The Global Connections of Gandhāran Art Edited by Wannaporn Rienjang Peter Stewart The Global Connections of Gandhāran Art Proceedings of the Third International Workshop of the Gandhāra Connections Project, University of Oxford, 18th-19th March, 2019 Edited by Wannaporn Rienjang Peter Stewart Archaeopress Archaeology Archaeopress Publishing Ltd Summertown Pavilion 18-24 Middle Way Summertown Oxford OX2 7LG www.archaeopress.com ISBN 978-1-78969-695-0 ISBN 978-1-78969-696-7 (e-Pdf) DOI: 10.32028/9781789696950 www.doi.org/10.32028/9781789696950 © Archaeopress and the individual authors 2020 Gandhāran ‘Atlas’ figure in schist; c. second century AD. Los Angeles County Museum of Art, inv. M.71.73.136 (Photo: LACMA Public Domain image.) This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. This book is available direct from Archaeopress or from our website www.archaeopress.com Contents Acknowledgements ���������������������������������������������������������������������������������������������������������������������������� iii Illustrations ���������������������������������������������������������������������������������������������������������������������������������������� iii Contributors ��������������������������������������������������������������������������������������������������������������������������������������� iv Preface ������������������������������������������������������������������������������������������������������������������������������������������������ vi Wannaporn Rienjang and Peter Stewart Part 1 Global perspectives Gandhāra perceptions: the orbit of Gandhāran studies �������������������������������������������������������������������� 1 Warwick Ball Part 2 The Graeco-Roman connection On the crossroads of disciplines: Tonio Hölscher’s theory of understanding Roman art images and its implications for the study of western influence(s) in Gandhāran art ���������������������������������29 Martina Stoye Roman sarcophagi and Gandhāran sculpture �����������������������������������������������������������������������������������50 Peter Stewart The transmission of Dionysiac imagery to Gandhāran Buddhist art �����������������������������������������������86 Tadashi Tanabe Part 3 Asian influences Buddha on the Rocks: Gandhāran connections through the Karakorum mountains��������������������105 M. E. J. J. van Aerde, A. D. L. Mohns, and A. G. Khan Buddhist temples in Tukhāristān and their relationships with Gandhāran traditions ����������������135 Shumpei Iwai More Gandhāra than Mathurā: substantial and persistent Gandhāran influences provincialized in the Buddhist material culture of Gujarat and beyond, c� AD 400-550����������������������������������������156 Ken Ishikawa Part 4 Gandhāra and China Cross-cultural Buddhist monastery ruins on the Silk Road and beyond: the layout and function of Buddhist monasteries reconsidered �������������������������������������������������������������������������������������������207 Joy Yi Lidu The sinicization and secularization of some Graeco-Buddhist gods in China ������������������������������234 Juping Yang Part 5 Epilogue De-fragmenting Gandhāran art: advancing analysis through digital imaging and visualization ���251 Ian Haynes, Iwan Peverett, Wannaporn Rienjang with contributions by Luca M. Olivieri i ii Acknowledgements The Gandhāra Connections project could not exist without the generosity and collegiality of scholars from different countries and institutional contexts. This is especially true of the proceedings of the project’s annual international workshops, which are produced on an extremely tight timetable and require from the contributors immense efforts of writing and sourcing of photographs. The editors are deeply grateful to all who have participated in this work and in the gathering of March 2019 which underpins the present volume. This is the third of five projected sets of conference proceedings, which are intended to reach the widest possible audience through parallel print and online, open access editions. As always, we should like to thank David Davison and his colleagues at Archaeopress for making this demanding format successful, particularly in the challenging context of the Covid-19 restrictions of spring 2020. We are also most grateful to Sarah Knights Johnson for her assistance in checking the text before publication. Above all, we should like to record our profound thanks to the supporters of Gandhāra Connections in its successive phases: the Bagri Foundation, Neil Kreitman, and Richard Beleson. Their generosity and enthusiasm have been fundamental to the whole initiative. Illustrations The contributing authors and editors have endeavoured to ensure that the images used in this open access, academic publication are available to be reproduced for academic purposes and, where required, that the necessary permissions have been obtained from copyright holders. If any omissions are brought to our notice we will be happy to include appropriate acknowledgement online or in any future edition. iii Contributors Marike van Aerde completed her doctoral degree from Leiden University, where she is now Researcher and Lecturer in Archaeology. Her research explores the archaeological evidence of trade routes connecting the Indian Subcontinent, East Africa, and the Mediterranean during the early first millennium AD. Her ongoing project, ‘Routes of Exchange, Roots of Connectivity’ focuses on the documentation and analysis of ancient rock carvings across the Karakorum mountain range in Pakistan (including ancient Buddhist rock art), the study of West Indian port sites, and the diversity of ceramics datasets from East Africa. Warwick Ball has excavated in Iran, Afghanistan, Jordan, Iraq, Ethiopia, and Libya. Positions have included Acting Director of the British Institute of Afghan Studies, Director of Excavations at the British School of Archaeology in Iraq and Founder Editor of the journal Afghanistan. His publications include Rome in the East: The Transformation of an Empire (2000), The Monuments of Afghanistan: History, Archaeology and Architecture (2008), Archaeological Gazetteer of Afghanistan (Revised and updated edition, 2019), and, as coauthor/editor, The Archaeology of Afghanistan (Revised and expanded edition, 2019). In 2020 he was awarded an honorary Doctorate of Letters by the University of St Andrews. Ian Haynes received his doctorate from Oxford University, and has taught at Oxford, London and Newcastle Universities. He is now Professor of Archaeology at Newcastle University. He has a long-standing research interest in the archaeology of cult practice and has published widely on the topic. Ian is also an active proponent of visualization/provocation as a research vehicle and of the application of digital technologies to the study of antiquity. He formed NU Digital Heritage to advance the availability of laser and white light scan images to researchers. In 2019 he won an ERC Advanced Grant to pursue a major research project in Rome, ‘Rome Transformed’. Ken Ishikawa is a DPhil candidate in Archaeology at Wolfson College, University of Oxford. His dissertation focuses on Buddhist rock-cut caves and monasticism in Early Historic Gujarat. Shumpei Iwai is Associate Professor of Archaeology at the Ryukoku University and Associate Curator of Central Asian Art at the Ryukoku Museum, Kyoto. His research interests focus on the archaeology of Central Asia during the first millennium AD. He has been actively involved in International Projects on Gandhāran and Central Asian Art including UNESCO funded projects for the preservation of the Buddhist Monastery of Ajina Tepa in Tajikistan and the restoration and preservation of the Buddhist site of Bamiyan. Abdul Ghani Khan completed his MA (Gold Medal) and MPhil degrees at Hazara University. His archaeological fieldwork and research focus on the study of ancient rock carvings, especially Buddhist carvings, from the Karakorum region of northern Pakistan, his home country. In collaboration with Marike van Aerde’s research project at Leiden University, he is currently planning new fieldwork in Gilgit-Baltistan to preserve and study as yet unrecorded rock carvings. Alexander Mohns conducted a comprehensive analysis of figurative Buddhist rock art from the Karakorum mountain range in Pakistan for his BA thesis, under the supervision of Marike van Aerde at Leiden University. His MA focuses on applied archaeological techniques in Sri Lanka, and he is collaborating with Van Aerde and Abdul Ghani Khan to prepare new fieldwork in Pakistan. Luca M. Olivieri is Director of the ISMEO/Ca’ Foscari Italian Archaeological Mission in Pakistan and Associate Professor of Archaeology and Cultures of Gandhāra and the Silk Roads at the Department of Asian and North African Studies of the Ca’ Foscari University of Venice. He has been working in Swat for more than thirty years. His main long-lasting project is the ongoing excavations at the urban site of Bazira/Barikot. His principal interests include excavation, rock-art, and heritage management methodologies. In 2017 he was awarded with the Sitara-i-Imtiaz of Pakistan for his three decades of archaeological work in Swat. iv Iwan Peverett is a Visiting Researcher at Newcastle University and Director of New Visions Heritage Ltd. His passion for developing the use of future technologies within the heritage sector has seen his work gain international recognition, published and featured in television programmes. Iwan has also been at the forefront of developing new work practices to assist in the integration of digital technologies and communication. He was a key member of the major urban model project team that received an award for technical innovation from BAFTA Scotland and the Merchant City Festival and he also conceived the initial concept as well as project-managing the delivery and digital elements of a project for the Samsung Digital Discovery Centre at the British Museum. This project was mentioned as an exemplar digital heritage project in the UK Government’s Culture White Paper on Cultural Heritage. Wannaporn Rienjang obtained her doctorate in Archaeology from University of Cambridge. She is now Lecturer in Archaeology, Museum and Heritage Studies at the Faculty of Sociology and Anthropology, Thammasat University and a project consultant for the Gandhāra Connections Project at the Classical Art Research Centre, Oxford. Her research focuses on the art and archaeology of Greater Gandhāra, Indian Ocean Trade and ancient working technologies of stone beads and vessels. Peter Stewart is Director of the Classical Art Research Centre and Associate Professor of Classical Art and Archaeology at the University of Oxford. He has worked widely in the field of ancient sculpture. His publications include Statues in Roman Society: Representation and Response (2003) and The Social History of Roman Art (2008). Much of his research concerns the relationship between ancient art traditions, particularly between Gandhāran art and Roman sculpture. Martina Stoye is Curator of South and South-East Asian Art at the Museum für Asiatische Kunst, Berlin. She studied Indology in Tübingen and Berlin, with specialization in the archaeology and history of art of South Asia. After graduating, she worked as a research associate for the history of Indian art at the Freie Universität Berlin as well as a project manager for several exhibitions in the context of the Festival of India in Germany. Much of her research concerns the relationship between Gandhāran Buddhavita scenes and Roman iconographies. In 2007/2008 she contributed to the Gandhāra exhibition at the Bundeskunsthalle Bonn. In her role as a museum curator she is in charge of a major Gandhāra collection. Tadashi Tanabe is Research Fellow of the Japan Society for the Promotion of Science and the International Research Institute for Advanced Buddhology, Soka University, and Visiting Researcher at the Institut für Indologie und Tibetologie, Ludwig Maximilians Universität, Munich. His research focuses on Greek and Roman influence on Gandhāran art and the iconographical study of Gandhāran Buddhist relief panels. His monograph, Courtesans in Gandhara (in Japanese), is expected to be published in 2020. Juping Yang is Professor of Ancient History and Director of the Center for Studies on Ancient Civilization on the Silk Road at Nankai University in Tianjin. He was a joint fellow of the Center for Hellenic Studies of Harvard University (CHS) and the German Archaeological Institute (DAI) in 2013-2014. He has published widely on the relationship between ancient China and the Graeco-Roman world and is currently director of the key project ‘Hellenistic Civilization and the Silk Road’, funded by the National Social Science Foundation of China. Joy Lidu Yi obtained her PhD at the University of Toronto in Canada and is Associate Professor of Chinese Art and Archaeology at the College of Communication, Architecture and The Arts at Florida International University, Miami. She is particularly interested in image-making, religious space and liturgy of Buddhist sanctuaries on the Silk Road. She is the author of Yungang: Art, History, Archaeology and Liturgy, the first ever comprehensive examination in any western language, which provides a social art history of Yungang: a fifth-century rock-cut court cave complex, UNESCO World Heritage site, and one of the greatest Buddhist monuments of all time. v (vi-vii): DOI: 10.32028/9781789696950-1 Preface Wannaporn Rienjang and Peter Stewart This volume presents the edited proceedings of the 2019 international workshop of the Gandhāra Connections Project at the University of Oxford’s Classical Art Research Centre. For this, the third of the project’s main annual conferences, we turned to a topic of fundamental importance for the understanding of Gandhāran art: the remarkable relationships of this regional tradition with the artistic currents of Asia and the Mediterranean. Naturally, these global connections constitute the point of departure for a project generated within a research centre that is principally dedicated to the art of the Graeco-Roman world. The first two workshops addressed the chronology and the regional geography of Gandhāran art – topics that underlie any wider exploration of the world of Gandhāran art (Rienjang & Stewart 2018; Rienjang & Stewart 2019). The geographical and cultural scope of the current volume is very much broader. We are not only concerned with the ‘influence’ of Greece and Rome on Gandhāra, much though this has dominated the subject in the last 150 years, but also with the role of other parts of Central and South Asia, and notably with the rich legacy of Gandhāra itself as it contributed to the development of art in China and beyond. Gandhāran art was created in a region that has been called the ‘Crossroads of Asia’, an area emblematic of cross-cultural connections and global links in antiquity. Of course, that reputation partly reflects our very modern concerns.1 In an era of globalization and global communication (never mind the imperial environment in which Gandhāran art was rediscovered), it has been tempting to invent this tradition in our own, modern image. Nevertheless, by any standards, the extent to which the artists and patrons of Buddhist Gandhāra borrowed from and influenced other cultures and lands is quite remarkable. In the last thirty years or so, a consensus has emerged that Gandhāran art drew upon a variety of artistic repertoires – Greek, Roman, Parthian/Persian, and Indian. A more sophisticated appreciation of the complexity of Gandhāran sculpture has emerged as a result; for example, there is greater interest in the complicated relationship of Gandhāran art with India, and even perhaps with the classical world through India. This awareness of multiple inter-cultural links, replacing a presumed binary, linear relationship with the classical world (with the ‘West’, as it is often misleadingly called), has been neatly summarized by Michael Falser, who describes it as a ‘new post-partition pluralistic tendency’, associating its origins particularly with the work of Mortimer Wheeler and Maurizio Taddei (Falser 2015; see also Stewart’s contribution to this volume). The vision of Gandhāran art as a confluence of styles and iconographies has been presented in some detail by perhaps the last major monograph on the subject, Lolita Nehru’s Origins of the Gandhāran Style: A Study of Contributory Influences, which was published in 1989. In Falser’s view, Taddei’s work contributed to the pluralization and diversification of Gandhāra’s position in the global history of art, releasing it from some of the ideological assumptions that hitherto prevailed. But importantly he advocates a more subtle and critical approach to Gandhāran art, which is perhaps impeded even by pluralistic cultural labelling. That is to say, applying a range of alternative labels – Greek, Parthian, Indian, etc. – can have its pitfalls as well. Falser makes a contrast between global history of art and world art history. Global art history is an approach that questions classifications and emphasizes transcultural processes. In contrast, traditional world art history is an additive container The modern filters that help to determine our vision of Gandhāra are part of the subject of our fourth workshop, originally planned for March 2020 (but delayed to 2021): ‘The Rediscovery and Reception of Gandhāran Art’. 1 vi Wannaporn rienjang and peter SteWart: preface which expands to include other cultures and approaches rather than thinking about the dynamic processes of interaction between them. To think in terms of global art history, inevitably means trying to evade from some of the constraints posed, even subconsciously, by cultural nomenclature. It is impossible to escape those names altogether and they will frequently be encountered in this volume. As far as possible, however, we need to keep in mind that they largely are modern inventions, impositions on the ancient material. The interconnected world of ancient art that is the matrix for the studies presented here constantly resists easy classification. The application of the term ‘globalization’ to the ancient world is controversial and in various respects anachronistic. Nevertheless, our modern, everyday experience of globalization, should at least help us to recognize the conceptual challenges that the global movements of culture pose. The ‘English’ language, like fine ‘china’, or the ubiquitously consumed Coca Cola, are rooted in their countries of origin but have a global life of their own. Just so, to think of Gandhāran art only in terms of East and West, or Greekness and Romanness, or hybridity (Graeco-Buddhist, Indo-Parthian) would be forcefully reductive. The purpose of this book is neither to explore ancient globalization nor global ancient art and it deals head-on with some of the traditional cultural relationships that have defined the past study of Gandhāran art, especially, in fact, the artists’ appropriation of the classical tradition and some of the very direct connections which appear to have existed across vast distances of the ancient world. None the less, the examples and arguments presented here should be viewed against the background of an ancient world whose complexity defies any straightforward ‘story of art’. References Falser M. 2015. The Graeco-Buddhist Style of Gandhara - a ‘Storia Ideologica’, or: How a Discourse Makes a Global History of Art. Journal of Art Historiography 13: 1-52. Nehru L. 1989. Origins of the Gandhāran Style: A Study of Contributory Influences. Delhi: Oxford University Press. Rienjang W. and Stewart P. (eds), 2018. Problems of Chronology in Gandhāran Art: Proceedings of the First International Workshop of the Gandhāra Connections Project, University of Oxford, 23rd-24th March, 2017. Oxford: Archaeopress. Rienjang W. and Stewart P. (eds), 2019. The Geography of Gandhāran Art: Proceedings of the Second International Workshop of the Gandhāra Connections Project, University of Oxford, 22nd-23rd March, 2018. Oxford: Archaeopress. vii Part 1 Global perspectives (1-25): DOI: 10.32028/9781789696950-2 Gandhāra perceptions: the orbit of Gandhāran studies Warwick Ball Introduction If one draws a north-south line through the exact middle of a map of Eurasia, the region of Gandhāra is seen to be exactly in the centre (Figure 1). Of course, such a device is entirely artificial and completely meaningless: in antiquity such a concept of ‘Eurasia’ (let alone its boundaries) would have been unknown. But it does at least illustrate how – and perhaps explain why – Gandhāra received, transmitted, and spread influences over much of that area. Gandhāran art exerts a fascination that is very different from other art categories: ancient Near Eastern art, for example, or modern art or even (dare one say) mainstream classical art. For the interest in Gandhāran art is that it comprised classical and classical-derived art that has been found thousands of miles away from where it should be (i.e. the Mediterranean) (Figure 2). Classical art in the Mediterranean requires no explanation; classical art in Gandhāra defies explanation. It is this more than anything else that has prompted misinterpretations, abuses, misappropriations – and its allure. Figure 1: Map of Eurasia illustrating the central position of Gandhāra (Image: author). 1 The Global ConneCTions of Gandhāran arT Figure 2: A Gandhāran relief in the Lahore Museum (top) compared to a Roman sarcophagus in the Antalya Museum (bottom) (Photos: author). 2 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies Myth and politics in British India The rediscovery of Gandhāran art coincided with the British expansion into north-western India from the mid-nineteenth century onwards: the opening of the Manikyala Stūpa and nearby stūpas as well as first forays into Taxila in the early parts of the century (Errington 2017a; 2017b), Charles Masson’s investigations in Afghanistan together with others by European officers in Sikh employment there and in the Sikh kingdom from the 1830s, the excavations at Takht-i Bāhī from the 1870s onwards, culminating in the excavations by John Marshall at Taxila after 1912 to name just some highlights (Errington & Curtis 2007). These followed on directly from the first British forays in the region (see below). The same period saw an upsurge in neo-classical architecture in Britain – or rather neo-Roman, as it identified far more closely with imperial Britain than Greek architecture did – reaching its height in the late Victorian period. This made Britain little different from other European imperial powers: neoclassical architecture characterized the capitals of other colonial empires as well, from St Petersburg to Paris (Amsterdam being a notable exception). But for the British there was an additional dimension: for the first time the British empire in India overlapped with that of Alexander of Macedon (Figure 3). And here was an art form that brought all those elements together: a neo-classical upsurge and the outermost expansion of two great empires. Never mind that the classical elements in Gandhāran art – as we now know – were not a direct result of Alexander’s campaigns; they provided a link back through time with one of the greatest empires and empire-builders of antiquity. No wonder the art proved so irresistible - and so collectable (Ball 2017). Of course, I do not suggest for one moment that this was a part of some deliberate thought out policy in the halls of power in Whitehall and Calcutta. But it does provide the background for why Gandhāran art became so collectable. And therein lies its tragedy: so much of Gandhāran art in museums (more so in private collections) are the result of, at best, uncontrolled excavation and stripping of sites or, at worst, looting (and faking). Hence, so much of it is divorced from its context, physical as well as social, and the problems of Gandhāran styles, provenance, and chronology are the biggest problems that continue to plague Gandhāran art to this day (problems which these workshops have done much to solve). Figure 3: Map showing the British expansion into the north-western Indian subcontinent overlapping with the campaign of Alexander of Macedon (Image: author, compiled from various sources). 3 The Global ConneCTions of Gandhāran arT A high point of Victorian Britain’s identification of its Indian Empire with the classical past was marked by the Delhi Durbar of 1877 when Queen Victoria was officially proclaimed Empress of India by the Viceroy, Lord Lytton. The ceremonial chosen for the occasion deliberately evoked Roman triumphs, reinforced by Lytton’s proclamation of Queen Victoria as, quite literally, Caesar: the Indian title chosen for her was Kaiser-i Hind (Cohn 1983: 201). The Lytton family themselves appropriately encapsulated British classicizing. His son-inlaw was the architect Sir Edwin Lutyens, responsible (in collaboration with Herbert Baker) for the design of New Delhi which is notable for its neo-classical elements. Lytton’s father was the eminent Victorian writer, statesman and man of letters, Sir Edward Bulwer Lytton, a noted Hellenophile who was even offered the crown of Greece (which he declined) in 1862 following the abdication of King Otto of Bavaria. Bulwer Lytton was awarded a full state funeral on his death in 1871 and buried in Westminster Abbey (see Lutyens 1979; Mitchell 2003; Oswyn Murray’s introduction to Bulwer Lytton 2004: 4). The identification with Alexander was a real one for the British, and both Pierre Briant and Jonathan Lee emphasized that European imperial expansion was seen as the continuation of a process begun by Alexander (Briant 2002: 901; Lee 1996: 74-8; also Briant in Haagsma et al. 2003). A glance in the British Library catalogue at titles about Alexander between 1700 and 1950 shows a sharp increase of titles between 1850 and 1940, corresponding exactly to the high point of the Indian Empire (Figure 4) (Ball 2012a, especially Appendix 2). The main increase was after the 1870s, rising to a peak in the 1890s, falling again in the 1910s and rising to another peak in the 1930s. It is possible to match this almost exactly with the British expansion into ‘Alexander territory’ in the north-west of India and related events in British imperial expansion: the annexation of the Figure 4: Graph showing the number of ‘Alexander the Great’ titles for every decade in the British Library Catalogue (compiled by Wendy Ball). 4 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies Punjab in 1849 following the Sikh Wars; the Delhi Durbar of 1877 and the proclamation of Queen Victoria as ‘Caesar’ of India; the 1880s and 1890s reflecting the main period of campaigning in the North-West Frontier when no less than eighteen campaigns were fought, such as the Hunza-Nagar Campaign of 1891, the Chitral Expedition of 1895, the Siege of Malakand, and the Tirah Campaign of 1897, all campaigns into regions of purported Alexandrine descendants. The next high point in the 1930s reflect a re-awakened interest in the Frontier from new campaigns there – following a lull when greater attention was focussed on the First World War – when ten campaigns were fought between 1919 and 1939, including the Third Afghan War of 1919. It fell to almost nil after 1950, corresponding to Britain’s departure from the scene following independence of India and Pakistan. The rediscovery of Gandhāran art reflected this pattern. Even after 1950, ‘looking for Alexander’ remained a strong archaeological motive for Britain. The establishment of the British Institute of Afghan Studies in Kabul in 1972 was due partly at least to pressure with this object: to answer the specific question of whether the site of Old Kandahar ‘overlies the Hellenistic city of Alexandria (or Alexandropolis) in Arachosia’ (Whitehouse 1978).1 It is significant too that members of the governing council of the Institute included such prominent classicists as John Boardman and Peter Fraser. Perhaps Gandhāran art received its ultimate aberration with British efforts to ‘see’ descendants of Alexander’s ‘Greek’ army as part of the art. Sir Olaf Caroe, perhaps the last of the remarkable soldier-scholaradministrators of the Frontier, in the 1940s recognized ‘Grecian blood’ in the young warriors of the Pathans, further emphasizing the point by comparing a photograph of a Pathan tribesman with an image of Alexander the Great alongside (Figure 5) (Caroe 1958: 44). Such wishful thinking was applied above all to the Kalash Kafirs of the Hindu Kush. Many aspects of their culture and ways of life were distinct, but most of all it was their physical appearance that gave rise to the myths of their Alexandrine descent (Bellew 1891; 143-6; Adamec 1985: 343-65; Vogelsang 2002: 32-5). It was their purported association with Alexander that formed the basis of Kipling’s famous story, ‘The Man who would be King’ published in 1888, and made into a film by John Huston in 1974 with the discovery of a lost ‘Greek’ city with Greek architectural styles which, while dissimilar to Gandhāran styles, were inspired by the discoveries of ‘Greek’ art in the region.2 Today, the fiction that the Kalash are descendants of Alexander’s army has become, if not exactly established fact, at least accepted wisdom that is related unquestioningly in numerous travel accounts and websites. The 1997 BBC Television series In the Footsteps of Alexander the Great, for example, featured a fireside meeting in Chitral with ‘the descendants of the last survivors of the Macedonian army which had burst across Asia like a meteor’ (Wood 1997: 8). A 2009 Daily Telegraph article refers to the Kalash as descendants of Alexander, as do numerous Greek and Macedonian websites, and many Kalash themselves now believe this.3 So widespread is this belief that ‘Hellenic Aid,’ a Greek NGO that is Greece’s official overseas development assistance programme, has established a Development Education Centre in Bomboret, the main Kalash community in the Kalash Valley of Chitral in Pakistan to encourage the Greek language and civilisation and to finance young Kalash to travel to the ‘Greek homeland’ for education.4 One need hardly say that modern scientific archaeological, linguistic, ethnographical and DNA researches on the Kalash have shown no traces of any connections with Alexander or the Greeks.5 It was also a response to the French having found their own ‘Alexander’ at Aï Khanoum: the momentum for a British presence in Afghanistan gained following an address to the British Academy by Paul Bernard in 1967, see Bernard 1967. 2 The lost ‘Greek’ city of Sikandargahr is solely John Huston’s, and does not appear in the Kipling story. Coincidentally, however, in the story itself Kipling does refer the two soldiers to John Wood’s Journey to the Source of the Oxus which actually describes a visit to a real lost Greek city, Aï Khanoum (Wood 1872: 259-60), although the Greek nature of the site was not recognized until the French excavations there beginning in the 1960s. Perhaps researchers in the cinema industry are often more thorough than we often credit them. 3 ‘Taliban targets descendants of Alexander the Great,’ Daily Telegraph, 21st September 2009. 4 See, for example, the online Express Tribune, 6th March 2020 with the headline ‘saving the lost sons of Alexander’ <https:// tribune.com.pk/story/728463/last-of-the-macedonians-kalasha-dur-saving-the-lost-sons-of-alexander> (last consulted 23rd April 2020. 5 See Bashir & Israr-ud-Din 1996; Young et al. 2000, with references. 1 5 The Global ConneCTions of Gandhāran arT Figure 5: Sir Olaf Caroe’s perception of physical likenesses between the modern people of the North-west Frontier region and the ancient Macedonians. (After Caroe 1958: 44). Did Gandhāran art influence later classical art in the west? The spread of art is always a two-way flow, and Gandhāran art transmitted as well as received. The transmission is usually viewed as towards the east: to India and to Inner Asia, China, and Japan. But did it go west as well? One variant of the Corinthian capital that is thought to have been a uniquely Gandhāran development from the first century onwards is of a figure – in a Gandhāran context the Buddha figure – framed by the acanthus leaves of the capital (Figure 6, a-c), perhaps inspired by the legend of Gautama seated under the Bodhi tree (e.g. Errington & Cribb 1992: 205). This treatment is little known in the classical art of the west, so that when it appears in late classical and early Gothic art of Italy and France (Figure 6, e), where it becomes a regular motif, it might be evidence of an east-west flow. Although plausible, it is difficult to substantiate. Such developments can in any case evolve quite independently. Versions of the form occur in the late Hellenistic ‘Palazzo delle Colonne’ at Ptolemais in Libya (although it might be later) as well as at Messene and Antioch-in-Pisidia (Lyttelton 1974: 56, fig. 15, pls. 63 and 67). Although employed differently, the use of busts in capitals might have been inspired by Egyptian art, such as at Hellenistic Dendera (Figure 7). In the National Museum of Syria in Damascus are several unlabelled basalt Corinthian capitals with a figure emerging from (rather than framed by, as in the Gandhāran examples) the acanthus (Figure 6, d). The capitals were unlabelled, but looked as if they might have originated in the Hauran in southern Syria, probably of the first or second century AD.6 Personal observation from a visit in 2013; the capitals were in the basement of the Museum, not on display, hence unlabelled; the Museum at the time was in any case undergoing renovation and was closed. 6 6 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies a b c e d Figure 6 a-e. Figures in Corinthian capitals from various locations. a: Surkh Kotal (Paris, Musée Guimet); b: Tepe Kalan, Haḍḍa (Paris, Musée Guimet); c: Lahore Museum; d: Damascus, National Museum; e: Abbey of Saint-Gilles, France. (Photos: author.) Another example comes from the architecture of the eastern provinces of the Roman Empire. In a seminal study in 1974, Margaret Lyttelton discussed an essentially eastern variant of Roman architecture dubbed ‘Roman baroque’ (Lyttelton 1974). Using the parallel of the conventional definition of baroque, as applied to the architecture of seventeenth- and eighteenth-century Europe, the ‘baroque’ style was applied to the more flamboyant forms of Graeco-Roman architecture, particularly in the areas of decoration, in contrast with the more strict, ‘Classic’ forms. While Roman baroque occurred throughout the empire – most famously at Hadrian’s Villa at Tivoli and in the fantasy houses depicted in the Pompeii 7 The Global ConneCTions of Gandhāran arT Figure 7. Capitals from the Temple of Hathor, Dendera, Egypt. (Photo: author.) frescos – it was more a feature of the eastern provinces, particularly Syria.7 Nearly all of the examples in Lyttelton’s study were drawn from here. Features of this style are the highly elaborate, flowing and curved façades, often conveying a sense of movement. Entablatures are alternatively recessed and breaking forward, as are the pediments. Decoration is liberal to the point of profligacy – indeed, such façades can appear almost organic. This is almost literally true for the most famous baroque façades, those at Petra, which are cut from the living bedrock. Niches and miniature pediments are used as additional embellishments, usually having no function other than decorative. Often, such niches are framed by pilasters and a pediment, a decorative feature known as the ‘Syrian niche,’ one of the most distinctive Syrian features of Roman baroque architecture (Figure 8). This is a niche set into a wall – often to house a statue – framed by a pair of engaged colonnettes supporting a miniature pediment. They were used to great – and elaborate – effect to adorn façades, most notably in the temenos walls of Baalbek and Palmyra. One authority includes the Syrian niche as part of a general category of ‘elaborated walls’ in Roman architecture. While eastern examples, such as Baalbek or Palmyra, are cited as the most imposing examples surviving, it is not known where they originated or why (MacDonald 1986: 203-7). Although MacDonald (1986: 220, 232) denies the emphasis on the East, viewing such buildings as the Temple of Venus at Baalbek or the Petra façades, which Lyttelton cites as typifying the baroque style, as not belonging to the baroque tradition. 7 8 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies Figure 8. So-called ‘Syrian niches’ in the Temple of Jupiter Heliopolitanus, Baalbek, Lebanon. (Photo: author.) Gandhāran architecture, although not included in Lyttelton’s study, is almost textbook classical baroque in these terms. In a discussion of 2000 an attempt was made to place the Roman elements of it into a broader eastern context generally and Gandhāran content specifically (Ball 2000: 433-45). The tradition of framed niches first occurs in the Buddhist architecture of Mauryan India, where they are used – like the Syrian niche – simply as a repeated, recurring decorative motif to adorn wall surfaces. They occur, for example, at the second-first century BC façades at Bharhut, Sanchi, and Bhaja, becoming characteristic of the cave architecture of Karli and Ajanta between the first and fifth centuries (e.g. Rowland 1977: figs. 15, 35, 38, 42, 46, 60; Michell 1989: 319-21; Allchin 1995: figs. 12.14, 12.15, 12.18). The importance of the framed niche in Gandhāran art is iconographical, deriving from the reverence for the Buddha image. This is one of the commonest features of Buddhist art, where the Buddha image typically occurs within a frame. The earliest occurrence is on the gold Bimaran reliquary, the current consensus for whose date is the second half of first century to first half of the second century AD for the deposition of the casket, based on a combination of stylistic analysis and the dates of the coins buried with it, although a precise date still remains controversial (Zwalf 1996: 348-50; Carter 1997; Cribb 2017; Cribb 2018a). We do know that by the time of Kaniṣka, images of the Buddha were beginning to appear on Kushan coins (Cribb 1984; 1999/2000), i.e. after AD 127/8 according to the most commonly accepted date (Cribb 2018b). Thereafter, it becomes characteristic in Gandhāran art, where Buddhist narrative reliefs or rows of seated Buddhas in niches are divided by Corinthian, or occasionally Persepolitan, columns. The best examples of the niche form in Gandhāran architecture are from the category defined as ‘inhabited caitya arches’ (Figure 9) (e.g. Zwalf 1996: pls. VI, 141, 143, 270-6). The tradition of using repeated rows of niches purely for decorative effect to relieve wall surfaces became a feature of Parthian and Sasanian architecture. It occurs, for example, on the Parthian palaces at Nisa, Ashhur and Qalʾa-i Yazdegird, as well as on Sasanian palaces at Firuzabad and Ctesiphon (e.g. 9 The Global ConneCTions of Gandhāran arT Figure 9. Stūpa at Jaulian, Taxila. (Photo: author.) Herrmann 1977: 34, 57, 67-72, 126-7). Such niches are a far cry from the ‘Syrian niches’ of Roman baroque or the repeated niches of Gandhāra, and any resemblance might well be illusory. But John Boardman has emphasized the underlying unity of the Achaemenid Empire over this entire region, that imposed a unity that long outlasted its collapse, and that the ‘remarkably durable arts of Persia’ not only survived but underpinned the formation and movement of other forms (Boardman 1994: 108).8 The importance of the Achaemenid Empire from the sixth to fourth centuries for internationalism and the transmission of artistic ideas cannot be emphasized too strongly. Against this background, the western spread of Gandhāran art to influence elements of Roman art is just as plausible as the eastern spread of classical art to influence Gandhāra. Gandhāra in Islamic literature Gandhāra occupies a rather unusual position in the early Islamic literature for the region. The spelling of the name has also been the source of considerable confusion. The Arabic form of the place name Gandhāra is ‫ قندھار‬which is transcribed as Q.ND.HAR. This, of course, is the same spelling as Qandahār in Afghanistan: indeed, the name Qandahar/Kandahar probably derived from Gandhāra (Achaemenid Gandara) rather than Alexander/Iskandar (from its ancient name of Alexandria Arachosia) as was earlier thought. It has been suggested that Qandahar received its name from Gandhāran refugees fleeing the Hephthalite invasions in 8 See also the remarks on the essential unity of this region by Curtis 1998-99. 10 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies the fifth century, but this is entirely speculative (Fischer 1967; Bernard 1974; Ball 1988). There was in any case a substantial Hephthalite presence at Kandahar, when it formed the centre of the Hephthalite (or perhaps Turk) dynasty of the Ratbils, which lasted until the tenth century (Ball et al. 2019: 364-5). An examination of the occurrence of the name in the early Islamic literature reveals not fewer than seven quite separate locations under the name ‫ قندھار‬/Q.ND.HAR/Gandhāra/Qandahar (Figure 10): Gandhāra itself; Qandahar in southern Afghanistan; Gandhāra, modern Qandabil, in Sind; Ghandhar in Gujerat; Kandhar in the Deccan; Qarajang corresponding to the modern province of Yunnan in China; and Qimar, modern Cambodia.9 This has led to some confusion in interpretation of the sources, especially the earliest mentions of the name. For example, al-Balādhuri’s account of Abbād ibn Ziyād’s raid on ‘al-Qunduhār’ in 53/673, the earliest Arab incursion into the region, is almost invariably interpreted as referring to Qandahar in Afghanistan (Le Strange 1905: 347; Minorsky 1942: 152, n.3; Bosworth 1968: 43; 1978: 356; Helms 1983: 242; Barthold 1984: 74); Caroe and Inaba are virtually alone in interpreting it as referring to Waihind in Gandhāra (Caroe 1958: 98-9; Inaba 2010). However, a careful analysis of the sources and the context of the raid shows that this – and other references to early Arab incursions – refer to Gandhāra/Qandabil in Sind; the earliest mentions of the name as definitely applying to the city of that name in Afghanistan are as late as the accounts of the Mongol invasions in the thirteenth century (Ball 1988: 118-20). Figure 10. Map showing the locations where the name ‘QNDHAR’ occurs in the Islamic sources. (Drawing: author.) Raverty (1892: 257, n. 200) recognizes three Gandhāras: Qandabil in Sind, Ghandhar in Gujerat, and Gandhāra; LongworthDames (1927) recognizes three: Gandhāra, Qandahar and Kandhar in the Deccan; Minorsky (1942: 152-3, n. 3) five: Qandahar, Ghandhar in Gujerat, Gandhāra, Qarajang, and Qimar/Cambodia; Mir Husain Shah (1982: 3) four: Qandahar, Gandhāra, Qandabil in Sind, and Kandhar in Deccan; Inaba (2010) three: Qandahar, Gandhāra, and Sind. Note that all are transcribed as Qandahar rather than Gandhāra. See also Ball 1988. 9 11 The Global ConneCTions of Gandhāran arT Why were there so many ‘Gandhāras’ in the early Islamic sources? Without analysing or quoting all references to the name,10 the following translations of the Ghaznavid and Seljuk poetic references give some indication: Farrukhi Sistani, d.499/1037-8. Your dwelling through its beautiful people is like paradise/ And your palace like Qandahar in respect of its idols [i.e., beauties]. Due to the handsome servants of your palace, every partition of your citadel is a Qandahar. Azraqi, d.ca.465/1072-3. Walk graciously towards me; for the sake of your gracious walk I threw up a thousand domes from Kashmir and Qandahar. Manuchihri, eleventh century. O mightly prince consume with sugar as long as desired(?) the lips of the beauty who is from Qandahar. Every moment the Padshah of Qayrawan sends loads of treasure I to him, with every breath the Shahriyar of Qandahar sends his tribute. You make 300,000 cities better than Qayrawan; you make 100,000 gardens better than Qandahar. The lovesick in spirit play with the Turks of Chigil and Qandahar. ‘Uthman Mukhtari, d.1149. Deal swiftly by fire and water as with the shrines of Kannauj and deal with the idols of the temples of Qandahar. Destroy root and branch the temples of the Ganges, then make for the idol temples of Qandahar.11 Leonard Harrow suggests that the only conclusions to be drawn from the use of the name ‘Gandhara/ Qandahar’ in these contexts are the stock images of distance, beauty, exoticism, and so forth.12 For example, Qayrawan in the extreme west is contrasted with Qandahar in the extreme east by Manuchihri, and ‘Turks’ are probably used as an image for barbaric beauty and cruelty, in much the same way as the nineteenth century image of the noble savage. The use of Qayrawan and Qandahar as images for distance is further amplified by the use of the place name ‘Chigil.’ Chigil lies in the Tien Shan Mountains, roughly corresponding to present-day Kyrgyzstan, and the use of its name for imagery occurs frequently in Persian poetry (e.g. Minorsky 1970: 297-300). These can be found in Ball 1988. For Farrukhi Sistani see Dabirsiyaqi 1968: 134 and 153; for Azraqi see Sa’id Nafisi: 706; for Manuchihri see Dabirsiyaqi 1968: 22, 32, 170 and 291; for Mukhtari see Homai 1962: 656 and 707. I am grateful to Leonard Harrow for selecting and translating these passages for me. See also Harrow 1973 and Tetley 2009. 12 Pers. comm. in Ball 1988: 124. I am very grateful to Leonard Harrow for earlier discussions on the poetic use of the name Qandahar by these poets. 10 11 12 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies It is also worth noting the association of ‘Turks’ with Chigil and Gandhāra: Chigil is in a Turk speaking area, but Gandhāra too was often associated in the Islamic mind with Turks: the term ‘Turk’ was often loosely applied to most non-Muslims in this part of the world by the sources with little regard to ethnic origins, but especially to the Hunnic rulers of Gandhāra and the succeeding Western Turk rulers. References to ‘idols’ and ‘temples’ by Farukhi Sistani and Manuchihri and to ‘domes’ by ‘Azraqi’ further suggest Buddhism (i.e., ‘domes’ = stūpas?). As a stock image for exotic beauty, distance and nonIslam (i.e. Buddhism) therefore, Gandhāra forms an entirely appropriate metaphor. Might this suggest a possible ‘reverse derivation’ of the name Gandhāra to the early Islamic mind as Qand Vihāra: ‘City/Fort of the [Buddhist] Monastery?’ Gandhāran art in Islamic architecture13 Deborah Klimburg-Salter has emphasized the massive increase of late Gandhāran art and architecture between the seventh and ninth centuries in Afghanistan, perhaps a result of the expansion of the Turk Empire in Central Asia, and that the Buddhist centres of Bamiyan, Kabul and Ghazni were part of a complex Buddhist communications network in eastern Afghanistan between the seventh and tenth centuries (Klimburg-Salter 2008: 131-159; 2010: 173-186). This has been confirmed by excavations at Tepe Sardar at Ghazni, as well as recent excavations of the Buddhist religious complexes of Tepe Narenj and Qol-i Tut in Kabul, which remained active probably as one of the main centres of Buddhism in Afghanistan at least until the ninth century (Paiman & Alram 2013; Taddei & Verardi 1978; Filigenzi 2009). The latter remained in use until the end of the eleventh century, well into the period of the Ghaznavids (Paiman 2018). Ghazna therefore was an important centre under the Western Turks, and its Buddhist Turk rulers overlapped with its Islamic Turk ones. In 961 Alptigin captured Ghazna on behalf of the Samanids, but it was retaken in 964 by the local non-Muslim ruler, Lawik, who held it for a year before it was retaken by Abu Ishaq, Alptigin’s successor (Bosworth 1963: 37-39). The independent Ghaznavid state centred on Ghazna was then firmly established by Sebuktigin in 977. The Turk newcomers from the Samanid state – a military elite, not a mass migration – would have merged seamlessly with the older Turk elites, well established there for centuries. Indeed, the long established Turk presence in eastern Afghanistan was probably one of the factors for Alptigin and his successors coming there in the first place; much of the Ghaznavid elite would have been Islamized local (former Buddhist) families of Ghazna. Ghazna, now a Muslim capital, had long been a religious centre for Turk princes. And, like the Turk princes in the past, the long tradition of patronage of the arts was continued by the newcomers. Thus, it is important to emphasize first, that non-Muslim rule in Ghazna overlapped with Muslim rule, and second, that the arrival of Alptigin marked ethnic Turk continuity at Ghazna, not a break. This continuity is reflected in the arts. First, a building type that reflects the Turkish tradition of patronage is the madrasa� Its institutional background in Seljuk Iran is well known. It has long been suggested that the Central Asian Buddhist vihāra or monastery inspired the madrasa, both as an institution for religious teaching and architecturally as a building type (Hillenbrand 1994: 174-5). Excavated examples of Central Asian Buddhist monastery complexes such as Ajina Tepe and Ak Beshim have been cited as architectural models. This might be so, but the Central Asian monasteries were generally abandoned in the eighth century or earlier. In eastern Afghanistan, however, there would have been many Buddhist monastery complexes still extant at the time of the arrival of the Ghaznavid Turks; such monastery complexes, furthermore, were firmly rooted in existing Turk architectural tradition, and the late flourishing of the Kabul and Tepe Sardar monasteries have been emphasized. In this context, the excavated building at Danestama in the central Hindu Kush might be viewed as a link: compare, for example, its plan with that of Guldarra (Figure 11) (Le Berre 1970; Ball 2019, site 231). 13 This is drawn mainly from Ball, forthcoming. 13 The Global ConneCTions of Gandhāran arT Figure 11. Plan of the Ghurid madrasa at Danestama in Afghanistan (left; after Le Berre 1970, from Ball 2019) compared to that of the Buddhist monastery of Guldarra (right; after Fussman & Le Berre 1976, from Ball 2019). At first sight, the decoration of Ghaznavid and Ghurid architecture appears to mark a complete break with the Graeco-Buddhist architectural decoration of the past: Islamic architecture with its emphasis on abstract patterns and calligraphy, the Buddhist with its emphasis on figurative art, the Buddha image in particular. But were the two styles so far apart? Common to both was a horror vacui, the compulsion to cover – almost to swamp – every available space of a wall surface with decoration. Ghaznavid and Ghurid buildings are particularly noted for their surface decoration, ‘whose ornateness consistently exceeds anything known in the rest of Iran in the Saljuq period … architectural decoration to a pitch of technical mastery never to be exceeded’ (Hillenbrand 2000: 129). For it must be remembered that the newly arrived Ghaznavid patrons would have employed craftsmen – or at least a craft tradition – who were used to decorating Buddhist buildings. Such craftsmen could not, of course, cover the new buildings with human images (although see below), but they would have found decorating wall surfaces to be not only natural, but irresistible. Such richness applies only to the general spirit of decoration. Are there specific decorative elements in Ghaznavid buildings that can be traced back to Buddhist architecture? A common element throughout Buddhist architecture are various forms of scroll patterns, vegetal motifs, and garlanding: compare, for example, the many stucco examples from Tepe Sardar14 and elsewhere in Buddhist Afghanistan15 with the marble friezes from the walls of the Palace of Mas‘ud16 or the painted and stucco decorative friezes from Lashkari Bazar (Figure 12) (e.g. Schlumberger & Sourdel-Thomine 1978: planches 39-40, 61d-e). Such decorative elements, of course, are common elsewhere in Islamic architecture, and can ultimately 14 E.g. the many examples illustrated on the IsIAO website <http://ghazni.bradypus.net/buddhist-cornices> (last accessed 5th March 2020) such as Inv. no. TS00945. 15 Such as Shotorak, e.g. Tissot 2006: 326, K.p.Sho.854.60. 16 E.g. Tissot 2006: 478-9, especially Isl.p. Gh. 1337.5 or Isl.p. Gh. 1339.7. 14 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies b a Figure 12. a: Ghaznavid decorative elements: top two Lashkari Bazar (after Schlumberger & Sourdel-Thomine 1978), bottom two Ghazni (Copyright: IsIAO archives Ghazni/Tapa Sardar Project 2014). b: Buddhist decorative elements: top two Tepe Sardar (Copyright: IsIAO archives Ghazni/Tapa Sardar Project 2014), bottom Shotorak. be traced back to classical forms (Allen 1988),17 as can the same elements in Buddhist architecture. But it is difficult to imagine that craftsmen working on the palace at Ghazni using second-hand copy-books while ignoring first-hand examples on their doorstep. Some closer examples can also be cited. A particular element in Ghaznavid architecture are multi-lobed blind arches, either singly or in series, that appear in relief carvings from Bust and especially Ghazni. At the latter the motif occurs on both the marble orthostats on which the monumental inscription is written and the many individual fragments on gravestones and other pieces (Figure 13, above) (e.g. Bombaci 1966: figs. 10-21, 137-42).18 An origin for this motif must surely be the multi-lobed blind arches that typically frame seated Buddha images and are an almost universal feature of Gandhāran architecture, particularly at Tepe Sardar (Figure 13, below) (Taddei & Verardi 1978: figs. 134-6). It is worth noting too that there is at least one example at the Ghazni palace that is a multi-lobed arch framing a human figure (Figure 13, above right), a figure furthermore very similar to the Lashkari Bazar paintings of ‘courtiers’ which in turn have been related to Buddhist styles (discussed below). Tepe Sardar was one of the most elaborately decorated stūpas in Afghanistan and one of the latest, remaining in use as late as the ninth century. In other words, much of it would still have been visible when construction of the Ghaznavid capital nearby commenced, and a monument of this size and richness would not have gone unobserved by the craftsmen. Repeated blind arches and its use in Gandhāran architecture must surely be relevant. Here, it took two forms: either the repeated multi-lobed arches framing seated Buddha images discussed above, or repeated arches framing scenes of Buddhist iconography separated by IndoCorinthian or Persepolitan columns and topped by a pseudo-classical entablature (Figure 9).19 In finding Especially Chapter 1 and figs. 17, 18, 21-26. See also Allen 1988: fig. 67. 19 The motif is so common in Buddhist architecture that it would take too much space to give an exhaustive catalogue here, but the many examples illustrated in Fussman 2008 should suffice. 17 18 15 The Global ConneCTions of Gandhāran arT Figure 13. Multi-lobed blind arches. Top left: Palace of Mas‘ud, Ghazni (Copyright: IsIAO archives Ghazni/Tapa Sardar Project 2014); top right: framed figure Ghazni (Copyright: IsIAO archives Ghazni/Tapa Sardar Project 2014); bottom: Tepe Sardar (top, after Taddei & Verardi 1978). the motif used almost continuously in Afghanistan from Gandhāran through to Islamic architecture it is difficult to escape the conclusion that the one has evolved from the other.20 Very little survives of Ghaznavid painting, the lines of ‘courtiers’ in the royal audience hall of the South Palace at Lashkari Bazar being almost the sole complete examples. These have been convincingly related to the late Gandhāran art of Afghanistan and the seventh century Buddhist cave paintings at Kucha and 20 On the other hand, the two occurrences might be entirely unrelated: ‘classic’ repeated multi-lobed blind arches also occur, for example, at the twelfth century Cistercian abbey of Melrose in the Scottish Borders (personal observation). 16 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies Figure 14. Plans of Ghaznavid and Ghurid stellate minarets. (Drawings: author.) Qizil in Xinjiang (Schlumberger & Sourdel-Thomine 1978: 61-5),21 particularly those that depict donors at the latter (e.g. Bussagli 1978: 80). Considerable discussion has revolved around the origin of the distinctive stellate plans of Ghaznavid minarets (Figure 14) (Hutt 1977; Pinder-Wilson 2001; Flood 2002). Plausible prototypes occur at Ghazni itself. These are the stellate plans of some votive stūpas and fire altars at Tepe Sardar (Figure 15) (Filigenzi 2009: fig. 6). Votive stūpas could be square, circular, or stellate in plan. Strictly speaking, many of the votive stūpas were ‘stepped square’ in plan, but ‘stellate’ in overall appearance. But many votive stūpas used a fully fledged stellate plan (Figure 15, bottom left) (Ball et al. 2019: 411, fig. 6.68). The survival of Gandhāran architecture into the Islamic period and the establishment of Turk dynasties in eastern Afghanistan from the fifth century through to the Mongol conquest, princes who furthermore patronized both the Buddhist and the Islamic arts, must be regarded as a continuum, and not as two separate histories. 21 See also in general Bussagli 1978. 17 The Global ConneCTions of Gandhāran arT Figure 15. Stellate fire altar (top left) and votive stūpas at Tepe Sardar. (Copyright: IsIAO archives Ghazni/Tapa Sardar Project 2014.) Gandhāra to China – and back The debt of the Buddhist art of China and Japan to Gandhāran art has long been acknowledged (e.g. Motamedi 1975; Yang and Yi in the present volume). But was there a reverse flow, of the Buddhist art of China returning to influence, in turn, late Gandhāran art? Once again, the Turks are key players and Bamiyan is a key site. Some of the paintings of the Bamiyan Buddhas have recently been dated by C14 to the middle of the seventh century (Shoten 2006), suggesting that the main period of construction was some time in the first half of that century. Hence, ‘As one of the possible directions for further study, one might consider that only the Yabghu of the Western Turks, the nominal overlord of northwest [sic; northeast] 18 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies Afghanistan as far as Kapiśā, had the means to construct such a monumental complex at the beginning of the seventh century’ (Klimburg-Salter 1989: 134). This coincides with a time which saw a late flowering of Gandhāran art which continued into the early centuries of Islam. Bamiyan represents perhaps the greatest high point of this late flowering. But in the architecture and art generally of Bamiyan there exists a paradox. For Bamiyan was only a minor principality in the Western Turk federation, the regional capital of which was at Qunduz in north-eastern Afghanistan, where no monuments on such a scale have been discovered; the main capitals were at Chach (Tashkent) and Suyab (Bishkek). Other principalities were located at Kāpiśā (Kabul) and Ghazna, where there are also important Buddhist remains as we have noted, albeit not on the scale of Bamiyan. The background is the establishment of the Turk empire on the inner Asian steppes in the middle of the sixth century covering a vast area stretching from the borders of Manchuria in the east to the Talas River in the west.22 It soon split into two ruled by the two brothers Muhan (553-72), who ruled the eastern half of the empire and Ishtemi (553-?), who ruled the western half. By about 555 the empire of the Western Turks probably extended as far as the Aral Sea and possibly even as far as the Volga. Ishtemi was succeeded as emperor of the Western Turks by his son Tardu, who ruled until 603. The kagan of the Eastern Turks, Taspar (572-81), was converted to Buddhism by a Chinese monk and Buddhism then spread among the Turks. During the time of Taspar, the Gandhāran monk Jinagupta translated many of the sacred Buddhist texts into both Turkish and Chinese at the great monastery at Kucha in the Tarim Basin. The kagans were regarded as semi-divine. The Western Turk Empire absorbed much of the Central Asian territories and expanded rapidly, particularly with the defeat of the former Hephthalite Empire centred in northern Bactria. By 616/7 a Turk army had even penetrated deep into Sasanian Iran as far as Rayy, just south of Tehran. In 619 the most powerful of the Western Turk kagans succeeded to the throne, Tong Yabgu Kagan. Under Tong Yabgu the empire reached its greatest extent, incorporating the Tarim Basin, Ferghana, Bactria, and parts of Afghanistan and northern Pakistan, with Tong Yabghu himself advancing as far as the Indus in 625. This resulted in a shift of gravity for the empire, with Tukharistan becoming a centre. Tong Yabghu was furthermore known to favour Buddhism, and there was a consequent upsurge of Buddhist art in the Hindu Kush area. The construction of the great Buddhas of Bamiyan was a part of the Turk patronage of Buddhism. After the death of Tong Yabghu in 630 the kaganate declined, with its final collapse following a defeat by the rising Tang Empire. There is literary evidence for a dynastic shrine among the Eastern Turks (Klimburg-Salter 1989: 136), and Deborah Klimburg-Salter has suggested that Bamiyan may have been a dynastic centre for the Western Turks, citing in particular the paintings of the ceremonial scene and the line of donors in the thirtyeight metre Buddha.23 This is much the same way as Surkh Kotal, similarly not a capital, functioned for the Kushan kings several centuries previously, symbolized by a similarly iconic standing statue, so perhaps a continuation of such a tradition of dynastic centres? The tradition of dynastic centres in the region was a strong one: Rabatak was another possible Kushan dynastic centre (Rosenfield 1967; Ball 2019: sites 944, 1223), as was Khalchayan in Uzbekistan (Pugachenkova 1965), and the continuity of Kushan traditions to late antiquity has been emphasized (Ball et al. 2019: 379-84). The carving of the Sasanian rock relief of Rag-i Bibi not far from Surkh Kotal and the identification of the surrounding region of ‘Kadagstan’ as a possible special enclave for the Sasanian kings might be a part of this dynastic continuity. 22 23 For summaries see Sinor 1990 and Ball 2012b: chapter 3. Notably by Klimburg-Salter 1989: 134-6. 19 The Global ConneCTions of Gandhāran arT Figures 16-17: The colossal Buddhas of Yungang. (Photos: author.) In this respect the concept behind the colossal Buddhas at Bamiyan resembles that of the earlier Northern Wei colossal Buddhas at Yungang in China. Before the emergence of the Turks as a major steppe power, an eastern Turk tribe, the Tabgach (or Tabgaj; Chinese Toba), had spread across northern China and Inner Mongolia in the late third century AD. They entered China in the fourth century AD as a part of the Xianpei confederacy, a loose alliance of steppe tribes who emerged after the collapse of the Xiongnu. The Toba or Tabgach Turks24 emerged as the leaders of this confederacy and founded the Northern Wei Dynasty of China in 386 centred on Pingcheng, modern Datong, which they made their capital in 398. The Northern Wei rulers were enthusiastic supporters of Buddhism – the 24 There is some dispute as to whether the Toba or T’o-pa were really Turk. However, see Sanping 2005, who confirms their Turk language. Tabgach also remained a Turkish name, such as in the Karakhanid ruler Tabghach Bughra Khan in 1069 20 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies first Chinese dynasty to be so – and Emperor Toba Dao was the first to elevate Buddhism to an official state level. In particular, the Toba kings embraced the peculiarly Mahāyāna element in Buddhism where the secular ruler is invested with the authority of the Buddha. This element took spectacular form in the Buddhist caves constructed throughout the fifth century by the Northern Wei dynasty at Yungang outside their capital at Pingcheng (Figure 16), as well as at the Longmen caves near their later capital of Loyang, where each Toba king erected a giant statue of the Buddha representing the Buddhist authority invested in himself: each giant Buddha statue, in effect, represented the authority of both the Buddha and the emperor. Joy Lidu Yi, in her seminal study of the caves, emphasizes that (Yi 2018: 48 and 72),25 the creation of Yungang was a highly complicated process. The first-phase cave temples functioned in practical terms as memorials to the Northern Wei court. Worshipping a Buddha in the five Tanyao caves was much the same as demonstrating respect and loyalty to the emperors... The five Tanyao caves promote the Buddhist ideal that the emperor is the living Tathāgata … They best exemplify the close association between imperial power, imagemaking, politics, and Buddhist rock-cut cave art under the patronage of the imperial family. The most striking feature of these cave temples is the gigantic size of the image inside each cave temple. ... The scale of the caves and the images is unprecedented and has never subsequently been matched’ This was the first time that giganticism – colossal Buddha statues – had been introduced into Buddhist art. The association of the emperor as Buddha was recognized long ago by Wolfram Eberhard, who further regarded the expression of this in cave architecture as entirely logical given the Toba Turk supposed origins in a mythical cave in the Ötükän Forest (Eberhard 1952: 145-6).26 The decision by Toba Dao to combine secular and religious authority in the person of the ruler and translate this into massive, visible architectural statements is a precedent that surely would not have been lost on the later Buddhist Turk rulers of Bamiyan. Combining secular and religious authority in the person of the ruler expressed by a gigantic Buddha statue was continued by the Emperor Shomu of Japan. In 741 Shomu ordered the construction of the Todai-ji or ‘Great Western Temple’ in Nara, the centrepiece of which was a gigantic bronze Buddha, completed in 752.27 Shomu’s motivation and its expression was identical to that of Toba Dao, three and a half centuries previously and the Western Turk kagans a century and a half previously, all three adopting the same element in Mahāyāna Buddhism. That the rulers of Yungang, Bamiyan, and Nara expressed their authority in terms of giganticism is no surprise. Giganticism has been used to express power ever since the Great Pyramids. The colossal Buddhas of Yungang are mainly seated Buddhas, a common iconographical stance in Buddhism. That erected by Toba Dao, the first Northern Wei Emperor to elevate Buddhism to state level, however is a See also Yi in this volume The mythical cave was ‘discovered’ in the far north-east of present day Inner Mongolia in 443 by an expedition sent by Tuoba Dao, who left an inscription at the entrance. See Kessler 1993: 70. 27 Deanna MacDonald, ‘Todai-ji’ <https://www.khanacademy.org/humanities/ap-art-history/south-east-se-asia/japan-art/a/ todai-ji> (last accessed 5th March 2020). 25 26 21 The Global ConneCTions of Gandhāran arT standing Buddha (Figure 17), and it is this image that is used in the great Buddhas of Bamiyan as well as at Kakrak. Whether the construction of the 53 metre Buddha can be attributed to Ishtemi or Tong Yabghu – or any of the other kagans – is impossible at the moment to say. Conclusion The proceedings of the second Gandhāra Connections workshop in 2018 were entitled The Geography of Gandhāran Art (Rienjang & Stewart 2019). The contributors to that workshop and its resultant publication were concerned with provenance and variations of objects and styles within the geographical region of Gandhāra. This paper has attempted to explore the ‘geography’ of Gandhāran art outwith the region of Gandhāra: expanding its ‘geography’ to cover much of the known ancient world. Of course much of this might not be real, but whether or not such spreads and influences were real is hardly the point: it is the various, occasionally strange, ways that Gandhāran art has been received, perceived, interpreted, misinterpreted, and passed on that are discussed here. Hovering in the background to these perceptions is perhaps the most extraordinary journey of all: that an art form originating in the Mediterranean came to influence the art of quite unconnected kingdoms thousands of miles to the east, and beyond. In accepting that central fact of Gandhāran art, even its unlikeliest descendants – from a face peering out of a Corinthian column in a Romanesque French church to rows of arches decorating an early Islamic palace to a giant Buddha in Japan – become less unlikely. References Adamec L. W. (ed), 1985. Historical and Political Gazetteer of Afghanistan.Vol. 6. Kabul and Southeastern Afghanistan. Graz: Akademische Druck- u. Verlagsanstalt. Allchin F. R. 1995. The Archaeology of Early Historic South Asia: The Emergence of Cities and States. Cambridge: Cambridge University Press. Allen T. 1988. Five Essays on Islamic Art. Solipsist Press. Ball W. 1988. The Seven Qandahars: The Name Q.ND.HAR in the Islamic Sources. South Asian Studies 4: 115-142. Ball W. 2000. Rome in the East: The Transformation of an Empire. London; New York: Routledge (2nd edition 2016). Ball W. 2012a. Some Talk of Alexander: Myth and Politics in the North-West Frontier of British India. Pages 127-158 in R. Stoneman, K. Erikson, and I. Netton (eds), The Alexander Romance in Persia and the East. Groningen: Groningen University Library. Ball W. 2012b. Sultans of Rome. The Turkish World Expansion. London: East & West Publishing. Ball W. 2017. The Sasanian Empire and the East: A Summary of the Evidence and its Implications for Rome. Pages 151-178 in E. Sauer (ed), Sasanian Persia: Between Rome and the Steppes of Eurasia. Edinburgh: EUP. Ball W. 2019. Archaeological Gazetteer of Afghanistan. Revised Edition. Oxford: Oxford University Press. Ball W. forthcoming. Buddhist Survivals in the Architecture of Afghanistan 1000-1250. In R. Hillenbrand (ed), The Architecture of the Greater Iranian World, 1000 - 1250. Edinburgh. Ball W. Bordeaux O., MacDowall D., Sims-Williams N., and Taddei M. 2019. From the Kushans to the Shahis. Pages 344-459 in The Archaeology of Afghanistan. From Earliest Times to the Timurid Period. Original edition ed. Raymond Allchin and Norman Hammond, revised and updated by Warwick Ball with Norman Hammond. Edinburgh. Barthold W., translated by S. Soucek 1984. An Historical Geography of Iran. Princeton: Princeton University Press. Bashir E. and Israr-ud-Din (eds), 1996. Proceedings of the second international Hindu Kush cultural conference. Karachi. Belenitsky A. 1968. Central Asia. New York: World Publishers. 