorak (Rosenfield 1967: 104). From this fragmentary sculpture it is impossible to discern the exact nature of the missing divinity, but if the warrior be supposed to represent the war-god Skanda, the missing statue could be Siva. 3. Fragment of a halo Discovered in an open space in front of the chapels A, B, and C at Khair Khanah (Hackin 1936: PI. XXIII, 32). A ribbon is represented in relief, incised with a nail-like pattern. Considering the presence of the tenon attached to Surya's back and Rao's view (Rao 1914:1-2, 311-318) that all Surya images had haloes regardless of their origins, this halo might have belonged in fact to the Surya statue mentioned above. One difficulty, however, is how to relate the ribbons on the Surya itself and on this halo. 4. Urdhvaretas Ganesa, h. 91 cm. From Gardez. This is the object still worshipped at Daram Sal Hinduwa or Pir Rathan Nath Dargah in Kabul. It was published by G. Tucci in 1958 and discussed again 223 by].C. Agrawala in 1968 (Tucci 1958: 328, fig. 40; Agawala 1968: 166, fig. 1). ~he statue, previously settled in the niche of the temple, now stands free for removing to a new location. The height of the pedestal and figure altogether is 71 cm., and of the rough unpolished base, 20 cm. The crown is damaged on its uppermost part, but still show a narrow decoration in front. This decoration, as we shall observe later, consists of a standing object in the centre and a crescent as well as two opposing makaras below the crescent. A pair of ears behind the shoulders are probably those of an elephant. The statue originally had four arms, but only the front arms, to the elbows, are extant. The sacred thread depicting a snake, viz., naga-yajnopavita, crosses the torso from the left shoulder to the right flank. These four arms, as well as the tiger skin wrapped around his waist and thighs are characteristic of Vinayaka, as mentioned in the Vishnudharmottara purana. An erect linga is shown under the bloated stomach. This piece and the next are important, as there are no early examples known except for the Gane§a image at the entrance of the Cave No.6 at Udayagiri, Madhya Pradesh (Agrawala 1968:167, fig. 3). Another significant fact is that this Ganesa figure has two lines of inscription in acute-angled Brahmi, giving the name of the donor king and the date of donation (Illus. 34; Chapter XV). The conspicuous feature is that the back of the head is carved with a clear representation of ribbons, which reminds us of that of the Khair Khanah Surya. 5. Urdhvaretas Ganesa This is also an object still worshipped in a Hindu Temple in Shor Bazaar at Kabul; it is said to have been found at Shakar Darah, about 15 km. north of Kabul. Agrawala published this piece in 1968 and M.N. Dhavalikar discussed it and the Gardez Ganesa (fig. 4) in a recent article (Agrawala 1968: 167, fig. 4; Dhavalikar 1971: 331-36). But the exact information regarding its discovery and the relation with the site is not available. Dhavalikar also referred to other Siva and Surya images from Shakar Darah in the same paper, but no information other than his description has reached the present writer. In general, the statue is worn. It apparently wore the tripartite crown and four arms can be recognized. The right front arm is holding a lotus flower, the left front arm is missing from the elbow, the other two rest on the heads of the attendants flanking Ganesa. The iconography of this statue is almost identical with that of No.4, including the sacred thread and the erect linga and tiger skin. The identity of the attendants is not known, but they could, with their long curly hair, be children. The pedestal, except for a narrow band at its top, is unfinished. According to Rao, the four-armed Ganesa represents Bala Ganapati and Bhakti-Vighnesvara, although no example with the combination of tiger skin, erect linga and the sacred thread, is known (Rao 1914: 1-1,52 ff.; Banerjea 1956: 359-361; Sivaramamurti 1950: 29-30). In the Ganesa pieces from India, regardless of region and period, there are no examples similar to the Gardez Ganesa with its short stubby trunk curled on to itself, and the Shakar Darah Ganesa, with its short trunk in the air and not holding anything. 6. Durga Mahisasuramardini, h. 34 cm. 224 The so-called Scorretti Marble of unknown provenance was published in 1955 by D. Schlumberger (1955: 110-119, pIs., I-II). The goddess Durga has her right foot on the back of the buffalo Mahisasura. Her hand is on the bffalo's mouth, pulling the head towards her. Durga stands behind him. Her buttocks and left leg are carved in full. The four legs of Mahisasura are lost, except for the right front hoof. The hoof is divided into two parts by a vertical line at the centre and between the hoof and leg are vertical incisions. The body of Mahisasura is complete, as is the right hand and back part of the body of Durga. Part of the trisula which Durga has planted in the rump and the arched tail are carved in relief. Durga wears a wide bracelet with pearl designs on the right wrist, and a ring on her little finger. The folds of the drapery on the left leg are carved in relief and part of a loincloth covers the buttocks. 7. Durga Mahisasuramardini, h. c. 60 cm. (lllus. 33, right) This is said to have been unearthed near Gardez. So far no publication of this object has come to the present writer except for the photographs published in 1968 by J. Auboyer (1968: PI. 99) and in 1970 by M. Taddei (1970: fig. 138,2-5), It represents the eight-armed Durga who is killing Mahisasura. Durga has her right foot planted on Mahisasura's back, her left front arm holding his head and her right front arm stabbing him in the neck before cutting off his head. Another right arm holds his tail. The other left hands are holding unidentified objects. Two right hands are missing. Another right arm, partly missing, must have held the trisula. Mahisasura's hair is combed back behind the ears and over the shoulders, and he is holding a dagger. The hoofs of Mahisasura and Durga's trisula and bracelets are represented in a characteristic fashion. The central prong of the trisula has a leaf design which is now greatly worn. The left hand holding Mahisasura's head and the right arm holding the dagger wear wide bracelets with floral designs, different from the other bracelets. The drapery over Durga's right leg is stylized in radial lines. This piece and No.6 are similar in showing Durga approaching Mahisasura from the back and stabbing his rump with a trisula. 8. Head of Durga, h. 6 cm. (lllus. 32, above left) This was collected in 1960 on the suface level of the mound called Qat' a Amir Muhammad at Tagao, Laghman (Fischer 1964: 35-42, fig. 5-8), and was originally part of a full figure. The head is carved so as to face three quarters to its left and the back of the head connects to the unfinished block which was intended to be a halo. It is somewhat worn, but the third eye and tripartite crown are clearly detailed. The central leaf of the crown has the same decoration as the crown of the painted Bodhisattva at Fonduqistan. The right and left leaf designs of the crown have identical decorations of beads arranged in multiple circles. On the right side of the head there seems to be a floral ornament but it is too worn to be clearly identified. Wavy hair can be seen under the crown behind the ear and on the forehead. On the top of the head, the hair is laboriously incised in numerous parallel lines, tied at the back of the head. It is difficult to be sure about the identification with Durga, but as the third eye and the crown show the female type among this marble group, so this piece could well be the goddess of Siva, and therefore 225 Durga. It is not certain for Fischer to intend to identify this head as that of the Scorretti Marble. 9. Head of Siva, h. 28 cm. (lllus. 32, below right) This head is said to have been found at Gardez and was published by D. Schlumberger in 1955 (1955: 114-116, fig. 2), It is the largest head among these marble sculptures. The delicate narrow face with a pointed chin wears a tripartite crown and has a third eye. Since the left side of the face is badly worn, it must have been above the ground for a considerable time. The curly hair, divided at the centre of the forehead, is represented by delicate lines and with clearness. At the bottom of the centre of the crown is a crescent and on the crescent a vertical object. The horn of the crescent and the top of this object are represented as if tied with ribbons. The eyebrows are notable, drawn in a single curved arch beyond the end of the eyes. A similar representation can be seen in the Khair Khanah Surya and in the Vishnu from Taxila. 10. Torso weaing a dhoti, h. 65 cm. Found at Tagao and published by Schlumberger in 1955 (1955: 114), The head, arms and legs below the knees are missing and it is in a generally worn condition. A shallow relief at the back indicates the fold of the drapery pulled towards the centre. The broken top part on the back could have been part of the hair or the head-dress. The figure originally could have had eight arms (my reexamination of the object in the Kabul Museum in 1987). 11. Pedestal with goddess relief, h. 20 cm. This is said to have been collected from Tapa Skandar, and published in 1964 by B. Dagens (1964: 35, pI. XXII). Two-thirds of the lower section is unfinished. The upper section is polished and has a female figure in relief in the centre of this part. Both arms are raised as if to support something, most probably the two feet of the main deity which is now missing. Some fragments of sculpture are still visible on the pedestal, the right fragment being a hoof or a part of the drapery. Dagens compared this pedestal with one of the sculptures from Sahr-i Bahlol in the Peshawar Museum, No. 1427 and with one of unknown origin in the Lahore Museum, No. 777 (Lyons, 1. and H. Ingholt 1957: no. 400, 158; no. 346, 148) and unduly tried to attribute this to Buddhist iconography (Taddei 1973: 69). 12. Uma Mahesvara, h. 81.5 cm. (Inus. 31) This is the only object among the marble sculptures which was unearthed during regular excavation in 1970 at the temple A of Tapa Skandar by the Kyoto University Archaeological Mission, and was published in 1972 by Kuwayama (1972: 5-14; 1974: 76 f.), It is one of the best preserved piece among the marble statues hitherto known. It is carved from one block of white marble, and represents the four-armed, three-eyed Siva seated on Nandin, flanked by his consort Parvati and Skanda standing at the left side of his mother goddess. The figures are carved on the pedestal which divides into two parts, 226 upper and lower. The upper, narrow part, well polished horizontally, bears a three-lined inscription of eighty-four acute-angled Brahmi letters including the punctuation marks. The height of this part measures about 5.8 cm. and the width 43 cm. This clear inscription gives this piece even more importance. The inscription tells us that the name Siva is given as Mahesvara and therefore the consort takes the name Vma. The lower part below this inscribed panel remains unfinished and rough as are the cases with the Khair Khanah pedestal (No.2), Surya (No.1), the Gardez and Shakar Darah Ganesas (Nos. 4 and 5), and the Gardez Durga (No.7). On the whole it is frontally oriented with a depth of only 18 cm. Of Nandi, only head and hoofs are visible from the front. He like Mahesvara has the third eye in the centre of the forehead. A flower pendant hangs from his neck. The hoofs are divided into two by a line and the hairs of the ankles are hinted at by incisions. Mahesvara, seated on Nandin, wears a bead necklace, a yajfiopavita and a dhoti, but the torso is bare. Over the dhoti a belt is worn. The folds of the dhoti are drawn towards the centre where the erect linga is shown. The drapery is stylized. The torso is skilfully executed in a reversed triangular shape. Of the two right arms, the front one is raised, although the wrist and hand are missing. The back right hand rests on the right knee. Compared to these right arms, the position of the left arms is unnatural due to the position of Vma. The front arm is made dwarfish, with the hand immediately below the shoulder, pointing down and resting on Vma's right shoulder; thumb and index finger are damaged. The back left arm is not visible from the front; it is represented as holding the trisula without a shaft, which is seen behind Vma's head. The trisula is unique-at least no other such examples have come to the attention of this writer-in that the central prong is in the shape of a leaf, with the left and right prongs emanating from the root of the central one. The head of Mahesvara can be divided into two, the face and the raised pentagonal chignon. The face is semi-oval with a short nose, long arched eyebrows and ears long enough to compare with the whole length of the face itself. Small ear-rings are attached to the large holes of the ear-lobes. The crown has a semi-circular decoration in front, especially similar to that of the Gardez Siva head. Beads adorn the edge, and from the top of the front decoration, the strung beads go towards the top of the head. On the right side of the chignon a small crescent is depicted, of a type not otherwise known to us. The hair flows down the back behind the ears and over the shoulders, in several braided locks. The ribbons attached to both the ends of the diadem are tied at the back of the head, the rest dangling. The lower part of the ribbons is incised with nail-shaped, shallow curved lines. The back of the head is treated with elaboration. Considering that Vma and other statues do not show a similarly detailed finish and that worshippers would not see the back of the statue, the reason must have been the importance of the main deity itself. The same can be said of the other marble statues, viz., the Khair Khanah Surya, the Gardez GaneSa and so on. The consort Vma rests her right arm on Mahesvara's left knee as she leans towards him. Her left hand rests on her waist. Vma is much smaller than MaheSvara, unlike the Hariti figure of Gandharan Pancika-Hariti composition. Another point of interest is 227 Uma's left hand, which is flat without any sculptured relief and entirely different from Mahesvara's hands and feet; the details of the goddess' left side are neglected. Uma's crown has a double, not triple, leaf decoration; the left leaf is larger and appears to be the central ornament of the crown. A string of beads decorates the base of the crown. The hair is long and flows over the left shoulder. On the left shoulder and arm are observed the rippling folds of the blouse, which recall those of Hariti from Skarah Dheri now in Lahore Museum (Bachhofer 1939: II, pI. 150). A square fringe of blouse on the breast overlapped by the double necklaces is represented by the strung-pearls or beads design. A long skirt covering the whole body below the waist shows Y-shaped folds at its hem and covers one-third of the pedestal on its proper left part. A scarf is also worn, covering the belly and both forearms. A snake-like hem falls between the legs. The small naked figure can be Skanda, since it does not represent Ganesa. He wears nothing but for a beaded band, bracelets and round ear-rings. Wavy hair is shown on both sides of the head, and there is no representation of hair at the central part of the head. His face is characteristic with the low forehead and eyes and eyebrows which are exactly like those of Urna and Mahesvara. U rna's bracelets are composed of both rings and strung beads, while Mahesvara' s and Skanda's bracelets are simpler rings. The back left hand of Mahesvara, holding the trisula, has a finger-ring on the little finger. A distinctive feature of this statue is the well-preserved vermilion, blue and black pigments. Vermilion is used for Mahesvara's hairline on the forehead, the upper eyelids, lips, chin, ears, neck, part of the crown, nipples, navel, waist, belt, right arm elbow and part of the trisula. On Uma, vermilion is applied more clearly to the hairline, eyelids, ears, neck, and lips. Mahesvara's garment was probably painted witl1 blue pigment (lapis lazuli) which remains in parts. The eyelids of Uma and Mahesvara are also painted blue. Black pigment can be seen on the hair incisions and again on the eyelids. The inscription is not in strictly even lines, partly because Uma's skirt flows over the upper right part of the pedestal, as we noted above. The inscription is of eighty-four characters, twenty in the first line, forty-five in the second, and eighteen in the third, leaving a long empty space. Besides the twelve above-mentioned pieces, at least eleven more are known, not including the four fragments discovered in 1970 and 1972 at Tapa Skandar, which brings the total to fifteen pieces. Of those fifteen pieces, two are of Siva, two are lingas, four are of Vishnu and two are only known as white marble statues. The four pieces from Tapa Skandar are each 10 cm. long; three of them hold lotus flower, the other being a part of the shoulder or the arm with a beaded bracelet. 13. Siva In a private collection in Kabul. From Deh Mir to the north of Tapa Skandar. Unpublished and said to be a triml1rti. 14. Head of Siva 228 Also unpublished and from Daruntah .. 15. Linga, h. 75 cm. Discovered at Tagao by A. Foucher (1942: 34, 172, pI. XXXI, 60), the phallic form of the top part is represented by shallow incision. Recently this piece was added to the exhibition in the Kabul Museum. 16. Ekamukhalinga, h. 42 cm. (Illus. 32, below left) Probably from Hund according to M. Taddei (1962: 288-310, figs. 1-4,7,9). The upper part is polished, the lower octagonal part unpolished; the lower end is flat. From the face the nose is missing. The head has a well developed chin and a third eye as well as clearly defined ear-rings. The wavy hair above the round face is carefully carved, but there is no dimensional representation. The ear-rings and necklace are of bead design. The chignon is peculiar and rendered carefully. 17. Brahma, h. 98 cm. This was discovered in many pieces in the northern part of the hall at the palace of Sultan Mas'ud III at Ghazni (Scerrato 1959: 29, 39,40, fig. 39; Taddei 1970: 257, fig. 166). The torso is missing. The three-headed Brahma is represented standing on a small lotus throne. At the foot of Brahma, on both sides, are small standing statues. There is a floral design adorning the edge of the semi-circular halo. The chignon of Brahma, twice the length of his face, is elaborate, but worn. The representation of necklace, bracelet, drapery, halo and pedestal are completely different from any pieces under discussion here. Scerrato, who discovered this piece, assumed that it was brought from India during the Ghaznavid raids, for the following reasons: firstly, according to Fihrist, Sultan Mahmud of Ghazna brought the statue from Somnath and used it at the entrance of the Arusu'l Falaq Mosque which commemorated the victory at Somnath. Secondly, according to Utbi, gold was taken from the war trophies of Somnath to be used to build the above mosque. Lastly, the damaged condition of the statue was due to the fact that it was walked over. The present writer agrees with the above view, since the statue has the same characteristics as the Surya statue found at Ajmer now in the Rajputana Museum and as the Surya statue of Rajasthan style dated to the twelfth century housed in the National Museum, New Delhi, No. 66.14 (Rao 1914: 1-2, pI. XC, 1). 18. AJain Tirthankara, h. 62 cm. This is also said to have come from the Ghaznavid palace a few years before the Italian Mission began to work. Formerly in the collection of Zahir Shah at Karez Amir, but later shifted to the Kabul Museum. The head is broken and all the surface is greatly worn. The style comparable with the Brahma noted above leads this statue to the same origin as No. 17 (Fischer 1969: 339, pI. 9). 19. A Hindu deity Not examined by the present writer. Fischer (1964: 35, note 1) and Le Berre 229 report it to have been found at Massamud, east of Qal'a Amir Muhammad in Laghman. 20-22. Three Vishnu statues Two pieces of Vishnu are reported by K. Fischer to have come from near Gardez. No. 20 is a fragment, 10 cm. high, of Vishnu caturmurti now in a private collection at Kabul. No. 21 is also of caturmurti, 40 cm. high, with four arms. No. 22 is a torso in the Ata Hakim of Ghazni (Fischer 1969: 340, pI. 13). 23. Vishnu caturmurti, h. 28 cm. D. Barrett published this piece as found at Attock (Barrett 1957: 54-59, fig. 1 a, b). On both sides of the main face are boar's heads, and on the back of the halo a head in shallow relief. The face and the crown of the main deity are worn out. The crown possibly had a trefoil. The braided hair falls over the shoulders. The ear-ring is unique: the part attached to the ear is the side view of a single ring, from which the circled pearls hang down. The wide three-string necklace supports a badly worn pendant. The halo has a decoration of leaf scroll at its margin and lotus petals in the centre. The face at the back of the halo has a double triangular ornament in the crown. The ear-rings and necklace are the same as those of the main deity, but more simply depicted. The above-mentioned twenty-three pieces of marble sculpture are without doubt Brahmanical (except No. 18). The temples or shrines where these statues were worshipped is not known except for the two cases, namely Khair Khanah and Tapa Skandar, which have been regularly excavated. Apart from such excavated examples, L. Edelberg and J. E. van Lohuizen-de Leeuw reported traces of Hindu temples in the upper Kunar Valley, northeast ofJalalabad. Fischer also reported that at Islampur the white-marble architectural fragments of a Hindu temple were used in the building of a wooden mosque (Edelberg 1957: 199-207; Van Lohuizen de Leeuw, 1959: 3-11; Fischer 1969: 357). From the photographs published by the above authors and the architectural fragments collected at Karez Amir, such marble fragments of architecture, most of them being "Corinthian pilasters", are irrelevant to any architectural element at Khair Khanah or at Tapa Skandar. 2. Views on the Marble Since Hackin first discovered the Surya in 1934, there have been various attempts as to the dating and stylistic comparisons of the marble sculptures. By examining these views and their problems, a systematic discussion of their dating will be given below. Hackin and Marshall. Six groups of buildings were uncovered at Khair Khanah. In the centre of the buildings were three similar chapels on a straight north-south line, built over a three-room chapel of the earlier period which had been buried with stone and mud brick. The chapels, A, B, and C were from the later period, with mud-brick walls on the stone masonry and under a common flat roof. Surya and other pieces had been enshrined in 230 these later chapels. The plan of three aligned chapels was unique. Hackin, nevertheless, related this plan to that of Bhumara, a small-scale Siva temple in Central India (Hackin and Carl 1936: 6-7). The Bhumara temple consists of a single room of stone; on the low terrace are the garbhagrha, where an ekamukhkhalinga is placed, surrounded by pradaksinapatha and mandapa, as well as two symmetrical shrines (Banerji 1924: pIs. Ib, II, XVII). Hackin stated that the plan and direction is similar between these two temples and that, moreover, they have the common features such as pradaksinapatha and mandapa. But pradaksinapatha and mandapa cannot be any clue to comparison and in fact it is quite different between them, Khair Khanah having three aligned chapels and Bhumara having only one. Hackin also stated that the Surya statues from both these sites were similar, each flanked by Danda and Pingala (Banerji 1924: pI. XVI a; Hackin and Carl 1936: chap. III). According to Hackin, two types of Surya representations existed in India. In the first, Surya rides in a war-chariot, either without attendants or with two female archers who dispel the darkness. In the second, Surya is flanked by his bearded secretary, Pingala, and the shaven warrior, Danda, both these attendants deriving from Azizos and Momimos, the attendants of the sun god of Edessa, and therefore of western origin. Hackin assumed that the attendants of Surya of both sites should follow the type of western origin. Hackin also found a similarity between Surya's costume and hair-style and those of the 4th-century Sassanian kings, such as Shapur II, Ardashir II and Shapur III. Therefore he attributes the Khair Khanah Surya to the 4th century works. He also dated the Khair Khanah site itself to the 5th century, following R. D. Banerji's dating of the Bhumara temple. But Marshall, contradicting himself on the dating of the Bhumara temple, seems to date this temple to the sixth century and accepts Hackin's dating of Surya (1937 499-501). Hackin based his dating on the costume of the Sassanian kings and the Siva temple in Central India. But it is now obvious that there is no connection between the costume of Surya and that of the Sassanian kings. The diadem of the Sassanian crown was used consistently up to the seventh century and was not particular to the fourth century. Comparing Danda and Pingala with those of Bhumara, one can admit a similarity, but the Khair Khanah Surya is so peculiar in its general composition and bears no feature common to others. Hackin's view is based purely on the iconographical resemblance. As to the dating of Bhumara, too, there are not any fixed clues which produced so widely different datings between Banerji and Marshall. Apart from such unreasonable datings, one notable point is the use of floral motifs for architectural ornaments (Banerji 1924: pI. VII c). Characteristic is the almost circular leaf with pronounced stem. It is more important to find that such motifs can be seen in Afghanistan, viz., B,amiyan, Fondukistan and Tapa Sardar. This type of design is not found among the sculptures of the Kushan period at Mathura or Amaravati or Nagarjunakonda. Even from the Gupta periods, only the decoration of the halo of the First Sermon from Sarnath shows a slight tendency towards the development of this leaf scroll motif. 