22 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies Bellew H. W. 1891. An Enquiry into the Ethnography of Afghanistan (repr. Graz, 1973). Bernard P. 1967. Aï Khanum on the Oxus: A Hellenistic City in Central Asia. Proceedings of the British Academy 53: 71-95. Bernard P. 1974. Les noms anciens de Qandahar. Studia Iranica 3: 171-185. Reprinted in Afghanistan 28/1, 1975: 1-15, and in Afghanistan 33/1, 1980: 49-63. Boardman J. 1994. The Diffusion of Classical Art in Antiquity. London: Thames and Hudson. Bombaci A. 1966. The kūfic Inscription in Persian Verses in the Court of the Royal Palace of Mas‘ūd III at Ghazni. Rome: IsMEO. Bosworth C. E. 1963. The Ghaznavids: Their Empire in Afghanistan and Eastern India 994-1040. Edinburgh: University Press. Bosworth C. E. 1968. Sīstān Under the Arabs: From the Islamic Conquest to the Rise of the Saffārids (30-250/651684 A.D.). Rome: IsMEO. Bosworth C. E. 1978. Kandahār. Encyclopaedia Iranica 4: 535-8. Briant P. 2002. From Cyrus to Alexander. A History of the Persian Empire. Winona Lake: Eisenbraun. Bussagli M. 1978. Central Asian Painting from Afghanistan to Sinkiang. Geneva: Rizzoli. Caroe O. 1958. The Pathans. 500 B.C. – A.D. 1957. London: Macmillan. Carter M. L. 1997. A Reappraisal of the Bīmarān Reliquary. Pages 165-209 in R. Allchin, B. Allchin, N. Kreitman and E. Errington (eds), Gandharan Art in Context. East-West Exchanges at the Crossroads of Asia. Delhi: Regency Publications. Cohn B. S. 1983. Representing Authority in Victorian India. Pages 165-209 in E. Hobsbawm and T. Ranger (eds), The Invention of Tradition. Cambridge: Cambridge University Press. Cribb J. 1984. The Origin of the Buddha Image: The Numismatic Evidence. Pages 231-34 in B. Allchin (ed), South Asian Archaeology 1981. Cambridge: Cambridge University Press. Cribb J. 1999/2000. Kanishka I’s Buddha Image Coins Revisited. Silk Road Art and Archaeology 6: 151-189. Cribb J. 2017. Dating the Bimaran Casket: Its Conflicted Role in the Chronology of Gandharan Art. Gandhāran Studies 10: 57-91. Cribb, J. 2018a. The Date of Bimaran Casket and its Significance. Page 47-65 in J. Stargardt and M. Willis (eds), Relics and Relic Worship in Early Buddhism: India, Afghanistan, Sri Lanka and Burma. London: British Museum Press Cribb J. 2018b. Numismatic Evidence and the Date of Kanişka I. Pages 7-34 in W. Rienjang and P. Stewart (eds), Problems of Chronology in Gandhāran Art: Proceedings of the First International Workshop of Gandhāra Connections Project, University of Oxford, 23rd – 24th March, 2017. Oxford: Archaeopress. Curtis J. 1998-99. Keynote Speech: British Near Eastern Archaeology and Museology on the Eve of the Millennium. British Association for Near Eastern Archaeology Newsletter 11 and 12: 9-10. Dabirsiyaqi M. (ed) 1968 (1347 shamsi). Manuchihrī. Tehran. Eberhard W. 1952. A History of China. London. Errington E. 2017a. Charles Masson and the Buddhist Sites of Afghanistan. Explorations, Excavations, Collections 1833-1835. London: The British Museum Press. Errington E. 2017b. The Charles Masson Archive: British Library, British Museum and Other Documents Relating to the 1832-1838 Masson Collection from Afghanistan. London: The British Museum Press. Errington E. and Cribb J. (eds) 1992. The Crossroads of Asia. Transformation in Image and Symbol in the Art of Ancient Afghanistan and Pakistan. Cambridge: Ancient India and Iran Trust. Errington E. and Curtis V. (eds) 2007. From Persepolis to the Punjab. Exploring ancient Iran, Afghanistan and Pakistan. London: The British Museum Press. Filigenzi A. 2009. The Buddhist Site of Tapa Sardar. Pages 41-57 in A. Filigenzi and R. Giunta (eds), The IsIAO Italian Archaeological Mission in Afghanistan 1957-2007. Fifty Years of Research in the Heart of Eurasia. Rome: IsIAO. Fischer K. 1967. Alexandropolis metropolis Arachosias. Zur Lage von Kandahar an landverbindungen Iran und Indien. Bonner Jahrbücher 67: 129-232. 23 The Global ConneCTions of Gandhāran arT Flood F. B. 2002. Between Ghazna and Delhi: Lahore and its Lost Manāra. Pages 102-112 in W. Ball and L. Harrow (eds), Cairo to Kabul: Afghan and Islamic Studies presented to Ralph Pinder-Wilson. London: Melisende. Fussman G. 2008. Monuments bouddhiques de la région de Caboul. Paris: Diffusion de Boccard. Haagsma M., Boer P. den, Moormann E.M., Sancisi-Weerdenburg H. 2003. The Impact of Classical Greece on European and National Identities. Amsterdam: J.C. Gieben. Harrow L.W. 1973. A Study of the Imagery of the Eleventh Century Panegyrist Azraqi of Herat. M.Phil. dissertation, School of Oriental and African Studies, London. Helms S.W. 1983. Kandahar of the Arab Conquest. World Archaeology 14/3: 342-354. Herrmann G. 1977. The Iranian Revival. Oxford: Elsevier-Phaidon. Hillenbrand R. 1994. Islamic Architecture: Form, Function, and Meaning. Edinburgh: Edinburgh University Press. Hillenbrand R. 2000. The Architecture of the Ghaznavids and Ghurids. Pages 124-206 in C. Hillenbrand (ed), Studies, in Honour of Clifford Edmund Bosworth. Vol 2. The Sultan’s Turret: Studies, in Persian and Turkish Culture. Leiden: Brill. Homai (ed). 1962. ᶜUthmān Mukhtārī. Tehran. Hutt A. 1977. The Central Asian Origin of the Eastern Minaret Form. Asian Affairs 64: 157-162. Inaba M. 2010. Kandahar, iii. Early Islamic Period. Encyclopaedia Iranica, vol. 15, fasc. 5: 477-478 (<http:// www.iranicaonline.org/articles/kandahar-early-islamic-period>, last updated 20th April 2012). Kessler A. T. 1993. Empires Beyond the Great Wall: The Heritage of Genghis Khan. Los Angeles: Natural History Museum of Los Angeles County. Klimburg-Salter D. 1989. The Kingdom of Bāmiyān: Buddhist Art and Culture of the Hindu Kush. Naples and Rome: IsMEO. Klimburg-Salter D. 2008. Buddhist Painting in the Hindu Kush ca. VIth to Xth centuries. Pages 131-159 in E. de la Vaissière (ed), L’Islamisation de l’Asie Centrale. Processus locaux d’acculturation du VIe au XIe siécle. Paris: Association pour l’avancement des études iraniennes. Klimburg-Salter D. 2010. Corridors of Communication Across Afghanistan 7th to 10th Centuries. Pages 173-186 in V. Marigo et.al. (eds), Paysages de centre de l’Afghanistan: paysages naturels, paysages culturels. Paris. Le Berre M. 1970. Le monument de Danestama en Afghanistan. Revue des Études Islamiques 38: 45-53. Le Strange G. 1905. The Lands of the Eastern Caliphate. Cambridge: Cambridge University Press. Lee J. L. 1996. The ‘Ancient Supremacy’: Bukhara, Afghanistan and the Battle for Balkh, 1731-1901. Leiden: Brill. Longworth-Dames M. 1927. Kandahār. Encyclopaedia Iranica, vol. 2: 711-12. Lutyens M. 1979. The Lyttons in India: Lord Lytton’s Viceroyalty. London: J. Murray. Lyttelton M. 1974. Baroque Architecture in Classical Antiquity. London: Thames & Hudson. MacDonald W. L. 1986. The Architecture of the Roman Empire. 2 Vols. New Haven: Yale University Press. Michell G. 1989. The Penguin Guide to the Monuments of India, vol. 1, Buddhist, Jain, Hindu. London: Penguin Books. Minorsky V. I. 1942. Marvazī on China, the Turks, and India. London: The Royal Asiatic Society. Minorsky V. I. 1970. Ḥudūd al-ᶜĀlam, The Regions of the World. A Persian Geography, 372 A.H.-982 A.H. London: Printed at the University Press, Oxford for the Trustees of the ‘E. J. W. Gibb Memorial’. Mir Husain Shah 1982. Al-Rukhaj in Hudud-al-Alam. Afghanistan 35/3: 1-10. Mitchell L. 2003. Bulwer Lytton: The Rise and Fall of a Victorian Man of Letters. London and New York: Hambledon and London. Motamedi H. 1975. The Influence of the Buddhist Art of Afghanistan on the Buddhist Art of Japan. Afghanistan 27/4: 32-35. Murray O. 2004. Introduction. A Lost School of History: Ancient Greece in the Age of Reform. Pages 1-34 in B. Lytton, Athens: Its Rise and Fall. London: Routledge. 24 WarWiCk ball: Gandhāra perCepTions: The orbiT of Gandhāran sTudies Paiman Z. 2018. Le monastère de Qol-e-Tut à la lumière des fouilles archéologiques. Afghanistan 1/1: 95121. Paiman Z. and Alram M. 2013. The Tepe Narenj Buddhist Monastery at Kabul: Buddhist Art During the First Muslim Raids against the Town. Paris: Diffusion De Boccard. Pinder-Wilson R. 2001. Ghaznavid and Ghurid Minarets. Iran 39: 155-186. Pugachenkova G. A. 1965. Khalchayan. Tashkent. Raverty H. G. 1892. The Mihrān of Sind and its Tributaries. Journal of the Asiatic Society of Bengal 61: 155508. Rienjang W. and Stewart P. (eds) 2019. The Geography of Gandhāran Art: Proceedings of the Second International Workshop of the Gandhāra Connections Project, University of Oxford, 22nd-23rd March, 2018. Oxford: Archaeopress. Rosenfield J. M. 1967. The Dynastic Art of the Kushans. Berkeley and Los Angeles: Berkeley: University of California Press. Rowland B. 1977. The Art and Architecture of India. Buddhist. Hindu. Jain. Harmondsworth and New York: Penguin Books Saᶜid Nafisī, undated. Azraqī. Tehran. Sanping C. 2005. Turkic or Proto-Mongolian? A Note on the Tuoba Language. Central Asiatic Journal 49/2: 161-74. Schlumberger D. and Sourdel-Thomine J. 1978. Lashkari Bazar: une residence royale ghaznévide et ghuride. Planches. Paris: Diffusion de Boccard Shoten A. 2006. Radiocarbon Dating of the Bamiyan Mural Paintings (Recent Cultural Heritage Issues in Afghanistan, vol. 2). Tokyo: National University Cooperation, Nagoya University Museum. Sinor D. 1990. The Establishment and Dissolution of the Türk Empire. Pages 285-316 in D. Sinor (ed), The Cambridge History of Early Inner Asia. Cambridge: Cambridge University Press. Taddei M. and Verardi G. 1978. Tapa Sardār: Second Preliminary Report. East and West 28: 33-136. Tetley G. E. 2009. The Ghaznavid and Seljuk Turks. Poetry as a Source for Iranian History. London and New York: Routledge. Tissot F. 2006. Catalogue of the National Museum of Afghanistan 1931-1985. Paris: UNESCO. Vogelsang W. 2002. The Afghans. Oxford: Wiley-Blackwell. Whitehouse D. 1978. Excavations at Kandahar 1974: First Interim Report. Afghan Studies 1: 9-39. Wood J. 1872. Journey to the Source of the Oxus. London: J. Murray. Wood M. 1997. In the Footsteps of Alexander the Great. A Journey from Greece to Asia. London: BBC Books. Yi J.L. 2018. Yungang: Art, History, Archaeology, Liturgy. London: Routledge. Young R., Coningham R., Batt C., and Ali I. 2000. A Comparison of Kalasha and Kho Subsistence Patterns in Chitral, NWFP, Pakistan. South Asian Studies 16: 13-142. Zwalf W. 1996. A Catalogue of the Gandhara Sculpture in the British Museum. 2 Vols. London: The British Museum Press. 25 Part 2 The Graeco-Roman connection (29-49): DOI: 10.32028/9781789696950-3 On the crossroads of disciplines: Tonio Hölscher’s theory of understanding Roman art images and its implications for the study of western influence(s) in Gandhāran art Martina Stoye The most striking characteristic of Gandhāran art and one of the main reasons motivating the study by European archaeologists of this Indian style of art has been, since the first major explorations of its archaeological sites in the nineteenth century, the clear presence of western, classical antique components. These can be seen in both narrative and decorative images, in terms of style as well as motif (Figures 1-4). They are so conspicuous, in fact, that they are even observable to a layperson.1 Attempts on the parts of archaeologists and historians to identify a clearly defined origin of Gandhāran art and credibly to date its products beyond doubt have, however, proven challenging: especially when it came to developing a coherent overall account, universally valid when applied to all phenomena of Gandhāran art. Early approaches tried to derive the traits of Gandhāran art so familiar to a European exclusively from one narrowly defined segment of western antique art, soon leading to fierce controversies: some saw Hellenistic art (that is, that of the pre-Imperial, non-Roman period) as Gandhāran art’s sole originator (Burgess, Grünwedel, Foucher),2 whereas followers of a ‘Roman School’ recognized predominant Roman sources (that is, of the Imperial and early Christian period) (Fergusson, Smith, Rowland, Buchthal, Wheeler, Soper, Ingholt, Ahrens, Seckel, Ackermann).3 Even though current research, owing to numerous excavations and individual studies, has been able to differentiate further, much controversy remains to this day (see further Stewart’s contribution to the present volume). But at least research has been able to work out reasons behind these irritations. That is to say, in Gandhāran art, ambiguities abound; the GraecoRoman influences in Gandhāran art have not only been blended into a predominantly native Indian frame, but Iranian and further Central Asian components have also found their way in.4 Because each individual There are countless small Corinthian/corinthianizing capitals on pilasters that separate Buddhist narrative scenes (Zwalf 1996: vol. 1, 50, 61 [ch. 8 § 3]; vol. 2, nos. 451-455, 127, 131, 139 etc.; Kurita 2003: vol. 2, e.g. nos. 630, 644, 645) as well as several monumental capitals of this kind showing Buddhist figures emerging out of the acanthus leaves (Zwalf 1996: vol. 1, 61 [ch. 8 § 3]; vol. 2, nos. 456, 457; Kurita 2003: vol. 2, nos. 516, 632-634). In addition, besides acanthus friezes (Zwalf 1996: vol. 2, nos. 449, 199, 219 etc.) or laurel leaf wreathes running along narrative scenes (Zwalf 1996: vol. 2, nos. 448, 177, 205, 206, 208 etc.; Stoye 2007; 2010b), there are plenty of decorative friezes of chubby putti bearing undulating garlands (Zwalf 1996: vol. 2, nos. 414-426; Kurita 2003: vol. 2, nos. 646-7, 653ff), all of the aforementioned motifs once adorning stūpas along with other more indigenous decor. There are figure groups borrowed from the Dionysiac sphere (Zwalf 1996: vol. 1, 34 n. 31; Kurita 2003: vol. 2, nos. 558, 560ff). Similarly, subjected to the pious Buddhist sphere were leogryphs (Berlin, inv. no. I 86), tritons, ketoi, (Zwalf 1996: vol. 2, nos. 340ff; Kurita 2003: vol. 2, nos. 695ff), centaurs (Zwalf 1996: vol. 2, no. 435; Kurita 2003: vol. 2, 705ff; fig. 3) or atlant-like caryatides (Zwalf 1996: vol. 2, nos. 355-378; Kurita 2003: vol. 2, nos. 447-459). Even in scenes of the Buddha’s lives, westerninspired figures and props appear, which did not exist in previous North Indian Buddhist imagery, such as Herakles-type males (Zwalf 1996: no. 293; Morgan 2019), a city goddess with turreted crown (Zwalf 1996: vol. 1, 44, nos. 176-178, 300), various figures with drapery billowing over their heads in velificatio (Mevissen 2011: 92ff; Tanabe 1998), the Trojan horse (Zwalf 1996: vol. 2, no. 300; Stewart 2016), cornucopiae (Zwalf 1996: vol. 2, nos. 93, 95-98: Kurita 2003: vol. 2, nos. 479ff), kantharoi (Zwalf 1996: vol. 2, nos. 98; Kurita 2003: vol. 2, 541) or tripods with lion paws (Stoye 2004) – just to name only the most obvious examples out of many possible ones. 2 Burgess 1899; Grünwedel 1901; Foucher 1905-1951: vol. 2, 401ff, 866-867, and 443,1; 1942-1947: vol. 2, 306-354. 3 Fergusson 1876: 177-182; Smith 1889-1893: 118-119; Rowland 1936; 1938; 1942; 1943; 1945; 1946; 1956a; 1956b; 1958; 1967; Buchthal 1942-3; 1943; 1945; Wheeler 1949; 1951; 1954; Soper 1951; Ingholt 1957; Ahrens 1961; Seckel 1964; Ackermann 1975. 4 E.g. pilasters with capitals of a Persepolitan order, frequent as scene dividers in some of the more peripheral Buddhist decor bands (Zwalf 1996: vol. 2, nos. 379ff, 455; Kurita 2003: vol. 2, nos. 642-643). Some iconographic elements suggest the absorption of Iranian concepts: the flames seen on the Buddha’s shoulders on some occasions, the frequent representations of a certain type of fire-altar in scenes of worship (Soper 1949; Tanabe 1981; Verardi 1987; 1988; 1994), the pictorial allusions to Pharro and Ardokhsho (otherwise known from coins) in some of the so-called ‘fertility couples’ of the Gandhāran imagery (Zwalf 1996, I: 1 29 The Global ConneCTions of Gandhāran arT Figure 1. Corinthianizing capital with Buddhist figures from Gandhāra. Berlin, Museum für Asiatische Kunst, inv. I 71. (Photo: copyright Staatliche Museen zu Berlin, Museum für Asiatische Kunst / Foto: J. Liepe.) Figure 2. Fragment of a putto-cum-garland frieze from Gandhāra. Berlin, Museum für Asiatische Kunst, inv. I 207. (Photo: copyright Staatliche Museen zu Berlin, Museum für Asiatische Kunst / Foto: I. Papadopoulos.) element was modified in order to create a harmonious whole, it is difficult retroactively to distill each respective, distinct ingredient. When it comes to the western components, most today agree that the influences from outside came in several impulses of differing cause, perhaps even through a continuous flow of ideas from the west (whatever one wants to understand by ‘west’, be it from the Middle or Near East, from Asia Minor or the Mediterranean).5 Even though this minimum consensus has been reached, current research on Gandhāran art is still searching for an elucidation of the details of this process. The fact that we continue to gather for workshops still occupied with the same topic that has been under discussion for 150 years, shows that a final conclusion that all can agree with has yet to be reached. Ever since the discovery of Aï Khanoum in the 1960s, it seems to me that those who believe that the origins of Gandhāran art can only be explained as a result of Hellenistic precursors in Asia itself predominate. This long-sought discovery of a Hellenistic city in Afghanistan (even though it is quite far away from Gandhāra) has been followed by the assumption that Gandhāran culture only became what it was through the centuries long westernized influences of its surroundings. From then on, theories 44) or the impact of Sasanian art in some later Buddhist icons (Zwalf 1996: 70); on Iranian influence on Gandhāran Buddhism in general cf. Scott 1990. 5 As W. Zwalf (1996: vol. 1, 67) noticed quite rightly, ‘a number of the more recent supporters of the Romano-Buddhist view accepted an initial role by a Hellenistic, even Iranised, art before the Roman influence … became dominant’. Other authors, in particular those who reflected on Gandhāran art after the discovery of Aï Khanoum, considered the western component of Gandhāran art solely as an overlay of various Hellenistic and Hellenized layers (e.g. a Hellenistic-Bactrian layer superimposed with Hellenized Parthian, as in Schlumberger 1970). Still other authors, accepting such premises, nevertheless thought the inclusion of an additional (last) Roman layer as self-evident, although the view was expressed that the latter layer in particular was difficult to discern (Nehru 1989: 4; Bussagli 1996: 244-245). 30 Martina Stoye: on the croSSroadS of diSciplineS Figure 3. Fragment of a centaur sculpture from Gandhāra. Berlin, Museum für Asiatische Kunst, inv. I 218. (Photo: copyright Staatliche Museen zu Berlin, Museum für Asiatische Kunst / Foto: J. v. Bruchhausen.) Figure 4. Fragment of a sculpture of Herakles-like bearded man with child from Gandhāra. Berlin, Museum für Asiatische Kunst, inv. I 214. (Photo: copyright Staatliche Museen zu Berlin, Museum für Asiatische Kunst / Foto: J. v. Bruchhausen.) that gave room to the possibility of Roman influence were vehemently rejected by many a renowned researcher, often even without further examination.6 Thus, Gandhāran culture has been described as a product of the greater Central Asian region, developing out of the historic after-effects of the conquests of Alexander, a late echo of the kingdoms of the Diadochi, especially of the Seleucid Empire, from which the Graeco-Bactrian kingdom split off, itself one of several important conditions for the creation of the Indo-Bactrian and Indo-Greek kingdoms. A series of kingdoms with a Hellenistic touch followed: the Indo-Scythians, the Indo-Parthians and the Kushans, although the mass production of Gandhāran art would flourish only under the latter. As a prologue to the hypotheses I will be presenting in this paper, I would like, to a certain extent, to reconsider this narrative of a long-lasting continuity – not by denying the history of developments that I have just laid out, but rather, by focusing on the breaks in the story. Especially because compared to the relatively late appearance of Gandhāran art in terms of pictorial traditions, these breaks may have meant the drying up of possible visual sources much too early. E.g. Dehejia 1997: 185 ‘A few early scholars also devoted their attention to narrative sculpture, explaining much in terms of Roman influence … Such proposals, belonging to the history of the study of Indian art, scarcely need refutation here.’ 6 31 The Global ConneCTions of Gandhāran arT For example, the Mauryan Empire expanded into eastern Afghanistan only after Alexander had arrived (for his very short interlude) in India. Aï Khanoum had already been abandoned in the 140s BC, by the latest around c. 130 BC (Mairs 2016: 61, 147, 151), and the last Indo-Greeks can only be traced to the decades around the turn of the first century BC to the first century AD (Cribb 2008: 65). I would also like to emphasize that there is still no archaeological evidence for extensive narrative relief art from those phases that could convincingly be regarded as a real precursor of Gandhāran art.7 The only image traditions that undeniably extend from these early Hellenistic enclaves to Gandhāran culture are the pictures found on coins (e.g. Errington & Cribb 1992: 52-88). The types of figures on coins and their range of actions are, however, so reduced that it would be impossible to see them as the main stimulating agent for the wide range of motifs and in particular of the visual narratives in second- and third-century Gandhāra. I must admit that many an Indo-Parthian was in possession of a good Greek education (Dani 1999: vol. 2, 197, 203) and, as we all know, that Parthian and Śaka preludes to Gandhāran art existed (Marshall 1960: 17-39; Fabrègues 1987; Nehru 1989: 68-94). In my opinion, however, neither of their image types not even the so-called toilet trays (Marshall 1951: vol. 2, 493-498; Francfort 1979; Dar 1979; Tanabe 2002; Lo Muzio 2002; 2011; 2018; Falk 2010; Pons 2011) adequately account for the later unified character of the mass production of Gandhāran art proper (i.e. for the sudden introduction of entire series of iconographies, even image programmes).8 I would also like to point out that the art of Gandhāra in its typical form, as reproduced over a period of 200 to 300 years, first developed at the end of the first century AD and flourished under the Great Kushans. Mass production began in the second and third centuries,9 so at exactly the same time as the flourishing and fullest extent of Roman Imperial art (see also Stewart’s chapter in this volume). This fact alone must therefore allow a renewed, in-depth look at the simultaneous production of art in the Roman Empire. It even seems imperative, especially now that at the end of the twentieth and beginning of the twenty-first century, classical archaeologists have made multiple brilliant contributions towards a new understanding of the language of Roman images.10 How Roman art is assessed and looked at is now fundamentally different from the approach at the time scholars first studied Roman prototypes for Gandhāran art, in the first half of the twentieth century. There are many possible reasons why research on Gandhāra has not taken much notice of these new contributions. Perhaps it is because these theories were mainly developed in German and not immediately translated into English, or because Gandhāran art scholarship never considered a revision of its old views of the possible impacts of Roman art (from the first half of the twentieth century) necessary, once the decision had been made to regard Gandhāran art as having grown predominantly out of regional Hellenistic sources. As announced in the title of this paper I would now like to draw attention to an important, academic theory for understanding Roman art and its possible implications for the study of western influence(s) in Gandhāran art. In the course of my PhD studies on the Buddhavita (life of the Buddha) in Gandhāran art and its relationships to ancient western representations of human life, I scrutinized a series of Buddhavita scenes11 and tried to relate them to late Hellenistic, Roman Republican, and Roman Imperial Compare the excavations of Barikot conducted by Luca Maria Olivieri (personal communication). Even if we were willing to accept the possibility of toilet trays being produced into the time of the Great Kushans, as suggested e.g. by Lo Muzio 2011: 338-339; 2018: 124. 9 Confirmed from a numismatist’s point of view also by Cribb (2008: 68; 2009: 69) ‘Die numismatischen Belege zeigen, dass von der zweiten Hälfte des 1. Jh. n. Chr. an ein freier Austausch zwischen Gandhara und einem Gemisch aus iranischen, römischen und indischen Einflüssen gegeben waren, die in die Entwicklung der gandharischen Bildhauerkunst eingegangen sind.’ 10 Scholars who summarize the more recent research history of Roman art studies speak of an historical turn – even of a seismic shift – in scholarship in the late 1960s and 1970s (Sinn 2000: 35-36; Hölscher 2000: 147; Hölscher 2002: 24; Lang 2002: 70-71; Stewart 2008: 4-5; Borg 2015: 3). 11 See <https://lisa.gerda-henkel-stiftung.de/martina_stoye> (last checked 18th May 2020). The scenes of the Buddhavita studied were: the miraculous birth and the ensuing first bath, the ride of young Siddhārtha to school, the first lesson, the 7 8 32 Martina Stoye: on the croSSroadS of diSciplineS imagery, amongst them biographical Roman sarcophagi as well as certain image types found in Roman triumphal iconographies. From a very careful consideration of the potential image-forming power of the still extant Buddhist narrative tradition against that of potential western models (of which I found surprisingly many previously unnoticed or undiscussed), a consistent pattern of iconography building (conflating various sources) soon emerged, which – once observed – seemed to apply also to many more image types. It also appeared to be valid for designs of decorative patterns.12 Since there are no parallels for this mode of image-building in early Indo-Buddhist imagery, I then started to search for explanations in the secondary literature on western antiquity and its pictorial languages, particularly in more recent books, which had not been used for Gandhāran research so far. Thus, one day I came across a book written in German by a German professor of classical archaeology on the semantics of Roman imagery, which really struck me: this was Tonio Hölscher’s Römische Bildsprache als semantisches System (Hölscher 1987). Through his suggestions with regard to Roman image-language, all my observations on the collage-like, but in terms of content meaningful, use of heterogeneous western models in ‘my’ iconographies suddenly seemed to make perfect sense. Hölscher offered very similar observations of a classical archaeologist on very different, but contemporaneous objects from a distant, but nevertheless connected context. I even got the impression that if one were to transfer the view of this classical archaeologist on the reception and adaptation of pre-Roman models in Roman art to the reception and adaptation of pre-Gandhāran models in the art of Gandhāra, then many contradictions and irritations regarding Gandhāran art could easily be reassessed (e.g. with regard to the multiple roots, to the text-image relationship, to our views on style etc.) – even without much talk about chronology. To me this book had a powerful effect. I should like to share its main thoughts and the implications for dealing with Gandhāran art in the following paragraphs. Arising from a lecture at the Heidelberg Academy of Sciences and Humanities in 1984, Hölscher’s slim volume was first published in German in a very modest edition in 1987. But in German archaeological circles, it soon became considered one of the most important and stimulating books ever written on Roman art,13 seminal for a deep understanding of Roman imagery,14 a book that should be compulsory reading for anyone interested in Roman art.15 Strangely enough, it took many years for this important work to be translated into English – until 2005, to be precise. The credit for accomplishing this task goes to an Oxbridge team, Anthony Snodgrass and his German wife Annemarie Künzl-Snodgrass, who provided a very beautiful, sensitive, and faithful translation,16 while Jaś Elsner contributed an insightful introduction, with helpful lists of further reading and a glossary.17 I should now like to introduce you to Hölscher’s ideas.18 Hölscher’s aim is to offer a new means of understanding Roman art by viewing Roman image-making as an embracing of older, Greek forms from heterogeneous backgrounds in order to express Roman ideals. He argues that in all periods of Roman art wedding, the Mahāparinirvāṇa scene (all of them newly developed in Gandhāran art and not known in previous Buddhist imagery). 12 On the elongated laurel wreath running along Buddhist narrative scenes as a migrated and then thoughtfully adapted decorative element, ideas were presented in Stoye 2007 and 2010b. 13 Stewart 2006: 210. 14 Newby 2006: 275. 15 Humble 2007: 126. 16 A. Snodgrass was Laurence Professor of Classical Archaeology at Cambridge from 1976 to 2001. A. Künzl-Snodgrass was a Language Teaching Officer in the Department of German at the University of Cambridge, and is a Fellow of Jesus College. Preface to the English edition in Hölscher 2004: xiii; Balty 2006: 636. 17 Jaś Elsner is Humfry Payne Senior Research Fellow in Classical Art and Archaeology at Corpus Christi College, Oxford, and the author of various studies on Roman art history. Foreword, Chronology and Glossary by Jaś Elsner in Hölscher 2004: xv-xxxv. 18 For this purpose, I will use several compilations of long quotations from this translation. As my own English is quite mediocre, whereas the translation is so brilliant, I thought that this procedure would serve best to achieve true reflections of Hölscher’s thoughts. 33 The Global ConneCTions of Gandhāran arT there was a profusion of types in circulation. Each time a new work of art was created, well-established image-types were selected from older art work (from whatever the source), thereby freely transferring some important aspect of the latter’s well-known core message to the content of the newly created artwork. Now let us read Hölscher’s own words in their English translation (Hölscher 2004). [5] ‘one basic and fundamental element of Roman art … [is] the indelible stamp of Greece…’; [6-7] ‘It is beyond argument that Roman art, in each of its fields and in as many different ways, rests on Greek foundations…’; [11] ‘Close inspection of Roman art reveals a picture of bewildering diversity. In every period of Roman history, the most varied stylistic phases of Greek art – from Late Archaic to late Hellenistic – are picked up and exploited’; [14] ‘In the choice of model, the extent of pluralism is remarkable’; [103] ‘…the whole range of forms from the Late Archaic to the late Hellenistic became available [for combinatory use]’; [11] ‘…we find established types of scene and figure, fundamental paradigms of scenic and figural composition, which derive from different epochs of Greek art but which are used, side by side, [at one time]…; [16] ‘The patterns … developed in Greek art are thus appropriated and exploited with breath-taking flexibility – sometimes for the whole composition, sometimes for single figures and groups, sometimes for yet smaller details’; [104] ‘The preconditions and beginnings of this language of imagery are not to be sought in Rome, but in Greek art of the second century BC … [But] from the second century BC onwards, Rome took part in this Greek process’; [105] ‘This handling of artistic forms was quickly transmitted by Greek artists to Rome’; [104] ‘Then, however, the phenomena were developed into a much more rigorous form, and to more far-reaching effect, in Rome and for the circumstances of the Roman Empire’; [111] ‘…by Augustan times at the latest, the repertory of forms must have achieved a certain completeness’; [104] ‘and … [it] came to fulfil a specific function’; [86-87] ‘…structures of form which had once, in the course of Greek history, been fundamental expressions of entire epochs, now acquired a new function as part of a system with an entirely different basis. In the centuries from the Archaic to the Hellenistic periods, Greek art had gone through a rapid and radical process of change, deeply stamped with the collective experiences of the successive epochs’; [87-88] ‘The unity of the various products of a given epoch thus predominated relatively strongly, … the new system of visual language, which was shaped in late Hellenistic times and remained in operation throughout the Roman Imperial period, represents not merely a new phase of development, but a fundamental break … formal resources, which in the past had been developed one after the other, were now available for use together. Out of a diachronic development, there came into being a synchronic range.’ Or, as Peter Stewart put it in his 2006 review of Tonio Hölscher’s book: ‘Greek forms [were] abstracted from their “diachronically different” origins and made synchronically available to [the] artists. The artists could pick and choose from the entire formal spectrum of past Greek art according to the requirements and expectations that surrounded particular works’19 (Figure 5). But what were these requirements and expectations? What was the intention behind the employment of such heterogeneous models? Can any intention be perceived at all? On what did the choice of models depend? Hölscher puts it this way: [18] ‘For if the choice of models does not depend on the taste and style of different periods, social groups or individuals, then on what does it depend? Is it a learned form of game-playing? Is it a symptom of Roman culture’s poverty of invention? A chaos of forms?’; [77] ‘… how [did] the use and adaptation of the models [take] place, and with what thinking behind them?’ On the basis of his observations on the application of diverse image-types on certain monuments and on artistic judgements in Roman literature, Hölscher concludes that different artistic types must have carried with them specific ideological meanings. The choice of one particular artistic model was not 19 Stewart 2006: 211. 34 Martina Stoye: on the croSSroadS of diSciplineS Figure 5. Simplified scheme of the strata from which Roman image-language draws its models. (M. Stoye; based on the work of T. Hölscher.) so much influenced by the prevailing aesthetic taste of the day, but was instead determined by wellestablished content-related associations corresponding with those visual forms. Or in short: in Roman art, the selection of certain models was strongly dictated by subject-matter, long associated with those models. To give you some examples: ‘for traditionally dignified figures […] the noble forms of High Classicism […] were preferred.’ This applied to images of gods and heroes. Images of such noble figures were meant to convey majesty (maiestas), dignity (gravitas), sanctity (sanctitas), and authority (auctoritas). Thus, most depictions of Roman gods and heroes were based on classical models, resulting in dignified, graceful figures with beautiful, ideally proportioned bodies and ideal, de-individualized faces of immaculate beauty, culminating in a seemingly timeless youthfulness and completely balanced expression.20 (Does that not, by the way, remind us of the Buddha image)? Classicism imbued with balance and gravitas (dignity) was also used to stage the great official ceremonies of the state.21 In contrast, the more naturalistic, animated, and emotional forms of the Hellenistic period (with their theatrical pathos, their inclination for dramatic passion) were very often chosen for ecstatic or wild figures, e.g. figures of the Bacchic revel (satyrs, dancing maenads, fauns, animals), for bucolic atmospheres or (typically) for battle scenes. [69] ‘No classicising taste, however strong, could have led to the search for fifth- and fourth-century [i.e. Classical] models for these motifs. The subject decided in advance the choice of representational possibilities.’ Hölscher characterizes the general composition of a Hellenistic battle scene, so readily adopted in Roman images whenever a great victory had to be underlined, as follows: [23-4] ‘[In battle-scenes of the Hellenistic period] …there is a multiple 20 21 Hölscher 2004: 69, 96-97, 105. Hölscher 2004: 47ff (chapter 6). 35 The Global ConneCTions of Gandhāran arT interrelation of actions within a coherent surrounding space… Each figure has its place and its role within this overall spatial composition; [26] ‘…forms are pushed together into masses, and thus can join together in collective movements… All this is presented with a pathos which was hitherto unthinkable… [here a] pain-distorted countenance … [there an] upraised arm … [27] pitiful gestures, as if displaying … the emotion that the viewer of the picture should feel… For all that, sharply contrasting effects are sought after… [29] [there is a] complete separation of victors and enemies … [44] the pathos of the fighting and the suffering of the defeated … [27] means no diminution of the glory of the victor: his position remains unaffected. On the contrary, he is elevated so far above the conquered.’ Does that not remind us of the Gandhāran scene of Mara’s attack, a scene that is meant to represent the greatest victory in the Buddha’s career? (I realize that this is a very associative link, but compelling nonetheless). Keeping this in mind, let us now move on to one of Hölscher’s major examples. Hölscher utilizes one monument in particular as his starting point, the Ara Pacis. It is his master example, the monument from which he unfurls his whole theory.22 The Ara Pacis was an altar to ‘Augustan peace’ erected between 13 BC and 9 BC by the Roman Senate to honour Augustus. It comprises a sculpted marble enclosure around the sacrificial altar itself, where important sacrifices would have been performed. The west side of the altar is decorated with mythological reliefs that refer to Rome’s mythical origins. The most intact relief on the east side celebrates abundance and fertility with a scene featuring a goddess often identified as Tellus. On the side walls one finds great processional friezes. They represent a solemn ceremony, with the Emperor, high office-bearers and the Imperial family23 participating, perhaps depicting the ‘inauguratio’ of the Ara Pacis itself.24 The participants of the procession walk from east to west on both sides toward the western (main) entrance of the Ara Pacis. Hölscher utilizes the Ara Pacis to show that even on one and the same monument (itself erected in the space of only two years) we can find a variety of heterogeneous models applied: [50] ‘…the great frieze with its imposing state ceremony follows the Parthenon frieze… this resemblance lies primarily in the overall composition and the handling of relief. The type of scene, the “Classical procession”, determines the overall appearance; it conveys for the event a quality of solemnity, of the dignitas and auctoritas of the state’s leading personalities and the religious establishment. [54] ‘The subject of a solemn state ceremony was … persuasively embodied by the form of composition used in the great frieze of the Ara Pacis’. Its general style of composition presented itself as the most perfect model for a solemn procession of the highest order also within its new, Roman context. And once designed, this type of scene, the ‘Classical procession’ with all its solemnity [55-56] ‘…retained its fundamental validity [as a scene type] for centuries.’25 [77] ‘Yet the individual figures in the procession nevertheless stand very much in different traditions. While the men in togas in some ways closely resemble figure-types from the time of the Parthenon frieze, the ruler’s imposing wife and the young mothers of the Imperial house are closer to the Late Classical and Hellenistic forms which emphasise the figure; the figure of Livia has rightly been placed along with the draped female figures of the late Hellenistic phase... Next come the flamines with their specifically Roman dress, for which one could not in any strict sense turn to older typological patterns, but only to reality. Thus a generally Classical type of scene was enacted with figure-types of different provenances, whose choice was once again dictated by their subjects… These heterogeneous figure-types, however, are not placed abruptly side by side but, through the execution of detail, are assimilated to their neighbours.’ Hölscher 2004: 76ff. Hölscher 2004: 49. 24 <https://www.bluffton.edu/homepages/facstaff/sullivanm/italy/rome/arapacis/arapacis.html> (last consulted 19th May 2020). 25 Hölscher 2004: 55-56: ‘the great relief scenes of the so-called Ara Pietatis, of the Arch of Titus, the Arch of Trajan at Beneventum, and of the triumphal arch of Marcus Aurelius, form an unbroken chain from the Ara Pacis… although over time the appearance of these derivative representations moved further and further away from the forms of Classical Greek relief.’ 22 23 36 Martina Stoye: on the croSSroadS of diSciplineS The case of the Aeneas panel on the Ara Pacis is used by Hölscher to refine this analysis further. It shows a prefigurement of the founding of Rome. Bottom left is the white sow, who according to prophesy would show Aeneas where Alba Longa was to be founded. In the background (top left) is a miniature model of a temple, which Aeneas had brought with him from Troy and which contained the Penates, the household gods of Rome. Aeneas is about to make a sacrifice by offering at a rustic altar. Before him are two attendants to the ritual, one with a bowl and jug, the other leading the sacrificial sow. According to Hölscher, the scene as a whole stands in the tradition of Hellenistic landscape [81] ‘…because the sacrifice of the Sow of Lavinium to the Penates was to be set in an idyllic sacral landscape, for which the only convincing tradition was that derived from Hellenistic art. The figure of Aeneas on the other hand follows Classical forms, because only thus could he acquire the qualities of auctoritas and pietas necessary for his role of ancestor and model for the Emperor. His ‘Classicism’ is thus … founded on content ... Yet already the sacrificial attendants are distanced vonce again from this “Classical” form… [16] Again, the group of the second sacrificial assistant with the pig follows a Hellenistic model, as preserved for instance in paintings from Delos … Finally, the depiction of Aeneas himself – while generally classical in type – is enriched by details developed only in later periods. Hence the sharply drawn folds around the legs appear in similar form on the late Hellenistic Poseidon of Melos. The patterns of representation, figural types and formulae developed in Greek art are thus appropriated and exploited with breath-taking flexibility – sometimes for the whole composition, sometimes for single figures and groups, sometimes for yet smaller details.’ [18] ‘In principle, therefore, we must distinguish between – on the one hand – modes of representation, figural types and formulae for detail which may be traced back to different epochs in Greek art, and – on the other hand – a conception of relief, together with a specific craft technique, whereby the heterogeneous elements of the work are presented in a unified “style”… The same principle applied for sculptures in the round’ (Hölscher 2004: 59ff). The iconography was therefore constructed as follows: first an appropriate scene type was selected, followed by the selection of appropriate models for the main figures until the main scene was finished. This could then be expanded to include further figures, regardless of whether or not the ingredients came from heterogeneous sources. Most important was that the forms selected made sense in terms of content: the visual form should be able to transport values and qualities, which the newly created motif could then be associated with. It was a kind of visual language, made up of well-established image codes that could be combined at will according to the needs of the subject matter at hand. An overall stylistic finish was then given to the whole (Figure 6). Let us now turn to Gandhāra. Can Hölscher’s depiction of a language of images in Roman art be usefully applied to Gandhāran art? Gandhāran art did develop at the time when this was already the principle of design in the Roman Empire. My experience with the iconographies that I studied indicates that it can. Have not so many studies already clearly shown that the image-creations of Gandhāran art used a similar method of composition, whereby motifs from very different sources were combined in order to create something completely new? If we allow for this possibility, how closely do we want to adhere to Hölscher’s ideas concerning the selection of motifs according to subject matter? And if we really want to accept that Gandhāran art itself adopted the semantics of the Roman visual language that Hölscher describes, what conclusions ensue for the further study of Gandhāran art? Let us first look again at the Birth Scene. Harking back to an already published article on the western source for one of its core motifs, the deva with the swaddling cloth (Stoye 2008; 2010a), I would like to expand on this study of its central motif and draw attention rather to the combinatory character of the whole iconography with regard to its heterogeneous sources. I will then show that the same principles of image-construction seem to have been at work in depictions of the first bath, using published and unpublished work of my own (Stoye 2004; Stoye, forthcoming). 37 The Global ConneCTions of Gandhāran arT Figure 6. Simplified scheme of the steps in the creation of a new meaningful iconography in Roman image-language (M. Stoye; based on the work of T. Hölscher.) Reliefs of the birth of Siddhārtha Gautama Let me quote selectively from my earlier publication in order to explain the example (Stoye 2010a): [160] ‘in the centre of the composition … Mâyâ, the mother of the Bodhisattva, is depicted in a charming stance … under the foliage canopy of a tree, her right hand grasping a branch of the tree above her head while the future Redeemer emerges half-length from her right flank. Her left arm rests on the shoulders of a female assistant, who supports her and acts as a midwife by pressing Mâyâ’s belly with her hand; [161-162] ‘The blessed child entering the light of day is always shown … head upwards, with outstretched arms moving away from his mother, casting his eyes in the direction of a male deva who approaches the mother from her right side…. Very often this deva, who receives the child on a swaddling cloth, is shown mid-stride, one foot forward, as if the artist had intended to show the motion, the speed and the impetus of the deity hurrying to be the first to receive the Saviour-to-be, an honour in itself… This striding out is often [not always] combined with a bowing down of the deva’s torso, which gives the figure a sense of enthusiastic devotion. This deva is always shown reaching out for the baby. Without exception, his outstretched hands are completely covered with a cloth, the ends of which fall down on both sides in gentle folds.’ Often there are additional figures added on both sides: [159] ‘male devas and spectators behind Indra and female assistants beside Maya’s midwife, bringing various props.’ [162] ‘Whatever the variations from relief to relief, the pose of Mâyâ or of the child is never abandoned, neither does Indra/the deva do anything else but receive the holy child on the cloth.’ Obviously this group was the central and binding element of the iconography, its quasi-canonical part, whereas the figures in its periphery were optional and could be handled with flexibility: different types of figures could fill their positions (Stoye 2010a: 159-160.). For a long time, the form given to the birth scene by Gandhāran artists had been explained exclusively in reference to Buddhist textual tradition (Stoye 2010a): [162] ‘But there are two fundamental problems with regard to the text-image relationship: firstly, close scrutiny of the textual record shows its details not to be congruent in a strict sense with the visual form chosen by artists… Secondly,… if we take into consideration 38 Martina Stoye: on the croSSroadS of diSciplineS what has been made known through philological research about the history and structure of the textual sources in question, we have to admit that the relevant passages exist in textual layers which may possibly be considerably younger than our iconography.’ [168] ‘[Thus,] we must ask: where, then, did the Gandhâran iconographers derive their well-defined form of the nativity scene from, which must have been very meaningful, even sacrosanct to them, as they clung to it without allowing any change. Beyond the Indian textual tradition, was there any other precursor?’ In my article (Stoye 2010a: 168-173), I tried to show that Gandhāran iconographers, starting from a still very simple narrative of the birth, in their search for an adequate image formula for a new anthropomorphic iconography of the Buddhist nativity, scoured western visual repertoires for meaningful equivalents, and came across depictions of the birth of Dionysos. And, considering the well-established connection of this theme’s image-formula to the subject of the divine birth, they found the application of its central scheme to be an exceedingly suitable tool to portray the Buddha also as a divine child who, like Dionysos, was born in a supernatural way. (The artists’ familiarity with Dionysiac imagery in Kushan India has been convincingly demonstrated – see also Tanabe’s paper in the present volume.) In the adjustment of the image to Buddhist needs, Dionysos became the Buddha. Sitting Zeus was replaced by standing Māyā and Hermes was transformed into Indra. In elaborating the details of their image further on, the artists then took recourse to even more visual motifs, some from an Indian background, some of other western inspiration than the image core. And they combined them – as it seems – without hesitation. By applying the śālabhañjikā pose to the figure of Māyā, they utilized an ancient Indian motif. On the one hand it solved the problem of depicting her standing parturition in a pleasing, even elegant way. At the same time, through its common Indian association with prosperity, the artists underscored the auspiciousness of this significant moment in the Buddhist narrative of salvation. The midwife-assistant, again, has its precedents in the western visual tradition. The lady behind her recalls female Hellenistic figures: the peacock fan in her hand is an Indian prop, the way she balances it in her hand, however, alludes to the western visual personifications of victory (Nike, Victoria) with palmleaf in their hands). Of Indian inspiration are the male figures in the picture’s left periphery. Nevertheless, a persistent treatment of folds in all figures betray the finishing in a unifying style. The same patchwork approach becomes apparent when examining other Gandhāran iconographies. I would now like to draw attention to the iconography of the First Bath as another example (Figure 7). Once again, I reproduce here the observations made in my other publications on the topic (Stoye 2004: 169-171; Stoye, forthcoming). The First Bath of the baby Siddhārtha: the tripod with lion-paws Three dozen Gandhāran examples of the scene known to us share the same composition: the baby Bodhisattva is shown in the centre and in frontal view standing on a three-legged table. Two male deities in three-quarter view, one to his left, the other to his right, pour water over the boy from a pitcher held over his head. Two kneeling women in richly pleated garments assist with the bath by supporting the standing new-born on either side. (Later, often not so carefully worked reliefs tend to simplify the details and eliminate the kneeling women and occasionally the three-legged support.) Until now, this basic iconography has been explained solely in the context of the Buddhist texts. The focus of discussion has been the depiction of the two male bathers as gods rather than nāgas, as was the case in the pictorial tradition of Sārnāth. The bathers have thus been identified as Indra und Brahma, and the Gandhāran iconography has been linked to the Lalitavistara, the only text that offers such an interpretation. Yet even this text is inconsistent and in one passage refers to the bathers as the nāgas Nanda and Upananda. While the texts prove unsatisfying in explaining even some of the central features of the images, other iconographic elements such as the kneeling women or the three-legged support, which were especially important in the early formulations of the representation, go entirely unmentioned in the texts. 39 The Global ConneCTions of Gandhāran arT It is precisely when textual and pictorial traditions diverge that art-historical formal analysis can be introduced to account for idiosyncrasies of the visual formulation. For it is conceivable that discrepancies between image and text are the result of artists expressing religious concepts in modes unique to the pictorial tradition, following their own semantic system of visual codes. In the case of Gandhāran art, the pictorial repertoire of the western art tradition often served as a source of inspiration. This was obviously the case here. The three-legged support in the scene of the First Bath can be recognized as one such western motif that reveals Gandhāran sculptors adopting not simply a Mediterranean form of furniture but a meaningful iconographic formula. The tripod in the Gandhāran reliefs, though occasionally of a rather simple shape, usually consists of a horizontal plate with three legs in the form of the paws of a beast of prey. In a few of the reliefs these legs have a special decorative form, which I call leontocephalopod (i.e. with a lion’s head and foot), following Schwendemann 1921. Here the lion-foot leg ends just above the knee joint with a lion protome, so that the leg’s projection becomes both the thigh of the lion leg and the chest of a beast of prey. This type of furniture has no prototype in early Buddhist art; yet furniture with legs in animal form, particularly lion-foot legs, exist in large numbers in Greek and Roman Art. The equivalent piece is the tripod with a round plate (Greek: tripous), widespread in the Mediterranean antique world where it was always categorized as a table (Greek: trapezai, Latin: mensae), rather than a type of seating or footstool (as may be surmised by its depiction in the First Bath scenes). The tripod is seen with increasing frequency on monuments in the Greek cultural area from the fourth century BC onwards, with either completely plain supports or simple, delicately shaped theriomorphic legs without protomes. By the Hellenistic Period (third and second centuries BC) the tripod is almost always seen with animal legs as a standard furnishing in banquet scenes, where it is shown next to the reclining diners and holds serving dishes. These Hellenistic lion-foot legs often looked rather muscular, but they always ended at the top with the thigh section and never included a lion protome. The adornment of the animal-leg form with further zoomorphic decoration, especially with the heads of lions, panthers, leopards, griffins, and related fantasy forms, became popular only in the Roman period. It must have been shortly before the first half of the first century BC that a table leg with a lion’s paw and protome was developed, probably in neo-Attic workshops. Numerous reliefs from all parts of the Roman Empire and countless finds of individual legs in this design from marble tables, the latter investigated by Moss 1988, testify to the popularity of the leontocephalopod type of table leg during Roman times.26 The golden age of this particular type of furniture was the first century AD. In the second century and thereafter, archaeological evidence shows a decrease in the number of such tables. Representations of the leontocephalopod table on reliefs first appeared during the Augustan Period (27 BC14 AD) and continued into the beginning of the fourth century. The leontocephalopod table leg form is thus distinctly Roman. Because the earliest Gandhāran depictions of the First Bath date from the second half of the first century AD, and the leontocephalopod table was popular in the western world in the first half of the first century, one may recognize the middle of the first century AD as the point in time when the transfer of this motif from the Mediterranean pictorial world to Gandhāra occurred. By virtue of its refined design and expensive materials the leontocephalopod table clearly stood out from the mass of objects for daily use. In wealthy Roman households it was a luxurious piece of furniture; in temples and sanctuaries it served as an appropriate cult table. Its use as a sacred table in particular appears to have been of importance in the adoption of the motif for the First Bath scene in Gandhāra. The use of the tripod table either as an altar for offerings or as a centrepiece for religious rituals can be attested to by numerous Roman sources, both visual and textual. More importantly, in the Mediterranean world the table could also be used as the base for a cult image and can be seen depicted as such. Of particular interest as a counterpart to the Gandhāran motif is a relief from Ostia of a high priest of Kybele, the only Roman representation of a tripod that includes the 26 Stoye (2004: 170): ‘According to Moss’s research, the earliest datable example of a small leontocephalopod tripod table comes from Pompeii and was manufactured before 42 BCE (fig. 10); one-third of all extant tables were excavated in the cities around Vesuvius (before 79 CE); those with inscriptions make clear that many originate from the time of Julius Claudius (27 BCE-68 CE); the remaining examples are more evenly distributed.’ 40 Martina Stoye: on the croSSroadS of diSciplineS Figure 7. Gandhāran relief of the First Bath of Siddhārtha, probably from the Swāt region, c. early second century AD. Japan, private collection. (Photo: after Kurita 1988, courtesy of Isao Kurita; previous academic publication in Stoye 2004: fig. 4.) figure of a child-god, similar to the figure of the infant Siddhārtha Gautama in the Gandhāran scene of the First Bath. The Ostia relief provides a new understanding of what was likely intended in Gandhāra when the Bodhisattva was placed on the table with lion’s feet: namely, the designation of Siddhartha as venerandum. That is, immediately after leaving the womb – and well before the enlightenment that he would achieve as an adult – the future Buddha is depicted as divine, very much in the manner of a cult image. The First Bath of baby Siddhārtha: the motif of the two symmetrically kneeling woman If we now take into consideration the two women kneeling on both sides of the centrally standing baby on its tripod (not mentioned in the Buddhist texts), but appearing in many Gandhāran scenes of the First Bath,27 further meaningful references to Roman visual models seem to emerge. On each side of baby Siddhārtha (as venerandum) a kneeling woman is shown in profile view. They are conceived as mirroring each other: the frontally standing baby provides the axis for their symmetry. Usually (but not always) the tiny figure towers above their heads. In those cases, the gaze of the slightly tilted back heads of the ladies is directed towards the baby. Their posture is ‘semi-kneeling’: Only one knee touches the ground (the sole of the corresponding foot is then vertical to the ground, only the toes are on the ground). The other knee is upright, on the level 27 For variations see Zwalf 1996, I: 152. 41 The Global ConneCTions of Gandhāran arT Figure 8. Frieze with victories from the Basilica Ulpia, Rome. Munich, Glyptothek, inv. GL 348. of the ladies’ hips, the corresponding feet firmly set on the ground. The moderately bent arms are slightly lifted and the ladies’ arms are stretched out towards the adored protagonist, apparently to hold the standing new-born child by one forearm. This pair of women (as an image type) has no predecessors in earlier IndoBuddhist art. Their formally quite stereotyped appearance does not find any detailed explanation in Buddhist textual tradition. However, kneeling figures have a long and complex pictorial tradition in Graeco-Roman art. This is discussed in detailed in my article (Stoye, forthcoming). The symmetrical kneeling of two women in the Gandhāran bath scene is astonishingly close to the Roman decorative motif of symmetrically kneeling victoriae (female personifications of victory) that particularly adorned one of the most important imperial monuments of the Trajanic period (AD 98-117), the Basilica Ulpia on the Forum of Trajan in Rome (Figure 8).28 Its symmetrically arranged pairs of victoriae may have had their predecessors on the so-called Campana reliefs, a type of architectural decorative panels made from clay or terracotta, widely used in the first century BC and first century AD to embellish architraves on buildings, mainly in Latium, of a more common type. Their strict symmetrical arrangements in particular were recognized through archaeological research as typically Roman (Borbein 1968: 179). The marble version of symmetrically kneeling goddesses of victory in the aforementioned imperial Basilica Ulpia is in all probability the noble culmination of a motif that had already been part of the Roman tradition for longer. The associations connected with the pictorial programme there are quite interesting in our context. The inauguration of the Basilica took place ten years after Trajan’s first triumph over the Dacians. As Packer writes in his book about the Forum of Trajan (Packer 1997: 4-5): ‘Construction … began in A.D. 106-107, and the buildings … were substantially complete by AD 112, the year when, according to an inscription found in Ostia, the complex was officially dedicated.’ Trajan styled himself here as optimus princeps (Zanker 1970: 531). The image-programme refers to his victories (Zanker 1970: 527). It celebrates his victoriousness in general. According to Zanker, on the Forum of Trajan everything is dominated by the one moment of the great victory, which includes all future and past victories and the resulting eternal presence of the new God (= Trajan). Trajan tried to address expectations of salvation and latent religious needs by concentrating gazes on the ruler as a supernatural being in order to give the empire greater cohesion (Zanker 1970: 543). The building must have received outstanding public attention: it is exaltedly praised in an ancient source and its depiction also adorned Roman Imperial coins of the period. The Forum is where numerous imperial acts took place (Packer 1997: 5). It must definitely have been among the places to be visited first by the embassies from India (= the Kushan empire?), of which we find mention in various ancient sources (Cobb 2018: 120). It is from this context that the similar motif of the two kneeling victories originates. Excavations at Trajan’s Forum in 1931-32 revealed fragments of an impressive architrave frieze, once placed over the inner columns 28 Trajan’s family name was Ulpius. 42 Martina Stoye: on the croSSroadS of diSciplineS (Photo: copyright Staatliche Antikensammlungen und Glyptothek München, Foto: Renate Kühling.) of the building, to which also the frieze fragment in the Munich Glyptothek once belonged (Fuchs 2002: 142). On the relief frieze, a symmetrically arranged representation framed by acanthus-covered balusters is repeated: the centre is occupied by a kneeling pair of victories, who decorate a thymaterion between them with laurel wreaths and at the same time scatter incense into the flames. On either side of the central grouping is a group of bull sacrifices: a half-dressed victory presses her knee into the back of a bull which she has forced to the ground (Fuchs 2002: 142) (in the Basilica Ulpia this scheme was repeated sixty-five times around the nave [Packer 1997: 233]). While the bull-sacrificing victoria/nike is historically based on a long Greek tradition, the image scheme of the goddesses of victory kneeling in front of a sacrificial device – in particular in its symmetrical arrangement with two victoriae – is not based on a classical Greek model, but is of Roman origin (Fuchs 2002: p. 143, in particular notes 19, 20). The Roman kneeling women are shown in profile view. They are conceived as mirroring each other. In the Trajanic frieze the thymaterion provides the axis for their symmetry. The gaze of the slightly tilted-back heads of the ladies is directed towards the ritual object. The kneeling women on both sides of the thymaterion are shown in ‘semi-kneeling’ pose: only one knee touches the ground (the sole of the corresponding foot is then vertical to the ground, only the toes are on the ground). The other knee is upright, on the level of the goddesses’ hips, the corresponding feet are firmly set on the ground. The moderately bend arms are slightly lifted and the goddesses’ arms are stretched out towards the thymaterion. If we accept Hölscher’s theory of Roman image-language and allow the possibility that a similar image construction method might have been at work in Buddhist Gandhāra (which was contemporary to the Roman Empire), which might have been inspired not only by locally transmitted Hellenistic but also by newly infiltrated Roman image-types, then we might open up to imagine the process of creation of this iconography as follows: Gandhāran iconographers, starting from a still quite simple narrative of the miraculous events immediately after the Saviour’s birth (the seven steps, proclamation of the final ‘victory’ of the Buddha-tobe, i.e. his escape from the cycles of rebirth in this last birth, the first bath of the new-born baby), in their search for an adequate image-formula for this moment pregnant with the preview of future victory, might have come across the image-formula of the pair of kneeling victories, devoting themselves to the solemn and focussed worship of a venerandum. This image-formula would not only have provided a beautiful visual form for underlining the importance of the small but central figure of the new-born baby, but (if Hölscher’s idea can be applied to Gandhāra) could also be read as encapsulating an undertone of the victoria aeterna of a princeps optimus. In designing the new scene the Gandhāran artists thus might have maintained the poses of the western female personification of victory (semi-kneeling, in profile view, slightly lifting up their arms towards a venerandum in between them), but then might have dropped their wings (as these would not have been adequate in their transfer to a Buddhavita scene) and replaced the thymaterion by a frontally standing baby Siddhārtha, thus providing a new axis-like venerandum between them and lending the scene 43 The Global ConneCTions of Gandhāran arT in their adaptation to a Buddhist context a ‘high-end’ solemnity of quasi-imperial status and a flavour of victoriousness of the highest kind. I admit that all this is an assumption. It is, however, an intellectually tested assumption developed from Tonio Hölscher’s ideas and earlier research in the field of Roman studies unnoticed so far by Gandhāran researchers and it makes, according to my view, surprisingly good sense, especially if we can apply the principles supposed and described here not only to one case of Gandhāran narrative but to a whole series of iconographies, as shown by my PhD studies. After having come to know Hölscher’s book, it seems to me that the genesis of Gandhāran iconographies works in much the same way described by Hölscher for Roman art. If we allow that this principle of image creation as described by Hölscher is also at work in Gandhāra, what are the conclusions and points of departure for the evaluation of narrative Gandhāran art in general? There are, I think, important conclusions with regard to the relationship between text and image, with regard to the diverse roots of Gandhāran iconography, and with regard to style as a tool for dating. And with regard to a potential consideration of the reliefs as evidence for everyday life in Gandhāra. First, Hölscher’s theory, with its emphasis on the decisive formative role of long-established visual models in the building of iconographies seems capable of inspiring Gandhāra researchers to modify their strongly textoriented views of the text-image relation in Buddhist imagery - and perhaps to allow for the possibility that visual models could have shaped an iconography just as much or perhaps even more decisively than many a narrative detail of Buddhist texts. Of course, there was a narrative core from which the image-designers began. An image is, however, a different medium from a text, following its own rules: an image needs to focus and condense a legend in a very different manner from a text. The main message needs to be visible at a glance. That image and text in Indian art often do not agree should in my opinion not be explained, as is often the case, as the result of a textual tradition having been lost. Image-creation always began with a very simple narrative or dogmatic core. But the image was then composed of older image types with all of their underlying established connotations, cleverly and meaningfully combined for a new context, just as described by Hölscher’s image-language (compare n. 11). Second, once we have identified the visual precursors, we can begin to decipher further nuances and deeper meanings added to Gandhāran reliefs. If we allow the possibility that the semantic field of a model played a crucial role in its selection, then any meanings underlying the image type may have acted as a kind of sounding board within the new iconography. Because combinatory use from diverse sources was common, the connotations incorporated into the image mean that the image could be interpreted on various levels. On the basis of my experience with the iconographies that I studied, I should even go so far as to assert that the choice of motifs was often so sophisticated that a consistent bilingual reading was possible. In most cases, I would even say that a bilingual visual opus was deliberately created that in all its subtlety must have appealed to the pious Buddhist as well as to the cultured Hellenistic, or Hellenized person. Third, considering the long-standing debate on the origins of Gandhāran art as Hellenistic or Roman, by making use of Hölscher’s theory of Roman image-language we can start to relax a lot: if we assume that the same method of image-making that Hölscher describes was also in use in Gandhāra, namely that diachronic elements were merged into something synchronous, then it becomes quite clear that we should expect elements of dissimilar date and heterogeneous origin in the different segments of Gandhāran art. All were present together: Hellenistic elements, Indo-Scythian, Indo-Parthian, Iranian and – I would argue – also Roman. It is exactly this kind of motif-reception and combination that is described by Hölscher. When we find elements of differing origin, therefore, it does not need to be a source of conflict. Instead, this is altogether in keeping with the style of the time (first to third) whatever we decide to call it. 44 Martina Stoye: on the croSSroadS of diSciplineS Why, though, does Gandhāran art on the whole look so dissimilar to Roman art? This can also be explained by Hölscher. On the left of Figure 9 can be seen the developmental phase model of Roman art according to my previous sketch. Transferring this to Gandhāran art would require the following amendments, as one can see on the right. In addition to Hölscher’s preliminary phases we would also need to include the regional antecedents to Gandhāra in the heterogeneous motif pool. Moreover, the underlying Buddhist religion set different priorities when choosing forms, leading in the end, of course, to a different appearance. Fourth, using Hölscher we may have to re-evaluate how we understand the different styles of Gandhāran art. Not only motif, but whole stylistic scenes could be transported from the diachronic to the synchronic system. Therefore, we need to consider the possibility that very differently worked pieces may not necessarily date to different time periods. We can see this on the Ara Pacis: the stylistic differences – on the one hand classicistic, on the other hand using Hellenistic forms – very clear in Hölscher’s comparison of the two procession friezes (Hölscher 2004: 78-79), arise from their differing intentions, not their date. Both are dated to 13 BC to 9 BC. Fifth, once a motif had been incorporated, over time its appearance could move further and further away from the form from which it had originally derived, through local developments taking their own directions (without further exchange with the original source). Thus, after a certain period of intra-Gandhāran transmission, an imported and then carefully adapted motif would still show similarity to the original visual idea in its iconographic structure, though less and less similarity in stylistic appearance. Sometimes the imported motifs that were incorporated at the beginning of the second century were later discarded again, perhaps because they did not mean enough to the simple believer, or because they were a kind of sophisticated elaboration considered dispensable in the often very reduced versions on small votive stūpas of the later second or third century. Interestingly some of them were also no longer passed on in further pictorial developments, e.g. in Northern India (of the fourth and fifth centuries), for example the aforementioned pair of kneeling women. Sixth, if we accept that the semantics of the Roman image-language played a certain role in Gandhāran iconography, then even further perspectives open up, with which I have not dealt today, but which may Figure 9. Simplified scheme of the strata from which Gandhāran image-language draws its models (M. Stoye; based on the work of T. Hölscher.) 