231 Schlumberger (1955). In relation to the publication of the Scorretti Marble, he discussed the marble sculptures such as the Gardez Siva, the Tagao linga and torso, and the Khair Khanah Surya. He pointed out the rarity of marble as a material for sculpture in Mghanistan and, further, of three-dimensional representation, and identified the Scorretti Marble with the non-Buddhist icon, Mithra Tauroktonos. He further stated that these four pieces except for the Scorretti Marble were not Buddhist but Hindu and placed them between the end of the Hephthalites and before the Islamic invasion, since marble was never used during the Kushan, Kushano-Sassanian and Hephthalite periods when Buddhist sculpture was being produced. As to Hackin's dating of the Khair Khanah Surya, Schlumberger emphasized the importance of the two coins of Napki MaIka found in the room F at Khair Khanah as for dating. He accepts Ghirshman (1948: 51-54) whose dating is the seventh century, discarding Hackin's view based on V. A. Smith (1906: 235, pI. XXV 2, 3) and concluded that the Khair Khanah Surya was from that date. Schlumberger could not identify the Scorretti Marble as Durga, but he was correct in giving it a non-Buddhist theme and material, and dating it not earlier than the Hephthalite period. It is, however, clear that the question of the Napki MaIka coins from Khair Khanah and their relation to the dating of the Surya image are somewhat difficult. Goetz (1957: 13-19). He supposed that the Scorretti Marble and the other marble sculptures would fill the gap in Afghan archaeology, and attributed these objects to Saivism which had played an important role after the Kidara Kushans. Apart from the Tagao linga and the Gardez Siva which are obviously connected with Saivism, he rightly supposed that the Tagao torso is either six- or eight-armed Siva, not Vishnu, because the multiple-armed Vishnu only became popular after the 10th century, while Siva statues were already common in the tenth century. Goetz therefore concluded that the Scorretti piece was a Hindu sculpture of Durga, the consort of Siva, and contemporary to the Durga Mahisasuramardini of Ellura cave No. 17 of the seventh century, though with some inevitable stylistic differences. But the present writer has doubts on his way of dating based on the Ellura dating, for the dating of Ellura caves has not always been fixed; for example, Coomaraswami dated them to the sixth century . Goetz believed that the style of the marble pieces was entirely Indian, with no similarity to Gandharan Buddhist art or the Sassanian art observable on the representation of Siva on Later Kushan coins. This entirely Indian style of the marble sculptures was explained by him as coming from Kasmir where the Kanauj school of the Gupta sculpture flourished after Lalitaditya Muktapida's victory over Yasovarman (720-753 according to Goetz) whose capitals were Gwalior and Kanauj, a centre of the latest Gupta style. It is not surprising, as Goetz further explained, that this style prevailed in Afghanistan since this area was under Kasmir control even before Lalitaditya, at the time of the king Candrapida. Goetz maintained that the Kanauj school of Gupta sculpture in Kasmir, being different from the original Kanauj school and a provincial copy of the Gupta style 232 and not purely Indian, gave an immediate impetus to the creation of the white marble statues in Afghanistan. Goetz, who was familiar with Kasmir art, based his view on the historical facts concerning Lalitaditya in the Rajatarangini (Stein 1900: I, book IV, vv. 126-371). The question is whether Goetz's view is acceptable. He, in any case, did not try to make a comparative study of the sculptures themselves, except for Siva's crown, which he considered to be "purely Kasmiri style", brought in directly from Kasmir. His view that the dissemination of the Kanauj school of Gupta sculpture is directly and only associated with political events is not convincing to the writer. As for the similarity of crowns, the writer cannot agree with him either, at least so far not enough examples from Kasmir have come to light to prove his view. Barrett (1957: 54-59; Burgessl897: 14, CLII-CLIII). Barrett agreed with Schlumberger regarding the Surya from Khair Khanah. He maintained that the Gardez Siva head is of Kasmir style, and dated it to the second half of the eighth century. No comment was given on the Tagao torso. He did not agree that the Scorretti Marble was Mithra Tauroktonos and he defined it, like Goetz, as Durga Mahisasuramardini, dating from the eighth century. In discussing the pose of the Durga figure, he pointed out the Durga figure from Elephanta, now in the Prince of Wales Museum, Bombay. He compared the trisula with the examples in the Brahmanical Cave at Aihole and Ellura caves Nos. 14,17,21, and 27, dating these examples from between the seventh and ninth centuries. Barrett shows two pieces from the British Museum collection which have come from Buner and Attock. He also publishes four schist Brahmanical sculptures in the British Museum and wooden sculptures from Kasmir Smast, classifying all of them as Kasmir style. He does not say anything about their chronological locations despite refering to the historical account of the Kasmir region, including Lalitaditya, that historical factors should not be separated from art style. Tucci (1958: 327-28, fig. 40). He first introduced the Gardez GaneSa in Pir Rathan Nath Dargah. He did not refer to other pieces in his writing, but endeavoured to date the piece by deciphering the two-line inscription on the pedestal. According to Tucci, the inscription states that this great and beautiful Mahavinayaka was consecrated by the renowned Shahi king, the illustrious Shahi Khingala, Paramabhattaraka-maharajadhiraja, in the 13th of tithi of the white fortnight of jyeshtha under the nakshatra Visakha and the lagna of the lion in the year eight. As he said, it is impossible to know to which era this inscription refers. Tucci then pointed out that the king's name Khingala could be associated with Deva Shahi Khingila as found on coins and Narendraditya Khinkhila in the Rajatarangini (Cunningham 1962: 278-79, pI. VII, 11; Rapson 1969: 29; Stein 1900: I, 521 -Book 1, v. 347). In the Rajatarangini, this king is mentioned as dedicating shrines to Bhutesvara. His consecration of shrines to Bhutesvara was interpreted as being the same act as performed by king ASoka when he was attacked by the mlecchas, and by doing so was given a son who in 233 turn attacked the mlecchas. Therefore Khimgala was intended to be Khinkhila, following the Rajatarangini. This king is a successor of Mihirakula, but according to de la Vallee Poussin and Rapson, he comes before Toramana, judging from the coins (de la Vallee Pousin 1935: 14; Rapson 1969: 29). Tucci did not mention the date of the king. He tried to identify Khirpgilo-tyana as Udyana; and hence Khirpgala as the king of Udyana. Even if this could be read Udyana, there is no mention in the Rajatarangini that this king ruled Udyana, or Swat. Tucci stated that the script dated to the end of the fifth and beginning of the sixth century, without giving substantial reasons. But the earliest known acute-angled Brahmi is dated to the late sixth century (Buhler 1959: 68) and as this inscription closely resembles that of the Skandar Inscription (Yamada 1972: 15-22). Tucci's comment is therefore unacceptable. As to this Ganesa, it has already been mentioned that R. C. Agrawala published his view. He observed this statue as the combined style of two elements, representation of the god as two-armed and wearing a tiger-skin derived from the Gupta Mathura school on the one hand, and representation of erected linga derived from such sculptures as the Ganesa image of Udayagiri on the other. He assumed that this Gardez Gane§a was a result of the Gupta art descended from both these regions. Agrawala gave no mention of its date (Agrawala 1968: 168). Fischer (1969: 37-38). In his article publishing the Durga head from Qal'a Amir Mohammad, Tagao, he stated his view on the style. The facial expression, the eyes and third eye, could have been influenced by Gupta Indian art, while the curly hair and the headdress could be from late Gandharan Buddhist art, and this style must have developed simultaneously with Kasmir art. The white marble sculptures created in Mghanistan show a combined style of Hindu art and he dated this style to the Turki Shahi or the Hindu Shahi. Taddei (1962: 288-310). In his article publishing the ekamukhalinga, he mentioned characteristic post-Gupta elements such as the necklace and chin as seen in tke Bodhisattva image at Elephanta and at the Vihara I of Ajanta, and dated the ekamukhalinga to the eighth century. He considered this linga to be a specimen not of a style of that region, but of an original style evolving from the tradition of the Gandhara regions. He dated the white-marble sculptures including this linga to the eighth century when the Shahi dynasties were ruling. He did not try to make a more exact dating. But the present writer tends to assign especially tke ekamukhalinga to a style other and later than the group mentioned below of the white marble sculptures known from east Mghanistan. 3. Details of the Sculptures Apart from Hackin's and Tucci's views which concentrate mainly on fifth or sixth century datings, the views proposed by the others, namely Schlumberger, Goetz, Barrett, 234 Fischer and Taddei, mainly resulted from stylistic comparisons between the marbles and the post-Gupta or the late Gupta sculptures. Their contributions are valid to the extent that they shed fresh light on the situation of our marbles in Indian art. But such comparisons are in the present writer's view not always understandable, even though they could be acceptable in tke broad sense, because they passed over each marble statue without close examination and, moreover, they ignored the fact that the marble sculptures in question had been produced in the land of Afghanistan or its vicinity. Hence one should check whether the marble sculptures are related to other sculptures or not, after a first and rudimentary examination of particular elements in the marbles themselves. It has also been assumed that the marble pieces are from the Shahi period and with this in mind it has been attempted to place the styles in the post-Gupta schools. But the Shahi period can be divided into two successive periods, namely the Turki Shahis and the Hindu Shahis. The discussions hitherto published have never mentioned whether these pieces belonged to both Shahi periods or only to one, and if so, which. The published discussions, however, cannot answer this problem insofar as they are concerned exclusively with conceptional comparisons without any detailed observations of the marble pieces themselves. Among the marble pieces available for discussion, the best preserved is the Skandar Umamahesvara with many important details intact. Taking this material as the basis for discussion, we will compare the details of the pieces with those of the Skandar example, then a more reasonable conclusion will be expected, admitting that almost all the marble pieces from east Mghanistan tend to show their contemporaneity. Pedestal (Illus. 31). The frontal surface of the pedestal of Umamahesvara is divided into two: the upper and the lower. The upper part is polished horizontally, and recedes slightly from the chiselled lower surface which is unpolished and uneven. The lower end forms a pointed edge towards the front when this part is looked at from both right and left sides. As the example excavated at Khair Khanah shows, this unfinished part must have been buried and the pointed bottom seems to have facilitated the implantation of the statue. It is not clear whether the narrow polished band in front was at all intended to be inscribed, or simply to act as a finishing element to the statue. The spacing of the characters on the Gardez Ganesa is well composed, whereas that of the Umamahesvara is rather disorganized, due in part to the overlapping edge of the garment with the uninscribed space of about one and half lines. This type of pedestal, with or without inscription or relief on it, is found in the following pieces: (1) The Skandar Umamahesvara, (2) the Khair Khanah Surya, (3) the Khair Khanah pedestal with warrior, (4) the Gardez Ganesa, (5) the Shakar Darah Ganesa, (6) the Gardez Durga, and (7) the Skandar pedestal with relief, as well as (8) the Tagao linga. Of the above, Nos., 3, 5 and 7 as well as No.1 have the same finish and style, while the Shakar Darah Ganesa has not such a broad band but a thin band simply denoted by a line. In the Khair Khanah Surya the middle part, with the driver Aruna and horses, belongs to this polished band which extends to the lower section where it joins 235 the unfinished part. The pedestal in this sense could be regarded as being similar to the Shakar Darah Ganesa. It is worthy of note that, the existing pedestal examples being ten pieces, eight of these have the same type. The exceptional examples are those of the two statues from the Ghaznavid palace, which are supposed to have been brought from Rajasthan. Hence all eight pieces discussed here should be considered as one and the same group. Trifula (Illus. 31and 33, left). Iconographically the trisula is held by the images related to the Sivaite school, as far as the examples here are concerned. Such images are limited to the Skandar Vmamahesvara and to the Gardez Durga Mahisasuramardini. As noted previously, the back left hand of MaheSvara holds directly the prongs of this weapon which is seen above Vma's head. The Gardez Durga is observed to hold her trisula in one of her right hands which is now missing. The form of the trisula is composed of three prongs represented with the whole forming a square. Peculiar are two points: the central prong which represents a leaf or a herring-bone design, and two pierced holes, on both sides of the central prong, which distinguish each prong at its root. The two prongs other than the central one remain roughly carved and plain. In the example published by Schlumberger, the "Scorretti Marble", the remains of the trisula can be seen on Mahisasura's rump; the trisula may be here the same type as that of the Gardez Durga, because it can be admitted that one type is common to those examples of Mahisasuramardim. In the Khair Khanah Surya, Dan.da is observed to have a spear which, though not a trisula, is decorated with the same design as those of the said trisulas. Furthermore, this ornamentation is also seen on the fringes of the saddlecovers worn by the two horses. In this connection, it does not seem to be unjust to show two examples, even if they are not made of marble: one from Fonduqistan and the other from Toqquz Sarai. At F onduqistan this design is depicted in incision on the clay figures, the so-called princely couple, now in the Kabul Museum, which was found in situ in the niche E as a main object for worshippers. The design in question is seen along the lapel of the neckband (Hackin 1959: figs. 190, 192). Fortunately enough, this niche and the worshipping objects can be dated by a cremation urn, more precisely by the coins in this urn, as stated later. The Toqquz Sarai example of this motif is found on a small wooden sculpture forming a spear-like shape, which is reported as from among the abundant sculpture of Buddhist origin in a temple (Hambis 1961: PI. XLIX, nos. 116,117). Crown with the lunar symbol (Illus. 32, below right). The following figures wear crowns: (1) The Khair Khanah Surya, (2) the Gardez Ganesa, (3) the Shakar Darah Ganesa, (4) the Qal'a Amir Mohammad Durga, (5) the Gardez Siva, (6) the Skandar Vma and Mahesvara, and (7) the Attock Vishnu as well as the Ghazni Brahma. From these examples, the Attock Vishnu and the Ghazni Brahma should be omitted because they do not belong to trends simiiar to that of the Afghan 236 marble statues. Among these, there is pointed out a striking similarity between the Gardez Siva and the Skandar MaheSvara. The common feature is seen in the frontal and central decoration of the crown, regardless of whether the crown bears a triple or single decoration on its diadem. This decoration forms a semi-circular shape, surrounded by strung beads, which is filled up with many small square units pierced in the centre by a small hole. Covering this semi-circle, but smaller than it, a combined motif is attached, which consists of two elements, a vertical object and a crescent. The former is carved as standing on the centre of the crescent. The horns of the crescent and the upper portion of the object are observed to be tied with ribbons or strings. As well as this complex, just below the crescent is carved the combed hair, and on either side of it the hair is treated in loops. In the case of the Gardez Ganesa, though the precise expression is unclear because of the missing top, the hair treatment below the crescent is replaced with two makara-like carvings which are set symmetrically on each side of a central large bead. In spite of such a small difference between the Gardez GaneSa and the others, the central leaf of the crown is admittedly a semi-circular shape and the lunar symbol is also the same as those of the Skandar Mahes'vara and of the Gardez Siva. As for the Sakar Darah Ganes'a, one can imagine a semi-circular leaf in the central decoration, but its worn condition cannot give a precise indication. The crown of the Skandar Vma has two leaf decorations as noted previously. The left one is slightly different from the central one of Mahesvara in its form, but if the lunar symbol be removed from those of Mahesvvara and the Gardez Siva, the same device can be admitted between them. The central decoration of the crown of the Qal'a Amir Mohammad Durga is different from that of Vma, but it links with the decoration of the crown of the Bodhisattva figure represented in painting on the facade of the niche Eat Fonduqistan. Regarding the lunar symbol seen on these crowns, attention should be called to the fact that the same type appears in Khotanese iconography such as the Vairocana Buddha, Mahesvara, and GaneSa (Williams 1973: 109-154). Ribbons of the crown. The ribbons are attached to the sides of the diadem and tied at the back, the ends hanging down along the back of the head and shoulders. The lower half of the ribbons has horizontal nail-shaped incisions, while the upper has multiple vertical lines. This device is common among the marble pieces which have crowns except for the following: the Sakar Darah GaneSa whose crown is greatly worn, the Qal'a Amir Mohammad Durga of which the back of the head is not carved, and the Gardez Siva of which the back of the head is destroyed. In spite of the fact that there is a great deal of difference between the crown of the Khair Khanah Surya and those of the Skandar Mahesvara and the Gardez Ganes'a, the rendering of the ribbons shows a remarkable similarity, but this design is much coarser in the Surya example than in the Mahesvara. The same coarse device is seen on the Gardez GaneSa, while that of the Gardez Durga remains only on the upper part of the left shoulder. A fragment of a halo from Khair Khanah is also to be 237 added to the list of this type of ribbons, regardless of whether this halo was origin~lly attached to the Surya or to the main deity now lost from the pedestal with a warrior. It should be noted that the backs of these statues are usually rudimentarily executed, whereas the emphasis is on the elaborately finished back portion of the head. The significance of this is not clear to the writer, but it is just in this point that there is no connection between these marble pieces and the Gandharan Buddhist carvings. Hair, drapery, ring, boot, and others. Besides the above major four points, the treatment of hair, drapery, finger-ring and some ornaments have certain similarities. The top of the Qal' a Amir Mohammad Durga, the Skandar Uma and the warrior on the Khair Khanah pedestal, have their hair represented by parallel V-shaped lines from the front to the back of the head. Danda in the Surya piece has undulating lines at both sides, similar to the hair of the child on the proper left of the Skandar Uma. Even if it is not meant to represent hair, perhaps being some kind of head covering, the lines are concentrated on the sides only, leaving the centre top of the head without any decoration, and are the same in both cases. The following have the drapery carved in V-shaped lines: the Skandar Umamahesvara, the Khair Khanah Surya, the Gardez and Shakar Darah GaneSas, the Gardez and Scorretti Durgas, the Skandar pedestal as well as the Khair Khanah pedestal. The following wear the same type of finger-ring on their little fingers. the Mahesvara's hand holding the trisula, the Gardez Durga' s hand holding the sword, and the Shakar Darah Durga's right hand (probably holding the sword). The same representation of hoofs is found on Nandi of the Mahesvara and on the Mahisasura in the Gardez and Scorretti marbles. Furthermore, the same device is observable in the central leaf design of the Qal'a Amir Mohammad Durga's crown and the armlet on the front left arm of the Gardez Durga. The above discussion of the detailed elements in the marble pieces shows that some examples share absolute common characteristics, and that, if some similar peculiarities are pointed out, they seem to interrelate to each other. There are, of course, differences among the examples, such as the triple leaf decoration of the Gardez Siva and the single leaf of the Skandar Mahesvara, and also the rendering of their eyebrows. Some, like the Skandar Umamahes'vara, are frontally executed, while others, like Durga, are carved entirely three-dimensionally. Even those frontally oriented have emphasis on the back treatment. As noted above, in this sense, the connection of the marble pieces with the Gandharan schist carvings should not be accentuated, though it is properly admitted that there is a relation in the general composition between the Skandar UmamaheSvara and the schist Pancika-Haritis. After observing the interrelationship between the examples, we can group them under a common tradition, which suggests that they were carved in a limited period of time. But it remains difficult, of course, to fix each of them in their chronological arrangement. Among the details discussed above, the most important for their dating is the ribbons of the crowns, which are were usually attached to the Sassanian crowns. The 238 chronology of the Sassanian dynasty provides firm ground for dating. The following section, therefore, discusses the treatment of such ribbons in view of dating the marble pieces. 4. Treatment of Ribbons All Sassanian ribbons are represented flowing, as seen in the royal cliff reliefs, on silver bowls and on coins. Following is a classification of the Sassanian ribbons (Pope 1938; Herzfeld 1920; Paruck 1924; Dalton 1964; Lukonin 1967; GobI 1968). Type A: The end of ribbon, after the knot, has nail-shaped horizontal incisions; commoner than the type C. Type B: The end of ribbon, after the knot, can be divided into two sections, vertical incisions on the upper and the nail-shaped horizontal ones on the lower, as seen on the backs of marble pieces: the most common. Type C: Ribbons without a linear design: very rare. Up to the time of Bahram II (276-293) these three types appeared on all the above-mentioned Sassanian monuments. Particularly among the coins, type A was common and continued to be popular during the period of the first five kings, from Ardashir I to Bahram II. After Bahram II, beginning with Narseh (293-302), the representation of type A was replaced by type B. Type C can be seen only on some coins both of Shapur I (243-273) and of Hormizd I (273-274). The earliest known silver plates or bowls date from Bahram I (274-276) on which type B ribbons are depicted. The example from the reign of Bahram II shows type A. The ribbons represented in the cliff reliefs are of two types, A and B, except for the vertically incised example at Naqs-i Rajab. According to this, the earlier king, Ardashir I, used exclusively type A and the next king, Shapur I, preferred type B at Naqsh-i Rustam, while he also has the above-noted vertical type at Naqsh-i Rajab. Then type B became common after Shapur I with the exception of Taq-i Bustan where Ardashir II wears type A. The following is a classification of all ribbons in the chronological order found in Sassanian art: Kings and date Cliff reliefs. Ardashir 1(227 -243) A Silver Plates A Coins Shapur I (243-273) B A,C Hormizd I (273-274) Bahram I (274-276) C Bahram II (276-293) Narseh (293-302) Hormizd II (302-309) A B A B B B Shapur II (309-379) Ardashir II (379-383) A B B B B B B A B B Shapur III (383-388) 239 B Bahram IV (388-399) B Yazdegard I (399-420) B Bahram V (420-438) B B Yazdegard II (438-457) B Hormizd III (457-459) B B Peroz (459-483) B Balas (483-488) B Kavad I (488-531) B Khusrow I (531-579) B Hormizd IV (579-590) B Bahram IV (590-591) Khusrow II (591-628) B B B B B Yazdegard III (632-651) B B Type A was used during the early Sassanian period especially on the coins from Ardashir I to Bahram II. During the reign of Shapur I, type B was introduced and became popular after him in the representation of the royal relief on the cliff. In spite of the fact that type B came to be commonly used in the cliff reliefs, it was not until the end of the fourth century that type B completely overcame the other types of representation in all three classes of artefacts. Although conventionalized, it remained, particularly on the coins, until the middle of the seventh century when the last Sassanian kings lost their powers. The examples discussed above were forcibly limited to the dated objects. Further examples may be found, too, on the facade of Taq-i Bustan from the late Sassanian period. Among the silver objects, too, type B can be observed in the specimens of the late or post-Sassanian times (Pope 1938: 4, pIs. 159 B, 230 A). Such representation of ribbons is also found among the Sassanian type coins such as those of the KushanoSassanians, Hephthalites, and so forth, circulated in the regions east of the Sassanian empire. It is therefore not strange to find such ribbons on objects other than coins. In this connection, Bamiyan and Fonduqistan will be discussed below, and the farthest illustration will come from Kasmir. (1) Bamiyan This motif is frequently represented in the wall paintings and clay mouldings. Both types Band C are common in the Bamiyan valleys, especially in the niche of the 38m. Buddha (Kodera et al. 1971A: 43; 1971b) and its environs and in the caves at the foot of the 55 m. Buddha as well as at Kakrak. The most concentrated are in eleven caves at the foot of the larger Buddha where such ribbons are depicted in clay moulding. Here the caves I, II and XI, which have an octagonal plan surmounted by a dome, should be specially noted (Hackin 1933: 20-23, figs. III, V, VI). There are two bands between the dome and the walls. The lower band is composed of a row of trilobate arches with pointed tops, in which a seated Buddha must have been set. Between each arch were moulded kirtimukha. The upper band consists of a row of pointed or carinated arches, but in this case the arches are not trilobate but simple. 240 Kirtimukha is also observable. The type B ribbons are found fluttering at the pointed tops of the arches (Hackin 1933: pIs. XXX, XXXII, XXXIV). Related to this kind of ribbons, a clay frieze of foliage scroll must be specially mentioned (Hackin 1933: pI. XXXI), because this foliage scroll has the peculiarity that its leaves tend to be almost curled, as seen in the Dhamekh stupa at Sarnath. And in Bamiyan this trend is observed further on kirtimukhas (Hackin 1933: pI. XXXIII, 35, 36; pI. LXXXII 91). Such foliage scrolls are also used for the decoration of arches in the caves V, XI, D, F, and G (Hackin 1933:5-11). Hackin's 'sanctuaire' of the cave F must be considered as having the same tendency as the above-mentioned caves at the foot of the larger Buddha, since it has the same architectural aspects, such as the domed octagonal plan with pointed trilobate arches on the upper and the pointed simple arches on the lower band between the dome and walls. Cave D, having a single row of arches, forms an octagonal plan surmounted by a dome, and the pointed trilobate arch is decorated with a foliage scroll (Hackin 1933: pI. XI, 12). Caves V and G have also similar floral decoration, the former having it on the pointed trilobate arches and the latter having the painted foliage on the domed ceiling. It is noteworthy that the clay seated Buddha from Cave G now in the Musee Guimet has the same device at the hem of the drapery as that of the Skandar Uma. So far seven caves at Bamiyan have an octagonal plan surmounted by a dome. Of these, the decoration of the arches in caves III and VII no longer remains. The other five caves have simplified floral designs and kirtimukha, except for Cave F. In at least two of these caves the same type of ribbons as those of the Sassanian kings occur. A similar combination of foliage scroll, kirtimukha, and the type B ribbon can be admitted to have coexisted in the Bamiyan valley. The mud brick vihara no. 17, of the later stage of Tapa Sardar, with a rectangular plan revealed sculptures of unbaked reddish clay. The objects published by Taddei clearly show the characteristics shared by Bamiyan. Foliage scroll patterns, apparently coarser than those in Bamiyan, decorate the trilobate arches of unbaked clay and the curled leaf design of the foliage scroll is a main factor of the accessories worn by clay figures from the vihara (Taddei 1968: figs. 40, 42, 47, 60). Other notable elements among the later clay sculptures is the treatment of the hem of drapery of the seated Buddha (Taddei 1968: fig. 28). The folds of the hem which well recall the Y-shaped pattern of the marble sculpture show Tapa Sardar's relationonship with other contemporary art works. TheY-shaped hem is also reported from the other viharas such as no. 37, the naga images from there are strikingly close to those of Fonduqistan. (2) Fonduqistan This fortress-like Buddhist temple gives us two types of ribbons. One is the sumptuous crown of the sculptured Bodhisattva now in the Musee Guimet and the other is the crown of the painted Bodhisattva holding a blue lotus housed in the Kabul Museum (Hackin 1959: figs. 174, 189 and 196). The former was an attendant to the proper right of the central figure in the niche D which is one of the niches surrounding the stupa, and the latter was on the facade in which the niche E opened. The significant fact of Fonduqistan 241 is that it provides the date of its construction, which is lacking at the above-mentioned sites, Bamiyan and Tapa Sardar. From the niche E, a clay sculpture of probably an influential couple of the temple was found. Under their pedestal was unearthed a cremation urn containing ashes and two Sassanian silver coins as well as another kind of silver coin and some copper coins. Hackin identified the two Sassanian-type coins as being drachms of Khusrow II and left the others unidentified. Hence Hackin's dating of Fonduqistan was not earlier than 590. Ghirshman reexamined these silver coins and found on the obverse the Arabic legend 'Bismillahi' and the letters AIR which could set the place of issue at Susa. According to the further letters, reading hftsih, he dated the coin in the thirty-seventh year of the Islamic calendar, or 627. He also found on the obverse two punched heads of animals commonly used in Tokharistan, and thus interpreted the coins as having been countermarked by minor kings of Bamiyan and Fonduqistan. On the copper coins he also read the Tokharian scripts CPI 0 SAHO and a symbol mark of the Hephthalites, and therefore identified them as issued by the petty kings of the Bamiyan valley (Ghirshman 1948: 28-32). GobI proposed a much later dating for Fonduqistan by examining the countermarks struck on the above Sassanian coin, which he identified with his classified countermarks nos. 10 and 3, and he dated these to 70 and 63 H. respectively. The later of these corresponds to 689 (GobI 1967: II, 313-314). Thus the niche E cannot be earlier than 689. The same dating can be given to the stupa court, since all the niches around the stupa were constructed at the same time. As discussed earlier, the two types of ribbon are typical to the painted clay Bodhisattvas found in these niches and each niche is covered with the arch supported by clay-moulded pilasters (Hackin 1959: fig. 189). The ornamentation such as rough foliage scroll is same as at Bamiyan (Hackin 1959: figs. 157- 158). Bamiyan, Tapa Sardar and Fonduqistan share the same trends in architecture and sculpture. Type B ribbon appears in asociation with the above factors. Considering the date of Fonduqistan, these factors are datable to the time from the end of the 7th century to the early 8th. (3) Ushkar (fig. 27) Also on the terracotta head from Ushkar is the flapping ribbon motif. Cunningham (1871: 99-103) states that Ushkar is to be identified with Hushkapura which is mentioned in the Rajatarangini (Stein 1900: book I, 30, n. 168). Hushkapura can be traced back to the period of Huvishka, as it yielded the Kushan objects. Significant is that Lalitaditya Muktapida, according to the Rajatarangini, established a Vishnu temple and a large vihara with a stupa called Muktasvamin at Hushkapura (Stein 1900: book IV, v. 188). If the terra cotta head carne in fact from Ushkar, important is the chronology of Lalitaditya which differs among the various authors (Tripathi, 1937: 201 ff.;Ray 1931, I, 112: Gupta 1947: 14 ff.; Goetz 1969: 8 ff.; Stein 1900: Introduction, 67 ff.). A discussion follows as to his date. The chapter on Kasmir in the second section of vol. 221 of the Tangshu, it is clearly mentions that in the eighth year of the Kaiyuan era, Zhentuolabili was assured by 242 the Chinese court to be the king of Kasmir. He was recorded as a king who had offered some strange medicine to the Chinese court and that, after the death of Tianmu, his brother Muduobi became king of Kasmir (Chavannes 1901: 166). Hence it could be proposed that the order of succession of the Kasmir kings was Zhentuolabili, Tianmu and Muduobi. Furthermore, the Rajatarangini informs that the kings Candrapida, Tarapida, and Lalitaditya followed one another (Stein 1900: book IV, 130). Chavannes and Levi (1885: 350) assumed that Candrapida was Zhentuolabili. Therefore, Tianmu is identified with Tarapida, Muduobi with Muktapida. According to the Rajatarangini, Candrapida ruled for eight years and eight months, Tarapida for about two years, and Lalitaditya for thirty-six years (Stein 1900: I, Introduction, Appendix I, 136). Zhentuolabili was appointed by the Chinese court in the eighth year of the Kaiyuan era and hence by 720 had become king of Kasmir. Tarapida therefore is from 728 to 730, Lalitaditya between 730and 766. Particularly concerning Lalitaditya, it is stated in vol. 975 of the Cefu Yuangui that Muduobi of Kasmir sent the great master monk Wuliduonian, or Wuliduo (according to the Tangshu) , to the Chinese court (Chavannes 1901: 209). It indicates that in the twenty-first year of the Kaiyuan era, or 733, Muduobi or Muktapida was already enthroned and the above recorded the official report from Kasmir, of his succession, to the Tang court. Accordingly it could be assumed that the Vishnu temple and the Buddhist temple Muktasvamin at Hushkapura were constructed between 730 and 766. A group of terra cotta heads from Ushkar (Kak 1933: 152-154) could be also dated to this time span, and the ribbon shown on the head reproduced here is supposed to have been used in the middle of the eighth century. This fact shows that the Sassanian ribbon types discussed here were used in Kasmir and its environs, namely in the regions geographically close to those where the marble sculptures discussed here were found, at least until the middle of the eighth century. The ribbons seen on the marble pieces are of the same type as one of the Sassanian types which was predominant in the later period of the Sassanian rule. This trend is confined not only to the Sassanian empire proper but also seen in the regions east of the empire, such as Bamiyan, Fonduqistan and even the more eastern lands. In F onduqistan and Bamiyan ribbons are combined with rough foliage scroll. At Bamiyan the ribbon and the foliage scroll are further combined with trilobate arches. Tapa Sardar has also trilobate arches decorated with only foliage scrolls. In Kasmir the ribbons become coarser, but are related to a religious site dated to the eighth century, and would seem to be the latest examples of the ribbons under discussion. The date of these elements at Bamiyan and Tapa Sardar cannot be definitive, whereas at Fonduqistan the silver drachm makes it possible to define the date of the site. The Brahmanical sculptures with representations of ribbons should be duly supposed to have been carved and worshipped in the same periods as the non-marble and nonBrahmanical art production seen in Bamiyan, Ghazni, and Kasmir. Nothing prevents us from stating the above fact, if the following is added too: similar treatment of the hem of the drapery, the repeated Y-shaped patterns, was also observed among the sculptures 2+3 uncovered at Bamiyan, Fonduqistan and Tapa Sardar. Moreover, the herring-bone or leaf-shaped designs are common amongst the objects found in the niche E at Fonduqistan and from Tapa Skandar, Khair Khanah, and so on. Furthermore, a decoration of the crowns seen on the Durga and on the painted Bodhisattva of the niche E at Fonduqistan is one of the clear evidences which places these artistic activities in the same period. It should be stressed that during the periods just before the advent of the Muslim power the Brahmanical statues were carved in marble and the Buddhist in clay and that religious difference must have dictated the choice of materials. It should be noted, however, that in the detailed elements such as ribbons, drapery, decoration of crowns, these sculptures disclose the fashion of the time which goes beyond the religious difference. 5. Brahmanical Images and the Hindu Shahis There have been various views on the dating of the marble pieces, as being from the Turki Shahis or the Hindu Shahis, with no definite dating on either. Among the pieces in the Kabul Museum, for example, all pieces except for the Khair Khanah Surya are regarded as being from the Hindu Shahi period (Anonym 1964:18-19). Nevetheless, these objects do not fit well with historical background. The following is a discussion that they do not belong to the Hindu Shahi period despite their clear orientation towards Brahmanism. Biruni states in the Ta'rikh ai-Hind, regarding the presence of the Shahis at Kabul: The Hindus had kings residing in Kabul, Turks who were said to be of Tibetan origin. The first of them, Barhatakin, came into the country and entered a cave in Kabul, which none could enter except by creeping on hands and knees .... Some days after he had entered the cave, he began to creep out of it in the presence of the people, who looked on him as a newborn baby. He wore Turkish dress, a short tunic open in front, a high hat, boots and arms. Now people honoured him as a being of miraculous origin, who had been destined to be king, and in fact he brought those countries under his sway and ruled them under the title of a shahiya of Kabul. The rule remained among his descendants for generations, the number of which is said to be about sixty.... The last king of this race was Lagaturman, and his Vazir was Kallar, a Brahman. The latter had been fortunate, in so far as he had found by accident hidden treasures, which gave him much influence and power. In consequence, the last king of this Tibetan house, after it had held the royal power for so long a period, let it by degrees slip from his hands. Besides, Lagaturman had bad manners and a worse behaviour, on account of which people complained of him greatly to the Vazir. Now the Vazir put him in chains and imprisoned him for correction, but then he himself found ruling sweet, his riches enabled him to carry out his plans, and so he occupied the royal throne. After him ruled the Brahman kings Samand (Samanta), Kamalu, 24+ Bhim (Bhima), ]aipal (Jayapala), Anandapal, Tarojanapala (Trilocanapala). The latter was killed AH 412 (1021), and his son Bhimapala five years later (1026). This Hindu Shahiya dynasty is now extinct, and of the whole house there is no longer the slightest remnant in existence (Sachau 1888: II, 13-18). The above passage states that the Hindu Shahi dynasty, which followed Brahmanism, took over the Shahis of Kabul, a Turkish dynasty at Kabul. The chronology of the Hindu Shahis is only clear as far as the last king is concerned, and nothing is known of the date of the first king, the Brahman wazir of the Turkish dynasty. The only definite source which can give a clue as to the date of Kallar is the Rajatarangini. In the chapter on the Kasmir king Sankaravarman, a remarkable king called Lalliya Shahi is recorded as the king who resided in the strong fortress at Udabhandapura and was the staunch v.232). supporter of Alakhana, king of Gurjara (Stein1900: V, 206, vv. 152-55; 2 Udabhandapura is Hund as Stein rightly claims (Stein 1900: II, 336, note ]), despite Caroe's identification (1962: 97-98) to a tepe near Lahore (in proximity of Hund). It is on the north bank of the Indus to the north of Attock. According to Stein, the date of Sankaravarman covers 883-902, and his successor, Gopalavarman, bestowed the rebellious kingdom of the Shahis upon Toramana, son of Lalliya and gave him the name Kamaluka (Stein 1900: I, 137, book v. 233). Stein states that Kamalu, the third ruler in Biruni's list of the Hindu Shahis, was the same person as Kamaluka contemporary to Gopalavarman (902-904). Moreover, Cunningham and Seybold as well as Stein had the same view on the identification of Kallar with Lalliya who was recorded in the Rajatarangini to have resided in Udabhandapura during the reign of Samkaravarman (Stein 1900: II, 336, note]; Seybold 1891: 700; Cunningham 1975: 82 ff.). For this sreason, if we follow the datings of kings in the Rajatarangini identified by Stein, Kallar or Lalliya, the first ruler of the Hindu Shahis, must have resided at Udabhandapura by 883, the first regnal year of Samkaravarman and that Kamalu or Kamaluka, the third king of the Hindu Shahis, was enthroned between 902 and 904. So Kallar or Lalliya must have been in his position before 883. According to the Ta'rikh-i Sistan, Ya'qub b. Lais, the second ruler of the Saffarid Dynasty, was declared amir in Sijistan in AH 248 (861), and a few years later invaded the Khalijids in Kirman and Rutbil in ar-Rukhaj as well as the Shah of Kabul. During this expedition he entered Ghazni from the south via Kandahar and gained complete control of the Kabul regions in AH 257 (870). This occupation had a great influence over the regions as seen by the fact that the war trophies gained from Bamiyan and the heretical idols stolen from the great temple at Kabul were sent as gifts to the Caliph in Baghdad. In general, Islamic influences are believed to have reached the Kabul region earlier, the above occupation being the first decisive blow by the Muslims (Majumdar, ed. 1962: 167-68; Ghafoor 1965-66: 4-12). After this invasion the Kabul region was ruled by 'Amru b. Lais, succeeding his brother Ya'qub who died in 879. In this year the Hindu Shahi regained control of the Kabul regions, particularly Logar and]alalabad. According to the Jami' ul-Hikayat by Muhammad 'Ufi, Fardghan was sent to Zabulistan as chief magistrate during the reign of Amru. 245 There was a large Hindu place of worship in that country which was called Sakawand, and people used to come on pilgr.image from the most remote parts of Hindustan to the idols of that place. When Fardghan arrived in Zabulistan he led his army against it, took the temple, broke the idols in pieces, and overthrew the idolaters .... When the news of the fall of Sakawand reached Kamalu, who was Rai of Hindustan, he collected an innumerable army and marched towards Zabulistan (Elliot 1869: II, 172). According to the above, it may be assumed that, (1) Kallar already resided at Vdabhandapura during the reign of the Kasmir king Sankaravarman and its terminus ante quem is 883. (2) In 879, the Hindu Shahi recovered the territory including Kabul, Logar and Jalalabad (3) The Hindu Shahi who recovered the territory is Kallar, if no. 1 is right. (4) Accordingly, Kallar abandoned Kabul and moved to Vdabhandapura before 879. (5) Possibly Kallar abandoned Kabul when Ya'qub b. Lais conquered there in 870. (6) In consequence, Kallar took the throne from Lagaturman before 870. But if Kamlu in the Jami al-Hikayat is the same as Kamalu in Biruni's list and as Kamaluka in the Rajatarangini, Stein's dating Kamaluka looses ground, because Kamlu, Rai of Hindustan, existed during the reign of 'Amru b. Lais (879-900), and Gopalavarman (902-904) bestowed the kingdom of the rebellious ~ahi upon Toramana, or Kamaluka. Taking into the life span of Kallar into consideration the Hindu Shahis definitely appeared during the second half of the ninth century, not earlier, despite the conflicting dates of the rulers. There is a gap between the date of the Hindu Shahis and that of the marble pieces. It is not therefore possible to attribute these pieces to the Hindu Shahi period. They should be attributed to the Shahi period before the Hindu Shahis originated by the Brahman wazir Kallar, that is, the Turki Shahis. This is true for all pieces, except for the ekamukhalinga and Vishnu caturmurti found in the Gandhara region as well as the two sculptures from Ghazni: there are no common features between such objects and the pieces found from the Tagao-Kabul-Gardez regions. The technique of elaborate chiselling of the Peshawar ekamukhalinga's lower octagonal part is quite different from that of rough carving of the Vma Mahesvara. Between the hair and accessories, particularly of the Attock Vishnu, and those found in eastern Afghanistan is a technical gap. Although no definite dating is possible, it seems that the objects from Gandhara must be from a later period, probably from the time when the Hindu Shahis moved from Kabul to V dabhandapura. There is no mention among Biruni's records that during the sixty generations from Barhatakin to Lagaturman anyone followed Brahmanism, although the Hindus were ruling. Even if Biruni correctly says that Barhatakin was the founder of the Turki 2+6 Shahi, he gives no date. Only the Rajatarangini states that during the reign of Lalitaditya in the middle of the eighth century one of his high officials was a Shahi (Stein 1900: I, book IV, vv. 142-43, note 140-43). Since this date is the mid-eighth century, it is quite certain that this Shahi is not of the Hindu Shahis but of the Turki Shahis, who were ruling the Kabul regions. The Chinese Buddhist pilgrims who came to the Kabul-Gandhara-Kasmir regions in the eighth century recorded that the regions were ruled by the Turks. Wukong arrived at Gandhara in the twelfth year of Tianbao (753). He recountedthat the Kapisi country had its eastern capital in Gandhara and that the king resided in winter here and in summer in Kapisi. Then he entered Kasmir during the second year of the Zhide era (756) and stayed until 760. Wukong recorded that Buddhist temples were dedicated in Kasmir by the Turkish royalties such as the Yelitegin temple founded by a Turkish prince and the temple founded by a Turkish queen. After returning to Gandhara in 760, he also observed two temples with the same names as above, built by the Turkish prince and his queen (Levi and Chavannes 1885:349-57). According to Huichao, who is believed to have stayed in Kapisi and Gandhara during the first half of the eighth century, the kings and armies of both these countries were all Turks, and those under the their rule were the Hu, or Iranians, among whom also were the Brahmans. This coincides with Biruni's notice that the rule of the Turkish dynasty covers the Hindus-the writer assumes that the word Hu used by Huichao represents non-Turkish people, but that the 'Hindus' used by Biruni denotes the people living in this part of India. Huichao also describes that the Turkish rulers of Gandhara were followers of the Triratna and many Buddhist temples were dedicated by the members of that ruling class. The record of the king holding pancapari~ad twice a year suggests his deep devotion to Buddhism (Fuchs 1939: 18-25). According to Xuanzang, followers of Brahmanism were found from Kapisi to the east countries into all of India. He referred to the Deva temples and heretics, and although this might include the J ains (Fischer 1969: 339, 343), their influence must have been quite large in the second quarter of the seventh century. In Kapisi, beside Diganbaras, the naked ascetics, there were described by him pasupatas, or those who cover themselves with ashes, and those who make chaplets of bones, which they wear as crowns (Beal 1884: I, 55.). The latter Saiva sect, that of the Svetasvataras, had a town or castle called Xibiduofaluo- about 40 Chinese miles from the capital of Kapisi, the modern Begram or Abdullah-i Borj (Beal 1885: I, 61). That this town is possibly identifiable with Tapa Skandar has often been argued by the author (See note 20.). Xuanzang also described how popular Buddhism was in Laghman. There there were only five Deva temples and about a hundred heretics, and that in Jalalabad there were about ten Buddhist temples with a small Buddhist population against numerous Deva temples and their devotees (Beal 1885: 91). In Gandhara the majority of the population believed in Brahmanism with more than a hundred Deva temples. Even their exact locations were given by Xuanzang: outside of the gate of Pushkalavati there was one Deva temple and on the top of a high mountain, north of Purushapura, was a statue of Bhima, consort of Mahesvara, and at the foot of this mountain was a shrine of Mahe§vara which was kept by pasupatas 247 (BeaI1885: 98,109,113). In the account of Ghazna, followers of the -deity Zhuna were highly influential, along with some remaining Buddhists. Xuanzang does not state whether the rulers themselves of these regions worshipped Brahmanical idols, except in the above GhaznaGardez regions, where the Zhuna was worshipped by having rich offerings presented to him and prayers said to him, not only by the inhabitants of these regions but also by votaries of all classes from the neighbouring countries (Beal 1885: 284-285). Although Zhuna has not been well explained as to its religious character, Brahmanism was becoming more popular in any way. It could be assumed that the Shahi Khingala of Uddiyana, of the Gardez Ganes'a inscription, was one of these followers. Gradually among the ruling class an interest towards Brahmanism grew when it became an ascendant religion against the declining Buddhism. A ruler of the ninth century such as Kallar could obtain political powers and ru]e over the large Brahmanical population, patronizing and utilizing Brahmanical powers. The Durga of Tapa Sardar unearthed by the Italian Archaeological Mission is an unquestionably unique piece in that it occupied a position in the Buddhist context and that it is made in clay, not carved from marble. The religious difference must have dictated the choice of materials. Taddei (1973: 203-213) says that the Tapa Sardar Durga shows local Hindu-Buddhist syncretism in the latest phase of Buddhism in that area. Brahmanism and Buddhism thus coexisted especially during the seventh-eighth centuries just before the Muslim power covered east Afghanistan. Postscript 1. The dating of the Khair Khanah site itself has been proposed by the author aftcr the original publication of this article. The construction of the later temple in which the Surya image was enshrined is dated between 608 and 629 (Kuwayama 1975: 93-107). 2. In 1974, the author found two marble sculptures of unknown provenance at an antique shop in "Chicken Street" in Kabul. One of them is a single block of marble with a slight remnant of carved drapery and the other, now housed in the National Museum of Afghanistan, is possibly a caturmukhalinga, but the probable phallic top is missing. The lower part of the linga is surrounded by four crowned heads of different characters, one of which may represent Sri Devi and all of whose faces are missing except some of the eyes and eyebrows. These are carved on the square pedestal. All four sides have the polished upper band on the coarsely chiselled lower base. Hence this piece should be attributed to the same group as discussed here. The total preserved height, 36 cm.; the height of pedestal, c. 16 cm. including the 4 cm.-high polished band; the width of both front and back, c. 23 cm., and the width of the sides, c. 15 cm. 248 xv The Inscription of Ganesa from Gardez and a Chronology of the Turki Shahis 1. More about the Mghan Marbles. Of the marble Brahmanical sculptures found in the areas of east Afghanistan from KapisiKabul through Gardez to Laghman, only a few were found under controlled excavations, such as the seated Surya image and others dug in 1934 by the French Archaeological Delegation at Khair Khana in a northern suburb of modern Kabul (Hackin and Carl 1936: PIs. XI-XVI and XXIII, 32) and the seated Uma Mahesvara image on the bull Nandin and other fragmentary pieces found in the years between 1970 and 1978 by the Japanese Archaeological Mission of Kyoto University at Tapa Skandar 30 km to the north of Kabul (Kuwayama 1972: Illus. 1 and 1974, Illus. 21). In addition to the paucity of the regularly excavated sculptures, only the Gardez Ganesa and the Tapa Skandar Uma Mahesvara bear inscriptions, the latter of which is a hymn dedicated to the Lord Mahesvara and does not contain any clues for precise dating (Yamada 1972; Gupta and Sircar 1972-73; Mirashi 1975). This situation has raised many difficulties for locating this group of sculptures in a proper chronological and historical context. Only because the sculptures are Brahmanical in character, some scholars have attributed them to the time of the Hindu Shahis who supplanted Lagaturman, the last ruler in the line of the Turki Shahi, in the middle of the ninth century, or more precisely in 843 as recently suggested by Abdur Rahman on the basis of his interpretation of the Hund slab inscription (1979: 52, 309-316). While several scholars, including Verardi (1977), had published from time to time selected marble pieces to give them their own dating, the material available until 1970 was dealt with as a whole only by the present writer (1976). According to that analysis, all of the marble sculptures carry a common artistic tradition which is partly shared with contemporary examples of Buddhist art in east Afghanistan. The common characteristics between the two different groups of religious sculptures can give to the whole group of marble Brahmanical sculptures a seventh-eighth century chronological framework, the roughly latter half of which covers the reigns of the Turki Shahis from the point of view of the political history of the Kabul Valleys. This framework excludes the possibility of attributing them to the time of the Hindu Shahis. The precise discussion did not address the question whether the Afghan marble pieces are comparable with other Brahmanical sculptures produced in the geographical proximity of the Kabul Valleys. A recent suggestion given in 1989 by Siudmak in a personal communication on his short visit to Kyoto is appreciable in this connexion: the 249 pentagon-shaped mukuta of the Tapa Skandar Mahesvara evidently shows close similarity with the central head of the Siva Triurti from Pandrethan in the Sri Pratap Singh Museum, Srinagar, which is, according to him, dated to the first half of the seventh century (Siudmak 1989: fig. 8 on p. 47). His suggestion has led me to investigate further the relationship between some of the Afghan marbles and the stone sculptures from Pandrethan where similarities seem closer than at other precincts around Srinagar. Decorative elements similar to those of the Tapa Skandar Uma and the Tagao Durga can now be pointed out on the crowns of two such stone sculptures as a Padmapani and a Saivite deity in the same museum (Siudmak 1989: figs. 12 and 13 on p. 50). Also highly suggestive for a comparison of at least two Afghan marbles-the Tapa Skandar Uma Mahe§vara and the Shakar Dara Ganesa-with the Kasmiri stone sculptures is the comprehensive contribution recently published as his doctoral thesis by Pran-Gopal Paul (1986). Of the two Brahmanical marble statues with inscriptions, the standing GaneSa (fig. 1) from Gardez, now in the Pir Rathan Nath Dargah in Kabul, is the most important for providing a signpost for the relative stratigraphy of the group of about two dozen sculptures revealed in east Mghanistan. This short note specifically concerns itself with the names of a king in the inscription of this statue. The king turns out to be chronologically significant in the eighth century stratum when dealt with in relation to other relevant historical factors. 2. The Ganesa Inscription Special attention was drawn to the inscription when first made public in the ditto copy of the Preliminary Report of the Indian Archaeological Delegation to Afghanistan by T. N. Ramachandran and Y. D. Sharma. Although this report has not been available to me, the same statue and inscription were later published by Tucci at the beginning of his mission to Swat in the footnotes of his article entitled 'Preliminary Report on an Archaeological Survey in Swat' which appeared in the East and West, Vol. 9, No.4 (p. 328). The statue has striking importance for the two-line Acute-Angled Brahmi inscription in Sanskrit (Illus. 34) on the polished uppermost part of its plinth, which was read by Tucci in that report based on a photograph provided by Fernando Scorretti. A reproduction of his interpretation follows: [Line 1] Olp salpvatsare a~ratame sam 8 jye~tha-masa-sukla-pak~a-tithau ttrayodaSyam sudi 10-3 ri~e visakhe subhe silphe ci .. [Line 2] mahat-prati~thapitam idalp mahavinayaka-paramabhanarakamaharajadhiraja-srl-~ahi khilpgalo-tyana-~ahipadai[b] (Floral Mark) The inscription states, according to Tucci, that this great and beautiful Mahavinayaka was consecrated by the renowned $ahi king, the illustrious $ahi Khilpgala, Paramabhanaraka-maharajadhiraja, in the 13th of tithi of the white fortnight of jye~tha 250 under the na~atra Visakha and the lagna of the lion in the year eight. Tucci especially remarks that it is impossible to know to which era this inscription refers, but he thinks that the characters seem to be of the end of the fifth and beginning of the sixth centuries. Stimulating is his cautious reading of the name otyana (Uddiyana), attached to the name of the king Khirpgala in the last line. This is important not only as the first inscribed mention of Uddiyana but also the first mention of a king's name in that country. Tucci thinks the reading is not absolutely sure, even with the support of Dr. B. Ch. Chhabra, yet of considerable interest is how and whether Sri Sahi Khirpgala is related with similar names known from certain coins (Deva Shahi Khingi[la]) (Cunningham 1962: .265, 278-279, PI. VII, fig. 11) and in the Rajatarangini (Khinkhila Narendraditya I) (Stein 1900: I, 52, taranga 347). Tucci finds no plausible reason to dissociate the king in the inscription from those on the coins and in that Kasmiri chronicle. D. C. Sircar also read the inscription, based on the photograph attached to the original report, to dissent from Tucci's view at several points (Sircar 1963: 44-47). First of all, he thinks that the characters belong to the Siddhamatrika alphabet of the sixth or the seventh century and that the inscription is clearly not much later than the middle of the seventh century: the use of the tripartite form of the letter y was noticed in eastern India in such early seventh century inscriptions as the Patiakella plate of SambhuYaSas (dated 602) and the Dubi plates of Bhaskaravarman (c. 600-650) and in Rajasthan rarely in late seventh century epigraphs like the Dhulev plate of Bhetti dated in the Harsha year 73 (679). His reading follows: [Line 1] [Siddham] [/1*] Sa[rp]vatsare a~tatame sarp 8 Jye~tha-masa-sukla­ ttrayodaSyarp su-di 10-3 ji(ri)k[~]e visakhe subhe sirp[he] chi [traka] [Line 2] mahat-prati~hapitam=idarp maha-vinayaka paramabhanarakamaharajadhiraja-srl-~ahi-khirpgalautyata-~ahi-padai[l:t] (Floral Mark) pak~a-tilau(thau) Sircar delved into several different points of the inscription, among which one of the more crucial is the last passage containing Tucci's otyana and the Shahi king Khirpgala. He revised the reading of the latter as Khirpgala, rightly pointing out that the vowel-mark with the letter I should be au, neither ai nor o. According to him, moreover, the two akSaras following lau can be read as tyiita, not tyiina, since the letter n which only appears in the word vinayaka has the left and right lower limbs engraved shorter than those of t which occurs several times in the same inscription. Hence his doubt about Tucci's Uddiyana associated with a ~ahi king. He maintains that there was no khirpgalotyana ~ahi but khirpgala otyata sahi and that otyata was a second name of Khirpgala, probably a title not connected with the territory over which he ruled. He gives such similar instances in the Gilgit manuscripts as Patoladeva-~ahi Vajraditya Nandin, Sri-deva-~ahi Surendravikramaditya Nanda (Nandin), Sahanu~ahi Patola-~ahi Sri-nava-Surendraditya Nandideva, all of them bearing double names and having ruled over the Darada country in the upper valleys of the Kishanganga in the seventh century. Although the possibility still exists that the image 251 itself was carried from the Swat Valley to Gardez, Sircar is therefore sceptical of the existence of any ~ahis of Swat who might have occupied the Gardez region within the dominions of the ~ahis of Kapisi or Kabul, probably the strongest amongst the ~ahi houses if any ~ahi house actually ruled there. The problem of the date of the king was further dealt with by M. K. Dhavalikar (1971: 331-336). He does not find any reason for the identification of Khinkhila Narendraditya in the Rajatarangini with the king Khirpgala for the reason that the latter is explicitly recorded as a ~ahi king. Yet he inclines to identify Khirpgala with the issuer of those coins bearing the legend Khinga[la] or Khingi[la] whom Vincent Smith ascribed to the end of the fifth century (Smith 1906: PI. XXVII, fig. 1). Dhavalikar does not give any definite date to the king of the inscription, agreeing with Tucci that there are too many uncertainties, but expects the Ganesa statue can be dated to the early sixth century at the latest on the stylistic peculiarity of the sculpture. His 'stylistic' points, however, are specifically based on Indian iconographic trends of the post-Gupta period which seem to me other than stylistic comparisons and ignore possible relations to otl1er marble sculptures in east Mghanistan. 