45 The Global ConneCTions of Gandhāran arT enrich further research, e.g. the allusion to certain key virtues through certain image-types in Roman art. Such scenes, for example, formed a crucial part of biographical scenes (they occur in the pictorial world of sarcophagi as well as in so-called historical reliefs of triumphal arches). If we take Hölscher’s account as valid for Gandhāran art, too, then not only the form but also the deeper meaning of such image types might have transferred itself in a modified form to Buddhavita scenes. References Ackermann H.C. 1975. Narrative Stone Reliefs from Gandhāra in the Victoria and Albert Museum in London. IsMEO Reports and Memoires XVII. Rome: IsMEO. Ahrens D. 1961. Die römischen Grundlagen der Gandharakunst (Orbis Antiquus 20). Münster: Aschendorff. Balty J.C. 2006. Review of T. Hölscher, The Language of Images in Roman Art translated by A. Snodgrass and A. M. Künzl-Snodgrass. L’Antiquité Classique 75: 635-637. Borbein A.H. 1968. Campanareliefs: typologische und stilkritische Untersuchungen. (Mitteilungen des Deutschen Archäologischen Instituts. Römische Abteilung. Ergänzungsheft 14.) Heidelberg: Kerle. Borg B.E. 2015. A Companion to Roman Art. Chichester: John Wiley and Sons. Buchthal H. 1942-3. The Foundations for a Chronology of Gandhara sculpture. Journal of the Royal Asiatic Society 3-4: 137-48. Buchthal H. 1943. The Common Classical Sources of Buddhist and Christian Narrative art. Transactions of the Oriental Ceramic Society 19: 21-32. Buchthal H. 1945. The Western Aspects of Gandhara Sculpture. Proceedings of the British Academy 31: 151-176. Burgess J. 1899. The Gandhara Sculptures. London: W. Griggs. Also published in The Journal of Indian Art and Industry 8 (1900): 23-40. Burgess J. 1900. The Gandhara Sculptures: Some Recent Acquisitions. The Journal of Indian Art and Industry 8: 73-92. Bussagli M. 1996. L’art du Gandhara. Paris: Librerie Générale Française. Cobb M.A. 2018. Rome and the Indian Ocean Trade from Augustus to the Early Third Century C.E. (Mnemosyne, Supplements, History and Archaeology of Classical Antiquity). Leiden: Brill. Cribb J. 2008. Die Chronologie Gandharas anhand der Münzen. Pages 64-69 in Gandhara, das buddhistische Erbe Pakistans. Legenden, Klöster und Paradiese. Bonn: Kunst- und Ausstellungshalle Bonn. Dani A. H. 1999. History of Civilizations of Central Asia. 2 vols. (Multiple History Series). Delhi: Motilal Banarsidass. Dar S. R. 1979. Toilet Trays from Gandhara and Beginning of Hellenism in Pakistan. Journal of Central Asia 2/2: 141-184. Dehejia V. 1997. Discourses in Early Buddhist Art: Visual Narratives of India. New Delhi: Munshiram Manoharlal. Errington E. and Cribb J. (eds) 1992. The Crossroads of Asia. Transformation in Image and Symbol in the Art of Ancient Afghanistan and Pakistan. Cambridge: Ancient India and Iran Trust. Fabrègues C. 1987. The Indo-Parthian beginnings of Gandhara sculpture. Bulletin of Asia Institute N.S. 1: 33-43. Falk H. 2010. Libation Trays in Gandhara. Bulletin of the Asia Institute 24: 89-113. Fergusson J. 1876. A History of Architecture in All Countries: From the Earliest Times to the Present Day, vol. 3, History of Indian and Eastern Architecture. London: John Murray. Foucher A. 1905-1951. L’art gréco-bouddhique du Gandhâra: Étude sur les origines de l’influence classique dans l’art bouddhique de l’Inde et de l’Extrême-Orient. 2 vols. Paris: Imprimerie Nationale. Foucher A. 1942-1947. La vieille route de l’Inde de Bactres à Taxila (Mémoires de la Délégation Archéologique Française en Afghanistan, 1). 2 vols. Paris: Éditions d’Art et d’Histoire. Francfort H.P. 1979. Les palettes du Gandhāra (Mémoires de la Délégation Archéologique Française en Afghanistan, 23). Paris: Diffusion De Boccard. Fuchs M. 2002. Römische Reliefwerke. (Glyptothek München, Katalog der Skulpturen, vol. 7). München: Verlag C.H. Beck. Grünwedel A. 1901. Buddhist Art in India. (Translated from its German edition in 1900 by Agnes C. Gibson, revised and enlarged by J. Burgess.) London: Quaritch. 46 Martina Stoye: on the croSSroadS of diSciplineS Hölscher T. 1987. Römische Bildsprache als semantisches System (Abhandlungen der Heidelberger Akademie der Wissenschaften, Philosophisch-Historische Klasse). Heidelberg: Universitätsverlag Winter GmbH. Hölscher T. 2000. Bildwerke: Darstellungen, Funktionen, Botschaften. Pages 147-165 in A.H. Borbein, T. Hölscher, P. Zanker (eds), Klassische Archäologie: eine Einführung. Darmstadt: Wissenschaftliche Buchgesellschaft. Hölscher T. 2002. Klassische Archäologie: Grundwissen. Darmstadt: Wissenschaftliche Buchgesellschaft. Hölscher T. 2004. The Language of Images in Roman Art (translated by A. Snodgrass and A. Künzl-Snodgrass , with an intro. by J. Elsner). Cambridge: Cambridge University Press. Humble N. 2007. Review of T. Hölscher, The Language of Images in Roman Art translated by A. Snodgrass and A. M. Künzl-Snodgrass. Classics Ireland 14: 123-126. Ingholt H. 1957. Gandhāran Art in Pakistan. New York: Pantheon Books. Kurita I. 2003. Gandhāran Art (Revised and enlarged edition). 2 vols. Tokyo: Nigensha. Lang F. 2002. Klassische Archäologie: eine Einführung in Methode, Theorie und Praxis. Tübingen: Francke. Lo Muzio C. 2002. L’artigianato di lusso nel Nord-Ovest di epoca indo-greca, saka e partica: i “piattelli per cosmetici”. Pages 77-83 in P. Callieri and A. Filigenzi (eds), Il Maestro di Saidu Sharif. Alle origini dell’arte del Gandhara. Rome: Istituto italiano per l’Africa e l’Oriente. Lo Muzio C. 2011. Gandharan Toilet-Trays: Some Reflections on Chronology. Ancient Civilizations from Scythia to Siberia 17: 331-340. Lo Muzio C. 2018. On the Relationship between Gandhāran Toilet-Trays and the Early Buddhist Art of Northern India. Pages 123-134 in W. Rienjang and P. Stewart (eds), Problems of Chronology in Gandhāran Art. Oxford: Archaeopress. Mairs R. 2016. The Hellenistic Far East: Archaeology, Language, and Identity in Greek Central Asia. Berkeley: University of California Press. Marshall J. 1951. Taxila. 3 vols. Cambridge: Cambridge University Press. Marshall J. 1960. The Buddhist Art of Gandhāra: The Story of the Early School, Its Birth, Growth and Decline (Memoirs of the Department of Archaeology in Pakistan 1). Cambridge: Cambridge University Press (repr. 1980). Mevissen G. 2011. The Great Renunciation: Astral Deities in Gandhāra. Pakistan Heritage 3: 89-112. Morgan L. 2019. Heracles’ Track to the Indus: Ancients and Moderns in the Swat Valley. Unpublished lecture, recorded at <http://www.carc.ox.ac.uk/GandharaConnections/Webcasts.htm> (recording last consulted 19th May 2020). Moss C. 1988. Roman Marble Tables. 3 vols. PhD Dissertation, Princeton University. Nehru L. 1989. Origins of the Gandhāran Style: A Study of Contributory Influences. Delhi: Oxford University Press. Newby Z. 2006. Review of T. Hölscher, The Language of Images in Roman Art translated by A. Snodgrass and A. M. Künzl-Snodgrass. Journal of Roman Studies 96: 274-275. Packer J.E. 1997. The Forum of Trajan in Rome. A Study of the Monuments. 2 vols. Berkeley: University of California Press. Pons J. 2011. From Gandhāran Trays to Gandhāran Buddhist Art: The Persistence of Hellenistic Motifs from the Second Century and Beyond. Pages 153-175 in A. Kouremenos, S. Chandrasekaran and R. Rossi (eds), From Pella to Gandhara. Hybridisation and Identity in the Art and Architecture of the Hellenistic East (BAR International Series 2221). Oxford: Archaeopress. Rowland B. 1936. A Revised Chronology of Gandhāra Sculpture. The Art Bulletin 18: 387-400. Rowland B. 1938. Buddha and the Sun God. Zalmoxis, revue des études religieuses 1: 69-84. Rowland B. 1942. Gandhara and Late Antique Art: The Buddha Image. American Journal of Archaeology 39: 489496. Rowland B. 1943. St. Peter in Gandhara: An Early Christian Statuette in India. Gazette des Beaux-Arts, 6th ser., 23: 489–96. Rowland B. 1945. Gandhāra and early Christian art: Buddha palliatus. American Journal of Archaeology 49: 445448. Rowland B. 1946. Gandhāra and Early Christian Art: The Home-Arcade and the Date of the Bimaran Reliquary. The Art Bulletin 28: 44-47. 47 The Global ConneCTions of Gandhāran arT Rowland B. 1956a. Gandhara, Rome and Mathura: The Early Relief Style. Archives of the Chinese Art Society of America 10: 8-17. Rowland B. 1956b. The Vine Scroll in Gandhāra. Artibus Asiae 19: 353-61. Rowland B. 1958. Rome and Gandhara. East and West 9: 199-208. Schlumberger D. 1970. L’Orient hellénisé: l’art grec et ses héritiers dans l’Asie non méditeranéenne. (L’Art dans le Monde.) Paris: Albin Michel. Schwendemann K. 1921. Der Dreifuss: ein formen- und religionsgeschichtlicher Versuch. Jahrbuch des Deutschen Archäologischen Instituts 36: 98-185. Scott D. A. 1990. The Iranian Face of Buddhism. East and West 40: 43-78. Seckel D. 1964. The Art of Buddhism. New York: Crown Publishers. Smith V. 1889-1893. Graeco-Roman Influence on the Civilization of Ancient India. Journal of the Asiatic Society of Bengal 58: 107-198, 61: 50-76, 62: 84-87. Soper A. 1949. Aspects of Light Symbolism. Archives of Asian Art 12: 252-283. Soper A.G. 1951. The Roman Style in Gandhāra. American Journal of Archaeology 55: 301-319. Stewart P. 2006. Review of T. Hölscher, The Language of Images in Roman Art translated by A. Snodgrass and A. M. Künzl-Snodgrass. The Classical Review 56: 210-211. Stewart P. 2008. The Social History of Roman Art. Cambridge: Cambridge University Press. Stewart P. 2016. The Provenance of the Gandhāran ‘Trojan Horse’ Relief in the British Museum. Arts Asiatiques 71/1: 3-12. Stoye M. 2004. Der Dreifuss in gandhārischen Szenen zum Ersten Bad des Siddhartha Gautama: Überlegungen zu Herkunft und Bedeutung eines visuellen Zitats (The Tripod in Gandhāran Scenes of the First Bath of Siddhartha Gautama: Reflections on the Provenance and Meaning of a Visual Quote) [with English Summary]. Artibus Asiae 64, 2: 141-176. Stoye M. 2007. Serta Laurea zu Ehren Buddhas – Zur gestreckten Lorbeerblattgirlande im Reliefdekor gandharischer Stūpas. Berliner Indologische Studien 18: 241-381. Stoye M. 2008. The Deva with the Swaddling Cloth: On the Western Origins of Gandhāran Birth Iconography and their Implications for the Textual and Art History of the Buddhist Saviour’s Nativity. Pages 1-25 in C. Bautze-Picron (ed), Religion and Art: New Issues in Indian Iconography and Iconology. (Volume 1 of the proceedings of the 18th conference of the European association of South Asian Archaeologists, London, 2005.) London: The British Academy. Stoye M. 2010a. The Deva with the Swaddling Cloth: On the Western Origins of Gandhāran Birth Iconography and their Implications for the Textual and Art History of the Buddhist Saviour’s Nativity. Pages 157-192 in C. Cüppers, M. Deeg and H.Durt (eds), Birth of the Buddha. Proceedings of the Seminar Held in Lumbini, Nepal, October 2004 (LIRI Seminar Proceedings Series 3). Kathmandu: Lumbini International Research Institute. Stoye M. 2010b. Festivity and Sacred Aura Thoughts on the Origins and Meaning of the Garland Moulding in the Ornamentation of Gandhāran stūpas. Pages 345-352 in P. Callieri, L. Colliva et al. (eds), South Asian Archaeology 2007: Proceedings of the 19th Meeting of the European Association of South Asian Archaeology in Ravenna, Italy, July 2007. 2 vols. (British Archaeological Reports: International Series 2133.) Oxford: Archaeopress, 2010. Stoye M. forthcoming. The Symmetrical Pair of Kneeling Women in Gandhāran Scenes of the First Bath of Siddhārtha Gautama and Visual Parallels in Western Art. Indo-Asiatische Zeitschrift. Tanabe K. 1981. Iranian Background of the Flaming and Watering Buddha Image in Kushan Period. The Ancient Orient Museum 3: 69-81. Tanabe K. 1998. The Goddess of Night in the Gandhāran Great Departure Scene of Siddhārtha. Silk Road Art and Archaeology 5: 213-232. Tanabe K. 2002. Greek, Roman and Parthian Influences on the Pre-Kushana Gandharan ‘Toilet-Trays’ and Forerunners of Buddhist Paradise (Pâramitâ). Silk Road Art and Archaeology 8: 73-100. Verardi G. 1987. The Buddha’s Dhūnī. Pages 369-383 in M. Yaldiz and W. Lobo (eds), Investigating Indian Art. Proceedings of a Symposium on the Early Buddhist and Hindu Iconography held at the Museum of Indian Art. 48 Martina Stoye: on the croSSroadS of diSciplineS Berlin, 1986. (Veröffentlichungen des Museums für Indische Kunst 8.) Berlin: Museum für Indische Kunst, Staatliche Museen zu Berlin, Stiftung Preussischer Kulturbesitz. Verardi G. 1988. Tematiche indiane di alcune iconografie gandhariche. II, Buddha, Agni, i lakṣaṇa, il miracolo de Sarasvati e alri problemi connessi. Pages 1533-1549 in G. Gnoli and L. Lanciotti (eds), Orientalia Iosephe Tucci Memoriae dicata 3. Roma: Istituto Italiano per il Medio ed Estremo Oriente. Verardi G. 1994. Homa and Other Fire Rituals in Gandhāra. Napoli: Istituto Universitario Orientale. Wheeler M. 1949. Romano-Buddhist Art: An Old Problem Restated. Antiquity 23: 4-19. Wheeler M. 1951. Roman Contact with India, Pakistan and Afghanistan. Pages 345-381 in W. F. Grimes (ed), Aspects of Archaeology in Britain and Beyond: Essays Presented to O.G.S. Crawford. London: H.W. Edwards. Wheeler M. 1954. Rome beyond the Imperial Frontiers. London: Penguin Books. Zanker P. 1970. Das Trajansforum in Rom, Archäologischer Anzeiger 85 (1970): 499-544. Zwalf W. 1996. A Catalogue of Gandhāra Sculpture in the British Museum. 2 vols. London: British Museum Press. 49 (50-85): DOI: 10.32028/9781789696950-4 Roman sarcophagi and Gandhāran sculpture1 Peter Stewart The subject of this paper is an artistic relationship that should not exist: an apparent, but still mysterious connection between two contemporary spheres of sculptural production separated by a distance of some 5,000 kilometres. At one extreme was the tradition of Roman marble sarcophagi, which where made in various large centres of production across the empire, particularly the city of Rome itself (e.g. Koch & Sichtermann 1982). Although it had very ancient antecedents, this tradition blossomed gradually in the course of the second century AD, stimulated by a shift of preference from cremation to inhumation. The sarcophagi were containers for the single or multiple remains of dead. Commissioned during the lifetime of the deceased or by their next of kin after their death, such marble caskets were a minority choice, the preserve of those wealthy enough to afford them. In Italy they were typically deposited inside mausolea, and could be richly decorated with sculptural reliefs, including mythological narratives whose themes resonated with the contemplation and lamentation of death. At the other, eastern extreme were the sculptures made to decorate monumental, Buddhist stūpas in Gandhāra. The tradition of Gandhāran religious sculpture appears to have originated little earlier than about the mid-first century AD. Its heyday was in the second and third centuries, and although the dating of individual pieces is notoriously challenging, it would be reasonable to assume that the examples mentioned below fall within that broad time-frame.2 The stūpas also were funerary monuments of a kind, intended to commemorate the life and parinirvāṇa of the historical Buddha within the townscapes and landscapes of the Gandhāran region and, strictly speaking, to contain his bodily relics. Consequently, they and their sculptures served as foci of veneration, circumambulated (as some Roman tombs possibly were also), by monks and lay-people who might contemplate the imagery as part of their devotions (pradakṣiṇa).3 That iconography included narrative scenes telling stories from the Buddha’s lives. It is important to emphasize from the outset that the connection posited between these two traditions does not represent, and certainly does not explain, the whole story of how Graeco-Roman art and Gandhāran art were implicated with each other. That is a problem which requires very much more research, which cannot be reduced simply to a historical moment, and which will continue to present problems of sometimes imponderable complexity. This paper does contend, however, that Roman sarcophagi contribute to understanding that story. Classical connections In the context of this volume and the project that gave rise to it, there is no need to say much about the place of classical art in the study of Gandhāra. From the outset, Graeco-Roman art has been perceived as fundamental to the global connections of Gandhāran art. As Warwick Ball has observed in The footnotes and references for this paper were completed during the coronavirus Covid-19 emergency of March-April 2020 and have been checked to the best of my ability without library access. A few pertinent publications have been cited for the reader’s benefit even where it was not possible for me to consult them (as indicated). For comments on versions of this paper I am grateful both to participants in the Gandhāra Connections workshop and to seminar and lecture audiences at Oxford, the Faculty of Classics in Cambridge, the Institute of Classical Studies in London, and the Department of Art History in the University of Toronto. 2 On the issue of chronology see Rienjang & Stewart 2018, with further documentation and references. Where no dates are suggested in the captions of this chapter, ‘c. second-third century AD’ should be assumed. 3 The evidence for circumambulation of Roman tombs (including the large stūpa-shaped mausolea of the emperors Augustus and Hadrian) – as opposed to funerary pyres – seems to me a little tenuous, but worth recording in this connection. See Winfeld-Hansen 1965: 35-63 (non vidi); Davies 1997: 52-8. For the notion that Gandhāran stūpas were actually influenced by Roman architectural technology see Kuwayama 1997. 1 50 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure his contribution to this volume, it was the European officials, soldiers, and travellers of British India, who first studied Gandhāran sculptures and who noticed their inclusion of some classical artistic conventions. Gandhāra was the North-West Frontier of British India, and it was also the eastern limit of the ‘classical world’, where Alexander the Great had crossed the River Indus in 326 BC, and where his Hellenistic Greek successors had continued to rule, preserving a patchy colonial inheritance of Greek culture till more than two centuries later. Perhaps the (often classically educated) explorers of the Raj were attracted by the idea of a familiar and authoritative classical tradition in India4 and perhaps, as Ball suggests, Gandhāran culture offered a sort of proxy for the presence of Alexander himself. At any rate, what Europeans increasingly realized in the last few decades of the nineteenth century was that Gandhāran Buddhist sculptures frequently exhibited uncanny reflections of the iconographical and stylistic repertoire of classical art, which was not only familiar to them from antiquities but also from its pervasive legacy within the modern art of the west. The examples that follow will offer ample illustration of those resonances, but they are more numerous, more varied, and harder to categorize than the present study can convey. Gandhāran art appeared to imitate Graeco-Roman art in respect to: its stylistic tendencies; its choice of devices from the repertoire of classical naturalism; the poses, gestures, and compositions of figures; their outfits (many figures even wear the Greek himation, though not the Roman toga as is sometimes claimed); and sometimes its specific iconography, in the case of figures such as ‘Herakles’-Vajrapāṇi, or the carefree ‘Dionysiac’ characters in scenes of carousing. In very rare cases more complex iconography was assimilated for Buddhist purposes, as with the ‘Trojan Horse’ scenes mentioned below. Even the anthropomorphic image of the Buddha himself, with his stereotypical face of ideal restraint and detachment, almost certainly borrows from the ethically charged ideal faces of Graeco-Roman art, such as images of the youthful god Apollo. It was probably invented in the first century AD, appropriating components of the classical tradition which made sense for the new religious context.5 Of course, resemblances of this kind are illustrated through carefully selected examples, and examples are, by definition, tendentious. It is sometimes observed that the Gandhāran style is much more diverse and less classical-looking than often believed, for we (that is to say, those of us influenced by a familiarity with classical imagery) assess the inconsistently published corpus of Gandhāran art through filters created by our experience and expectations. In fact, over the last half-century there has been a growing tendency to recognize the pluralism of Gandharan style, seeking to balance the perception of direct Graeco-Roman influence against other contributory factors: connections with South Asia, for example, or with other parts of Central Asia, or with the Hellenized artistic traditions of the Persian/Parthian world, which had absorbed and mediated classical art. Such a reasonable open-mindedness about the global connections of Gandhāran art must be the right approach, although the tidy cosmopolitanism it implies has its own problems (Nehru 1989; Falser 2015 for critical perspective). The move beyond a binary comparison of classical and Gandhāran art has been helpful, and increasingly chimes with our awareness of global cultural complexity through modern globalization. Like many twenty-first century commentators, I am constitutionally inclined to the fluidity and connectivity which this approach implies. On the other hand, there may be a case for arguing that the Graeco-Roman component is more pervasive than is sometimes recognized. This is because the past tendency to illustrate classical influence through very selective examples of Greek and Roman art does not do justice to the diversity of the classical tradition itself. That is to say, some of the less obviously classical works of Gandhāran art On the other hand, some responses were distinctly negative about the poor reflection of classicism represented by Gandhāra, in contrast to the distinctive beauties of Hindu art. On the early reception of Gandhāran art see the introductory discussion in Zwalf 1996 and esp. 23 n. 6. 5 Among the vast bibliography on this subject see e.g. Cribb 1984; DeCaroli 2015: esp. 27-42 (with an historiographical focus). Controversy has mainly surrounded where the first Buddha images were created, rather than the fact that Gandhāran images drew on Graeco-Roman traditions. 4 51 The Global ConneCTions of Gandhāran arT may appear less distant from the Graeco-Roman tradition if considered against the full bandwidth of artistic production across the Greek and Roman world, rather than merely famous exemplars. The reasons for the affinity between Graeco-Roman and Gandhāran art are still unclear (elucidating this problem is the principal objective of Oxford’s Gandhāra Connections project). Even today, after a century and a half of research on Gandhāran art, and inoculated as we hope we are against some of the Eurocentric and imperialist biases of his age, we can still appreciate the astonishment expressed by Rudyard Kipling, referring to the Lahore Museum in the opening pages of his novel, Kim: ‘In the entrance-hall stood the larger figures of the Greco-Buddhist sculptures done, savants know how long since, by forgotten workmen whose hands were feeling, and not unskilfully, for the mysteriously transmitted Grecian touch.’ A ‘Grecian touch’ it seemed to be indeed, for the most obvious explanation of the ‘classicism’ of Gandhāran art was, and is, to link it to the legacy of Alexander. It was already evident in the nineteenth century, from classical literary sources as well as Graeco-Bactrian and ‘Indo-Greek’ coinage, that the rulers of Bactria and Gandhāra in the Hellenistic period were usually men of Greek descent, the leaders of colonial communities who preserved Greek language and institutions, and styled themselves as victorious kings in the mould of the Ptolemies or Seleukids, or Alexander himself. The last century confirmed this evidence with plentiful finds of texts and artefacts, and particularly the site of Aï Khanoum, perhaps the Hellenistic city of Alexandria on the Oxus, which appeared to transplant Greek material culture to north-eastern Afghanistan, and thrived there until the later second century BC.6 The echoes of Greek cultural influence in Central and South Asia resounded for centuries and the notion that Gandhāran Buddhist art emerged from this thoroughly Hellenized matrix remains attractive. The early explorers of this tradition called it ‘Graeco-Buddhist’ art and that rather awkward, hybrid term still survives, particularly in popular writing on Gandhāra. But already within the early decades of study, problems were recognized in the idea of Hellenistic continuity. There appeared to be a chronological gap between Hellenistic rule in Central Asia and the emergence of Gandhāran religious sculpture (see also Martina Stoye’s contribution to the present volume). The existence and extent of that gap was contested, but we can now be reasonably confident that it is a matter of several generations between the last ‘Indo-Greek’ rulers in Gandhāra itself, in the first half of the first century BC, and the earliest datable narrative sculptures (which come from the Swat Valley). There is a span of about a century and a half between these and the last substantial remains of Greek art and architecture in Afghanistan. Moreover, the earliest Gandhāran sculptures in Swat are the least classical-looking – their linear, abstract style more akin to the sculpture of contemporary India (Filigenzi 2012). Although it is hard to identify the earliest material in the Peshawar Basin, there is some reason to believe that it may have shared these characteristics (Naiki 2019). In short, the works of Gandhāran sculpture which most readily recall Graeco-Roman traditions flourished in a slightly later period, and particularly in the second and third centuries. Indeed, they cannot confidently be placed before about the end of the first century. A ‘missing link’ was once thought to be provided by a very different sculptural form – the so-called toilet-trays or palettes, conceivably libation dishes, which were discovered at Sirkap (Taxila) and other sites and have been much admired, collected, and probably faked, since (Francfort 1979; Falk 2010). The dishes are decorated in a variety of ways, but some of them have manifestly classical imagery and the form of the dishes themselves is very plausibly derived from metal objects produced 6 For a recent critical discussion of the culture of this Hellenistic ‘Far East’ see Mairs 2016. 52 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure in the late Hellenistic eastern Mediterranean (Parlasca 1981; 1983 [non vidi]; Boardman 1994: 1168). Their importance rests on thirty-two provenanced finds from Sir John Marshall’s excavations at Taxila between 1913 and 1930, which he placed between the second century BC and the mid-first century AD, principally towards the end of this range (Marshall 1951: vol. 2, 493-8). However, even assuming that these small objects are considered to have anything to do with the emergence of the tradition of Gandhāran architectural sculpture, their chronology has been seriously challenged in recent years, so that they no longer obviously bridge the gap in evidence for classicizing art in the region (Erdosy 1990; Lo Muzio 2011).7 To be clear, I am not attempting an argument from silence. If not the toilet-trays, then other artforms perhaps, including perishable objects such as textiles, wood-carvings, the highly fragmented tradition of wall-painting, may have transmitted a continuous Hellenistic artistic tradition to or within Gandhāra. Daniel Schlumberger, who regarded the Gandhāran phenomenon as one aspect of a general post-Hellenistic survival of Hellenized art in Asia, persuasively evoked the impact of new discoveries, including the sites of Dura Europos and Surkh Kotal, on our perceptions of ancient artistic geography, and the notion of an (as yet) largely invisible Central Asian genealogy for Gandhāran Buddhist sculpture remains inherently plausible (Schlumberger 1960). Indeed, the pre-existence of a Hellenized visual culture in Gandhāra – a predisposition to see artistic subjects in a classical mode – is helpful in understanding even the Roman imperial influence proposed here. Consequently, I do not wish to argue dogmatically against the idea of Gandhāran art as a Hellenistic legacy, but merely to suggest that the facts force us to adjust our premises and look at the contemporary connections of Gandhāran art. For it is, fundamentally, a product of the period of the Roman Empire and, as we shall see, some of its closest affinities can be found in Imperial sculpture. This was the position adopted by a number of scholars in the twentieth century who dissented from the prevailing Hellenocentric perception of Gandhāran art. Particularly notable contributions were made by Mortimer Wheeler, the famous Roman archaeologist who had served as Director General of the Archaeological Survey of India and later as a consultant to Pakistan’s archaeological service (e.g. Wheeler 1949); the Warburg Institute medievalist Hugo Buchthal (Buchthal 1943; 1945), whose contributions received such vituperative criticism that he did not return to Gandhāran material after the 1940s;8 Benjamin Rowland (e.g. Rowland 1936: esp. 392-5; 1942; 1956); John Rosenfield (Rosenfield 1967); and Alexander Soper (e.g. Soper 1951); not to mention Ananda Coomaraswamy, whose reaction against philhellenic admiration for Gandhāran art led him to dismiss it as a mere imitation or adaptation of Roman provincial art, at that time widely regarded as inferior to Greek art (e.g. Coomaraswamy 1913: 53-4; Bracey 2020: 36-8).9 These contributions had a limited influence at the time, which was no doubt partly due to the immense cultural authority of ancient Greek art, accompanied by a general disdain of Roman art before the 1980s. Roman art was regarded as highly derivative. Because the artistic production of the late Roman Republic and Empire had inherited the Hellenic tradition, adopting the whole package of past Greek styles and A key factor, in this respect, is Erdosy’s down-dating of Sirkap (once supposed the Indo-Greek city at Taxila) to the ŚakaParthian-Kushan periods. 8 Robin Cormack, pers. comm. For an example of the response see Marshall 1946. Even ‘Romano-Buddhist’ advocates like Soper were not entirely sympathetic to some of Buchthal’s arguments (Soper 1951: 303). His particular Achilles’ heal was his excessively late floruit for Gandhāran sculpture. Denying second thoughts on his own part, Marshall concludes (Marshall 1946: 122): ‘I have no shadow of doubt as to the correctness of my chronology. I fear that no amount of wishful thinking on Dr. Buchthal’s part can alter the hard facts revealed by the spade.’ 9 The notion of Gandhāran art as, in some sense, ‘provincial Roman’ recurred in the work of Coomaraswamy and some others and is very intriguing (even if misleading). It implies a vision of Roman provincial culture as almost entirely detached from the geography of the Roman Empire itself. This is in contrast to the tight relationship between political geography and cultural diffusion which dominates in our understanding of Roman art. 7 53 The Global ConneCTions of Gandhāran arT iconographies, it appeared to lack a very clear-cut cultural identity of its own (this has been termed ‘the problem of Roman art’; see esp. Brendel 1979). In respect to its formal characteristics, Roman art was thoroughly imbued with Greek conventions, even if these were put to use for a new, vastly expanded clientele and sometimes for new purposes. This fact has two consequences that continue to be relevant for the study of classical influence in Gandhāran art, one practical and the other psychological. It is usually difficult or impossible to state that a particular motif or formal trait in Gandhāran art is ‘Roman’ in origin rather than ‘Greek’. The motif of the Eros/Cupid holding a garland, for example, which is discussed further below, is common in the art of the Roman world, but originates in Hellenistic Greek art (Bromberg 1988). Is the source of ‘influence’ therefore Hellenistic or Roman? If the former, then this imagery is evidence for the Hellenistic heritage of Gandhāran art, whether directly or via the Hellenized visual koine of the Middle East and Central Asia. In that case Roman and Gandhāran imagery could be regarded merely as art-historical ‘cousins’, sharing a common ancestry in Hellenistic art which would explain the shared visual vocabulary. The second, psychological, obstacle is that, where the origin of a motif is uncertain, a general Hellenocentric bias tends to give priority to Greek art as the default cultural source; ‘Graeco-Roman’ is treated as ‘Greek’. Consequently, the burden of proof has always lain with those who wish to argue for the importance of Roman influence. For descriptive purposes it may indeed be accurate to say that the motif is Greek (i.e. invented by the Greeks). Yet if the Gandhāran artists inherited such a motif as a result of familiarity with the sculptural production of the Roman Empire, for instance, the label ‘Greek’ becomes quite misleading. It may be intended to describe the art-historical ancestry of the image, but subtly assimilates the process of influence to the Hellenistic period, de-emphasizing the particular Roman social, political, and economic context in which the cultural exchange occurred. For example, Boardman, while very open-minded about the possibility of contemporary Roman contacts as a potential factor in the development of Gandhāran art, also dismisses the contrast between Greek and Roman – with some justice – as a semantic problem, ‘rather unreal, given that Greeks are the intermediaries’ in trade (Boardman 1994: esp. 122-3). The problem is that ‘Greek’ influence (as a late Hellenistic legacy) has very different historical implications from ‘Greek’ influence as a contemporary result of the phenomenon of Roman imperial expansion. Notwithstanding these problems, in recent years a broad consensus has developed around the assumption that the Roman Empire is at least part of the story of Gandhāran art. With this openness to the consideration of contemporary global connections, there is the possibility – and need – to examine in more detail what, why, and how the Gandhāran artists might have borrowed from the art of the Roman world. The revisionist authors mentioned above sometimes used specific examples of Roman sarcophagi as part of their attempts to trace Gandhāran classicism to the Roman Empire. In particular, Alexander Soper argued interestingly about the surprising resemblance to the Italian figural sculptures (Soper 1951: 305): I know only one part of the Western world in the century of Kushan magnificence that produced an art comparable in interests and methods to the Gandhāran sculpture analysed above: the west Mediterranean area centering on Rome. In the frieze sarcophagi that made up the bulk of the Roman sculptural output of the second century, there is the same manipulation of the human figure to the exclusion of almost everything else. There is the same delight in producing the illusion of a stage peopled by living actors. In consequence there is a like variety of poses, among which the rear view is exploited as a valuable accent. There is the same crowding of planes between foreground elements that stand out almost in the round, and the background. Soper, however, only touches on a few controvertible examples, while other contributions have been similarly brief. There is scope for considerable further study of this hypothetical connection, for which the present paper aims to offer additional material and directions. My aim in the remainder of the chapter is therefore to 54 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure revisit the connection with Roman sarcophagus manufacture, expanding and refining the sorts of comparisons made by Soper and others, as well as challenging them. I shall review a number of examples of apparent visual resemblances between Roman sarcophagus reliefs and Gandhāran sculptures, questioning a few of them, before turning to more subtle resonances between the two artistic cultures. I shall begin with more generic decorative elements before examining specific narrative scenes, followed by paratactic compositions. Finally, I shall sketch the context in which this putative artistic transmission may have taken place. Iconographical echoes of sarcophagi in Gandhāran reliefs? Cupids and garlands Perhaps the most conspicuously Graeco-Roman motif in the relief decoration of stūpas comes from the friezes of infant boys supporting an extended leafy garland or series of garlands between them. But as we shall see, this iconography has rather different classical antecedents from most of the examples that follow. The figures are often called putti by analogy with the classically-derived infants of Italian Renaissance art, and they clearly imitate, directly or indirectly, the Erotes/Cupids of Graeco-Roman art (Soper 1951: 306, 317 notes 39 and 41; Ingholt 1957: 152-4; nos. 374-80; Boardman 1994: 130-1; Stančo 2013: 116-7, 124-33). In the Gandhāran context these were evocations of veneration and celebration, apparently often surrounding the base or drum of the stūpa, and perhaps prompted by the use of real garlands to adorn these monuments. The general imagery originated in the Hellenistic Greek Mediterranean and was readily adopted into Roman funerary sculpture in the late Republic (e.g. Gatti 2005: esp. 170-3; note also the continuous garland held by cupids on the early imperial Mausoleum of the Julii at Saint Rémy de Provence). The iconography is, however, most abundantly attested in art of the second century AD. Erotes with garlands of leaves, flowers, and fruit were frequently used to decorate Roman marble sarcophagi. Indeed, the ultimately Roman source of this imagery has been accepted even by scholars more generally inclined to scepticism about the role of Roman art in the formation of Gandhāran art (Boardman 1994: 131 – ‘We can be reasonably sure that this motif is a new arrival and has nothing to do with any Bactrian heritage’). Notwithstanding the difference in the structure of the garlands, the components are so close to Roman friezes as to abolish most doubts as to their general ancestry, and sarcophagi in particular present the closest parallels (Figures 1 and 2; cf. Figure 2 in Martina Stoye’s chapter in this volume). In the Roman imperial context, the garlands are usually discreet swags. They are frequently composed from assorted fruits (as if from the contents of a cornucopia) but can be leafy. Some of the tighter, more formal arrangements of laurel-leaves, for example, may inform the dense, rhomboid patterning of Gandhāran garlands which are dominated by foliage or indistinct floral forms. Figures occupy the spaces above each garland: busts and masks, animal faces, eagles, rosettes, often gorgoneia – there was considerable latitude in how the sculptor finished off such elements which were roughed out after quarrying. In Gandhāra, human busts, birds, or winged demigods occupy the equivalent spaces. In both traditions the ‘putti’ hold the garlands with raised hands or over shoulders, with similarly varied, contrapposto poses. However, one routine detail is diagnostically important: in Gandhāran art fruit regularly appears in the form of clusters hanging from the middle of each swag. This motif is rare in sarcophagi made in Roman Italy (see Herdejürgen 1996: 164-5, no. 159, pls. 81.1, 83.1; 169-70, no. 169, pls. 99.2, 103.1), whereas it is ubiquitous in those manufactured in the same period (around the middle to second half of the second century AD) in Asia Minor, which were widely used in the Roman Near East and Egypt (Koch & Sichtermann 1982: 499-500 and regional survey 484-579, with relevant plates). As a result of this distribution, it seems probable that Gandhāran artists were tapping into an iconography endemic in this part of the Roman Empire rather than further west, whether directly from sarcophagus artists or indirectly (Boardman 1994: 131 suggests that it came via non-monumental art and possibly travelling artists). For this reason, I treat the garland friezes as a somewhat different case to those that follow. 55 The Global ConneCTions of Gandhāran arT Figure 1. Part of a Gandhāran garland frieze from Jamālgarhī. London, British Museum, inv. 1880.59 (Photo: copyright the Trustees of the British Museum.) Figure 2. Roman sarcophagus with cupids holding garlands. Found at Tarsus, Cilicia, early third century AD. New York, Metropolitan Museum of Art, 70.1, Gift of Abdo Debbas, 1870 (Photo: Museum, CC0 1.0 licence). It is regrettable that the doubtful and fluid chronology of most Gandhāran sculpture prevents any attempt to make a closer analogy specifically with the eastern sarcophagi, given their relatively well dated floruit (from around the 140s AD). The absence of datable putti-and-garland friezes before this time might have been strongly indicative of a link, but for the fact that so few sculptures before this period are datable. ‘Atlantes’ The so-called ‘Atlas’ figures – atlantes – had a rather similar position within the structure of a Gandhāran stūpa (Figure 3; cf. Yang Juping’s paper in the present volume), but their relationship with the classical world is more complex. Rows of these supporting figures decorated the bases or drums of stūpas, serving partly as conventional architectural devices, like the atlantes and caryatids of western art, and holding up the entablature above them with hand, or shoulder, or head (cf. Vitruvius, De Architectura, 6.7.6). However, like the putti they are also presumably conceived of as minor characters in a hierarchical vision of the Buddhist pantheon. They were well built, naked or semi-naked, winged demi-gods, usually 56 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Figure 3. Gandhāran frieze of ‘atlas’ figures from Sahri Bahlol. Peshawar Museum, inv. 1323. (Photo: after Ingholt 1957: no. 381 [Islay Lyons].) understood as yakṣas (Stančo 2013: 62-81 on the iconography). The name ‘Atlas’ is partly a matter of architectural terminology, but the figures do in fact resemble some representations of the (wingless) giant of that name in Graeco-Roman art, so there is perhaps a loose mythological affinity between these supporting characters and the Atlas of classical legend. They sometimes also echo the muscular physique and facial hair of Herakles/Hercules figures in classical art, as well as late Hellenistic or Roman Imperial crouching barbarians. Other atlas figures are more distinctively Gandhāran or Indian, sometimes with bestial, satyr-like faces, or youthful features, or more elaborate and idiosyncratic hair, wearing long locks and moustaches. Many – by no means all – of the atlantes are among the most naturalistic and classical-looking figures of Gandhāran sculpture. The classical Atlas begins to be represented routinely in the form of such muscular, crouching, supporting figures in the Roman imperial period.10 Significantly, perhaps, we encounter them as feet for some of the earliest Roman metropolitan sarcophagi, around the 130s AD (Figures 4 and 5).11 The figures on these funerary monuments have the same structural (and hierarchical) role as the atlantes of Gandhāran stūpa sculptures. They are by no means typical; they were probably by the same artists and were found together with a cupid-and-garland sarcophagus in the same mausoleum near Rome’s Porta Viminalis. But a little later we can see how their role could be elaborated in a famous mid- to late-second-century sarcophagus from Velletri in Italy, on which a series of atlantes ‘hold up’ the lowest register, providing part of the architectural framing for a mythological tour de force (Velletri, Museo Civico; Lawrence 1965; Thomas 2011: 399-404) (Figure 6). The virtuosic Velletri sarcophagus is quite idiosyncratic and cannot be taken to exemplify any broader trends, but with its multi-tiered structure, populated niches, and cupids holding garlands of fruit on the roof, it rather uncannily mirrors the setting of supporting figures in stūpa decoration, as represented for instance on the miniature stūpas at Jauliāñ, near Taxila (Figure 7; cf. Figure 9 in Ball’s contribution to this volume). The architectonic logic involved is analogous in the two traditions.12 10 For an association with the representation of subject barbarians, as exemplified for example on a late Hellenistic stele from Erythrai, now in Munich, Glyptothek, inv. Gl. 509, see Schneider 1986: esp. 28-9, 45-50 (stele at 29 n. 88). 11 For the context and the accompanying garland sarcophagus (a satyr holding the garlands as well as the cupids) see Herdejürgen 1996: 37-39; 126-7, no. 78, pls. 10.5, 11.2, 32.3, 35.1-3, 34.1-3, 112.1 and 3. 12 Musei Vaticani, Museo Gregoriano Profano, inv. 10450 (Orestes) and 10437 (Niobe); Benndorf & Schöne 1867: 286, no. 415 and 296, no. 427; Robert 1890: 168-71, no. 155, pl. 54; Robert 1919: 381-3, no. 315, pl. 100; Zanker & Ewald 2012: 376-80, 372-4. A third, contemporary sarcophagus from Rome with a relief of the triumph of Dionysos, now in the Museo Civico in Verona, also has satyr-like atlantes on its feet, but they are carved in a different marble and their antiquity has been doubted; Matz 1968: 198, no. 83, pls. 106.1-2, 107.1-2, 108.1. 57 The Global ConneCTions of Gandhāran arT Figure 4. Roman sarcophagus with scenes from the myth of Orestes, from a mausoleum near the Porta Viminalis, Rome, c. 130s AD. Musei Vaticani. (Photo: agefotostock/ De Agostini/G. Nimatallah.) Figure 5. Roman sarcophagus with scenes from the myth of Niobe; same site and date as Figure 4. Musei Vaticani. (Image: after Robert 1919: pl. 100; copyright Universitätsbibliothek Heidelberg, Open Access <https://digi.ub.uni-heidelberg.de/diglit/ asr3_3/0233/image>.) 58 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Figure 6. Roman sarcophagus from Velletri, c. mid- to late second century AD. Velletri, Museo Civico. (Photo: Artokoloro/Alamy.) Figure 7. Stone and stucco subsidiary stūpa (D4) at Jauliāñ. Third-fifth century AD. (Photo: copyright Teseum <https://www. flickr.com/photos/teseum/43329938785>, CC BY-NC 2.0.) 59 The Global ConneCTions of Gandhāran arT The Wooden Horse Let us now turn from such relatively generic figures to specific narrative compositions. In at least one rare case a narrative which is to be encountered on a few Roman sarcophagi is imported relatively unchanged into the Gandhāran repertoire. The myth of the Wooden Horse, which was used to smuggle Greek warriors into the long-besieged city of Troy, is attested in two Gandhāran reliefs. One is a tiny fragment published in 1990 while in a private Pakistani collection and reportedly found near Pitao, a few miles north-east of Dargai, on the north of the Peshawar Basin (Khan 1990).13 This includes a crude figure similar in dress and pose to representations of the Trojan priestess Cassandra in Roman wall-paintings (Sparkes 1971: 66; LIMC 3.1: 813-7; 3.1: 589-92),14 but otherwise departs from known classical iconography in that a man emerges from the opened neck of a horse on a wheeled platform, as if unzipping himself from a costume. More enlightening – but unusual – is a small relief in the British Museum, which was found near Hund on the Indus (Figure 8; British Museum, inv. OA 1990.10-13.1; Zwalf 1996: vol. 1, 233-4, no. 300; Stewart 2016, with full bibliography at 6, n. 16). In this much discussed scene, the wheeled Wooden Horse is clearly represented in front of the walls of the city. At least three male figures accompany the horse towards the gate. Another opposes its progress, lunging at it with a spear. In the classical myth this role belongs to the priest Laocoon who, like Cassandra, foresees the trouble that will follow from the Greeks’ trickery. A half-naked female, adorned with jewellery in the Indian manner, stands in front of the gateway, raising her arms in distress. She ought to be Cassandra, but may have been metamorphosed into a city-goddess or another character in this relief. Assuming that this and the other fragment were used to adorn Buddhist monuments (the Hund relief is probably part of a stair-riser from a stūpa), the story of the ‘Trojan Horse’ has presumably been reinvented as a Buddhist narrative, as Alfred Foucher suggested (Foucher 1950). Figure 8. Gandhāran relief with scene of the ‘Wooden Horse’, from Hund, c. second century AD (H. 16.2 cm). British Museum, inv. OA 1990.10-13.1. (Photo: copyright the Trustees of the British Museum.) 13 14 Khan 1990. Naples, Museo Nazionale Archeologico, inv. 9040, 9010. 60 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Figure 9. Detail from a drawing of the Tabula Iliaca Capitolina by Feodor Ivanovich. Marble relief found at Marino, near Rome, c. first century AD. (Image after Theodor Schreiber, 1895: Atlas of Classical Antiquities, London and New York, Macmillan, pl. XCIII, courtesy of <www.mediterranees.net>.) Figure 10. Detail of a Roman sarcophagus lid with scenes from the Trojan War, c. late second century AD. Oxford, Ashmolean Museum, inv. AHMichaelis.111. (Photo: copyright Ashmolean Museum, University of Oxford.) What is especially interesting here is that both the narrative and the specific imagery employed to represent it have been appropriated from Graeco-Roman culture. The figures, while simple, are rendered in a naturalistic style, with poses that recall classical art (including that of the otherwise Indianized ‘Cassandra’). The ‘Laocoon’ figure at least wears a classical tunic and cloak. Moreover, the closest parallels anywhere for the composition itself are Graeco-Roman representations of the Trojan Horse story, both of them made in the Roman Empire. One is the roughly first-century Tabula Capitolina Iliaca, a virtuoso miniature sculpture showing the whole story of the Trojan War, found at Marino near Rome, which includes two scenes of the Horse (Figure 9; Rome, Museo Capitolino, inv. MC 316; Sadurska 1964; Squire 2011). The second comparandum is a later second-century sarcophagus lid in Oxford (Figure 10; Michaelis 1882: 566-88, no. 111; Robert 1890: 73-5, fig. 64). The compositional relationship with the sarcophagus lid is especially close, to the extent that such scenes on sarcophagi may be the ultimate source for the Gandhāran artist. We will return to the comparison below. However, it is important to note that the subject is uncommon and variable on extant sarcophagus reliefs, and the Oxford relief is uniquely close to its Buddhist relative, so in every respect this is a peculiar case.15 15 Only a small handful of Wooden Horse reliefs is attested, of which the next closest is a lid relief fragment in (or formerly in) Berlin: Robert 1890: 75, fig. 65; cf. drawing of it in the British Museum, inv. 2011,5012.46. The main emphasis in sarcophagus iconography is on the violent aspects of the sack of Troy (Zanker & Ewald 2012: 74-6, 350-3). 61 The Global ConneCTions of Gandhāran arT The Great Departure and the adventus/profectio In his argument for the influence of Roman sculpture on Gandhāran art, Hugo Buchthal drew attention to the remarkable compositional similarity between Siddhārtha’s ‘Great Departure’ on horseback, as he renounces his princely lifestyle and leaves the palace, and Roman scenes of imperial adventus and profectio – triumphal arrivals and departures at cities – as commemorated on coins from the mid-first century onwards (Figures 11 and 12; Buchthal 1945: 12-3). He observed that elements of the narrative existed in earlier, Indian art, as in sculptures from Sanchi and Bharhut, but that the compositional feel of the Gandhāran scenes was informed by Roman art. He further illustrated the cross-cultural resonance with an ancient Christian sarcophagus relief representing the arrival of Christ into Jerusalem, which was considered to be derived from the imperial scenes – a religious appropriation of the imagery analogous to that of the Gandhāran Buddhists (Figure 13). There is certainly an echo among these different narratives, in the pose of the horses and their riders, the riders’ hand gestures, and the various attendants or other figures which crowd around the emperor/Christ/Siddhārtha (including sometimes prostrate barbarians beneath the hooves of the emperor’s horse). In some cases, a soldier holding a standard behind the emperor particularly recalls the umbrella-holding servant behind Siddhārtha, for instance. The resemblance is rather stronger with the Christian scenes. Here the yakṣas who lift the feet of the horse, Kanthaka, recall the diminutive youth who honours Christ’s arrival by placing a garment beneath the front legs of the colt. The disciples behind him look like the prince’s servant, or perhaps the figures of Brahmā and Mārā(?) in front of the future Buddha in the Gandhāran scenes. Is this coincidence or influence? The similarity is very striking, but the confirmation bias that affects such comparisons of composition is illustrated by a debate about the genealogy of the Roman iconography itself. Thomas Mathews has convincingly challenged the consensus that the Christian scenes are derived directly from the imagery of secular power, pointing out that a much closer analogy for Christ’s entry is provided by the private ‘adventus’ scenes showing wealthy Romans returning to their estates from the Figure 12. Reverse of an aureus of the Emperor Trajan showing the emperor’s profectio, AD 114-117. New York, American Numismatic Society, inv. 1944.100.43617. (Photo: ANS, public domain <http://numismatics.org/ collection/1944.100.43617>.) Figure 11. Gandhāran relief showing the Great Departure of Siddhārtha. From Loriyān Tangai. New Delhi, National Museum. (Photo: National Museum, from Google Arts & Culture <https://g.co/arts/Efhe596csyAaJgS29>.) 62 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure hunt (Mathews 1999: 24-7; Jensen 2015: 24-33 on the issue). For example, the small figure who watches Christ’s arrival from a tree, taken to be the Zacchaeus mentioned in the Gospel of Luke 19:1-6 (a motif ingrained in the iconography for centuries to come), is descended from the harvester (or harvesting cupid) who was represented in the pagan scenes of the nobleman’s homecoming. Significantly, for our purposes, Mathews illustrates this iconography with a sarcophagus of around AD 300 in the Bode Museum (Figure 14; Bielefeld 1997: 101-2, no. 13, pl. 37.1-5). But this is not common sarcophagus imagery and our evidence from both Christian and non-Christian sarcophagi is relatively late. Ultimately it seems more plausible to suppose that the classical ancestry for the Gandhāran compositions lies in a broader range of compositionally similar Roman scenes which share the same structural logic, representing elevated figures trotting by on horses, obsequiously attended by servants or supportive bystanders. To the tradition of adventus scenes and Christ’s entry into Jerusalem we might add the funerary reliefs made in Italy to commemorate young equites (i.e. ‘knights’, or members of the upper-class equestrian order), such as the monument for Titus Flavius Verus erected by his mother in third-century Ostia (Figure 15; Musei Vaticani, Museo Gregoriano Profano, inv. 10659; Benndorf & Schöne 1867: 381-2, no. 545). An unresolved question remains about when the Great Departure in this composition emerged in Gandhāra. A case can be made for dating one fragment from Ranigat early, perhaps in the first century AD, because of its stylistic associations with other pieces, including those thought to be the earliest at this site (Naiki 2019: 48, fig. 14). But the sculpture has no useful context of its own and shows only the lower part of the scene. If its dating was more convincing it would provide additional evidence against a specific source in the repertoire of Roman sarcophagi. Figure 13. Detail of Roman sarcophagus relief showing Christ’s entry into Jerusalem. From the Vatican area of Rome, fourth century AD. Musei Vaticani, Museo Pio Cristiano, inv. 31461. (Photo: University of Michigan Library: Art Images for College Teaching/Allan T. Kohl, CC0.) 63 The Global ConneCTions of Gandhāran arT Figure 14. Roman sarcophagus relief including the return of a man from the hunt. Found at Vicopisano. Made around the end of the third century AD. Berlin, Bode Museum (Skulpturensammlung und Museum für Byzantinische Kunst), inv. 6596. (Photo: SMB-digital [Rosa Mai], CC BY-NC-SA 3.0 DE licence, by permission). Figure 15. Funerary relief of the eques Titus Flavius Verus. From Ostia, c. end of the second century AD. Musei Vaticani (Museo Gregoriano Profano, inv. 10659). (Photo: Victoria and Albert Museum, London; by John Henry Parker, c. 1864-8.) The First Bath of the Buddha and Prometheus Another key scene from the life of the Buddha has been proposed as an adaptation of classical iconography. Martina Stoye has drawn attention to Graeco-Roman comparisons for the tripod table on which the new-born Siddhārtha stands, and she has suggested how religious connotations may have been transferred from the classical iconography (Figure 16; Stoye 2004; she elaborates on these arguments in her contribution to the present volume, asking whether the transmission of iconography is explicable through Hölscher’s concept of an ancient image-language). In addition to the compelling comparisons she makes, or even as an alternative source of inspiration, we might point to an iconography used for mythological sarcophagi. This is the imagery of the creation of the first man by Prometheus.16 16 See Robert 1919: 436-449, pls. 117-118; Koch & Sichtermann 1982: 183-4, with list of examples 183 n. 1; more generally, LIMC 7.1, 531ff; 7.2, pls. 420-30. 64 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Figure 16. Roman Prometheus sarcophagus from Arles, c. AD 240 (reused as the tomb of Saint Hilary of Arles). Paris, Musée du Louvre, inv. MA 339. (Photo: Bildarchiv Foto Marburg, <www.fotomarburg.de>.) Figure 17. Fragment of a Roman sarcophagus relief, c. late second century AD. Madrid, Museo del Prado, inv. E000140. (Photo: copyright Museo Nacional del Prado.) Prometheus appears like a sculptor on at least nine sarcophagus reliefs, creating the first man out of clay with the aid of the goddess Athena (Figure 16).17 The diminutive human figure is usually on a pedestal, and in one case he stands on a tripod.18 On some sarcophagi Athena endows man with a soul in the form of a butterfly (Figure 17).19 The compositions recall the First Bath of the Buddha with its assembly of divinities reverently attending the proceedings, Indra and Brahmā pouring water over the child (Figure 18).20 It is notable that on the Louvre sarcophagus there are also child-like humans wrestling and playing 17 18 19 20 Paris, Louvre, inv. Ma 339, c. AD 240; Baratte & Metzger 1985: 115-118, no. 47. Paris, Louvre, inv. Ma 445, c. third century. For the example in the Prado see Schröder 2004: 493-496 (non vidi). Peshawar Museum, inv. 2071; Ingholt 1957: 53, no. 16. 65 The Global ConneCTions of Gandhāran arT Figure 18. Gandhāran relief showing the First Bath of the future Buddha. Peshawar Museum, inv. 2071 (Photo: after Ingholt 1957: no. 16 [Islay Lyons].) (lower middle and right) which are reminiscent of Gandhāran scenes of another incident in the early life of Siddhārtha – the wrestling competition (e.g. Peshawar Museum, inv. 1906; Ingholt 1957: 56, no. 30). Yet intriguing though this comparison is, the subject is relatively uncommon on extant sarcophagi, none of them very early (they begin in the late second century), so while a classical origin is suggested there is not yet a compelling argument for sarcophagi in particular offering a source of ideas to the Gandhāran artists. Śibi and Marsyas A similar case is presented by the Śibi Jātaka, one of the stories of the Buddha’s previous lives which entered the Gandhāran repertoire of stūpa decoration, at least in a small way.21 The story recounts how the bodhisattva, as King Śibi, exhibited unlimited compassion by sacrificing his own flesh to save the life of a dove from a hawk. In the famous relief in the British Museum he is represented patiently sitting with support from the queen while a servant crouches to flay his leg with a large knife (Figure 19). Another weighs it out in the divine presence of Indra and Brahmā. A very similar motif appears in Roman scenes of the story of the satyr Marsyas, who was rewarded with flaying at the hands of a Scythian slave after unsuccessfully competing with the god Apollo in a musical contest (see e.g. Apollodorus, Bibliotheca, 1.4.2; Ovid, Metamorphoses, 6.382-400). The horrific scene is represented in various media, including sculpture in the round, but the Gandhāran imagery particularly echoes the composition of Marsyas sarcophagi (Figure 20).22 In the example illustrated here, a young nature-god stands by watching from 21 22 At least two secure examples are known in relief: Neelis 2019: 178, nos. 153-4; 181. Palazzo Doria Pamphilij example: Robert 1904: 259-60, pl. 67. 66 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure the right, in a pose and dress strikingly similar to the bystanders on the British Museum relief (Robert 1904: 259-261, pl. 67, fig. 207). Around thirty sarcophagi with this iconography are extant, so this is among the more popular mythological scenes on metropolitan Roman sarcophagi, although many of these present Marsyas as a minor element within scenes of the Muses (Robert 1904: 242-267, pls. 64-69; Koch & Sichtermann 1982: 158-9; Zanker & Ewald 2012: 235-236). The Śibi scene is rare in Gandhāra, but the compositional resonances are nevertheless compelling. The ascetic and the herdsman Figure 19. Gandhāran relief of the Śibi Jātaka. London, British Museum, inv. 1912,1221.1 (Photo: courtesy of the Warburg Institute, London.) Some narrative motifs were easily transferable between stories. A commonly recurring figure in Gandhāra is the seminaked ascetic who sits on a makeshift seat within a simple booth constructed from bundles of reeds or similar vegetation. Figure 21 is a typical example, showing in this case the Buddha visiting an ascetic with Vajrapāṇi, but the same figure assumes a variety of guises in Buddhist iconography (cf. Figure 34).23 His pose and the construction of the shelter are so close to Graeco-Roman imagery as to leave no doubt about their ultimate ancestry. The motif had no doubt been used across a variety of classical media. Indeed, a somewhat similar, if more elaborate, hut appears in the riverine landscape of the Figure 20. Drawing of the punishment of Marsyas scene on the side great Nile Mosaic at Palestrina (probably of a Roman sarcophagus in the Palazzo Doria Pamphilij, Rome. From c. 100 BC). But in this form we commonly the Via Aurelia at Rome, c. AD 230. (Image: after Robert 1904: pl. 67; encounter it on Roman bucolic sarcophagus copyright Universitätsbibliothek Heidelberg, Open Access <https:// digi.ub.uni-heidelberg.de/diglit/asr3_2/0236>.) reliefs, particularly favoured in the third and fourth centuries, and sometimes as a comparatively marginal, genre element in mythological reliefs that have a pastoral setting (Koch & Sichtermann 1982: esp. figs. 124, 125, 126). The herdsman on his own, resting on a rustic seat with crossed legs and an animal nearby, often milking, is a still more common and enduring figure in classical bucolic art (Himmelmann 1980: 124-9, pls. 60, 62b, 73). Indeed, he is a regular part of the cast in reliefs of the moon-goddess Selene visiting her sleeping, mortal lover, Endymion (Robert 1897: 53-111, pls. 12-25; Koch & Sichtermann 1982: 144-146, figs. 156-61) – one of the very most common subjects on metropolitan sarcophagi. The sarcophagus illustrated here is a relatively late and very high-quality 23 For the relief in Figure 21, perhaps standing for Siddhārtha’s first see meeting with the Brahmans, see Ingholt 1957: 63, no. 53. 67 The Global ConneCTions of Gandhāran arT Figure 21. Gandhāran relief representing the Buddha meeting an ascetic in his hut. Peshawar Museum, inv. 2066. From Hoti Mardan. (Photo: after Ingholt 1957: no. 54 [Islay Lyons].) Figure 22. Roman sarcophagus of Iulius Achilleus: a bucolic scene in relief. Rome, Museo Nazionale Romano, inv. 125802. From the city of Rome, c. 260s AD. (Photo: Bildarchiv Foto Marburg, <www.fotomarburg.de>.) exemplar of a tradition that extends back to second-century reliefs: the later third-century sarcophagus of Iulius Achilles (Figure 22).24 It amply demonstrates how close the posture and dress of the ascetic and herdsman could be, and how similar in shape and construction the booth of bound reeds or wattle. Muses sarcophagi and Buner reliefs We could continue in this vein, gathering more or less suggestive similarities between Gandhāran reliefs and the narratives on Roman sarcophagi, but a common theme has emerged from the arguments involved. In general terms, the above examples reinforce the impression of a relationship between the traditions of Roman Imperial art and Gandhāran stūpa sculptures, sometimes revealing remarkable echoes. However, this does not in itself amount to proof of a direct relationship with Roman sarcophagi. Against such a relationship is the relatively late chronological centre of gravity of some of the Roman iconographical conventions, mainly around the later second century AD. That is not necessarily an impediment, since the paucity of dated Gandhāran evidence usually prevents us from demonstrating that a particular iconography was embedded there much earlier, but it hardly favours the argument. Similarly, the comparative rarity of some of the scenes, either among Roman sarcophagi or Gandhāran works, undermines the notion of substantial communication of visual ideas, even if this may be due to the accidents of survival. It is helpful at this point to shift attention to more subtle, but in certain respects more persuasive, illustrations of this putative relationship. It will already be clear that we are not dealing with some straightforward imitation of foreign models on the part of Gandhāran artists, but with a more thorough assimilation and transformation of the classical iconographical and stylistic repertoire, which sometimes makes the sources seem elusive. The classical art historian considering Gandhāran art is repeatedly struck by a feeling of déjà vu rather than an instant recognition of Greek or Roman models: we might 24 Giuliano 1979: 312, no. 187. 68 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure think of them metaphorically as family resemblances manifesting a shared artistic DNA, even if the explanation for the genetic connections is obscure. This phenomenon is evident in the compositional patterns transmitted from Graeco-Roman art to Gandhāran scenes, which can be both very approximate and, just sometimes, implausibly precise. This is illustrated by the arrays of standing figures that characterize the so-called Buner reliefs (Figures 23 and 25).25 These small reliefs (named after the provenance of some of them in the northern part of the Peshawar Basin), are stair-risers from stūpas. They are among the most expressly ‘classical-looking’ sculptures of Gandhāra (Rowland 1956; Ingholt 1957: 160-161, nos. 411-415). They feature paratactic compositions of more or less frontal, male and female figures. Their identity is generally ambiguous. They are presumably Buddhist worshippers, and the presence of wine-cups in some reliefs suggests celebratory ritual. The dress of the figures often appears classical in inspiration. It is hard to judge just how contemporary or realistic this might have appeared to Gandhāran viewers. In any case, the participants are dressed in tunics and mantles resembling the Graeco-Roman himation, and they regularly wear them in a classical manner. The gestures and deportment are Graeco-Roman too, and the figures stand in classical contrapposto. As others have observed, these paratactic galleries are a common feature of Roman art in various media, as well as earlier imperial sculpture (Rowland 1953: 77, with n. 17 citing a silver cup from the Marengo Treasure in Turin; Rowland 1956), but we encounter them most commonly in sculptural reliefs, and notably on sarcophagi representing the Muses (Wegner 1966; Koch & Sichtermann 1982: 197-203, figs. 260-266). The latter are significantly larger than the Gandhāran reliefs, but they show similar frontal compositions of these deities of the arts, accompanying the god Apollo and sometimes Minerva/Athena (Figures 24 and 26). Muses are among the most common subjects on second- to third-century, metropolitan Roman sarcophagi, with more than 200 extant examples. Always allowing for the difference in scale, the broad compositional similarity is interesting and suggestive in itself. What they appear to reveal is a stylistic habit of naturalistic figural representation that is deeply ingrained in the Gandhāran sculptors’ practice. However, a closer look at the Muses in particular reveals surprisingly specific correspondences in gestures and poses. For example, direct comparisons could be drawn between the poses of figures in Figures 23 and 24.26 Starting on the left, each relief has: i. Standing figure, facing right, with crossed legs, left arm crooked. ii. Female standing figure in contrapposto, weight on left leg; head in three-quarters view angled to proper right; wearing long tunic and himation worn in Graeco-Roman manner, with ‘arm-in-sling’ pose. iii. The next two Gandhāran figures do not closely correspond (although they are nevertheless very classical in form). The fifth figure in each case stands in contrapposto, wearing a tunic and himation; head directed aside in three-quarters pose; right elbow bent to raise hand to gesture at chest-level. iv. Sixth Gandhāran figure matches the tenth on the sarcophagus in gender, pose, gestures, and dress, except that her himation is draped at waist level rather than across the chest and she is angled a little more to the viewer’s left. v. The furthest Gandhāran figure on the right of the fragment corresponds to the eighth on the sarcophagus inasmuch as they are both in profile facing to the right, shoulders in three-quarters view. 25 26 On Figure 23 see Zwalf 1996: 302. Figure 25: Ingholt 1957: 161, no. 414. Figure 24: Germoni 2009: 400-404. 69 The Global ConneCTions of Gandhāran arT Figure 23. Gandhāran stair-riser relief of standing figures, from Takht-i-Bāhī. H: 13.7 cm. London, British Museum, inv. 1900,0414.13. (Photo: copyright the Trustees of the British Museum.) Figure 24. Roman sarcophagus with a relief of Muses, found at the Isola Sacra necropolis, Ostia, in 2008. Made c. late second century AD and used for the burial of a young child. H: 44 cm. Museo Ostiense, inv. 59954 and 59955. (Photo: copyright Eric Vandeville/akg-images.) Figure 25. Gandhāran stair-riser relief of ‘donors’. Peshawar Museum, inv. 24 (Photo: after Ingholt 1957: no. 414 [Islay Lyons].) Figure 26. Detail of a Roman sarcophagus relief showing the Muses (known in the city of Rome in the sixteenth century). Late second century AD. Vienna, Kunsthistorisches Museum, Antikensammlung, inv. I 171. (Photo: copyright Kunsthistorisches Museum, Vienna.) 70 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure An even more surprising echo exists between the central parts of a Buner relief in Peshawar and a Muses sarcophagus in Vienna (Figures 25 and 26).27 It is as if the pivotal figure of the goddess Athena has been metamorphosed into the Gandhāran turbanned figure who turns leftwards with a bunch of foliage in his left hand. They have a similar crooked right arm, hand on hip; their mantles wrap around their waists and hang on the left arm in a similar manner; although the male figure is bare-torsoed, his necklace recalls the sagging folds of Athena’s tunic above her breasts; the shape of the headdress even distantly recalls Athena’s helmet. Meanwhile the figures on the left side of the photographs have the same crooked right arm and rightwards gaze; those on the right stand in a similar contrapposto with bent left arm. The Muses sarcophagi themselves are very diverse in the details of poses and costume, and the correspondences illustrated here with selected examples do not apply to the whole corpus (although some other sarcophagi come rather closer in particular details). These might be remarkable coincidences generated by the Gandhāran artists’ deep familiarity with Roman artistic conventions in general, rather than the specific imitation of galleries of the Muses. They do, however, strongly imply a rather closer relationship with the tradition to which the sarcophagi belong than that illustrated by the narrative scenes above. Sarcophagi as proxies or direct sources? But what sort of relationship? Indeed, it is highly unlikely that any Roman marble sarcophagi themselves reached Gandhāra. They could, of course, travel long distances by sea; many sarcophagi made in Asia Minor or Greece were imported to the city of Rome, sometimes partially carved, to be used alongside those made in Italian marble. Italy made and exported its own monuments too. Yet, allowing for the fact that we have vanishingly little evidence for elite burial practices in Gandhāra in our period (De Marco 1987; Olivieri 2019), there is no reason to think that sarcophagi were in demand and no evidence whatsoever for their acquisition. So we would have to ask by what means the imagery that we know best from Roman sarcophagi might have been conveyed to the Gandhāran artists. At first sight, the most plausible explanation is that the extant sarcophagi are merely representative of a larger classical visual culture, which is more sporadically attested in other forms. Sarcophagi amount, after all, to the single largest corpus of Roman marble sculptures. From this perspective, we ought to regard the coincidence between sarcophagus imagery and Gandhāran scenes as a proxy for a wider complex of influence. Indeed, we have some evidence to suggest that the narrative reliefs on sarcophagi were synthesizing images from discrete mythological vignettes which had been in use in sculpture or other media before the production boom. Roman sarcophagi started to be made in substantial numbers only in the first half of the second century AD, and their rich repertoire of mythological scenes appears to be a sudden innovation just as much as the narrative imagery of Gandhāran Buddhism, but its antecedents can sometimes be discerned. For example, a scene of the hero Orestes leaving the sanctuary at Delphi, which appears on Orestes sarcophagi, is attested as early as mid-first century on the Sebasteion at Aphrodisias in Asia Minor (Smith 2013: 246-248, D34; Smith 2019: 161-164). A scene on several sarcophagi perhaps to be identified as Achilles ordering the sacrifice of Trojan prisoners, is parallelled both on a silver cup from Manching and in a plaster-cast – presumably moulded from an identical silver vessel – found at Memphis (Strocka 2015). It is likely that plaster-casts taken from silver relief decoration were important among the various portable vehicles of imagery that could have served as sources for the sarcophagus-sculptors (Froning 1980). 27 For Figure 25 see Ingholt 1957: 161, no. 414. 71 The Global ConneCTions of Gandhāran arT In light of these observations, it will be helpful to consider examples of imagery that are more exclusively associated with sarcophagi, and which argue for Gandhāran sculpture’s more directly link with that tradition. Death-bed scenes One of the most commonly produced narrative scenes in Gandhāran art was, unsurprisingly, the Mahāparinirvāṇa – the scene of the death of the Buddha (Ingholt 1957: 92-94, nos. 137-142). The composition was remarkably repetitive and consistent, allowing for variations in detail (to the extent that it would be possible Figure 27. Gandhāran Parinirvāṇa scene. Berlin, Ethnologisches Museum, inv. I 80. to organize the extant scenes into (Photo: courtesy of the Warburg Institute, London.) types and sub-types, as has been done for sarcophagus reliefs and other Roman sculptures). Certain elements are consistent (Figures 27 and 30). The enrobed Buddha lies on his right side on a couch with elaborate, turned legs and a sheet falling from beneath the mattress, like a valance. It is flanked by trees (sometimes occupied by their grieving divinities) and surrounded by a crowd of distressed followers, containing a mixture of noblemen and disciples with shaven heads. The strong-man bodhisattva Vajrapāṇi is among them with his vājra. These mourners raise their arms in alarm or pluck or beat their heads in anguish. In front of the couch are two models of emotion: one disciple sits in impassive meditation (viewed either frontally or from behind). Another companion Figure 28. Roman sarcophagus with representation of a young girl’s death-bed and mourners. Later second century AD. London, British Museum, inv. 1805,0703.144 (Photo: copyright the Trustees of the British Museum.) 72 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Figure 29. The Roman Portonaccio Sarcophagus. From Portonaccio, north of Rome, c. 180s AD. Rome, Museo Nazionale Romano (Palazzo Massimo), inv. 112327. (Photo: Deutsches Archäologisches Institut, Rome, D-DAI-ROM-2006.1365.) Figure 30. Gandhāran relief with Parinirvāṇa scene from Loriyān Tangai. Kolkata, Indian Museum. (Photo: courtesy of the Warburg Institute, London.) 73 The Global ConneCTions of Gandhāran arT collapses sideways in distress, like one of the fallen warriors of classical art. The scene corresponds quite closely to the traditional textual accounts of this pivotal moment (note esp. the Pali Mahāparinibbāna sutta, DN 16). The contrast in emotional response and physical self-control among the bhikkus is made explicit. But some of the compositional choices probably derive from the Graeco-Roman iconographical tradition. Death-bed scenes were not uncommon in Roman art, where they appear exclusively in funerary contexts. The scenes are both mythological – the death of the hero Meleager or Patroklos – and ‘realistic’.28 The latter are particularly relevant here, because the typical Roman death-bed scene is an image of mourning, in which expressive, grief-stricken mourners surround the body of the deceased woman or child on a funerary couch. They tear their hair and raise arms in the air in an uncontrolled way. These scenes appear only on Roman sarcophagi and other funerary reliefs (Figure 28).29 Their similarity to the Gandhāran sculptures – which share some of the same gestures, the compositional structure, and details such as the decoration of the couch – hints that the artists may have drawn specifically on the sarcophagus repertoire. General compositional similarities Figure 31. Fragment of a Gandhāran figure of Hārītī. Oxford, Ashmolean Museum, EA1997.3. (Photo: copyright Ashmolean Museum, University of Oxford.) The compositional tendencies of some Roman sarcophagus reliefs are echoed in those of Gandhāra, and this also may be a more compelling relationship than that of the narrative scenes. There is, particularly, a fondness for the crowding of figures in reliefs where the height of the representational field permits it. The effect can be dazzling in both Roman and Gandhāran sculpture, presenting a virtuoso display of interacting and overlapping figures which requires a visual effort to disentangle (compare, for example, Figure 29, the famous Portonaccio Sarcophagus, and Figure 30, a Parinirvāṇa scene from Loriyān Tangai). This is not to claim that such is the dominant style in either tradition, and there are significant Figure 32. Detail from a fragment of a mythological(?) scene on a Roman sarcophagus, c. mid-third century AD. London, British Museum, inv. 1958,0202.1 (Photo: author.) 28 In contrast, the so-called Totenmahl or funerary banquet scene, which is among the most common subjects in Hellenistic and Roman provincial funerary art, shows the deceased as a ‘living’ diner. Interestingly, although this iconography has no purpose and is not adopted in Gandhāran Buddhist art, it is used on some ‘toilettrays’: Marshall 1951: 494-5, nos. 63-4, pl. 144. 29 For Figure 28 see Walker 1990: 17-18, no. 6. 74 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Figure 33. Lid of a Roman Endymion sarcophagus with decorative end panels (Helios and Selene) and panel for inscription. Made in Rome c. 230s AD and found near Bordeaux. Paris, Musée du Louvre, inv. Ma 1335. (Image: after Robert 1897: pl. 18; copyright Universitätsbibliothek Heidelberg, Open Access <https://digi.ub.uni-heidelberg.de/diglit/asr3_1/0195/image>.) Figure 34. Gandhāran scene of the Syāma Jātaka, with ascetics’ huts and decorative end panel. Peshawar Museum, inv. 1891. (Photo: after Ingholt 1957: no. 5 [Islay Lyons].) technical differences, inasmuch as Roman sculptors undercut the figures, whereas Gandhāra artists were usually preoccupied with the outermost plane of the relief and simply projected their figures from a relatively consistent background plane (Pearson 2010: esp. 9-12). Smaller groups of overlapping figures are more common in Gandhāran art, frequently followers of the Buddha or witnesses to his actions, like a chorus to the narrative. Clusters of figures are typically represented obliquely, fanned out to overlap one another on different registers. It is a compositional approach repeatedly encountered in Roman art, and not exclusive to sarcophagi.30 Yet the most striking similarity is with the sarcophagus reliefs. Moreover, the gestures and facial expressions are sometimes very close to those of the Gandhāran bystanders (Figures 31 and 32; cf. Fig. 2a&b in the chapter by Haynes et al. in the present volume).31 Sarcophagus lids We have repeatedly looked at the long, low reliefs that characteristically decorated stair-risers on Gandhāran stūpas. In their format these are closer to the sculpted lids of Roman sarcophagi than to the main chest, so it is interesting to note that they have compositional similarities to the sculptures on the lids.32 It is perhaps these more general resonances that are most persuasive of all in demonstrating an 30 Indeed it may be a distinctively graphic style. Note especially a c. sixth-century AD fragmentary drawing on papyrus in the Museo Egizio, Florence (Museo Archeologico Nazionale), which appears to shows Christ with a crowd of shaven-headed adherents in overlapping three-quarters poses; but for the cross in his nimbus it might as well be a Buddhist scene. 31 For Figure 31 see: Jongeward 2019: 168, no. 137; Figure 32: Walker 1990: 59, no. 75. 32 This is to say nothing of the similarities in subject matter. Gandhāran stair-risers sometimes favour less overtly religious scenes, whose superficially secular subject-matter reflects common themes in Roman art, e.g. playful hunting figures and ‘Dionysiac’ scenes. 75 76 Figure 36. Gandhāran stair-riser reliefs from Jamālgarhī, showing the punctuation of narratives with trees and portals. Peshawar Museum. (Photo: courtesy of the Warburg Institute, London.) The Global ConneCTions of Gandhāran arT Figure 35. Lid of sarcophagus in Figure 4, showing the role of trees and gates in the composition of lid reliefs. (Photo: agefotostock/ De Agostini/G. Nimatallah.) peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure affinity between Roman sarcophagi and Gandhāran sculptures, and I let the illustrated examples speak for themselves (Figures 33 and 34; 35 and 36). Like the sarcophagus lids, the Gandhāran reliefs are bracketed by framed figural motifs which act like bookends: acanthus leaves, for instance, and, in the case of the sarcophagi, frequently masks (Figures 33 and 34). The structure is similar, with punctuating motifs such as doors or gates and trees (Figures 35 and 36) breaking up the scenes or marking a transition between different episodes of a story. The general stylistic character of comparatively early friezes from both traditions is extraordinarily similar.33 Craft practices The similarities above argue for a closer potential relationship between Gandhāran and Roman sculpture than is implied by any notion of the ‘copying’ of ‘models’. A package of ideas has been transmitted which includes not only subjects, motifs, and stylistic devices, but also transferable methods for composing scenes as the artist planned and carved the relief down from the surface of the stone. It might be objected that there are also fundamental differences of technical approach to which we have briefly alluded. Gandhāran sculptors hardly ever used the drill, which was a fundamental tool of the Roman sculptor, its marks often much in evidence on the marble. Nor, normally, did they undercut the figures in high relief, a basic naturalistic technique facilitated by the drill, which grinds away the crystalline marble and avoids the percussive shock of the hammer and chisel.34 In her insightful critique of the influence of Roman sarcophagi, Stephanie Pearson has used these and other technical traits to suggest that the transmission of classical imagery to Gandhāran sculpture relied neither on sarcophagi nor the expertise of their artists (Pearson 2010).35 Yet, as she acknowledges, the use of the slate-like, metamorphic schist as the primary stone for Gandhāran sculpture really precludes the routine use of the drill. Moreover, as Pia Brancaccio has recently discussed, some sculptors of the Swat Valley did conspicuously but selectively employ the drill, as well as a range of classical iconographical motifs. She argues that this may have constituted a deliberate use of imported Roman methods as a distinctive selling point: ‘It is likely that given the extensive use of drills in Roman sculpture at the beginning of the Common Era, the mechanical innovation presented here could be read as a technical citation enhancing the authenticity of the Graeco-Roman repertoires represented by Gandhāran artists.’ (Brancaccio & Olivieri 2019: 138-141; quotation from 141.) If we adopt a wider perspective on Gandhāran production of narrative art, one less obvious methodological similarity to sarcophagus sculpture emerges. Roman sculpture was highly repetitious. Besides precise copying of models (in the reproduction of the Emperor’s portrait image, for example), Roman sculptors frequently reproduced looser compositions and motifs in which they were well versed. They probably improvised in their carving to a large extent, drawing upon a profoundly familiar repertoire which could be generated freehand and from memory. This accounts for the combination of homogeneity and variety in Roman sculpture. Roman sarcophagi are especially repetitious. While there is much diversity and novelty among them, there seems to have been surprisingly little premium on originality, in contrast, perhaps, to the improvisational variety of Roman wall-painting, where constant variations in the use of the decorative repertory seem to have been expected. In past generations this has contributed to the rather disdainful view of Roman sarcophagi as ‘mass-production’. The fact that most Roman figural sarcophagi can be sorted into iconographical types which reproduce more or less On the similarities and differences in narrative approach see Nehru 1989; Taddei 2015. In not carving the sides of figures realistically, Gandhāran artists rather resemble the co-called Coptic sculptors of Late Antique/Byzantine Egypt. 35 I am grateful to Stephanie Pearson for providing me with a copy of her thesis, much of which I agree with even though we reach opposing conclusions in certain respects. 33 34 77 The Global ConneCTions of Gandhāran arT predictable iconographies has been seen as evidence of ‘production to stock’, though the consensus on this is shifting (e.g. Russell 2011). (It seems that speed and economy resided in the artists’ knowing very well what they were doing once commissioned rather than prefabricating coffins to await customers.) It is striking that the corpus of narrative sculpture – which also includes many examples of innovation and idiosyncracy – can very largely be divided into stock scenes like the Parinirvāṇa or the Birth of the Buddha, which were reproduced with a common set of elements by artists working in differing styles and with differing levels of ability. In other words, the repetitious production of Gandhāran sculptural iconography echoes that of the Roman Empire. It suggests that, whether by coincidence or influence, there were profound similarities in the organization and methods of production, rather than merely visual analogies between the two traditions. Trade as a vehicle for influence? The evidence outlined above argues for a strong relationship between Roman imperial sculpture and Gandhāran art. To what extent sarcophagi are specifically formative in this relationship, or merely the conspicuous traces of a larger sculptural tradition in Rome remains, of course, open to debate. For the sake of argument, however, let us assume that the reader has found this range of circumstantial evidence for the influence of sarcophagi convincing. How should we account for an influence which is in some respects counter-intuitive and defies geography, particularly since the relationship I have sketched seems to involve almost exclusively the narrative sarcophagi of metropolitan Rome rather than those of the eastern Empire? Trade is normally invoked as the mechanism by which Gandhāran artists (and perhaps their customers) might have become familiar with the visual culture of Rome. It is certainly plausible to suppose that trade links were at the root of the global connectedness of the ancient world, of which we are increasingly conscious. Perhaps trade was a precondition for some of the migrations of imagery that we detect (compare the contribution by Van Aerde and her colleagues in this volume). There is very abundant evidence for trade between the Roman Empire and southern India in the early centuries AD. Indeed, the economic value of such trade for the Empire must have been massive. It has been estimated (conservatively) that the tax revenue on imports from India via Roman Egypt in the second century could have amounted to some 230 million sesterces – sufficient to fund a third of the cost of the imperial army each year (Wilson 2015: 23-24). In contrast, the picture of trade with Gandhāra is less clear, partly because of the very much smaller quantity of Roman artefacts found there, though the spectacular Roman and Indian imports cached at Begram in Afghanistan have frequently been used as an illustration of the connection – they have been made to carry rather a lot of weight in this respect. The content of trade between the Roman Empire and Central Asia or northern India did not leave the same kind of material precipitate as we find in Kerala and Tamil Nadu. Nevertheless, the trade connections are well documented, particularly by sea, as most commercial exchange seems to have happened through the Red Sea and Indian Ocean, following the route described by the first-century AD Periplus Maris Erythraei (for a conspectus of trade with India, Central Asia, and beyond see Graf 2017; cf. McLaughlin 2014 and 2016). The evidence of long-term trade-contacts is reinforced by specific sources alluding to diplomatic contact between the Roman and Kushan Empires which, after all, shared a common intervening enemy in the form of the Parthians. We are told about embassies to the emperor Hadrian (by ‘kings of the Bactrians’, perhaps around the time of Kaniṣka’s accession c. AD 128: Historia Augusta, Hadrian, 21.14) and Antoninus Pius (‘from the Indians, Bactrians, and Hyrcanians’ in the years around AD 150: [Aurelius 78 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Victor], Epitome de Caesaribus, 15.4). Some Kushan coin series imitated early imperial Roman coins, and perhaps some of the Roman imagery of power was adopted by the Kushans (Rosenfield 1967).36 All of this is to say that the Romans and the people of Kushan Gandhāra inhabited the same world, despite the vast distance between them, and this is the context for the emergence of Gandhāran sculpture. As Buchthal elegantly expressed it (long before the word ‘globalization’ was invented): ‘More than any other aspect of ancient art the Gandhara school of sculpture makes us conscious of the inherent unity of artistic achievement in a world which was essentially one – though it extended far beyond the reach of Greek and Roman arms, to the limit of ancient geographical knowledge.’ (Buchthal 1945: 23.) Yet such connectedness is not sufficient to explain Gandhāran art. The early Buddhist art of southern India proves the point, for despite the demonstrably huge scale of Roman contacts with that part of Asia, the Roman influence on the Buddhist Sculpture of Andhra, for example, is much harder to discern. Some strong arguments have been made to suggest that here too both Buddhist and secular sculpture were drawing upon Roman traditions (Zin 2014; Zin 2016), but the echoes are not so obvious as those of Gandhāra, and in some cases it requires the eye of faith to discern them! There are other reasons to doubt the role of imported objects as vectors for Roman artistic ideas, although that scenario is often proposed (e.g. Boardman 1994: 131 and Mairs 2014, both recognizing a role for travelling craftsmen; Pearson 2010: 16-17 arguing for pattern-books instead). While exposure to some elements of iconography might have come from portable objects, including the elusive drawings or ‘pattern-books’ which are often invoked (with negligible evidence) to explain the spread of Roman sculptural imagery, it is inconceivable that the examples of imitation described above could have come about in this way. What we can see in Gandhāra’s inheritance of Roman traditions is the adoption not only of motifs but of an entire body of know-how, which – by the time we see it in extant, datable works – has been thoroughly internalized.37 The consequence is that Gandhāran artists never, or almost never, made ‘errors’ when imitating classical imagery. The original imagery was assimilated and transformed to make new sense within an innovative form of Buddhist art. The ‘Wooden Horse’ relief above (Figure 8) represents an example of the deft, nuanced, and economical manner in which this transformation was effected. The spear of ‘Laocoon’ on the Gandhāran relief has taken the place of the rope in the Ashmolean’s sarcophagus lid, emphasizing resistance rather than capitulation: the Horse (so far) is being kept outside the walls of the city, not dragged in. The fact that it is to the right of the gate underlines that this is a different moment (emphasized by the Gandhāran tendency to ‘read’ narrative reliefs from right to left), and the figure of ‘Cassandra’ seems to bar the gate itself. We might well suppose, as Foucher speculated, that this is a counterfactual version of the classical myth, in which the bodhisattva-Laocoon is successfully exposing the deception to save the city (Foucher 1950). Be that as it may, the artist has shown himself to be completely comfortable with the Graeco-Roman idiom – in this case comprising naturalistic style, costumes, and composition. This imagery has not been copied from a gem or a pot. The same impression is given by the full range of Gandhāran sculpture, and not just those that are particularly polished and sophisticated. Perhaps the foreign conventions fell on fertile ground because the visual culture of Gandhāra had been informed for centuries by contact with the Greek world. In that For Kushan connections see further Graf 2017: esp. 492; Thorley 1979. Pearson’s argument (2010) is precisely that know-how was not being communicated because of the marked differences in technique and practice between Roman and Gandhāran sculpture. I would contend that departure from Mediterranean techniques in the ongoing tradition of schist stūpa-sculpture is not a problem for the theory of mobile artists, whereas the seamless assimilation of classical norms does challenge the idea of learning from objects. 36 37 79 The Global ConneCTions of Gandhāran arT respect at least the traditional view of Gandhāran art as a Hellenistic legacy may be accurate. In any case, the ‘Roman’ component in this art was completely absorbed. The role of the Roman artist I would propose, therefore, that we reconsider an explanation already entertained by Soper and others: that the movement of people, not objects, was crucial for the Roman imperial contribution to Gandhāran sculpture. Soper’s hypothetical scenario is anecdotal: merely intended as a model for explaining the extant evidence. He imagines a diplomatic mission to Rome securing ‘at the capital the services of a master sculptor, who was escorted back to Peshawar to lend a properly imperial note to the rapidly expanding art of Gandhāra’ (Soper 1951, 305). I am instinctively very resistant to the positivistic, simplifying tendency underlying such an explanation. It might be considered a ‘rabbit out of the hat’ explanation for a phenomenon so subtle and complex that it has defied convincing analysis for a century and a half. It is still the case, as Soper himself observed seventy years ago, that no known evidence exists to prove the hypothesis. It is not directly attested by any of the evidence reviewed so far in this chapter, nor by other evidence adduced in the past to suggest the presence of artists from the empire in Central Asia, notably the inscription of the painter ‘Tita’ (interpreted as ‘Titus’) at Miran (Stein 1921: vol. 1, 529-531). There are some works of Gandhāran sculpture which would be indistinguishable from Roman craftsmanship if one encountered them in marble, but not many (Figure 31 is a candidate). If Roman artists were present in Gandhāra we cannot see them, and perhaps we should not expect to (Soper 1951: 306). Insisting on individual, influential agents coming from one place to another and causing change – a change, moreover, that we are mainly seeing reflected in somewhat later production – seems at odds with the non-linear, non-bipolar, global network of contact and influence which is an increasingly attractive model for explaining the complexity of ancient world culture. Nevertheless, the movement of artists remains perhaps the most satisfactory explanation for the Gandhāran adoption of Roman artistic idioms. It is not hard to imagine situations in which the migration of artists took place, whether the personal enterprise of individuals or demand from a powerful patron, maybe bringing artists to Gandhāra to fulfil a large and ultimately very influential project. Such a migration is famously invoked by the apocryphal third-century Acts of the Apostle Thomas, in which Jesus sells the apostle Thomas as a skilled slave to an agent of the Indo-Parthian king of Gandhāra, Gondophares or ‘Gundaphorus’ (James 1924; Soper 1951: 305). The blatant fictionality of the account might positively discourage us from envisaging any such situation in real life, but in fact there is nothing inherently unlikely about the possibility of immigrant craftsmen from the Roman Empire.38 On a larger scale we have explicit evidence for a similar situation in AD 66, when the Emperor Nero hosted his client king Tiridates I of Armenia to Italy for his coronation. According to Dio, Nero despatched him to Armenia with gifts of artisans to rebuild his capital at Artaxata (Dio Cassius [epitome], 63.7). The king also hired more himself at high prices, but the Roman general Corbulo prevented them from leaving the Empire. In the case of Gandhāra, we probably should not hypothesize a single moment of contact; rather a sustained period of strong interactions between craftsmen of the two empires. But the example of Tiridates shows us the potential importance of top-down demand from single patrons. 38 Pia Brancaccio has suggested to me the theoretical possibility not only of Roman migrants as such, but also Roman slaves from Central Asia returning to their home region after obtaining freedom. The majority of Roman artists appear to have been slaves or former slaves. 80 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Whatever scenario we imagine, we should recognize an underlying economic reality behind the explosion of classicizing Gandhāran sculpture. This applies as much to the traditional notion of enduring, latent, Hellenistic influence as it does to the picture of fresh Roman contacts endorsed here. Gandhāran Buddhist sculpture came into existence, in its archaeologically conspicuous form, when and where it did because of a demand for an artistic solution to a particular need. In the first and second centuries AD there existed in Gandhāra an affluent and apparently substantial community of Buddhist donors who believed that their lives, and more particularly their future lives, could be made better by converting their wealth into durable, commemorative monuments. The monuments were made to last, since that was one source of merit alongside the more ephemeral day-to-day devotions that have left little or no trace, and they were richly adorned in the imagery that reminded devotees whom they were venerating and why, and provided them with a visual focus for their visits to the reliquary sites. This religiously motivated materialism is part of the reason why the Gandhāran region today bristles with the remains of Buddhist stone sculpture and architecture, even after generations of excavation and looting, whereas other aspects of Gandhāran society are relatively poorly evidenced by archaeology. Such monuments had long existed in South Asia, and the venerable stūpa sites at Bharhut and Sanchi, for example, were enhanced by narrative sculptures (albeit without yet the anthropomorphic image of the Buddha himself) which provided a precedent. But nothing in Buddhist art catered for the scale and intensity of demand for Buddhist sculpture that we encounter in second-century Gandhāra. In the context of this demand, we can imagine, the artists and patrons of Gandhāran art were oriented towards the place which was, at that moment in history, the biggest centre of sculptural production on the planet: the city of Rome.39 References Baratte F. and Metzger C. 1985. Musée du Louvre. catalogue des sarcophages en pierre d’époques romaine et paléochrétienne. Paris: Éditions de la RMN. Benndorf O. and Schöne R. 1867. Die antike Bildwerke des Lateranensischen Museums. Leipzig: Breitkopf and Härtel. Bielefeld D. 1997. Die stadtrömischen Eroten-Sarkophage, 2nd fascicule, Weinlese- und Ernteszenen (Die antiken Sarkophagreliefs 5.2.2). Berlin: Mann. Boardman J. 1994. The Diffusion of Classical Art in Antiquity. London: Thames and Hudson. Bracey R. 2020. The Gandharan Problem. Pages 27-50 in J. Elsner (ed), Empires of Faith: Histories of Art and Religion from India to Ireland. Cambridge: Cambridge University Press. Brancaccio P. and Olivieri L.M. 2019. Regional Workshops and small stūpas in the Swat Valley: An Analysis of the Evidence from Gumbat, Saidu Sharif, and Pānṛ. Pages 121- 142 in W. Rienjang and P. Stewart (eds), The Geography of Gandhāran Art. Oxford: Archaeopress. Brendel O. 1979. Prolegomena to the Study of Roman Art (expanded from Prolegomena to a Book on Roman Art. Memoirs of the American Academy in Rome 21 [1953]: 9-73). New Haven and London: Yale University Press. Bromberg C. 1988. The Putto and Garland in Asia. Bulletin of the Asia Institute. New Series 2: 67-85. Buchthal H. 1943. The Common Classical Sources of Buddhist and Christian Narrative Art. Journal of the Royal Asiatic Society (1943): 137-48. 39 As far as I am aware, no centre of sculptural production in this period came close to Rome in respect to the volume of output, though no systematic comparison has been carried out. The other contenders beyond the Roman Empire would have been in Han China, Mathura, and India. For an overview of the Roman imperial sculptural trade see Russell 2013: esp. 256-320 on sarcophagi. G. Koch (1993: 1) has estimated that there are between 12,000 and 15,000 extant Roman sarcophagi, representing an original total production of up to three-quarters of a million in the period AD 120-310. Up to half of these may have been made in Rome. This is, however, at the limit of what is plausible (Russell 2011: 127-8). Even the more modest marble sarcophagi may have required two or three man-years to carve. 81 The Global ConneCTions of Gandhāran arT Buchthal H. 1945. The Western Aspects of Gandhara Sculpture (Henriette Hertz Trust Annual Lecture on Aspects of Art). London: British Academy. Coomaraswamy A. 1913. The Arts and Crafts of India and Ceylon. London and Edinburgh: T.N. Foulis. Cribb J. 1984. The Origin of the Buddha Image - the Numismatic evidence. Pages 231-34 in B. Allchin (ed), South Asian Archaeology 1981. Cambridge: Cambridge University Press. Davies P.E. 1997. The Politics of Perpetuation: Trajan’s Column and the Art of Commemoration. American Journal of Archaeology 101: 41-65. DeCaroli R. 2015. Image Problems: The Origin and Development of the Buddha’s Image in Early South Asia. Seattle: University of Washington Press. De Marco G. 1987. The Stūpa as a Funerary Monument: New Iconographical Evidence. East and West 37/1: 191-246. Erdosy G. 1990. Taxila: Political History and Urban Structure. Pages 657-74 in M. Taddei and P. Callieri (eds), South Asian Archaeology 1987: Proceedings of the Ninth International Conference of the Association of the South Asian Archaeologists in Western Europe. Rome: Istituto Italiano per il Medio ed Estremo Oriente. Falk H. 2010. Libation Trays from Gandhara. Bulletin of the Asia Institute n.s. 24: 89-113. Falser M. 2015. The Graeco-Buddhist Style of Gandhara – a ‘Storia Ideologica’, or: How a Discourse Makes a Global History of Art. Journal of Art Historiography 13: 1-52. Filigenzi A. 2012. Orientalised Hellenism versus Hellenised Orient: Reversing the Perspective on Gandharan Art. Ancient Civilizations from Scythia to Siberia 18: 111-141. Foucher A. 1950. Le cheval de Troie au Gandhâra. Comptes rendus des séances de l’Académie des Inscriptions et Belles-Lettres 94: 407-12. Francfort H.-P. 1979. Les palettes du Gandhara (Mémoires de la Délégation archéologique française en Afghanistan). Paris: Diffusion de Boccard. Froning H. 1980. Die ikonographische Tradition der kaiserzeitlichen mythologischen Sarkophagreliefs. Jahrbuch des Deutschen Archäologischen Instituts 95: 322-341. Gatti C. 2005. Il sepolcro dorico e La Tomba dei Festoni sulla via Appia: due esempi di ricostruzione ottocentesca. Archeologia Classica 56: 155-87. Germoni P. 2009. Fiumicino-Isola Sacra: vecchi e nuovi rinvenimenti. Bulletino della Commissione Archeologica Communale di Roma 110: 398-404. Giuliano A. (ed). 1979. Museo Nazionale Romano. Le sculture, vol. 1.1. Rome: De Luca. Graf D. 2017. The Silk Road Between Syria and China. Pages 1-118 in A. Wilson and A. Bowman (eds), Trade, Commerce, and the State in the Roman World. Oxford: Oxford University Press. Herdejürgen H. 1996. Stadtrömische und italische Girlandensarkophage, 1st fascicule, Die Sarkophage des ersten und zweiten Jahrhunderts (Die Antiken Sarkophagreliefs 6.1.1). Berlin: Mann. Himmelmann N. 1980. Über Hirten-Genre in der antiken Kunst. Opladen: Westdeutscher Verlag. Ingholt H. 1957. Gandhāran Art in Pakistan. New York: Pantheon Books. James M.R. 1924. Apocryphal New Testament. Oxford: Clarendon Press. Jensen R.M. 2015. Allusions to Imperial Rituals in Fourth-Century Christian Art. Pages 13-48 in L.M. Jefferson and R.M. Jensen (eds), The Art of Empire: Christian Art in its Imperial Context. Minneapolis: Fortress Press. Jongeward D. 2019. Buddhist Art of Gandhara in the Ashmolean Museum. Oxford: Ashmolean Museum. Khan N.A. 1990. A New Relief from Gandhāra Depicting the Trojan Horse. East and West 40: 315-9. Koch G. 1993. Sarkophage der römischen Kaiserzeit. Darmstadt: Wissenschaftliche Buchgesellschaft. Koch G. and Sichtermann H. 1982. Römische Sarkophage. Munich: Beck. Kuwayama S. 1997. A Hidden Import from Imperial Rome Manifest in Stupas. Pages 119-71 in R. Allchin et al. (eds), Gandharan Art in Context. Cambridge: Ancient India and Iran Trust. Lawrence M. 1965. The Velletri Sarcophagus. American Journal of Archaeology 69: 207-22. LIMC. Lexicon Iconographicum Mythologiae Classicae. Zurich: Artemis (1981-1989). 82 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Lo Muzio C. 2011. Gandharan Toilet Trays: Some Reflections on Chronology. Ancient Civilizations from Scythia to Siberia 11: 331-340. Mairs R. 2014. Models, Moulds and Mass Production: The Mechanics of Stylistic Influence from the Mediterranean to Bactria. Ancient West and East 13: 175-195. Mairs R. 2016. The Hellenistic Far East: Archaeology, Language, and Identity in Greek Central Asia. Oakland, University of California Press. Marshall J. 1946. Review of Buchthal 1945. The Journal of the Royal Asiatic Society of Great Britain and Ireland (April, 1956): 116-22. Marshall J. 1951. Taxila: An Illustrated Account of Archaeological Excavations 1913-1934. 3 vols. Cambridge: Cambridge University Press. Mathews T. 1999. The Clash of Gods: A Reinterpretation of Early Christian Art Princeton: Princeton University Press. Matz F. 1968. Die dionysischen Sarkophage, part 2 (Die antiken Sarkophagreliefs 4.2). Berlin: Mann. McLaughlin R. 2014. The Roman Empire and the Indian Ocean: The Ancient World Economy and the Kingdoms of Africa, Arabia and India. Barnsley: Pen & Sword. McLaughlin R. 2016. The Roman Empire and the Silk Routes: The Ancient World Economy and the Empires of Parthia, Central Asia and Han China. Barnsley: Pen & Sword. Michaelis A. 1882. Ancient Marbles in Great Britain. Cambridge: Cambridge University Press. Naiki S. 2019. Geographical Differences and Similarities in Gandhāran Sculptures. Pages 41-57 in W. Rienjang and P. Stewart (eds), The Geography of Gandhāran Art. Oxford : Archaeopress. Neelis J. 2019. Making Places for Buddhism in Gandhāra: Stories of Previous Births in Image and Text. Pages 175-185 in Rienjang & Stewart 2019. Nehru L. 1989. Origins of the Gandhāran Style: A Study of Contributory Influences. Delhi: Oxford University Press. Olivieri L.M. 2019. The Early-Historic Funerary Monuments of Butkara IV: New Evidence on a Forgotten Excavation in Outer Gandhara. Rivista degli Studi Orientali 92: 231-257. Parlasca K. 1981. Review of Francfort 1979. Gnomon 53: 694-7. Parlasca K. 1983. Griechisch-römische Steinschälchen aus Ägypten. Pages 151-60 in Das römischbyzantinische Ägypten: Akten des internationalen Symposions 26.-30. September 1978 in Trier. Mainz am Rhein: Zabern. Pearson S. 2010. Reworking Western Models: The Importance of Sculptural Technique in Gandharan Relief. MA Thesis, University of California, Berkeley. Rienjang W. and Stewart P. (eds) 2018. The Chronology of Gandhāran Art: Proceedings of the First International Workshop of the Gandhāra Connections Project, University of Oxford, 23rd – 24th March, 2017. Oxford: Archaeopress. Rienjang W. and Stewart P. (eds) 2019. The Geography of Gandhāran Art: Proceedings of the Second International Workshop of the Gandhāra Connections Project, University of Oxford, 22nd – 23rd March, 2018. Oxford: Archaeopress. Robert C. 1890. Die antiken Sarkophag-reliefs, vol. 2, Mythologische Cyklen, Berlin, G. Grote’sche Verlagsbuchhandlung. Robert C. 1897. Die antiken Sarkophag-reliefs, vol. 3, part 1, Einzelmythen: Actaeon - Hercules. Berlin: G. Grote’sche Verlagsbuchhandlung. Robert C. 1904. Die antiken Sarkophag-reliefs, vol. 3, part 2, Einzelmythen: Hippolytos - Meleagros. Berlin: G. Grote’sche Verlagsbuchhandlung. Robert C. 1919. Die antiken Sarkophag-reliefs, vol. 3, part 3, Einzelmythen: Niobiden - Triptolemos ungedeutet. Berlin: G. Grote’sche Verlagsbuchhandlung. Rosenfield J. M. 1967. The Dynastic Art of the Kushans. Berkeley and Los Angeles: University of California Press. Rowland B. 1936. A Revised Chronology of Gandharan Sculpture. Art Bulletin 18: 387-400. 83 The Global ConneCTions of Gandhāran arT Rowland B. 1942. Gandhāra and Late Antique Art: The Buddha Image. American Journal of Archaeology 46: 223-36. Rowland B. 1953. The Art and Architecture of India. Harmondsworth: Penguin. Rowland B. 1956. Gandharan, Rome and Mathura: The Early Relief Style. Archives of the Chinese Art Society of America 10: 8-17. Russell B. 2011. The Roman Sarcophagus ‘Industry’. Pages 119-147 in J. Elsner and J. Huskinson (eds), Life, Death and Representation: Some New Work on Roman Sarcophagi. Berlin: De Guyter. Russell B. 2013. The Economics of the Roman Stone Trade. Oxford: Oxford University Press. Sadurska A. 1964. Les tables iliaques. Warsaw: Państwowe Wydawnictwo Naukowe. Schlumberger D. 1960. Descendants non-méditerranéens de l’art grec (part 1). Syria 37: 131-66. Schneider R.M. 1986. Bunte Barbaren: Orientalenstatuen aus farbigem Marmor in der römischen Repräsentationskunst. Worms: WernerscheVerlagsgesellschaft. Schröder S.F. 2004. Catálogo de la escultura clásica del Museo del Prado, vol. 2. Madrid: Museo Nacional del Prado. Smith R.R.R. 2013. The Marble Reliefs from the Julio-Claudian Sebasteion (Aphrodisias 6). Darmstadt/Mainz: Zabern. Smith R.R.R. 2019. Mythology and Sarcophagi at Aphrodisias and Rome. Pages 161-175 in C. Hallett (ed), Flesheaters: An International Symposium on Roman Sarcophagi, University of California at Berkeley, 18-19 September 2009. Wiesbaden: Reichert. Soper A.C. 1951. The Roman Style in Gandhara. American Journal of Archaeology 55: 301-19. Sparkes B. 1971. The Trojan Horse in Classical Art. Greece and Rome 18: 54-70. Squire M. 2011. The Iliad in a Nutshell: Visualizing Epic on the Tabulae Iliacae. Oxford and New York: Oxford University Press. Stančo L. 2013. Greek Gods in the East: Hellenistic Iconographic Schemes in Central Asia. Prague: Karolinum Press. Stein A. 1921. Serindia: Detailed Report of Explorations in Central Asia and Westernmost China. Oxford: Clarendon Press. Stewart P. 2016. The Provenance of the Gandhāran ‘Trojan Horse’ Relief in the British Museum. Arts Asiatiques 71: 3-12. Stoye M. 2004. Der Dreifuss in gandharischen Szenen zum Ersten Bad des Siddhartha Gautama: Überlegungen zu Herkunft und Bedeutung eines visuellen Zitats (The Tripod in Gandharan Scenes of the First Bath of Siddhartha Gautama: Reflections on the Provenance and Meaning of a Visual Quote) [with English Summary]. Artibus Asiae 64/2: 141-176. Strocka V.M. 2015. Der Manchinger Silberbecher. Bonner Jahrbücher 215: 323-352. Taddei M. 2015. Narrative Art between India and the Hellenistic World. Journal of Transcultural Studies 6: 34-74. Thomas E. 2011. Houses of the Dead? Columnar Sarcophagi as ‘Micro-Architecture’. Pages 387-435 in J. Elsner and J. Huskinson (eds), Life, Death and Representation: Some New Work on Roman Sarcophagi. Berlin: De Guyter. Thorley J. 1979. The Roman Empire and the Kushans. Greece and Rome 28: 181-190. Walker S. 1990. Catalogue of Roman Sarcophagi in the British Museum (Corpus Signorum Imperii Romani, Great Britain 2.2). London: British Museum. Wegner M. 1966. Die Musensarkoge (Die antiken Sarkophagreliefs 5.3). Berlin: Mann. Wheeler R.E.M. 1949. Romano-Buddhist Art: An Old Problem Restated. Antiquity 23: 4-19. Wilson A. 2015. Red Sea Trade and the State. Pages 13-32 in F. De Romanis and M. Maiuro (eds), Across the Ocean: Nine Essays on Indo-Mediterranean Trade. Leiden and Boston: Brill. Winfeld-Hansen H. 1965. Les Couloirs annulaires dans l’architecture funéraire antique’. Acta ad Archaeologiam et Artium Historiam Pertinentia 2: 35-63. Zanker P. and Ewald B. 2012. Living with Myths: The Imagery of Roman Sarcophagi. Oxford: Oxford University Press. 84 peTer sTeWarT: roman sarCophaGi and Gandhāran sCulpTure Zin M. 2014. Nur Gandhara? Zu Motiven der klassichen antike in Andhra (einschliesslich Kanaganahalli). Tribus: Jahrbuch des Linden-Museums 64: 179-204. Zin M. 2016. Buddhist Narratives and Amaravati. Pages 46-58 in A. Shimada and M. Willis (eds), Amaravati: The Art of an Early Buddhist Monument in Context. London: British Museum Press. Zwalf W. 1996. A Catalogue of the Gandhara Sculpture in the British Museum. 2 Vols. London: The British Museum Press. 85 (86-101): DOI: 10.32028/9781789696950-5 The transmission of Dionysiac imagery to Gandhāran Buddhist art Tadashi Tanabe Introductory remarks Within the corpus of Gandhāran sculpture, there are several relief panels depicting so-called Dionysiac or Bacchanalian scenes.1 These show scenes of banqueting, men and women fraternizing, amorous couples, music performances, and wine-making. Such images were undoubtedly influenced both iconographically and stylistically by Greek and Roman art. In this paper, I shall address this Gandhāran Dionysiac imagery and attempt to clarify how such a nonBuddhist imagery was transmitted from the Roman Empire to Gandhāra. First, I shall explain typical Dionysiac images in Gandhāran relief panels; second, I shall survey their depiction in Greek and Roman art; and third, I shall make a comparison of the two corpora, with the purpose of clarifying some peculiar aspects of Gandhāran Dionysiac imagery. Finally, I shall attempt to answer two questions raised by Peter Stewart in his paper in the present volume: how artistic ideas were transmitted, and whether this occurred through the movement of objects or of people. Dionysiac or Bacchanalian images from Gandhāra In this section I will enumerate and briefly survey five relief panels depicting Dionysos and his followers, the so-called thiasos (Figures 1-5). For simplicity I use the name ‘Dionysos’ provisionally in describing the Gandhāran examples for, as we shall see, the Graeco-Roman tradition was transformed selectively in its new Buddhist context. Figure 1. To the viewer’s right on this relief panel a banquet scene is represented, probably celebrating the marriage of Dionysos and Ariadne. Dionysos has a cantharus cup in his right hand and is seated with his bride Ariadne on his knee. Her right hand is placed on the god’s shoulder. He is surrounded by two women, probably maenads, each holding a wine-cup or plate. A man is standing behind him. To the viewer’s left, a man shoulders a leather wine skin while another man scoops up the wine from that bag in both hands. Both wear an I express my deepest gratitude to Dr Peter Stewart for having kindly invited me to the workshop, ‘The Global Connections of Gandhāran Art’, Oxford, 18th-19th March 2019. 1 Figure 1. Gandhāran banquet scene. H. 29.2 cm, c. second to third century AD. Tokyo National Museum, inv. TC-740. (Photo: after Tokyo National Museum et al. 2003: fig. 129.) 86 Tadashi Tanabe: The Transmission of dionysiaC imaGery To Gandhāran buddhisT arT Figure 2. Gandhāran banquet scene. H. 19.2 cm, c. second to third century AD. Tokyo National Museum, inv. TC-705. (Photo: Integrated Collections Database of the National Museums, Japan <https://colbase.nich.go.jp>.) exomis, the Greek costume for lower-class males. Behind his back stands a woman, probably a maenad. All the men depicted on this relief panel are bearded and all wear Greek- or Roman-style costume. Unfortunately, all the heads of Dionysos as well as the standing men and women (or maenads) are not original, having broken away and been subsequently restored. During the process of restoration, Dionysos was mistakenly bearded, since his face must have been youthful and without a beard when he married Ariadne.2 Figure 2. On this relief panel, all the figures are depicted dancing or playing music at a banquet. To the viewer’s left, a drunken Dionysos or Silenus is supported on both sides by a bearded man (likely not a satyr) and a woman (supposedly a maenad). ‘Dionysos’ is bearded and has a tendril of grape-vine tied around his head. To the viewer’s right, a female, probably a maenad, is playing a woodwind instrument and a male figure wearing Indian costume is dancing and beating a drum. A grape-vine is featured on both sides of this relief panel, symbolizing Dionysos. Figure 3. Set between a pair of leonine feet is an example of a Dionysiac scene that is both typical and remarkable. It is composed of two pairs of couples facing one another (Ingholt 1957: 157, fig. 398). The shape of this panel, designated by W. Zwalf as a ‘stair panel’ modified by a narrow inward curve, is unique and reminds us of the siṃhāsana (lion-throne) of the seated Buddha and bodhisattva (Zwalf 1996: vol. 1, 299). To the viewer’s left, a maenad wearing a shawl and a crossed marriage-belt (cestus) is sitting with a bald and bearded man, probably Silenus. He wears a wreath of vine leaves around his head and is offering to his female partner a drink of wine from a shallow bowl. To his left, a beardless and barechested young man is also sitting with the other maenad who is touching his shoulder with her right hand. This young man is supposedly Dionysos but I am not of the firm conviction that this identification is accurate owing to the fact that the face of the man is broken. Figure 4. This relief panel depicts two pairs of male and female figures holding a banquet under a grapevine symbolizing Dionysos. A bearded male figure wearing a topknot together with a female figure Cf. the marriage of Dionysos and Ariadne on the bronze gilt krater from Derveni, c. 330 BC, Archaeological Museum, Thessaloniki. 2 87 The Global ConneCTions of Gandhāran arT Figure 3. Gandhāran Dionysiac scene. H. 24 cm, c. second to third century AD. Lahore Museum, inv. 1914. (Photo: after Tokyo National Museum et al. 2002: fig. 17.) supports a cantharus cup. The male figure is attempting to make her drink the wine. On the left of them, another bearded man wearing a turban is holding a tendril of vine in his left hand and in his right a cup probably filled with wine. Next to him stands a female figure. She has a jug in her right hand and appears to be talking to him. Figure 5. In the central zone of this vertical relief panel, five medallions formed by two symmetrical intertwining vines are depicted. Tendrils, leaves and grapes grow out from the branches of the vine-scroll, forming these medallions’ enclosing images. In the top medallion we can see a seated man drinking wine from a rhyton held in his right hand. Pictured in the next medallion below is an amorous couple, whose iconographic importance resides in that the male figure is touching the female’s breast. In the third medallion from the top, a man can be observed carrying a basket of grapes on his back. Figure 4. Gandhāran banquet scene. H. 40.5 cm, c. second to third century AD. Tokyo National Museum, inv. Tc-626. (Photo: author.) This male figure holds a bunch of grapes, checking whether it is appropriate to harvest the grapes or not. The medallion below this encloses a male figure with a child on his shoulders trampling grapes in order to extract juice from them. In the lowest medallion an archer is shown. This archer aims an arrow at some wild animal that is not depicted. 88 Tadashi Tanabe: The Transmission of dionysiaC imaGery To Gandhāran buddhisT arT Figure 5. Gandhāran Dionysiac relief with peopled vine-scroll. H. 124 cm, c. second to third century AD. Boston, Museum of Fine Arts, Charles Amos Cummings Fund 39.36. (Photo: Courtesy of the Museum of Fine Arts, Boston). According to these five examples of Gandhāran Dionysiac imagery, the dramatis personae are Dionysos, Silenus, Pan, a maenad, a lion or panther, and vine that is symbolic of Dionysos.3 These characters are engaged in making and drinking wine, playing music, dancing, and love-making or fraternizing. Most figures are more or less borrowed from or modelled after Dionysos and his thiasos; but some male figures do not belong to the thiasos. Most participants wear Greek or Roman costume. These motifs seem to lack sacred as well as divine significance and rather stress sensual pleasures. What is more, it must be emphasized that the god Dionysos himself appears but rarely in Gandhāran Dionysiac scenes. And even where Dionysos is depicted, it is not as a cultic object but simply as one character in the drama. Generally speaking, Gandhāran Dionysiac scenes are indifferent to the profound religious significance or distinguishing traits of the Dionysiac cult and his mysterious rituals. Dionysiac or Bacchanalian imagery in Greek and Roman art Dionysiac imagery in Greek and Roman art is precisely classified in the Lexicon Iconographicum Mythologiae Classicae (LIMC 1986: vol. 1, 420-423, 541-542).4 This repository has collated many figurative motifs pertaining to Dionysos/Bacchus. For the purposes of this paper I shall simply draw attention to some of the most frequent relevant themes in imagery that portrays Dionysiac myth, cult, and ritual. Specific examples will be compared briefly in the next section. In addition to these examples, there are two reliefs in a Japanese private collection depicting Dionysos and thiasos. Their provenances are unknown. One of them represents a fraternizing couple under an Indian arch, the other features wine making, fraternizing, drinking wine scenes. Cf. Tokyo National Museum et al. 2003: fig. 131; Tanabe 2006: fig. 78. 4 Many of the images and basic information from LIMC are also available on the Digital LIMC Database <weblimc.org>. 3 89 The Global ConneCTions of Gandhāran arT Figure 6. Table support with Dionysos and Satyr, c. 170-180 AD. Athens, National Archaeological Museum (Photo: author, by permission). 1. The independent statue or image of Dionysos and his thiasos The naked and youthful Dionysos is frequently represented as an independent figure, for example in Roman imperial sculpture. A typical scheme shows the god standing, holding the staff known as the thyrsos in his left hand, with a wine jug held in the right. A wreath of ivy and grapes adorns his long hair. A characteristic example of this type is preserved in the Palazzo Altemps in Rome – a second-century work found on the Gianicolo Hill (Museo Nazionale Romano, inv. 60920; Paribeni 1932: no. 271). The god is frequently accompanied by a panther in images of this kind, and on occasion he is shown riding on the animal’s back. A famous example is provided by the late fourth-century BC pebble mosaic from Pella (now in Pella Archaeological Museum; Siganidou & LilimbakiAkamati 2008: 64, fig. 45). Other creatures and Dionysiac companions may surround the otherwise isolated figure of the deity. For example, a sculpted table support of around the 170s AD in Athens shows a nude Dionysos holding a rhyton ending in a panther protome. Next to him, the goat-footed god Pan holds a stick or weapon known as a lagobolon. A young satyr climbs up a vine and plucks grapes with his right hand (Figure 6). 2. Myths of Dionysos Specific mythological narratives centred on Dionysos are represented in many classical works of art, notably in the rich repertoire of figure-decorated pottery made in Athens and South Italy between the sixth and fourth centuries BC. The Dionysiac retinue is represented taking part in specific mythological scenes. For example, in the scene of the birth of Dionysos on a late fifth-century BC Apulian red-figure krater in Taranto, the infant Dionysos emerges from the right thigh of Zeus and is received by an attendant nymph. Olympian deities, the thiasos, Pan, and Silenus surround them (Taranto, Museo Nationale Archeologico, inv. IG. 8264; Dell’Aglio & Zingariello. 2015: 47; Stoye 2008: figs. 10-11). A more unusual scene is represented on a famous Attic black-figure cup in Munich, made by Exekias around 530 BC. A bearded Dionysos wearing an ivy wreath sails his ship or boat. A vine is growing up the mast. Encircling his ship are seven swimming dolphins, probably denoting his encounter with Tyrrhenian (Etruscan) pirates, whom Dionysos transformed into dolphins. The theme of the scene is his bringing the gift of the vine and wine-making from Naxos to Athens, also included in the Hesiodic 90 Tadashi Tanabe: The Transmission of dionysiaC imaGery To Gandhāran buddhisT arT Hymn to Dionysos, of uncertain date, but possibly originating as early as the Archaic Greek period, the sixth century BC (Zanker & Ewald 2012: 143, pl. 131; Boardman 2014: 8; Knauß 2017: 119). 3. Harvesting grapes, wine-making, and Dionysos in a vineyard Let us now look at three sample representations of the god in connection to the vintage and wine-making. Figure 7. Dionysos is depicted in a vineyard sitting on a cross-legged chair (sella curulis) while drinking wine from his distinctive high-handled cup, the cantharus, and tended to by satyrs harvesting grapes. The drinking Dionysos and wine-making satyrs are common themes in Greek vase-paintings but here the painter combines various elements, creating a lively pattern of baskets, twisting vines, and ripe and juicy grapes. Figure 8. To the viewer’s right a satyr is carrying a basket filled with grapes to be brought to the wine press. Another satyr stands in this basket set on two wine tubs. Behind them stands a bearded Dionysos, watching their wine making. The god has a cantharus in his left hand. Behind Dionysos, a satyr shoulders a basket filled with grapes. Figure 9. This beautifully curved sarcophagus is known as the Farnese Sarcophagus.5 Satyrs and maenads grace its sides, the latter harvesting grapes while the former interrupt their work by flirtatiously pulling at their garments and exchanging amorous glances with them. Figure 7. Dionysos and satyrs harvesting grapes. Attic black-figure amphora, c. 540-530 BC. Boston, Museum of Fine Arts, Boston, inv. 63.952 (Henry Lillie Pierce Residuary Fund and Francis Bartlett Donation of 1900). (Photo: Museum of Fine Arts, Boston). Figure 8. Dionysos and Satyrs making wine. Attic red-figure krater, 450-440 BC. Athens, National Archaeological Museum (Photo: author, by permission). P.R. Crowley has approached this sarcophagus from many angles. According to him the sarcophagus is evidently less interested in the mythological protagonists as it is in the supporting cast of the thiasos and Dionysos, who are conspicuously absent (Crowley 2018: 42). 5 91 The Global ConneCTions of Gandhāran arT Figure 9. Harvesting grapes. The Farnese Sarcophagus, AD 225. Boston, Isabella Stewart Gardner Museum. (Photo: courtesy of the Isabella Stewart Gardner Museum.) Sarcophagi housed in the Capitoline Museum and the Archaeological Museum of Thessaloniki also adopt a similar Dionysiac vision of the grape harvest as its theme (Nielsen 2018: 34-35, 61). Bacchanalian revelry was a popular theme in ancient Roman sarcophagi reliefs. 4. Procession and Triumph of Dionysos and Satyr The procession of the god and his retinue was a frequent and enduring subject of Dionysiac imagery, whose tradition has been explored by Boardman (Boardman 2014). Two representative images from the Roman Imperial period are reproduced here as examples. Figure 10. This extremely well-preserved Roman marble sarcophagus depicts Dionysos seated on a panther in the centre, but he is somewhat overshadowed by four large standing figures who represent the Four Seasons from left to right: Winter, Spring, Summer, and Autumn. These figures are shown as sturdy youths. Around these five central figures other Bacchic figures and cultic objects are featured, all carved on a smaller scale. On the rounded ends of the sarcophagus are two other groups of large Figure 10. Dionysos seated on a Panther with his thiasos, c. AD 260-270. New York, Metropolitan Museum of Art, inv. 55.11.5. (Photo: Museum, CC0 licence.) 92 Tadashi Tanabe: The Transmission of dionysiaC imaGery To Gandhāran buddhisT arT Figure 11. The Triumph of Dionysos on the Pashley Sarcophagus, from Crete, c. 130s-140s AD. Cambridge, Fitzwilliam Museum. (Photo: Fitzwilliam Museum.) figures, similarly intermingled with lesser ones. On the left end, Mother Earth is portrayed reclining on the ground, accompanied by a satyr and a youth bearing fruit. On the right end, a bearded male figure, probably to be identified with the personification of a river-god, reclines in front of two winged youths, perhaps representing two additional seasons. One of them is probably drinking wine. Figure 11. The main frieze of this sarcophagus shows Dionysos returning from the East in triumph with his traditional entourage of satyrs and maenads. The god’s chariot is drawn by centaurs. Pan dances ahead and an elephant leads their way. Silenus lurches drunkenly across the centre of the scene. On the small frieze above, satyrs and maenads recline on couches, drinking and demanding wine. These Dionysiac scenes are suitable subjects for the decoration of a sarcophagus because elements of the myth and worship of Dionysos are focused on rebirth. Figure 12. Stone pilaster. Basilica, Leptis Magna. (Photo: Squarciapino 1974: tav. LXV-III 1-4.) 93 The Global ConneCTions of Gandhāran arT 5. The peopled vine-scroll Finally let us look at an example of the so-called peopled vine-scroll, extensively represented here by the Basilica in Leptis Magna (Figure 12).6 On the bottom of this pilaster stands a krater ornamented with vine leaves and ivy. From its mouth two vines grow up and intertwine symmetrically to form seven medallions. In the first, lowest medallion on the krater, Dionysos rides a panther. In the next medallion a satyr holds a vine-leaf in his left hand. In the third medallion a dancing maenad is rendered. The fourth medallion shows Silenus wearing a tunic covered with wool or animal skin, bearing a stick in his left hand and raising his right. In the fifth medallion a maenad is depicted, but her head is broken away. In the sixth medallion Pan is represented with hind legs and wearing a cloth over the upper body. The last or top medallion is decorated with an Eros picking up grapes. Among the corpus of Dionysiac imagery in Greek and Roman art, there are a great number and variety of motifs. Most of them are cultic objects clearly related to the adoration of Dionysos. Some of them are something like emblems indicative of Dionysiac mysteries; they tell us the birth and life stories of this god, presenting Dionysos as the deity of vines and wine. As a rule, Greek and Roman Dionysiac imagery is sacred and divine, showing a profound religious significance. Although small in number, there are, needless to say, secular and sensual images among the corpus of Greek and Roman Dionysiac imagery. But in general, the Greek and Roman Dionysiac imagery seems to be religious in motivation. Comparison between Gandhāran art and Greek and Roman art In this section I will compare the Dionysiac imagery depicted on Gandhāran sculpture with that of Greek and Roman art. First, as we have seen, amongst the Greek and Roman repertory there are independent and free-standing sculptures of Dionysos (Figure 6) or of the god accompanied by a thiasos. However, the independent and freestanding image of Dionysos is not known among Gandhāran Buddhist sculptures to the best of my knowledge. Moreover, there is no Gandhāran Buddhist relief panel that either faithfully or apparently depicts the myth of Dionysos. Moreover, emblematic images illustrating Dionysiac Mysteries, such as those found on some Roman mosaics, are not known to Gandhāran Buddhist relief panels (Figure 13).7 Taking this fact into account, we may tentatively conclude that it is highly unlikely that Gandhāran Buddhists under the rule of the Kushans embraced the cult of Dionysos. Faith in these gods was likely rare, and their cults likely non-existence, because no temple of Dionysos has ever been identified in Gandhāra or its environs. More concretely, it can be said that Gandhāran Buddhist sculpture has nothing to do with the cult of Dionysos or Bacchus. Second, as regards the motifs of harvesting grapes and wine-making, there are a great deal of examples in Greek and Roman Art. Greek vases are occasionally decorated with Dionysos seated and satyrs harvesting grapes or making wine (Figures 8 and 9). As we have seen, maenads, satyrs, and erotes are quite regularly represented on Roman sarcophagi, as is Dionysos himself (Figures 9 to 11). In contrast, Gandhāran relief panels are not decorated with maenads or satyrs in the context of grape-harvesting and wine-making, but in the company of some unidentifiable local, male figures.8 Even in wine-drinking scenes, reliefs show men wearing in Indian attire (Figures 2 and 4). One good example is to be found in the collection of the Musée Guimet in Paris (Arts Asiatiques 55 [2000], 139, fig. 4), on which a male figure is found stamping on grapes in a tub (Figure 14). He carries a child on his shoulder, which adds to his On the term, ‘peopled vine-scroll’, see Toynbee & Ward-Perkins 1950: 1-2; Rowland 1956: 353, n. 1. Horn (1972) has made a comprehensive survey of the mosaic in the collection of the Römisch-Germanisches Museum in Cologne. He investigated various depictions on the mosaic and clarified the symbolic significance of Dionysos mysteries. 