3. ANew Interpretation of the Inscription. During the third campaign in 1974 of the excavations at Tapa Skandar, I had a chance to visit the Pir Rathan N ath Dargah in Kabul with the guidance of a Hindu clerk in the Gosho Company in that city. The statue in question had temporarily been removed for relocation within the temple and was at that time standing free against a wall. The photographs shown in figs. 1 and 2 were taken on that nice opportunity which also allowed me to make a good ink rubbing of the inscription (Illus. 34). I rendered a copy of the rubbing to Prof. Hideaki Nakatani who provided the following detailed reading. His reading enabled us, against Sircar, to identify the name of the king as khirpgalauQiyana. He admits there are many mistakes in the inscription of either engraver or writer such as the absence of a horizontal stroke in the letters th of tithau and f of sri. Further, the words a~tatame, jye~tha, ri~e, visakhe, Mahavinayaka and paramabhaneraka should have been engraved as ~tame, jye~tha, rik~e, vaisakhe, mahavinayakam and paramabhanaraka respectively. The most serious mistake is a~ta-tame that seems to expose a low level of knowledge about Sanskrit. His reading and translation follow: [Line 1] sarpvatsare a~tatame sarp 8 jye~tha-masa-sukla-pak~a-tithau ttrayodaSYarp su. di. 10. 3. ri~e visakhe subhe sirphe (citra-) [Line 2] (k)e mahat prati~thapitam idarp maha-vinayaka paramabhanerakamaharaja-dhiraja-sri-~ahi khirpgalauQyana-~ahi-padail). On the thirteenth day of the bright half of the month of Jy~tha, the [lunar] mansion being the Visakha at the auspicious time when the zodiacal sign Lion was bright on the horizon (lagna), in the year eight, this great [image] of the 252 Mahavinayaka was consecrated by the supreme lord, the great king, the king of the kings, sri ~ahi Khirhgala, the king of O<;lyana. According to Nakatani, closest to the scripts of this inscription as a whole are the styles called by Sander as the second Gilgit-Bamiyan type and those represented by the first half, or the medical part, of the Bauer Manuscripts, usually assigned to the earlier half of the sixth century. The latter is a forerunner of the Gilgit-Bamiyan type II which later developed into the Sarada scripts and is dated by Sander herself to the time between the sixth and the tenth centuries (Sander 1968). The Ganesa inscription itself includes scripts related to each of these styles. To review previous datings, palaeo graphically Tucci thought that the inscription was datable to sometime between the end of the fifth century and the beginning of the sixth century. Sircar assigned it to the sixth or seventh century which was also accepted by Petech (1988, 187-189). The datings given by Tucci, Sircar and Petech to the styles of the letters of this inscription therefore all fall in the earlier part of Sander's chronology. However, there are several elements that seem to converge towards a later date for this inscription. One is the form of the letter y which was the only reason taken up by Sircar for his dating. Nakatani finds a form similar to this letter of this inscription in the manuscripts preserved in Buddhist monasteries in Japan and, based on the Bonji Kichoshiryo Shusei (Tokyo, 1980), he believes it attributable to the eighth century. Nakatani also suggests the other point from which this inscription might broadly be dated is the word paramabhattaraka. After being examined throughout the Buddhist inscriptions in India and Pakistan collected and edited by M. Sizutani (1978), this particular phrase does not seem to have appeared in any epigraph of the Kushan or Gupta periods, but often occurs much later in Pala times. This fact draws the inscription further from rather than closer to the Gupta period. The other important point for dating is the samvat. The readers of the inscription such as Sircar and Petech take the year eight as the regnal year of the king. Although there is no particular reason for ousting their idea, one may be allowed to suppose the possibility that the samvat eight is referrable to either the Vikrama era or the Laukika era the actual date being shortened to a single digit. Since the statue itself stylistically fits well with other Brahmanical sculptures of east Mghanistan in the seventh and the eighth centuries as demonstrated elsewhere (Kuwayama 1976), the candidates for the real year under the Vikrama era might be either 608 or 708, which would in turn be identical with 665 and 765 respectively reckoned from the vernal equinox of 57 Be. If reckoned on the basis of the Laukika era, the possible year might be either 3708, which is equal to 632/633 (= 3708 - 307613075), or 3808, which is identical to 7321733 (= 3808 - 3076/3075). 4. Stratigraphy of Khingala Before discussing the real year of consecration of the Mahavinayaka image by a Turki or Kabul Shahi who ruled over Swat, I shall below deal with the question whether 253 kings from other sources with similar names might be one and the same. In addition to the names on coins and in the Kasmiri chronicle, two other kings with similar names have been noticed by Petech from both Chinese and Muslim sources. He quotes al-Ya'qui (1969. 479) who mentions Khinkhil (Khinjal or Khinjil), an Arabic transcribed form of Khingila or Khirp.gala, in the time of al-Mahdi (775-785). Al-Ya'qubi says that al-Mahdi sent to kings of various countries messengers who asked for their submission and that many of them did submit among whom were the Kabul Shah called Khinkhil and others. The Chinese historical sources give the name of a local king in Kapisi as Xingnie, which may phonetically be restorable in Middle Chinese as *x[i]all-llar/*henger/Khingar or Khingal, as Petech rightly mentioned. In Volume 198 of the Jiu Tangshu , Volume 99 of the Tang Huiyao and Volume 222a of the Tangshu the account for the eighth year of the Xianqing era (658) states that Kapisi has as king Hejiezhi, a possible Middle Chinese form of which might be *yaryartSie<*ghar-ilci. He is the twelfth king of the line inaugurated by Xingnie. According to the Cefu Yuangui (Vol. 970), Kapisi reported to the Tang court-on the occasion of their tribute on the eleventh month of the fourth year of the Yonghui era (653)-that the princely heir had succeeded to the throne in their country. This account most probably indicates that in 653 Hejiezhi (*ghar-ilci) had already ascended the throne in Kapisi. Consequently we have several similar names of a king or kings in the historical context of the Northwest and eastern Afghanistan such as Sri ~ahi Khirp.gala in the inscription, Deva ~ahi Khingila and Khingi or Khinga on coins, Khinkhila Narendraditya I in the Rajatarangini, Khinkhil in Ya'qubi's Ta'rikh and Khingal in the Chinese sources of the Tang period. Petech thinks that Khirp.gala was not a personal name but an eponym based on the name of the dynasty which, as Chinese sources inform, passed down from generation to generation in Kapisi. However, it does not seem reasonable to regard Khirp.gala as a dynastic name. The Rajatarangini (Stein 1900: 1,52, Taranga 347) says that 'his [Gokar1).a's] son Narendaditya who bore the second name of Khinkhila consecrated shrines to Bhute§vara .... ' In this case Khinkhila clearly indicates a personal name. In the same way, Deva ~ahi Khingila also is a distinct personality among the many Hephthalite kings known from their coins-such as Toramana and Mihirakula. I have elsewhere discussed the withdrawal of the Hephthalite power from the Northwest in the late sixth and the seventh decades of the sixth century when the West Turks invaded the homeland of the Hephthalites in Tokharistan (Kuwayama 1990). Due to this invasion the ties of the Hephthalites in the Northwest, headed by the tegin, with their headquarters in Tokharistan were cut. It is believable that, as a consequence, the vast region extending from Kapisi in the west to Gandhara in the east fell into the hands of a local dynasty. Xuanzang mentions in Volume 1 of the Da Tang Xiyu ji that on his visit there in 630 the king of Kapisi belonged to a family of Chali which is usually assigned to the ~atriya, one of the Indian catur-varrya. In the manuscripts of the Da Tang Xiyu ji datable to the Tang period found in Dunhuang or in others preserved in existing Buddhist monasteries in Japan such as Chusonji and Ishiyamadera, Chali is replaced by Suli which usually means Sogd. Yet the existence of a Sogdian king of Kapisi 254 does not seem to have any historical reality and this word might presumably be derived from a copyist's fallacy, at least if one does not take seriously the statement of the Suishu that the king of Cao (Kapisi as noted below) had the family name Zhaowu which was usually held by local Sogdian dihqans. As discussed below, one may find evidence of confusion in the account of the Cao country in the existing Suishu. Even if we take Chali as representing the Iranian Kshathriya, this statement in the Da Tang Xiyu ji may at least indicate that his house was not a descendant of the nomadic Hephthalites but indigenous. Based on the Tang sources mentioned above, there is a strong possibility that this house was inaugurated by Xingnie, i.e., Khingal. If we take it as right that the king Hejiezhi (*ghar-ilci) was the twelfth king who ascended the throne in or before 653, Xingnie (Khingal) most probably founded his dynasty of Kapisi in the late sixth century, released from the yoke of the Hephthalites. In this context it may be right to think that the king of Kapisi whom Xuanzang first met in his capital city, modern Begram, in 630 and again at Udabhandapura, a town on the Indus in his easternmost territory, in 643 or 644 was in the line of the Xingnie Dynasty. The late sixth century date of Khingal, the first king of the local Kapisian dynasty in the Chinese sources, is thus ostensibly close to that of Deva ~ahi Khingila on the coins and Khinkhila Narendraditya in theRajatarangini. But Khingal must, in fact, be quite distinct from both of them. First of all, the coinage of Deva ~ahi Khingila clearly belongs to a peculiar series of Hephthalite coins, the obverse of which traditionally bore a characteristic bust of a king. It is now evident that the first king of the Kapisian dynasty is quite different from the Hephthalites and later than them. Therefore he cannot be identical with Deva ~ahi Khingila on the coins. On the other hand, in the Rajatarangini Kalhal)a attributes Khinkhila Narendraditya to a dynasty called Gonancfiya which was much earlier than the Karkota Dynasty. The Karkota Dynasty was founded by Durlabhavardhana Prajfiaditya who was, according to the author of the Rajatarangini, on the throne between the years 3677 and 3713 of the Laukika era (601/602-637/638) and contemporaneous with Xuanzang who passed Kapisi and Kasmir some sixty years after the Hephthalites' withdrawal. Specifically, Xuanzang was received in Kasmir by Prajfiaditya in 630. Stein found confusion in Kalhana's stratigraphy of the kings before the Karkota Dynasty and properly tried to correct their order. According to him, Khinkhil Narendraditya 'must be recognized in the Ephthalite ruler who calls himself on his coin Deva ~ahi Khiilgila' and 'the thoroughly un-Indian character of the name KhiIikhila and the references (in the Rajatarangint) to some religious endowments attributed to this king seem to lend support to the belief in his historical existence' (Stein 1900: I, 80). If we accept Stein, Khinkhil Narendraditya cannot be identical with Khingal (Xingnie) for the same reason as mentioned above: the former is Hephthalite and the latter is nonHephthalite. Nevertheless, this conclusion does not solve the question whether any Hephthalite or HUl)a king really resided in and ruled over Kasmir. In Song Yun's account of his travel to the Hephthalite kingdom we do have a clear reference to a Hephthalite king (tegin) of Gandhara who was fighting by Jhelum against Kasmir which is called Jibin in the Luoyan Qielan ji (Kuwayama 1990: 95-97) but we have no solid clue to identify Khiilkhil in the Kasmir chronicle with Khingila on the Hephthalite coins. 255 There is other evidence for separating Xingnie (Khingal) from the Hephthalites: the difference in coinage. The local dynasty inaugurated by Khingal supposedly issued coins which were quite different from those bearing the peculiar busts of kings of the Hephthalites. In fact they appear to be one of the coin types called by GobI as Nspk MLK and by Harmatta as Nezak Shah on the obverse of which is depicted a king wearing a crown surmounted by a bull's head (GobI 1967: I, 132f.; II, 71£.; Harmatta 1969). This specific coinage has mostly been attributed and dated to the lineage of the Hephthalites by scholars, especially numismatists, who think, with the hypothesis in mind that Jauvla, Jabula, Janbula or Jabuvlah on the Hephthalite coins is identical with Zabul, that the Hephthalites governed over even Zabulistan and Kapisi (de Morgan 1923-1936; Marquart 1901; Ghirshman 1948; Bombaci 1957; GobI 1967; Mitchner 1975; Wink 1990). No written document however supports the assumption that the Hephthalites after gaining power in Tokharistan extended their rule beyond the Hindukush southwards to Kapisi and Zabulistan. Such Chinese sources contemporary with the Hephthalites as the Song Yun's narrative in the Luoyan Qeilanji (Account of Buddhist Monasteries in Luoyang) and the original paragraphs of the existingWeishu (History of the Northern or the Later Wei Dynasty) only inform us of the existence of a Hephthalite king in the Northwest who invaded southeastwards through the regions between the Hindukush and the Karakorum to rule over that area as the Kidara Kushans had also done (Kuwayama 1990:109f.). Coins with a bull-head crown were apparently distinctive to the Kapisi Dynasty founded by Xingnie (Khingal). This Khingal Dynasty in Kapisi should be supposed to be identical with a dynasty of which an account was given in the Chinese written documents on the Kapisi country in the Sui period (581-618). As I demonstrated elsewhere (Kuwayama 1975: 93-107), the country phonetically given the name as Cao in Sui Chinese represents neither Caojuzha nor Caoli, usually assigned to Zabulistan, but Jibin of the Tang period which was first identified with Kapisi by Levi as early as in the last century (Levi and Chavannes 1895; Levi 1896 and 1897) and later by Shiratori (1917) . This bull-head crown of the Kapisi king who was contemporaneous with the Sui Dynasty is actually mentioned in the Account of the Western Regions of the Beishi which was edited by Li Yanshou in 659 as quoted from the Weishu, the Zhoushu and the Suishu. The existing Suishu (Vol. 83), however, tells that the king wears the golden fish-head crown which was later copied in the Cefu Yuangui (Vol. 960), too. The bull-head crown is revived in the Tongdian, edited by Dou You between 766 and 768, even though the rest of the account of the Cao country in Volume 92 of this document is quite the same as that of the existing Suishu. Replacement of the fish-head by the bull-head in the Tongdian and the mention of the bull-head in the Beishi clearly prove that the Suishu used by Li Yanshou and by Du You in their editions in the mid-seventh and the mid-eighth centuries respectively actually recorded a bull-head crown, not a fish-head crown. Therefore the Suishu which had been available before the middle of the eighth century at the latest presumably described the bull-head crown of the Kapisi king in the late sixth and the early seventh centuries. This supposition might also give cause to think that the Suishu as it was first edited in 256 629-630 in the early Tang period only mentioned the bull-head crown. Remusat (1829: 211) rightly referred to the bull-head crown as early as the last century and GobI also called attention to it, but he seems tohave failed to attribute it to the crown on the coins referred to above (GobI 1967: I, 135). The Cao country of the Suishu appears to be one of the earliest mentions of Kapisi as it emerged onto the historical stage after the withdrawal of the Hephthalites' power from both sides of the Hindukush. The Nezak Shah coins bearing the bull-head crown are therefore quite different from the Hephthalite series, belonging, as they seem to do, to the local dynasty inaugurated by Khingal. I think that there is no positive evidence for proving that the Nezak Shah coins with the distinct bull-head crown were issued by some descendants of the Hephthalite kings whose coins, including those of Deva ~ahi Khmgala, apparently differ from the Nezak Shah coins in the styles of king's bust and crown. Therefore Xingnie (Khingal) is different from a personage derived from the Hephthalites. 5. Sri Shahi of U ddiyana According to the new reading of Prof. Nakatani provided in Section 3 above, Sri $ahi Khirpgala bore the epithet Otyana Shahi, i.e., Shahi of Uddiyana, to which Tucci first referred. The hypothesis of Petech, who accepted Sircar and attributed otyata to a local derivative form of a personal name, Udyaditya, now needs revision (Petech 1988, 187-189). The king bearing the name Sri $ahi Khirpgala in the inscription is a definite personality in the eighth century, perhaps one of the Turki Shahis in the Kabul Valley, or the Kabul Shahis. Judging from the styles of the letters used in the inscription, we have proposed above the possibility that the inscription itself is attributable to sometime later than the dates proposed by Tucci and Sircar, i.e., to the seventh or the eighth century and, if the sarpvat eight belonged to one of the aforesaid eras, Sri $ahi Khirpgala could have consecrated the Ganesa image in either 665 or 765, by the Vikrama era, or in either 632/633 or 7321733, by the Laukika era. With this in mind the last, but not least important, discussion follows in support of dating the king and his Ganesa statue to the eighth century. Turning our eyes to Chinese Tang records, we know that Kapisian missions came to the Tang court in the years 619 (Jiu Tangshu, Vol. 198),637 (Jiu Tangshu, Vol. 198; Tongdian, Vol. 192, 8; Cefu Yuangui, Vol. 970), 640 (Cefu Yuangui, Vol. 970), 647 (Cefu Yuangui, Vol. 970), 648 ( Cefu Yuangui, Vol. 970), 651 (Cefu Yuangui, Vol. 970), 653 (Cefu Yuangui, Vol. 970), 654 (Cefu Yuangui, Vol. 970), 658 (Jiu Tangshu, Vol. 198),670 (Cefu Yuangui, Vol. 970), 671 (Cefu Yuangui, Vol. 970), 692 (Cefu Yuangui, Vol. 970), 710 (Cefu Yuangui, Vol. 970), 719 (Jiu Tangshu, Vol. 198) and so on (Kuwayama 1991A). We do not know how long the Kapisian king Hujiezhi (*ghar-ilci) ruled after his accession in 653 and therefore whether he sent such missions as those in the years following 670. Yet the year 719 is apparently the year of the first tribute of the Turkish king of Jibin whom the Emperor of the Tang Dynasty, Kaozong, admitted on this occasion as 257 Qaradachi Tegin, a new Turkish official title signifying one who assumes control over the country. This Turkish title, never found on kings of Jibin before him, leads one to postulate that something politically important happened in the Kapisi-Kabul regions in the years between 658 and 719. Jibin and Xieyue (Zabul in the Muslim sources), on the other hand, sent their embassies together to the Tang court in 710. According to Huichao who travelled through the Kabul Valley in or just before 72 7 (the year of his arrival in Kucha), the kings and their troops of both Jibin (Kapisi-Kabul) and Xieyue (Zabul) were all Turkish and the king of Zabul was politically independent from the king of Jibin, although he was a nephew ofJibin's king. Taking into consideration this relation between Jibin and Zabul, there is the possibility that already at the time of the embassies from Jibin and Zabul in 710 the king of Jibin was Turkish. The event in the background of these accounts was thus most probably the beginning of the Turki Shahi Dynasty, viz., the Turkish usurpation of the local dynasty in the line of Khingal, which, as examined above, happened sometime between 658 and 710 (in 666, according to Rahman 1979: 47). As I demonstrated elsewhere (Kuwayama 1991A), the Turkish usurpation of the local Xingnie dynasty founded at Begram in Kapisi took place in the Kabul region. In the Da Tang Xiyu ji Xuanzang alludes to a country to the south of Kapisi, called Fulishi-sthana. The name may be restorable as ilrdhasthana signifying the "high region". More important is that the king who ruled the country is Turkish. There is no significant indication in any written sources that the West Turks in Tokharistan crossed over the Hindukush through Kapisi to the area south of it. It would therefore be impossible for him to be ethnically relevant to the West Turks in Tokharistan. He probably had been a minor power under the rule of kings of the Khingal line in Kapisi for a long time when Xuanzang came to the south of the Hindukush. According to Huichao, on the other hand, the Turkish king killed the king of Jibin and became chief of the kingdom when the Turkish forces became stronger in that region (Fuchs 1938,445). This statement well fits with Biruni's explanation that a Turkish noble ascended the throne in Kabul (Sachau 1888, II, 10). There is no reason to exclude the possibility that Fulishi-sthana is Kabul and the adjacent regions. The importance of Kabul gradually increased to become greater than ever in parallel with the several waves of Muslim invasions from the south. Actually, after the end of the seventh century Jibin often sent its missions to the Tang court together with Zabul (Xieyue). Associated missions of this kind had never appeared before that time. Such association may be political and military, suggesting that Jibin and Xieyue stood firm against the Muslim invasions. Therefore it is Kabul in the time after the Turkish usurpation that is indicated by Jibin, the Tang Chinese denomination. The Turkish king of Jibin must actually have also ruled over the vast regions extending to the west of the Indus, as the king of Kapi§i at the time of Xuanzang did Apparently twenty years after 719, in 738, a Jibin king with the Turkish title Khorasan Tegin Shad sent an envoy to the Tang court reporting that he was succeeded to his son, Fulinjipo (*F[H]ROM KESARO) had succeeded him because of his old age. The Khorasan Tegin Shad, so aged in 738, must have been the first Turkish ruler and the 258 usurper. He is possibly identified with Barhatakin of Biruni (Sachau 1888, II, 10) and the aye of Huichao (Fuchs 1939,444-445). Volume 198 of the Jiu Tangshu and Volume 99 of the Tang Huiyao tell that in Jibin the king F[H]ROM KESARO was succeeded by his son Bofuzhun in 745 and that in the same year the Tang Emperor accepted Bofuzhun as the king of both Jibin and Uddiyana. As mentioned earlier, Sircar doubted whether the Shahi of Swat really existed and if he occupied the Gardez region within the dominions of the Shahis of Kapisi or Kabul, probably the strongest among the Shahi houses. Tucci suggested that Uddiyana inscribed on the Gardez Ganesa was the first mention of that country in inscriptions. Really Uddiyana hardly appears on the historical stage even in Chinese official history recorded by the Tang court. The only exception is their first tribute in 642 to the Tang court of Emperor Taizong's regime (Cefu Yuangui, Vol. 970). The second and last reference to Uddiyana in Chinese documents is the establishment of the Kabul Shahi's rule of that country in 745 or just before that year. If we take this fact into consideration, it is needless to think that the Shahi ofUddiyana in the inscription was an independent Shahi other than the Turki Shahi in Kabul as Sircar supposed. Weare therefore allowed to take the king in the inscription on Sri Shahi Khirpgala, the Shahi of Uddiyana as either Bofuzhun or one of the Turki Shahis later than Bofuzhun whose actual name is not known. In either case, the possible year of consecration of the marble Mahavinayaka image falls only in 765 of the four candidates mentioned above, because the Turki Shahi Bofuzhun extend his rule to Swat for the first time in 745 at the latest and the stylistic characteristics of the image do not allow us to attribute it to the ninth century. If we take the first case as right, the name Bofuzhun would possibly be a Turkish title since it is not phonetically comparable with Khingala in the inscription, and this Khirpgala-Bofuzhun would have installed the marble Mahavinayaka image as the king of Uddiyana twenty years after his accession. In this case, we are left without any clue to positively identify the Kabul Shah called Khinkhil or Khinjil-whom Ya'qubi mentions as contemporary with Al-Mahdi (775-785)-with Sri Shahi Khirpgala, Shahi of Uddiyana. The question whether Khinkhil is identical Khirpgala-Bofuzhun would simply depend on how long the latter was on the throne. There is, however, no clue for the duration of Bofuzhun's rule. The documents on the Western Regions recorded by.the Tang court refer to two further missions of Kabul-Gandhara in 748 and 753 and eventually end with the arrival at Changan in 758 of three Buddhist monks from Kabul-Gandhara, Middle India and Kasmir. In these documents there is no mention of the name of a king in Kabul-Gandhara. The only other document on Kabul-Gandhara that is left for the period in question is the narrative of Wukong who travelled to Kabul, Gandhara and Kasmir between 753 and 764, but also says nothing of the king's name, although some of the Buddhist precincts in Kapisi and Gandhara that had been visited by Xuanzang more than a century before him were referred to. Lastly, if the year eight did represent the regnal year as Sircar and Petech stated, the starting point would be the year of Bofuzhun's accession to the throne of both Kabul and Swat, i.e., 745. 259 XVI Dating Yasovarman of Kanauj on the Evidence of Huichao The purpose of this article is to fix more exactly the date of Yasovarman of Kanauj. His date has usually been given based on the interpretation of the accounts on Yasovarman in the Gaudavaho ofVakpati and his political relationship with Lalitaditya Muktapida of Kasmir in the Rajatarangini. Yet in the account of Huichao' s pilgrimage in India and Central Asia and other Chinese historical sources of the Tang Dynasty some more informtion about this problem can be expected. Of vital importance is the date given .by Huichao himself of his arrival at Kucha which also affords us an answer as to when he stayed in both Tokharistan and Gandhara. The date of Huichao's stay in Gandhara may, in turn, lead us to the framing of the date of his stay in Kanyakubja. In Kanyakubja, as Huichao tells us, an unnamed victorious "king of Central India" had resided.