8 However, a man who looks like Silenus is depicted to the left of a grape-stamping scene in Lahore: Falk 2009: fig. 1. 6 7 94 Tadashi Tanabe: The Transmission of dionysiaC imaGery To Gandhāran buddhisT arT Figure 13. Mosaic from a Roman house at Cologne, third century AD. Cologne, Römisch-Germanisches Museum. (Photo: author.) Figure 14. Gandhāran wine-making scene. Paris, Musée Guimet Museum, inv. MA 6354. (Photo: courtesy of Katsumi Tanabe, by permission). 95 The Global ConneCTions of Gandhāran arT weight to quicken the trampling process. To the viewer’s right, a scene of filtration of grape-juice is illustrated. But this active male figure is not a satyr. The process of making wine in Gandhāran relief panels is borrowed from a Roman iconographic type, including trampling a tub, but curiously Dionysos is never represented. Third, the procession scene is quite often depicted in the Greek and Roman art (Figures 10 and 11). However, as far as Gandhāran relief panels are concerned, no such procession or Triumph scenes of Dionysos have been attested thus far. It was in the era of the Roman Empire that the so-called peopled vine-scroll was created. As we have seen, this consists of medallions formed by two symmetrical intertwining vines, with one or two figures depicted within each. The conspicuous difference between the Roman peopled vine-scroll and that of Gandhāran Dionysiac relief panels lies in their figural imagery. In the former, both Dionysos and thiasos are depicted (Figure 12), whereas in the latter only thiasos-like, but probably local, figures are identifiable. Taking this into account, J. Boardman notes that the subject matter of Gandhāran Dionysiac relief panels is generally misunderstood, at least in terms of identifying the identities of Greek divinities and associated behaviours, but the origins are nonetheless quite unmistakable (Boardman 2014: 45-47). Furthermore, Gandharān Dionysiac relief panels do not emphasise any mythological aspects or significance relatable to Dionysos. Rather they prefer to highlight drinking and drunken scenes of fraternization, illustrative of the happy, joyful life of the Dionysiac paradise. This means that Gandhāran relief panels retain only one function of Dionysos, namely, as the god of wine. Therefore, it can be said that only certain motifs of Greek and Roman Dionysiac imagery were intentionally selected and hence accepted by Gandhāran Buddhists, having in all likelihood undergone a process of reinterpretation in the context of Gandhāran Buddhism. The purpose of this kind of active selection by Gandhāran Buddhists seems to have been to concretely visualize sensual pleasures to be obtained in the Buddhist after-life, as I already proposed in my previous paper read at the Conference of the South Asian Archaeology and Art 2018. Thus, the repertory of Dionysiac imagery transmitted to Gandhāra was apparently limited and those motifs related to the cult, myth, and ritual of Dionysos were neither transmitted nor adopted by the Buddhists of the region. How were the artistic ideas of Dionysiac imagery transmitted to Gandhāra? Through movement of objects or people? Everybody admits that Dionysiac images depicted on Gandharān relief panels must have been derived from either Greek (Hellenistic) or Roman Imperial art. However, there remains the controversial matter among art historians of whether these Greek and Roman characteristics of Gandhāran sculpture reflect a Greek tradition resulting ultimately from Alexander’s conquests – if they are the product of subsequent cultural contacts with later traditions of the Hellenistic East – or if they are due to the immigration of contemporary artists from the Roman Empire (see the contributions of Stoye and Stewart in the present volume). This problem has been elaborated in previous studies. Alfred Foucher maintained that the Indo-Greeks emigrating from the Graeco-Bactria to Gandhāra played an important role in transmitting Greek artistic ideas and techniques to Gandhāra (Foucher 1917: 111-137).9 Foucher also suggested that the person who harmonized two traditions, between Greek and Indian, in Gandhāran art was an artist by his Greek father and a Buddhist by his Indian mother (Foucher 1922: 467). According to John Marshall A pair of gold clasps representing Dionysos and Ariadne on a monster and Pan lifting a rhyton were found at Tillya Tepe in Bactria. Behind Ariadne is a hovering Nike holding a wreath above the head of the couple (Hiebert & Cambon 2011: 286-287, pl. 216). Dionysiac motifs were found in Bactria but the depiction of these motifs is not the same as those from Gandhāra. Therefore, Greek craftsmen might have come to Gandhāra from Bactria but sculptors in Gandhāra intentionally selected Dionysiac imagery and they created their own Dionysiac imagery. 9 96 Tadashi Tanabe: The Transmission of dionysiaC imaGery To Gandhāran buddhisT arT the Parthians not only had vast numbers of Asiatic Greeks within the borders of their own Empire, but were able to enjoy at least in times of peace both commercial and cultural contacts with the GraecoRoman world. Therefore, he thought that the Parthians carried out the renaissance of the Hellenistic tradition in Gandhāran art (Marshall 1960: 26-32). More recently, John Boardman, Elizabeth Errington, Joe Cribb, and Ladislav Stančo have also emphasized the Greek influence on Gandharan relief panels.10 On the other hand, Hugo Buchthal, Benjamin Rowland, Mortimer Wheeler and others attributed the western elements in Gandhāran art to the influence of Rome.11 There is indeed the possibility that the transmission of such imagery is to be attributed to the movement of people, that is to say, artisans and craftsmen. It is demonstrable that there were Greeks living in Gandhāra and environs. According to Richard Salmon, many inscriptions of the Kushan period are found on objects donated to Buddhist monasteries, particularly on reliquaries. In many donative inscriptions we find donors with distinctly non-Indic names, usually of Greek or Iranian origin (Salomon 2018: 4042). Apart from the names of the Indo-Greek kings, we have a few Greek names in Kharoṣṭhī inscriptions inscribed on a relic casket, a seal, and a silver vessel, which were found in Gandhāra. For example, Theüdora in Gāndhārī = Theodoros in Greek (cf. Thaudama in Gāndhārī = Theodamas, Theiodamos, Theodemos in Greek) (Konow 1929: 4, 6; Falk 2001: 308; 2002: 53, in Barrate 2002), as well as Demetria (i.e .Demetrios) and the likely mixed Graeco-Indic name, Helaüta (i.e. the Greek Helios with Sanskrit -gupta). We can therefore presume that some Buddhists with Greek and Iranian names donated relief panels with Dionysiac imagery to Buddhist temples. Contrastingly, we have no concrete evidence that attests to a Greek sculptor or craftsman involved in artistic activity in the region. But this is in harmony with the absence of Greek Dionysiac mythological scenes among Gandhāran Dionysiac relief panels. Such typical Dionysiac themes as the Birth of Dionysos, Dionysos sailing, Dionysos turning pirates into dolphins, Dionysos’ discovery of the sleeping Ariadne, and the Triumph of Dionysos, have not been found in Gandhāran Buddhist art as of yet. Similarly Latin names of craftsmen, that is transcriptions of a name originally written in Latin, virtually never appear in extant Gāndhārī inscriptions. Only one Kharoṣṭhī inscription, written on a fresco mural found from Mīrān in Central Asia, appears to mention such a figure; it reads: 1 Titas̱a eṣā ghali12 2. hastakrica [bhaṃma]ka13 3. 3 1000. This inscription was accurately translated by M. L’Abbé Boyer: ‘This is the fresco of Tita who received 3,000 bhaṃmaka (coins) (Boyer 1911: 417)’. Marc Aurel Stein alternatively translated it: ‘This fresco is [the work] of Tita, who has received 3,000 Bhaṃmakas [for it]’ (Stein 1921: 529-531). The most interesting word of this inscription is the western name Tita. This name is recognized as a transliteration of the Latin Titus.14 Certainly it is no surprise that a painter who bore a Latin name was employed in Mīrān where Roman-style figures such as a putti or youthful genii shouldering a garland are depicted on walls. B. Rowland assumed that the Mīrān murals were produced 10 Errington and Cribb say that images such as Atlas, the triton, Eros shouldering a garland, Dionysos and his thiasos, shown in Gandhāran art were influenced by Greek art because the Greek homeland had been embraced by the Roman empire and the arts of the eastern Mediterranean area remained essentially Greek, even during Roman rule, and thus set the standards for the development of Roman art (Errington & Cribb 1992: 37). Stančo places emphasis on Greek influence from Bactria and agrees with the assumption that Dionysiac images must be the work of Greek artists who fled from Bactria to Gandhāra before Bactria was invaded by nomads or their successors (Stančo 2012: 86-87). 11 Buchthal, Wheeler, and Rowland compared Gandhāran art with Roman art in the light of several iconographic motifs. Buchthal said that the Roman achievement was accepted in Gandhāra in its entirety (Buchthal 1945; Wheeler 1949: 1954: 183202; Rowland 1958). 12 Boyer translated ghali is khaḍī (Boyer 1911: 415). Molesworth’s dictionary explains that this word signifies a species of steatites used to rub over the writing-board or to whitewash walls; it is also an unctuous and whitish stone, a sort of pipeclay (Molesworth 1857: 193; Turner 1966: 198). On this basis Boyer argued that the word means ‘fresco’. Turner incidentally gives Sanskrit khaṭikā and Prakrit khaḍī in the sense of ‘chalk’. 13 According to Boyer this word is Skt. bharman. This word means gold or money. It could denote coinage. 14 According to Stein, this name is a noun which could not be etymologically or phonetically explained as being indigenous to any Indian and Iranian language during the period covered by the ruined Mīrān temple (Stein 1921: 530). 97 The Global ConneCTions of Gandhāran arT in conjunction with an atelier of artists trained in the Mediterranean tradition, in this case under the direction of Tita (Rowland 1974: 33-36). Stefan Baums mentions that this is the only potential classical name among the Gāndhārī sources from Central Asia (Baums 2018: 43). It is not clear whether Tita is an immigrant from the Roman Empire or an indigenous man of Mīrān bearing a Latin name. Mario Bussagli suggested that Tita and the sculptor of reliefs depicting the Buddha accompanied by Vajrapāṇi from Mardan, Pakistan were one and the same person (Bussagli 1963: 21-23; Filigenzi 2006: 72). In any case, a painted frieze representing nude putti or youthful genii shouldering a garland was found in Mīrān with this Kharoṣṭhī inscription found on the lower part of the right fore-leg of a white elephant. The motif of Erotes shouldering a garland is a quite popular and a favourite theme in both Roman and Gandharān art. Therefore, the painter who made this mural at Mīrān must have known Roman iconographic motifs and in view of the Gāndhārī inscriptions, we can assume that Tita emigrated most probably from the Roman Empire or Roman East (West Asia) to Mīrān by way of the Gandhāra region. As regards the route from the Roman Empire to Gandhāra and environs, according to the Periplus of the Erythraean Sea, ships from the Roman Empire or Alexandria in Egypt lay at anchor at the ancient market port of Barbarikon, situated at the mouth of the Indus River, and freight was transported from Barbarikon to a metropolis in the headwaters of the Indus river, i.e. in Gandhāra (Schoff 1974: 37). Marshall maintained that many objects, such as silver wares, gold jewellery, engraved gems and others, were imported from the Mediterranean to Taxila (Marshall 1951a: 616-633, 675-676; Marshall 1951b: pl. 188-1, 2, 5, 191-96, 97, 98, 207-11). However, this assertion must be carefully scrutinized. David Whitehouse has stated that Taxila was an active participant in the exchange network that brought the products of Central and Eastern Asia to the Indian Ocean, whence they were shipped to Egypt, and that the Kushans’ Barbarikon was the gateway to the Roman world (Whitehouse 1989: 95). Matthew Cobb states that Palmyrene merchants were trading via the Red Sea in the third century and that Roman trade in the Indian Ocean appears to reach a peak broadly in the latter half of the first century, after which time there appears to be a decline in volume (Cobb 2015: 373-374). As the treasures of Begram show, artefacts of the Roman Empire were carried to Greater Gandhāra through commerce along the Indus river and her tributaries. Several medallions were found at Begram, on one of which appears a drunken Dionysos and his thiasos.15 It is likely that ‘itinerant’ Roman craftsmen and artisans could have followed the same route to Gandhāra (Rowland 1960: 8). The participation of Roman artisans is attested by an extremely realistic rendering of Herakles-Vajrapāṇi and of Tyche-Ardoxsho, both made of clay, which were excavated at Tape-Shotor in Haḍḍa (Boardman 2015: 188-189, figs. 122, 123). In addition, as Rowland mentioned, a Gandhāran peopled vine scroll (Figure 5) housed in the Boston Museum of Fine Arts derived from the Roman peopled vine-scroll typical of the eastern Roman school. It is inconceivable that such a heavy and bulky relief panel was physically transported from the Roman Empire to Gandhāra; moreover, the object is made of Gandhāran schist, which would confirm its provenance beyond any doubt. Although we have only scanty evidence of Roman artisans, I am nevertheless certain that Roman artisans must have come to Gandhāra and transmitted both their techniques and Dionysiac images or motifs to the region. There is also the possibility that the movement of objects can explain the presence of such imagery. A significant number of objects are known to have been exported from the Mediterranean area to the East, including Mesopotamia, Iran, Central Asia, India, China, and Mongolia (cf. Boardman 1994; 2015). These comprise mostly luxury pieces, such as silver vessels, and some depict Dionysiac images. All relevant pieces cannot be dealt with in this paper, however, some examples will suffice for reference. 15 Hansen et al. 2009: 399-400, cat. nos. 330-334. Cat. no. 333 is a medallion depicting Dionysos and his thiasos. 98 Tadashi Tanabe: The Transmission of dionysiaC imaGery To Gandhāran buddhisT arT A bronze head of Silenus was found at Begram, in addition to many plaster or gypsum casts of late Hellenistic metalwork. Several clay or terracotta moulds for plaster casts have been found in Gandhāra (Cambon and Jarrige 2010, fig. 221; Tanabe 2015: 63-64, figs. 1a, b, 7, 8,10-14, colour plate 4). Silver artefacts depicting Dionysiac themes were even found from as far afield as China and Mongolia, including for instance, a silver plate with Dionysos riding a panther, two stem cups with Dionysiac figures and a grapevine, and a medallion or phalera with a satyr attacking a maenad or goddess (Watt et al., 2004: 149, 184-185, figs. 59, 90; Sofukawa and Degawa 2005: fig. 95; Polosmak et al., 2011: 110-117, figs. 4. 40a, b; 4. 42). These finds were transported from west to east by overland and sea trade. The land route here denotes the so-called Silk Road through the Parthian Royal Road and the sea route traverses the Erythraean sea.16 Conclusion Based on my investigation of Gandhāran Dionysiac imagery it is now possible to answer the question posed by Peter Stewart: how were the artistic ideas of Dionysiac imagery transmitted to Gandhāra? Through movement of objects or people? It is highly conceivable that both ideas and the repertory of figures associated with Dionysiac imagery were not only transmitted as components of Hellenistic and Roman luxury objects, but also through the immigration of Roman artisans to Gandhāra. However, with the present state of knowledge it is very difficult to decide which of the two played the more decisive role, objects or people. The most important fact to be gleaned is that the Gandhāran Dionysiac imagery is nothing but a limited adaptation of many Dionysiac themes, deliberately appropriated and exploited by Gandhāran sculptors, most likely including Roman craftsmen and their local apprentices, in order to visualize the themes with which Gandhāran Buddhists desired to embellish their monasteries, such as the Buddhist afterlife, the Buddha’s life stories, and so forth. One recalls the proverb, ‘necessity is the mother of invention’, an expression duly applicable to this case concerning the Gandhāran absorption of Greek and Roman art. References Baratte F. 2002. East and West: A Central Asian Silver Hoard from the Parthian Era. London: Melisende. Baums S. 2018. Greek or Indian? The Questions of Menander and Onomastic Patterns in early Gandhāra. Pages 33-46 in H.P. Ray (ed), Buddhism and Gandhara: An Archaeology of Museum Collections. London and New York: Routledge. Boardman J. 1994. The Diffusion of Classical Art in Antiquity. London: Thames and Hudson. Boardman J. 2014. The Triumph of Dionysos: Convivial Processions, from Antiquity to the Present Day. Oxford: Archaeopress. Boardman J. 2015. The Greeks in Asia. London: Thames and Hudson. Boyer A.-M. 1911. Inscriptions de Miran. Journal Asiatique 17: 413-430. Buchthal H. 1945. The Western Aspects of Gandharan Sculpture. London: Cambridge University Press. Bussagli M. 1963. Painting of Central Asia. Geneva: Skira. Cambon P. and Jarrige J.F. 2010. Gerettete Schätze - Afghanistan - Die Sammlung des Nationalmuseums Kabul. Bonn: Kunst- und Ausstellungshalle der Bundesrepublik Deutschland. Cobb M. 2015. The Chronology of Roman Trade in the Indian Ocean from Augustus to Early Third Century AD. Journal of the Economic and Social History of the Orient 58: 362-418. Crowley P.R. 2018. The Farnese Sarcophagus in Antiquity. Pages 31-61 in C. Nielsen (ed), Life, Death & Revelry: The Farnese Sarcophagus. Boston and London: Paul Holberton Publishing. 16 Schoff proposed locations for the Parthian stations on a map and Young mentions that the Parthians played an important role on the Roman trade. Palmyrenes especially participated in the furthest stages of the trade with India, as well as in the caravan trade to the Gulf (Schoff 1914: 16, map; Young 2001: 136-148). 99 The Global ConneCTions of Gandhāran arT Dell’Aglio A. and Zingariello A. 2015. Marta: Il Museo Nazionale Archeologico di Taranto. Taranto: Scorpione Editrice. Drachenfels D. and Luczanits C. (eds) 2008. Gandhara: The Buddhist Heritage of Pakistan. Legends, Monasteries, and Paradise. Mainz: Philipp von Zabern. Errington E. and Cribb J (eds) 1992. The Crossroads of Asia: Transformation in Image and Symbol in the Art of Ancient Afghanistan and Pakistan. Cambridge: Ancient India and Iran Trust. Falk H. 2001. Annexe. Names and Weights Inscribed on Some Vessels from the Silver Hoard. Journal des savants 20: 308-319. Falk H. 2009. Making Wine in Gandhara under Buddhist Monastic Supervision. Bulletin of the Asia Institute. New Series, 23: 65-78. Filigenzi A. 2006. From Saidu Sharif to Miran. Indologica Taurinensia 32: 67-89. Foucher A. 1917. The Beginnings of Buddhist Art and Other Essays in Indian and Central-Asian Archaeology. Paris: Paul Geuthner; London: Humphrey Milford. Foucher A. 1922. L’art Gréco-Bouddhique du Gandhāra. 2 vols. Paris: Imprimerie Nationale. Hansen S., Wieczorek A., and Tellenbach M. 2009. Alexander der Große und die Öffnung der Welt Asiens Kulturen im Wandel. Regensburg: Schnell & Steiner. Hiebert F. and Cambon P. 2011. Afghanistan: Crossroads of the Ancient World. London: The British Museum Press. Horn H. G. 1972. Mysteriensymbolik auf dem Kölner Dionysosmosaik. Bonn: Rheinland Verlag. Ingholt H. 1957. Gandhāran Art in Pakistan. New York: Pantheon Books. Knauß F. 2017. Die Kunst der Antike: Meisterwerke der Münchner Antikensammlungen. München: Beck. Konow S. 1929. Kharoshṭhī Inscriptions with the Exception of Those of Aśoka. Calcutta: Government of India. Lehmann-Hartleben K. and Olsen E. C. 1942. Dionysiac Sarcophagi in Baltimore. Baltimore: Walters Art Gallery. LIMC. Lexicon Iconographicum Mythologiae Classicae, vol. 3, 1986. Zürich and München: Artemis Verlag. Marshall J. 1951a. Taxila: An Illustrated Account of Archaeological Excavations Carried out at Taxila under the Orders of the Government of India between the Years 1913 and 1934, vol.2. Cambridge: Cambridge University Press. Marshall J. 1951b. Taxila: An Illustrated Account of Archaeological Excavations carried out at Taxila under the Orders of the Government of India between the Years 1913 and 1934, vol. 3. Plates. Cambridge: Cambridge University Press. Marshall J. 1960. The Buddhist Art of Gandhāra: The Story of the Early School: Its Birth, Growth and Decline. reprinted in 1980. Cambridge: Cambridge University Press. Molesworth J.T. 1857. A Dictionary, Maraṭhi and English. Second edition. Bombay: Bombay Education Society’s Press. Nielsen C. 2018. Life, Death and Revelry: The Farnese Sarcophagus. London: Paul Holberton Publishing. Paribeni R. 1932. Le Terme di Diocleziano e il Museo Nazionale Romano, 2nd edn. Rome: Libreria dello Stato. Polosmak N.V., Bogdanov Ye.S., and Tseveendorzh D. 2011. Dvadtsatyi Noin-Ulinskii kurgan [The Twentieth Noin-Ula Tumulu]. Nobosibirsk: Inpholio. Rowland B. 1949. The Hellenistic Tradition in Northwestern India. The Art Bulletin 31: 1-10. Rowland B. 1958. Gandhāra Art in Rome. Artibus Asiae 21/ 3-4: 282-284. Rowland B. 1960. Gandhara Sculpture from Pakistan Museums. NewYork: The Asia Society. Rowland B. 1974. The Central Asia. New York: Crown Publishers. Salomon R. 2018. The Buddhist Literature of Ancient Gandhāra: An Introduction with Selected Translations. Somerville: Wisdom Publications. Schoff W.H. 1914. Parthian Stations by Isidore of Charax. Philadelphia: Commercial Museum. Schoff W.H. 1974. The Periplus of the Erythraean Sea. New Delhi: Munshiram Monoharlal Publishers. Siganidou M. and Lilimbaki-Akamati M. 2008. Pella: Hauptstadt der Makedonen. Athens: KulturministeriumKasse für Archäologische Mittel und Enteignungen. 100 Tadashi Tanabe: The Transmission of dionysiaC imaGery To Gandhāran buddhisT arT Skinner M.C. 2017. Marks of Empire: Extracting a Narrative from the Corpus of Kuṣāṇa Inscriptions. Unpublished PhD dissertation. University of Washington. Sofukawa H. and Degawa T. 2005. China: Crossroads of Culture. Tokyo: Tokyoinshokan Printing. Squarciapino M. 1974. Sculture del Foro Severiano di Leptis Magna, Roma: L’Erma di Bretschneider. Stančo L. 2012. Greek Gods in the East: Hellenistic Iconographic Schemes in Central Asia. Prague: Karolinum Press. Stein M. S. 1921. Serindia: Detailed Report of Explorations in Central Asia and Westernmost China, vol. 1. Oxford: Clarendon Press. Stoye M. 2008. The Deva with the Swaddling Cloth: On the Western Origins of Gandharan Birth Iconography and their Implications for the Textual and Art History of the Buddhist Saviour’s Nativity. Pages 1-25 in C. Bautze-Picron (ed), Religion and Art: New Issues in Indian Iconography and Iconology. (Volume 1 of the proceedings of the 18th conference of the European association of South Asian Archaeologists, London, 2005). London: British Academy. Tanabe K. 2006. Sex and Death through the Origin of the Buddha Image. Kyoto: Yanagiharashuppan, in Japanese. Tanabe K. 2015. A Study in Clay or Terracotta Casts: On the Model of the Figures Depicted on Gandharan Toilet-Trays. Ars Buddhica 340: 63-77 (in Japanese). Tokyo National Museum, NHK, and NHK Promotions. 2002. The Art of Gandhara, Pakistan. Tokyo: NHK and NHK Promotions. Tokyo National Museum., NHK, and NHK Promotions. 2003. Alexander the Great: East-West Cultural Contacts from Greece to Japan. Tokyo: NHK and NHK Promotions. Toynbee, J. M. C and Ward-Perkins, J. B. 1950. Peopled Scrolls: A Hellenistic Motif in Imperial Art. Papers of the British School at Rome, 18: 1-43. Turner R.L. 1966. A Comparative Dictionary of the Indo-Aryan Languages. London: Oxford University Press. Watt J.C.Y., Jiayao A., Howard A.F., Marshak B.I., Bai S., and Feng Z. 2004. China Dawn of a Golden Age, 200-750 AD (exhibition catalogue). New York: Metropolitan Museum of Art. Wheeler R.E.M. 1949. Romano-Buddhist Art: an Old Problem Restated. Antiquity 89: 4-19. Wheeler M. 1954. Rome Beyond the Imperial Frontiers. Harmondsworth: Penguin Books. Whitehouse D. 1989. Begram, the Periplus and Gandharan Art. Journal of Roman Archaeology 2: 93-100. Young G.K. 2001. Rome’s Eastern Trade: International Commerce and Imperial Policy, 31BC-AD305. London and NewYork: Routledge. Zanker P. and Ewald B.C. 2012. Living with Myths: The Imagery of Roman Sarcophagi. Oxford: Oxford University press. Zwalf W. 1996. A Catalogue of the Gandhāra Sculpture in the British Museum. 2 Vols. London: The British Museum Press. 101 Part 3 Asian influences (105-134): DOI: 10.32028/9781789696950-6 Buddha on the Rocks: Gandhāran connections through the Karakorum mountains1 M. E. J. J. van Aerde, A. D. L. Mohns, and A. G. Khan ‘It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.’ (Arthur Conan Doyle, A Scandal in Bohemia). Introduction This paper takes the Gandhāra region as a starting point to explore its wider connections as part of ancient trade networks from c. 300 BC onwards, often referred to as the early ‘Silk Road’.2 While taking a bottom-up approach to the archaeological record, we focus especially on the role and spread of the diverse Buddhist imagery along the trade routes that have been found throughout the Karakorum mountain range. In the nineteenth and early twentieth centuries, Gandhāra scholars worked from a near- exclusive focus on Greek and Roman influences on the art of the region. During such early excavations, as evidenced by their reports, materials were often taken out of their original archaeological and environmental contexts upon discovery.3 On the one hand this enabled the detailed art-historical studies of what was labelled ‘Graeco-Buddhism’,4 while on the other hand it remains an obstacle when we wish to analyse patterns of connectivity based on the much wider archaeological contexts from sites that were excavated during these early periods. Even within such a larger scope, however, the unique character of Gandhāran Buddhist art is undeniable. The arrival of Mediterranean migrations in this region does seem to coincide with the development of naturalistic sculpting techniques that would so significantly shape the repertoire of especially Buddhist material culture in Gandhāra.5 As a result of the unique character of its Buddhist material culture in particular, the Gandhāra region itself has in the past been studied mainly as a unique or even isolated region, too. But, as is evident from its extensive archaeological record, not only has the material culture of Gandhāra spread out well beyond its regional borders – the very region itself, in a geographical sense, was ideally positioned between the Hindu Kush and Karakorum mountain passes and the north-western ports of the Indian Subcontinent to enable routes of exchange and trade and their naturally resulting processes of connectivity. Gandhāra would have functioned as a natural nodal point, a crossroads, within a network of interaction, migration, and trade that went not The authors wish to thank Peter Stewart and Wannaporn Kay Rienjang for inviting Marike van Aerde to speak at the third Gandhāra Connections workshop ‘The Global Connections of Gandhāran Art’ at Oxford, 18th-19th March 2019, where part of this research was first presented. Van Aerde’s research project at Leiden University is generously supported by the Byvanck Fellowship (2017-2019). We also wish to thank Jason Neelis and Murtaza Taj, whose Applied Field Workshop of Gilgit-Baltistan Petroglyphs, hosted by the Karakorum International University (KIU) in July 2019, was attended by co-authors Alexander Mohns and Abdul Ghani Khan. We are particularly grateful to Muhammad Zahir (Hazara University) for his collaboration and expertise. Our gratitude is likewise due to the late Harald Hauptmann for sharing his data and experience concerning the Karakorum petroglyphs with Van Aerde (Heidelberg 2017-2018). We furthermore wish to thank Mike Kneppers for his work on the zoomorphic petroglyphs, and Beatriz Gomez de Silva for her assistance with the GIS charts. 2 We choose not use this term, as ‘Road/Roads’ incorrectly implies linear connections, which is not in line with the currently available archaeological data. The term ‘Silk’ is another issue: it refers to a traditional focus on luxury goods (which was based on fragmentary reference only, such as Pliny, Natural History, 12.84), while the archaeological records of the ‘Silk Roads’ trade exchanges in fact contain substantially more utility material, such as food stuffs, transport ceramics, spices and herbs, etc. For that reason, we here refer to ancient routes of exchange or trade networks. See also Van Aerde & Zampierin 2020 on the (misinterpretation of) archaeological statistics of early Afro-Eurasian trade routes. 3 Cf. Cunningham 1970; Marshall 1918; 1951. 4 As first coined by Foucher 1905. The concept is still in use today, e.g. see recently, Boardman 1994; 2015; Beckwith 2015. 5 Cf. Van Aerde (2018: 203-230) for a review of Gandhāran sculpture at Taxila in particular, where these contexts are discussed more widely. See also Rienjang & Stewart (eds.) 2018; 2019 for various recent explorations on Gandhāran art and sculpture. 1 105 The Global ConneCTions of Gandhāran arT only beyond the region but also beyond the Indian Subcontinent; and Buddhist art, it seems, became part of these processes.6 When we turn to the full (currently available) record of the output of Gandhāran material culture through these wider connections, sculpture and monumental architecture make up only a fraction of the quantity; ceramics are perhaps not the first or most artistically striking example that comes to mind when we think of Gandhāra, but pottery used for exchanging goods such as spices, herbs, oils, and rice constitutes by far the most numerous finds; in the ceramics assemblage from Taxila, for example, amphorae can be found with parallel types from Gujarat ports like Somnath and Devnīmorī, Arikamedu in the Tamil south, but also from Roman Palmyra, Petra, and Jerusalem, thus concretely linking Gandhāra with a widespread network of international trade exchange.7 In this chapter, we propose to study early Buddhist iconography within this same, wider context of the routes that led to and from the Gandhāra region. In particular, we focus on the Karakorum mountain range and the currently available dataset of rock carvings from that region. Specifically, after a brief introduction to the Karakorum archaeological record and state of research, this paper will: 1) examine the diversity of Buddhist iconography encountered by means of a database of all relevant anthropomorphic Buddhist carvings and their analysis; 2) use the above-mentioned data, in combination with statistical analyses of stūpa and animal carvings, to determine distribution patterns by means of GIS, and subsequently identify specific routes and nodal points along them. Throughout these analyses, we have found that certain patterns in Buddhist carvings coincide with specific nodal points and/or changes in the empirically determinable routes; in the concluding section of this paper we will elaborate on the implications of this and offer several new points for future discussion. On the one hand, we hope to contribute new knowledge concerning Buddhism depicted in the Karakorum range in particular, and on the other hand, we wish to contribute concrete new data about the wider connected networks that linked ancient Gandhāra directly to these mountains. The Karakorum carvings: state of research The Karakorum mountain range borders modern-day India, Pakistan, and China. It includes the GilgitBaltistan region in Pakistan, the Ladakh region in India, and the south-western Xinjiang region of China, and is part of the western edge of the Himalayas along with the Hindu Kush range bordering Pakistan and Afghanistan in the West (Figure 1). From 1979 onwards, the Karakorum Highway enabled accessibility to remoter mountain regions for archaeological campaigns. The first collections of petroglyph documentation were conducted by Jettmar and Dani, overseen by the German Research Council and subsequently by the Heidelberg Academy.8 The long-running project entitled ‘Rock Carvings and Inscriptions along the Karakorum Highway’ was initiated in 1983 as a collaboration between the Department of Archaeology and Museums (DOAM), One specific archaeological example of the apparent link between Buddhism and trade across the Indian Subcontinent is found in the diverse merchant patron portraits (with recognizable attributes and iconography from Arabian, Mauryan, Kushan, Bactrian, and Mediterranean origins) at Buddhist stūpas throughout Gandhāra and at Buddhist monasteries during Kushan times. Many of these portraits have now been separated from their original stūpas and remain unpublished. Cf. collections at the National Museum of Delhi; Kurita 2003 (for private collection records). For a more in-depth exploration of the connection between Buddhist religion and trade, see especially Neelis 2011; 2014b: 3-17; 2014b: 45-64. 7 For Taxila ceramics: Marshall 1951, vol. 3: pls. 121-128. For Devnimori ceramics: Mehta 1966. For Somnath ceramics: Nanavati et al. 1971. For Arikamedu ceramics: Begley et al. 1996; 2004. Cf. Van Aerde & Zampierin 2020 for a review of Mediterranean ceramics at Arikamedu and Berenike specifically. 8 Documented in the Materien zur Archäologie der Nordgebiete Pakistans (MANP) catalogues, volumes 1–11, edited by Gérard Fussman, Karl Jettmar, Ditte König et al. between 1989 and 1994, and subsequently from 2003 to 2011, under the auspices of the Heidelberg Academy. 6 106 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains Figure 1. Map of Central Asia. The research area is shown within black lines (Imagery: Google Earth, 2018) Government of Pakistan, and its regional branch at Gilgit, and the Heidelberg Academy, initially under the guidance of Jettmar and from 1989 of Hauptman.9 Since that time, fieldwork in the region has been rare and has prioritized digital documentation techniques and cultural heritage preservation.10 The Karakorum carvings must be considered to be under immediate threat. Rescue campaigns prior to the planned construction of the Diamer-Basha dam were restricted: the completion of the dam will flood an estimated 37,051 carvings found on 5,928 rock clusters.11 Moreover, the Buddhist carvings in particular have been targeted in situ across the Gilgit-Baltistan region and many have been damaged and/or hidden beneath paint, a practice that is continuing today and makes the study and continued preservation of these carvings increasingly hard.12 Because of this current state of the research, the MANP catalogues remain the primary available data; as a result, to date no systematic or statistical analyses are published of the currently catalogued data and so in-depth interpretative studies of the carvings, their contexts, and their implications remain a lacuna.13 Hauptman 2017-2018, personal communication. E.g. recent applied fieldwork hosted by Karakorum International University (KIU), supervised by J. Neelis and M. Taj (2019), as well as cultural heritage campaigns of the National College of Art at Lahore (at present unpublished). Currently no substantial excavations have been conducted at Karakorum sites, only petroglyph recording and surveys. 11 Cf. Schrader 2011: 1; Yusuf 2011; personal communication with Prof. Hauptmann, 2017. Cf. Van Aerde 2019: 459. Upon completion, the dam is estimated to submerge over twenty-four villages and households of c. 25,000 people. It would likewise submerge 110 kilometres of the Karakoram Highway. Cf. Khan 2018: 1. 12 Personal experience (KIU, Gilgit), Mohns and Khan 2019, and additional personal communication with Muhammad Zahir (Hazara University). 13 Van Aerde (2019: 455-480) provides a partial statistical analysis of Buddhist stūpa carvings. Khan (2018, MPhil thesis, supervised by Muhammad Zahir at Hazara University), provides a full statistical analysis of stūpa carvings from the DiamerBasha region. So far, publications apart from the MANP catalogues have mainly focused on selected case-studies from the Karakorum dataset, in particular pertaining to Buddhist art/iconographical studies and inscriptions. Cf. Dani 1983; 1995; Jettmar 1985; Jettmar 1993 and1989; Carter 1993; Fussman & Jettmar 1994; Zwalf 1996; Rhie 1999; Thewalt 2008; Hauptmann 2008 and 2009; Neelis 2014b. 9 10 107 The Global ConneCTions of Gandhāran arT In this chapter, we offer initial steps towards a more statistical as well as interpretative understanding of the currently known Karakorum carvings by focussing on the full dataset of anthropomorphic Buddhist carvings, and subsequently offering, in addition, cross-references related to stūpa and animal carving clusters. The existing documentation offers catalogues according to survey location, but does not include specified categorizations or subsequent interpretations of the carvings: combined, these records offer multiple thousands of carvings for analysis, including early Buddhist imagery of stūpas and anthropomorphic figures, as well as caravan and hunting scenes, a wide variety of animal species, and a diverse range of inscriptions.14 A first analysis of Buddhist imagery at selected field stations already indicated several remarkable location spikes and distribution clusters, and the preliminary emergence of patterns, which are explored in greater depth here.15 First of all, stūpa imagery make up by far the majority of Buddhist imagery among the Karakorum carvings, and the distribution of the carvings is almost exclusively found on rock clusters along the Indus river banks. From among the stūpa carvings, direct parallels of Gandhāran stūpa architectural types could be detected, featuring specific decorative designs not found on early stūpa architecture beyond that region.16 However, the majority of the carvings are small and of a basic style, of which parallels are found in the (less thoroughly documented) Hindu Kush mountains.17 Another remarkable parallel, and so far the only similar stūpa carving recorded beyond the Indian Subcontinent, is found at the Hoq cave on Socotra island (Yemen): quite recently, a Belgian team of geologists discovered, unexpectedly, hundreds of rocks carvings and inscriptions within one of the island’s sea caves, left by maritime traders from the first half of the first millennium AD who used Hoq cave as shelter while crossing the Indian Ocean.18 Carved on a wall near a Sanskrit inscription that mentions the Gujarat Satraps, several small stūpa carvings were found (Figure 2).19 Figure 2. A Stūpa carving from Hoq cave and parallels from among the MANP volumes. (Images after: De Geest 2012: 252-254; Bandini-König & Von Hinüber 2001). 14 As documented in the MANP volumes: Bennmann & König 1994; Fussman & König 1997; Bandini-König 1999; 2003; 2005; 2007; 2009; 2011; 2013; Bandini-König & Von Hinüber 2001; Bennmann 2001. 15 These findings are discussed in Van Aerde 2019: 455-480. Initial conclusions and implications were presented at the European Association of Archaeologists (EAA) conference in Barcelona in 2018, as part of the session entitled Advancing Global Rock Art as an Archaeological and Community Resource. 16 Van Aerde 2019: 465. 17 Kotera et al. 1971: 40; fig. 38; Van Aerde 2019: 466. 18 Socotra is briefly mentioned to that purpose in the Periplus Maris Erythraei. Recorded inscriptions from the Ho cave include Brahmi and Sanskrit, Ethiopian Ge’eze script, Palmyrene, and pre-Islamic Arabic, all from the first millennium AD. Cf. De Geest 2012: 232-253. 19 De Geest 2012: 252-254. 108 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains This presents new evidence that not only (large quantities of) utility ceramics travelled from Gandhāra, via the ports of Gujarat, across the Indian Ocean, but that Buddhism (specifically, Buddhist imagery) also travelled as far as the southern edge of the Arabian Peninsula. In type, these stūpas show distinct similarities with the early, basic stūpa types known from the Karakorum carvings – and, as a result, they present a concrete indication of an apparently tangible connection between Buddhism and the expanding trade networks that flourished across and beyond the Indian Subcontinent from the first century BC to the first century AD.20 Consequently, statistical and interpretative studies of the Buddhist carvings of the Karakorum range are highly relevant for our understanding of the spread of Buddhism as a religion and of its practical application as ‘lived religion’ along these mountain routes.21 Moreover, they provide substantial archaeological evidence for in-depth study of the connecting patterns between Buddhism and trade routes. And lastly, as an integral part of the complete dataset of Karakorum rock art, along with animal scenes and inscriptions, they provide invaluable data to help reconstruct the actual, physical routes that led from the Gandhāra region to the East. Our first step, to this end, is to provide the full overview and interpretative analysis of all currently recorded anthropomorphic Buddhist carvings. In the subsequent part, where we turn to distribution statistics, we will also include data concerning the record of stūpa carvings as well as our initial progress in cataloguing all recorded zoomorphic carvings according to taxa and distribution. Anthropomorphic Buddhist carvings The full dataset presented in the Appendix consists of all recognizably Buddhist images that feature full or partial elements of anthropomorphic depiction. The current dataset consists of 118 individual carving entries. Among these, alongside the anthropomorphic images, some depictions of animals, stūpas, and inanimate objects are likewise included when they constitute integral parts of the anthropomorphic scene in question. For the purpose of our analysis, we have devised four main categories: the Buddha, bodhisattvas, humans (devotees), and spirits/celestials. In our Database 2 additional categories are included: animals as part of an anthropomorphic scene, and inanimate objects as part of an anthropomorphic scene. The statistical distribution of the four main categories is given in Figure 3.22 The Buddha In total thirty individual, clearly defined images of the Buddha were found, from four different locations: Chilas (four), Thalpan (twenty-one), Shing Nala (four) and Shatial (one jātaka scene).23 While all these recorded depictions feature the recognizable attributes of the Buddha (such as the uṣṇīṣa and the ūrṇā), it is immediately evident that they do not follow a single or even remotely uniform stylistic pattern. In fact, they seem to be evidence of a remarkable variety of different depicted attributes and characteristics, as well as what appear to be diverse carving methods, as will be explored in more detail below. As likewise raised by Neelis 2011; 2014a: 3-17; 2014b: 45–64. The concept of ‘lived religion’ was suggested by Jörg Rüpke (2016) for Roman religious contexts, but could be more widely applicable. 22 Fussman & Bandini-König 1997 (Shatial data); Bandini-König & Von Hinüber 2001 (Shing Nala data); Bandini-König 2003 (Chilas Bridge data); Bandini-König 2003; 2005 (Thalpan data). 23 At Shatial there are two additional possible Buddha figures, but they are only partially preserved and unclearly defined, providing only an abstract outline without iconographical details, and for that reason they are not included in our present database. The only clearly defined Buddha at Shatial is from rock 34, which is part of a depiction of a jātaka scene, containing an image of the Buddha in one of his previous lives as a bodhisattva. This is an important distinction, as in our database (Appendix, Table 1) we aim to maintain depictions of bodhisattvas and the Buddha as distinct categories, and have therefore grouped this carving under ‘jātaka scenes’. 20 21 109 The Global ConneCTions of Gandhāran arT Figure 3. Graph showing the distribution of anthropomorphic Buddhist petroglyphs (by M.E.J.J. Van Aerde & A.D.L. Mohns, 2019). Bodhisattvas In Buddhist cosmology, a bodhisattva is recognized as an individual who is on his or her way to attain enlightenment.24 Unlike the image of the Buddha, bodhisattvas can include any living being, either animal or human. Moreover, the jātaka tales of the Buddha’s former lives portray the Buddha Shakyamuni as a bodhisattva. For this reason, images of bodhisattvas are deliberately different from depictions of the Buddha; they are mostly recognizable by their specific characteristics of adornment or in the context of the depicted scene of which they are part. In total fifteen bodhisattvas were found within the dataset. The largest concentration is located at Chilas (eight), followed by Thalpan (six) and Shatial (one). Among these, Avalokiteśvara and Maitreya are the most clearly identifiable bodhisattvas in the dataset, each with three depictions, followed by the bodhisattva Mañjuśrī in two carvings at Thalpan and Chilas. In addition, there are three recognizable jātaka scenes that depict the Buddha as bodhisattva, one each at Chilas, Thalpan, and Shatial. The remaining four bodhisattvas are unidentified; three of them (64:16, 84:2, and 8:1) seem to be depictions of either Mañjuśrī or Maitreya, but the lack of clearly determining characteristics/attributes (such as Maitreya’s water flask) forces us to classify these images as unidentified. Humans (devotees) The majority of human images within the dataset are representative of practitioners and devotees of Buddhism. In total there are twenty-one human depictions within the dataset that cannot be classified as the Buddha or bodhisattvas. These images are found at Chilas (seven), Thalpan (seven), and Shatial Bodhisattvas occur in Buddhist scriptures since the earliest jātaka texts and the Buddhacarita, and are part of some of the earliest known examples of anthropomorphic Buddhist material culture. Cf. Nagar 1993; Shaw 2006; Krishnan 2009. 24 110 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains (seven), but none at Shing Nala that can be clearly identified as such. Several of these human images, especially at Chilas, are related to/clustered alongside stūpa depictions (e.g. groups 30:1, 30:2, 64:18 and scene 31:A). Human depictions associated with the Buddha are usually part of a specific scene from the Buddha’s life (e.g. 194:K, depicting the Buddha’s first sermon to his five future disciples at Sarnath). In some cases, the human figures associated with the Buddha are less clearly identifiable (e.g. 195:I, which contains an image of the Buddha alongside what appear to be several attendants who cannot be identified as bodhisattvas but may represent devotees/Buddhist practitioners).25 Spirits/celestials The number of carvings depicting spirits and/or celestial beings from Buddhist mythology is significantly lower than the aforementioned categories. The dataset contains only four, from Thalpan (three) and Chilas (one). Nonetheless, they do form an important role in determining the context of the image clusters of which they are part (e.g. 195:W, which depicts the scene from the life of the Buddha known as the ‘Temptation of Māra’, with two celestial beings flanking the Buddha who can be identified as the daughters of Māra, which is directly reminiscent of the well-known descriptions of this passage).26 A note on non-anthropomorphic depictions In many of the above-mentioned carving groups, non-anthropomorphic images constitute an integral part of the specific scenes depicted. These include animals, smaller inanimate objects, and stūpas, and in some cases their presence is crucial for identifying specific Buddhist scenes (e.g. 30:B from Chilas, which depicts the Tigress Jātaka, and 30:X from Thalpan, which depicts the Ṛṣipañcaka Jātaka. The identification of these scenes relies directly on the specific animals depicted, namely, the tigress and cubs in 30:B and the pig, crow, pigeon, snake, and deer in 30:X). In addition, the presence of recognizable stūpas in direct relation to anthropomorphic figures helps us to interpret practical aspects of these carvings as well, as stūpas themselves are objects of Buddhist devotion and worship, and their functionality thus translates, by association, to the anthropomorphic figures with which they are depicted. While all our basic interpretations of the relevant carvings are included in Table 1 (Appendix), many of the depictions necessitated an in-depth comparative analysis because of the rather complex variety encountered, in terms of iconographical content, rendering techniques, and the more elusive aspects of artistic style and its related chronological nuances. While style and iconography cannot be regarded as an exact or empirical means to determine chronology, it cannot be ignored that many of the Karakorum anthropomorphic Buddhist depictions seem to suggest a long-term continuation of a Buddhist presence and the changing artistic influences that went along with these processes up until at least the eighthninth century AD. At Thalpan and Chilas Bridge especially, a variety of iconographies and styles are evident. In line with this, we here discuss five selected carving groups, representing all four above categories, from either Thalpan or Chilas specifically.27 Some of these figures appear to be sweeping the ground around the depicted stūpa and Buddha, while others kneel down beside it. Based on this visual reference only, they may be devotees and/or attendants. With the current data it is not possible to tell whether these figures were a later addition to the stūpa and Buddha carving or created simultaneously to form a particular scene (195:I presents the clearest example of such a potential scene). 26 The demon Mara and his daughters occur in many Buddhist traditions and texts, including the Buddhacarita and the Mārasaṃyutta passage in the Samyutta Nikaya scripture. On the representation of Māra throughout Buddhist art and literature, see Guruge 1991: 183-208. 27 Owing to size limitations we cannot include our full analyses here. This chapter offers a selection that we deemed most suited to the context of this particular publication. We refer to the full database (Table 1) of anthropomorphic Buddhist carvings for further reference. 25 111 The Global ConneCTions of Gandhāran arT Figure 4. Carving 64:16 (Chilas); Buddha Vipaśya seated in dhyānamudrā. (After Bandini-König 2003, table 2.) Figure 5. Carving 64:18 (Chilas); Siṇhoṭa (devotee). (After Bandini-König 2003, table 1.) Figure 6. Seated Crowned Buddha. Kashmir, c. tenth century AD. (After Pal 1975: fig. 32). Figure 7. Chinese seated Buddha. Gilt bronze, dated c. AD 338. (After Rhie 1999: fig. 2.2). 112 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains 1. The Buddha: 64:16 (35 x 101 cm, Chilas) Carving 64:16 of a seated Buddha in dhyānamudrā constitutes a unique image within the dataset (Figure 4). Most depictions of the Buddha present him as an ascetic in plain robes, but here we find him depicted as a crowned and richly adorned figure also known as the ‘Bejewelled Buddha’.28 In a public lecture in 2011, Hinüber argued that this is an image of the Buddha Vipaśya, which is not frequently shown in Buddhist material culture, based on the inscription associated with the image (64:17).29 Also associated with this image is a human individual (64:18; Figure 5), which appears to be a devotee and, quite possibly, the donor or patron of these carvings; this person can be identified as Siṇhoṭa from the same inscriptions associated with this particular Buddha image.30 This human figure is depicted wearing garments that can be recognised as Central Asian or possibly related to more Southern regions.31 This Siṇhoṭa figure holds an incense burner in his right hand and a mala, bead necklace for meditation, in his left. The figure has been rendered in a simplistic, nearly abstract manner with little elaboration of his garments or personal adornments, apart from the objects in his hands and his headgear.32 This image of the Buddha Vipaśya itself can be compared to several bronze sculptures from Kashmir, such as the seated crowned Buddha in Figure 6.33 Remarkably similar clothing and crown iconography is apparent in these Kashmir sculptures and is only known from this region, and in some other parts of eastern India.34 These bronze sculptures from Kashmir have been generally dated to the eighth-tenth century AD, which indicates the period in which Buddhism truly flourished throughout these regions and that coincides with the Tang Dynasty era in China, which led to a great flourishing of trade networks as well.35 Apart from these similar clothing and adornment styles used for Buddhist iconography in Kashmir, this particular Buddha Vipaśya also shows another clothing style comparable to Buddhist gilt bronze figurines known from various Chinese sites, from the early to mid-first millennium AD (Figure 7), namely, the specific shape of the round, drooping drapery of the lower robe and the pleated folds across the wrist, which is also observed in 30:22, 64:14, 172:1, 194:65, and 38:13 in this dataset.36 These Chinese bronzes, in turn, may have had some manner of influence from much older sculptures, such as those known from the first to fourth century AD Kara-Tepe monastery in Uzbekistan, which is well-known for its Gandhāran sculptural influences and where perhaps the oldest example of this round, drooping drapery style was found.37 Lastly, another remarkable comparison is found in the terracotta figure of a Bejewelled Buddha found at the settlement and monastery at Fondukistan in Afghanistan, dated from the second-seventh century AD, which is likewise known for its Gandhāran Buddhist art; the three-fold triangular shape of the jewelled chest garment is remarkably similar to the one worn by our Bejewelled Buddha carving from Chilas.38 Kim 1997: 235. Inscription 64:17 in Bandini-König 2003: 191-196. Cf. Von Hinüber’s public lecture on ‘Bronzes of the Ancient Buddhist Kingdom of Gilgit’, Metropolitan Museum of Art, New York, 2011. 30 As mentioned by Hinüber in New York 2011 (but remains unpublished). 31 Jettmar 1987: 22. 32 Parallels for devotees or possible stūpa donors can be found at Thalpan stone 63:1, which features a nearly identical figure holding prayer beads and incense burner (Bandini-König 2003: table 1, as well as 116:5, depicting a kneeling abstract human figure beside a large elaborate stūpa and a seated Buddha. Bandini-König 2005: table 83. Other possible devotees are 30:2 and 30:21 at Thalpan (Bandini-König 2003: table 1). 33 Pal 1975: 106 (fig. 32): seated crowned Buddha, Kashmir, dated to c. tenth century AD. 34 Krishnan 1996: 132. 35 On the archaeology of Tang Dynasty, its capital Chang’an, and its connection to global trade networks, cf. Liu 2010: 62-108; Hansen 2016: 579-881. 36 For full references see Table 1 (Appendix) entries on these particular carvings: from Chilas, Thalpan and one at Shing Nala. Cf. Bandini-König & Von Hinüber 2001; Bandini-König 2003; 2005. 37 Rhie 1999: 34. 38 The ‘Buddha adorned with three-cornered cloak’ was found at the Ghorband Valley, Fondukistan Monastery, Niche D, and dated to c. seventh century AD. It is made of unbaked clay, 72 x 24 cm. First published in Klimburg-Salter 1989: pl. XXXII. Currently at Musée Guimet, Inv. no. MG 18960. (Owing to copyright, no image of this famous work is reproduced here.) 28 29 113 The Global ConneCTions of Gandhāran arT These parallels may indicate a form of continuity, then, within both the Kashmiri and wider Karakorum regions; Kashmiri Buddhist material culture is generally regarded to have been strongly influenced by the Gandhāran style from the early first millennium AD onwards and this may be similarly true for Buddhist imagery throughout the Karakorum.39 We might speculate, then, that this process of adaptability and addition of diverse stylistic and iconographical aspects would have continually been accumulated, up to the eighth-ninth centuries at least, by artisans depicting Buddhist scenes such as that represented in carving 64:16 at Chilas. 2. Bodhisattva: 63:6 (53x133cm, Chilas) Carving 63:6 (Figure 8) presents an image of a bodhisattva standing upright, surrounded by a decorated halo reminiscent of leaves sprouting from Figure 8. Carving 63:6 (Chilas): a Figure 9. Carving 63:4 (Chilas): the lotus flower pedestal upon which bodhisattva (Maitreya). (After a bodhisattva (Avalokiteśvara). he stands. His hands are in abhayamudrā Bandini-Konig 2003, table 3.) (After Bandini-Konig 2003, table 3.) and he wears an elaborate crown, a sash across his torso, and decorated garments covering his lower body; he is moreover adorned with jewellery, a beaded necklace and bracelets. In his left hand he carries a water vessel, which is known as a kamaṇḍalu. This iconography, especially in respect to the kamaṇḍalu, garments, and headdress, clearly identify this figure as Maitreya.40 In the dataset, an almost identical carving can be found (64:19) in close proximity to this one and to carving 63:4, which portrays the bodhisatva Avalokiteśvara and shares several stylistic similarities, including the lower garment, decorated halo, and crown (Figure 9). It is noteworthy, however, that this Maitreya stands on a lotus flower that is placed in turn upon a pedestal, giving the impression of a physical sculpture rendered in a two-dimensional carving. Another unique aspect is the Maitreya’s pectoral area, which has been rendered to give the appearance of a muscular upper torso, whereas the Avalokiteśvara of 63:4 is portrayed with a very slender physique. The three-pointed and leaf-like crown worn by both this Maitreya and Avalokiteśvara show a remarkable similarity to crowns found in the tombs at the site of Tillya Tepe in Afghanistan, specifically tomb VI, which have been dated to the first century BC-first century AD, coinciding with the particular flourishing of Gandhāran Buddhist art at the time.41 Another parallel is the Maitreya figure 64:19 at Chilas, which has iconography nearly identical to that of 63:6.42 Another comparison for the Maitreya in particular is a bronze sculpture from the Swat region (Figure On the development of Buddhist art in Kashmir see Kaul 2005: 159-171; Rhie 1999: 162-239. Cf. Pal 1975; Krishnan 2009. Kim 1997: 4; Iida 2016, trans. of Taisho vol. 14, 454 (on the descent of Maitreya Buddha and his enlightenment). Maitreya is known as the next Buddha to succeed the most recent one, Buddha Śakyamuni, who was also known as Siddhārtha Gautama. 41 Hiebert & Cambon 2008: 284. The crown from tomb VI at Tillya Tepe in currently in the National Museum of Afghanistan, Kabul. Owing to copyright restrictions, an image of the famous crown is not reproduced here. 42 Bandini-König 2003: table 4 (see also Table 1 in Appendix). 39 40 114 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains Figure 10. Standing Maitreya from the Swat Valley. Bronze, inlaid with copper and silver. Lost; formerly in Berlin. (After Barrett 1962, fig 20.) 10); this bronze was originally located in Berlin but has been lost since the Second World War.43 Despite its lack of halo, the bronze in the photograph shows remarkably similar iconography and stylistically identical elements, such as the shape of the water vessel, the lower decorated drapery, the sash, beaded necklace, position of arm bracelets, muscular torso, and the leaflike pointed crown and headgear, with tassels that flow down from the side of the crown onto the Maitreya’s shoulders. Some of these iconographical similarities can also be found in the seated Maitreya statue dated to Kushanera Gandhāra, first-second century AD, currently at the National Museum of Tokyo.44 Although seated and rendered in a more clearly naturalistic style, this Maitreya wears a similar lower garment and has a muscularly defined physique, but especially the beaded necklace, arm bracelets, and the tassels that fly down in a wide angle from the smaller headgear/crown are comparable to the Maitreya carving at Chilas, while they do not seem to be typical attributes of Gandhāran Maitreya sculpture overall.45 3. The Buddha and bodhisattva: 194:151 and 194:152 (42 x 57cm and 7 x 11cm, Thalpan) The carvings 194:151 and 194:152 are part of one scene, depicting the Buddha seated with his hands held in varadamudrā, the wish-giving mudrā. He wears ascetic robes that are carved to give the impression of a much lighter fabric texture compared to the double or multiple pleated technique found in many of the other Buddha carvings at Thalpan (e.g. 30:22, 172:1, 194:65, 195:429). The double-lined halo is of a simple design, undecorated, and the Buddha is seated upon a lotus flower with two petal rows. The face and body present a more naturalistic rendering of the anthropomorphic physique than the other anthropomorphic carvings in the data set, even offering a suggestion of subtle perspective to give a more three-dimensional quality to the image; the seated Buddha is shown at a nearly three-quarter angle, while all other anthropomorphic Buddhas in the dataset are rendered frontally. The contours of the body are very round and give the impression of an overall smooth rendering, in part by means of the thinner carving lines used compared to the majority of the other anthropomorphic carvings. The Buddha’s facial features seem reminiscent of the Gupta style recognizable from the third to sixth century AD across the Indian Subcontinent, which emphasizes fleshy lips and the roundedness of the face (the fleshy lips can also be observed in carvings 30:22 at Thalpan and 38:13 at Shing Nala, but without any of the other characteristics of naturalism, perspective, and thinner carving lines).46 Accompanying this Buddha is a richly adorned bodhisattva figure rendered in the same style and also at a subtle three-quarter angle, and his head tilted slightly to the right. He wears an elaborately adorned, pointed crown (which, in turn, is reminiscent of the bodhisattva crowns discussed above), lightly rendered garments and multiple types of jewellery such as earrings and bracelets. In his right hand he holds a delicate, thin flower, while in his left hand he holds a very recognizable rendering of the Vajra (lightning bolt), which Filigenzi 2015: 106-107; Barrett 1962: fig. 20. National Museum of Tokyo, currently on display, no further details known (it is listed only as Kushan, from Gandhāra, north India). Owing to copyright restrictions, no photograph of the sculpture is reproduced here. 45 Cf. Harle 1994: 59-71; Rhie 1999: 1-4; Kim 1997; Krishnan 2009 on various Maitreya portrayals. The recognizable attributes here are the crown, physique, water flask, and prayer beads. In this case also the tassels of the headgear and specific details of the jewellery and clothing provide a parallel. 46 Harle 1994: 87-122. 43 44 115 The Global ConneCTions of Gandhāran arT makes the figure directly identifiable as the bodhisattva Vajrapāṇi, the well-known yakṣa disciple, attendant, and guardian of the Buddha.47 Especially with the rise of Vajrayāna Buddhism mainly across the north of the Subcontinent from around the third century AD (coinciding with the rise of the Gupta Empire), similar depictions of Vajrapāṇi become very common among Buddhist imagery. In respect to its technical rendering, the particular style of drapery is recognizable as the so-called ‘clinging technique’ of the garments around the legs and arms made popular in Sarnath during the Gupta era from around the late fourth century onwards.48 The smaller, close folds of Vajrapāṇi’s arm drapery are likewise indicative of the Gupta tradition.49 An interesting parallel may be found in the Vajrapāṇi depicted as part of the Buddhist wall paintings known from the Ajanta caves in Maharashtra (first phase dated c. second century BC up to the early fifth century AD, and the second phase c. fifth to seventh century AD).50 Figure 11. Carvings 194:151 and 194:152; the Buddha and bodhisattva (Thalpan). (After Bandini-König 2005, table 4.) While this Ajanta painting (Figure 12), most likely from the late Gupta era, is notably more rich in detail and elaborate in appearance than carving 194:152, both depictions of Vajrapāṇi are rendered in recognizable naturalistic style; they are positioned in a slight three-quarter angle with head tilted to the right; they both hold a thin, delicate flower in their right hand (the left hand of the Ajanta figure is too damaged to recognise a Vajra); they both wear a remarkably similar, elaborate crown with pointed tips and lush decorations; and they each wear long earrings and a lower garment tied at the waist by a thin sash and the fabric rendered according to the ‘clinging technique’, giving the appearance that the garment clings tightly to the upper legs. Consequently, both the Ajanta painting and the Thalpan carving reflect recognizable elements of late Gupta style and technique in their respective depictions of the yakṣa bodhisattva Vajrapāṇi and his attributes. 4. Celestials: 95:428 and 195:430 (11x30cm and 15x30cm, Thalpan) This carving group depicts an important scene from the life of the Buddha, the temptation of Māra: the demon Māra attempts to prevent the Buddha’s enlightenment by tempting him with earthly pleasures, therefore he sends Figure 12. Cave 1, Ajanta Caves: Mural of Vajrapani, 25. (Photo: copyright Archaeological Survey of India, ASI.). 47 On Vajrapāṇi in narrative Buddhist reliefs, see e.g. Santoro 1979: 293341; Zin 2009: 73-88. 48 Rani Tiwari 1998: 73. 49 Ganguly 1987: 74, 108; Harle 1994: 111-117; Rani Tiwari 1998: 73-76. 50 See most recently, Spink & Yaguchi 2014: 16-64. The Ajanta complex features thirty caves with elaborate paintings and carvings. The excavation reports of the Archaeological Survey of India (ASI) are currently unpublished. 116 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains Figure 13. Carvings 95:428 and 195:430: the Buddha and the daughters of Māra (Thalpan). (After Bandini-König 2005, table 94.) his daughters to seduce him, but the Buddha resists them and gains his enlightenment (Figure 13).51 The Buddha in this scene (carving 195:429) holds his arm positioned in bhūmisparśamudrā, touching the earth in order to call upon the earth goddess to witness his enlightenment. The celestial beings known as the daughters of Māra (carvings 195:428 and 195:430) are depicted as half-naked women in dancing positions, flanking the Buddha.52 They have their backs turned to him and their hands held joined above their heads; this could be interpreted either as their dancing or as their flight from the Buddha as he reaches enlightenment. The carving style of the two celestials seems more naturalistic than that of the Buddha figure they flank, although judging by the similar thickness of the carving lines and the rounded shapes of the body silhouettes, all three figures seem part of a single scene and were most likely carved simultaneously. Especially for the celestials, a basic attempt has been made at creating proportional body parts and natural postures, by means of perspective and overlap. A remarkable comparison for the dancing posture of the daughters of Māra is known from Gandhāran relief panels.53 Figure 14. Carving group 30:D, Sibi jātaka scene (Thalpan). (After Bandini-König 2003, table 53.) For an overview of the Buddha’s encounters with Māra, see Guruge 1991: 183-208. In the Samyutta Nikaya scripture 4.25, v. 518, the three daughters of Māra are described as stripping naked in front of the Buddha in order to tempt him. This passage seems to correspond to the specific scene depicted by the rock carving discussed here. 53 A particularly notable parallel is provided by a schist relief in the Museum Rietberg, inv. no. RVI 25. However, the unpublished relief was acquired from antiquities market and does not have a verified collection history or archaeological provenance. It was 51 52 117 The Global ConneCTions of Gandhāran arT In this relief, the celestial woman to the Buddha’s left holds her hands raised and joined above her head in the same way as the celestials in the Thalpan carving, and also shows similar posturing of the legs in a suggestion of dancing and contrapposto. The Buddha in the Gandhāran relief is also depicted in bhūmisparśamudrā and wearing a similar, folded plain ascetic robe as the Buddha in the carving. However, the daughter of Māra in the relief is fully clothed, although her body shape is visibly emphasized, whereas the celestials of the carving are naked apart from their jewellery and small, leaf-like skirts. So, while several Figure 15. Gandhāran relief of the Śibi Jātaka. London, British Museum, elements of the scene directly match, in inv. 1912,1221.1 (Photo: courtesy of the Warburg Institute, London.) particular the visual narrative elements of the Buddha and the dancing celestials required to tell this tale, there are distinct differences in the rendering of garments and attributes. 5. Śibi jātaka: 30:38 and 30:39 (51x30cm and 40x64cm, Thalpan) The scene catalogued as 30:D consists of two anthropomorphic figures (30:38 and 30:39), flanked by several stūpa carvings on the same rock surface, and can be identified as a carving of the Śibi Jātaka, which tells of a king (a former life of the Buddha) who, out of compassion, feeds part of his own flesh to a hawk in order to save the life of the dove that the hawk wished to eat. The scene from our dataset offers a simple, visual version of the story: the first anthropomorphic figure (30:38) is recognizable as King Śibi himself, who is seated and has raised his right arm while brandishing a knife over his own left arm. Beside him stands another man who holds a set of scales (30:39), used to measure the right amount of flesh that the king offers the hawk in exchange for the dove, which is depicted within one of the scales for balance (the right scale in the image, on the figure’s left). The opposite scale contained an undefined round mass, which seems to represent the king’s flesh. In respect to clothing and attributes, King Śibi can be recognized according to his rank by his more elaborated robes and the adorned headgear he wears, which is a stark contrast with the nudity of the second figure. The Śibi Jātaka is often depicted in Buddhist material culture, e.g. at the above-mentioned Gupta-era Ajanta cave paintings, as well as in several relief panels from the Gandhāra region, around the first to third century AD. One particular Gandhāran relief is strikingly similar (Figure 15; cf. also Stewart’s chapter in the present volume).54 This image shows a more elaborated visual rendering of the jātaka, including many more individuals present at the scene: to the left of the scale-bearer we see a bodhisattva carrying the vajra, while King Śibi is once again seated to the left of the panel, in similar adorned garments and headgear, while attendants cut flesh from his exposed leg, and with a bird seated directly below him (which could indicate either the hawk or the dove). But most interesting about this particular relief is the depiction of the scales: the detailed rendering of their various components is directly comparable to those in carving 30:39. This would indicate that this particular type of scales, with measuring parts apparently made out of cloth or hide that are tied with rope to a long stick, would have been in use and thus the most recognizable displayed in the ‘Next Stop Nirvana’ exhibition at Museum Rietberg, Zurich, 2019. 54 London, British Museum, inv. no. OA 1912,1221.1; Zwalf 1996: 85, pl. 136. 118 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains type of weighing tool for a visual narrative in both the regions where the relief was made (Gandhāra, northern Pakistan) and the Karakorum carvings to the north-east. Taken overall, the Karakorum dataset of anthropomorphic Buddhist carvings consists of highly diverse depictions in varying styles that contain recognizable, iconic components from Central Asian and Chinese material culture, the Gandhāra region, and the wider Indian Subcontinent throughout the first millennium AD. From our current analyses, pertaining to the individual anthropomorphic images of the Buddha and bodhisattvas, we have been able to discern two images that contain Central Asian styles (63:4, 194:65), one image that is comparable to Buddhist sculptures known from Kashmir (64:16), one image that can be compared to sculpture known from the Swat region (63:6), one image that is most comparable to parallels from the post-Gupta period (194:151, 152), and two images that show evidence of a combination of stylistic components from China, Central Asia and regions such as Kashmir and Swat (64:14, 30:22), with an additional three images that seem to lack a clear stylistic identification (30:18, 38:13, 172:1). We must note here, however, that while some depictions give clear indications for a certain stylistic influence from a particular region, this is in most cases based on the presence of several important recognizable characteristics or attributes, such as the type of crowns worn by the Buddha or bodhisattvas. But alongside these indications, in many instances, the same images also contain indications of different stylistic influences at the same time. Essentially, in most cases we seem to encounter a kind of fusion or mixture of multiple styles, sometimes even from multiple time periods, converging in these carvings. It appears to be this flexibility and adaptability that produced this particular and, indeed, unique dataset of Karakorum anthropomorphic Buddhist depictions. Distribution Following the selected analyses above, in this final section we ask what new insights can be gleaned from a wider statistical analysis of the distribution of the Karakorum carvings throughout the mountain range. Studies concerned with Buddhist art and the spread of the religion have primarily highlighted examples of Buddhist carvings from among these datasets, but such a selected focus does not allow for a comprehensive overview and, subsequently, cannot give an empirical interpretation of the carvings’ statistical distribution through the mountains and the implications this might have for humanenvironment relationships, specific indications of physical routes, and emerging patterns and/or variables of human travel, presence, and religious practice throughout the Karakorum. To that purpose, we combined three statistical datasets of 1) anthropomorphic Buddhist carvings, 2) stūpa carvings, and 3) zoomorphic carvings for our analysis here.55 The first noticeable aspect is quantity. In total, 118 identifiable anthropomorphic Buddhist images have so far been documented among all known carving sites. In comparison, a total of 485 stūpa carvings have been identified in the Diamer-Basha reservoir area (Chilas, Shing Nala, Gichi Nala, Oshibat, Dadam Das).56 Combined with stūpa carvings recorded at Thalpan (937) and Shatial (156), this dataset comes to a current total of 1,578 stūpa carvings.57 The total number of zoomorphic carvings currently documented at all known carving locations is 2,976, with a 55 Data pertaining to the anthropomorphic Buddhist carvings was gathered and analysed by Alexander Mohns (2018) – is this also a bibliographical reference? If so, please could you provide details?, and the statistical analysis of zoomorphic carvings was conducted by Mike Kneppers (2019) – is this also a bibliographical reference, if so, please could you provide details? , both as part of Van Aerde’s project at Leiden University. The statistical analysis of stūpa carvings in the Diamer-Basha region was conducted by Abdul Ghani Khan and supervised by Muhammad Zahir at Hazara University, Pakistan (2018) at Hazara University, Pakistan. The additional statistical data for stūpas and Shatial and Thalpan is from Van Aerde 2019: 455-480. 56 From this total, 352 stūpas can be securely identified, the remaining 133 carvings are less well preserved or somewhat unclear in their design, even though their general form and shape is highly reminiscent of Buddhist stūpas (Khan 2018, MPhil research conducted under supervision of Muhammad Zahir). Cf. on the interpretation and implications of Karakorum stūpa carvings: Van Aerde 2019 460-460. 57 All data based on MANP vol. 1, 2, 6–11, pertaining to Oshibat, Shatial, Thalpan and Chilas field stations: Bennmann and König 1994; Fussman and Bandini-König 1997; Bandini-König 2003; 2005; 2007; 2009; 2011; 2013. 119 The Global ConneCTions of Gandhāran arT Figure 16. Graph showing the distribution of stūpa, zoomorphic, and anthropomorphic Buddhist petroglyphs (compiled by Van Aerde, Mohns, Khan, Zahir & Kneppers, 2018-2019). taxa distribution of 96.7% Mammalia, 2.4% Aves, and 1.0% Reptilia.58 In the graph in Figure 20, these basic statistics are combined with those of the anthropomorphic Buddhist carvings from the first section of this paper. Two patterns are immediately evident: first of all, the fact that anthropomorphic Buddhist carvings are noticeably rare compared to the stūpa and zoomorphic carvings throughout the dataset, and secondly, the fact that of all three categories the largest number of carvings is found at Thalpan (even if the total number of anthropomorphic carvings at Thalpan is significantly lower than the number of stūpas and zoomorphic images, it is still the largest of its category). A third interesting pattern is the fact that the second largest recorded cluster of zoomorphic carvings is at Hodar, where no anthropomorphic Buddhist carvings were found, and there is a lower number of stūpa carvings (122) compared to the high number of zoomorphic carvings (715); we will return to this point below. Anthropomorphic Buddhist carvings are currently documented only at four locations: Thalpan, Chilas, Shing Nala and Shatial, of which the large majority are found at Thalpan (thirty-seven) and Chilas (twenty). For this reason, previous studies of Buddhist art especially have hypothesized the emergence of a Buddhist petroglyph shrine at these locations, which are in each other’s direct vicinity.59 At both Chilas and Thalpan, we indeed find remarkably large and elaborately carved depictions of the Buddha, Kneppers 2019 (MA thesis, full database in progress) – this is not listed in reference. Please could you provide details?. For a more detailed discussion of these hypotheses see Van Aerde 2019: 460-463. Cf. Von Hinüber 1989a; 1989b: 41–72, 73–106; Fussman 1993: 1–60 (This is not listed in reference. Please could you provide details?); Fussman 1994: 57–72. Most interpretative approaches have been mainly ethnoarchaeological, focused on cultural and/or ethnic iconographical categories among the variety of carving types (Jettmar 1989: XXII). 58 59 120 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains bodhisattvas, and jātaka scenes that are not found at any other locations, in terms of size and detail/ quality, and most likely reflecting a diverse chronology as well, as indicated by the selected iconographical analyses above. We also find patron dedications, such as the inscription and depicted figure of the devotee Siṇhoṭa at Chilas, carved right beside a large and elaborate carving of the Buddha Vipaśya.60 Such examples certainly argue for a prominent Buddhist presence or function at these locations, although more extensive excavation and survey would be necessary before sucha hypothesis can be further substantiated. But the currently known carvings themselves can shed more light, as well. We see a very large quantity spike of stūpa carvings at especially Thalpan (937). Moreover, the subsequent three largest stūpa recordings are at Chilas (199), Shing Nala (168) and Shatial (156) – thus encompassing all four locations where anthropomorphic Buddhist carvings have been recorded. Moreover, only at these four locations (Chilas, Thalpan, Shing Nala and Shatial) do we encounter notably large stūpas, as well as a much greater number of medium-sized stūpas; all other locations feature small and medium stūpas only.61 However, in respect to chronology, to the degree that this can be determined based on recognizable style and the stūpas’ architectural features, 80% of all documented stūpas coincide with early stūpa designs known from around the first century AD, usually linked to Gandhāran architectural types.62 However, at both Thalpan and Chilas, a higher percentage of what may be recognized as sixth century AD stūpa architecture is found than at other locations, which, in turn, would coincide with the Gupta-style Buddhist imagery discovered at the same locations (such as, most noticeably, the Buddha and bodhisattva group 194:151 and 194:152 at Thalpan, discussed above).63 On the basis of these data, we can identify, 1) a noticeable cluster at both Thalpan and Chilas for anthropomorphic Buddhist carvings of diverse style and chronology, and 2) at especially Thalpan a very substantial cluster of stūpa carvings, including rare large-sized carvings, and mainly of relatively early architectural style. These statistics are relevant not only in terms of petroglyph quantity, but also in relation to their varying sizes, the diversity of iconography, style and manufacture technique, and subsequent indications of chronology. The dataset of zoomorphic carvings adds yet another pattern to these observations. When we look closely at the many different Mammalia taxa and, especially, their relation to depicted human scenes and context throughout the dataset, distinct clusters of mounted animals, pack animals, and animals that directly accompany human figures become evident, the most prominent examples being, in plain terms, horses, camels, yaks, donkeys, and dogs.64 Remarkably, these specific clusters appear most notably (and most numerously) at Thalpan, and subsequently at Chilas, Shing Nala, Shatial and, finally, Hodar. At these stations specifically, the presence of significant numbers of mounted and/or domesticated zoomorphic carvings may indeed indicate a more prominent presence of such domesticated animals which may have inspired their remaining depictions. Moreover, it is noteworthy that at the other locations, which are situated in between the above-mentioned five, carvings of wild animals, such as eagles, leopards and many different kinds of bovid species (Bovidae) are far more numerous while domesticated animals are rare.65 When these data are joined together to reflect their physical locations per satellite, in basic summary, the cluster patterns become clear (Figure 17). 64:18, Bandini-König 2003: tafel 1. Khan 2018: fig. 4.8 and table 4.4. At Chilas we find 7 large stūpas, and at Shing Nala 31. 62 Khan 2018: fig. 4.9 and table 4.5. Cf. Fussman & Bandini-Kӧnig 1997: 40; Dani 1989: 91; Arif 2001: 35. 63 Khan 2018, table 4.6. See the discussion of Gupta influences from carvings 194:151 and 194:152 at Thalpan at the above section. 64 These statistics were initiated by Mike Kneppers as part of Van Aerde’s research project in 2018-2019 and the finalisation of the full data analysis is currently ongoing: specific zoological taxa determination is forthcoming. So far, only one article has been published on the animal species encountered among the Karakorum carvings, but this does not offer any statistical overview or a scientifically correct taxa determination: König 2004: 73-172. 65 In many cases mounted or domesticated animals are rare at these locations or even entirely absent (full analysis forthcoming, Van Aerde & Kneppers 2019 – this is not listed in reference, please could you provide details?). 60 61 121 The Global ConneCTions of Gandhāran arT Figure 17. Summary overview of wild/domestic animal carvings and the presence of Buddhist carvings. (Image: Van Aerde 2019, GoogleEarth). In this image, the four locations of Shatial, Thalpan, Chilas, and Shing Nala clearly stand out. Moreover, the distances between these four seem relatively evenly distributed, which might indicate specific travel distances that were most suitable/practical between station posts or nodal points along these routes, for example, where travellers could stay overnight, meet other caravans or pilgrims on the road, after specific periods of travel. A less speculative observation from these data is the apparent relation between the routes, which are indicated by the carving locations, and their vicinity to the Indus and Gilgit rivers. All so-far recorded Buddhist carvings, anthropomorphic and stūpa-related, are situated along riverbeds, and only sporadic zoomorphic carvings have been found further up the mountain slopes.66 The vicinity of water and the lower, more accessible terrain near these rivers are crucial variables for practical travel through a mountain range such as the Karakorum; if indeed transit stations emerged along these specific routes it is probable that they would develop along the sequence indicated by the currently documented carvings at these locations. The additional evidence of domesticated animals specifically found at these stations likewise supports this type of functionality. At such places, we can speculate, travellers could gather and rest while on their journeys, and thus they were also the most likely locations for people to practice the rituals of their religion. If, for example, merchants who were Buddhist followers travelled these routes from as early as the first century AD, and they left their initial marks in the form of relatively simple stūpa and Buddhist carvings, perhaps the initial spread of Buddhist imagery and ideas in these regions was coincidental with the spread of the routes themselves and the trade conducted through them. Over time, then, some of these locations (and Thalpan most particularly) apparently developed beyond their practical function; as evident from the appearance of 66 Kneppers 2019 (MA thesis), full data analysis forthcoming. 122 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains more elaborate Buddhist carvings that can be dated in terms of iconography to as late as the sixth century AD, perhaps these initial transit posts gradually developed into Buddhist mountain shrines or monasteries and in that capacity became destinations for pilgrims specifically. The evidence of the rock carvings in these locations, which is currently the only archaeological material available, seems to support this hypothesis. This is also in line with many theories concerning the spread of Buddhism from Gandhāra and the textual evidence of Chinese pilgrims travelling to these mountains.67 But to expand this investigation, excavation and survey would be necessary to further document these ancient routes on a larger scale, as well as study the potential materials and carvings left behind in more detail. Another angle of investigation would turn to the differing pattern evident from the Hodar location: in regard to its large quantity of zoomorphic carvings and clusters of domesticated animals, as well as its general distance between the other four locations, it seems at first similar to Thalpan, Chilas, Shing Nala and Shatial. However, there is no evidence of anthropomorphic Buddhist carvings here and a relatively low number of 122 stūpas. This might indicate that Hodar was a station post like the others, but with a somewhat different practical function: a gathering place specifically intended for pack animals and caravans, we could speculate, in the vein of road stables. The distribution chart for Hodar (Figure 18) provides more detail in terms of the petroglyphic spread across the landscape, and is noteworthy for its lack of Buddhist carvings. This, too, would call for further investigations: if similar clusters exist throughout the mountain range to the north, we might continue to recognise distinctions between such travel stations based on the patterns evident from the types and styles of carvings and their statistical distributions – and this, in turn, would allow us to better identify potential locations for excavation and closer investigation. Based on the data at present, however, the Thalpan location can already be identified as such a specific place of interest. The distribution chart in Figure 19 allows us a closer look at the range of carvings documented at Chilas Bridge in the Thalpan region, which indeed seems to reflect a remarkably high amount of human activity at this location from at least the first century AD onwards. Its geographical location, at a central crossing of the Indus river, may be an important environmental factor in both facilitating and necessitating access at this particular location. While not divergent in terms of the categories of carvings encountered here, the sheer quantity of petroglyphs at Thalpan, and the Chilas Bridge section, is so far unique in the region. As these clusters indicate, that location yields by far the most Buddhist petroglyphs, with numerous stūpas and anthropomorphic figures, as well as domestic zoomorphic carvings, as were found at Hodar.68 It is important to note here, however, that there is a clear variety among the carvings’ potential chronologies as well: whereas most of the stūpas and possibly also many of the zoomorphic carvings seem to date back to the first century AD, the most elaborate anthropomorphic Buddhist carvings show clear signs of a much later chronology, as explored in the previous section. So then, the bustle of human activity that is implied by the many diverse petroglyphs of Thalpan/Chilas Bridge, and especially its remarkable range of Buddhist carvings, should be read as a chronicle of time passing, as well. This, too, would support the hypothesis that the Thalpan area may have started out as one of the most active field stations for travellers/merchants in these mountains from the first century AD, and that (at least in part) because of these apparent practical merits of the location, it eventually developed into a Buddhist shrine and continued to thrive as such, as late as the sixth century AD. Conclusion To return to the quotation at the start of this chapter, studies of Gandharān art and the archaeology of the connected regions still face obstacles in attempts to theorize before sufficient data has been gathered. In many cases, this is due to circumstances of documentation or lack of access to the relevant materials. But it can also lead to substantial misunderstandings of the empirical evidence. Especially Cf. Arif 2001: 29-32; Hauptmann 2009: 8-9; Neelis 2014a: 3-17; 2014b: 45-64. In the present distribution chart (Figure 20) we include the specific data for (domestic) zoomorphic and anthropomorphic Buddhist carvings at Thalpan/Chilas Bridge only. Our database for the stūpa carvings is currently in progress and would, in fact, be too numerous to be included in full in a single chart. 67 68 123 The Global ConneCTions of Gandhāran arT Figure 18. Distribution chart of zoomorphic carvings at Hodar, by faunal category. (Van Aerde & Kneppers 2019.) Figure 19. Distribution chart of anthropomorphic Buddhist and zoomorphic carvings at Chilas Bridge. (Van Aerde & Gomez de Silva 2019.) when wider patterns and processes are concerned, when merely a handful of objects are presented to support a particular theory of widespread connectivity, in reality that theory may not in any other way be supported by the full or at least statistically viable datasets of the available evidence, and as such can 124 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains Figure 20. Alexander Mohns exploring undocumented Buddhist carvings near Gilgit (2019). become unintentionally misleading. To avoid these pitfalls, for our research presented here we chose to turn to all available documentation of the Karakorum carvings, of which we conducted, for this first time, comprehensive statistical analyses and subsequent interpretations of these data. Based on our findings we could, albeit tentatively, raise several new points of discussion concerning: 1) the spread of Buddhism and Buddhist imagery as evident from the Karakorum carvings; 2) reconstructions of the actual routes that transgressed these mountains during the first millennium AD; and 3) Gandhāra’s wider connection to its surrounding regions. Concerning the Buddhist carvings in particular, we encountered a great diversity of influences, or rather a mixture of them that is difficult to label, including contemporary iconographical elements from Gandhāra, Kashmir, and the Gupta Empire. This also relates to the fact that these carvings seem chronologically diverse, too: it appears that most stūpa carvings, especially those of the simpler carving type, were introduced as early as the first century AD and show particular parallels with Gandhāran stūpas, whereas the more elaborate and larger Buddhist carvings appear to have been added in later times, when, as we might hypothesize, certain station posts along the Karakorum routes had gained a distinct religious significance and functionality. As part of this process, then, these Buddhist petroglyphs developed a mixture of styles and influences, throughout various time periods, that seems unique to the Karakorum in particular. Concerning the routes themselves, at this point in our analysis certain recurring variables are already noteworthy. The vicinity of rivers seems a crucial component for the maintenance and success of such routes throughout the ancient world.69 The additional evidence of domestic zoomorphic carvings at 69 See also Van Aerde & Zampierin 2020, which discusses the variable of river vicinity concerning the data of ancient trade routes and connecting ports in both Egypt and South India. 125 The Global ConneCTions of Gandhāran arT specific station posts, where also the most Buddhist carvings were encountered, likewise supports the significance of these particular locations as gathering places and nodal points, and can help us continue to chart out the Karakorum routes beyond their current documentation. At present, the rock carvings are the only available archaeological material, but it seems highly likely that large quantities of goods, e.g. spices, medicinal herbs, rice, oils, ores, and pigments, would have travelled these same routes, from Gandhāra to Kashmir to Xinjiang, and eventually even all the way into the Tang Dynasty, transported in (ceramic or other) containers and reliant on a complex bureaucratic system of exchange throughout the entire region, necessitating contact and exchange between multiple different cultural spheres and political contexts. And alongside, Buddhism seemed to have travelled the same routes, perhaps unintentionally at first, in the company of merchants who were Buddhist devotees, and eventually came to leave a distinct mark on the Karakorum region. As a result, it seems, the Karakorum can be regarded as a passageway for early Buddhism in both a figurative and a literal sense. For that reason, also, it is crucial to ensure the continued documentation, study, and preservation of these petroglyphs, of which many remain unpublished and threatened today, so that our understanding of this region’s rich past will likewise be able to grow. References Arif M. 2001. Study of Petroglyphs of Buddhist Period along the Silk Road between Shatial and Khunjarab Pass Northern Areas, Pakistan. University of Karachi: Thesis Submitted for PhD Degree in General History. Bandini-König D. 1999. Die Felsbildstation Hodar mit Beiträgen von Gérard Fussman, Harald Hauptmann, Oskar von Hinüber, Thomas O. Höllmann, Ruth Schmelzer und Hellmut Völk (Herausgegeben von H. Hauptmann), Band 3. Mainz: MANP. Bandini-König D. 2003. Die Felsbildstation Thalpan I: Kataloge Chilas-Brücke und Thalpan (Steine 1–30). Bearbeitung der Inschriften durch Gerard Fussman et al. (Herausgegeben von H.Hauptmann), Band 6. Mainz: MANP. Bandini-König D. 2005. Die Felsbildstation Thalpan II: Katalog Thalpan (Steine 31–195). Bearbeitung der Inschriften durch Oskar von Hinüber und Thomas O. Höllmann (Herausgegeben von H. Hauptmann), Band 7. Mainz: MANP. Bandini-König D. 2007. Die Felsbildstation Thalpan III: Katalog Thalpan (Steine 196–450). Bearbeitung der Inschriften von Gérard Fussman, Oskar von Hinüber, Nicholas Sims-Williams (Herausgegeben von H. Hauptmann), Band 8. Mainz: MANP. Bandini-König D. 2009. Die Felsbildstation Thalpan IV: Katalog Thalpan (Steine 451–811). Bearbeitung der Inschriften durch Philip T. Denwood et al. (Herausgegeben von H. Hauptmann), Band 9. Mainz: MANP. Bandini-König D. 2011. Die Felsbildstation Thalpan V: Kataloge Ziyarat, Thakot, Khomar Das, Gichoi Das, Dardarbati Das. Bearbeitung der Inschriften von Gérard Fussman, Oskar von Hinüber, Nicholas Sims-Williams (Herausgegeben von H. Hauptmann), Band 10.Mainz:MANP. Bandini-König D. 2013. Die Felsbildstation Thalpan VI: Kataloge Ba Das, Ba Das Ost, Gali, Gukona, Mostar Nala, Ke Ges, Ame Ges und Drang Da. Bearbeitung der Inschriften durch Gérard Fussman, Oskar von Hinüber und Nicholas Sims-Williams (Herausgegeben von H. Hauptmann), Band 11. Mainz: MANP Bandini-König D. and von Hinüber O. 2001. Die Felsbildstationen Shing Nala und Gichi Nala, mit Beiträgen von W. Bernhard Dickoré und Günther A. Wagner (Herausgegeben von H. Hauptmann), Band 4 Mainz: MANP. Barrett D. 1962. Bronzes from Northwest India and Western Pakistan. Lalit Kala 11: 35-45. Beckwith C.I. 2015. Greek Buddha: Pyrrho’s Encounter with Early Buddhism in Central Asia. Princeton: Princeton University Press. Begley V., Francis P., Mahadevan I., Raman K.V., Sidebotham E., Slane K.W., and Will E.L. 1996. The Ancient Port of Arikamedu: New Excavations and Researches 1989-1992, Volume 1. Pondichéry: Ecole Francaise d’Extreme-Orient. Begley V., Francis P., Mahadevan I., Raman K.V., Sidebotham E., Slane K.W., and Will E. L. 2004. The Ancient Port of Arikamedu: New Excavations and Researches 1989-1992, Volume 2. Paris: Ecole Française d’Extreme-Orient. 126 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains Bennmann M. 2001. Die Felsbildstation Badam Das, mit Beiträgen von oskar von Hinüber, Jason Neelis und Nicholas Sims-Williams (Herausgegeben von H. Hauptmann), Band 5. Mainz: MANP. Bennmann M. and König D. 1994 Die Felsbildstation Oshibat, mit Beiträgen von Gerard Fussman, Oskar von Hinüber und Nicolas Sims-Williams (Herausgegeben von H. Hauptmann), Band 1. Mainz: MANP. Boardman J. 1994. The Diffusion of Classical Art in Antiquity. London: Thames and Hudson. Boardman J. 2015. The Greeks in Asia. London: Thames and Hudson. Carter M. 1993. Petroglyphs at Chilas II: Evidence for a pre-iconic phase of Buddhist art in Gandhara. Pages 349-366 in A.J. Gail and G. J. R. Mevissen (eds), South Asian Archaeology 1991: Proceedings of the 11th International Conference of the Association of South Asian Archaeologists in Western Europe, Berlin 1-5 July 1991. Stuttgart: Franz Steiner. Cunningham A. 1970. Coins of Alexander’s Successors in the East: The Greeks and the Indo- Scythians. Delhi: Indological Book House. Dani A. H. 1983. Chilas, the City of Nanga Parvat (Dyamar). Islamabad: Published by Author. Dani A. H. 1989. History of Northern Areas of Pakistan (Upto 2000 AD). Lahore: Sang-e-Meel Publications. (revised 2001; repr.2007). Dani A. H. 1995. Human Records on Karakorum Highway. Lahore: Sang-e-Meel Publications. De Geest P. 2012. The Inscriptions from Hoq: Texts and Contexts. Pages 232-253 in I. Strauch (ed), Foreign Sailors on Socotra: The Inscriptions and Drawings from the Cave Hoq. Vergleichende Studien zu Antike und Orient 3. Bremen: Hempen Verlag. Filigenzi A. 2015. Art and Landscape: Buddhist Rock Sculptures of Late Antique Swat/Uḍḍiyāna. Vienna: Verlag der Österreichischen Akademie der Wissenschaften Foucher A. 1905. L'Art Gréco-Bouddhique du Gandhâra. Paris: École française d’Extrême-Orient. Fussman G. 1994. Chilas-Thalpan et l’art du Tibet. Antiquities of Northern Pakistan, Reports and Studies 3: 57–72. Fussman G. and Jettmar K. 1994. Antiquities of Northern Pakistan: Reports and Studies 3. Mainz: Verlag Philip von Zabern. Fussman G. and Bandini-König D. 1997. Die Felsbildstation Shatial, mit Beiträgen von Oskarvon Hinüber, Thomas O. Höllmann, Karl Jettmar und Nicolas Sims-Williams (Herausgegeben von H. Hauptmann), Band 2. Mainz: MANP Ganguly D.K. 1987. The Imperial Guptas and their Times. New Delhi: Abhinav. Guruge Ananda W.P. 1991. The Buddha’s Encounters with Mara, the Tempter: Their Representation in Literature and Art. Indologica Taurinensia 17–18: 183–208. Hansen V. 2016. The Silk Road: A New History with Documents. Oxford University Press. Harle J.C. 1994. The Art and Architecture of the Indian Subcontinent. Yale University Press. Hauptmann H. 2008. Rock Art in the Upper Indus Region. Pages 352-357 in C. Luczanits (ed), Gandhara – The Buddhist Heritage of Pakistan: Legends, Monasteries, and Paradise. Mainz: Phillip v. Zabern. Hauptmann H. 2009. Felsbilder und Inschriften am Karakorum Highway. Heidelberg: Heidelberg Academy. Hiebert F. and Cambon P. (eds) 2008. Afghanistan: Hidden Treasures from the National Museum, Kabul. Washington D.C.: National Geographic. Iida S. 2016. Descent of Maitreya Buddha and his Enlightenment. Taishō 14: 454. Jettmar K. 1985. Zwischen Gandhara und den Seidenstrassen, Felsbilder am Karakorum Highway: Entdeckungen Deutsch-Pakistanischer Expeditionen 1979–1984. Mainz: Verlag Philip von Zabern. Jettmar K. 1989. Antiquities of Northern Pakistan, Reports and Studies 1. Mainz: Verlag Philip von Zabern. Jettmar K. 1993. Antiquities of Northern Pakistan, Reports and Studies 2. Mainz: Verlag Philip von Zabern. Kaul A. 2005. Buddhism in Kashmir. Indologica Taurinensia 31: 159-171. Khan A. G. 2018. An Analysis of Buddhist Rock Carving in Diamer-Basha Reservoir Area, Gilgit-Baltistan. Unpublished MPhil Thesis, Department of Archaeology, Hazara University. Kim I. 1997. The Future Buddha Maitreya. Delhi: D.K. Printworld Klimburg-Salter D. 1989. The Kingdom of Bamiyan: Buddhist Art and Culture in the Hindu Kush. Naples/Rome: Istituto Universitario Orientale. Kotera T., Maeda K., and Miyaji A. 1971. Bamiyan: Report of Survey in 1969. Nagoya: Nagoya University Press. 127 The Global ConneCTions of Gandhāran arT König D. 2004. Zu den Tierdarstellungen auf den Felsen am Oberen Indus. Pages 73-172 in G. Fussman, K. Jettmar and D. König (eds), Antiquities of Northern Pakistan Reports and Studies Vol. 3. Mainz: Verlag Philipp von Zabern. Krishnan Y. 2009. The Buddha Image: Its Origin and Development. Delhi: Munshiram Manoharlal Publishers. Kurita I. 2003. Gandharan Art: The World of the Buddha, Volumes 1-3. Tokyo. Liu X. 2010. The Silk Road in World History. Oxford; New York: Oxford University Press. Marshall J.H. 1918. A Guide to Taxila. Calcutta: Superintendent government printing, India. Marshall J.H. 1951: Taxila: An Illustrated Account of Archaeological Excavations Carried out at Taxila under the Orders of the Government of India between the Years 1913 and 1934, 3 Vols. Delhi; first published: Cambridge 1951. Mehta R.N. 1966. Excavation at Devnimori: A Report of the Excavation Conducted from 1960 to 1963 (No. 8). Department of Archaeology and Ancient History, Faculty of Arts, MS University of Baroda. Nagar S. 1993. Jatakas in Indian Art. Delhi: Parimal Publications. Nanavati S. J.M., Mehta R.N. and Chowdhary S.N. 1971. Somnath 1956, Being a Report of Excavations. Department of Archaeology and Ancient History, Faculty of Arts, MS University of Baroda. Neelis J. 2011. Early Buddhist Transmission and Trade Networks: Mobility and Exchange within and beyond the Northwestern Borderlands of South Asia. (Dynamics in the History of Religion). Leiden: Brill. Neelis J. 2014a. Networks for Long-Distance Transmission of Buddhism in South Asian Transit Zones. Pages 3-17 in T. Sen (ed), Buddhism across Asia: Networks of Material, Intellectual and Cultural Exchange. Singapore: Institute for Southeast Asian Studies. Neelis J. 2014b. Localizing the Buddha’s Presence at Wayside Shrines in Northern Pakistan. Pages 45-64 in P. Wick and V. Rabens (eds), Religions and Trade: Religious Formation, Transformation and Cross-Cultural Exchange between East and West. Leiden: Brill. Pal P. 1975. Bronzes of Kashmir. Graz: Akademische Druck- u. Verlagsanstalt Rani Tiwari U. 1998. Sculptures of Mathura and Sarnath: A Comparative Study (up to Gupta Period). Delhi: Sundeep Prakashan. Rhie M. 1999. Early Buddhist Art of China and Central Asia, Volume 3: The Western Chi’in in Kansu in the Sixteen Kingdoms Period and Inter-relationships with the Buddhist Art of Gandhara. Leiden: Brill. Rienjang W. and Stewart P (eds) 2018. Problems of Chronology in Gandhāran Art: Proceedings of the First International Workshop of the Gandhāra Connections Project, University of Oxford, 23rd-24th March, 2017. Oxford: Archaeopress. Rienjang W. and Stewart P. (eds) 2019. The Geography of Gandhāran Art: Proceedings of the Second International Workshop of the Gandhāra Connections Project, University of Oxford, 22nd -23rd March, 2018. Oxford: Archaeopress. Rüpke J. 2016. On Roman Religion: Lived Religion and the Individual in Ancient Rome. NewYork: Cornell University Press. Santoro A. 1979. Vajrapani nell’arte del Gandhara: ricerche iconografiche ed interpretative. Rivista degli Studi Orientali 53: 293-341. Schrader F. 2011. Pakistan Water and Power Development Authority, Cultural Hertitage Management Plan. Lahore: National development consultants. Shaw S. 2006: The Jatakas: Birth Stories of the Bodhisattva. Delhi: Penguin Global. Spink W.M. and Yaguchi N. 2014. Ajanta History and Development. Volume 6: Defining Features. Leiden: Brill. Thewalt V. 2008. Stūpas und verwandte Bauwerke in Felsbildern der Stationen Chilās I, Chilās II und Oshibat am Indus (Nordpakistan). Heidelberg: Heidelberg Academy. Van Aerde M. E. J. J. 2018. Revisiting Taxila: A New Approach to the Greco-Buddhist Archaeological Record. Ancient West & East 17: 203-229. Van Aerde M. E. J. J. 2019. Routes Beyond Gandhara – Buddhist Rock Carvings in the Context of the Early Silk Roads. Pages 455-480 in L. Yang (ed), Socio-Environmental Dynamics along the Historic Silk Road, Past Global Changes. Cham, Switzerland: Springer Open Press/Kiel University Press. 128 Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains Van Aerde M. E. J. J. and Zampierin D. 2020. A Lot of Pepper and a Little Garum: An Archaeological Comparison of the Roman Presence at Berenike and Arikamedu. Journal of Ancient West & East (AWE) 19: 145-166. Von Hinüber O. 1989a. Brahmi Inscriptions on the History and Culture of the Upper Indus Valley. Pages 4172 in K. Jettmar (ed), Antiquities of Northern Pakistan, Reports and Studies. Mainz: Verlag Philipp von Zabern. Von Hinüber O. 1989b. Buddhistische Inschriften aus dem Tal des Oberen Indus. Pages 73- 106 in K. Jettmar (ed), Antiquities of Northern Pakistan, Reports and Studies. Mainz : Verlag Philipp von Zabern. Yusuf S. 2011. Threatened Rock Carvings of Pakistan. Dawn, May 18 (digital issue: <https://www.dawn. com/news/629659/basha-dam-threatens-thousands-of-ancient-rock- carvings>, last consulted 29th April 2020). Zin M. 2009. Vajrapani in the Narrative Reliefs. Pages 73-88 in C. Frölich (ed), Migration, Trade and Peoples, Part II: Gandharan Art. London: The British Association for South Asian Studies. Zwalf W. 1996. A Catalogue of the Gandhara Sculpture in the British Museum. Volume 2, Plates. London: British Museum Press. Appendix Table of recognizably Buddhist images in Karakorum rock carvings that feature full or partial elements of anthropomorphic depiction. A Group Scene Contains Page Number 64:C 64: 13,14,15,16,18,19 G46+47, 1,2,4,32 64:E **1 64:20, 65:1 G46, 2 4 Photo Number I-a,b III-a,b,c IV-a,b.c V-a IX-a IX-b X-a,b X-c XI-a XII-b,c X-c XI-a,b XII-a,b,c,d XIII-a XIII-b **2 84:1,2 4,34 XV-a,b **3 8:1 2 I-c 19:A 19:1,2 G82, 49 XVI-b 30:C 30:31,32 G84, 50 30:D 30:38,39 G84, 53 XVII-c XVIII-a XXII-a XXII- b XXI-a,b XXV-b,c 6:A 30:A 30:B 63:A 63:B 64:D 6:1,2,3 30:1,2,21 30:6,10,11,12,13, 14,15,16,17,18,19 63:1,2 63: 4,6,8,10 G37, 1,14 G41, 1,21 G41,3,5,7,8,9,11 G44,1,30 G44,3,11,30 64:16,18 G45, 1,2 30:Q 30:140,141,142,143, G86,51 144 30:R 30:145,146,147,148, G86,51 149,150 XVII-c XXV- b,c 129 Place Description Chilas Bridge Boddhisattva and 2 stupas Chilas Bridge 2 devotees near elaborate stupa Chilas Bridge Tigress Jataka Chilas Bridge Devotee near stupa Chilas Bridge 2 Bodhisattvas, 1 stupa, 1 flower pot Chilas Bridge Devotee making offering to Bodhisattva Chilas Bridge 3 Bodhisattvas, 1 Buddha, 1 stupa, 1 devotee (*Contains Scene 64D) Chilas Bridge Buddha on pedestal (face is damaged) Chilas Bridge Buddha in Dharmacakra mudra. single figure Chilas Bridge Celestial deity with stupa (no group number listed) Chilas Bridge Bodhisattva, possibly Manjusri or Maitreya Thalpan 2 figures, 1 silhouette of Buddha, 1 unidentified Thalpan 2 figures, possible part of Jataka Thalpan Thalpan Thalpan Sibi Jataka or associated w/ stupa 30:40 1 large Buddha with 4 smaller Buddhas (noteworthy for folded robes and halos) 4 Buddhas, similar to 30:Q, possibly part of the same group The Global ConneCTions of Gandhāran arT 30:X 116:A 30:192,193,194,195, G88, XVII-c 196,197,198 54,58,61,62,63,65 XXVII-a 30:18 49 XVII-c XVIII-a XX-a 30:22 50 XVII-c XX-b XXI-a,c 30:30 50 XVII-c XX-c 116:1,2,3,4,5 G83,1,9,22,49 XI-c,d Thalpan 122:A 122:1,3 G84, 2, 52 Thalpan 126:A 126: 1,2 174:A **1 **2 **3 Thalpan Rsipancaka Jataka Thalpan Celestial being with crown holding prayer beads. Possibly associated w/ stupa Buddha in dhyana mudra (noteworthy for fleshy lips) associated w/stupa 30:26 Avalokitesvara with lotus Thalpan Thalpan G85, 1,2 XIV-a,b,c XV- a ??* Thalpan 174,1,2 G88, 2,10 XXIII-a Thalpan 194:I 194:55,56,57, 58,59,60,61 G90, 3,64,65 XXVI-a,b Thalpan 194:K G90,3,7,16,27 Thalpan G91,4 G92,1,5,69,73 XXVI-c XXVII-a XXVIII-c XXXI-a Thalpan Thalpan 195:W **1 **2 194: 65,66,67, 68,69,70,71,72 194: 151,152 195:124,125,12 6,127,128,129, 130,131 195:428,429,430 132:39 172:1 G94, 5, 9, 1 2 XXXIII-b XVII-c XXII-c Thalpan Thalpan Thalpan **3 **4 **5 176:1 194:103 195:170 3 3 5 XXIII-b XXVIII-a XXXI-b Thalpan Thalpan Thalpan **6 135:1 9 ?? Thalpan **1 Tafel D 1,3,4,16,20,23 V-a,b Shatial 31:A 34:125,126,127, 128,129,130-133, 134,135,146,170, 171 31:114,115 G38,3,22 IV-b Shatial **1 38:13 1 II-b,c Shing Nala **2 47:3 2 ?? Shing Nala **2 47:3 2 ?? Shing Nala **3 **4 47:5 47:6 3 3 IV-a IV-b Shing Nala Shing Nala **5 **1 **2 48:4 215:12 215:17 4 6 6 ?? Shing Nala Shatial Shatial 194:S 195:I 130 Buddha under Bo tree with ghostly devotee. Associated w. stupa 116:1 Elaborate Stupa with miniature Buddha associated Outline of halo and one image of Buddha with undefined body and halo One seated Buddha with arms beneath robes, and one undefined face Buddha on pedestal, associated with 4 stupas, upper part of face not preserved First sermon at Sarnath scene Mahayana, Buddha with celestial being Stupa with 4 associated beings in a tree, and one person (devotee) sweeping/praying Temptation of Mara scene Outline of seated Buddha Seated Buddha, noteworthy for folded robes Seated Buddha, unfinished Buddha in gift-giving mudra Outline of seated Buddha, no distinct features Adorned figure, unidentified, possibly Bodhisattva Sibi Jataka Small figure (devotee) making offering to stupa Seated Buddha in dharmacakra mudra, noteworthy for fleshy lips Seated Buddha, rippling robes, dhyana mudra Seated Buddha, rippling robes, dhyana mudra Seated Buddha, hands beneath robes Small seated Buddha, undefined features, hands beneath robes Halo, no Buddha Unidentified face Unidentified face Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains B Number Chilas Bridge 6:1 6:2 Orientation -------------N N Technique ------------ Patination ------------Middle Middle Size ----------29x70cm 40x61cm Type Inscription -----------------------Stupa Bodhisattva 6:6 6:3 30:1 30:2 N SW SW - Middle Middle Middle 28x83cm 96x213cm 40x67cm Stupa Stupa Man 30:21 SW Middle 19x44cm 30:6 SW Middle 32x83cm 30:10 SW Middle 29x14cm 30:11 SW Middle 32x13cm 30:12 SW Middle 30x29cm 30:13 SW Middle 32x104cm 30:14 SW Middle 21x62cm 30:15 30:16 30:17 30:18 30:19 63:1 63:2 63:4 SW SW SW SW SW E E S Middle Middle Middle Middle Middle Middle Middle Middle 11x33cm 6x19cm 7x22cm 11x21cm 16x39cm 47x55cm 53x135cm 33x112cm 63:6 63:8 63:10 64:16 S S S E Middle Middle Middle Middle 53x133cm 29x82cm 13x24cm 35x101cm 64:18 E Middle 29x61cm 64:13 E Middle 61x81cm 64:14 E Middle 30x80cm 64:15 E Middle 64x130cm 64:19 64:20 E - Middle - 27x104cm - 65:1 N Middle 49x62cm 84:1 N - - - 131 Interpretation/Remarks ---------------------------Manjusri according to inscription 30:3,4,5,22,28 30:3,4,5,22,28 Practitioner praying to Stupa Man 30:3,4,5,22,28 Practitioner praying to stupa Tree spirit 30:8,9,29 Some kind of female tree spirit Man 30:8,9,29 Man/possibly a celestial being Man 30:8,9,29 Man/possibly a celestial being Man 30:8,9,29 Man/possibly a celestial being Mountain 30:8,9,29 Mountain side, strangely side depicted w/ zigzagged lines Bodhisattva 30:8,9,29 Buddha as Mahasattva bodhisattva Tiger 30:8,9,29 Starving tiger cub Tiger 30:8,9,29 Starving tiger cub Tiger 30:8,9,29 Starving tiger cub Tiger 30:8,9,29 Starving tiger cub Tiger 30:8,9,29 Starving Tigress Man 63:3 Stupa 63:3 Bodhisattva 63:5 Avalokiteshvara according to inscription Bodhisattva Bodhisattva Maitreya Stupa Flowerpot Offering pot next to stupa. Buddha 64:17* Buddha Vipasya according to inscription Man 64:17* Man practicing devotion towards the Bodhisattva. With 64:16 is part of larger scene 64:C Bodhisattva 64:12 Avalokiteshvara according to the inscription Buddha Buddha in Dhyana mudra on a pedestal Stupa Associated with the large group Bodhisattva Maitreya Buddha Face broken off in dhyana mudra Buddha No group, dharmacakra mudra Stupa The Global ConneCTions of Gandhāran arT 84:2 N - - Bodhisattva 8:1 NE Middle 43x53cm Bodhisattva Thalpan I 19:1 ------------S -----------Middle ----------19x26cm -----------Silhouette 19:2 S Middle 25x27cm Silhouette 30:31 (C) W Middle 58x74cm Bodhisattva 30:32 (C) W Middle 54x75cm Bodhisattva 30:33 30:38 (D) W Middle 51x30cm Bodhisattva 30:39 (D) 30:140 (Q) W SE Middle Middle 40x64cm 64x102cm Man Buddha 30:141 (Q) 30:142 (Q) 30:143 (Q) 30:144 (Q) 30:145 (R) 30:146 (R) 30:147(R) 30:148(R) 30:192(X) SE SE SE SE SW SW SW SW W Middle Middle Middle Middle Middle Middle Middle Middle Middle 13x24cm 14x26cm 15x30cm 14x30cm 11x20cm 18x29cm 17x30cm 17x28cm 50x80cm Buddha Buddha Buddha Buddha Buddha Buddha Buddha Buddha Man 30:193(X) 30:194(X) 30:195(X) 30:196(X) 30:197(X) W W W W W Middle Middle Middle Middle Middle 20x25cm 12x4cm 17x9cm 24x8cm 32x17cm Ram Pigeon Crow Snake Pig 30:198(X) W Middle 20x70cm Tree 30:18 W Middle 38x52cm Bodhisattva 30:17 30:22 W Middle 55x78cm Buddha 30:30 W Middle 30x55cm Bodhisattva 30:28 Thalpan II 116:1 116:3 116:4 ----------SW SW SW -----------50x75cm 38x34cm 46x87cm -----------Stupa Tree Buddha 116:5 122:1 122:3 SW SW SW 20x19cm 79x146cm 22x31cm Man Stupa Buddha ---------- ------------ -----------Middle Middle Middle Middle Middle Middle 132 ----------- 30:21 ----------- Richly adorned Bodhisattva, manjusri or maitreya. Possibly Maitreya or Manjusri ----------------------------Silhouette of Buddha/deity in asana Robust silhouette, not entirely sure what it is. Left figure of S30:c. Adorned with necklace talking with figure on the right. Manjusri according to inscription King Sibi, part of the king sibi Jataka Man holding scales Central Buddha surrounded by 8 small buddhas. Rippling robes and fleshy lips Small Buddha Small Buddha Small Buddha Small Buddha Small Buddha Small Buddha Small Buddha Small Buddha Part of the Rsipancaka Jataka. This is the hermit Rsipancaka Ram Rsipancaka pigeon Rsipancaka Crow Rsipancaka snake Rsipancaka pig? (Not normal in the jataka) Tree under which the hermit sits. Maitreya according to inscription Buddha with folded robes and fleshy lips. Avalokitesvara (according to inscription) ----------------- Buddha under a tree in dhyana mudra, folded robes, associated w/ stupa Small Buddha associated w/ elaborate stupa Van aerde, mohns, and khan: buddha on The roCks: Gandhāran ConneCTions ThrouGh The karakorum mounTains 126:1 126:2 SE SE Middle Middle 21x41cm 39x79cm Silhouette Buddha 174:1 174:2 194:56 194:57 194:58 SE SE S SW S Middle Middle Middle Middle Middle 18x16cm 73x103cm 17x21cm 14x29cm 32x32cm Face Buddha Stupa Stupa Buddha 194:59 194:61 194: 65 S S NW Middle Middle Middle 28x38cm 22x42cm 35x43cm Stupa Stupa Buddha 194:66 194:67 194:68 NW NW NW Middle Middle Middle 8x11cm 10x12cm 10x10cm Man Man Deer 194:69 NW Middle 10x22cm Wheel 194:70 NW Middle 10x12cm Deer 194:71 194:72 194:151 NW NW E Middle Middle Middle 8x10cm 10x12cm 42x57cm Man Man Buddha 194:152 195:124 195:125 195:127 195:128 195:129 195:130 195:131 195:428 195:429 E NE NE NE NE NE NE NE N N Middle Middle Middle Middle Middle Middle Middle Middle Strong Strong 7x11cm 13x12cm 13x47cm 7x14cm 7x14cm 4x7cm 14x40cm 6x9cm 11x30cm 23x33cm Bodhisattva Man Stupa Man Man Man Throne Man Spirit Buddha 195:430 132:39 N W Strong Middle 15x30cm 19x32cm Spirit Silhouette 172:1 SE Middle 53x57cm Buddha 176:1 N Middle 48x41cm Buddha 194:103 W Middle 40x40cm Buddha 195:170 N Middle 17x23cm Silhouette 135:1 S Middle 38x55cm Celestial 133 Silhouette of a halo Buddha, upper body only, few details preserved. Face, possibly of Buddha Buddha in dhyanamudra Associated w/58 Associated w/58 Buddha on a pedestal in dharmacakra mudra, upper half of face is missing. Associated w/58 Associated w/58 Buddha w/ dharamacakra mudra. Scene of first sermon. 1 of 5 first disciples 1 of 5 first disciples Deer re-presenting deer at Sarnath scene. Wheel of suffering, meant to represent enlightenment Deer re-presenting deer at Sarnath scene. 1 of 5 first disciples 1 of 5 first disciples Buddha, Mahayana style. Varada mudra. Vajrapani (most probably) Daughter of Mara Buddha in Bhumisparsa mudra, scene of the temptation of Mara. Daughter of Mara Silhouette of Buddha in dhyana mudra. Buddha in dhyana mudra, folded robes. Seated Buddha, body is unfinished, fleshy lips. Buddha in Varada mudra. robes in clinging style. Silhouette of Buddha/ bodhisattva in dhyana mudra. Celestial being, possibly a prince or a Bodhisattva. Unclear. The Global ConneCTions of Gandhāran arT Shatial 34:125 34:126 34:127 34:128 -----------SW SW SW W ------------Picked Picked Picked Picked ------------Middle Middle Middle Middle -----------72x73cm 174x162cm 33x68cm 55x45cm 34:129 34:130 34:133 SW SW SW Picked Picked Picked Middle Middle Middle 40x60cm Man 50x40cm Man 410x207cm Stupa 34:134 SW Picked Middle 65x60cm Man 34:135 SW Picked Middle 46x36cm Man 34:146 34:171 SW SW Picked Picked Middle Middle 30x60cm Man 304x260cm Tree 31:114 NW 31:115 NW Shing Nala 38:13 -----------ESE Middle/ Strong Picked Middle/ Strong ------------ ---------Middle 47:3 S Middle 65x93cm Buddha 47:5 S Middle 50x63cm Buddha 47:6 S Middle 27x51cm Buddha 48:4 SSW Middle 89x57cm Silhouette Picked ----------------------Bodhisattva Building Flower Pot Man 64x32cm Stupa 38x13cm Man ------------------------- ----------135x210cm Buddha 134 -----------------Jataka scene of king Sibi Top of a temple/stupa Man figure holding a type of sickle with a ball in the middle. Kneeling w/ flowers Man with scales Very elaborate, with many decorations, associated with devotees. w/ sickle and ball and a type of torch w/ flames. Bent over in devotion with flowers. Kneeling w/ flowers Decorated tree, possibly bo tree w/ pedestal for Bodhisattva Devotee worshipping stupa ----------------Buddha in folded robes and fleshy lips Buddha in folded robes and fleshy lips. Buddha w/ halo and folded robes, hands under robes. Small Buddha, no defined features. Silhouette of halo (135-155): DOI: 10.32028/9781789696950-7 Buddhist temples in Tukhāristān and their relationships with Gandhāran traditions Shumpei Iwai Introduction Buddhism was introduced into the Tukhāristān region1 from Gandhāra around the first century AD. Thereafter, many Buddhist temples were constructed in the region and were prosperous until around the eighth century. Although it has actually been recognized that Tukhāristān Buddhism was formed under the influence of Gandhāran Buddhism, we can also find several original elements in architectural plans and Buddhist art of Tukhāristān. In addition, many studies have proposed that the temples in Tukhāristān tentatively declined in the latter half of the fourth century and became active again at the later fifth century or sixth century at the latest. In this paper, I shall firstly survey architectural plans and excavated finds from these ancient Buddhist sites in Tukhāristān, mainly belonging to the first to fourth centuries, in order to consider the early relationship with Gandhāran Buddhism. Then, I shall try to examine the hypothesis of the decline of Buddhism in Tukhāristān and to elucidate the reasons for the decline, if indeed this occurred. Buddhist sites in northern Tukhāristān This paper will concentrate on the sites dated to the first to fourth centuries to compare them with Gandhāran traditions (Figure 1). Since it is unfortunately very difficult to determine the detailed chronology of each Buddhist site, it is necessary to put these sites into this broad time span. I should like to concentrate on the sites where relatively scientific excavations have been carried out.2 Kara Tepa Kara Tepa is a very famous temple complex existing from the first century AD at Termez in Uzbekistan (Ставиский 1964; 1969; 1972; 1975; 1982; 1996). We can date the life of the sites by investigating excavated pottery and coins. There are two types of temple plan at the site. One is a plan consisting of caves and open-air buildings, which researchers call a ‘complex’. These complexes are concentrated especially at the southern part of the site. Another is an open-air temple with a large stūpa with a rectangular base and a rectangular monastery with a courtyard, which is very similar to Buddhist temples in Gandhāra (Figure 2). Firstly, we will examine the complex type. One of the characteristics of these caves is that they usually have a rectangular room surrounded by corridors perhaps intended for circumambulation (Figure 3). At the same time, we can find an open-air building with the same plan around the caves. In fact, there are many buildings with this plan in Tukhāristān and the Xinjiang Uyghur region in China, but it is relatively rare in Gandhāra, as we shall see later. Tukhāristān is an old name for the area including the north-east of Afghanistan and the south of Uzbekistan and Tajikistan. Although this area was called ‘Bactria’, now we know that it had already been called ‘Tukhāristān’ in the Kaniṣka period at the latest (Sims-Williams 2015). We therefore use ‘Tukhāristān’ in this article. 2 I have not dealt with the following sites: Zurmala Stūpa, a small shrine and a stūpa of Zar Tepa, and a newly discovered Buddhist temple at Chinghis Tepa. It is necessary to gather information about these sites and include them in future for more detailed analysis. 1 135 The Global ConneCTions of Gandhāran arT Figure 1. Map of Gandhāra and Tukhāristān. (Map: author.) In the later fourth century, a broad area of the southern part including the complexes was used as burial sites, suggesting that the temples had gone into decline. We can secure the period by some vessels from the burials which have small stamps and polish-line decorations over red slip (Figure 4). Coexistence of both stamps and polish-lines appears frequently around the fourth century and a type of carinated shallow bowl (Figures 4.11, 4.12) appears from the late fourth century onward (Сычева 1975). These facts indicate that the pottery assemblage from the graves of Kara Tepa belongs to around the late fourth century. We can confirm these trends on pottery with the excavations of Durman Tepe and Chaqalaq Tepe near Kunduz, and also with an excavation of Dal’verzin Tepa Citadel (Kuwayama 1975; 136 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Figure 2. Topographical map of Kara Tepa. (Map: after Ikeda 2018: figure on p. 13.) 137 The Global ConneCTions of Gandhāran arT Figure 3. Complex A and B at Kara Tepa. (Plan: after Ставиский 1975: pl. 2.) Iwai 2003). Furthermore, according to reports of Kara Tepa by V. Staviskii and his colleagues, the latest coins excavated with these pottery forms in the burials were the copper coins of the type of Warahran Kushan-shah. Moreover, G. Fussman indicates that many of the potsherds with inscriptions, most of which were written in Kharoṣṭhī, could be dated as late as the fourth century by the style of calligraphy (Fussman 2011). These data show that Kara Tepa temporally declined as a group of Buddhist sites at the end of the fourth century. On the northern side of the site, the Uzbek-Japanese joint team excavated a new open-air temple with a Gandhāra-style plan, as already mentioned (Figure 5). A large main stūpa is surrounded by small shrines, which is a typical plan of Buddhist temples in Gandhāra (Kato & Pidaev 2002; Fussman 2011; Пидаев 2016; Ikegami 2017; 2018; Iwamoto 2019). Excavators found a small, round-based, core-stūpa on the inside of the main stūpa. This fact indicates that the north temple was constructed in the relatively early period of Kara Tepa, the same as the complexes in the south part. However, some rooms of the rectangular monastery have squinches at the upper corners (Fussman 2011), suggesting that some parts of this north temple were constructed after the fifth or sixth century when Buddhism in Tukhāristān became active again. This continuity (or revival) of the site as Buddhist temples after the fifth century is corroborated by the fact that some inscriptions on potsherds can be dated to the fifth to seventh centuries by the style of calligraphy (Fussman 2011, vol. 1: 104-105). 138 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Figure 4. Pottery assemblage from a grave in Complex A at Kara Tepa. (Tracing after Ставиский 1975: pl. 36-38.) We will now turn to the Buddhist art of Kara Tepa. Buddhist sculptures from Tukhāristān are mostly made of limestone or clay, and sculptures from Kara Tepa are not exceptions. But when we pay attention to their details, such as wrinkles on the robes, and individual motifs on the reliefs, we see that they are generally based on Gandhāran sculptures. For example, the expression of the relief of the Buddha’s lifestory, ‘Maya’s dream’ (Figure 6), entirely depends on the Gandhāran tradition. On the other hand, features such as turned up, almond-shaped eyes and generally awkward, frontal expressions are often referred to as Graeco-Iranian or Oxus style, which actually appears to be different from Gandhāran style proper.3 As regards the so-called Oxus style (Figure 7), we can find one of the direct ancestors in the clay sculptures of the Khalchayan site (Figure 8) dated to around the first century BC. (Пугаченкова 1966).4 From these examples, it is possible to say that the Gandhāran tradition and the local cultures co-existed in Kara Tepa. These expressions sometimes look very similar to the sculptures from the Swat region. This point may be interesting, although not enough preparatory research has been carried out to analyse it here. 4 The characteristic style of trousers is also very important. We can find many donors in such style with Central Asian costumes in Buddhist sculptures, as we will mention later. 3 139 The Global ConneCTions of Gandhāran arT Figure 5. Plan of the north temple of Kara Tepa. (Plan: author, after Fussman 2011: pl. 18.) 140 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Next, we will examine the mural paintings from Kara Tepa. Recently, an Uzbek team found precious examples of Buddhist murals at a small shrine located on the west side of the large stūpa of the north temple (Пидаев 2016) (Figure 9). We can see a characteristic technique to emphasize light and shadow, especially around the figures’ eyes. It is difficult to compare them to the Gandhāran tradition because there are not so many earlier examples of mural paintings in Gandhāran art. On the other hand, these murals are apparently similar to the famous murals from Mīrān in China. Some art-historians have pointed out that the expression of the eyes and the use of vivid red colour in these paintings resemble each other (Yasuda 2018). In this way, the murals from Kara Tepa will be important Figure 6. Relief depicting ‘Maya’s dream’ from Kara Tepa. (Photo: after Kato and Pidaev 2002: fig. 6. Courtesy of Toho Shoten.) Figure 7. Deva from Kara Tepa. (Photo: after Kato and Pidaev 2002: fig. 11. Courtesy of Toho Shoten.) Figure 8. Donor wearing trousers from Khalchayan. Tashkent, Fine Arts Research Institute. (Photo: author.) 141 The Global ConneCTions of Gandhāran arT Figure 9. Mural paintings from the north temple of Kara Tepa. (After Пидаев 2016. Courtesy of Ш. Пидаев.) material for investigating the relationship among Gandhāra, Tukhāristān, and the Xinjiang Uyghur region. However, it is difficult to assert the date of the murals because the relationship between the large stūpa and the small shrine with the murals is not so clear. Since the room does not have a gateway on the stūpa side, it is possible that the room was associated with undiscovered structures located on the more western side. Excavators actually found some traces of structures on the west side of the room, and we should judge the relationship carefully. Fayaz Tepa Fayaz Tepa is a Buddhist temple located only 1 km from Kara Tepa. The Temple has a plan in Gandhāran style which consists of a main stūpa and a rectangular building which might have served as a monastery, small shrines, and an assemblage hall, although we cannot know the actual function of each building (Figure 10). The site was recently restored and reconstructed by the UNESCO Mission. Pottery inscriptions from the site were reported by Fussman, and he concluded that all the potsherds with inscriptions are datable to the first to fourth century from the style of calligraphy, except for just one piece dated to the 142 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Figure 11. Seated Buddha from Fayaz Tepa. Tashkent, State Museum of History of Uzbekistan. (Photo: author.) early fifth century (Fussman 2011). The greater part of them are written in Kharoṣṭhī characters, which Figure 10. Plan of Fayaz Tepa. means that the Buddhism practised in Fayaz Tepa was (Plan: author, after Fussman 2011: pl. 21.) strongly influenced by Gandhāran Buddhism, the same as Kara Tepa. In this way, we can assume that Fayaz Tepa also declined as a Buddhist temple around the end of the fourth century. The fact that the site was reused for burials after the fifth century (Альбаум 1974; Мкртычев 2013) would corroborate this hypothesis.5 We can find such Gandhāran influence in the sculptures from the site, especially from a room located at the centre of the presumed monastery (indicated by an asterisk * in Figure 10). The famous seated Buddha statue from the room also exhibits strong influence from Gandhāran art in the shape of the niche, the handling of the robe, and the Buddha’s wavy hair (Figure 11). In contrast, the up-turned, almond-shaped eyes of two monks are very similar to those of sculptures from Kara Tepa. Mural paintings were also found in the same room. The style of some of the murals looks different from that of Kara Tepa. Fortunately, we have some new examples of murals from Fayaz Tepa, which have never been reported so far (Figure 12). Although they have been stored in the Institute of Archaeology of Samarkand for a long time without any conservation treatment, a conservation project has been launched by an Uzbek-Japanese joint team led by E. Kageyama and preliminary results have been reported (Kageyama et. al 2017; 2018; 2019). The murals have many similarities with those excavated before in this room. At the same time, we can see some features in common with the newly discovered 5 Some scholars once advocated that the stūpa of Fayaz Tepa had been rebuilt in a cross-shape in the sixth to seventh century, and the site had become active again in this period. However, recent investigation revealed that there were no traces of renovation of the stūpa (Lo Muzio 2012; Мкртычев 2013). 143 The Global ConneCTions of Gandhāran arT Figure 12. Newly discovered mural paintings from Fayaz Tepa. Tashkent, Institute of Archaeology of Uzbekistan. (Photo: after Kageyama and Reutova 2018: figs. 5, 7. Courtesy of E. Kageyama.) murals from Kara Tepa, like the technique of emphasizing light and shadow around the eyes. Moreover, parts of a mural depicting a large stūpa and two small stūpas are important. They enable us to know the structure of chattrāvalis on the top of the stupa and ornaments which adorn the stūpa. Figure 13. Mural painting of two donors from Fayaz Tepa. Tashkent, State Museum of History of Uzbekistan. (Photo: author.) 144 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Figure 14. Mural painting of a donor with ram’s horns. Tashkent, State Museum of History of Uzbekistan. (Photo: author.) A famous mural of royal donors with Central Asian costume is also being restored by the team. While we can see its replica at the State Museum of History of Uzbekistan in Tashkent, nobody had seen the genuine mural. However, it became clear that the majority of the murals were being kept in the Institute of Archaeology in Samarkand. A close attention to details of the mural shows that there are some differences between the genuine mural and the replica and its drawing. We should use this new information for investigating the murals from the site. The date of the murals is very problematic. The Bactrian inscription ‘φαρο’ which was written at the upper part of a donor depicted at the inner wall of the entrance of the abovementioned room (Figure 13) might be evidence for dating the mural to around the second century, from its palaeographical style.6 On the other hand, Lo Muzio has supposed that the murals of the site could be dated to the fourth century because a technique used in their painting had been influenced by Sasanian art (Lo Muzio 2012). The variety of the styles in the murals may indicate that each was painted in a different period. In particular, a mural of donors excavated from the outside of the room, one of whom wears a head-cover surmounted by ram’s horns, looks to have been painted in a rather different style (Figure 14). Ram’s horns remind us of some headdresses of Kushano-Sasanian or Kidarite rulers (Lo Muzio 2012: 200-201) and might date the mural to the fourth century.7 In any case, there is no contradiction with the dating of the site to the late fourth century, whichever hypothesis we adopt for dating the paintings. Professor Yutaka Yoshida, pers. comm. Most researchers regard the donor with ram’s horns as a female (Grenet 2010; Lo Muzio 2012; Kageyama & Reutova 2017). However, if the ram’s horns actually relate to a ruler’s headdress, we cannot rule out the possibility that this donor was the ruler himself. In the Gandhāra Connections workshop of March 2019, Dr Joe Cribb kindly indicated, from the numismatic point of view, that the Kidarite ruler Peroz (‘Varahran’ in the inscription on his Kushano-Sasanian-style gold coin) is depicted with a headdress surmounted by ram’s horns. 6 7 145 The Global ConneCTions of Gandhāran arT Airtam Airtam was also located on the north bank of Amu-darya, but entirely disappeared after construction work. The site consists of a rectangular base of small stūpas8 and a shrine (Figure 15). It is rare to set a base and a large shrine side by side like this. In addition, G. Pugachenkova (1991/92: 27, 33) indicated that there were traces of a rectangular monastery at the north side and other Buddhist constructions, including a stūpa with a rectangular base, at the east side of the main structures. If the monastery was actually attached, Airtam also had a plan in Gandhāran style. Here, we should pay attention to the plan of the shrine. A rectangular room and a front chamber are surrounded by a corridor, the same as the caves and open-air buildings in Kara Tepa. It is certain that the main object of worship was not a Buddhist statue but a stūpa, because excavators found some pieces of chattrāvali in the main room of this shrine. Pugachenkova reported the existence of a building of the same plan from the lower stratigraphic layer of this site. It is therefore possible to assume that this type of plan is a kind of traditional one in this area. Actually, we can find many earlier and contemporary examples in Tukhāristān at Takht-i Sanguin (Litvinskii & Pichikian 1994), Surkh Kotal (Schlumberger et. al 1983), Diliberjin Tepe near Balkh (Кругликова 1986), and so on. For example, at Surkh Kotal, a main shrine and another two subsequently added shrines have the same plan, that is, a rectangular room with double enclosure wall. Many researchers point out that this characteristic plan of shrine originated from a Zoroastrian fire-temple in Iran, or more traditional religious buildings in the west Asian region (Yamamoto 1979; Pugachenkova 1991/92; Rhie 2002; Кызласов 2006; Ball 2016). If so, it might suggest that Buddhists in Tukhāristān had adopted a local religious tradition into Buddhist architecture. From the site, a famous limestone sculptural frieze of musicians was excavated (Figure 16). Although their turned-up eyes and somewhat awkward expression are very similar to the sculptures from Kara Tepa, we can find some elements shared with the Gandhāran tradition. For example, the rendering of a drum played by one musician is almost the same as that on Gandhāran reliefs. We know one more famous limestone sculpture which represents a pair of deities like Shiva and Pārvatī, with a Bactrian inscription carved on the base (Figure 17). It is said to have been excavated near the northern surrounding wall of the site. This motif is clearly Indian or Gandhāran style, rather than Graeco-Iranian or Oxus style. In the light of these friezes of musicians Figure 15. Plan of Airtam. (Image after Pugachenkova 1991/92.) 8 Pugachenkova indicated that this base was for a statue with Bactrian inscription, or for small stūpas (1991/92: 32). 