(1) identifying him with Yasovarman, his placement within a clearer chronological framework can also be deduced from other datable factors in Huichao's account itself and other Tang sources. Before going further, some words should be given about Huichao's account. In Cave 17 of Dunhuang Pelliot found in 1908 a rather short manuscript of 227 lines, now housed and listed in the Bibliotheque Nationale as the Pelliot chinois, Touen-houang No. 3532. Despite the loss of its beginning and closing parts which must have contained the title and the author's name, Pelliot (1908: 511-512; 1936: 275) identified it as a ninth century abridged reproduction of the Huichao Wang Wu Tianzhuoguo zhuan (Record of Huichao's Travels in the Five Indic Regions), recollecting that the title of this record-and only the title-survived as quoted by Huilin in the 100th volume of his Yiquijin Yinji (Dictionary of Phonetics and Meanings of the Specific Words in All the Translated Buddhist Sutras and Literature, completed in 817). In this dictionary Huilin clearly mentions that Huichao's narrative is in three-volumes and explains the pronunciation or meanings, or both, of some eighty-five specific words selected by Huilin himself from each volume. The Pelliot chinois, on the other hand, is not divided into any volumes and lacks most of the words quoted by Huilin. There is no doubt, however, that the description of the manuscript covers almost all of Huichao's itineraries, i. e., from somewhere in East India up to Kucha. Moreover, a couple of words selected by Huilin from the third volume are also found among the last lines of the manuscript. Hence the direct assumption has been made that it was based on the same three-volumed edition as used by Huilin. However, such a simple solution does not seem satisfactory. First, a basic question must be asked: Why does the manuscript lack most of the words quoted by Huilin? The manuscript could have included all of the words quoted by Huilin, if it were an abridged 260 version. Secondly, the manuscript contains a number of misused words along with grammatical irregularities, all of them being too unfamiliar in proper Chinese sentences to be used in so official a report as one registered in the Yiquijin Yinji. In fact, a recent linguistic approach to the words and phrases in the manuscript has shed new light on this problem, rightly concluding that it is not an abridged copy of the three-volumed edition but a rush one reproduced by someone at Dunhuang in the eighth century after the original draft of Huichao and that, based on this draft, the account was later edited into a more official form comprising the three volumes which Huilin might have used on editing the Yiquijin Yinji (Takata 1992: 197-212). The more extensive book has been lost, but this copy of the draft has survivedY) 1. A chronological framework based on Huichao Since the manuscript lacks its beginning part, the first country available in the itinerary of Huichao is Kusinagara. His itinerary as recorded in the manuscript is not always the same as that which he actually took. Although a couple of countries will have to be omitted as inserted from hearsay, as will be discussed later, the itinerary as it actually appears in the manuscript is as follows: An unidentifiable country in India to Kusinagara: 30-day march. Kusinagara to Varanasi: Distance and direction are lost. Varanasi to Kanyakubja: Perhaps 60-day march, to the west. Kanyakubja to a town of the South Indic king's residence ('Il; 90-day march to the south. A town of the South Indic king's residence toa town of the West Indic king's residence ('2): 60-day march to the north. A town of the West Indic king's residence to a North Indic country called J alandhara: 90-march to the north. Jalandhara to Takkadesa: 30 days to the west. Takkadesa to Xintuogula('3): 30-day march to the west. Xintuogula to Kasmira: 15 days to the north. Kasmira to Gandhara (Udabhandapura): 30-day march to the northwest. Gandhara to Laghman: 7-day march to the west. Laghman to Jibin('4): 8-day march to the west. Jibin to Zabul: 7-day march to the west. Zabul to Bamiyan: 7-day march to the north. Bamiyan to a town of the Tokhara king's residence (Warwaliz): 20-day march to the north. Warwaliz to a place in Badakhshan where the Tokhara king had fled: 30-day march to the east. Somewhere in Tokharistan to Wa'khan: 7-day march to the east. Wa'khan to Kharbandan (Tashkurghan): 15-day march to the east. Kharbandan to Kashgar: 30-day march to the northeast. Kashgar to Anxi (Kucha): 30-day march to the east. Kucha to Yanqi (Kara Shahr in later times): ?-day march to the east. *1 - Assignable to Badami. *2 - Either Alor or Brahminabad. *3 - Most probably the same as Sinhapura 261 in the Da Tang Xiyu ji. *4 - Kabul. See discussions in this chapter. Places in italics were inserted in the itinerary by hearsay. See Chapter 3. On line 217 of the manuscript Huichao says that he has arrived at Kucha during the first ten days of the 11 th month in the 15th year of the Kaiyuan era of the Tang Dynasty. Since the 14th day of the 11 th month in that year is identical with the 31st of December, 727, his arrival was in between the 18th and the 27th of December, 727. This date is basic for fixing other dates. In Tokharistan he had snowfall which inspired him to write the poem seen on line 196. Snowfall naturally means that he was there in the winter. This winter must be prior to December, 727 and is undoubtedly at least one year earlier than the time when Huichao arrived at Kucha. Earlier in his account, lines 155 to 156, Huichao says that going 20 days to the north after leaving Bamiyan he reached a town where the Tokhara king resided. This was called by him fudiye, a leading town in Tokharistan, identical with Warwaliz referred to by the Arab geographers and assignable to somewhere around Qal'a-ye Zal in the area of the lower Kunduz River (Kuwayama 1989: 120ff.).(3) Based on my experience in the Kunduz-Khanabad valleys in the latter half of the sixties, snowfalls seldom occur in December; therefore, Huichao may have been in Tokharistan in the beginning of 72 7 at the earliest, if he stayed in Warwaliz during the snowfall. In fact, Huichao does not refer to the specific place in T okharistan where he had the snowfall: it may not have beenfudiye. Lines 155 to 157 of the manuscript mention that by the time of Huichao's visit the Tokhara king had fled to Badakhshan, a 30-day march to the east of the capital of Tokharistan (Warwaliz), since the Arab troops had occupied it. Although the manuscript itself does not give any account as to how Huichao reached Badakhshan, a comparison of the 20-day march between Bamiyan to the Tokhara capital, shown in the manuscript, with the 30-day march from there to Badakhshan suggests that Huichao first arrived at Warwaliz and then proceeded to somewhere in Badakhshan. He also describes a 7-day march to Wa'khan from the Tokhara country-quite a short distance in comparison with the 30-day march between Warwaliz and Badakhshan. The short distance implies that the Tokhara country from where he proceeded to Wa'khan does not mean Warwaliz. The starting place for Wa'khan must be somewhere deep in the Badakhshan area, a place rather close to the western gateway to Wakh'an, modern Ishkashim. In such highlands snowfall could occur very much earlier than in the lower valleys of the Kunduz. If so, the date of his stay in Tokharistan could have been even as early as the autumn of 726. The date thus provided also frames that of his stay in Gandhara since we have good reason to suggest that Huichao was in Gandhara during winter also. Before reaching this hypothesis, Huichao's description of Gandhara and Jibin needs to be considered. On reaching Gandhara after a one-month march through the mountainous region from Kasmira, Huichao wrote his account in the section of Gandhara (lines 140-142 in particular) as follows: The king and his troops are all Turkish, while the local people are Hu barbarians.(4) This country (Gandhara) had originally been ruled by a king of 262 Jibin (Kapisi) under whom the father of the Turkish king (in the time of Huichao) was subject with his tribesmen and his troops; when the Turkish forces became more active afterwards, he killed the king of Jibin and became the chief of the kingdom; accordingly, this kingdom abutted on the frontier with the supreme king of the Turks [of the northern land], ... but this usurper was not in the same line as the northern Turks .... This town overlooks the big river of the Indus, located on its northern bank. ([ ] represents missing characters in the text and ( ) should be read as supplementing the meanings.) Huichao further describes Laghman, a seven-day march to the west of Gandhara, as being ruled by the Gandhara country (or the Gandharan king) without its own king. As for Jibin, an eight-day march to the west of Laghman, i.e., in the Kabul Valley, Huichao describes it as follows: This country is also ruled by the Gandharan king, who is accustomed to be in Jibin in summer, seeking a cool place to live, and in Gandhara in winter, seeking a warm place to live, since there are no snowfalls in Gandhara, it being warm and not cold, while the Jibin country has much snowfall in winter, wherefore it is cold; the local people of this country Gibin) are the Hu and the king and his troops are Turkish. The countries extending from Gandhara in the east to Jibin in the west in the Kabul Valley were controlled by the "Gandharan" king of Turkish stock. This statement gives the impression that there existed a king of Gandhara whose rule extended to the Jibin region where there was another king. Yet this was not true. The historical reality is rather that the Kabul Valley and Gandhara were ruled by one and the same king who resided in Jibin in summer and in Gandhara, or more properly at Udabhandapura on the north bank of the Indus, in winter. Huichao's account that the "Gandharan king" ruled from there to Jibin seems to suggest that Huichao happened to arrive in Gandhara right at the season when the king of the Kabul Valley was accustomed to reside there. In this connection the Turkish seizure of the Jibin kingship recorded by Huichao in the section of Gandhara also suggests that the event may not have happened in Gandhara itself but elsewhere in the king's extensive domains. Some thirty years later, Wukong, the vice-ambassador of the Tang Mission to the king ofJibin, arrived at Udabhandapura and clearly explained that this town of Gandhara was the eastern capital of Jibin. The "eastern capital" presupposes the existence of a "western capital". For Wukong, the Jibin country seems to have denoted a more extensive country in geographical and political terms, and a kingdom including the extensive Kabul Valley and having western and eastern capitals, the latter in Gandhara. In view of the fact that Huichao uses Jibin and Gandhara as two different localities, the whereabouts of Jibin needs to be further examined. The following argument reinforces doubts about S. Levi's identification of Jibin, so far as it is used in Tang China, exclusively with Kapisi which has long held the field as undebatable. 263 Huichao describes that when the Turkish forces became stronger the Turkish chieftain killed the king of Jibin and became chief of the kingdom. If the event did not happen in Gandhara but somewhere in the Kabul Valley, it would not be historically inconsistent with Biruni's legendary story about a Turkish nobleman who eventually usurped the throne and began reigning in Kabul. Sachau translated it as follows (1964, II: 13-18): The Hindus had kings residing in Kabul, Turks who were said to be of Tibetan origin; the first of them, Barhatakin (Barha Tegin, or Bori Tegin), came into the country and entered a cave in Kabul, which none couid enter except by creeping on hands and knees .... Some days after he had entered the cave, he began to creep out of it in the presence of the people, who looked on him as a newborn baby; he wore Turkish dress, a short tunic open in front, a high hat, boots and arms; now people honoured him as a being of miraculous origin, who had been destined to be king, and in fact he brought those countries under his sway and ruled them under the title of a shahiya of Kabul.. .. Apart from doubts about the strange mention of the Turks in Kabul as being of Tibetan origin, the rise of a shahiya of Kabul and his usurpation of power there strongly support the hypothesis that the Turkish seizure of the throne described by Huichao happened in Kabul, not in Gandhara. It also suggests that Barha Tegin is one and the same with the father of the Turkish king at the time of Huichao. In fact, on Xuanzang's way to and back from India, in 629 and 643, a Turkish ruler already resided in a country called Fulishi-sthana located between Kapisi and Ghazni, as recorded in Vol. 12 of the Da Tang Xiyu ji. Insofar as it simply records that the king of the Fulishi-sthna is of Turkish stock, the rule of the king seems to have been geographically and politically restricted as a local minor power, perhaps under the Kapisian hegemony (Kuwayama 1991: 282-283). Consistency with the above account of Biruni implies that the Fulishi country of Xuanzang may be equivalent to the Kabul region. The importance of the Kabul region may have gradually increased to become greater than ever parallel with the several waves of Muslim invasions from the south from 665 onward. On his way to Nalanda and back from there to China in 643 Xuanzang found that the Kabul region, the Fulishi country, was more or less peaceful ruled by a Turkish chieftain under the hegemony of the Kapisian king, but about two decades later the Turks of Kabul must have no longer been able to ignore the threat of the Arabs. In AH 46-47 (665-666) the Kabul Shah, the Turkish king of Kabul, was forced to fight against the Arab troops ofIbn Samurah (Murgotten 1969: 146-147). In the earlier stages of fighting against the Arabs the Kapisian king might have increased the Kabul Turks' military capacity to maintain the region, the southern frontier of the Kapisi kingdom. Yet, as events progressed, the voice of the Turkish minority probably increased, and gradually they themselves turned into a powerful threat, eventually seizing the Kapisian throne. As the Tang sources mention, after the end of the seventh century Jibin often sent 264 its missions to the Tang court together with Zabul. Associated missions of this kind had never appeared before that time, which suggests that Kabul and Zabul stood firm together against the Arab invaders. Jibin as used in Tang China, therefore, means Kabul after the Turkish usurpation that happened most probably soon after 666. Needless to say, the new king of Jibin, Barha Tegin, must have also ruled over the vast regions from Kabul to Udabhandapura where the second Turkish king, a son of Barha Tegin, resided on Huichao's arrival in winter. The above argument leads us to interpret Huichao as follows: the Turkish king of Gandhara was a ruler who governed from Gandhara to Kabul and was accustomed to stay at Udabhandapura in the winter season and at Kabul in the summer. Huichao called him the king of Gandhara because he happened to be residing in Gnadhara when Huichao visited there. This means that it was in winter that Huichao travelled through Gandhara. The king at the time must have been the son of Barha Tegin. Allowing for the date of his stay in Tokharistan, Huichao must have been in Gandhara some time between the late autumn of 725 and the early spring of 726, not later than that. Since Huichao came to Gandhara from the east beyond the Indus, his pilgrimage in the subcontinent was made before the date given above. This chronological framework allows us to discuss who was the king of Central India at Kanauj at the time of Huichao' s visit and to narrow down the date of this sovereign. 2. A king of Central India: Yasovarman Lines 10 to 48 of Huichao's account is his record on Central India, an original translation of which follows: Varanasi is reached within [... ] days. This country, (like Kusinagara,) is in decline without a king .... (Here were) the Five Bhikshus (to whom the Buddha first preached). The clay images (of that scene) are present in a stupa. (There is a stanbha) surmounted by a figure of a lion. The stone stanbha is as thick as to be able to hold five people together and very fine in quality.... This stanbha was made at the same time as the stupa. The name of the temple is Dharmacakra .... The heretics do not wear clothes but smear ashes on their bodies to serve the Mahadeva. In this temple there are an image of gilt copper and (images of) the five hundred (arhats). In old days in Magadha there was a king called Siladitya. He made these images. He made a (wheel) of gilt copper at the same time. The dharmacakra does not lack any part measuring exactly thirty bu (about 47 m) in circumference. This town (Varanasi) is on the northern bank of the Ganga so as to look down on the river. In the territory of the king of Magadha are the Four Great Sacred Stupas at the Deer Park here (Sarnath), Kusinagara, Rajagriha and Mahabodhi. In this country (Varanasi) both Mahayanic and Hinayanic teachings are being held. Thus could I reach the Mahabodhi Temple .... Within a two-month march from here, Varanasi, 265 westward one reaches a town of residence of the king of Zhong Tianzhu (Central India), called Kanyakubja. The territory of this Central Indic king is very vast, well-populated and prosperous. The king has nine hundred elephants, each of the other great chieftains (in this country) having two or three hundred. The king customarily fights a battle at the head of his army. He often makes war against the other four regions of India and always wins .... It is in this territory of Central India that there are the Four Great Stupas Three of them are on the northern bank of the Ganga. The first is in the Anatapindada Park of Sravasti where the monastery, housing monks, is still functioning. The second is in the Amula Park of Vais ali where the stupa is still standing but the temple is desolate without any monks. The third is in the town of Kapilavastu where the Buddha was given birth. The Tree is still existing, but the town has long declined. The stupa is still standing but there are no monks. There are also no inhabitants in the town. This town is located the most northerly of the above three and is left to run waste with bushes and woods, the road being infested with many footpads. Therefore those wishing to go there for worshipping have much suffering. The fourth is the stupa at the Holy Step of Three Courses (at Sankasya) located between the two Gangas a seven-day march to the west of the town of the Central Indic king .... There is still a temple with monks. ([ ] represents missing characters in the text and ( ) should be read as supplementing the meanings.) According to these paragraphs on Central India the king resides in Kanyakubja (Kanauj) ruling a very extensive territory and possessing a force of elephants much stronger than the other four regions; he himself always fights at the head of his army and gains victories over the kings of the others. Lines 41 and following show that his domain includes the locations of the 'Four Great Stupas' at Kapilavastu, Sravasti, Sankasya and Vaisali. Unfortunately Huichao does not refer to the name of this great victorious king. On the other hand, he clearly mentions Siladitya as a king of Magadha. Since Central India, or the Uttarapatha, usually includes Magadha, the question has arisen whether the Central Indic king is identical to Siladitya. Stein (1900: I, 89) first identified the nameless king of Central India wid1 Yasovarman which is phonetically equivalent to Yishafamo in Vol. 221a of the Tang shu. Yishafamo appears there as a king of Central India having sent a Buddhist monk of great virtue to the Tang court which received him in 731. Since Yishafamo is definitely the same as Yasovarman, Yasovarman was in the position in or before 731. Otani (1934: 154-156) agrees with Stein, but, strangely enough, prefers Siladitya as the better alternative taking into consideration that the king of Central India might not be absolutely contemporary with Huichao. Yet the following discussion proves that Otani's interpretation conflicts with the description by Huichao himself who clearly writes that the king of Central India is comporary with Huichao. Huichao shows that the Four Great Sacred Stupas at Sarnath, Kusinagara, Rajagriha and Mahabodhi are in the territory of Magadha, the old domain of Siladitya, and the 266 other Four Great Stupas at Kapilavastu, Sravasti, Sankasya and Vaisali are in Central India under the rule of the Central Indic king. The very same eight stupas were also noted by Wukong about three decades later without dividing them into the two categories. The four main events in the life of the Buddha were the Birth at Kapilavastu, the Enlightenment at Bodhgaya, the First Sermon at Sarnath and the Nirvana at Kusinagara, as Faxian rightly refers to the Four Great Stupas at these four places. But Huichao's Four Great Sacred Stupas excludes the stupa at Kapilavastu, which instead is included in the other category. Therefore, his two categories have no practical significance in the Buddhist terms, and should be taken as aimed to make clear the distinction between the ancient Magadha territory of Siladitya and contemporary Central India. For Huichao the king of Central India has nothing to do with Siladitya. It is the fact that the death of Har~avardhana Siladitya in the late forties of the 7th century caused his domain to split into minor local states and that no imperial unity had been achieved in the northern parts of India until Yasovarman came to power. The success ofYasovarman in extending his domain to the Gau<;la country is well-documented in the GaiicJavaho written by Vakpati and partly documented in the Rajatarangini and in the] ain sources such as the Prabhavaka-carita, the Prabandha-kofa and the Bappabhattasuricarita. Only Yasovarman can be taken as the proper candidate for the Central Indic king whose extensive territory and constant victories are quite clearly recorded Huichao. 3. Dating Yasovarman The date of Yasovarman has variously been given by scholars: 72 8-745 by Smith (1908: 784), some time between the second half of the seventh century and the first half of the eighth century by Pandit (1927: xcv-xcvi), 725-754 by Tripathi (1937: 195-197) and 700 (or 690-740) by Majumdar (1954: 131). Since Huichao arrived at Kucha in late 727, Smith's hypothesis does not hold well. Tripathi's date of accession, 725, seems to be based on the fact recorded in the Rajatarangini that it occured more or less contemporaneously with that of Lalitaditya Muktapida for whom he gives a date from 724 to 760, as also proposed by Utgikar (Pandit 1927: cclvi-cclx). The date of accession, 72 5, given by Tripathi does not seem appropriate. According to Huichao, the Central Indic king had already expanded his domains when Huichao was in Kanyakubja. For Huichao it took three months from Kanyakubja to a town of king's residence in South India, which might fit Badami of the Chalukyas; two months from there to a town of king's residence in West India, which might be either Alor or Brahminabad; three months from the town of West India to] alandhara; one month from there to Takkadesa; also one month from TakkadeSa to Xintougula in North India and half a month from Xintougula to Kashmir. From Kashmir he reached Udabhandapura after another month. In total it took eleven months and fifteen days, nearly one year, from Kanyakubja to Udabhandapura. If the durations of stay at each place are added together, the time he spent from Kanauj to Gandhara much exceeds one year. Based on the fact that Huichao was in Gandhara from the autumn of 72 5 at the earliest, he thus 267 would have stayed in Kanyakubja no later than 724. Tripathi's hypothesis seems to lose validity. The above chronology might be valid insofar as we literally accept Huichao's itinerary. A series of doubt arises, however, from the conspicuous difference between the description of the route shown above and that from J alandhara through Gandhara and Jibin to Tokharistan and farther. In the latter case the actual route he took can clearly be traced, since he states each city on the route one by one in due order, and it is quite convincing despite the existence of some countries known only by hearsay and the unconvincing locality of Xintuogula between Takkadesa and Kasmira. Yet in the case of the routes to South India and from West India and those in South and West India, he does not specify anything about the route and the main city in each part, only saying that after a certain number of months one reaches a town where the king of South (or West) India resides. Nothing of the route he might have taken is referred to. On the other hand, he clearly specifies the names of cities in the case of North and Central India where he must have actually visited, stating that going northward three months or so from a town where the West Indian king lived he reached a North Indian country called Jalandhara (Line 65) and also that he reached a town of Central India's king, called Kanyakubja (Line 21). Also the route he took on the way from J alandhara to Tokharistan in particular is traceable without a doubt. All these differences in the descriptions between South and West India, on the one hand, and North and Central India, on the other, may be taken as clues which show that he did not actually visit South and West India, even though he recorded a poem in the section of South India, on lines 57-58, where he is quite moved by coming so far from his homeland. His poem, however, does not include any words to specify the fact that he actually reached South India. I do not think it is going too far to say that Huichao's description about South and West India simply is hearsay. Without seeing South and West India he went from Kanyakubja northwestward to J alandhara. If so, the time taken from Kanyakubja through South and West India to J alandhara, i. e., eight months, should be omitted from the total eleven months and a fortnight which was taken from Kanyakubja to Gandhara. Three months and fortnight are the time which Huichao spent from Jalandhara to Udabhandapura via Takkadesa, Xintuogula and Kasmira. Although we are left uninformed about how long it actually took from Kanyakubja to J alandhara, he was in Kanyakubja at least three and a half months earlier than the late autumnal months in 725. Even if the durations of stay in each town along the route are taken into account, Huichao's stay in Kanyakubja possibly was not in 724 but in the earlier half of 725. Until the earlier half of that year, therefore, Yasovarman may have been finalizing his conquest to widen his political territory. According to the Cefu yuangui Indian missions, seemingly a joint venture of all Five Indic Regions, arrived at the Tang capital in the 3rd month of Year 3 in the Tianshou era (692). The Tang shu (Vol. 221a) describes that these missions were received in 69l. Despite no clue as to whether the missions came once or twice, it is believable that both sources refer to the same event. The names of the kings responsible for the missions in each source are common except for some negligible differences in the Chinese characters. 268 The list of kings is as follows: (1) Shiluoyiduo, the king of West India, is restorable to Siladitya, identifiable with either Siladitya IV of Valabhi or Yuvaraja Sryasraya Siladitya, one of the viceroyalty ofVinayaditya of No. 2. (2) Zhelouqi-faluopo (Tang shu), the king of South India, is restorable to CalukyaVallabha, clearly identifiable with Vinayaditya (681-696) of the Chalukyas of Badami, known as the virudas Sri-prithivi-vallabha. The Cefu yuangui lacks the last character. (3) Louqi-nuonuo (Tang shu), the king of North India, is restorable to Loke~ Nana or Loke~ Nanna, but unidentifiable. The Cefu yuangui lacks the first two characters. (4) Moluofamo (Cefu yuangui) , the king of East India, is restorable to Maravarma or Maravarman, but unidentifiable. The Tang shu writes the third character as 'zhi' 11 instead of 'fa' ft. (5) Dipoxinuo, the king of Central India, is restorable to Devasena. The date of the reception of the Indian missions and the regnal years ofVinayaditya lead us to believe that these kings in the South Asian Subcontinent are all contemporary. Up to now Devasena has not been known in any other sources in the context of the history of 'Central India' or geographical northern India after the death of Harsha. Can Dipoxinuo (Devasena) be an epithet of Yishafamo (Yasovarman)? If Dipoxinuo and Yishafamo had been one and the same king, the Tang sources would have selected only one name from the two. Since I believe that Devasena can possibly be taken as distinct from Yasovarman, the existence in 691 or 692 at the latest of Devasena as the Central Indic king is important for chronologically framing Yasovarman. Evidently Yasovarman came to the throne after 691 or 692 and was still on the throne in 731, the date of the Tang's reception of the mission sent under his own name. If the itinerary of Huichao discussed above is right, Yasovarman's accession is datable to some time after 691 or 692 and before 72 5. As to the Indian missions received by the Tang court after 69112, the Cefu yuangui (Vol. 971) and theJiu Tang shu (Vol. 198) describe that the emperor Xuanzong received several missions from Central India during the Kaiyuan era, such as those in the 5th month of Year 5 (717), the 1st month of Year 8 (720), the 7th month of Year 13 (725) and the 10th month of Year 19 (731). The last one is attested in the aforesaid Tang shu as the mission undoubtedly sent by Yasovarman. Therefore all these missions fall in the above framework between 692 and 72 5. Considering the intervals of each mission, the twenty-six year interval between Devasena's mission and the one in 717 is far longer than the others, such as three, five and six years that seem to represent a series of missions sent by one and the same king. It may be allowed to think, therefore, that 269 Yasovarman came to the throne in some year between 69112 and 717, or more prop~rly, in a year quite close to 717, already gained victory over the vast regions of not only Central India but also Gauda by 72 5, and still was in the position six years later in 731. 270 NOTES FOR CHAPTERS Chapter III 1. The Weishu, ch. 102 (Western Regions), says that the king of Da Yuezhi called Jiduoluo (*Kidara), brave and fierce, eventually dispatched his troops southward and invaded North India (modern Northwestern India), crossing the great mountains, to subjugate the five kingdoms which were located to the north of Gandhara. The five kingdoms are not mentioned in the above dynastic history. Nevertheless, in other passages which are believed to have been quoted from the Huisheng Xingzhuan (Account of Huisheng's Travels) one can find out the five kingdoms, such as Yarkand, Tashkurghan on the Pamir, Wa'khan, Chitral and Gandhara, that were under the rule of the Hephthalites. 2. The term Middle India or Madhyadesa in this paper is an English adaptation of Chinese Zhong Tianzhu, one of the Five Districts of India, which indicates modern Bihar and environs. 3. The passages quoted in this paper are original translations from classical Chinese unless any reference is made. 4. The Chu Sanzangji ji and the Gaoseng zhuan do not use the term more directly indicating Gandhara such as Gantuoluo, Gantuowei, Jiantuowei and Jiantuoyue. The Jibin in the Ayuwang zhuan (the Life of King Asoka) translated by An Faqin corresponds to Kasmirapura in the original Divyavadana. Paramartha also translated Kasmira to Jibin in the Apidamo Jusheshi fun (The Commentary of the Abhidharmakosha-fastra), and Xuanzang did so in the Chinese version of the Abhidharmakosha-sastra. The Chinese term Jibin therefore seems to have been considered as phonetically equivalent to Kasmira in Sanskrit at the time of translating Buddhist scriptures and to have become fixed as a specific term in something like 'translation manual'. On the other hand, the Jibin had already been used in the Hanshu, the most authoritative dynastic history, as denoting somewhere south of the Pamir, but no one could understand at the time of translating the scriptures where Jibin in the Hanshu actually was, except that Jibin was south of the Pamir. As a result, Jibin was to be confused. In the narratives of pilgrimage which were compiled by the pilgrims themselves or by ones who could use the original narratives, Jibin was rightly used as indicating Kasmira, as in the case of the Luoyang Qielan ji and the Faxian zhuan (Biography of Faxian). 5. The so-called Kanishka stupa near Purushapura is variously described in Chinese sources, with very precise measurements that however vary among the sources. See S. Kuwayama (1997) The Main Stiipa of Shiih-jf-ki pheri, Kyoto. 6. The date of translation is doubtful. It is given only in the Lidai Sanbao ji (The Dynastic Records of the Three Treasures) which was edited in 597, while the other catalogues of the Chinese scriptures edited in 594 and 606 respectively do not mention it. One of the catalogues even excludes this. 7. Attention should be drawn to the fact that Mihirakula as a destroyer of Buddhism in Jibin (Gandhara) appears exclusively in the Buddhist literature written in China. Who can be sure that the king whom Song Yun met was Mihirakula? It is highly possible that Chinese 271 Buddhists gave Mihirakula a leading character of destroying Gandharan Buddhism in order to stress on the fear of extinction of the True Law. 8. The Hephthalites occupied the Northwest, as the nomads like the Kushans had also done. I think it to be a fact that a second or third king claimed the territory extending to the Salt Range in the south and to the region around Jherum in the southeast. His claim brought him into conflict with a king of Kasmira who also wanted to be commercially rather than politically influential in the same regions. So the Luoyang Qielan ji tells that the king, relying on his bravery and force, has contested the territory with Kasmira Gibin in the original text), and the two were locked in battle for three years. The whereabouts of the military camp of the Hephthalite king should have been discussed. No one has ever imagined that the king resided outside Gandhara. Song Yun met him in his military camp which was located eight days' journey to the east of the Indus! The Luoyan Qielan ji says that, after five days' journey to the west of a monastery where Song Yun was sent by the king to lodge, he arrived at the place where the Tathagata cut and gave his head to an old Brahman, to which Faxian also refers with the name of Takshasila saying that Bodhisattva gave his head to a man. Again travelling westward from Takshasila for three days, Song Yun arrived at a great river called Xintou, or the Indus. Travelling again westward from the Indus for three days, he arrived at the town Boshafu, or modern Shahbaz Garhi, which is also identical with Jiantuowei, Gandavati, that Faxian describes. Both pilgrims in the fifth and sixth centuries took six days in all from Taxila to Shahbaz Garhi. This fact implies that the number of days in their itineraries does not mean how long they personally spent between the towns, but indicate the absolute distance that was officially admitted in their days in the Northwest. Comparing the distance between Taxila and Shahbaz Garhi with that between Taxila and the military camp of the Hephthalite king, I presume that the military camp was located somewhere around Jhelum, the place which was strategically important as a gateway to India situated between the eastern limit of the Salt Range and the southern foot of the Pir Panjal Range, through which the Kasmiris had got salt from Punjab as stated by Kshemendra (Samayamatrika 2.90). Chapter IV 1. The original draft of this article was read at the conference entitled "The Crossroads of Asia: Transformation in Image and Symbol", held by the Ancient India and Iran Trust between 5th and 7th October 1992 at Madingley Hall, Cambridge. I am greatly indebted to Dr Elizabeth Errington and Thomas Fitzsimmons for kindly checking the manuscript and to Professor Yasuke Ikari for extending his profound knowledge about the agnicayana rituals. Valuable suggestions were also given by Dr David W. MacDowall and Professor Maurizio Taddei in relation to the appearance of this type of structure in the South Asian Subcontinent. It is Dr Anna-Maria Quagliotti that gave up much of her valuable time at the mausolea in and around Rome. To all of these my sincere thanks are due. 2. Mitra 1971: 151. Mitra found another stupa with the spoke-and-hub pattern at Ratnagiri which dates from the fifth century AD at the earliest, more probably from the eighth century AD (Mitra 1981: 33ff.). The entire structure of the stupa was founded on an earlier structure with a square plan. As the square plan for stupas supposedly came to be used much later in Orissa than in other places, the eighth century date for the spokes proposed by Mitra is reasonable. I do not 272 know why the spoke-and-hub pattern was used at Ratnagiri at such a late period. The Ratnagiri example is beyond the scope of the present discussion primarily because of its late date, but the description of the spokes is included here for reference purpose. According to Mitra, the entire superstructure of the stupa (ibid., fig. 4) has disappeared. The surviving massive base or platform of solid brickwork and three courses of the radial walls around a solid hub above the platform belong to Period I of the stupa. The platform is bereft of a large part of its facing, which has a maximum extant height of 2.743 m. The core of the platform rises a further 1. 5 m in height, at which level it gives place to twelve spokes of the drum. These spokes of solid brickwork, one of which was missing, (ibid., pI. XIIIA), radiate from a central circular hub, also of solid brickwork, 3.29 m in diameter. The maximum number of extant brick courses in the spokes is only three, with a height of 23 cm; the available length of the best-preserved of the spokes is 1. 525 m. In all cases, their extremities have disappeared with the collapse of the upper edge of the platform. 3. Gupta 1985: 50. The latest report on the Sanghol excavations also very briefly describes the discovery of a third stupa (S-3) located on the south-east side of the monastery at SGL-11, the same area with stupa no. 2. It was found deleted down to the ground level. The photograph included with the report reveals that the stupa has at least two ring walls intersected by spokes radiating from a solid hub structure. The device of a circle within a square common to Stupas nos 1 and 2 does not seem to have existed with this stupa. See IAR 1989-90: 91,93, Pl. xxvii, b. 4. I am not able to identify this stupa with the one on the plan. 5. The discussion made here is only tentative. After the conference in 1992 I have again tried to make clear many of questions raised by the excavators' ambiguous descriptions. The excavation reports comprise no more than three sheets of plans, several photographic illustrations and 44 pages of descriptions, many of these materilas containing unclear explanations and often lacking details we really need. From them one can only realize that the site produced only several pieces of schist sculpture which are normally recovered in great number at Gandhara Buddhist sites, and also that baked bricks were familiar material for the builders of the Shah-ji-ki Dheri stupa and other buildings. The use of baked bricks as a building material in a Buddhist temple of Gandhara is highly unusual. Plentiful use of bricks may suggest a fairly late date for the site-but not because of the geographically difficult access to stone as the excavators supposed. Yet the discovery of the 'Kanishka Casket,' even if it was produced later than that monarch, resists assigning the entire site to a late date. The coexistence of such earlier and later factors does not support a single building activity. If so, how many rebuildings did the site undergo? Where should the earlier stupa be found? What is the later phase? To which phase of rebuildings is the 'Kanishka Casket' attributable? If those questions go unanswered, the casket cannot correctly be interpreted in the history either of the site itself or of Gandharan Buddhist monasteries. With these questions in mind, a reevaluation of the excavation reports has led in the direction of the following understanding: in Period I the main stupa was first built in the time contemporary to Kujula Kadphises as a circular stupa with a wheel-shaped pattern inside; In Period II it was then renovated in the time of Kanishka I as a square one with the casket deposited off center against the damaged radial walls of the previous structure; Eventually in Period III it was transformed into a cross-form with the projections annexed to the repaired square stupa sometime in between the 560s and the 630s. About my first understanding of "Kanishka Stupa", see Chapter V and Kuwayama 1997. 273 6 According to Erdosy Sirkap had been used by the time of Huvishka and abandoned by Vasudeva. See Erdosy 1990: 669 f. 7. Although built of squared kanjur, the base moulding of stupa no. 1 in Court A is not a simple, flatter type of torus-and-scotia which is chracteristic of the earliest stupas in Taxila but a later type common to monuments with the semi-ashlar masonry. 8. An excaption is Stupa No.2 of Peddaganjam. The diameter measures 11. 65 m. 9. As to the tumuli located in and around Rome, see Eisner 1986. 10. Dr. David MacDowall suggested a possibility that the Indians might have directly learnt about the wheel-shaped structure actually observing Augustus' mausoleum under construction, pointing out the fact that the first Indian embassy reached Rome in 26 BC only two years after the work started (personal communication at the conference). In fact, Augustus himself says in the Res Gestae that 'Embassies from kings in India were frequently sent to him; never before had they been seen with any Roman commander'. See Brunt and Moore 1967: 34, 35. According to Warmington (1974: 35-8), Strabo, Dio Cassius, Horace, and others classical writers describe the Indian embassies received by Augustus in the years 26 (or 25),21,17, and 14-11 Be. Chapter VIII 1. They are generally believed to be of Central Asian origins, such as from An (Parthia or Bukhara) and Kang (Samarkand) for the reason that their surnames may indicate their homelands. These names however do not always mean that Buddhism had spread to such parts of Central Asia in such earlier days. In the Chu Sanzangji ji, Vol. 2., Kang Senghui is mentioned as an Indian sramana. 2. The Gaoli text shows that Huijing died in the temple of the alms bowl. He is otherwise described to have passed away when they crossed the Spinghar Range to the south of modern Jalalabad. Therefore, I depend upon the texts of the Song, the Yuan and the Ming editions where Huijing is rightly replaced to Huiying. 3. However, the same biography says that Five Hundred Arhats in Jibin usually come and go to the Anavatapta pond. Xuanzang also recorded the legends of Five Hundred Arhats in Kasmir. This implies that Jibin in this context is clearly Kasmir. However it may be, Zhimeng worshipped in prayer the hairs, the teeth, the ushnisha bone and the clear shadow of the Buddha in Kaweiluowei (Kapilavastu). There is no evidence for the existence of such relics in Kapilavastu in any documents. The objects of worship such as these attracted no less Chinese to Nagarahara, among whom even Faxian and Song Yun as well as Xuanzang are not exceptional. Hence I surmise that Huijiao, the editor of the biography of Zhimeng, was quite careless of precise locations of North Indian kingdoms, or he had little knowledge about Indian geography in those days. Same would have been the case for the Jibin problems. 4. This may have been edited in China. The catalogue of the translated s in China, for example the Chu Sanzangji ji, book 2, does not give any date of translation and omits the first two letters, Fushuo, which mean 'the Buddha preached'. See T 55: lIe. 5. The text of the inscription written by Huiyuan himself in 413 is reproduced in the Guang Hongmingji, book 15. In this, Huiyuan says that the learned about the shadow from the 274 priests who had come from the Western regions and he also met a monk from Jibin and a sramana who had learned the vinaya in the South. The former may be Sanghadeva of the Jibin origin who entered into Lushan in 386 and the latter is certainly Faxian. Therefore, Sanghadeva actually saw the shadow of the Buddha in Nagarahara, and presumably it is highly possible that all monks related in any case to Jibin had well acquainted themselves with the shadow, that is, they all went to see it to Nagarahara. 6. Chavannes (1903: 406, fn. 3) identified Shemi with Chitral, but Stein (1928: 10f.) criticised Chavannes saying that the very existence of Buddhist remains in Chitral is opposed with the description that there are no followers of Buddhism in Shemi. Stein tried to put Shemi in the upper Kunar valley. 7. Chavannes,1903: 427, fn. p. 10: "tout Ie paragraphe qui suit concerne Ie royaume de Nagarahara et me parait tire de la relation de Tao-yao. Rien ne prouve que Song Yun et ses compagnons aient porte leurs pas aussi loin". 8. About two centuries later than Song Yun, Huichao described Bamiyan to be independent from any other countries which, therefore, never invaded it. "Der Konig dort ist ein Hu und untersteht keinem anderen Land. Da seine Truppen stark und zahlreich sind, wagen die anderen Lander alle nicht, ihn anzugreifen" (Fuchs 1939: 25). 9. Only the Gao-Ii text says he arrived in the first year of the Da-tong era of Western Wei. In other texts the Datong is replaced by a letter meaning' succeeding' or ' later', which I prefer to the Datong. 10. According to the Da Tang Xiyu ji, book 11, Valabhi is mentioned as the Northern Lala and Malwa as the Southern Lala. Yijin also mentions Lala as a country where Xuanzhao in the middle of the seventh century went in search of a magical medicine by the Imperial order. Based on the account that the capital city of Malwa is defended by the Mahi river on the Northwest, it can be identified with modern Baroda, and Valabhi is rightly the Western Lala and Malwa is the Eastern Lala. The territory of the whole Lala covers an area extending from the Gulf of Cambay. 11. Since Xuanzang points out that an Earlier King's temple is on the south bank of the large river (the Panjshir), that of Dharmagupta may be the same with the above temple. The only difference between the two sources is a letter meaning 'earlier'. If the biographer of Dharmagupta is right, the 'earlier' may suggest that there happened the dethronement before Xuanzang visited to Kapisi. 12. Damoxi-binduo is equal to Damoxi-tiedi by Xuanzang (Da Tang Xiyuji, book 12 and Da Tang Daciensi Sanzang-Jashi zhuan, book 5). Damoxi-tiedi is identical with (Skt.) Dharmasthiti. Marquart 1901: 225. 13. "When the Later Wei Emperor Taiwu was on the royal throne, he sent Dongyan to the Western Regions, and he returned and informed .... " (Tongdian 191) 14. If the itinerary of Song Yun in Gandhara and its environs is right as mentioned in the existing text, the military camp of the Hephthalite tegin was located five days' march to the east of Taxila, or eight days' march to the east of the Indus. He never resided in the Peshawar basin proper. In addition, Yamada sharply criticised the view that the Hephthalite tegin seen in Song Yun's narrative is identical with Mihirakula in various sources, Indian or Chinese. See Yamada 1963. 275 15. The Zhoushu, book 50, describes that the last Hephthalite envoy had come in the second year of the Emperor Ming's rule (558) just before the Turkish conquest which resulted that the Hephthalite ulaqs were dispersed and their tributes to China therefore were ended. 16. The Siva-typed deities have been found in the north of the Hindukush and in ancient Sogdian countries. The present writer believes that they may have been a reflexion of the icons in the south of the Hindukush which reached the north through Bamiyan from the middle of the sixth century on. Chapter X 1. The ancien sanctuaire and the trois sanctuaires at Khair Khana may be identical with Conglingshan in the Suishu and with Elunaoshan in the Da Tang Xiyu ji respectively. See a detailed discussion in Chapter XII. 2. When Jolian was first established in Period III of Taxila, it consisted of both the main stupa and the quadrangular monastery. In Period IV the stupa court was enriched with many smaller stupas and surrounded by rectangular rows of small chapels. 3. Marshall eventually assigns the round bastions of the Sirsukh enclosure wall to the reign of Wima Kadphises. He believes that the Kushans came with this type of bastion to the Northwest from Central Asia. But the bastions contemporary to or even earlier than the Kushans in the north of the Hindukush are square. Ghirshman presumes that the round bastions came from Sirsukh to Begram accepting Marshall's dating. According to Ghirshman, the Sirsukh semi-circular bastion was introduced from Rome much earlier than the Sassanians first adopted this type of bastion after Shapur 1's victory over Valerian. Schlumberger and Bernard did not follow their predecessor. The masonry and the style of bastioned city wall of Sirsukh, which I find a well developed type in the history of masonry in Taxila, does not seem to fit well with the discussions giveb by Marshall and Ghirshman. The present knowledge of archaeology in Taxila does not allow us to figure out in which level of the strata the bastioned city wall of Sirsukh was built. 4. The other sites should be dealt with in comparison to this chronological order. Balkh III and Kobadian V show the same tendency of pottery as Durman III. But in contrast to Durman the handles of bowls are not decorated with incision, but made of twisted clay. If the Durman incised handle is derived of the Balkh twisted one, Balkh III and Kobadian V are evidently anterior to Durman III. Chapter XI 1. I am quite sceptical of the view maintained since Cunningham (1962) that the Hephthalites occupied the areas extending from the Kabul Valleys to Zabulistan, which they called J awuda or Jabula. First, the names on coins such as Jabula, Jabuvla, or Zoobol cannot necessarily be place names. Second, no evidence is found for that the Hephthalites invaded Kapisi, Kabul and Zabulistan through the western ranges of the Hindukush, but Chinese sources only suggest that Northwestern India was occupied by them after coming through the area between the Hindukush and the Karakorum. Third, Sheboluo which is recorded as an older name for Gandhara in the Song 276 Yun's narrative in the Luoyang Qielan ji (Record of Buddhist Monasteries in Luoyang) is not only still to be identified but also quite inconsistant with the fact that this denomination is older than the Hephthalite presence in Gandhara. 2. The date of Xuanzang's first visit to the kingdom of Kapisi given here is based on establishing his departure date from Changan which I set at the very beginning of 628. Discussion on when he left for India has usually been divided between those believing it was the first year of Zhenguan (the twenty-third of January 627 to the tenth of February 628) and those believing it was the third year of this era (the thirtieth of January 629 to the seventeenth of February 630). Both of the dates are clearly written in various original documents regarding the life of Xuanzang and in his own memorials to the second Emperor of the Tang dynasty. My own solution however is quite different from the scholars who have long deemed either of the two to be right. See "The Date ofXuanzang's Departure for India", a chapter in Kuwayama 1988: 1-33. 3. I was informed of the existence of several kinds of simple stamp decorations at Surkh Kotal through the paper entitled 'Les origines de la ceramique kouchane' by Dr. J.-c. Gardin and read by Prof. G. Fussman as his substitute on the occasion of the International Conference of Kushan Studies held in Kabul in November 1982. Then my brash request to Dr. Gardin at the Tenth International Conference of South Asian Archaeology held in Paris in 1989 for information on the as yet unpublished corpus of pottery types from Surkh Kotal was promptly granted. For this genuine scholarly kindness extended to me I should like to express my heartfelt gratitude. 4. The single example, very small with a diametre of 15 mm, was collected on the surface of a hill within the Buddhist temple complex at Lalma by the Kyoto University Archaeological Mission during the excavation in 1965. It consisted of five petals of a flower radiating from the centre. I am not aware of the existence of stamped pottery at Tapa Shotor. The use of such pottery must have been quite limited, even if it did exist in Jalalabad. 5. For more detailed explanations of the 1976 and 1978 excavations at Tapa Skandar, see Kuwayama 1978a and 1980. 6. Tapa Skandar underwent the three periods. The last occupation, Period III, is confined to a place between the Inner Fort of Period II and the westward slope, where a large oblong-planned mansion, or 'qal'a', was built in the early seventh century. 7. The usage of stone for the walls differs between the two locations in a minor point. River boulders, Ghirshman remarks, are the usual material for the foundations of houses at Begram III, while at Tapa Skandar II they only came into use sometime after quarried rubble became popular: in certain sondages at Tapa Skandar a foundation wall faced with river boulders overlies another structure which has a bastion masoned with quarried rubble. If no such superimposition is eventually revealed in the Kapisi district, the initiation of Begram III might be sometime later than the earlier phase of Period II at Tapa Skandar. 8. According to the Da Tang Da Ciensi Sanzangfashi zhuan, in 10 juan (scrolls), (Great Tang Biography of the Master of the Trepitaka and Dharma)-Xuanzang's most extensive biography edited first by Huili in five scrolls and later by Yancong whose preface is dated 20th April 688Xuanzang came to Kapisi in company with a Hinayana Master of the Trepitaka called Prajnakara who was originally from TakkadeSa and had been staying in the Nava-sangharama at Balkh. When they arrived, the king of Kapisi came out from the city to see Xuanzang together with 277 monks. There were about one hundred monasteries in Kapisi, and the monks from there strived to be the first to have Xuanzang stay at their monasteri~s. Among them there was a Hinayana monastery called the Shaluo(lao)jia which was said to have been built by a hostage from the Chinese emperor. Since Prajfiakara did not want to stay at a Mahayana monastery and monks of this monastery courteously invited them, Xuanzang eventually decided to stay at this monastery, where he made a summer retreat. Supposedly Xuanzang arrived in Kapisi in either the end of spring, 629 or the beginning of summer and left for Lanpaka to the east after a three month retreat. 9. The location of the royal city on the south bank of the river needs a simple explanation since neither the Da Tang Xiyu ji nor the Biography localize it so specifically. The locations of Buddhist temples in Kapisi, as in other kingdoms, are usually described with both direction and distance from the royal city. However, exceptions to this pattern are the descriptions of three monasteries established by the king and the queen ruling prior to the contemporary king, which are recorded without any distances. They are simply said to be situated to the Northwest of the royal city and on the south bank of a large river. The description in this case strongly suggests that the city and the monasteries were too close to make it necessary to refer to the distance and that they all must have been on the south bank of a larger river. 10. Another solution for Xibiduofaluo-ci, *Spen-varas (white hair), was proposed by Marquart and de Groot (1915: 265-266) . 11. Congling in Chinese classics is usually assigned to the Pamirs. 12 . For the identification of Cao with Kapisi, see Kuwayama 1975: 96-101. 13. Jaukuta is a phonetically restituted form for the original Chinese Caojuzha (Mizutani 1970: 370). The section of the country Xieyue in the Western Region chapter of the Tangshu says that originally called Caojuzha or Caoju and also called Hedaluozhi in the time of the Xianqing era (656-660), this country was renamed as Xieyue by the Empress Wu (690-704). Evidently Caojuzha ( or Caoju ) is first given in the Da Tang Xiyu ji and another denomination, Caoli, is also recorded as an epithet in the same section of the Da Tang Xiyu ji. Later in the thrid decade of the 8th century Huichao followed the Empress Wu to call it Xieyue, but he gave another name Shehuluosatana which was locally being called by the people of the country. Furthermore, Xuanying gives a similar denomination Shewutusuotana (Xuanying Yinyi, Vol. 18). If Caoli is restituted as J awuli or J awula, Xieyue as J awul and Shehuluosatana or Shewutusuotana as Jawula[da]stan, Jaukuta may be isolated from all of the other denominations. 14. I would like to express my most sincere gratitude to Prof. Maurizio Taddei for his kind delivery to Kyoto in the very beginning of January, 1989 of both information on the chronological situation of the medallion-stamped pottery found in the Early Period II at Tapa Sardar and his latest opinion about its dating. I also thank him for having allowed me the opportunity to go through the manuscript of his unpublished paper on Tapa Sardar originally read in a series oflectures delivered in the College de France (Taddei: Paris). 15. It is unclear, in fact, whether or not Building E is attributable to Period II since no stratigraphical cross-section connecting the sanctuaries with Building E is given in the original report of the excavations. It is at least certain that the Building E complex was built on top of the floor extending to the east from the Inner Court, and it is also clear that on the western part of the floor of the Inner Court stood the foundation of the Ancien Sanctuaire. 278 16. For the withdrawal of Hephthalite power in both Tokharistan and the Northwest at the advent of the West Turks and its relevant political, geographical, and cultural implications, see Kuwayama 1990. Chapter XIII (*)The Zhongguo Shudian edition is used for the Dynastic History of China such as the Weishu and the Zhoushu. The Arabic numbers in brackets represent the volume and page number of this edition of the history. For example, (6. 2279) stands for (Vol. 6, p. 2279). For the Tang Gaoseng zhuan, see the T Shinshu Daizokyo, Vol. 50. 