146 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Figure 16. Sculptural frieze of musicians from Airtam. Saint Petersburg, State Hermitage Museum. (Photo: after Field & Prostov 1942: fig. 2, via <www.archive.org>.) Figure 17. Limestone sculpture of a pair of deities from Airtam. Tashkent, State Museum of History of Uzbekistan. (Photo: author.) and the Greek characters of the Bactrian inscription, the temple might have functioned during the Kushan period. Regarding the latter, Pugachenkova (1991/92) wrote that the inscription mentions the foundation of this temple at year 4 of ‘Hubishka’. Now, however, we cannot accept the reading easily because N. SimsWilliams has written that it was unduly speculative in its readings, and that the Airtam inscription was poorly preserved and did not seem to offer usable historical data (Sims-Williams 2012). Figure 18. Plan of Ushtur Mullo. (Plan: after Kato 1997: fig. 2-63.) Ushtur Mullo Ushtur Mullo is located on the East bank of Kafirnigan-darya, south Tajikistan. Unfortunately, we do not have enough information about the site. K. Kato introduces it in his comprehensive study of Central Asian Buddhist sites and writes that it has a square-based stūpa and a probable rectangular monastery with a courtyard (Kato 1997). And at the northern side of the monastery, a room with double enclosure wall was set up (Figure 18). 147 The Global ConneCTions of Gandhāran arT Dal’verzin Tepa Dal’verzin Tepa is an important ancient settlement located on the west bank of Surkhan-darya, about 100 km north of Termez (Пугаченкова и др. 1978). There were two Buddhist temples. One was in the suburbs of the shahristan (the first temple), and another was in the shahristan (the second temple). Both temples were active in the Kushan period, and as for the second temple, it had gone into decline at the end of the fourth century, which can be determined by the stratigraphic data of the pottery and Carbon-14 dating. Samples of the floor of the second Buddhist temple indicate AD 320-410 (76.4%) according to this dating (Soka University 2012). Regarding the first temple, it is highly possible that it stopped functioning earlier than the second temple, judging from the pottery. Although excavators found a square stupa-base in the first temple, the whole plan was not revealed (Пугаченкова и др. 1978: 90-97). Around the base many clay figures were found, the style of which is similar to those of Khalchayan sculptures. This fact supposes that these sculptures were influenced directly by the Graeco-Iranian tradition of this region. We need to note that some donors are wearing Central Asian costume like trousers, which is one of major traditional features of the region (Figure 19). On the other hand, some of the sculptures from the second Buddhist temple (Figure 20) have many Gandhāran elements, like the spiral shaped curls of hair and various ornaments of a bodhisattva, including a turban (Soka University 1996; 2012). We can assume from the difference that there were two routes of influence on the Buddhist art of Tukhāristān. One is the Graeco-Iranian (local) tradition of the region and another is the Gandhāran tradition. The difference of the styles between the first temple and the second temple possibly shows that the Gandhāran tradition of Buddhist art became gradually stronger in Tukhāristān. Figure 19: Donor wearing trousers from the first temple of Dal’verzin Tepa. Tashkent, Uzbekistan, Fine Arts Research Institute. (Photo: author) Buddhist sites in southern Tukhāristān Next, we will examine Buddhist sculptures from the south of Amu-darya (southern Tukhāristān). Several sculptures made of limestone have been found especially around Kunduz and Baghlan, although we only know for certain a few Buddhist temples there which belong to the Kushan period. This fact makes it difficult to know the date of the sculptures precisely from an archaeological point of view, while most art historians have thought that they were sculpted before the latter half of the fourth century by comparison with Gandhāran sculptures. Such limestone sculptures were excavated from Akhonzada Tepe (Tepe Ahingaran), Lili Tepe, Cham Qala and so on (Fischer 1958; Mizuno 1962). As we can see, a seated Maitreya with a small water bottle in his left hand totally depends on the Gandhāran tradition (Figure 21). Reliefs representing the Buddha’s life-story (the Great Departure, farewell of Kanthaka, and so on) 148 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Figure 20: Bodhisattva from the second temple of Dal’verzin Tepa. Fine Arts Research Institute, Uzbekistan. (Photo: author) Figure 21. Seated Bodhisattva from Akhonzada Tepe (Tepe Ahingaran). (Photo: after Mizuno 1962: fig. 125.) are also under the strong influence of a Gandhāran expressive scheme (Figure 22). On the other hand, we can sometimes find donors in Central Asian costume in these reliefs, who are certainly similar to those from northern Tukhāristān. A French team found a probable Buddhist site which is only 2 km east from Surkh Kotal in Baghlan (Schlumberger et. al 1983). There remained a rectangular platform surrounded by foundation stones of a pillar (Figure 23). On the platform, they found a wall-like structure constructed in sun-dried bricks and the foot of a huge clay figure. This situation clearly indicates that the structure was not a stūpa-base. In addition, the side face of the platform was covered elaborately by rectangular limestone ashlar, and from there, many limestone reliefs were excavated, which appear to have Gandhāran-style Buddhist Figure 22. Relief depicting the Buddha’s life-story from Akhonzada Tepe (Tepe Ahingaran). (Photo: after Mizuno 1962: fig. 123.) 149 The Global ConneCTions of Gandhāran arT motives (Figure 24). Excavators dated this site from the later second to the later third centuries. Recently, a very similar platform covered by limestone was also excavated at Tepe Zargaran in Balkh (Bernard 2006). This is possibly a base of a stupa, because excavators found a reliquary cell at almost the centre of the platform. If these sites are really Buddhist temples, we should study more about the relationship of Buddhist cultures between the north and the south of Amu-darya because the plan and construction materials are totally different. Characteristics of Buddhist sites in Tukhāristān To sum up the main points about the characteristics of Buddhist sites in Tukhāristān we have seen thus far, we can say the following. First of all, it is certain that some Buddhist temples exist in Tukhāristān from the Kushan period at the latest, and they were basically under the influence of Gandhāran Buddhism. We can find the evidence from various points of view. The rectangular stupa-base and a rectangular monastery with a courtyard, for example, must have been introduced from Gandhāra, because they are thought to have been developed there first (Kuwayama 1978; Behrendt 2003).9 Gandhāran style in Buddhist art and the use of Kharoṣṭhī characters in Buddhist temples in Tukhāristān also suggest a strong influence from Gandhāra. Figure 23. Base of a probable Buddhist site near Surkh Kotal. (Photo: after Schlumberger et. al 1983: pl. 47.128.) Figure 24. Relief of a turban, from a probable Buddhist site near Surkh Kotal. (Photo: after Schlumberger et. al 1983: pl. 66. 213.) On the other hand, there remain some local traditions in Buddhist art and architecture in Tukhāristān. For example, donors in Central Asian costume and the up-turned, almond-shaped eyes of statues and reliefs seem to be the local features inherited from the Hellenistic period in the region (Mkrtychev 2007). We have already mentioned that rectangular shrines with double enclosure walls might have a relationship with the local religious tradition associated with a kind of Zoroastrianism. Moreover, some of these traditions probably influenced Gandhāran temples and art. Regarding the donors in Central Asian costume, they frequently appear in Kāpiśī-style reliefs. Many researchers suppose that this must be a representation of Kushan people because they originated from Central Asia. If so, it is possible to Regarding the rectangular monastery with a courtyard, J. Shaw pointed out a possibility that earlier prototypical examples of this kind of monastery had existed in central India in around the second century BC (Shaw 2009: 123-124). 9 150 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Figure 25. Plan of a shrine at Gumbat. (Photo: after Faccenna and Spagnesi 2014: fig. 485, courtesy of ISMEO Italian Archaeological Mission in Pakistan.) assume that this costume might be one of the influences which Tukhāristān art gave to the Gandhāran tradition. The rectangular shrine with double enclosure wall would be one more example, although we cannot find many shrines of this type in Gandhāra. At Taxila, Building H of the Dharmarājikā complex is a shrine of this type (Marshall 1951: pl. 45; Behrendt 2003: fig. 1). We could date the building to around the first century AD because it was constructed in the so-called diaper masonry. Jaṇḍiāl C is an earlier example constructed in rubble masonry (Behrendt 2003: fig. 14), which has a main room and a front room with surrounding corridor. In Swat, we also have a few examples. Especially in Butkara I, the so-called Great Building has this kind of plan and appears to occupy an important place in the temple (Faccenna 1980, part 1: pl. 6; Behrendt 2003: fig. 97). This building was first constructed in almost the same period as Building H of Dharmarājikā, according to Behrendt (2003: 100). A small shrine at Gumbat also has exactly the same plan (Faccenna & Spagnesi 2014) (Figure 25). The construction of this shrine could be dated back to the later first century AD by carbon-14 dating (Olivieri & Filigenzi 2018: 80), and it is almost same time as Building H of Dharmarājikā and the Great Building of Butkara I. These examples might be a result of influence from the Tukhāristān region, or more simply, might be a reflection of the Iranian traditions which had been rooted before the Kushan period in both Tukhāristān and Greater Gandhāra. Although it is difficult to decide which one is correct, early examples no doubt exist in Tukhāristān and this tradition continues to exist until around the eighth century. Furthermore, the rectangular shrine with double enclosure wall also appears in the Xinjiang Uyghur region (Rhie 1999; 2002). We can find many examples of this type belonging to the third to fourth centuries onward. If we take them into consideration, it is clear that there was a strong relationship between Tukhāristān and the Xinjiang Uyghur region too. 151 The Global ConneCTions of Gandhāran arT The temporal decline of Buddhism in Tukhāristān As we have already mentioned, most Buddhist temples in Tukhāristān stopped their operation in the latter half of the fourth century. Afterwards, some of them became active again (like Kara Tepa) and new temples were built after the later fifth or sixth century (like Ajina Tepa) (Литвинский и Зеймаль 1971).10 This phenomenon used to be regarded as a result of a ‘social crisis’ in the whole Tukhāristān region caused mainly by the invasion of nomads from the north (la Vaissière 2004: 94-95). Researchers of the former Soviet Union had especially emphasized that the collapse of urban sites and emergence of small castles which they call ‘Zamok’ or ‘Gorodishiche’ concentrated on the late fourth to the early fifth centuries, and that this was the social crisis associated with a transition from Antiquity (Slavery) to Middle Ages (Feudalism) (Седов 1987: 114-116; Аннаев 1988: 8, 49-51; Брыкина (ред.) 1999: 5). However, with regard to the decline of Buddhism, we need to pursue more concrete reasons, because the decline of Buddhism in Tukhāristān looks like a temporary phenomenon.11 Of course, the situation was complicated and there was a combination of reasons. Sasanian intervention in Buddhism From recent studies, we know that the Sasanians temporarily established direct control over southern Tukhāristān in the later fourth century although it was not to be very long-term (Sims-Williams 2008: 92-93; Cribb 2010: 111-112). In this respect, we might be able to suppose simply that the Sasanian intervention in Buddhism caused the decline because Shapur II was said to be intolerant towards foreign religions. On the other hand, we should pay attention to the fact that there are no archaeological data suggesting any violent destruction of temples in Kara Tepa and Fayaz Tepa (Fussman 2011: vol. 1, 25; vol. 2, 261-262; Мкртычев 2013: 123). Thus, it would be more reasonable to suppose that the Sasanians did not have a positive relationship with Buddhism, rather than to regard them as its active devastator. The disappearance of a political group which connects the north and the south of the Hindukush This second reason might be more important for investigating the historical situation of the region. Now we know from a numismatic perspective that Sasanians soon lost hegemony in the south of the Hindukush (Alram & Pfisterer 2010; Cribb 2010). Although the relationship among Sasanians, KushanoSasanians, Kidarites, and Alkhan Huns is complicated and opinions on this subject vary among researchers,12they all agree to a certain extent that there was social turmoil caused by the invasion of nomads from the north in the late fourth century. Therefore, it is highly possible that such a confused situation led to a temporary interruption of main roads between Tukhāristān and Gandhāra. If Buddhism in Tukhāristān strongly relied on that of Gandhāra, as already mentioned, it must have been a great loss for Tukhāristān’s Buddhists that they could not accept Gandhāran monks and ideas. It meant that the influence of Gandhāran Buddhism was not reaching Tukhāristān. Afterwards, Buddhist temples in Tukhāristān became active again around the later fifth or sixth century. This may be closely related to the unification of both sides of the Hindukush by the Hephthalites and Alkhan Huns (Alram & Pfisterer 2010). On the other hand, a clay sealing with a Bactrian inscription from Swat implies another possibility (Rahman et. al 2006; Grenet 2010). The sealing bears the portrait of a king who is very similar to Kidarite kings on their coins, and the inscription identifies him as ‘Hun king’, 10 The Bamiyan site, located at a contact point between Tukhāristān and Gandhāra, is one of these newly constructed Buddhist temples. 11 The collapse of cities and the emergence of small castles does not seem to be concentrated in a short span of time when we carefully investigate pottery from the sites. F. Grenet pointed out that the social crisis was quickly followed by a new phase of urbanization, especially in Sogdiana (Grenet 2002: 203). 12 There is a big difference among historians particularly concerning the date of Kidarites in Tukhāristān (Enoki 1969-70; Grenet 2002; La Vaissière 2004) and numismatists (Alram & Pfisterer 2010; Cribb 1990; 2010). 152 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions ‘great Kushan king’, and ‘ruler of Samarkand’. If the king really belonged to the Kidarites, both sides of the Hindukush might have been reconnected more quickly by them. In this case, the temporal decline of Buddhism in Tukhāristān must be shorter than previously believed. Conclusions As we have seen, Tukhāristān Buddhism was basically under the influence of Gandhāran Buddhism. But we can also find local traditions in Buddhist art and architecture in Tukhāristān, some of which influenced Gandhāran temples and art. Although this evidence proves the tight connection between Tukhāristān and Gandhāra, Tukhāristān Buddhism may have been in trouble in respect to the acquisition of new information and trained monks when the connection was interrupted. It is very difficult to study the relationship between political groups and Buddhist sites from an archaeological point of view, and in the future we will need to gather further information to reach more specific conclusions. References Альбаум Л.И. 1974. Раскопки Буддийского Комплекса Фаяз-Тепе (По Материалам 1968-1972 гг.). Pages 53-58 in В.М. Массон (ed), Древняя Бактрия. Предварительные Сообщения об Археологических Работах на Юге Узбекистана. Leningrad: Nauka. Alram M. and Pfisterer M. 2010. Alkhan and Hephthalite Coinage. Pages 13-38 in M. Alram, D. KlimburgSalter, M. Inaba and M. Pfisterer (eds), Coins, Art and Chronology II. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Аннаев Т.Д. (1988) Раннесредневековые Поселения Северного Тохаристана. Tashkent: Fan. Ball W. 2016. Rome in the East: The Transformation of an Empire. 2nd edn. Abingdon: Routledge. Behrendt K. 2003. The Buddhist Architecture of Gandhāra. Leiden and Boston: Brill. Bernard P. 2006. Le stupa de Tepe Zargaran. Comptes rendus des séances de l’Académie des Inscriptions et Belles-Lettres 150: 1217-1241. Брыкина Г.А. (ed) 1999. Средняя Азия в Раннем Средневековье. Moscow: Nauka. Cribb J. 1990. Numismatic Evidence for Kushano-Sasanian Chronology. Studia Iranica 19: 151-193. Cribb J. 2010. The Kidarites: The Numismatic Evidence. Pages 91-146 in M. Alram, D. Klimburg-Salter, M. Inaba and M. Pfisterer (eds), Coins, Art and Chronology II. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Enoki K. 1969-70. On the Date of the Kidarites. Memoirs of the Research Department of the Toyo Bunko 27: 1-26; 28: 13-38. Faccenna D. 1980-81. Butkara I (Swāt, Pakistan) 1956-1962. 5 vols. Rome: IsMEO. Faccenna D. and Spagnesi P. 2014. Buddhist Architecture in the Swat Valley, Pakistan: Stupas, Viharas, a Dwelling Unit. Lahore: Sang-e-Meel Publications. Field H. and Prostov E. 1942. Excavations in Uzbekistan, 1937-1939. Ars Islamica 9: 143-150. Fischer K. 1958. Gandhāran Sculpture from Kunduz and Environs. Artibus Asiae 21-3/4 : 231-253. Fussman G. 2011. Monuments Bouddhiques de Termez I. Catalogue des inscriptions sur poteries. 2 vols. Paris: Edition-Diffusion de Boccard. Grenet F. 2002. Regional Interaction in Central Asia and Northwest India in the Kidarite and Hephthalite Periods. Pages 203-224 in N. Sims-Williams (ed), Indo-Iranian Languages and peoples. Oxford: Oxford University Press. Grenet F. 2010. A View from Samarkand: The Chionite and Kidarite Periods in the Archaeology of Sogdiana (Forth-Fifth centuries A.D.). Pages 267-281 in M. Alram, D. Klimburg-Salter, M. Inaba and M. Pfisterer (eds), Coins, Art and Chronology II. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Ikegami S. (ed). 2017. Kara Tepa Site: Preliminary Report for the Excavation in 2016 (Japanese with English summary). Tokyo: Rissho University. 153 The Global ConneCTions of Gandhāran arT Ikegami S. (ed). 2018. Kara Tepa Site: Preliminary Report for the Excavation in 2017 (Japanese with English summary). Tokyo: Rissho University. Iwai S. 2003. A New Chronology of Pottery in Post-Kushan Period Bactria. (Japanese with English summary.) Journal of West Asian Archaeology 4: 41-54. Iwamoto A. 2019. A Study on the Prosperity and Decline of Buddhist Sites in Northern Bactria: Kara Tepe and Zurmala. The Rissho International Journal of Academic Research in Culture and Society 2: 151-178. Kageyama E., Amridin B. and Abdullaev K. 2017. For Publishing the Murals from Fayaz Tepa, South Uzbekistan. (Japanese). Bulletin of National Research Institute for Cultural Properties, Nara 2017: 12-13. Kageyama E., Amridin B. and Abdullaev K. 2018. For Publishing the Murals from Fayaz Tepa, South Uzbekistan (2017). (Japanese.) Bulletin of National Research Institute for Cultural Properties, Nara 2018: 16-17. Kageyama E., Amridin B. and Abdullaev K. 2019. For Publishing the Murals from Fayaz Tepa, South Uzbekistan (2018). (Japanese.) Bulletin of National Research Institute for Cultural Properties, Nara 2019: 10-11. Kageyama E. and Reutova M. 2017. Mural Paintings from the Buddhist site Fayaz Tepa in South Uzbekistan. (Japanese.) Pages 5-10 in Research Institute of Cultural Properties of Teikyo University and National Academy of Sciences of the Kyrgyz Republic (eds), Proceedings of the Meeting for the Study of Silk Road in 2017. Tokyo: Teikyo University. Kato K. 1997. Studies on Buddhist Sites of Northern Central Asia. (Japanese.) Nara: The Nara International Foundation Commemorating the Silk Road Exposition. Kato K. and Pidaev Sh. (eds) 2002. New Finds of Uzbek Archaeology. (Japanese.) Osaka: Toho Shuppan. Кругликова И.Т. 1986. Дильберджин. Храм Диоскров. Moscow: Nauka. Kuwayama S. 1975. Kāpiśī Begrām III: Renewing its Dating. ORIENT 10: 57-78. Kuwayama S. 1978. Notes on the Introduction of Square Podium to the Taxilan Stupa. ORIENT 14: 23-34. Кызласов Л.Р. 2006. Городская Цивилизация Срединной и Северной Азии. Moscow: Vostochnaya Literatura. La Vaissière E. de 2004. Histoire des Marchands Sogdiens. Deuxième édition révisée et augmentée. Paris: Collège de France, Institut des hautes études chinoises. Литвинский Б. А. и Зеймаль Т. И. 1971. Аджина-тепе: Архитектура, Живопись, Скульптура. Moscow: Iskusstvo. Litvinskij B.A. and Pichikian I. R. 1994. The Hellenistic Architecture and Art of the Temple of the Oxus. Bulletin of the Asia Institute 8: 47-66. Lo Muzio C. 2012. Remarks on the Paintings from the Buddhist Monastery of Fayaz Tepe (Southern Uzbekistan). Bulletin of the Asia Institute, 22: 189-206. Marshall J. 1951. Taxila. 3 Vols. Cambridge: Cambridge University Press. Mizuno S. (ed.) 1962. Haibak and Kashmir-Smast. Kyoto: Kyoto University. Mkrtychev T. 2007. Buddhism and Features of the Buddhist Art of Bactria-Tukhāristān. Pages 475-485 in J. Cribb and G. Herrmann (eds), After Alexander: Central Asia before Islam. Oxford: Oxford University Press. Мкртычев Т. 2013. Буддийский монастрь Фаяз-тепе (Северная Бактрия) по исследованиям последних лет. Российская Археология 2013-2: 119–125. Olivieri L.M. and Filigenzi A. 2018. On Gandhāran Sculptural Production from Swat: Recent Archaeological and Chronological Data. Pages 71-92 in W. Rienjang and P. Stewart (eds), Problems of Chronology in Gandhāran Art. Oxford: Archaeopress. Пидаев Ш. 2016. Коратепадаги Будда Ибодатхонасининг Деворий Суратлари. Фан ва Турмуш 2016/3-4: 45-51. Пугаченкова Г.А. 1966. Халчаян. К Проблеме Художественной Культуры Северной Бактрии. Tashkent: Fan. Пугаченкова Г. А., Ртвеладзе, Е. В. и др. 1978. Дальверзинтепе. Tashkent: Fan. Pugachenkova G. A. 1991/92. The Buddhist Monuments of Airtam. Silk Road Art and Archaeology 2: 23-41. 154 shumpei iWai: buddhisT Temples in TukhārisTān and Their relaTionships WiTh Gandhāran TradiTions Rahman A., Grenet F. and Sims-Williams N. 2006. A Hunnish Kushan-shah. Journal of Inner Asian Art and Archaeology 1: 125-131. Rhie M. M. 1999. Early Buddhist Art of China and Central Asia, vol. 1. Leiden, Boston and Cologne: Brill. Rhie M. M. 2002. Early Buddhist Art of China and Central Asia, vol. 2. Leiden, Boston and Cologne: Brill. Schlumberger D., Le Berre M. and Fussman G. 1983. Surkh Kotal en Bactriane. 2 vols. (Mémoires de la Délégation Archéologique Française en Afghanistan, vol. 25.) Paris: Diffusion de Boccard. Седов А. В. 1987. Кобадиан на Пороге Раннего Средневековья. Moscow: Nauka. Shaw J. 2009. Stūpas, Monasteries and Relics in the Landscape: Typological, Spatial, and Temporal Patterns in the Sanchi Area. Pages 114-145 in A. Shimada, and J. Hawkes (eds), Buddhist Stūpas in South Asia: Recent Archaeological, Art-Historical, and Historical Perspectives. New Delhi: Oxford University Press. Sims-Williams N. 2008. The Sasanians in the East: A Bactrian Archive from Northern Afghanistan. Pages 88-102 in V. S. Curtis and S. Stewart (eds), The Sasanian Era. London: I.B. Tauris. Sims-Williams N. 2012. Bactrian Historical Inscriptions of the Kushan Period. Silk Road 10: 76-80. Sims-Williams N. 2015. A new Bactrian inscription from the time of Kanishka. Pages 256-264 in H. Falk (ed), Kushan Histories. Bremen: Hempen Verlag. Soka University Mission for Silk Road (ed) 1996. Dalvarzintepa DT25: Reports of 1989-1993 excavations. (Japanese with Uzbek summary.) Tokyo: Soka University. Soka University Research Center for Silk Road (ed) 2012. Dal’varzintepa Buddhist Temple. (Japanese with Uzbek summary.) Tokyo: Soka University. Ставиский Б.Я. (ed). 1964. Кара-Тепе -Буддийский Пещерный Монастырь в Старом Термезе. Moscow: Nauka. Ставиский Б.Я. (ed). 1969. Буддийские Пещеры Кара-тепе в Старом Термезе. Moscow: Nauka. Ставиский Б.Я. (ed). 1972. Буддийский Культовый Центр Кара-тепе в Старом Термезе. Moscow: Nauka. Ставиский Б.Я. (ed). 1975. Новые Находки на Кара-тепе в Старом Термезе. Moscow: Nauka. Ставиский Б.Я. (ed). 1982. Буддийские Памятники Кара-тепе в Старом Термезе. Moscow: Nauka. Ставиский Б.Я. (ed). 1996. Буддийские Комплексы Кара-Тепе в Старом Термезе. Moscow: Vostochnaya Literatura. Сычева Н.С. 1975. Керамика Кара-Тепе. Pages 88-148 in Б.Я. Ставиский (ed), Новые Находки на Кара-Тепе в Старом Термезе. Moscow: Nauka. Yamamoto Y. 1979. The Zoroastrian Temple Cult of Fire in Archaeology and Literature (I). ORIENT 15: 19-53, pls. 1-55. Yasuda H. 2018. Newly Found Murals from Kara Tepa: Comparison with Examples from Eastern Turkistan. Pages 44-51 in S. Ikegami (ed), Kara Tepa Site: Preliminary Report for the Excavation in 2017. (Japanese with English summary.) Tokyo: Rissho University. Zeymal T.I. 1999. On the Chronology of the Buddhist Site of Kara-Tepe. Pages 413-421 in M. Alram and D. Klimburg-Salter (eds), Coins, Art and Chronology. Essays on the pre-Islamic History of the Indo-Iranian Borderlands. Wien: Verlag der Österreichischen Akademie der Wissenschaften. After completing this article, I learned of the following report. This is a very important recent achievement and I should like to include the results in future work. Пидаев, Ш.Р., Баратова Л.С., Ким Донхун, Ли Чиён, Ульмасов А.Ф., Ли Усоп, Чон Юнхи, и Стоякин М.А. 2019. Археологические Работы Узбекистано-Корейской Экспедиции в Буддийском Культовом Центре Каратепа в Старом Термезе в 2011-2014 гг. Daejeon: National Research Institute of Cultural Heritage of Korea. 155 (156-204): DOI: 10.32028/9781789696950-8 More Gandhāra than Mathurā: substantial and persistent Gandhāran influences provincialized in the Buddhist material culture of Gujarat and beyond, c. AD 400-550 Ken Ishikawa Introduction This paper investigates a new distinctive form of ‘provincial’ Buddhist material culture that emerged at Devnīmorī (Mehta & Chaudhary 1966) in north Gujarat in c. AD 400 under the Kārdamaka line of the Western Kṣatrapas (c. AD 78-415), amid the contemporary dominions of the imperial Guptas (c. AD 320-550) and their allied Vākāṭakas (c. AD 250-500) in other parts of South Asia. The Buddhist monastic complex at Devnīmorī or ancient Paśāntika-paḷḷi (Sircar 1965: 337) is characterized by its monumental brick stūpa and distinctive terracotta ‘buddha’ images and ornaments adorning its exterior. The Buddhist material culture at Devnīmorī is often discussed under the heading of ‘Gupta’ art and architecture, which are well known to have come into existence out of the two preceding, and most influential, traditions of Buddhist material culture in the region of Gandhāra and at the city of Mathurā (in north India). In fact, Peshawar (ancient Puruṣapura) in Gandhāra and Mathurā were the former capitals of the Kushan empire (c. AD 30-330) in its heyday. In this paper, Mathurā is used more in the sense of a cultural capital. Although Buddhist art and architecture of Gandhāra and Mathurā under the imperial Kushans were geographically centred on Gandhāra and Mathurā, ‘Gupta’ material culture was more widespread or even pan-Indian in nature and, misleadingly, was not necessarily associated with the imperial Guptas. Ruling from the Magadha Kingdom in the lower Gangetic Valley, the imperial Guptas were the second pan-Indian state in history, long after the earliest unification of South Asia by the Mauryans (c. 322-185 BC). Despite this fact, almost no ‘Gupta’ material remains survive from the Gupta capital of Pāṭaliputra (modern Patna). In such circumstances, the Buddhist material culture at Devnīmorī, from north Gujarat, is even considered as one of the earliest and finest examples of ‘Gupta’ art and architecture, far beyond the Gupta heartland that stretched between eastern and central India. Hence, pan-Indian ‘Gupta’ material culture has often been discussed with reference to its problematic geo-political and socio-cultural relationship with its provinces (Harle 1974; Williams 1982). So panIndian ‘Gupta’ material culture merely stands as a generic model derived from its various distinctive regional manifestations. To disambiguate, the term ‘Gupta’ is used in this paper to qualify a spectrum of widespread/pan-Indian material culture during the Gupta and post-Gupta periods in South Asia with a loose spatio-temporal association with the Gupta empire, in a similar way to the use of the term ‘Gandhāra’ with its elastic geography and chronology. In addition, the term ‘province’ is also used loosely, not strictly as an imperial administrative unit, but rather as a remote region outside major cultural centres. In particular, this paper shall examine the earliest extant buddha images in Gujarat, produced at Devnīmorī, in terms of the extent of cultural transmissions from Gandhāra and Mathurā, where the earliest extant groups of Buddha images in the Indian subcontinent were created, but also of the innovative receptivity of this newly created regional material culture at Devnīmorī. Besides, I shall also reconsider the formation of homogenous yet heterogeneous Gupta material culture through imperial vs. interregional models of the emergence of a new provincial material culture as transculturation. 156 ken ishikaWa: more Gandhāra Than maThurā This theoretical approach may potentially have subtle implications for the ongoing debate over the origins of the first Buddha images in Gandhāra and Mathurā. Methodologically, while applying a conventional object-centred approach, with the main body of my materials being buddha images published elsewhere or documented by myself, I shall also contextualize objects with archaeology, epigraphy, numismatics, and Buddhology, wherever appropriate. Furthermore, I will trace the waves of Gandhāran influences observed at Devnīmorī, which within, or after, a century or so eventually reached Sārnāth and Ajaṇṭā and locally persisted at Śāmalājī in north Gujarat, Dhānk in Saurashtra in India, and Mīrpur Khās in Sindh in Pakistan. In this scholarly inquiry, I will also touch upon the archaeological visibility of Mahāyāna Buddhism in Gujarat with reference to problematic identifications of buddhas and bodhisattvas in Gandhāra. Although many of these buddhas represented in art are regarded as the historical Buddha Śākyamuni, there have also been spatiotemporally distant buddhas such as the Seven Buddhas of the Past plus Maitreya Buddha and other buddhas or tathāgatas such as Amitābha and Bhaiṣajyaguru, whose images are notoriously difficult to identify unless clearly labelled in their associated, well-preserved inscriptions. In other words, there is only one Buddha but are many buddhas. I do not employ the terms ‘buddha’ and ‘bodhisattva’ as proper nouns, even though I do not regard them as generic, simply to avoid confusion. The Buddha as a proper noun and ‘buddha’ or ‘awakened one’ as a noun with no capitalization of the initial ‘b’ are carefully distinguished in this paper. The Buddha is used when it is certain that he is the historical Buddha/Śākyamuni, including his universal manifestation in Gupta-Sārnāth, but the noun ‘buddha’ is otherwise preferred when the identification is unclear, for example, in late Gandhāra, where identifications of various past and Mahāyānist buddhas are problematic. Similarly, for some of the early Kapardin Buddha images from Kushan Mathurā or one related image from Bodh Gayā, which were inscribed as ‘bodhisattva’, I employ the designation of ‘Buddha/Bodhisattva’. Gandhāra or Mathurā: between the two different opinions There are two different opinions as to the sources of influences of Devnīmorī buddhas with reference to Gandhāra and Mathurā, both of which are to be challenged in this paper. Shah made the following remark: At Devni Mori particularly the Buddha figures, deriving some of their stylistic features directly from Gandhāra, already reflect all the principal characteristics that one normally associates with ‘Gupta’ Buddhas. And yet, they were created at least a century before the earliest dated Buddhas of Sārnāth (Shah 1972: 46). Schastok (1985: 35) then responded to Shah as follows: Gandhāran features appearing in the early 5th century at Devnī Morī are seen in this argument as reflecting direct contact with a pure Gandhāran style, but even a cursory look at Kuṣāṇa sculptures from both centres shows that a number of stylistic and iconographic features had already been exchanged during the Kuṣāṇa period. Thus some Gandhāran features were already part of Indian art at Mathurā before the 4th century and might be evidence of a link between North Gujarat and Mathurā rather than Gandhāra. I find both of the above arguments to be determinist and reductionist in limiting the sources of influences of Devnīmorī either to Gandhāra or Mathurā, a view that will be refuted. 157 The Global ConneCTions of Gandhāran arT Devnīmorī as a Buddhist archaeological site: a site profile of the monastic complex A Buddhist monastic complex at Devnīmorī in north Gujarat (the north-east of the state of Gujarat), now submerged by a dam (the Meshvo Dam) and thus inaccessible, is located in the valley of the Meshwo river in the alluvial plain of the central part of mainland Gujarat on the western foothills of the Aravalli range (Figures 1 & 2). North Gujarat was one of the main Buddhist regions in Gujarat, at least from the early Western Kṣatrapa period, because at least two other related major Buddhist sites are known in the region, namely Vadnagar and Taranga (Rawat 2011). The Buddhist site of Devnīmorī would have been a major, local or even regional ceremonial centre in the years in which they were active. Devnīmorī was excavated in the early 1960s by the M.S. Baroda University (Mehta & Chowdhary 1966) as an act of rescue archaeology for the construction of the Meshwo Dam across the Meshwo. While the now submerged monastic complex at Devnīmorī remains in situ underwater, major archaeological finds are mostly preserved at the M.S. Baroda University in Gujarat. Archaeological excavations at Devnīmorī (Mehta & Chowdhary 1966; Chowdhary 2010) revealed a Buddhist monastic complex dating to c. AD 400. These monumental remains at Devnīmorī are characterized by a considerable use of fired bricks and terracotta, which were produced from clay locally sourced from the river bed of the Meshwo running adjacent to Devnīmorī. This was a logical choice, according to Schastock, given the very small occurrence of stone, which was also locally available but not in proximity (Agrawala 1959: 63; Schastock 1985: 25-26). Figure 1. Map of Gujarat and Sindh showing the sites mentioned in the present paper (Map: ESRI and others; author.). 158 ken ishikaWa: more Gandhāra Than maThurā Figure 2. Map showing the environs of Devnimori (Image after Mehta & Chowdhary 1966; courtesy of M.S. Baroda University). Despite the fact that rock-cut caves constitute the main body of Buddhist monuments in Gujarat during the Western Kṣatrapa period (Nanavati & Dhaky 1969: 15), the regional tradition of Buddhist monuments in brick is still attested in Gujarat (Schastok 1985: 29). Such brick remains are found on the foothill of Mount Girnar in Saurashtra, peninsular Gujarat, and include the Rudrasena Vihāra, a brick courtyard monastery, possibly dating to the reign of king/mahākṣatrapa Rudrasena I (ruled AD 200-222) (Sompura 1969: 15-16, fig. 10), and the stūpa called Lakha Medi in the Boria hill, which could be as early as some of the earliest extant stūpas in India at Sāñcī, Andher and Sonar (Mitra 1971: 98; Le 1992: 99-100), or even earlier from the Mauryan period or Western Kṣatrapa period (Lahiri 2011: 124-126). The monastic configuration included a terraced stūpa, a courtyard monastery, a caitya hall and smaller funerary/votive stūpas (Figure 3). The monastic complex at Devnīmorī would have been much more extensive as one of the trenches revealed part of another vihāra (Vihāra II) approximately 150 metres 159 The Global ConneCTions of Gandhāran arT Figure 3. The configuration of the site of Devnimori monastic complex (Image after Mehta & Chowdhary 1966; courtesy of M.S. Baroda University). to the east of the mahāvihāra (Vihāra I) (Mehta & Chowdhary 1966: 65) and still remains largely unexcavated. The ritual focus of the monastic complex at Devnīmorī was the stūpa on a two-tier square platform, which is of the Gandhāran origin (Kato, Yatani & Masui 2017). Its exterior, though severely damaged, was adorned with buddha imagery and ornaments in terracotta, while its core was relatively undisturbed and contained a wide range of relic deposits including an inscribed relic casket, one of the two inscriptions of which refers to the stūpa as the mahāstūpa and the vihāra as the mahāvihāra (the terms which will be used for each in this paper) (Mehta & Chowdhary 1966: 32-66; 123-171). The mahāvihāra at Devnīmorī is a singled-storeyed courtyard monastery or vihāra with tiled roofs (Behrendt 2003: 170-171). Interestingly, the central cell at the back of the mahāvihāra is irregularly articulated, a feature described by the excavators as ‘a shrine room’ (Mehta & Chowdhary 1966: fig. 13). This peculiar trend at Devnīmorī appears to correspond to a similar pattern of rock-cut courtyard monasteries in the Western Deccan during the so-called ‘blank period’ of rock-cut monastic architecture between the third and late fourth century AD. As noted by Owen (2001), during this ‘blank period’, the innermost central 160 ken ishikaWa: more Gandhāra Than maThurā cells of rock-cut courtyard monasteries increased in their size, morphology, and ritual significance, anticipating the incorporation of the image shrine into the vihāra at Ajaṇṭā by the late fifth century AD. The site of Devnīmorī seems to show a fully ‘domesticated’ phase or state of Buddhism (see Strenski 1983; Fogelin 2015 for ‘domestication’), indicated by the existence of the courtyard monastery, which is considered to facilitate ‘law and order’ in analogy with a cloister in Christian monastic architecture (Schopen 1994: 547; Shaw 2007: 35; chs. 9 and 11; Shaw 2011: 115 for the issue of the relationship of courtyard monasteries with ‘domestication’). In this connection, the monastic community or saṅgha at Devnīmorī was possibly engaged in water management for irrigated agriculture, as evident from a series of reservoirs found in its vicinity (Mehta 1963; Sutcliffe, Shaw & Brown 2011: 784), suggesting ‘religious/ monastic governmentality’ (see Coningham et al. 2007; Shaw 2007; Gilliland et al. 2013; Chatterjee 2015; Shaw 2016 for ‘religious/monastic governmentality’). Some pilgrims, lay Buddhists and monks may also have been involved in long-distance trade since the symbiotic relation of the Buddhist saṅgha with trade is well known by this period in western India in relation to Indian Ocean trade (Ray 1989) as well as in Gandhāra (Neelis 2011). This may be as indicated by the discovery of the imported Roman bronze statue of Atlas found on the riverbed of the Mashwo (Indian Archaeology - A Review 1960-61: 58; Chawdhary 1964: fig. 75) as well as the presence of the Red Polished Ware used for the transportation of goods between inland and coastal sites, as well as abroad in the context of Indian Ocean trade (Pinto-Orton 1991; 2013). In fact, the mahāstūpa and the mahāvihāra, according to the Sanskrit inscription out of the two relic casket inscriptions, were constructed by two Mahāyāna monks (śākyabhikṣu), named Agnivarman and Sudarśana, who were superintendents of the construction (kārmāntika) (Sircar 1965: 337). This testifies to the diffusion of Mahāyāna Buddhism in Gujarat by the time of their construction. However, according to the Sanskrit text, the inscribed relic casket was fashioned by a mason named Varāha, a son of Sena, who appears to have been a lay Buddhist or Hindu, and the inscription also refers to the reign of a king (nṛpati) Rudrasena (Srinivasan 1968: 68). All these together suggest a complex social milieu and logistics behind the construction of the monastic complex. After the initial construction, most likely in c. AD 400 in the late Western Kṣatrapa period, the site clearly remained occupied at least until some time in the Maitraka period (c. AD 475-776), suggested by the finding of three undated Maitraka coins attributed to the late repairs of the mahāstūpa (Mehta 1965: 413), until the abandonment of the site at some unknown point. The mahāstūpa at Devnīmorī The mahāstūpa at Devnīmorī as a Gandhāran-type terraced stūpa The morphology of the mahāstūpa on a double square platform at Devnīmorī (Figure 4) can be traced back to similar cylindrical stūpas on square terraces that originally developed in Swāt, central Gandhāra, and Taxila in northern Pakistan during the first to sixth century AD (Chaudhary 1964: 109; Chaudhary 2010: 157-160; Faccenna & Spagnesi 2015; Kato, Yatani & Masui 2017) (Figure 5). Unlike the design of the main body of a stūpa in India proper, being merely a dome (aṇḍa) on a circular platform (medhi), Gandhāran stūpas are characterized by their three-tier structure of, from bottom to top, a square platform(s) (medhi), a cylindrical shaft, and a hemispherical dome (aṇḍa) (Karashima 2018: 474). South Asian stūpas are then typically topped with a superimposed structure that consists of a square pavilion (harmika), a pole (yaṣṭi) and an umbrella (chattra) (Karashima 2018: 474-475). The idea of the combination of a cylindrical stūpa with a square terrace in Gandhāra may have been conceptually inspired by the unusual circular brick temple in a rectangular enclosure at Bairāt in Rajasthan (Figure 1), which was dated by Brown to the 161 The Global ConneCTions of Gandhāran arT second century BC (Brown 1940: plate VI). Categorically the same Gandhāran stūpas are also regionally distributed in Sindh (Chaudhary 2010: 157-160), for instance, at the Kahujo-daro at Mīrpur Khās (Cousens 1914), Thūl Mīr Rukān (Cousens 1926: 7-11, pl. 10; Van Lohuizen-De Leeuw 1979: 156-8), Mohenjo-dāro (Marshall 1931) and Sudheranjo-dāro at Saidpur (Bhandarkar 1920). In north Gujarat, such stūpas can be found not only at Devnīmorī but also at Vadnagar (Rawat 2011) and more recently in Taranga hills and beyond but nowhere else in India proper. The double square-platformed stūpas at both Devnīmorī can be considered to be typologically comparable with those stūpas on similarly ‘setback’ double square platforms at Taxila (in the later development c. AD 300-400), to which the Devnīmorī counterpart is almost contemporary, rather than those on single or multiple square platforms (Kato, Yatani & Masui 2017: 2986). Such Gandhāran stūpas on double square platforms are known to correspond to those described in one version of the Chinese translation of the Mūlasarvastivāda Vinayakṣudrakavastu, Genben shuo yiqie youbu pinaiye zashi 根本説一切有部毘奈耶雜事 (T.1451:24.287a-292a; Odani 2003). The text describes a funerary/votive stūpa, both for deceased monks and for buddhas, and the description of the terraced stūpa therein resembles Gandhāran miniature and monumental stūpas (Odani 2003: 58-63). Similarly, the textual description of a great bejewelled stūpa that enshrines a body of a past buddha named Prabhūtaratna in the eleventh chapter Stūpasaṃdarśana (‘manifestation of stūpa’) of the Saddharmapuṇḍarīka resembles Gandhāran stūpas because of the presence of numerous arched niches over a cylindrical barrel and a high umbrella (Karashima 2018: 473-474). At Devnīmorī, while arched niches are Figure 4. The tentative reconstruction of the mahāstūpa at Devnimori at the Museum of the Department of Archaeology and Ancient History, M.S. Baroda University. (Photo: author, courtesy of M.S. Baroda University.) 162 Figure 5. The miniature terraced stūpa on a square platform from Swat. (Photo: courtesy of the Huntington Archive, Digital Database Collection, Scan Number 4871.) ken ishikaWa: more Gandhāra Than maThurā present, a high umbrella is not quite attested. However, some of Gandhāran examples show a high pole (yaṣṭi) with multiple umbrella-like discs (chattra) like a tower, which also match the textual description (Karashima 2018: 474): such example of a high tower-like stūpa can be seen at Cave 19 at Ajaṇṭā (Spink 2009: fig. 135) suggesting a Gandhāran influence. The terraced mahāstūpa at Devnīmorī is also regarded as a prototype of later Gupta terraced brick temples (Mukherjee 2008: 73) and in support of this argument, close parallels of certain motifs between the mahāstūpa at Devnīmorī and the brick temple at Bhitargaon from the fifth century AD have been drawn (Schastock 1985: 31, n. 45). Most significantly, the mahāstūpa at Devnīmorī displays terracotta buddha images and ornaments on its outer walls of the square platforms in a manner reminiscent of many terraced square-platform Gandhāran stūpas. Figure 6. Relic Casket II and its deposits from Devnimori. (Photo: author, courtesy of M.S. Baroda University.) The core of the mahāstūpa and relic deposits The core of the mahāstūpa (Figure 7) contained, from top to bottom, a buddha image, the aforementioned Casket II, a pot with eight Western Kṣatrapa coins inside, a collection of mostly broken brick ornaments Figure 7. The section of the trench into the core of the mahāstūpa at Devnimori (Image: after Mehta & Chowdhary 1966, courtesy of M.S. Baroda University). 163 The Global ConneCTions of Gandhāran arT used as the floor of the core, a group of eight more terracotta buddha images, and lastly another, but broken and ‘unfinished’, relic casket containing ash (Relic Casket I) in schist at the base (Mehta & Chowdhary 1966: 118-120). The inclusion of coins in relic deposits is rather characteristic of the relic cult in Gandhāra (Jongeward et al 2012). Other excavated artefacts at the monastic complex include various kinds of pottery shards, terracotta figurines, beads, stone, metal and glass objects and sixtynine coins though their archaeological contexts were not recorded systematically (Mehta & Chowdhary 1966: 69-118). Inscribed Casket II in schist, recovered from the upper part of the core of the mahāstūpa at Devnīmorī, is a short, cylindrical reliquary (Mehta & Chowdhary 1966: 118-120; pl. XXXI, fig. B) (Fig. 6), which is reminiscent of a category of similar Gandhāran caskets (Jongeward et al 2012: 268; though not strictly Gandhāran), of a kind also seen in the relic caskets from stūpa no. 2 at Satdhāra near Sāñcī, Madhya Pradesh, central India: Maisey 1847-1854). The casket contained miscellaneous objects (Figures 6 and 7). The term mahāstūpa denotes, according to Skilling (Skilling 2016: 23-4), a kind of stūpa, which is epigraphically stated to contain relic deposits. The Sanskrit inscription on Casket II dates the construction of the stūpa to the reign of a Rudrasena in the year 127 of the otherwise unknown Kathika era and thus Devnīmorī remains undatable with a precise absolute date. The dating of the mahāstūpa at Devnīmorī The dating of the mahāstūpa at Devnīmorī has been largely disputed. In terms of epigraphy, there has been no consensus regarding the controversial date of the mahāstūpa as suggested by the Sanskrit one of the two inscriptions (Sircar 1965; Mehta 1965; Srinivasan 1968) on the aforementioned reliquary (Casket II). This inscription records the date of the construction of the mahāstūpa by two monks in the year 127 of an unknown era of the Kathika kings but also mentions the name of the king Rudrasena, who was most likely one of the Western Kṣatrapas. The era was identified with the Śaka era by Sircar (1965), who also identified the Rudrasena as Rudrasena I (AD 200-222) giving the date of the construction of the mahāstūpa as early as AD 205-206, or alternatively with the Kalacuri era giving the date of AD 375 (Mirashi 1965). The palaeography (Srinivasan 1968), the use of classical Sanskrit (Mirashi 1965; Salomon 1998: 90), and the relatively early occurrence of the word śākyabhikṣu in both Indian and Chinese contexts in the Sanskrit inscription of Casket II and the philology of the other inscription on the same reliquary, in Buddhist Middle Indic resembling Pāli (von Hinüber 1985: 196-197), together suggest (Shizutani 1953; Schopen 1979; Cousins 2003: 232-239; Palumbo 2013: 3; Fukuyama 2014: 468-471) the reign of either Rudrasena III (c. AD 348-378?) or Rudrasena IV (c. AD 384-388), of the four Western Kṣatrapas with that name (Damsteegt 1978: 226; Jha & Rajgor 1992: 16); this is leaving aside the almost contemporary Rudrasena I and Rudrasena II of the Vākāṭakas. This dating range also agrees with the aforementioned chronologies of the architectural types of the ‘set-back’ double square-platformed stūpa in Gandhāra as well as of rock-cut courtyard monasteries in the Western Deccan during the third to fourth centuries AD. In respect to numismatic evidence, eight coins were found in the pot deposited in the core of the mahāstūpa and among them three belonged to the reigns of Rudrasena I, Viśvasena and Rudrasiṁha, which all together give a chronological range between AD 203 and 313 (Schastock 1985: 29), again suggesting the reinternment of the relic deposits. The numismatic evidence at Devnīmorī as a whole merely gives an impression of Western Kṣatrapa and subsequent Maitraka occupations (Mehta & Chowdhary 1966: 30, 106-116). Similarly, the pottery evidence at Devnīmorī such as Red Polished Ware, micaceous ware, and stamped-and-incised red ware merely indicate a single-phase occupation only datable broadly to the early centuries AD, in parallel with other sites in Gujarat and elsewhere in the Indian subcontinent (Mehta & Chowdhary 1966: 69-87; Shaw 2007: 107). 164 ken ishikaWa: more Gandhāra Than maThurā The peculiar archaeological context of the mahāstūpa at Devnīmorī further problematizes the uncertain date of its construction. As mentioned, the core of the mahāstūpa contained along with three reliquaries, nine terracotta buddha images, eight grouped at the bottom and one below the top as well as terracotta ornaments (Figure 7). These additional deposits in terracotta are identical with those that decorated the exterior of the mahāstūpa. It has been suggested by Gorakshkar that such a burial practice in respect to the eight buddha images can be compared with that of the bronze images of Seven Buddhas of the Past plus Maitreya Buddha of a later date, re-deposited into a pre-existing stūpa at Sopara (Gorakshkar 1991; Desai 2013). Thus the buried buddha images seem to have been appropriated as relics while other ornaments were used even as building materials for the floor of the core of the mahāstūpa. Although the excavators emphasized that the core of the mahāstūpa was undisturbed, scholarly consensus supports the reinternment of the relic deposits after some external damage to the mahāstūpa during later reconstructions and restorations, which was rather a common local practice, as reported from other stūpa sites in western India and eastern Pakistan (Van Lohuizen-De Leeuw 1979: 164-165; Williams 1982: 58-59; Schastock 1985: 29-30). Among numerous such examples, Van Lohuizen-De Leeuw listed stūpas at Sopāra in the Western Deccan, at Lakha Medi in Gujarat, Mainamati, Brahmānābād and Mīrpur Khās in Sindh, whose ostensibly reinterred relic deposits in the cores of their bodies included broken/repaired relic caskets, broken sculptures and brick ornaments (Van Lohuizen-De Leeuw 1979: 164-165) as also seen at Devnīmorī, mentioned above (Casket I). Devnīmorī buddha images General characteristics of Devnīmorī buddha images The Buddhist monastic complex of Devnīmorī is particularly significant for its characteristic terracotta buddha cult images (non-narrative and frontal imagery for worship, ritual, meditation and visualization) (Figures 10, 13 & 14) that once adorned the terraced brick mahāstūpa (Mehta & Chowdhary 1966; Chowdhary 2010), a manifestation of so-called Gupta material culture (Harle 1974; Williams 1982). Devnīmorī buddha images (Mehta & Chowdhary 1966; Chowdhary 2010) in terracotta relief, originally coloured with whitish cream, once decorated the double square-platform of the mahāstūpa alongside other additional terracotta ornaments. Uniform in size, approximately 67-68 cm in height, the buddha images are modelled in high relief with simple halos, backdrops, and single- or double-petalled lotus thrones. They were originally placed under respective caitya arches built in separate ornamental bricks. The common characteristics of Devnīmorī buddha images (Figures 10, 13 & 14) such as meditative downcast eyes, so typical of both Gandhāran and Gupta buddhas, the dhyānamudrā (the meditative handgesture) and the padmāsana (a cross-legged posture), give an exceptionally strong sense of meditative practice. Despite variations in hairstyles, upper garments, and lotus thrones, and the occasional absence of the ūrṇā, the overall configuration of the buddha images is highly standardized, clearly indicating a single intensive phase of production. Devnīmorī buddha images are examples of the finest Gupta terracotta imagery and testify to the fully-fledged Gupta style, based conservatively on Kushan and post-Kushan Mathurā buddhas and bodhisattvas and otherwise heavily influenced by late Gandhāran buddhas. The excavation report of Devnīmorī claims that twenty-six terracotta buddha images in fragments were recovered and that only twelve of them can be reconstructed fully (Mehta & Chowdhary 1966: pl. 37 A-D, pl. 38 A-D, pl. 39 A-D), though it also lists a thirteenth buddha image in full (Mehta & Chowdhary 1966: pl. 17 B) and a more recent, revised report lists yet another one (Chowdhary 2010: 80, fig. 39 B). Twenty heads were also recovered and twelve of them fit their corresponding torsos (Mehta & Chowdhary 1966: 141). It appears that at least thirty-two buddhas originally decorated the exterior of the mahāstūpa. Most 165 The Global ConneCTions of Gandhāran arT of the buddha images and heads are now preserved in the Museum of the Department of Archaeology and Ancient History, the M.S. Baroda University, but others are missing from this collection, including one example of a buddha head now in the collection of Los Angeles County Museum of Art (Pal 1986: 263). The stylistic dating of Devnīmorī buddha imagery As discussed earlier, the mahāstūpa is generally dated to c. AD 400, around the time of the conquest of Gujarat under the Western Kṣatrapas by the imperial Guptas and consequently, the style of fullyfledged Gupta buddha images at Devnīmorī was considered to be the result of the conquest, explained by the fact that the Guptas had acquired influence over the region (Williams 1982: 59; Schastock 1985: 30). However, whether Devnīmorī post-dates the conquest is debatable, and the date of the buddha images at Devnīmorī itself has long been disputed, owing partly to a poor understanding of the pre-existing material culture in Gujarat at a regional level (Williams 1982: 59; Schastock 1985: 30). Initially, a very early Gupta date was assigned to the Devnīmorī buddhas, largely on account of the late fourth-century date of the inscribed reliquary, as discussed above. Thereafter, the Devnīmorī buddhas started to be identified as independent progenitors of Gupta material culture (Shah 1972: 45-46). This theory was further challenged, but the ongoing role of traditional Mathurā as a source of influence has been overemphasized owing to the absence of earlier material culture in north Gujarat (Williams 1982: 59; Schastock 1985: 30), long before recent excavations at Vadnagar (Rawat 2011). Williams considered certain features of the terracotta buddhas and ornaments at Devnīmorī to be consistent with the parallel development in the other regions of Gupta India in the late fourth century AD, and thus suggested that the buddhas were somewhat later than the late fourth century AD (Williams 1982: 59-60). Certain indications of Gandhāran influence on Devnīmorī also meant that the site was given an early Gupta date (Shah 1972: 46: Schastock 1985: 30; Williams 1982: 59). However, Williams and Schastock acknowledged that overall Gandhāran influence was rather limited (Williams 1982: 59; Schastock 1985: 30). Both also argued that the emergence of the Devnīmorī buddhas in the fully-fledged Gupta style was a result of the conquest of Gujarat by Candragupta II c. AD 400 (Schastock 1985: 30; Williams 1982: 59), despite the fact that no Gupta coins had been found at Devnīmorī. However, more recent numismatic evidence makes the date of Candragupta II’s conquest as late as AD 407, and furthermore indicates that the Western Kṣatrapa rule persisted in north Gujarat under Rudrasiṃha III as late as AD 415 (Bhandare 2006). The chronology of buddha images in other parts of India may aid the relative dating of Devnīmorī buddha images within a typological sequence of epigraphically datable formative, fully-fledged, and mature, Gupta-style buddhas/tīrthaṅkaras at the major production centres in Mathurā, Vidiśā, central Magadha, Ajaṇṭā, and Sārnāth, including their hinterlands (Harle 1974; Miyaji 1980; Williams 1982; Huntington 1985). As a rule, earlier Gupta sculptural remains are fundamentally based on the influential Kushan/ post-Kushan Mathurā tradition, which was long-lasting and far-reaching (Rosenfield 1963: 24). However, all the Gupta Buddha images show one or more formative-Gupta characteristics: the ornamentation of the halo with floral and gem motifs, the garments with diaphanous drapery, hair curls, meditative eyes, elongated earlobes, the pronounced lower lip and/or three lines across his neck (Miyaji 1980: 16). Despite the paucity of datable buddhas/tīrthaṅkaras between the post-Kushan and fully-fledged/mature Gupta phases at Mathurā, some carvings could belong to this formative Gupta stage i.e. dating from prior to AD 400 (Miyaji 1980: 12-20; Williams 1982: 29; Koezuka 1984: 88-94). However, imperial Gupta material culture was certainly being formulated at Mathurā during the last quarter of the fourth century AD, considering the pillar fragment from a Śaivite shrine erected and inscribed in GE (Gupta era) 61 i.e. AD 380 or 381 under Candragupta II (Williams 1982: 29). 166 ken ishikaWa: more Gandhāra Than maThurā Apart from such a continuous cultural sequence at Mathurā itself between the post-Kushan and Gupta periods, pre-existing Kushan or post-Kushan heritage was certainly still influential in the Gupta material culture of Mathurā. Devnīmorī buddha images retain some features of the Kushan-Mathurā school, and yet are not as advanced as the Mankuwar Buddha image dated to GE 108/109 or 110 or AD 427/428 or 429, under Kumāragupta I’s reign (Williams 1982: 81, fig. 104), a group of Jain tīrthaṅkara images from Kankāli Tīlā, one of which is dated to GE 113 or AD 432/433 under Kumāragupta’s reign (Williams 1982: 68, figs. 60, 210), the Govindnagar Buddha image dated to GE 115 or AD 434/435 (Williams 1982: 68, n. 3; Schopen 1987: 267) and even the four buddha images placed at Stūpa I at Sāñcī by the mid-fifth century AD (Huntington 1985: 197-198). The formation of the early Gupta style in Vidiśā is more significant if we consider the transition from the three formative Gupta Jain tīrthaṅkara images in beige sandstone from Durjanapura (c. AD 376-80? under Rāmagupta; Williams 1982: 28-29, figs. 12, 13, 14) (Figure 8) to the fully-fledged Gupta Hindu rockcut cave-temple of Cave 6 at Udayagiri (82 GE or AD 401/2 under Candragupta II’s reign; Huntington 1985: 188-189). Both are epigraphically associated with the imperial Guptas themselves. Meanwhile, in central Magadha, one image of the seated Buddha in locally unavailable ‘dark reddish-brown stone’ (Huntington 1985: 14) or ‘red sandstone in imitation of the material commonly used at Mathura, though it is clearly not an import from Mathura’ (Asher 2008: 62), or otherwise ‘a yellowish buff’ stone (Williams 1982: 33), was found at the Buddhist centre of Bodh Gayā (Figure 9). It is inscribed as depicting a ‘bodhisatva’ and clearly shows a formative Gupta style. However, despite missing arms, this Buddha/Bodhisattva from Bodh Gayā is also very clearly based on the composition of the so-called Kapardin Mathurā Buddha/Bodhisattva images (Cifuentes 2013: 87-89), which are also occasionally inscribed as ‘bodhisatva’ (Rhi 1994)’: characteristically with the right hand raised, which would have shown the abhayamudrā; the left hand resting on the left knee and holding the hem of the drapery of the robe, which covers the left shoulder; the nimbus covered with a lotus open directly behind the head (this feature is visible in the coloured photo of Figure 9 but unrecognizable in black-and-white photos published elsewhere), which is a feature similar to a better preserved one of the Durjanapura tīrthaṅkara images (Figure 8) (Dr. Claudine Bautze-Picron; Dr. Yoachim Karl Bautze: pers. comm.). Despite such archaism derived from Kapardin Mathurā Buddha/Bodhisattva images, the Bodh Gayā Buddha/Bodhisattva is more advanced, i.e. Gupta, than the former, which is apparent from his characteristic Gupta meditative eyes. The inscription of the Bodh Gayā Buddha/Bodhisattva gives the year 64 in an unknown era under the reign of a Mahārāja Trikamala. On the basis of its palaeography, it has tentatively been dated in the Gupta era, thus giving the corresponding date of AD 383/4 (Damsteegt 1978: 156). The find-spot of the Buddha/Bodhisattva image in Magadha, not too distant from the Gupta capital of Pāṭaliputra (Patna), not only supports the dating to the Gupta era, but also indicates the image’s importance as a rare early specimen from the Gupta heartland proper. One Buddha image of the unknown origin in beige sandstone seated in the bhadrāsana (with legs pendant) on a padmāsana/siṃhāsana (a lotus/lion throne) (Museum für Asiatische Kunst, Berlin, acc. I 22; Revire 2016: vol. 1, 55-56; vol. 2, 36, fig. 2.34; Martina Stoye, pers. comm.), possibly from central or western India, also seems to fall into the formative stage of Gupta Buddha imagery. However, this buddha image may date slightly later than the other examples, probably to around c. AD 400, contemporaneous with Devnīmorī buddha images. As Revire correctly observed, this buddha image has an archaic yet unusual style: while the styles of the robe and the nimbus show the late Kushan features of the third century AD, the introduction of the bhadrāsana and the padmāsana (‘lotus throne’) is almost completely new (Revire 2016: 55-56). 167 The Global ConneCTions of Gandhāran arT Figure 8. Sandstone tīrthaṅkara figure from Durjanapura, Madhya Pradesh, reign of Rāmagupta (reigned c. AD 376380), Gupta period. Height: 66 cm. Bhopal Museum (Photo: courtesy of the Huntington Archive.) Figure 9. Sandstone Buddha/Bodhisattva figure from Bodh Gayā, Bihar, reign of Mahārāja Trikamala, AD 383/384?, Gupta period. Kolkata, Indian Museum. (Photo: courtesy of Joachim Karl Bauze) However, the drapery of this Buddha with no provenance seems more intricate and thus advanced, and even comparable to the much later Govindnagar Buddha, dated to AD 434/435, which wears a robe with a similar drapery. Another Buddha image possibly belonging to this phase can be recognized. This second, similar yet headless Buddha image, now in the Metropolitan Museum, New York (inv. 1992.191), claimed by the Museum to be from Uttar Pradesh and from the early fifth century AD, shows a similar type of drapery but also an archaic pedestal in the earlier Kushan or post-Kushan Mathurā style; it is also comparable to the Mankuwar Buddha, dated to AD 427/428 or 429 with a similar pedestal. The issue of an archaic revival of Kushan or post-Kushan Mathurā features is highly tricky in dating many early Gupta images. Overall, the early Gupta Jain tīrthaṅkara images from Vidiśā are regarded as anticipating, together with the Bodh Gayā Buddha/Bodhisattva image and fully-fledged Gupta-style buddha images at Devnīmorī soon after, the fully-fledged/mature Gupta Buddha images at Mathurā and Sārnāth that developed during the following fifth century AD. These examples from Vidiśā are considered to be rare Gupta prototypes that are largely based on Kushan-Mathurā Buddha/Bodhisattva and tīrthaṅkara images ( Miyaji 1980: 16-20; Williams 1982: 28-29, 33-34; Huntington 1985: 188). The relative dating of the Devnīmorī buddha images thus generates a time span between c. AD 376/380, with Durjanpura tīrthaṅkara images as a terminus post quem, and AD 427/428 or 429, with the Mankuwar Buddha image as a terminus ante quem. Considering the likely date of the construction of the mahāstūpa at Devnīmorī to be the late fourth century AD (despite a possible earlier history of the mahāstūpa ascribed to the first, earlier internment of Casket I and the pot containing eight coins in the third century AD), in the late Western Kṣatrapa 168 ken ishikaWa: more Gandhāra Than maThurā period, prior to the conquest of Gujarat by Candragupta II of the imperial Guptas, the Devnīmorī buddha images can be regarded as among the earliest fully-fledged Gupta remains. The mahāstūpa at Devnīmorī, in fact, is also the earliest ‘Gupta’ monument in brick (Harle 1974: 29). Even so, the idea that Devnīmorī or the Western Kṣatrapas were the progenitor of Gupta material culture has long been a subject of debate (Williams 1982 58-9). Although the role of western India in the formation of pan-Indian Gupta material culture is a notoriously problematic issue, we might further contextualize Devnīmorī by reconsidering the extent of the late Gandhāran influence as well as pre-existing material culture of Gujarat. Gandhāran influence on the Devnīmorī buddha images Although a very small number of motifs decorating the mahāstūpa at Devnīmorī originate in Gandhāra, most importantly the chequer pattern (Williams 1982: 59), I consider the overall Gandhāran influence found at the monastic complex of Devnīmorī to be very significant because of the form of the terraced mahāstūpa on a double-square platform; the short cylindrical reliquaries (though not exclusively Gandhāran as mentioned earlier); the buddha cult imagery; one of the architectural elements with a relief of a buddha on acanthus leaves - possibly part of a pilaster (Devnīmorī: Mehta & Chowdhary 1966: pl. 60, A and B)1; and the schist relief of the buddhapāda or the footprint of the Buddha with svastika symbols on its fingers (Devnīmorī: Mehta & Chowdhary 1966: 67, B; Gandhāra: e.g. Kurita 2003: figs. 786 and 788). However, the postulation of an indirect Gandhāran influence through earlier Mathurā rather than directly needs to be treated cautiously; for instance, the interaction between Gandhāra and Mathurā can be seen in the iconography of the Seven Buddhas of the Past and/or Maitreya (Behrendt 2014) or in the Indo-Corinthian pilasters (Mehta & Chowdhary 1966: figs. 55, B, C) of Gandhāran origin that appear at Mathurā and then at Devnīmorī.2 In particular, in what follows, I demonstrate direct Gandhāran influence on Devnīmorī sculptures in the form of the occasional wavy hairstyle, the monastic dress, and the lotus throne. Gandhāran wavy hairstyle The occasional occurrence of a Gandhāra-derived wavy hairstyle has long been recognized at Devnīmorī (Figures 10 & 11) as the most obvious example of Gandhāran artistic/iconographic influence beyond Greater Gandhāra (Sompura 1969: fig. 12). Although the vast majority of Devnīmorī buddha images have a series of curls known as Gupta curls, which slightly differ from other Gupta and late Gandhāran counterparts and which are even comparable to those from Andhra (Mori 2007: 285), one buddha image and one fragmentary buddha head excavated from the mahāstūpa at Devnīmorī exhibit a specific late Gandhāran variant of the Gandhāran wavy hairstyle (Mehta & Chowdhary 1966: pl. 40, A & B) (Figure 10). Although the early Gandhāran wavy hairstyle of buddhas in Gandhāra was rendered naturalistically in the Greco-Roman style (Rhi 2008b), the one at Devnīmorī constitutes a distinct late Gandhāran variant, consisting of a series of bow-shaped waves in a few radiant concentric circles, altering their direction one layer after another. This distinctive wavy hairstyle modelled in terracotta at Devnīmorī originates in one of the recognizable, though not yet systematically studied, variants of the late Gandhāran wavy hairstyle in Gandhāra, which appears mostly in stucco but occasionally in terracotta or stone. Gandhāran examples of the wavy hairstyle seen at Devnīmorī include the stucco buddha head in the Victoria and Albert Museum (inv. IM3-1931) and a stone buddha head in the British Museum (inv. OA 1889-174), which are both typically dated to fourth-fifth century AD (Zwalf 1996: 460). Williams 1 2 See also Cousens 1914: pl. 36, b for Mīrpur Khās; Zwalf 1996: 46 for Gandhāra. See also, for Mīrpur Khās: Moti 1964: fig. 13b, but also at earlier Mathurā, Vogel 1930: fig. LIII, c. 169 The Global ConneCTions of Gandhāran arT Figure 11. Detail of the sculpture in Figure 10 showing the Gandhāraderived wavy hairstyle. (Photo: author, courtesy of M.S. Baroda University.) Figure 10. One of the Devnimori buddha images. (Photo: author, courtesy of M.S. Baroda University.) observed the long continuation of the late Gandhāran wavy hairstyle in Gandhāra, as late as in the seventh century at Fondukistan (Williams 1982: 59), though her example merely shows the persistence of the late Gandhāran wavy hairstyle in general, but not necessarily of its distinctive variant seen both in Gandhāra and Devnīmorī. Gandhāran influence on the monastic dress of the Devnīmorī buddhas There are two types of monastic dress among the fourteen Devnīmorī buddha images published: nine are entirely clad (Mehta & Chowdhary 1966: pl. 37, A-D, pl. 38, B-D, pl. 39, A and pl. 42, B) (Figure 13) and the other five have only one shoulder covered and the other exposed (Mehta & Chowdhary 1966: pl. 38, A, pl. 39, B-D; Chowdhary 2010: 80. fig. 39, B) (Figure 14). In the more frequent first type (entirely clad), the drapery takes the form of concentric U-shaped lines, which apparently corresponds to the similar drapery from the late Kushan and post-Kushan periods onwards (Takata 1967: 334-342; Miyaji 1980: 18-9). Figure 12. The Devnimori-type, Gandhāran wavy hairstyle of a stucco buddha head from Hadda, Afghanistan. Kabul Museum. (Photo courtesy of the Huntington Archive, Digital Database Collection, Scan Number 10579.) In Gandhāra, the first type (entirely clad) is associated with meditation or, iconographically, the dhyānamudrā (a meditative hand gesture with hands resting on the lap) whereas the second type (with only one shoulder covered) is related to teaching or the dharmacakramudrā (a hand gesture of teaching that symbolises the turning 170 ken ishikaWa: more Gandhāra Than maThurā Figure 13. One of the Devnimori buddha images. (Photo: author, courtesy of M.S. Baroda University.) Figure 14. One of the Devnimori buddha images. (Photo: author, courtesy of M.S. Baroda University.) of the dharmacakra, the wheel of the dharma i.e. teaching of the Buddha or a buddha or, in some cases, an advanced bodhisattva) (Filigenzi 2005: 108-109). Although the combination of a type of dress and its corresponding mudrā may carry a specific meaning in Gandhāra, Devnīmorī buddha images with both types of dress all invariably show the dhyānamudrā. Two fragments of the buddha images in schist, possibly produced locally, with the first dresstype (entirely clad) but in the Mathurā style were also found at Devnīmorī (Mehta & Chowdhary 1966: 89: pl. 23, D and E). On the other hand, the second dress-type (with Figure 15. Detail of Figure 14, revealing the Gandhāran-type only one shoulder covered) largely reflects the inner dress. (Photo: author, courtesy of M.S. Baroda University.) adaptation of the Buddha/Bodhisattva images of this type in Kushan Mathurā. As observed by Uehara (Ishikawa & Uehara 2014), the second dress-type at Devnīmorī also shows the inner dress (Figure 15) not seen at Mathurā but very obviously depicted in late Gandhāran images of a preaching buddha, for example, the Muhammad Nari stele (Figure 26). This particular type of monastic dress might realistically depict the three garments prescribed in the Vinaya, i.e. the ‘robe’ (Pāli, uttarāsaṅga) subtly shown under the ‘upper garment’ (Pāli, saṅghāti) on the upper body and the ‘undercloth’ (Pāli, antaravāsaka) on the lower body (Griswold 1963: 87-88). 