1. It is important to appreciate the nature of the contents of the Western Region chapter of the existing Weishu in order to use it properly as authentic information about the Western Region kingdoms in the time of the Northern Wei Dynasty. Since it had lost many paragraphs when the compilers of the Beishi tried to use it as a source for their own Western Region chapter, they filled in the gaps in the remaining Weishu with the descriptions in the Zhoushu and the Suishu and eventually replaced the Western Region chapter of the original Weishu with the newly compiled chapter of the Beishi. The Western Region chapter of the existing Weishu is in fact an exact copy of that of the Beishi. That chapter of the existingT¥eishu therefore consists of mixed information derived from different sources of different times. However, fortunately, some of original paragraphs are easily accessible when the paragraphs quoted from the Zhousbu and the Suisbu are simply deleted from the existing Weisbu. This information about the Hephthalites in the existing Weisbu can thus be used as authentic and contemporary. 2. Identification of the Hephthalite headquarters in Tokharistan is fully discussed in Kuwayama 1989, 120ff. 3. For a chronological order of the Kapisian missions, see Kuwayama 1991A: 115, Table 3. 4. For the designation and illustration of each coin, see GobI 1967. 5. The twig-like ornament placed below the king's bust on some of the Hephthalite coins which is shared with the Napki coins has led scholars to assign the latter to the Hephthalites. Yet sharing this similar ornament does not necessarily lend support for the above identification, as the Napki coins also borrow some elements from Sassanian coins. For those who begin minting a new series of coins, the use of elements on coins anterior to them may not be unusual. 6. The wangshe cheng, a town of a king's residence, is usually taken as a Chinese translation of Rajagriha whenever it appears in Buddhist contexts. However, in this case, careful attention should be drawn to the fact that it is used in the Suisbu. The Sui court does ot seem to have received any information about India before Dharmagupta came to Changan. That is why Yancong and PeiJu edited the Record ofIndia in between 606 and 610 based on Dharmagupta's knowledge about India. Therefore, I do not think that Wijie and Du Xingman actually reached India or even climbed up Rajagriha. The section on the Hephthalite kingdom in the Western Region of the Zbousbu describes that the Hephthalite king resides in the town badiyan which means the wangshe cheng. This word badiyan is not a proper noun but simply means a town of a king's residence, or the capital of a kingdom. I believe that the wangshe cheng in the Suisbu is a copy from Wijie's Xifan ji and I do not hesitate to identify it with a leading town around which the Hephthalites customarily reside in tents as their winter home. I therefore suggest that Wijie 279 and Du Xingman visited the Hephthalite capital, although the Hephthalites had already declined on their visit. For a detailed discussion about the whereabout of the Hephthalite capital, or their qishlaq, see Kuwayama 1989, also referred to in Note 2. Chapter XVI 1. The ancient five divisions of the Subcontinent-Central, East, West, South and North India-as shown in the title Wu Tianzhuoguo (Five Indic Regions) do not always correspond to modern geographical divisions. "Central India" covers the area conceived as the center of Indian civilization, or the Ganga-Yamuna doab, often including western Bengal. "East India" includes most of Bengal and Orissa, while "West India" covers the middle and lower Indus Valleys and also the area to the southeast of them. "South India" is almost synonymous with the peninsular part of the Subcontinent, and "North India" includes the northwestern part of it extending from Panjab through northern Pakistan and Kashmir to eastern Afghanistan. The regional denominations of the Subcontinent used below should be read as the same as the above five unless otherwise mentioned. 2. The modern translated versions of the Pelliot 3532 manuscript includes Fuchs 1939, Sadakata 1971, Yang et al.1984 and Kuwayama 1992. 3. Yoshida (1993: 200) suggests that fudiye may share the original word with a Bactrian word bagodaiggo referred to by Davary (1982: 170-171), which appears on the reverses of the coins such as Emissions 244 and 245 (GobI 1967: III, PIs. 66 and 67). As for the interpretations of the legends, see also Hunbach 1966: 62-63 and GobI 1967: I, 167-168. 4. The Hu, a Chinese denomination of the western barbarians used by Huichao throughout his record, are not always the Sogdians or Iranians, as has been thought, since it is not acceptable to identify the stock in Gandhara and in the east of Kashmir with the Iranians. 280 REFERENCE Abbreviations ARASI: Archaeological Suroey of India, Annual Report. ASIR: A. Cunningham, Archaeological Suroey of India, Report V for the Year 187: (Calcutta 1875). BHC: Bulletin de Correspondence HelUnique, LXXIX. IA: Indian Antiquary. IAR: Indian Archaeology, A Review. CRAI: Academie des Inscriptions et Belles-Lettres, Comptes rendus. MIA: Materiali i Issledovaniya po Arkheologii. PBA: The Proceedings of the British Academy. T: The T. Shinshu Daizokyo (The Buddhist Canon, New Compilation of the Taisho era). 85 Vols., 1924-1932, Tokyo. Ackerman, H. Ch. (1975) Narrative Stone Reliefs from Gandhiira in the Victoria and AlbeTt Museum in London, Roma. Agrawala, R. C. (1968) Urdhvaretas Gal)esa from Afghanistan. East and West 18, 166-168. al-Fida'. Abu al-Fida' edited by M. Reinard, Paris, 1940. Allchin, F. R. (1968) Archaeology and the Date of Kanishka: The Taxila Evidence. Papen on the Date ofKani!ka, edited by A. L.Basham, Leiden, 4-34. Allchin, F. R. (1993) The Urban Position of Taxila and Its Place in the Northwest-Pakistan. UTban Form and Meaning in South Asia: The Shaping of Cities from PnhistoTic to Pncolonial Times, edited by H. Spodek and D. Meth Srinivasan, Washington, 69-81. Allchin, F.R. and N. Hammond (eds.) (1978) The ATchaeology of Afghanistan from eaTliest times to the Timurid peTiod, LondonlNew York/San Francisco. al-Ya'qubi, Ta'Tikh, edited by M. Th. Houtsma, Leiden, 1969. Anonym (1964) Guide du visiteuT, Musee national d'Afghanistan: Catalogue partie! des objets exposes, 2nd edition, Caboul. Auboyer,J. (1968) Afghanistan und seine Kunst, Prague. Altheim, F. (1969) Geschichte deT Hunnen 2, Berlin. Bachhofer, L., (1939) Early Indian Sculptun, 2 Vols., London. Ball, W. (1982)Anhaeological Gazetteer ofAfghanistan: Catalogue des sites archeologiques d'Afghanistan, avec la collaboration de Jean-Claude GaTdin, 2 Vols, Paris. Banerjea, J. (1956) The Development of Hindu IconogTaphy, Calcutta. Banerji, R. D. (1924), The Temple of Siva at Bhumara. Memoin of the Archaeological Suroey of India16, Calcutta. Barrett, D. (1954) Sculptures from AmaTavati in the BTitish Museum, London. Barrett, D. (1957) Sculptures of the Shahi Period. Oriental Art 3(2), 54-59. Barthoux, J. (1933) Les fouilles de Hadda 1: Stupas et sites. Memoires de la DeUgation aTcheologique franrcaise en Afghanistan 4, Paris. 281 -' ... Barger, E. and Ph. Wright (1941) Excavations in Swat and Explorations in the Oxus Territories of Afghanistan: A Detailed Report of the 1938 Expedition. Memoirs of the Archaeological Survey of India 64, Calcutta. Barrett, D. (1957) Sculptures of tile Shahi Period. Oriental Art 3, 54-59. Beal, S. (1884) Si-yu-ki: Buddhist Recm'ds of the Western World, translated from the Chinese of Hiuen Tsiang (AD 629), London. Beal, S. (1911) The Life of Hiuen-Tsiang, London. Begley, V. and R. D. de Puma (1991) Rome and India: The Ancient Sea Trade, Madison. Belenitski, A. M. and B. 1. Marshak, B. 1. (1971) L'art de Piandjikent a la lumiere des dernieres fouilles 1958-1968. Arts Asiatiques 23,3-39. Bernard, P. (1964) Les fouilles de Kohna Masdjid. Comptes rndus, Academie des Inscriptions et Belles-Lettres 1964, 212-221. Bernard, P. (1968) Chapiteaux corinthiens hellenistiques d'Asie central decouverts aAi Khanoum. Syria 45(1-2), 111-151. Bernard, P. and H.-P. Francfort (1978) Etudes de geographie histm'ique sur la plain d'Ai Khan.oum (Afghanistan). Publication de tURA 10, Memoire 1, Paris. Bernard, P. and F. Grenet (1981) Decouverte d'une statue du dieu solaire surya dans la region de Caboul. Studia Iranica 10,127-146. Biswas, A. (1973) The Political History of the Hunas in India, New Delhi. Bombaci, A. (1957) Ghazni. East and West 8(3),247-259. Bonji Kicho-Shiryo Shusei (A Collection of the Important Brahmi Scripts), 2 Vols., Tokyo, 1980. Bosworth, C. E. (1968) Sistan unde1' the Arabs, from the Islamic Conquest to the Rise of the Saffm'ids (30-2501651-864), Roma. Brunt, P. A. and]. M. Moore (1967) Res Gestae Divi August: The Achievements of the Divine Augustus, Oxford. Buhler, G. (1894) The Bhattiprolu Inscriptions. Epigraphia Indica 2,323-329. Buhler, G. (1959) Indian Palaeog1'aphy, reprint edition, New Delhi. Burgess, ]. (1887) The Buddhist Stupas of Amaravati and Jaggayyapeta in the Krishna District, Mad,'as Presidency, Surveyed in 1882, (Reprint edition, 1970, Varanasi). Burgess, J. (1897) The Ancient Monuments, Temples, and Sculptures ofIndia, London. Bussagli, M. (1962) Gandhara. The Encyclopedia of World Art 6, Roma, 18-37. Bussagli, M. (1984) L 'Arte del Gandhiira, Roma. Caroe, O. (1958) The Pathans 550 B.G.-AD 1957, London. Carl, ]. (1959A) Le fortin du Saka et Ie monastere du Gul Dara (Rapport de]. Carl, 1935). Memoires de la Delegation anh6logique franfaise en Afghanistan 8, Paris, 13 -18. Carl, J. (1959B) Le bazar de Begram. Memoires de la Delegation anh6logique franfaise en Afghanistan 8, Paris, 84-102. Carl,]. (1959C) Le Tepe Kalan du Koh-i Pahlavan. Memoires de la Delegation arch610gique franfaise en Afghanistan 8, Paris, 128-132. Carl,]. and]. Hackin, (1959) Le monastere bouddhique du Tepe Marandjan (1933). Memoires de la Delegation anh610giquefranfaise en Afghanistan 8, Paris, 7-12. Carter, M. L. (1985) Hsuan-tsang and the Colossal Buddhas at Bamiyan. Studies in Buddhist art of South Asia, edited by A. K. Narain, New Delhi. 282 Casson, L. (1989) The Periplus Maris Erythraei: Text with Introduction, Translation, and Commentary, Princeton. Chanda, R. (1919-1920) Some Unpublished Amaravati Inscriptions~ Epigraphia Indica 15,258-275. Chavannes, E. (1903A) Voyage de Song Yun dans l'Udyana et Ie Gandhara (518-522 p.c.). Bulletin de l'Ecole franfaise d'Etreme Orient 3, 379-441. Chavannes, E. (1903B) Documents sur les Tou-Kiue (Turcs) Occidentaux, Paris. Chavannes, E. (1905) Jinagupta (528-605 apres].-c.). T'oung Pao 2(4),332-356. Christensen, A. (1944) L'Iran sous les sassanides, 2me edition. Copenhagen. Collingwood, R. G. and 1. Richmond (1969) The Archaeology of Roman Britain, revised edition, London. Crompton, C. A. (1875) Report of the Excavations at Mound Shahji-ka Dheri, near Peshawar. Second Supplement to the Government Gazette, Punjab and its Dependencies, 18th November, 187~Lahore. 717-718. Cunningham, A. (1962) Later Indo-Scythians (Complete Works of Alexander Cunningham, edited by A. K. Narain), Varanasi. Cunningham, A. (1871) The Ancient Geography of India 1, Buddhist Period, London. Dagens,B. (1964A) Monasteres rupestres de la vallee de Foladi. Monuments pre-isfamiques d'Afghanistan, Memoires de la Delegation franfaise en Afghanistan 19, Paris, 43-60. Dagens, B., (1964B), Fragments du sculptures inedits. Monuments preislamiques d Afghanisan, Memoires de la Delegation franfaise en Afghanistan 19, Paris, 11-39. Dalton, O.M. (1964) The Treasure of the Oxus, 3rd edition, London. Dani, A.H. (ed.) (1968-69) Chakdara Fort and Gandhara Art. Ancient Pakistan 4, Peshawar. Dani, A.H. (1986) The Historic City ofTaxila, Paris/Tokyo. Dani, A.H. (1983) Chilas. The City of Nanga Parvat (Dyama1), Islamabad Davary, G. D. (1982) Baktrisch, Ein Wijrterbuch auf Grund del' Inschl'iften, Handschriften, Miinzen und Siegelsteine, Heidelberg. Deo, S. B. and]. P. Joshi (1972) Pauni Excavation 1969- 70, N agpur. de la Vallee Pousin, L. (1935) Dynasties et Histoil'e de l'Inde depuis Kani~ka jusqu 'aux invasions musulmanes, Paris. de Morgan,]. (1923-1936) Manuel de numismatique orientale de l'antiquite et du moyen age 1, Paris. Dhavalikar, M. K. (1971) A Note on Two Ganesa Statues from Afghanistan. East and West 21(3-4), 331-340. Doerfer, G. (1975) TU1'kische und Mongolische Elemente im Neupersischen 4, Wiesbaden. Dobbins, K. W. (1971) The Stiipa and Vihara ofKanishka I, Calcutta, 1971. Edelberg, L. (1957) Fragments d'un stupa dans la vallee du Kunar en Afghanistan. Arts Asiatiques 4,199-207. Eisner, M. (1979) Zur Typologie der Mausoleen des Augustus und Hadrian. Mitteilungen des Deutschen Archiiologischen Instituts (Rom) 86, 319-324. Eisner, M (1986) Zur Typologie del' Grabbauten im Suburbium Roms. Rijmische Mitteilungen, Erganzungsheft 26, Mainz. Elliot, H. M. (1869) The History of India as Told by Its Own Histol'ians, London. Enoki, K. (1951) Efutalu minzoku no Kigen (The Origin of the Hephthalites). Wada hakushi Kanreki Kinen Toyoshi Ronso (Studies in Oriental History in commemoration of the 283 Sixtieth Birthday of Dr. Wada), Tokyo, 133-150. Enoki, K. (1955) The Origin of the White Huns or the Hephthalites. East and West 6,234-236. Erdosy, G. (1990) Taxila: Political History and Urban Structure. South Asian Archaeology 1987, edited by M. Taddei, Rome, 657-674. Errington, E. (1987) Tahkal: The Nineteeth-Century Record of two Lost Gandhara Sites. Bulletin of the School of Oriental and African Studies University of London 50(2),301-324. Errington, E., J. Cribb Fabregues, C. (1987) The Indo-Parthian Beginnings of Gandhara Sculpture. Bulletin of the Asia Institute 1, 33 -43. Faccenna, D. (1962) Sculpture from the Sacred Area of Butkara I (Swat, W Pakistan) 2: Plates, Descriptive Catalogue by M. Taddei, Rome. Faccenna, D. (1962) Mingora: Site of Butk-ara 1. Report on the Campaigns 1956-1958 in Swat (Pakistan), Roma. Faccenna, D. (1964) A Guide to the Excavations in Swat (Pakistan) 1956-1962, I, IsMEO Reports and Memoirs 3(1), Roma. Faccenna, D. (1982) ButkaraI (Swat, Pakistan) 1956-1962, Roma. Felsbilder (1985) Zwischen Gandhiira und den Seidenstrassen Felsbilder am Karakorum Highway, Mainz am Rhein. Fischer, K. (1964) Une tete sivai'te en marbre de l'Afghanistan oriental. Arts Asiatiques 10(1), 35-42. Fischer, K. (1969) Preliminary Remarks on Archaeological Survey in Afghanistan. Zentralasiatische Studien 3,327-408. Fitzsimmons, T. (1994) Ceramis and the Chronology of Dilberdzhin Tepe and Zhiga Tepe (North Afghanistan). Zinbun: Annals of teh Institute for Research in Humanities, Kyoto University 29,33-60. Fitzsimmons, T. (1996) Chronological Problems at the Temple of the Dioscuri, Dil'berdzin Tepe (North Afghanistan). East and West 46(3-4),271-298. Fleet, J. F. (1885) The Legends on the Silver Coins of the Early Guptas and Others connected with them. Indian Antiquary 14,65-68. Fleet, J. F. (1886) The History and Date of Mihirakula. Indian Antiquary 15,245-252. Fleet, J. F. (1888) Inscriptions of the Early Gupta Kings and their Successors. Corpus Inscriptionum Indicarum 3, reprint ed edition, Varanasi, 1970. Fleet, J. F. (1889) The Coins and History of Toramana. Indian Antiquary 18,225-230. Forte, A. (1974) Divakara (613-688), Uno Monaco Indiano nella Cina dei Tang. Annali di Ca' Forscari 13(3). Foucher, A. (1901) Notes sur la geographie ancienne du Gandhara. Bulletin de I'Ecole Franfaise d'Extreme-Oreint 1, 322-369. Foucher, A. (1905) L'art greco-bouddhique du Gandhiira 1, Paris. Foucher, A. (1918) L'art greco-bouddhique du Gandhiira 2, Paris. Foucher, A. (1942-47) La vieille route de l'Inde de Bactres it Taxila, 2 Vols., Memoit-es de la Delegation archeologique franfcaise en Afghanistan 1, Paris. Frye, R.N. (1979) Islamic Iran and Central Asia (7th-12th centuries), London. Fuchs, W. (1939) Huei-chao's Pilgerreise durch Nordwest-Indien und Zentral-Asien urn 726. 284 Sitzungsberichte der Preussischen Akademie der Wissenschaften, phil.-hist. Klasse 30, Berlin, 426-469. Fujita, T. (1931) Echo Ogotenjikukokuden Senshaku (Huichao's Narrative of his Travels in Five India: Annotation and Commentary), Beiping. Fujiyoshi, M. (1987) Mappoka toshiteno Narendorayashasu - Shu Zui Kakumeito Tokugochojakyo (Narendrayasas as an Advocator of the Idea of the Final Dharma the Zhou-Sui Revolution and the Dehu-Millionaire. Toyoshikenkyu Oournal of Oriental Studies) 46(1),29-56. Fun9.ki, K. (1954) Efutalu ni kansuru Chugoku-shiryo ni tuite (Chinese Sources on the Hephthalites). Shi en 61,57-77. Fussman, G. (1974) Ruines de la vallee de Wardak.ArtsAsiatiques 30,65-95. Fussman, G. (1976) Monuments bouddhiques de la region de Caboul1: Le monastere de Gul Dara. Memoires de la Delr!gation archeologique franfcaise en Afghanistan 22, Paris. Fussman, G. (1987) Numismatic and Epigraphic Evidence for the Chronology of Early Gandharan Art. Investigating Indian Art, edited by W. Lobo. and M. Yaldiz, Berlin, 67-88. Fussman, G. (1993) Taxila: The Central Asian Connection. Urban Form and Meaning in South Asia: The Shaping of Cities from Prehistoric to Precolonial Times, edited by H. Spodek and D. Meth Srinivasan, Washington, 83-100. Gelder, J. M. van (tr.) (1963) Miinava Srautasutra belonging to the Maitriiya'lJi sa'fJlhitiis, reprint edition, Delhi, 1985. Ghafoor, M. A. (1965-66) Two Lost Inscriptions Relating to the Arab Conquest of Kabul and the North West Region of West Pakistan. Ancient Pakistan 2,4-12. Ghirshman, R. (1946) Begram: Recherches historiques et archeologiques sur les kouchans. Memoire de la Delr!gation archr!ologique franfaise en Afghanistan 12, Le Caire. Ghirshman, R. (1948) Les chionites-hephthalites. Memoire de la Delegation archr!ologique franfaise en Afghanistan 13, Le Caire. Ghosh, A. (1947 -48) Taxila (Sirkap) 1944-45. Ancient Pakistan 4,41-84. GobI, R. (1967) Dokumente ZU1' Geschichte der iranischen Hunnen in Baktrien und Indien, 4 Vois., Wiesbaden. GobI, R. (1968) Sasanidische Numismatik, Braunschweig. Godard, A., Y. Godard and J. Hackin (1928) Les antiquites bouddhiques de Biimiyiin. Memoire de la Delr!gation archr!ologique franfaise en Afghanistan 2, Paris-Bruxelles. Goetz, H. (1957) Late Gupta Sculpture in Afghanistan: the "Scorretti Marble" and Cognate Sculptures. Arts Asiatiques 4, 13-19. Goetz, H. (1964-65) A Kashmiri Lingam of the 10th Century. Artibus Asiae 23,275-279. Goetz, H. (1969) The Conquest of Northern and Western India by Lalitaditya-Muktapida of Kasmir. Studies in the History and Art of Kashmir and the Indian Himalaya, Wiesbaden. Gold, M. (1976) The Tarikh-e Sistan, Roma. Gupta, P. L. and D. C. Sircar (1972-73) Uma Mahesvara Image Inscription from Skandar (Mghanistan). Journal ofAncient India History (Calcutta) 4(1-2), 1-4. Gupta, S. P. (ed.) (1985A) Kushana Sculpturesfrom Sanghol (1st-2nd Century AD): A Recent Discovery 1, New Delhi. Gupta, S.P. (ed.) (1985B) Masterpieces from the National Museum Collection. New Delhi. Hackin, J. (1959A) Recherches archeologiques a Bamiyan en 1933. Diverses recherches archeoloqiques 285 en Afqhanistan (1933-1940), Memoire de la Delegation arcbeologique franfaise en Afghanistan 8, Paris, 1-6. Hackin, J" (1959B), Le monastere bouddhique de Fondukistan. Diverses recherches arcbeologiques en Afghanistan. Memoires de la Delegation arcbeologique franfaise en Afghanistan 8, Paris, 49-58. Hackin, J, and J, Carl (1933) Nouvelles recherches archeologiques a Biimiyiin 1930. Memoire de la Delegation arcbeologique franfaise en Afghanistan 3, Paris. Hackin, J, and J, Carl (1936) Recherches arcbeologiques au col de Khair Khana pres de Kabul - Fouilles J. Carl et J. Hackin, Memoires de la Delegation archeologique franfaise en Afghanistan 7, Paris. Hallade, M. (1968) The Gandhiira Style and the Evolution of Buddhist Art, London. Hambis, L. (ed.) (1961) Toumchouq, Mission Paul Pelliot 1, Paris. Hargreaves, H. (1911) Excavations at Shah-ji-ki Dheri. Archaeological Survey of India, Annual Report, F1'ontier Circle, 1910-11, Peshawar, 1911, 12-16. Hargreaves, H. (1914) Excavations at Shah-ji-ki Dheri, Annual Report of the Archaeological Survey of India 1910-11, Calcutta, 1914,25-32. Hargreaves, H. (1923-24) Frontier Circle (Hund, the Ancient Udabhanda), Exploration, Aurchaeological Survey of India, Annual Report 1923-24, 68-70. Harle, J,c. and A. Topsfield (1987) Indian Art in the Ashmolean Museum, Oxford. Haslett, P. (1876) Report on Archaeological Explorations at Tahkal near Peshawar. Punjab Government Gazette, Supplement, 30th March 1876,238-239. In Errington 1987, 320-321. Haussig, H. W. (1956) Quellen tiber die zentralasiatischen Herkunft der europaischen Awaren. Central Asian Journal 2, 21-43. Hawqal. Bibliotheca Geographicorum Arabicorum 2, Leiden, . Hermann, A. (1925) Die Hephthaliten und ihre Beziehungen zu China. Asia Major 2, 565-580. Herzfeld, E. (1920)Am Tor von Asien: Felsdenkmale aus Irans Heldenzeit, Berlin. Higuchi, T. ed. (1984) Biimiyiin, 4 Vols. Kyoto. Higuchi, T and S. Kuwayama (1970) Chaqalaq Tepe, Kyoto. Houtsma, M. Th. (ed.) (1969) Ibn-Wadhih qui dicitur Al-Ja'qubi, Historiae, 2 Vols., Leiden. Hulsewe, A. F. P. and M. A. N. Loewe (1979) China in Central Asia, the Early Stage: 125 BC- AD 23. An Annotated Translation of Chapters 61 and 96 of the History of the Former Han Dynasty. Sinica Leidensia 14, Leiden. Humbach, H. (1966) Baktrische Sprachdenkmaler 1, Wiesbaden. Huntington, S. L. (1985) The Art ofAncient India, New York/Tokyo. IA (1874) The Lanjadibba or Mound at Bhattiprol, Repalli, Taluqa. The Indian Antiquary 3, 124. IAR (1957-58) Excavation at Nagarjunakonda, District Guntur. Indian Archaeology 1957-58: A Reviw, 5-9. IAR (1961-62) Stupa-Mound and Early Sculptures, Ter, District Osmanabad. Indian Anhaeology 1961-62: A Review, 102. IAR (1967-68) Excavation at Ter, District Osmanabad. Indian Archaeology 1967-68: A Review, 25. IAR (1971-72) Excavation at Sanghol, District Ludhiana. Indian Archaeology 1971-72: A Review, 39-41. IAR (1972-73) Excavation at Sanghol, District Ludhiana. Indian Archaeology 1972-73: A Review, 286 28. IAR (1984-85) Excavation at Sanghol, District Ludhiana, Indian Archaeology 1984-85: A Review, 62-66. IAR (1985-86A) Excavation at Adurru, District East Godavali. Indian Archaeology 1985-86: A Review, 1-2. IAR (1985-86B) Excavation at Sanghol, District Ludhiana. Indian Archaeology 1985-86: A Review, 67-69. Ingholt, H. and I. Lyons (1957) Gandharan Art in Pakistan, New York. Irwin, J. (1979) The Stupa and the Cosmic Axis: The Archaeological Evidence. South Asian Archaeology 1977, edited by M. Taddei, Naples, 799-845. Jettmar, K. (1980) Neuentdeckte Felsbilder und -inschriften in den Nordgebieten Pakistans: ein Vorbericht. Allgemeine und Vergleichende Archiiologie, Beitrage 2, 151-199. Joshi, N. P. (1966) Mathura Sculptures: A Hand Book to Appreciate Sculptures in the Archaeological Museum, Mathura, Mathura. Joshi, N. P. and R. C. Sharma (1969) Catalogue of Gandhiiran Sculptur·es in the State Museum Lucknow, Lucknow. Julien, S. (1853) Histoire de la vie de Hiouen-thsang et de ses voyages dans l'Inde depuis l'an 629 jusqu'en 645, par Hoei-Ii et Yen-thsong, suivie de documents et d'iclaircissements giogmphiques tires de la relation originale de Hiouen-thsang, Paris. Kak, R. Ch. (1933)Ancient Monuments of Kashmir, London. Kawakatsu, Y. (1981) A propos de la pensee de Huisi. Bulletin de l'Ecole franfaise d'Etreme Orient 69,97-105. Kawakatsu, Y. (1987) Chugokuteki Shin Bukkyo Keisei eno Enerugi (Energy for the Birth of New Buddhism to Chinese People). Chugoku Chuseino Shukyo to Bunka, edited by M. Fukunaga, Kyoto, 501-537. Klimburg-Salter, D. (1989) The Kingdom of Biimiyiin: Buddhist Art and Culture of the Hindu Kush, Naples-Rome. Klimburg-Salter, D. (1988) Bamiyan: Recent Researches. East and West 38(1-4),305-312. Kodera, T., M. Maeda and A. Miyaji (1971A) Paintings and Architecture of Bamiya: Researches on the Iranian Style ofthe Buddhist Art. Gakujutsu Geppo 23(11), 711-715 (in Japanese). Kodera, T., M. Maeda and A. Miyaji (1971B) Biimian, Nagoya (in Japanese). Konow, S. (1929) Corpus Inscriptionum Indicarum 2 (1): Kharoshthi Inscriptions with the exception of those ofAtoka. Calcutta. Kottkamp, H. (1992) Der Stupa als Repriisentation des buddhistischen Heilsweges: Untersuchungen ZU'f· Entstehung und Entwicklung architektonischer Symboloik. Studien in Oriental Religions 25, edited by W. Heissig and H.-]. Klimkeit, Wiesbaden. Kreitman, N. (1992) Kanishka Casket and Related Designs. In Errington and Cribb 1992. 193-197. Kruglikova, I. T. (1974) Dilberdjin, Moskva. Kurita, I. (1988) Gandhiiran Art 1: The Buddha's Life Story, Tokyo. Kuwayama, S. (1972A) Dairiseki Hindu-zo wa Hindu-ocho no monoka. Tohogakuho 43, 1-54 (in Japanese). Kuwayama, S. (1972B) The First Excavation at Tapa Skandar. Archaeological Survey of Kyoto 287 University in Afghanistan 1970, Kyoto, 5-14. Kuwayama, S. (1974A) Excavations at Tapa Skandar, Second Interim Report. Kyoto UniVeisity Archaeological Survey in Afghanistan 1972, Kyoto, 5-13. Kuwayama, S. (1974B) Kapisi Begram III: Renewing its Dating. Orient 10, 57-78. Kuwayama, S. (1974C) Takishira Butsuji no Ganran Haichi (Historical Approach to the Monastic Layout in Taxila). Tohogakuho.Journal of Oriental Studies 46, 327-354. Kuwayama, S. (1975) Khair Khana and its Chinese Evidence. Orient 11, 93 -1 07. Kuwayama, S. (1976) The Third Excavations at Tapa Skandar. Japan-Afghanistan Joint Al-chaeological Survey in 1974, Kyoto, 5-15. Kuwayama, S. (1978A) The Turki Sahis and Relevant Brahmanical Sculptures in Afghanistan. East and West 26(3-4),375-407. Kuwayama, S. (1978B) The Fourth Excavation at Tapa Skandar. Japan-Afghanistan Joint Achaeological Survey in 1976, Kyoto, 5-12. Kuwayama, S. (1978C) Notes on the Introduction of Square Podium to Taxilan Stupas. Orient 14,23-34. Kuwayama, S. (1980) The Fifth Excavation at Tapa Skandar. Japan-Afghanistan Joint Archaeological Survey in 1978, Kyoto, 5-15. Kuwayama, S. (1981) Kapisi-koku Hen'nen Shiryo Ko (Chinese Documents on the History of Kapisi). Bukkyo-Geijutsu (Ars Buddhica) 137, 86-114. Kuwayama, S. (1982A) Kapisi-koku Hen'nen Shiryo Ko (Chinese Documents on the History of Kapisi). Bukkyo-Geijutsu (Ars Buddhica) 139,80-116. Kuwayama, S. (1982B) Kapisi and Gandhara according to Chinese Sources. Orient 18,133-139. Kuwayama, S. (1983) Indo eno Michi - Genjo to Kochi (A Passage to India: Xuanzang and Prabhakaramitra). Tohogakuho 55, 145-210. Kuwayama, S. (1984) Buddhist Temples in Taxila and Gandhara: A Historical Review of Monastic Arrangements. A Catalogue of the Exhibition of Gandhiiran Art of Pakistan. Tokyo. Kuwayama, S. (1985) Bamiyan Daibutsu Seiritsu ni kakawaru Futatsu no Michi (Two Itineraries Concerning the Date of the Colossi in Bamiyan), Tohogakuho 57,109-209. Kuwayama, S. ( 1987 A) Literary Evidence for Dating the Colossi in Bamiyan. Orientalia: Iosephi Tucci Memoriae Dicata. Serie Orientale Roma 56 (2), edited by Gnoli, Gh. and L. Lanciotti, Roma, 703-727. Kuwayama, S. (1987B) Tapa Shotor and Lalma: Aspects of the Stupa Court at Hadda. Annali (Istituto Universitario Orientale di Napoli), 47(2),154-176. Kuwayama, S. (1988) How Xuanzang Learned about Nalanda. Tang China and Beyond: Studies on East Asia from the Seventh to the Tenth Century. Italian School of East Asian Studies, Essay 1, edited by A. Forte, Kyoto, 1-33. Kuwayama, S. (1989) The Hephthalites in Tokharistan and the Northwest. Zinbun: Annals of the Institute for Research in Humanities-Kyoto University 24, 89-134. Kuwayama, S. (1990) The Buddha's Bowl in Gandhara and Relevant Problems. South Asian Archaeology 1987, Roma, 945-978. Kuwayama, S. (1991A) The Horizon of Begram III and Beyond: A Chronological Interpretation of the Evidence for Monuments in the Kapisi-Kabul-Ghazni Region. East and West 41(1-4),79-120. 288 Kuwayama, S. (1991B) L'Inscription du Ganesa de Gardez et la chronologie des Turki-~ahis. JournalAsiatique 279(3-4),267-287. Kuwayama, Sh. (ed.) (1992) Huichao's Wang Wu-Tianzhuguo zhuan:Record of Travels in the Five Indic Regions, Translation and Commentary (~Jm1±1L~~~~liJfJE), Kyoto (in Japanese). Kuwayama, S. (1996) Tapa Skandar. Enciclopedia dell'Arte Antica, Classica et Orientale: Second Supplemento 1971-1994, 5(Romana, Arte-Zuglio), Rome, 528-531. Kuwayama, S. (1997) The Main Stiipa of Shah-ji-ki l)heri: A Chronological Outlook, Kyoto. Legge, J. (1886) A Record of Buddhistic Kingdoms, Oxford. Le Berre, M. and Schlumberger, D. (1964) Observations sur les remparts de Bactres. Monuments pre-islamique d'Afghanistan. Memoires de la Deligation archiologique franfaise en Afghanistan 19,61-105. Le Berre, M. (1965) Prospection a Khisht Tepe. Le tdsor monitain de Qunduz. Memoins de la Deligation archiologique franfaise en Afghanistan 20, Paris, 83-88. Le Berre, M. (1987) Monuments pre-islamiques d'Hindukush centrale. Memoires de la Deligation archiologique franfaise en Afghanistan 24, Paris. Legge, J. (1886) A Record of Buddhistic Kingdoms being an Account by the Chinese Monk Fa-bien of bis Travels in India and Ceylon (AD 399-414) in Search of tbe Buddbist Books of Discipline, Oxford. Le Strange, G. (1905) Tbe Lands of the Eastern Caliphate: Mesopotamia, Penia and Central Asia from tbe Moslem Conquest to tbe time ofTimur, Cambridge. Levi, S. (1895) Note additionelle: Kipin, situation et historique. Journal asiatique 9(6), 371-384. Levi, S. (1896) Note rectificative sur Ie Ki-pin.Journal asiatique 9(7), 161-162. Levi, S. (1897) Note additionnelle sur les Indo-sythes.Journal asiatique 9(10),521, fn. 2. Levi, S. (1902) Notes chinoises sur l'Inde 1-2. Bulletin de l'Ecole franfaise d'Extreme Orient 2, 246-255. Levi, S. (1903) Notes chinoises sur l'Inde 3. Bulletin de l'Ecole franfaise d'Extreme Orient 3,38-53. Levi, S. (1904) Notes chi noises sur l'Inde 4. Bulletin de l'Ecole franfaise d'Extreme Orient 4, 543-579. Levi, S. (1905) Notes chinoises sur l'Inde 5. Bulletin de l'Ecole franfaise d'Extreme Orient 5, 253-305. Levi, Sand E. Chavannes (1895) Voyages des pelerins bouddhistes: L'Itineraire d'Ou-kong (751-790). Journal asiatique 9(6), 341-371. Lippe, A. (1970) The Freer Indian Sculptures. Smithonian Institution Ft·eer Gallety of Art Oriental Series 8, Washington. Litvinsky, B. A. and T. 1. Zeymal (1971) Adzhina Tepe, Moskow. Ltiders, H. (1912) A List of Bmhmi Inscriptions from the Earliest Times to about AD 400 with tbe exception of thos of Asoka. Appendix to Epigraphia Indica and Record of tbe Archaeological SU77Jey of India 10, Calcutta (Reprint Edition, Varanasi, 1973). Lukonin, v.G. (1967) Persia 2, Archaeologia Mundi, Geneva. MacDowall, D. W. (1968) The Shahis of Kabul and Gandhara. Numismatic Cbronicle 7, 189-224. Maclean, D.N. (1989) Religion and Society in Arab Sind, Leiden. Majumdar, R.c. (ed.) (1962) Tbe Classical Age. The Histoty and Culture of tbe Indian People 3, 2nd edition, Bombay. 289 Margabandhu, C. (1985) Archaeology of the Satavahana-Kshatrapa Times, Delhi. Markwart, ]. (1938) Wehrot und A rang. Untersuchungen zur My th isch en und Geschichtlichen Landeskunde von Osten, edited by H. H. Schaeder, Leiden. Marquart,]. (1901) Eransahr nach del' Geographie des Ps. Moses Xorenac'i, Berlin. Marquart, ]. and]. ]. M. de Groot (1915) Das Reich Zabul und der Gott Zun vom 6.-9. Jahrhundert. Festschl'ift Eduard SachaU, edited by G. Weil, Berlin, 248-292. Marshall,]. (1912-13) Excavations at Taxila.Archaeological Survey of India, Annual Report 1912-13, 1-52. Marshall,]. (1914) Shah-ji-ki-Dheri. Archaeological Survey of India, Annual Report of the DirectorGeneral ofArchaeology, Part 1, 1911-12, Calcutta. Marshall,]. (1937) A Book Review: Recherches Archeologiques au Col de Khair Khaneh pres de Kabul. Journal of the Royal Asiatic Society of Great Britain and Ireland 1937,499-501. Marshall, ]. (1951) Taxila: An Illustrated Account ofAl'chaeological Excavations carried out at Taxila, 3 Vols., Cambridge. Marshall,]. (1936)A Guide to Taxila, 3rd edition, Delhi. Marshall,]. (1960)A Guide to Taxila, 4th edition, Cambridge. Marshall, J. (1960) The Buddhist Art of Gandhiit'a, Cambridge. Marshall,]., A. Foucher and N. G. Majumdar (1940) The Monument of Sanchi, 2 vols. Delhi. Maspero, H. (1911) Sur la date et authenticite du Fou-fa-tsang-yin-yuan-tchouan. Melanges d'lndianisme, Paris, 13 Of. Masson, Ch. (1842) Nan'ative of Variuus Journeys in Balochistan, Afghanistan and the Panjab, 3 . vols., London. Matsuda, H. (1970) Kodai Tenzan no Rekishi-Chirigakuteki Kenkyu (A Study of History and Geography in the Ancient Tianshan), revised edition, Tokyo (in Japanese). Meunie, ]. (1942) Shotorak. Memoires de la Delegation archeologique franfcaise en Afghanistan 10, Paris. Meunie, ].(1959) Begram-Fouilles de 1938. Memoires de la Delegation archeologique franr;caise en Afghanistan 8, Paris, 103-106. Meyer, ]. ]. (1903) /Vemendra's Samayamiitrika (Das Zauberbuch der Hetraten). Altindische Schelmenbuchel'l, Leipzig. Minorsky, V. (1970) Hudud al-'Alam (The Regions of the World): A Persian Geogl'aphy 372 A.H.982 AD, 2nd edition, edited by C. E. Bosworth, London. Mirashi, V. V. (1975) A Note on the Uma MaheSvara Image Inscription from Skandar (Afghanistan). Journal of the 01'iental Institute, Baroda 25, 155 -15 6. Mitchner, M. (1975A) Some Late Kushano-Sassanian and Early Hephthalite Silver Coins. East and West 25(1-2), 157-165. Mitchner, M. (1975B) Who Were Napki Malik. East and West 25(1-2), 167-174. Mitra, D. (1971) Buddhist Monuments, Calcutta. Mitra, D. (1981) Ratnagil'i (1958-1961) 1. Memoirs of the Archaeological Survey of India 80, New Delhi. Mizuno, S. (ed.) (1967) Hazal'-sum and Fil-khana, Kyoto. Mizuno, S. (ed.) (1968) DU1'man Tepe and Lalma, Kyoto. Mizuno, S. (ed.) (1969) Mekhsanda, Kyoto. 290 Mizuno, S. (ed.) (1970) Basawal and Jelalabad-Kiibul. Kyoto. Mizuno, S. and T. Higuchi (ed.) (1978) Thareli, Kyoto. Mizuno, S. and T. Nagahiro (1956) Yunkang: The Buddhist Cave- Temples of the Pifth Century A. D. in Northern China, 16 Vols., Kyoto. Mizutani, Sh. (1970) Daito Sai'ikiki (The Da Tang Xiyu ji: AJapanese Version and Commentary), Tokyo. Murgotten, F. C. (1969) The Origin of the Islamic State, being a translation from the Arabic accompanied with annotations, geographic and historic notes of the Kitab PutE!; al-Buldan of aI-Imam abu-l 'Abbas Ahmad ibn-Jabir al-Baladhuri2, 1st AMS ed. (rep. from the 1924 ed.), New York. Moustamindy, Ch. (1969) La fouille de Hadda. Camptes rendus, Academie des Inscriptions et BellesLettres, janvier-mars, 1969, 119-128. Mostamindi, M . et Sh. (1969) Nouvelles fouilles a Hac;lc;la (1966-1967) par l'Institut Afghan d'Archeologie.Arts Asiatiques 19, 15-36. Mukherjee, B. N. (1964) Shah-ji-ki Dheri Casket Inscription. The British Museum Quarterly 28 (1-2), 39-46. Mustamindi, Sh. (1971) Preliminary Report of Hadda's Fifth Excavation Period. Kushan Cultun and History 2, ed. Falima Ayubi, Kabul, 43-50. Mustamandi, Sh. and M. (1967-68) The Excavation of the Afghan Archaeological Mission in Kapisa.Afghanistan 20(4), 67-79. Organ, R. M. and A. E. Werner (1964) The Restoration of the Relic Casket from Shah-ji-ki-Oheri. The British Museum Quarterly 28 (1-2). 46-51. Otani, Sh. (1934) Echo-Ogotenjikukokuden chu no ichi-ni ni tuite (On a Few Problems of the Accounts in Huichao's Wang Wu-Tianzhuguo zhuan). ada-sensei ShOju-kinen Chosen Ransh u (Festschrift for Prof. Oda: Papers on Korea), edited by R. Fujita, Keijo-fu (Seoul), 143-160 (in Japanese). Pal, P. (1975) Bronzes of Kashmir, Graz. Pal, P. (1986) Indian Sculpture: A Catalogue of the Los Angeles County Museum of Art Collection 1, Los Angeles/Berkeley/London. Pal, P. (1989) Art and Architecture ofAncient Kashmir, Bombay. Pandit, R. B. (ed.) (1927) The Gaucjavaho, A Prakrit Historical Poem by Vakpati. 2nd ed., edited by N. B. Utgikar, Poona (With supplementary notes by Utgikar, ccxxxi-cclxxii). Parlato, S. (1979) A Brahmi Inscription on a Mud-plaster Floor at Tapa Sardar, Ghazni, East and West 29(1-4),265-269. Paul, D. (1987) The Art ofNalanda. Development of Buddhist Sculpture 600-1200 AD, Amsterdam. Paul, P.-G. (1986) Early Sculpture of Kashmir', Before the Middle of the Eighth Century AD: An Approach to Art History and Epigraphy of the Jhelum Valley and its Peripheral Regions, Amsterdam. Pelliot, P. (1903) A review article, E. H. Parker, China, the Avars, and the Franks; The Ephthalite Turks; Chinese Knowledge of Early Persia. Bulletin de tEcole franfaise d'Extreme Orient 3 (1903),99-101. Pelliot, P. (1908) Une bibliotheque medievale retrouvee au Kan-sou. Bulletin de tEcole Ft'anfaise d'Extreme Orient 8, 501-529. Pelliot, P. (1914) Les noms propres dans les traductions chinoises du Milindapaiiha. Journal 291 asiatiques 11 (4), 379-419. Pelliot, P. (1936) Apropos du 'Tokharien.' Toung Pao, 32,259-284. Peppe, W. C. (1898) The Piprahwa Stupa Containing Relics of Buddha. Journal of the Royal Asiatic Society 1898,573-588. Petech, L. (1950) Northern India according to the Shui-ching-chu. Serie Orientale Roma 2, Roma. Petech, L. (1988) Note su Kapis! e Zabul. Selected Papers on Asian History. Serie Orientale Roma 60,187-194. Piggott, S. (1943) The Earliest Buddhist Shrines. Antiquity 17(65), 1-10. Przyluski, J. (1914) Le nord-ouest de l'Inde dans Ie Vinaya des Mula-Sarvastivadin. Journal asiatique 11(4), 493-568. Pope, A. U.,(1938) A Suroey of Persian Art, Oxford. Pugachenkova, G. A. (1966) Khalchayan, Tashkent. Pulleyblank, E. G., (1963) The Consonantal System of Old Chinese. Asia Major 9, 58-144, 206-265. Rahman, A. (1979) The Last Two Dynasties of the Sahis, Islamabad. Ramachandran, T. N. and Y. D. Sharma (1956) Archaeological Reconnaissance in Afghanistan: Preliminary Report of the Indian Archaeological Delegation, the Cyclostyle Edition, New Delhi. Rao, T. A. Gopinatha (1914) Elements of Hindu Iconography, 4 vols., Madras. Rapson, E. (1969) Indian Coins, reprint edition, Varanasi. Paruck, F. D. J. (1924) Sasanian Coins, Bombay. Ray, H. C. (1931) Dynastic History of Northern India 1, Calcutta. Rea, A. (1874) The Lanjadibba or Mound at Bhattiprol, Rapalli, Taluga. Indian Antiquary 3, 124. Rea, A. (1894) South Indian Buddhist Antiquities Including the Stiipa of Bhattipr'olu, Gudivada, and Ghantafala and Other' Ancient Sites in the Krishna District, Madras Presidency. Anhaeological Suroey of India, New Imperial Series 15, Madras. Remusat, A. (1829) Sur quelque peuples du Tibet et de la Boukharie, tire de l'ouvrage de Ma-touan-lin, et traduit du chinois. Nouveaux Melanges Asiatiques 1, Paris, 186-257. Robertson, D. S. (1943) Greek and Roman Architecture, Cambridge. Rosenfield, J. M. (1967) Dynastic Arts of the Kushans, Berkeley/Los Angeles. Rowland, B. (1942) Gandhara and Late Antique Art: the Buddha Image. American Journal of Archaeology 46,223-236. Rowland, B. (1967) The Art and Architectut'e of India, 3rd revised edition, Harmondsworth. Sachau, E. C. (1888) Alberuni's India. An Account of the Religion, Philosophy, Literatu1'e, Geography, Chronology, Astronomy, Customs, Laws and Astrology of India about AD 1030, 2 Vols., London. Sadakata, A. (1971) 'Echo-ogotenjikukokuden' Wayaku (A Japanese Version of Huichao's Wang Wu-Tianzhuguo zhuan). Tokaidaigaku Kiyo, Bungakubu (Bulletin of Tokai University, Faculty of Letters), 16,3-30 (in Japanese). Sander, L. (1968) Palaographisches zu den Sanskrithandschriften der Berliner Turfansammlung. Verzeichnis der orientalischen Handschriften in Deutschland, edited by W. Voigt, Supplement 8, Wiesbaden. Sarkar, H. D. (1960) Some Aspects of the Buddhist Monuments at Nagarjunakonda. Ancient 292 India: Bulletin of the Archaeological Survey of India 16 (1960),65-84. Sarkar, H. D. (1966) Studies in Early Buddhist Architecture of India, Delhi. Satyaprakash and P. R. S. Sharma (eds.) (1968) Baudhiiyanafulbasiitram with Sanskrit Commentary by Dwarka Nath Yajvan, English Translation and Critical Notes by G. Thibout. Dr. Ratna Kumari Publications Series 4, New Delhi. Scerrato, V. (1959) Summary Report on the Italian Archaeological Mission in Mghanistan - The First Two Excavation Campaigns at Ghazni 1957-1958. East and West 10 (1-4),23-55. Scerrato, V., (1960) A Short Note on Some Recently Discovered Buddhist Grottoes near Bamiyan, Mghanistan, East and West 11 (213),94-120. Scerrato, V. (1967) A Note on Some Pre-Muslim Antiquities of Jagatu. East and West 17(1-4), 11-24. Schlumberger, D. (1955) Le marbre Scorretti. Arts Asiatiques 2(2), 110-119. Schlumberger, D. and M. Le Berre (1964) Observations sur les remparts de Bactres. Monuments pre-islamiques d'Afghanistan. Memoires de la Delegation archiologique franfcaise en Afghanistan 19, Paris, 61-106. Schlumberger, D., M. Le Berre and G. Fussman (1983) Surkh Kotal en Bactriane I: Les temples, Memoires de la Delegation archiologique franfaise en Afghanistan 25, Paris. Shigematsu, T. (1919) Ehutalu Shuzoku Ko (On the Hephthalite Tribes). Shigakuzasshi 28(1), 20-50, 115-154. Shih, R. (1968) Biographies des moines eminents (Kao Seng Tchouan) de Houei-Kiao 1: Biographies des premier's traducteurs, Louvain. Shiratori, K. (1917) Keihin-koku Kou (On the Kingdom of Jibin). Toyogakuho 7(1). In the Shiratori Kurakichi Zenshu (the Collected Papers of Shiratori Kurakichi), Tokyo, 1970, 350-359. Shizutani, M. (1978) Indo Bukkyo Himei Mokuroku (A Collection of Buddhist Inscriptions in India and Pakistan), Kyoto. Siddiqui, A. D., (1935-36)ASIAR, 1935-36. Silvi Antonini, Ch. (1979) A Short Note on the Pottery from Tapa Sardar. South Asian Archaeology 1977, edited by M. Taddei, Naples, 847-864. Sims-Williams, N. (1983) Chotano-Sogdica. Bulletin of the School of Oriental and African Studies 46,40-51. Simms-Williams, N. (1997) New Light on Ancient Afghanistan: the Deciphe'l7JZent of Bactrian. An Inaugural Lecture Delivered on 1 Febnlary 1996, London. Sinha, B. P. and S. R. Roy (1969) Vaifiili Excavations 1958-1962, Patna!Allahabad. Sircar, D. C. (1963) The Early Medieval Inscriptions 1: Kabul Inscription of Shahi Khlii.gala. Epigraphia Indica 31 (1), 44-47. Sircar, D. C. (1963-64) Fragmentary Pillar Inscription from Arnaravati. Epigraphia Indica 35, 40-43. Sircar, D. C. (1965) Select Inscriptions 1, 2nd edition, Calcutta. Siudmak, J. (1989) Early Stone and Terracotta Sculpture. Art and Architecture ofAncient Kashmir, edited by Pratapaditya Pal, Bombay, 41-56. Sivaramamurti, C. (1950) Geographical and Chronological Factors in Indian Iconography. Ancient India 6,21-63. 293 Sivaramamurti, C. (1956) Amaravati Sculptures in the Madras Government Museum. Bulletin of the Madras Government Museum, New Series-General Section 4, Madras. Smith, V. A. (1901) The Jain stupa and Other Antiquities of Mathura, Delhi. (Reprint edition, Varanasi, 1969). Smith, V. A. (1906) Catalogue of the Coins in the Indian Museuim, Calcutta, Oxford. Smith, V. A (1908) The History of the City of Kanauj and of King Yaso-varman. Journal of the Royal Asiatic Society, 1908,765-793. Snodgrass, A. (1988) The Symbolism of the Stupa, Ithaca. Specht, E. (1883) Etudes sur l'Asie centrale d'apres les historiens chinois.Journal asiatique 8(2), 317-350. Spooner, D. B. (1908) Excavations at Shah-ji-ki Dheri. Annual Report of Archaeological Survey of India, Frontier Circle, 1907-08, Peshawar, 17-22. Spooner, D. B. (1912) Excavations at Shah-ji-ki-Dheri. Archaeological Survey of India, Annual Report 1908-09, Calcutta, 1912,38-59. Stavisky, B. (1972) Buddyiskyi Kultovii Tzentr v Starom Termeze 1965-1971, Moskva. Stein, M. A. (1900) Kalha1Ja's Riijatarangi1Ji. A Chronicle of the Kings ofKafmZT, 2 Vols., Westminster. Stein, M. A. (1905) White Huns and Kindred Tribes in the History of Indian North West Frontier. Indian Antiquary 34, 73-87. Stein, A. (1912) Excavations at Shah-ji-ki-Dheri. Annual Report of the Archaeological Survey of India, FTontier Cinle, 1911-12, Peshawar, 1-2. Stein, M. A. (192 8) Innermost Asia, 4 Vols., London. Stein, M. A. (1930) An Archaeological Tour in UppeT Swat and Adjacent Hill TTacts. Memoir of the Archaeological Survey of India 42, Calcutta. Stein, M A. (1973) A Contribution to the history of the Sahis of Kabul. East and West 23(1-2), 11-20. Taddei, M. (1962) An Ekamukhalinga from the N.W.F.P. and Some Connected Problems: A Study in Iconography and Style. East and West 13, 288-310. Taddei, M. (1964) A Linga-shaped Portable Sanctuary of the Sahi Period. East and West 15(1-2),24-25. Taddei, M. (1968) Tapa Sardar: First Preliminary Report. East and West 18, 108-124. Taddei, M. (1970) India, Archaeologia Mundi, Geneva. Taddei, M. (1972) Archaeological Mission in Afghanistan, IsMEO Activities. East and West 22(3-4), 379-384. Taddei, M. (1973) Una Kaumodaki-gada di arte Sahi nella raccolta dell'Istituto Universitario Orientale di Napoli. Arte Orientale in Italia 3 (Museo Nationale d'ATte Orientale,), Roma, 61-71. Taddei, M. (1974) A Note on the Parinirval).a Buddha at Tapa Sardar (Ghazni, Mghanistan). South Asian Archaeology 1973, Leiden, 111-115. Taddei, M. (1978) The Pre-Muslim Period: Settlement, Material Culture, Architecture and Art. In Allchin and Hammond 1978: 255-299. Taddei, M. (Paris) Le sanctuaire bouddhique de Tapa Sardar pres de Ghazni, Afghanistan. Taddei, M. and G. Verardi (1984) The Italian Archaeological Mission in Afghanistan: Brief Account of Excavation and Study. Studi di StoTia dell'arte in MemoTia di Mario Rotili, 294 Napoli, 41-70. Taddei, M. and G. Verardi (1985) Clay Stupas and Thrones at Tapa Sardar, Ghazni (Mghanistan). Zinbun: Memoirs of the Research Institute for Humanistic Studies (Kyoto University) 20, 17-32. Takata, T. (1992) Linguistic Aspects of the Wang Wu-Tianzhuguo zhuan and Characteristics of the Dunhuang Manuscript in Kuwayama 1992, 197-212. Tarzi, Z. (1976) Hadda a la lumiere des trois dernieres campagnes de fouilles de Tapa-e Shotor (1974-1976). Comptes rndus, Academie des Inscriptions et Belles-Lettres, juillet-octobre, 1976, 381-410. Tarzi, Z. (1977) L 'architecture et Ie decor rupestre des qrottes de Biimiyiin, 2 vols, Paris. Thewalt, V. (1983A) Rockcarvings and Inscriptions along the Indus: the Buddha Tradition. South Asian Archaeology 1983, edited by]. Schotsmans and M. Taddei, Naples, 779-800. Thewalt, V. (1983B) Jataka-Darstellungen bei Chilas und Shatial am Indus. Ethnologie und Geschichte: Festschriftfur KarlJetmar, edited by P. Snoy, Wiesbaden, 622-635. Toynbee,]. M. C. (1971) Death and Burial in the Roman WoTld, London. Tripathi, R. S. (1937) History of Kanauj to the Moslem, Benares. Tucci, G. (1958A) On a Sculpture of Gandhara. East and West 9(3), (also in Opera Minora II, Roma, 1971),227-230. Tucci, G. (1958B) Preliminary Report on an Archaeological Survey in Swat. East and West 9, 279-328. Uchida, G. (1961) An Introductory Inquiry to the Account of Pilgrimage in the Western Regions by Song Yun and Huijiao. Studies in History of Buddhism in honoUl· of the commem01'ation of the sixtieth birthday of Professor Tsukamoto, Kyoto, 113-116 (in Japanese). Uchida, G. (1967) A Restoration of Zhimeng's Youxing Waiguozhuan (The Account of Travelling Foreign Countries) with annotations. Kenkyu (Bulletin of the Faculty of Letters, Kobe University) 39, 19-34 (in Japanese). Uchida, G. (1972) Tokara Kokushi Ko (On the History of the Tuhuoluo Country). Tohogakkai S01·itsu Nijugoshunen Kinen Tohogaku Ronshu (Papers on East Asian Studies in commemoration of the Twentieth Anniversary of the Association of East Asian Studies), Tokyo. 91-110. Uchida, G. (1972) Gisho Saiikiden Genbun Koshaku (2). Toyoshi Kenkyu 31(3), 58-72. Uchida, G. (1973) Zui Haiku sen Saiiki Zuki Ibun Sanko (Reconstruction of theXiyu Tuji edited by Pei Ju). Fujuwara Kodo sensei Koki Kinen Shigaku Bukkyogaku Ronshu (Studies in the History and Buddhism in commemoration of the Seventieth Birthday of Prof. Fujiwara Kodo), Kyoto. 115-128. Uchida, G. (1975) Nishi Tokkutsu Shoseishi no Kenkyu (Studies in Early History of the West Turks). Kita Ajia shi Kenkyu (Studies in History of North Asia), Kyoto. van Gelder, ]. M. (tr.) (1963) Miinava Srautasiitra belonging to the Maitriiya1Jz sa'tJZhitii, Delhi (Reprint edition 1985). van Lohuizen-de Leeuw, ].-E. (1959) An Ancient Hindu Temple in Eastern Mghanistan. Oriental Art 5(2), 3-1l. Verardi, G. (1977) Notes on Afghan Archaeology II: Garesa Seated on Lion: - a New Sahi Marble. East and West 27(3-4),277-283. Veuve, S. (1974) La ceramique de Kohna Masdjid (Afghanistan): Travail d'etudes et de recherches 295 presente, en vue de l'obtention de la Mfritrise specialisee d'Histoire de l'Art et Archeologie, Vniversite de Bordeaux. Virgili, P. (1984) A Proposito del Mausoleo di Augusto: Baldassarre Peruzzi aveva ragione. Archeologia Laziale 4, Sesto incont1'o di studio del Comitato per l'anheologia laziale, Roma, 209-212. Vogel, J. Ph. (1929-1930) Prakrit Inscriptions from a Buddhist Site at Nagarjunikonda. Epigraphia Indica 20, 1-37. Vogel, J. Ph. (1947-48) Prakrit Inscriptions from Ghantsala. Epigraphia Indica 27, 1-4. Walker,]. (1941)A Catalogue of the Arab-Sassanian Coins, London. Wang, Y. (1984) A Record of Buddhist Monasteries in Lo-yang by Yang Hsuan-chih, Princeton. Warmington, E. H. (1974) The Commerce between the Roman Empire and India, 2nd edition, London and New York. Watters, Th. (1904-05) On Yuan-Chwang's Travel in India (AD 629-645), London. Williams, J. (1973) The Iconography of Khotanese Painting. East and West 23, 109-54. Wheeler, R. E. M. (1949) Roman-Buddhist Art: An Old Problem Restated. Antiquity 23,4-18. Wheeler, R. E. M. (1955) Rome beyond the Imperial Frontier, Harmondsworth. Wheeler, R. E. M. (1963) Gandhara Art: A Note on the Present Position. Le rayonnement des civilisations grecques et romaines sur les cultures periphiriques: Ie huitieme cong1'es international d'archiologie classique, Paris, 555-565. Wilson, H. H. (1841) Ariana Antiqua, London. Wink, A. (1990) AI-Hind The Making of the Indo-Islamic World 1: Early Medieval India and the Expansion of Islam, 7th-11th Centuries, LeidenlNew YorklKobenhaven/Koln, Witzel, M. (1980) Early Eastern Iran and the Atharvaveda. Persica 9,86-128. Wright, A. F. (1948) Fo-t'u-teng: A Biography, Harvard Journal ofAsiatic Studies 2,321-371. Wright, A. F. (1955) Biography and Hagiography: Hui-chiao's Lives of Eminent Monks. Silver Jubilee Volume of the Zimbun-kagaku Kenkyusyo, Kyoto. Yamada, R. (1955) Rengemen-gyo ni tsuite. Yamaguchi hakushi Kannki Kinen Indogaku Bukkyogaku Ronshu (Studies in Indology and Buddhism in commemoration of the Sixtieth Birthday of Dr. Yamaguchi), Kyoto, 110-123. Yamada, M. (1963) Mihirakura to sono Shuhen (Mihirakula and Relevant Problems). Bukkyoshigaku 11,44-58, 104-120. Yamada, M. (1972) Skandar Inscription of the Vma Mahesvara Image. Anhaeological Survey of Kyoto University in Afghanistan, Kyoto, 15-22. Yamada, M. (1990) HU1).a and Hephthal. Zinbun: Annals of the Institute for Reseanh in Humanities, Kyoto University, 23, 79-113. Yang, H. et al. (1984)The Hye Ch'o Diary, Memoir of the Pilgrimage to the Five Regions of India, Religions ofAsia Series 2, edited by L. R. Lancaster and]. L. Shastri, Berkeley and Seoul. Ya' qubi. Bibliotheca Geographicorum Arabicorum VII, Leiden. Yoshida, Y. (1993) Book Review on Kuwayama 1992. Oriento 36(1), 199-200. Yu, T. (1986) Yeda shi Yanjiu (A History of the Hephthalites), Jinan. Yule, H. (1873) Notes on Hwen Thsang's Account of the Principalities of Tokharistan.Journal of the Royal Asiatic Society 1873, 92-120. Zhou Z. (1963) Luoyang Qielanji Xiaoshi (The Luoyang Qielanji: Text and Commentary), Beijing, 296 183-227. Zurcher, E. (1968) The Yueh-chih and Kani~ka in the Chinese Sources. Papers on the Date of KaniJka, edited by A. L. Basham, Leiden, 346-390. Zwalf, W. (ed.) (1985) Buddhism: Art and Faith, London. Zou 1963. Luoyang Qielan ji Xiaoshi, Beijin. 297 ILL USTRATIONS 1 - Stupas on the main street at Sirkap, Taxila. N I ~ ~ '"0 o>-ij ...j ~. ~ Vl ::r' ~ S· crq & S S· (D ~ ('j a: (D Vl § 0- S o ~ ~ Vl M (D >-t (i)' :n 0100 N8 C>IIlItllUr.I ,'. 120 .""'< 700 toO 1100 UQO \600'I'AADt IpP,.".,..I." ,.",.UM •• ,I'/.ul flul.llt••f 21 tu",."rll!fI1ll1..-"sd,;a."fll,U h~1o "'MOIl,. ~q .......r mU\oIfU W ,<;.[ j/(m,,/urJI'J 1~1 ,oil ilPJltf)S d $_10'1 (lalil".I~' Z021,rJJ Altt •• t>t#lIMV,..IIIJ'fPOllllolJ;O",;a,n l lll,u",I>l'.IJi.,/ll",'''''''II/JtlJJ r~,.tlItJ"""UIi/,1· n.;,hi, .00D~ • • ,.tIII U~ lUll ,,,,-,,,. Ibl ~ _ o I..N ~ ~ "'0 0 ......... c;J ~ ::::1 0.. ::r ~ I-i ~ [J) ::r ~. ::::1 crq & (D ~ [J) q C? ~ ~. 0 ::::1 0 ......... s0 ::::1 I ... 7~ ,/ IrK '\ -_._¥» iV' ~ ~ [J) M (D I-i (D' :n SWAT. BOUNER GANIJHARA U])yjNA MERI])]ONAL ItllL>rnetre.J' ~), I 1 I f i : i « i ! f l i I i I 1 I l ~ ! ? I oI 3m ! i': i I , : i J"W-"_ ........:..I ~------------~--------------------------------.--~------------~ o 18m I 4 - Stupa IF, Sirkap (above) and Chir Tope C (below). o I ~ . -0 5 - Plans ofJolian showing three successive phases. r ~-- -- 6 - Plan of the stupa courts at Takht-i Bahi. f7 p:Jc e:y , · 0 0 00 0 0 (J 0 OOJ) 0 0 00 o ~Om 7 - Plan of tb e Stopa COUrt D T at areli. .~ 481) _____ ----~G&j ~~ t/ J~-------------------I7r) v~~ ----- 8 - Plan of Mekhkha Sanda monastery. 9 - Plan of Jamal Garhi monastery. 10 - Adoration of the patra: Lahore Museum (above) and Kabul Museum (below). ~- 9.Ter 10.Salihundam ..,/ ~ ~ 15.Alluru~14.K;dava1i ~ ~l1.Amaravati ~ ~.Adurru ~ ,r&1,.~..,>-I 17-30.Nagarjunakonda f<.12.Ghantasala 1 Bhattiprolu 13.Peddaganjam ... Stu pas with wheel-shaped pattern . .6. Other stupas referred to in the text. 11 - Map showing the distribution of stupas with and without the wheel-shaped pattern inside. 20m ~------~------~I 12 - Examples of the wheel-shaped structure: Augustus's mausoleum, Ghantasala, Kankali Tila,Dharmarajika (Taxila), Fil Khanah, Sanghol No. l. (From the above left clockwise.) -- I ~ ~ ...1"1 ~ [J J .;3'-' J-//~ 0..) d.;~tf(.... . 1 -- ~I~ I. . .~-.~~.:.· :)--~D "0 0. ~ ' 0 / : .. 0 :-. __ 0 l;'. __ _ a:~~~~~~~~ed within the square body :-s·········· °L rJ r·_···· . r : o·n._. o v- ········~¥-· :. .~-.. " .,: a circle with a 56 m ,;' . . touching the outermost of the extant radial walls ..-.-,.0.1-01' a circle with a 72 m diameter touching the inner sides of the small srupas located at the northwestern corner ~-~~~~------------ surface soil " - - - - 180 em ---",I --- )( v] 30"cm ....v.- - - - ... ____ -} ~ ll-lrrl radial wall IIIIII em (?) I 13- Plan of the Shah-ji-ki Dheri main stupa and its schematic cross-section of the stupa mound. 30cm 30m s a ] rt=n==:xun:~_- ,. -.----.-.-,. . ,...,.,. .}r·---~---~--------- ! ----1t~7Utl L___ - __ - __ - __ -_-_-_-_-_---'_~] 14 - Plans of stupa courts at Hadda: Lalma (above) and Tapa Shotor (below). 'Pilaster J ~ Mouldings of the plinth j - ' -_ _...J Tapa Kalan, Stupartl.141 -,-_ _....... 4 2 TapaKalan,No.33 6 lOOcm I ~ Tapa Shoto r Lalma I Coarse plaster Loose soil Floor of the later i3tage ............... ":::Ii Stupa No.13 Main stupa FlC)Or"o'f"'1;ne"'earr:re'i" stage Stupa No.7 ~ Ie ~ IB ~ IA o 20cm t:1::I::=r=r::::r:::±======::::::ll 15 - Mouldings from stupas at Hadda (above) and cross-sections showing the superimposed floors between the stupas at Lalma. " III! L-__~____ 1~~__~__~200~ (r~) ,,- / (' ,----' \" I ,,-- j ( \ ) ) I I ) ( \ ( \..) 16 - Routes of Song Yun in the Hindukush (above) and a map showing a route from Gandhara to the Punjab (below). till ?O_ _-"--_--J200km ' - - _ - ' -_ _I..... III) ' - - _ - ' -_ _100"-,_ _,,--_~200km 17 - Maps showing the routes taken by Buddhist monks before (above) and after (below) the mid-sixth century. 18 - The 55 m. and the 38m. Buddhas, Bamiyan, photogrammetrically surveyed in 1970 and 1976. 3SClCl 4SClCl~t 3000 40km ~----~----~----~--~I 19 - Distribution of monuments in the Kapisi-Kabul region. tX 3 5 o 30m I'---_~_---'-_---'I 20 - Monuments with round bastions: 1. Shotorak, 2 and 5. Begram, 3. Khair Khanah, 4. Tepe Maranjan, 6. Saka, 7. Bagh Gai, 8. Kohna Masjid, 9. Tapa Skandar. 1 2 3 4 5 6 7 8 9 10 11 12 21- Stamp-impressions in north Afghanistan. Durman Tepe (1-3, 11, 15, 16, 18-21,23), Balkh (4, 5, 7,8, 10, 12, 14, 17,22), Kara Tepe (9, 13), Shahri Bannu (6) and Chaqalaq Tepe (photographs 1-12). • ~~1? ~ __ ~~p~v '~w ~~~:-r--4 _-.,....-0C>_<=>_ 0 c:> _ ~I ~::;::? 7G~~ 12 ~\ltt~~k&1 C1? )[')['4 w)Kw 22 - Pottery types in the Kunduz region: Durman III (1-4), Durman IV (5), Chaqalaq Tepe II (6-14, 22-26), Durman III-IV (17-21), Chaqalaq Tepe III (15, 16,27). Scale 1:6. 23 - Plan of the excavated monuments at Tapa Skandar. Black dots and white circles indicate the find spots of stamp-impressed potsherds and pieces of marble Brahmanical sculptures respectively. • ~ ... ': r2 ". ~ ,,·41 37 ~, ~··t2 ~43 .~~u4""'4 :. 't~ .~ . '47 ~ 48 ,~I~. fi'))~; 49 ~50 "Il~l 24 - Medallion stamps. Floral designs. Begram Bazar (9, 32, 34, 39), Begram III (2, 23, 29, 33), Gul Dara in Logar (12), Kham Zargar (3), Saka (1, 6,10,21,24,26,27,50,53), Shotorak (7,14), Tapa Sardar (4,8,22,28,36,51), Tapa Skandar II (5, 11, 13, 15-17,25,30, 37, 38,41-45,47,49), Tepe Maranjan (18-20, 31, 35,40,46,48, 52), f.~a~ o· '4 I • • 3 l.~~~,;;' ';. ~ l\tj)~~ '''~~7 8 9 a 10 ~ ... • . . ~ ~ , ~.' , ". ' 17 16 29 30 32 ~3 rq 48 " '1 . .' .. • " ~.'~.I-' ~~';,:''''''<' ~' \ , ••::i. 50 I \ ' 51 25 - Medallion stamps. Animal designs. Begram Bazar (1, 8, 43, 46), Begram III «9, 17, 46), Ghundai Paisa (37), Gul Dara in Logar (3, 4, 10,45), Kham Zargar (6, 12), Shotorak (31), Tapa Sardar (26-28,36,42), Tapa Skandar II (2,7, 16, 19-23,25,29, 30, 33-35, 38, 39,44,47-51), Tepe Maranjan (5, 11, 13-15, 18, 24, 32,40,41). /] \' i ] 26 - Plan around Stupa 64 and miniature city wall 69, Tapa Sardar (Taddei and Verardi 1984) (above). Plan and cross-section showing the interrelation between the constructions H, F, G, J and c, Tapa Skandar (below). 30m '---------', 27 - A plan of Tepe Maranjan revised by Fussman (Fussrnan and Le Berre 1976) and modified by Kuwayama (above) and the layout of the precinct and monastic buildings, Gul Dara in Logar (Fussrnan and Le Berre 1976, PI. II). 20m I---_L.-...----l' 28 - General plan of Shotorak (Meunie 1942: 10) with its detailed plan of Court D (Meunie 1942: 21). D6 TK121 ~~~~~~~==~~==~~D4. O~!______~____~l~ 29 - Stupas with deep niches (D2, D3, D4 and D6 at Shotorak) and octagonal tiers (D4 at Shotorak and TK121 at Hadda). Not to scale for TK121. (Meunie 1942: figs. 4-7 and Barthoux 1933: fig. 89). - -"''''.. Buildings built in the time of the later shrines EARLY SHRI~T£ ~ ....t - - - Terrace ---+"::~ built in the time of the early shrine o, 30m f 30 - Plan and section of main buildings at Khair Khanah. (Drawn by Kuwayama after hackin 1936: Plans I and II). (' I - l ) / / c:....! 31 - Line drawings of the Uma-Mahesvara statue, Tapa Skandar. 32 - Marble Brahmanical images (1). Head of Durga from Qal'a Amir muhammad, Tagao (above left), head of Siva from Charsada (above right), Ekamukhalinga from Hund (below left) and head of Siva probably from Gardez. 33 - Marble Brahmanical images (2). Surya from Khair Khanah and Durga Mahishasuramardini from Gardez. 1 2 34 - Inscription of the Ganesa statue from Gardez. An ink rubbing.