171 The Global ConneCTions of Gandhāran arT Uehara also pointed out peculiar details of the upper garment that appear on the first dress type (entirely clad) of Devnīmorī buddhas: the pronounced layers of the long sleeves of the upper garment visibly overlay the thighs (Ishikawa & Uehara 2014). This peculiar feature is also seen in one later buddha bronze from Swat, now in the Ashmolean Museum (inv. EA1995.115) (Ishikawa & Uehara 2014), along with the same concentric U-shaped lines of the drapery and the Gandhāran-style wavy hairstyle. Meanwhile, the occurrence of the second dress type (with only one shoulder covered) contradicts the general trend of the Gupta period (except for Ajaṇṭā), during which the first dress-type (entirely clad) was preferred. This peculiarity can be interpreted both as an archaic feature inherited from early Swat or KushanMathurā Buddhas/Bodhisattva images of around the first to third century AD and as a reflection of the late Gandhāran adaptation of this feature from the third and fourth centuries AD. Although the aforementioned Bodh Gayā Buddha/ Bodhisattva image also shows the second type (with only one shoulder covered), the Devnīmorī buddha images seem to be less archaic. Figure 16. The Gandhāran inside-out, single-petalled lotus throne of one of the Devnimori buddha images. (Photo: author, courtesy of M.S. Baroda University.) Gandhāran influence on the lotus thrones of the Devnīmorī buddha images There are three types of lotus thrones at Devnīmorī: 1) the single-petalled type (Mehta & Chowdhary 1966: pl. 39, B; Chowdhary 2010: 81, fig. 40, A); 2) the double-petalled type (Mehta & Chowdhary 1966: pl. 37, A & C; pl. 39, A) (Figure 18); and 3) the inside-out singlepetalled type (Mehta & Chowdhary 1966: pl. 37, B & D, pl. 38, A-D, pl. 39, C & D, pl. 42, B; Chowdhary 2010: 80, fig. 39, B). Of these, the inside-out, single-petalled lotus throne (Figures 16 and 17), which is the most frequent at Devnīmorī, is of a particular type that, in my view, consists Figure 17. The inside-out, single-petalled lotus throne of a late Gandhāran preaching buddha image. Taxila Museum. (Photo: courtesy of the Huntington Archive, Digital Database Collection, Scan Number 10271.) 172 ken ishikaWa: more Gandhāra Than maThurā of the three horizontal layers of lotus components such as, from the bottom, inside-out petals, filaments in the form of vertical lines (rather than kuśa grass spread underneath certain buddhas in early Swāt, mature Gandhāra, and post-Kushan Mathurā) and an oversized stigma as a cushion. Inside-out single-petalled lotus thrones almost identical to those at Devnīmorī are seen underneath some images of the late Gandhāran preaching buddha (for Gandhāran examples, see Harrison & Luczanits 2011; 207, figs. 16 and 18). Figure 18. The double lotus throne of one of the Devnimori buddhas. On the other hand, the single(Photo: author, courtesy of M.S. Baroda University.) petalled and double-petalled lotus thrones at Devnīmorī can also be compared with other late Gandhāran counterparts, which are more three-dimensional in nature (once again underneath late Gandhāran preaching buddha images).3 These are more conventional and standardized, like the Andhran examples.4 Brown has even shown the resemblance to the lotus thrones of one Mucilinda (Mucalinda in Pāli) Buddha from Andhra, which precedes the Gupta period, and a Śiva from Mandhal, from the early Vākāṭaka period (Brown 2004: 6, figs. 5.18, 5.19). According to Harrison and Luczanits, the lotus throne or padmāsana is not traceable to the earliest groups of Buddha images in Kushan-period Gandhāra and Mathurā in a strict sense (Harrison & Luczanits 2011: 81). However, the Buddha on the lid of the so-called ‘Kaṇiṣka’ reliquary (Jongeward et al 2012: 82-83, figs. 3, 32a-b, 276, no. 253), more recently dated to the reign of Huviṣka c. mid- to late second century AD (Errington & Falk 2002), is technically and conceptually seated on a padmāsana, with a stigma of the flattened full-blown lotus engraved on the lid, while attended by Indra and Brahmā both exhibiting the añjalimudrā. Similarly, the lotus throne was not unattested at late Kushan or post-Kushan Mathurā: one small Buddha image on a double-petalled padmāsana (height: 38 cm), with a halo having the typically Mathurā-school scalloped edge, was found at Chaubara (Lucknow Museum inv. B 23; Foucher 1905: 685, fig. 552), in which the Buddha is attended by what appear to be Indra and Brahmā (but seated on lotuses!) adopting the añjalimudrā, more or less like the Buddha over the ‘Kaṇiṣka’ reliquary. In contrast, the lotus throne was well documented in South India, predominantly in buddha images in Andhra Pradesh of the late second to the early third century AD, as well as among seated late Gandhāran buddha and bodhisattva images, which are ‘generally’ dated to the third to fourth centuries AD (Harrison & Luczanits 2011: 81-83). Lotus thrones of buddhas in India proper certainly became manifest in pan-Indian Gupta material culture, especially at Sārnāth (Figure 19) and Ajaṇṭā. The inscription on the Sārnāth Buddha image dated to AD 477 refers to his single-petalled lotus throne as a padmāsana in the sense of a lotus throne For example, for the single-petalled type, see Harrison & Luczanits 2011: fig. 9; for the double-petalled type, see Harrison & Luczanits 2011: fig. 12; Loriyan Tangai/Indian Museum, Kolkata: inv. A23485/5090) 4 For example, Stone 1994: figs. 110, 112, 118. 3 173 The Global ConneCTions of Gandhāran arT rather than the cross-legged posture of the same name (Rosenfield 1963: 12-13). The earliest occurrence of the padmāsana in Gupta India proper, at least in a Buddhist context, was possibly among Devnīmorī buddha imagery c. AD 400 but it may be challenged by the two depictions of Brahmā on a lotus over the Varāha image at Cave 5 at Udayagiri (Mitra 1963: 100) and over the severely damaged image of Viṣṇu as Śeṣaśayana at Cave 13 (Willis 20014: 31), both of which can be dated to the fifth century AD but possibly as early as c. AD 400 (Huntington 1985: 192-193). Within a century or so after Devnīmorī, the padmāsana spread, in the Buddhist context, to Sārnāth (Figure 19) and Ajaṇṭā. The doublepetalled padmāsana of the Gandhāra/Devnīmorī type also reached Sārnāth (Huntington 2000: 35, fig. 3) and the Western Deccan, where it is ubiquitous. Even in the Hindu-Gupta context, within a century or so after Udayagiri, padmāsanas appeared underneath Śiva at Mandhal under the Vākāṭakas (Brown 2004: 68, fig. 5.20) and then Brahmā at Deogarh (Huntington 1985: 207, fig. 10.29). The Gupta text of Kumārasambhava (86.2) by Kālidāsa, indeed, describes Brahmā as sitting upon a padmāsana (‘padmāsanasthā’) when worshipped. One key aspect of this development is an innovative combination of an Indic siṃhāsana (‘lion throne’) and a Gandhāran padmāsana (‘lotus throne’) within a single throne that appears among Buddha images in the socalled bhadrāsana posture c. AD 400, as seen in the aforementioned Buddha image with no provenance (Museum für Asiatische Kunst, Berlin, inv. I 22; Revire 2016: vol. 1, 55-56; vol. 2, 36, fig. 2.34) and then at Sārnāth (e.g. inv. 1880.7, British Museum) and Ajaṇṭā (e.g. the main cult image of a buddha juxtaposed onto the rock-cut Figure 19. Sandstone Buddha image from Sārnāth, AD 476/477, Gupta period. Sarnath Site Museum. (Photo: courtesy of the Huntington Archive, Digital Database Collection, Scan Number 55207) 174 ken ishikaWa: more Gandhāra Than maThurā stupa in Cave 26: Spink 2009). This certainly is a significant late Gupta innovation. In this connection, it is noteworthy in a conceptual sense that the Gilgit manuscripts of the Saddharmpuṇḍarīka, a siṃhāsana is considered to be located in a calyx of a lotus (Schopen 1977: 182). In this connection, the lotus throne of the buddha at Museum für Asiatische Kunst, possibly dating to c. AD 400, seems extremely experimental, since such a lotus throne is normally used for buddhas or bodhisattvas seated cross-legged or standing on it but here used as a seat. A pair of lions are somewhat detached from the structure but they definitely derive from those of the Indic-type pedestal in the Kushan and post-Kushan Mathurā style. On the other hand, the Sārnāth version of the siṃhāsana/padmāsana ‘(lion/lotus throne’ or vice versa) shows the adaptations of a classic Indic siṃhāsana in the style of Kushan and post-Kushan Mathurā and of a classic Andhra-style padmāsana footrest, which originates in the square pedestals of the Buddha’s footprints (buddhapāda) (Harrison & Luczanits 2011: 81). The lotus throne, however, was the most significant in the late Gandhāran context among late Gandhāran preaching buddha images, as exemplified by the Muhammad Nari stele (Harrison & Luczanits 2011) (Figure 26). In this context, Rhi associated images of the late Gandhāran buddha on a lotus with the textual account of the practice of image-making of buddhas on lotuses (Rhi 2003: 167-170) while Harrison and Luczanits shed light on the soteriological significance of the image-making tradition to be reborn on a lotus (Harrison & Luczanits 2011: 116-117). Overall Gandhāran influence on the Devnīmorī buddhas As attested above, the wavy hairstyle, the inner dress. and the lotus throne of the Devnīmorī buddhas dating to c. AD 400 are incidentally and specifically associated with almost contemporary late Gandhāran buddhas, most importantly, preaching buddhas, that are generally dated to the third to fourth century AD but evidently not considerably earlier than Devnīmorī buddhas. Substantial and persistent Gandhāran influences beyond Devnīmorī Mīrpur Khās in Sindh In the neighbouring region to Gujarat, to the north-west in Sindh, at the remote site of Mīrpur Khās (Figure 1), now reportedly destroyed, one comparable set of material remains to Devnīmorī is known from its Kahujo-daro stūpa. It has a Gandhāran-style terraced brick stūpa on a square platform and Buddha imagery which is Gupta-style but with some clearly late Gandhāran influences (Cousens 1914). The stūpa facing the west rests on a one-tier square platform, whose front i.e. western face had a slight projection, with three small inner cell shrines and traces of a pair of stairs leading up to the terrace (Cousens 1914: 83). Made in terracotta, plastered and decorated with polychrome, the Mīrpur Khās buddha images in high relief are set against square panels with ornamental edges. They once adorned the square platform of the stūpa, three on each of the three side faces of its one-tier square platform, numbering nine in total (Cousens 1914: 86) (Figures 20, 21, 22 & 23). Buddhist material culture in Sindh shows stylistic similarities with that in north Gujarat: Van Lohuizende Leeuw conducted a comparative analysis of similar motifs used in carved bricks between various sites in Sindh (including Mīrpur Khās and Sudheranjo-dāro) and Devnīmorī (Van Lohuizen-de Leeuw 1979: 167-168). Likewise, the Mīrpur Khās buddha images are also comparable to those at Devnīmorī and the sizes are almost equivalent to each other; for instance, the buddha image in the Victoria and Albert Museum mentioned below (IM13-1931) measures 68 cm in height. 175 The Global ConneCTions of Gandhāran arT 20 21 22 23 Figures 20-23. Buddha images from Mīrpur Khās, Sindh. Mumbai, Chhatrapati Shivaji Maharaj Vastu Sangraha. (Photos: courtesy of the Huntington Archive, Digital Database Collection, Scan Numbers 6752, 6744, 6764, 6755.) 176 ken ishikaWa: more Gandhāra Than maThurā The practically uniform Mīrpur Khās buddhas are again invariably shown meditating, with downcast meditative eyes, seated in the padmāsana and exhibiting the dhyānamudrā, as at Devnīmorī. They consistently wear a garment, with both shoulders covered and, unlike the figures from Devnīmorī, there are no buddha images with only one shoulder covered. As on the Devnīmorī buddhas, the ūrṇā is occasionally absent, and there are variations in hairstyles and thrones (Cousens 1914: 86-87). At Mīrpur Khās, the double-petalled lotus throne of Devnīmorī type also occurs (Chandra 1964: figs. 1, 3b) (Figure 20), but the majority of thrones at the site constitute one distinctive type of the single-petalled (Figure 21) or double-petalled (Figure 22) lotus thrones (Chandra 1964, figs. 2b, 3a), which are rather closer to those of later buddha bronzes from the Swāt Valley from the following centuries. The Mīrpur Khās buddhas certainly follow their regional prototypes from Devnīmorī. They are also fundamentally comparable to Gupta-Sārnāth Buddha images but are also substantially influenced by late Gandhāran buddhas. Out of the nine buddha images at Mīrpur Khās, seven were still in situ at the time of the excavation, while two had already been removed (a buddha image and a buddha head) by Woodburn, as stated in the notes to his published drawings (Woodburn 1897). The buddha image is identified with the one now in the Victoria and Albert Museum. The buddha head, with the upper part of the panel and the halo, may be identical to the one also now in the same museum (inv. IM141931), but, if so, the head is likely to have been removed from one of the nine images (Cousens 1914: 86). Finally, five of the remaining buddhas are now in Chatrapati Shivaji Maharaj Vastu Sangrahalaya (CSMVS), along with numerous terracotta ornaments, as well as the famous dvārapāla with polychrome decoration (Chandra 1964: fig. 5b; Mukherjee 2008: 68). As for the dating of the Kahujo-dāro stūpa at Mīrpur Khās, the uninscribed reliquary and its undated deposits do not provide any evidence. The Mīrpur Khās buddha images show somewhat later features, especially in the diaphanous drapery of the upper garment and their elaborate halos/thrones, which are comparable to Buddha/buddha images from Sārnāth (and Ajaṇṭā) from the late fifth century AD. Two such Gupta-Sārnāth Buddha images are dated to GE 154 or AD 473/474 under the reign of Kumāragupta II (Rosenfield 1963: 10, fig. 1) and to GE 157 or AD 476/7 under the reign of Buddhagupta (Rosenfield 1963: 11, fig. 2). All these images from Sārnāth (and Ajaṇṭā) can be used as a basis for comparison with the Mīrpur Khās buddha images. Along with the aforementioned Jain image from Mathurā under the reign of Kumāragupta I, the three dated Buddha images from Sārnāth testify that they were produced directly under the imperial Guptas during the fifth century AD. Huntington dates the Mīrpur Khās buddha images to the mid- to late fifth century AD (Huntington 1985: 205), which seems cautious and appropriate, if we take into consideration Williams’s argument that there was no significant time lag between the centre and the ‘provinces’ in Gupta India (Williams 1982: 34). Another piece of evidence to support such an argument is the pan-Indian motif of a pair of long-necked haṃsas or geese coming out of the mouths of a pair of makaras, as decorative elements of the throne of a preaching buddha. A pair of long-necked haṃsas occurs in a portable terracotta plaque (16 x 15 cm) at Mīrpur Khās (Chandra 1964: fig. 4b) but also in rock-carvings at Ajaṇṭā (e.g. Spink 2008: fig. 81), Aurangabād (e.g. Huntington Archive no. 55845: Cave 7) and Nāsik (e.g. Huntington Archive no. 26856: Cave 23) in the Western Deccan (Dr. Claudine Bautze-Picron, pers. comm.). The diaphanous drapery of the upper garment and the elaborate halo of the Mīrpur Khās buddha images are comparable to those of Gupta Sārnāth from the late fifth century AD, but subtle V-shape wrinkles on the drapery are also related to Gupta Buddha images from Mathurā. However, the Mīrpur Khās buddhas reflect the proportional blending with its own Gandhāran heritage. The Gandhāran hairstyle of one of the Mīrpur Khās images (Figure 20) again is one of the most obvious Gandhāran elements (Chandra 1964: fig. 1) and the undersized halos at Mīrpur Khās (Figures 20, 21, 22 & 23) unlike their oversized Gupta counterparts can also be seen to reflect general Gandhāran influence. Similarly, the rim of the triangular rays of the Mīrpur Khās buddhas (Figures 20, 21 & 22) seems to derive from the halos of Gandhāran bodhisattvas (e.g. 177 The Global ConneCTions of Gandhāran arT Bautze-Picron 1990: 84, 91 n. 72, figs. 13 & 15; Zwalf 1996: 55, 58, 66), rather than to the scalloped halos from Kushan Mathurā that Huntington alludes to (Huntington 1985: 197-198). Two of the Mīrpur Khās buddhas (Chandra 1964: fig. 3a; Gorakshakar 1991: 83, fig. 2) subtly show an explicitly Gandhāran folded hem: with the zigzag hemline to the proper left of the Gandhāran-style ‘droopy’ semicircular front hem (Figure 21). This folded hem is absent from the Sarnāth Buddha images, but common in Gandhāra, where it is depicted aside the droopy semicircular front hem.5 However, similar zigzag hems can be seen in standing images from Mathurā already in the early fifth century AD as exemplified by the aforementioned Givindnagar Buddha image. Therefore, the zigzag hemline at Mīrpur Khās in the late fifth century AD can be taken either as the result of pan-Indian influence or the inheritance of the Gandhāran aesthetics through Mathurā. As for the ‘shortened legs’ of the Mīrpur Khās buddhas, this peculiarity can be explained by their undersized thrones (Figures 20, 21, 22 & 23), just like those of late Gandhāra, which are commonly narrower than the horizontal extent of the seated buddha, as in the central preaching buddha in the Muhammad Nari stele (Harrison & Luczanits 2011: 199, fig. 4) (Figure 26). It seems that placing a late Gupta buddha on such a small late Gandhāran lotus throne at Mīrpur Khās resulted in the shortening of the legs as a provincial feature (Figure 23). Śāmalājī and Dhānk in Gujarat In the period following Devnīmorī, i.e. after c. AD 400, Gandhāran influences can be traced within Gujarat itself at Śāmalājī near Devnīmorī in north Gujarat as well as at Dhānk in inland Saurashtra. At the Hindu site of Śāmalājī, dated to the beginning of the sixth century AD by Schastok (Schastok 1985), some late Gupta Hindu images possibly show some Gandhāran influences in their zigzag drapery, drapery loops, and leaf ornaments in female coiffures. This was argued by Shah with comparison to a late Gandhāran Hārītī image from the Peshawar Museum (Shah 1960: 60-62) possibly dating as late as the fifth century AD (Lyons & Ingholt 1971: 39). However, Shah’s view that there was direct Gandhāran influence on Śāmalājī was later rejected by Schastok, who regarded Śāmalājī as an example of indirect Gandhāran influence, owing to the chronological gap between Gandhāra and Śāmalājī and the potential role of Mathurā as a mediator of the influence (Schastok 1985: 33-35). At Dhānk in Saurashtra (Figure 1), during my fieldwork I have discovered a relief in a weathered condition depicting a Buddhist triad with a preaching buddha in the bhadrāsana (with his legs pendant), possibly attended by a pair of bodhisattvas (Figure 24). This new evidence is complemented by the much larger image of the buddha in the bhadrāsna, again from Dhānk, that was reported with a photograph (neg. 210.39, American Institute of Indian Studies, Gurgaon; Ray 2004: 55, fig. 4) but now unfortunately untraceable. Such paradoxically provincial yet pan-Indian kinds of buddha images date probably 5 Figure 24. The stele of the preaching buddha image in a triad with a pair of bodhisattvas from Dhānk, Gujarat (c. 50 x 50 x 20 cm). (Photo: author, courtesy of M.S. Baroda University). For example, a Gandhāran buddha in schist from the British Museum: inv. 1895,1026.1. 178 ken ishikaWa: more Gandhāra Than maThurā Figure 25. Buddha image in a triad with a pair of bodhisattvas in Cave 26 at Ajaṇṭā, late fifth century AD, the Vākāṭaka period. (Photo: courtesy of the Huntington Archive, Digital Database Collection, Scan Number 8616.) Figure 26. The Muhammad Nari stele from Muhammad Nari. Lahore Museum. (Photo: courtesy of the Huntington Archive, Digital Database Collection, Scan Number 9650.) 179 The Global ConneCTions of Gandhāran arT from the late fifth to the early sixth century AD, from the time of the Mairtaka dynasty, though the latter piece was dated by Ray to the fourth century AD (Ray 2004 55, fig. 4), which seems dubious and too early. Spink dates buddha images in bhadrāsna at Ajaṇṭā to the late fifth century AD (Spink 2009: figs. 66, 98) but bhadrāsna buddha images at Ajaṇṭā are generally understood to date somewhat later (Owen 2001: 38). It can be seen clearly that these Dhānk buddha images relate on one hand to Sārnāth and on the other to Ajaṇṭā from the late fifth century AD onwards, especially in respect to the predominance of the bhadrāsna, which was the least common pose in Gandhāra. Although the ultimate origin of the iconography of the preaching buddha is certainly Gandhāran, the Buddhist triad found at Dhānk is stylistically related to those at Ajaṇṭā. Yet some provincialized Gandhāran stylistic features such as the broad shoulders and the drapery are still seen in the larger buddha found at Dhānk. Ajaṇṭā in the Western Deccan and beyond Figure 27. The imported sandstone Kushan-Mathurā Buddha/ Bodhisattva found at Vadnagar, Gujarat. Second century AD. Vadnagar Museum. (Photo: author, courtesy of M.S. Baroda University.) At Ajaṇṭā in the Western Deccan, Gandhāran influences are seen provincialized in such a distant region. The connection between Ajaṇṭā and Gandhāra is evident especially in the Gandhāranstyle murals of standing buddhas added to the pillars of Cave 10 (Krishna 1981) but – relevant here – also significant in some of the buddha images at the site (Miyaji 1985; Fukuyama 2014). Almost simultaneously with Sārnāth and Mīrpur Khās, the first ever locally created buddha images emerged at Ajaṇṭā in the late fifth century AD (Spink 2009: 33-35): for example: the standing buddha at Cave 19, the seated buddha in the bhadrāsana at Cave 26, and the seated buddha in the padmāsana at Cave 11), that are essentially parallel to the Sārnāth Buddha images. Ajaṇṭā was then directly under the Vākāṭakas, who allied themselves with the Guptas. According to Spink, the buddhas at Ajaṇṭā variously date between AD 468 and 480 with very speculative precise dates based on external evidence, assigned to each of them (Spink 2009: figs. 22, 57, 66, 70, 71, 81, 84, 90, 98, 122, 130, 132, 147, 148, 173, 178, 181, 182). Although it is impossible to verify these exact dates, it can be said that these buddha images were produced in the dating range of AD 468-480 or later, which is more or less contemporaneous with the dated Sārnāth Buddha images mentioned earlier. However, different chronologies of the monastic complex of Ajaṇṭā also exist (Fukuyama 2014: 77-87). At Ajaṇṭā, Gandhāran influences are clearly seen in images of the preaching buddha in a triad with a pair of bodhisattvas as also seen in late Gandhāra but often in the bhadrāsana, which was rare in Gandhāra (Miyaji 1985; Fukuyama 2014) (Figure 25), even more clearly than Devnīmorī and Dhānk. It appears that the late Gandhāran preaching buddha images of the third to fourth century AD became influential in Buddhist rock-cut cave complexes in the Western Deccan, most prominently at Ajaṇṭā, from the late fifth century AD onwards. As discussed above, both the inner dress and the lotus throne of the Devnīmorī buddha images dating to c. AD 400 are incidentally and specifically associated with the almost contemporary late Gandhāran preaching buddhas less than a century before Ajaṇṭā. 180 ken ishikaWa: more Gandhāra Than maThurā Despite such clear influences from late Gandhāran preaching buddha images at Ajaṇṭā, all the buddha images at Devnīmorī invariably show the dhyānamudrā. However, the preaching buddha images at Ajaṇṭā clearly inherited the mudrā of the Gandhāran preaching buddhas, which resembles the contemporary dharmacakramudrā shown by Sārnāth Buddha images. The dharmacakramudrā mainly symbolizes the Buddha’s first sermon (Sounders 1960: 94) (though not exclusively). However, the dharmacakramudrā at Ajaṇṭā is considered a distinct variant and identifications of buddhas are unclear: Huntington even identifies them as the double image of Śākyamuni/Vairocana (Sounders 1960: 94; Huntington 2000: 34-36). Certain iconographic elements of the ‘palace’ and ‘lotus-pond’ compositions of the late Gandhāran preaching buddha images (Harrison & Luczanits 2011: 96-101, figs. 10-12, 16) also recur at Ajaṇṭā, for instance, a pair of putti with wings holding a circular wreath or an umbrella over the head of the buddha with a new addition of a crown also being offered to him instead, though the ‘palace-type’ architecture was largely lost or simplified (Fukuyama 2014: part 2) (Figure 25). The substantial recurrence of the late Gandhāran double-petalled lotus throne of the Devnīmorī type, of Gandhāran origin, occasionally growing out of a pond, accompanied by a pair of nāgas (Figure 25), is also geographically and chronologically significant in the transmission and the distribution of the iconography of the late Gandhāran buddha from Gandhāra to the Western Deccan, possibly through Gujarat. The double-petalled lotus throne of the Devnīmorī/Ajaṇṭā type, of Gandhāran origin, eventually became prevalent during the Pāla and Sena periods in eastern India, and in Bengal in the eighth to thirteenth century AD. The problematic identifications of preaching buddhas and bodhisattvas in Mahāyāna Buddhism The above case-studies of the Devnīmorī buddha images and other related examples demonstrate strongly that late Gandhāran influence was substantial and far-reaching. I have shown that the Devnīmorī buddha is stylistically, chronologically and geographically related to the late Gandhāran preaching buddha. It is evident that the iconography of the late Gandhāran preaching buddha played an exceptional role in the cultural transmission of Gandhāran Buddhist material culture to Gujarat and beyond. A further investigation on the late Gandhāran preaching buddha may assist the identification of the Devnīmorī buddha images, for which only limited evidence is available. Therefore, this section revisits the issue of the obscure identity of the late Gandhāran preaching buddha and discusses the implications it may have for identification of Devnīmorī buddha images. Identifying the late Gandhāran preaching buddha There is an identification problem regarding the late Gandhāran preaching buddha showing his characteristic preaching mudrā, on a lotus throne, in a triad with a pair of bodhisattvas. This appears on over forty steles in Gandhāran Buddhist art (Miyaji 2011: 129), in both simple and complex compositions, as exemplified by the Muhammad Nari stele (Figure 26). Identification of the preaching buddha remains an open question, despite tremendous art-historical and Buddhological investigations hitherto carried out (Harrison & Luczanits 2011: Miyaji 2002). Luczanits and Harrison categorized extant remains of the late Gandhāran preaching buddha images into three main compositional types defined by such representations as lotus ponds, palaces, and emanations, of which the palace type was selected as the main specimen owing to its iconographic richness and complexity, suitable for its comparison with textual evidence (Harrison & Luczanits 2011: 88-106, 117-118). In short, the palace type as a composite stele, as in the Muhammad Nari stele (Figure 26), is characterized by a central preaching buddha on a lotus, who is exhibiting the Gandhāran dharmacakramudrā in a triad with a pair of bodhisattvas; he is surrounded by buddhas and bodhisattvas in tiers in a palace-like 181 The Global ConneCTions of Gandhāran arT architectural composition or mandorla. Having been identified initially as a depiction of the Miracle of Śravāstī (Foucher 1905), the composite images of the late Gandhāran preaching buddha in a palace-like mandorla have been seen increasingly as a product of Mahāyāna Buddhism, primarily on account of the presence of bodhisattvas, whose precise identities are as problematic as the main buddha. Proposed Mahāyāna identifications of the late Gandhāran preaching buddha, established through text-image parallels, include: Amitābha or Amitāyus in Sukhāvatī (Huntington 1980; Fussman 1987: 73; Quagliotti 1996); Akṣobhya in Abhirati (Schopen 1987: 273-274, n. 50); the cosmological Śākyamuni of the Mahāyāna Buddhist imagination, not as the historical Buddha but as his Mahāyāna manifestation that appears in multiple Mahāyāna texts (Howard 1986: 56; Miyaji 2002: 147-151); Vairocana in the *Buddhāvataṃsaka-sūtra (Giès & Cohen 1996: 341-344); or a generic buddha (Fussman 1999: 548-551; Rhi 2008a; 2011). Among these above identifications, there are certain common views that the palace-type represents a buddha-field (Skt. Buddhakṣetra) (Fussman 1999: 548-551; Rhi 2008a; 2011; Harrison & Luczanits 2011) as an embodiment of a buddha (Miyaji 2002: 143, 153; Harrison & Luczanits 2011), whether certain or generic, or a theophany (Rosenfield 1967: 235-238; Rhi 1991, 148; 2003: 174-175; 2006, 171) or a visionary experience (Luczanits 2008: 47-49). As is evident from an apparent lack of scholarly consensus, the identification of the late Gandhāran preaching buddha remains highly obscure and therefore existing scholarship on this subject needs to be reviewed in order to identify the Devnīmorī buddha. Methodological problems The above history of the exceptionally large body of existing research on identifying the late Gandhāran preaching buddha, with no convincing result, urges us to review the methodology. Recent interdisciplinary iconographic projects between art history and Buddhology by Miyaji (2002: 144), Harrison and Luczanits (2011: 115), and Rhi (2018: 255-6) have shown that text-image parallels cannot be drawn convincingly, and that references to Mahāyānist literature cannot help to identify the late Gandhāran preaching buddha with any particular buddha in his buddha-field. Nonetheless, even with no definitive text-image parallels drawn, but only partial associations, Harrison and Luczanits were inclined to hypothesize that the late Gandhāran preaching buddha was Amitābha attended by Avalokiteśvara and Mahāsthānaprāpta. Their justification involved intentionally compromising on the crucial discrepancy in the identification of one of the individual bodhisattvas as Maitreya (Harrison & Luczanits 2011: 115-116, 118), as clearly demonstrated by Miyaji (Miyaji 1985), which contradicts the iconographic programme of the Amitābha triad. They supported such a hypothetical identification by arguing the case for the abstraction of the individual identities of bodhisattvas into a symbolic meaning or meanings of the triad as a whole, which is the permanence of the dharma as clearly underlined in the Larger Sukhāvatīvyūha (Harrison & Luczanits 2011: 115). Similarly, Miyaji’s identification of the triad as Śākyamuni flanked by Avalokiteśvara and Maitreya is not textually attested in the corresponding Mahāyāna context; that is to say, there is no attestation of Śākyamuni/Avalokiteśvara/Maitreya in Mahāyānist materials thought to be contemporaneous with the production of these images, but only in the Vajrāsanasādhana, a still Mahāyānist (but from a slightly later period in the development of Indian Mahāyāna) but also tantric text in the Sādhanamālā (Miyaji 2002: 114). He then admitted common disagreements between artistic and textual representations, which are recognizable in Buddhist art and iconography between the second century BC and the sixth century AD (Miyaji 2002: 128), rather than justifying the time lag. 182 ken ishikaWa: more Gandhāra Than maThurā Consequently, there is a clear scholarly trend towards accepting the gap between textual and visual images but extracting underlining symbolic theological meanings. Such a shift is even seen in the latest paper by Rhi on the subject, which takes a very conservative view on any identifications of the late Gandhāran preaching buddha with his flanking bodhisattvas, and considers Gandhāran Buddhist art as highly conceptual (Rhi 2018: 256-7). There seems to be a shift of methodological focus from the study of the iconography to that of general symbolism in respect to the late Gandhāran preaching buddha, but this scholarly trend seems to have come to a standstill. The visionary/psychedelic experience of samādhi In order to go beyond iconography and symbolism, as typically discussed in the field, a fundamentally different approach needs to be undertaken to advance the study of the late Gandhāran preaching buddha. In doing so, I will highlight a socio-ritual role of the Gandhāran preaching buddha in actual practice in the contemporaneous Buddhist community, in particular, in the context of the practice of visualization. To date, only the narrative aspect of the visionary experience of the late Gandhāran preaching buddha has been discussed. but not with reference to the actual practice of visualization. The elaborate steles with complex compositions depicting this buddha have been recognized as a theophany but also as a cosmological vision of infinite lotuses, buddhas, bodhisattvas, and buddha-fields attained through samādhi (‘concentration’); the depiction has been described by art historians in such psycho-spiritual terms as ‘visionary’ (Luczanits 2008: 47-49) and ‘mystical’ (Miyaji 2002: 148). Like the art historians, the Buddhologist Harrison argued that the elaborate descriptions of other worlds i.e. buddha-fields, as found in the Larger Sukhāvatīvyūha and several other Mahāyāna texts, are closely associated with the parallel development of the concept of samādhi and the practice of visualization in the early Mahāyāna Buddhism. In turn, accounts of visualization practice can help explain the background to the emergence of ‘Pure Land’ literature, such as versions of the Sukhāvatīvyūha (Harrison 2003: 120-128). In fact, these visionary aspects of the elaborate textual descriptions of the Mahāyānist cosmology in the Gaṇḍavyūha-sūtra have been understood as reflecting actual visualization practice and been described accordingly as ‘entopic’ (Osto 2018) or ‘fractal’ or ’psychedelic’ (Fox 2015: 263). In cognitive anthropology, however, the shamanic, entopic vision denotes the residual image effect of an abstract nature (Lewis-Williams 2002), and therefore the inherently eidetic vision of samādhi can be better explained as psychedelic or visionary. In this connection, Osto (2018: 1880190) analysed a specific, early Mahāyānist visualization practice called pratyutpanna-samādhi, in which a practitioner visualizes any of myriad/infinite buddhas of the present, in the socio-ritual context. Pratyutpanna-samādhi is considered as an adaptation of the earlier practice of buddhānusmṛti (‘commemoration of the Buddha’) in mainstream Buddhism and involves prolonged visualization of a buddha in his buddha-field (Harrison 2003: 120). The idea is that through mentally constructing an image of a buddha in a buddha-field according to specific guidelines prescribed, a meditator captures a vision of a buddha either in a waking or dreaming state that assures him of reaching this very buddha-field (Harrison 2003: 120). Pratyutpanna-samādhi is highly relevant to discussion of the late Gandhāran preaching buddha, owing to its association with Pure Land Buddhism. The early Mahāyāna pratyutpanna-samādhi meditation/ visualization text of the Pratyutpanna-saṃmukhāvasthita-samādhi-sūtra has survived in its entirety only 183 The Global ConneCTions of Gandhāran arT as Chinese and Tibetan translations, while there are fragments in Sanskrit and Gāndhārī (Harrison, Lenz & Salomon 2018: 117-120). The earliest extant translation the Banzhou sanmei jing 般舟三昧經 (T.416-419) is attributed to Lokakṣema in AD 179 (Harrison 1998: 1-2; Harrison, Lenz & Salomon 2018: 118-119). A recently identified group of fragments of a manuscript of the Pratyutpanna-saṃmukhāvasthita-samādhi-sūtra in Gāndhārī language and Kharoṣṭhī script is palaeographically dated to the first or second century AD, but radiocarbon dates of its related Gāndhārī manuscripts suggest an even earlier date of the first century BC (Harrison, Lenz & Salomon 2018: 121-123). The very early Buddhist practice of buddhānusmṛti adopted in the Pratyutpanna-saṃmukhāvasthita-samādhi-sūtra, dates back even further (Harrison 1978; Williams 2009: 209-212). According to Harrison, this work is considered to be ‘a work of Pure Land Buddhism’ if not its pure product and contains the earliest datable reference to Amitābha (Harrison 1998: 2-3). In fact, the object of visualization meant therein can be any buddha or buddhas of the present in any direction to any buddha-field and Amitābha is ‘merely adduced as an example’ (Harrison 1978: 43-44; 1998: 2-3). Therefore, regardless of any possible certain identities of the late Gandhāran preaching buddha, the study of pratyutpanna-samādhi provides an understanding of a wider socio-ritual context of the iconography, which is also useful for the analysis of the Devnīmorī buddha images. Osto (2018) argued that altered states of consciousness of samādhi, by analogy with psychedelic experiences, may have been induced by such ‘mind-altering techniques’ as ‘fasting, sensory and sleep deprivation, intense concentration, visualisation meditation and hypnosis’, with the aid of the ‘set and settings’, described in the Pratyutpanna-saṃmukhāvasthita-samādhi-sūtra (Osto 2018: 188-190). The resulting vision of samādhi is accordingly described as ‘dream-like’ or ‘mind-only’ state, but not as a physical experience by means of superhuman powers (Harrison 1978: 46). Harrison argues that these visualized mental images are ‘no mere hallucinations’ but doctrinally a manifestation of the Mahāyānist concept of śūnyatā or emptiness lacking their intrinsic material nature (Harrison 1998: 2-3; Williams 2009: 212-213). The question then arises as to whether the depiction of the late Gandhāran preaching buddha as a vision of samādhi was merely to recount this visionary experience or if the imagery had a socio-ritual function in the actual visualization practice. The use of buddha images in the practice of visualization and the origins of Buddha images One crucial point that Osto (2018) did not discuss regarding the set and the settings in the visualization practice of pratyutpanna-samādhi described in the Pratyutpanna-saṃmukhāvasthita-samādhi-sūtra is the references therein to the making of (physical) images of a buddha, either as a sculpture or a painting (Ominami 1975: 734; Harrison 1978: 38-39). The consequent possibility, as argued by Harrison, is that these buddha images may have been used as aids for visualization (Harrison 1978: 38-39: Harrison 1998: 1-2), which are of direct relevance to the study of the ritual role of the late Gandhāran and Devnīmorī buddhas images. Meanwhile, the idea of samādhi and the practice of visualization add an important perspective to continuous discussion of the origins of the Buddha/Bodhisattva images in South Asia. The first ever sculpted or drawn/painted buddha image may have been made for the particular purpose of visualization, and thus supposed to be conceptually identical with a ‘mental image’ (Harrison 1978: 3839). Harrison points out that the transition from aniconism to anthropomorphism certainly took place by the beginning of the second century AD, as attested by the date of the Banzhou sanmei jing, which itself makes references to buddha images (Harrison 1978: 38-39). 184 ken ishikaWa: more Gandhāra Than maThurā To put the practice of visualization into context, Osto’s (2018) methodology of connecting the Pratyutpannasaṃmukhāvasthita-samādhi-sūtra with the Gaṇḍavyūha-sūtra, which is generally dated to AD 200-300 (Osto 2004: 60), shows a slight chronological gap. He then even misses a crucial transition between the visualization of a buddha and the seeing of a buddha in the sense of darśana, a ritual act of seeing. This developed towards c. AD 400, as clearly observed, for instance, in the Guan Fo Sanmei Hai Jing 觀佛三昧海經 (The Sūtra on Ocean-Like Samādhi of the Visualization of the Buddha), the earliest of so-called visualization texts that survive in Chinse translations, with their typical emphasis on visual imagery. There the visualization of a buddha is considered as a skilful means for the ‘seeing’ a buddha (Ominami 1975: 735). The Guan Fo Sanmei Hai Jing, which is generally dated to the fifth century AD, explains three successive stages of visualizing, recollecting and ‘seeing’ buddhas (Ominami 1975: 735). Although the attribution of the Chinese translation of the Guan Fo Sanmei Hai Jing to Buddhabhadra (AD 359-429) has been disputed (Ogasawara 2019: 194; Yamabe 2019: 397), some suggest its Indian origin (Yamabe 2019: 418). Therefore, the practice of visualization continued into the fourth and fifth centuries AD, which is our study-period for the late Gandhāran preaching buddha and Devnīmorī meditating buddha images. In this connection, Miyaji notices the general transition from narratives to cult images with reference to the late Gandhāran preaching buddha as well as Gupta buddhas as a relatively late development (Miyaji 1993: 425-426). However, the earliest buddha and/or bodhisattva images in Swat and Mathurā already have the nature of cult images in a triad with either Brahmā and Indra or a pair of whiskbearers, respectively, just like later triadic compositions with a pair of bodhisattvas, with narrative elements either stripped or simplified. Therefore, consideration of the possible non-narrative, socioritual function of buddha images for visualization in the first half of the first millennium AD in South Asia needs to be incorporated into discussion of the origins of the Buddha images. It is therefore highly likely that the earliest Buddha images from the latter half of the first century AD to the early second century AD already had different functions for exchange-oriented ritual by worshippers (Karashima 2013: 181-184) and for visualization by meditators (Harrison 1978: 38). Superhuman powers and supernatural miracles of the Buddha Equally relevant to discussion of late Gandhāran and Devnīmorī buddhas is the notion of yogic superhuman powers (eg. Sanskrit abhijñā: Pāli abhiññā) and supernatural miracles (e.g. Sanskrit prātihārya: Pāli pāṭihāriya) caused by buddhas. The Muhammad Nari stele (Figure 26) has long been considered to depict a miracle scene of multiplication (Foucher 1905) or emanation (Harrison & Luczanits 2011) or the emission of light (Miyaji 1993). Such supernatural miracles can be categorized into ṛddhiprātihārya in Sanskrit or iddhi-pāṭihāriya in Pāli (‘supernatural miracles’) in the traditional list of the three types of pāṭihāriyas (‘miracles’) (Goshima 2015: 1). The category of iddhi-pāṭihāriya corresponds to that of iddhi-vidhā-ñāṇa ‘the wisdom of diverse supernatural powers’) in the traditional list of six kinds of abhiññās (‘wisdoms’) (Clough 2012: 77). In these lists, both iddhi-pāṭihāriya and iddhi-vidhā-ñāṇa are treated as mundane and even achievable by non-Buddhists. In contrast, the third pāṭihāriya and the sixth abhiññā are elevated as transmundane i.e. profound, clearly distinguished from the rest and reserved only for Buddhism. As such, superhuman powers are also said to have been viewed negatively by the Buddha himself and displaying such powers before householders was prohibited by him (Goshima 2015: 2). With such conservatism, the term pāṭihāriya meaning ‘miracle’ started to possess a connotation of ‘indoctrination’ or ‘instruction’ (but still through superhuman abilities and supernatural miracles) from early on (Goshima 2015: 30). In both mainstream and Mahāyāna forms of Buddhism, the use of supernatural miracles had to be justified by the purpose of conversion of non-Buddhists: Fiordalis argues 185 The Global ConneCTions of Gandhāran arT that this aspect can be regarded as a ‘religious’ role (Fiordalis 2012: 122). All this also suggests that the iconography of the late Gandhāran preaching buddha may have a double meaning of his miracle and his teaching and may also have been useful for proselytizing Mahāyāna Buddhism in Gandhāra. Yet all these ambivalent attitudes paved the way for some major multiplication/emanation miracle stories in Mahāyāna Buddhism by the first or second century AD (e.g. in the Smaller and Larger Sukhāvatīvyūhas and the Saddharmpuṇḍarīka) (Miyaji 2002; Harrison & Luczanits 2011). In mainstream Buddhism, the multiplication aspect of the Great Miracle at Śrāvastī was added only at a later stage in the Divyāvadāna to the pre-existing popular miracle story of the mango tree, due possibly to such earlier Mahāyāna influence: this development was also seen in art at Sārnāth and Ajaṇṭā in the late fifth century AD (Miyaji 2002). Likewise, in Gandhāra, the Great Miracle at Śrāvastī was represented only as the so-called Twin Miracle (Miyaji 2006). Miracles of light Miyaji argued that the Muhammad Nari stele (Figure 26) depicts Śākyamuni, who emits light from his ūrṇā: his analysis shows that visual points of astonished bodhisattvas surrounding him all focus on the ūrṇā (Miyaji 1993). The scene was identified by him with ‘the miracle of great light (大光明)’ in the prologue of the Chinese translation of the Saddharmpuṇḍarīka by Kumārajīva, dated to AD 406 (Karashima 2015: 166). His argument based on the internal visual evidence is significant in exploring the broader conceptual meaning of light as a supernatural miracle through his superhuman power. In the Saddharmpuṇḍarīka (Vaidya 1960), at Rājagṛha, having taught the mahānirdeśa (‘great instruction’), the blessed one (bhagavat) with his disciples was seated cross-legged (paryaṅka) on a throne of the great dharma (mahādharmāsana) and went into a samādhi called anantanirdeśapratiṣṭhāna (‘the abode of infinite instructions’). Then there fell over them a rain of divine flowers and it trembled the buddha-field in six ways, and a diverse assembly looked at him in amazement. (2, 1) He then emitted light from his ūrṇā: illuminated and revealed 18,000 buddha-fields in the eastern quarter with all beings, monks, nuns, lay Buddhists, yogis, yoga practitioners, bodhisattvas the mahāsattvas, buddhas, the blessed ones, and bejewelled stūpas of past buddhas (3, 1). Most importantly, Mañjuśrī explains to Maitreya the meaning of the scene, namely that the miracle of light anticipates the tathāgata’s teaching of the great dharma (11, 1), i.e. the saddharmpuṇḍarīka (‘the white lotus of the sublime dharma’) (13, 1). In the Nidānaparivarta prologue of the Saddharmpuṇḍarīka in Sanskrit, datable to c. AD 100 (Karashima 2015: 163), the light emitted by the blessed one is simply called ‘miracle-light’ (prātihāryāvabhāsa) (3, 1) rather than ‘great light (大光明)’. This light in fact consists of a single ray of light (raśmi) which is omitted from his ūrṇā (‘hair’) on the forehead in the Sanskrit original (3, 1). This single ray of light emitted from the ūrṇā is conceptually similar to the divine eye as light (āloka) but also as sight (āloka), as described as one category of superhuman powers (abhiññā in Pāli) in the Visudhimagga: the divine eye can see as far as light can reach with its all-pervading light and all-seeing sight (Fiordalis 2011: 108). Miyaji also makes references to other Mahāyāna texts that include not identical but broadly similar descriptions of miracles of light (Miyaji 1993), but he does not give details of their differences in nuance nor his view on the chronological development of such an idea. In this respect, his comparison of the Muhammad Nari stele from the third to fourth century AD with the prologue of the Saddharmpuṇḍarīka from c. AD 100 generates a significant time lag despite its long-term influence. 186 ken ishikaWa: more Gandhāra Than maThurā Such a chronological gap can be filled by looking at a more developed form of the story of the miracle of light in Mahāyāna Buddhism towards the third century AD, around the time of the late Gandhāran preaching buddha of the third to fourth century AD. In particular, the Larger Prajñāpāramitā (‘The Transcendence of Wisdom’) material, also known as ‘the light-emission group’, that characteristically includes miracle stories of light, exemplifies the chronological phase of c. AD 150-250 (Katsusaki 2015: 31). There also is growing evidence that the Prajñāpāramitā scripture originates in Gandhāra, in the Gāndhārī language, at least in its earliest datable phase of AD c. 50-150, despite the apparent superiority of the Prajñāpāramitā scripture as a physical object of worship over Buddha images in this phase (Karashima 2013). In fact, as Okada has argued, the story of the miracle of light (raśmyavabhāsa) in the Larger Prajñāpāramitā shows striking parallels with that in the Saddharmpuṇḍarīka but constitutes a later chronological development: according to him, the author of the Pañcaviṃśatisāhasrikā-prajñāpāramitā was aware of the content of the Saddharmpuṇḍarīka (Okada 2017: 101). Therefore, this related but later version of the miracle of light could potentially be chronologically more appropriate for discussion of the Muhammad Nari stele. On this matter, the prologue of the partially preserved version of the Gilgit manuscripts of the Aṣṭādaśasāhasrikā Prajñāpāramitā in Sanskrit (Conze 1962; Zacchetti 2005) is particularly pertinent, both chronologically and geographically, to late Gandhāra. This collection contains some descriptions useful for analysis of the late Gandhāran preaching buddha, such as miracles of light, emanations, and a palace, which will be discussed below. Miracle stories of light can be found throughout the Aṣṭādaśasāhasrikā Prajñāpāramitā but especially at the beginning of the prologue. The blessed one (bhagavat), seated on a siṃhāsana (‘lion-throne’), first entered into a samādhi called samādhi-rāja (‘the king of concentration’) and looked at a buddha-field through his divine eye. He then issued ‘sixty hundred thousand niyutas of kotis’ of rays of light from each of his body parts and hair pores. These rays then illuminated the trichiliocosm and other world-systems ‘as numerous as the sand of the Ganges’ in all the ten directions, and ever being exposed to the light ‘becomes fixed in supreme perfect awakening’ (anuttarasamyaksaṃbodhi) (LPG 2r). The text then describes a miracle of emanations (vigraha). The blessed one put out his tongue, covered the entire trichiliocosm with it, smiled, and then his tongue emitted ‘many hundreds of thousands of niyutas of kotis’ of rays of light. From each of these rays arose a golden, bejewelled and thousands-petalled lotus with an emanation of the standing and seated tathāgata on it. These emanated tathāgatas then instructed the dharma i.e. the six pāramitās (‘perfections’) in the world-systems in the ten directions. All the beings who had heard the dharma became ‘fixed in supreme perfect awakening’ (LPG 3r). Furthermore, there is a description of a miracle of palace. The blessed one then entered into another samādhi called siṃhavikrīḍita (‘lion’s play’) again on a siṃhāsana and then trembled the trichiliocosm and world-systems in the ten directions in six ways. All the humans and devas looked at the tathāgata (‘thus-come/thus-gone’), rejoiced, and worshipped him, and then offered him divine flowers, garlands, incenses, ointments, powders, cloths, flowers, filaments, the bark, leaves, decorations, umbrellas, flags, banners. These offerings were then transformed into a summit-palace (kūṭāgara) as immense as the trichiliocosm through his superhuman power (adhiṣṭhāna) (LPG 3r-5r). These miracle scenes in the Aṣṭādaśasāhasrikā Prajñāpāramitā, discussed above, undoubtedly show many parallels with the Saddharmpuṇḍarīka, but with further elaboration. It appears that by the mid-third century AD, most of the miraculous elements that are present in the Muhammad Nari stele such as light, emanations, and palace should have been common enough to anticipate the Gandhāran preaching buddha in the third to fourth century AD, if the text came earlier than the image. 187 The Global ConneCTions of Gandhāran arT Yet these miracles in the Aṣṭādaśasāhasrikā Prajñāpāramitā again typically anticipate the teaching of Śākyamuni as seen in the Saddharmpuṇḍarīka (Okada 2017: 100): one tathāgata Ratnakāra explains that Śākyamuni, who is tathāgata, arhat, and samyaksaṃbuddha, ‘stands, abides, and remains’ in the worldsystem of Sahā in the western direction and teaches prajñāpāramitās to bodhisatvas the mahāsatvas (LPG 5v). The comparison between the above two prologues certainly implies the existence of a common generic model of a story that can take any form with a broadly similar base plot but with further modifications and elaborations. This character applies to many elements, such as differences in thrones, samādhis, entities, assemblies, bodhisattvas, types of the dharma and so on, just like all the similar life stories of past buddhas, but with different names, under different bodhi trees, from different kalpas etc. Meanwhile, consistent emphasis on the real-life benefits of the exposure to the light issued from the blessed one shows an additional soteriological aspect, which was lacking in the Saddharmpuṇḍarīka (Okada 2017: 100) but is present rather in earlier material of Pure Land Buddhism (Amitābha literally means ‘infinite light’) (Ishida 1997: 11). In the Aṣṭādaśasāhasrikā Prajñāpāramitā, all the humans who have been exposed to the light attain anuttarasamyaksaṃbodhi, which is buddhahood in Mahāyāna Buddhism. This soteriological aspect can be seen as a confluence of different Mahāyāna ideas. The description of such instant awakening may also explain a lack of the depiction of the diverse assembly in the Muhammad Nari stele, which is questioned by Harrison and Luczanits. They deny Miyaji’s identification of it as the depiction of the miracle of light in the Saddharmpuṇḍarīka owing to the absence of the depiction of śrāvakas (Harrison and Luczanits 2011: 114). In fact, it was not only śrāvakas who were missing but the entire diverse assembly, described above. If all those in the diverse assembly who had been exposed to the light and heard the teaching of the dharma attained anuttarasamyaksaṃbodhi or ‘supreme perfect awakening’, they should all look like bodhisattvas. Therefore, this may explain the predominant presence of bodhisattvas in the Muhammad Nari stele. However, there are still obvious discrepancies, as should be expected. For instance, the siṃhāsana of the blessed one, described in the Aṣṭādaśasāhasrikā Prajñāpāramitā, is absent in Muhammad Nari stele. In the text he is not on a lotus but on a lion-throne (siṃhāsana). However, as mentioned earlier, according to the Gilgit manuscripts of the Saddharmpuṇḍarīka, a siṃhāsana is considered to be located in a calyx of a lotus (Schopen 1977: 182) and thus it could have been technically challenging to depict a lotus and a siṃhāsana in one image, which only appears in India proper from the Gupra period onwards. Another important text belonging to the period immediately preceding the time of the late Gandhāran preaching buddha is the aforementioned Gaṇḍavyūha, whose principal buddha is called Vairocana: such proto-Vairocana (‘one who shines forth) is comparable to the blessed one in the Aṣṭādaśasāhasrikā Prajñāpāramitā, who is frequently described as shining forth (virocate) (vi√ruc). In-depth comparative studies of related Mahāyāna texts belonging to this chronological period may enhance our understanding of the concept of the buddha behind the late Gandhāran buddha image. Different ideas of buddha-bodies Discussion of supernatural miracles and superhuman powers leads us to different ideas of bodies of buddhas that developed in Buddhism over time, but especially towards the time of the late Gandhāran and Devnīmorī buddhas. In the Mahāyāna context, the initial distinction and contrast between the two buddha-bodies of dharmakāya (dharma body) and rūpakāya (form body) developed into tripartite divisions of trikāya (three bodies), i.e. dharmakāya plus two forms of rūpakāya that are saṃbhogakāya (enjoyment/bliss body) and nirmāṇakāya (emanation/transformation body) towards the fourth century 188 ken ishikaWa: more Gandhāra Than maThurā AD (Williams 2009: 177, 179). In trikāya, particularly relevant to the late Gandhāran preaching buddha is the saṃbhogakāya (or saṃbhogikāya), which is the blissful and luminous buddha-body in the mainstream Yogācāra doctrine of trikāya or the three bodies as described in the Mahāyānasūtrālaṃkāra, which possibly dates to the fourth century AD, and then into its later tantric manifestation (Tucci 1930; Hakamaya 1986; Griffiths 1990: 111-112, n. 12). The saṃbhogakāya is a buddha-body of bliss and light, which entails visionary experiences, enjoyed by a buddha, as well as by other advanced bodhisattvas, by attaining buddhahood through self-identification with a buddha (Williams 2009: 181). It is also the ‘glorified’ body of a buddha with his physical major and minor characteristics (mahāpuruṣalakṣaṇa), seated on a lotus and preaching the Mahāyāna to his assembly of bodhisattvas in an akaniṣṭha heaven until the end of saṃsāra (Williams 1989: 180-181). Such embodiment shows some similarities with the earlier Theravādin idea of a mind-made body (manomayakāya) explained in such scholastic path manuals as the Paṭisambhidāmagga, which roughly dates to the second century BC, and the Vimuttimagga , which possibly dates to the first century AD; manomaya-kāya is considered as a ‘hollow’ body permeated with the bliss (sukhatā) and lightness (lahutā) in order to enter into the fourth jhāna and cultivate supernatural powers categorised as iddhis and abhiññās (Clough 2011: 82-83). The concept of saṃbhogakāya clearly demonstrates the maturity of visionary, experiential, and magical aspects of non-ontological Buddhist cosmology that developed in Mahāyāna Buddhism, which was later inherited by Vairocana, whose body is the saṃbhogakāya in the *Buddhāvataṃsaka-sūtra, whose first comprehensive Chinese translation dates to AD 418-421 (William 2009: 132-138, 175). Huntington alludes to the saṃbhogakāya of Vairocana in relation to the Muhammad Nari stele (Huntington 1980: 659) (Figure 26). Existing scholarship tells us further that much earlier than the Mahāyānist idea of the three bodies, the polemic against the conventional relic worship gave a rise to different ideas of buddha-bodies or embodiments of buddhas – vajrakāya and tathāgatagarbha are innovations associated with the second century AD, even if only appearing in Chinese translations a century or two later. in light of the dating of the Mahāparinirvāṇamahāsūtra by Radich (Radich 2012; Radich 2015; Jones 2016). Radich has explored the historical development of the idea of immortality that was initially ascribed to cosmically distant buddhas, but by the fourth century AD applied to Śākyamuni embodied as a permanent and indestructible buddha-body of adamant known as vajrakāya in contemporary with the trikāya doctrine. According to Radich, the idea that the Buddha has a vajrakāya comes shortly before tathāgatagarbha (Radich 2015: 171) and emerged out of the Mahāyānist reinterpretation of the dharmakāya as the selfidentity of the Buddha with the absolute (dharmatā, dharmadhātu, tathatā; Radich 2012: 272-273). His research has also established a close link between the buddha-body of adamant and mental states of adamant, which is most manifest in the samādhi, as frequently seen in certain Mahāyāna texts of around AD 400 (Radich 2012: 274-280). This reinforces the above discussion of the relationship between samādhi and embodiments of buddhas. Another relevant point made by Radich concerns the personhood of the buddha of the vajrakāya in relation to stūpa/relic worship, namely that the indestructible relics of the Buddha contained in stūpas, which were venerated collectively as a dhātu (‘element’) but not conventionally as relics or śarīrāṇi (‘body’), attributed buddha-nature or what later came to be known as tathāgatagarbha internally to sentient beings rather than externally to a stūpa (Radich 2012: 280-281). In light of the revised chronology of tathāgatagarbha texts by Radich (2012; 2015) and Jones (2016), it is apparent that there was no single idea of buddha-bodies that was universally accepted in South Asia 189 The Global ConneCTions of Gandhāran arT at the time of the late Gandhāran preaching buddha and Devnīmorī buddha images towards the fourth century AD in South Asia, though sectarian affiliations or geographical origins of different Mahāyānic doctrines of buddha-bodies and embodiments require further Buddhological inquiry. In response to the work of Radich (2012), we may need to decode underlying meanings of buddha-bodies and embodiments, for instance, the permanence and immortality of cosmically distant buddhas or Śākyamuni. In relation to the mahāstūpa at Devnīmorī, its relic casket inscription refers to the relics contained therein as śarīrāṇi (‘śarīra’ in the compound daśabala-śarīra-nilaya or ‘a receptacle of relics of the daśabalaʼ [‘the one who is endowed with ten (superhuman) powers’] – possibly an epithet of the Buddha but not exclusively (Srinivasan 1968: 69) – instead of a dhātu. Since śarīra and dhātu do not represent very different ideas of ‘relics’ at stūpas, it is likely that the mahāstūpa was a conventional stūpa with relics (śarīrāṇi) embodying the presence the Buddha or a buddha as a person while simultaneously manifesting him as the dharma as the dharmakāya (Boucher 1991: 1516, 27). A similar form of the embodiment of the Buddha is also attested at the site level at Devnīmorī by the excavated steatite seal of the so-called ye dharmā hetuprabhavā ‘creed’ (Mehta & Chawdhary 1966: 122, fig. 36). This creed is known to establish stūpas and the presence of the Buddha just as relics do (Boucher 1991: 4; Ghosh 1967; Hinüber 1985 ; Strong 2004: 10). The changing meaning of the dharmacakramudrā The significance of the late Gandhāran preaching buddha showing the dharmacakramudrā (Figure 26), which means ‘the hand gesture of the wheel of the dharma’, including a variety of forms that developed over time within South Asia and across Buddhist Asia. Saunders explains its Gandhāran variant in these terms: ‘the right hand, with the fingers rather close together and the palm turned inward, loosely envelopes the joined ends of the thumb and index of the left hand: the other fingers are negligently closed’ (Saunders 1960: 94, 231-232). This mudrā can otherwise be contextualized in the post-Gandhāran development of this mudrā, inherited from Gandhāra, at Sārnāth and Ajaṇṭā, with a clear association with Śākyamuni’s first sermon at Sārnāth (Nakanishi 2013). The dharmacakramudrā, first emerged in late Gandhāra and became the most popular of all mudrās during the Gupta period. The Gandhāran version of the dharmacakramudrā, as shown by the late Gandhāran preaching buddha, which is the earliest of its kind, has been noted as a variant, according to the criteria of the standard Sārnāth model; ‘the right hand with gathered fingers and the palm turned inward, loosely enveloping the joined ends of the thumb and index finger of the left hand, whose other fingers are negligently closed’ (Sounders 1960: 94). In Gandhāran narrative art, the Buddha’s first sermon was shown in the narrative with different mudrās in close association with a dharmacakra and a pair of deer (Zwalf 1996: cat. no. 199; vol. 1, 181-183; vol. 2, 121; Huntington n.d.: 11-12), that indicates the location of Sārnāth (Nakanishi 2013). The Buddha’s sermon is otherwise represented with a dharmacakra without the deer (Zwalf 1996: cat. nos. 145 and 200; vol. 1, 183; vol. 2, 122), which possibly symbolizes some other teaching. Despite its typological importance, the relationship between the late Gandhāran and Sārnāth versions of the dharmacakramudrā has not been discussed adequately. In particular, the dharmacakramudrā shown by the late Gandhāran preaching buddha may be distinguished from the Buddha’s first sermon at Sārnāth because of a lack of specific iconographic references to Sārnāth such as a pair of deer. According to the Saddharmapuṇḍarīka, there were two distinct types of dharmacakras in early Mahāyāna Buddhism: while one of them is the conventional dharmacakra i.e. the Śākyamuni’s first sermon at Sārnāth, the other is the foremost and maximum dharmacakra, reinterpreted as the ‘true’ teaching of the Buddha, i.e. the ekayāna (‘one vehicle’) rather than the triyāna (‘three vehicles’), which was previously taught only 190 ken ishikaWa: more Gandhāra Than maThurā provisionally by Śākyamuni at Sārnāth (Fujichika 1995: 705-706). In the Chinese Mahāyāna context, towards the beginning of the fifth century AD, the sermon of the Buddha at Sārnāth was considered as the second sermon while the first was at Akaniṣṭha (Chappell 1983; Huntington 2000: 37). It is therefore likely that the dharmacakramudrā shown by the late Gandhāran preaching buddha symbolized the Mahāyānist dharmacakra taught by Śākyamuni with his transcendental Mahāyānist identity or by any other Mahāyānist buddha. In addition, the Saddharmapuṇḍarīka also refers to turning of the dharmacakra by other buddhas (Fujichika 1995: 706) just like all accounts of past buddhas. Therefore, the dharmacakramudrā may symbolize any teaching of the dharma by any buddha or advanced bodhisattva but often with its historic reference to the Buddha’s first sermon at Sārnāth. The late Gandhāran preaching buddha and Devnīmorī meditating buddhas Turning back to the late Gandhāran preaching buddha, we have seen that archaeologists have struggled to identify him conclusively either as a specific buddha or as a generic figure, but this ‘unknowable’ nature of the concept of ‘buddha’ is in fact nothing new; indeed it is an ancient one in the Mahāyānist discourse, in which the question of whether there is one Buddha or many buddhas was debated (Williams 2009: 180). The notion of the multiplication of the Buddha is as old as the cult of the Seven Buddhas of the Past dating back at least to the mid-third century BC at the time of Aśoka, who doubled the size of the stūpa dedicated to Konākamana (Koṇāgamana in Pāli and Kanakamuni in Sanskrit), one of the Seven Buddhas of the Past (Gombrich 1980: 67) at Nigali Sagar. From the Buddhist perspective of the absolute, no buddhas are intrinsically indistinguishable from one another (Williams 2009: 180; Radich 2012: 273), which implies that any buddha can otherwise be self-identified with Śākyamuni. In this respect, Strong argues that relics of past buddhas were venerated exclusively in association with Śākyamuni as the Buddha of the present (Strong 2004: 49). Strong’s view also resonates with an account of a stūpa, described in the eleventh chapter Stūpasaṃdarśana (‘manifestation of a stūpa’) of the Saddharmapuṇḍarīka in the Mahāyāna context: a great bejewelled stūpa that enshrined a body of a past buddha named Prabhūtaratna became personified as a person (Prabhūtaratna) and conversed with Śākyamuni (Karashima 2018: 472-473). The cult of past Buddhas was certainly popular in north Gujarat as attested by a fragment of a relief depicting four buddhas in the natural rock-shelter of Jogida in the Taranga Hill (Rawat 2009: 20/pl. 2, 97; Rawat 2011: 231): this sculpture possibly belongs to the Maitraka period and would have represented the Seven Buddhas of the Past under their respective trees, as most commonly depicted in the Western Deccan. Likewise, the presence of eight terracotta buddha images interned into the mahāstūpa at Devnīmorī as relics suggests the cult of the Seven Buddhas of the Past plus Maitreya Buddha in north Gujarat, as discussed earlier. The relics in the inscribed reliquary excavated from the core of the mahāstūpa at Devnīmorī are referred to as belonging to a the daśabala (‘the one who is endowed with ten [superhuman] powers’) (Srinivasan 1968: 68-69), who may well be Śākyamuni or some past buddha. As Strong argued, Śākyamuni and the Seven Buddhas of the Past would have been venerated in conjunction (Strong 2004: 49), and this would also have been the case in north Gujarat. In this connection, Karashima identified the rock engraving of a stūpa topped with a high pole with multiple disks accompanied by a pair of stūpa-like buddhas at Hodar in Gilgit in Pakistan (Hauptmann 2008: 353) as Śākyamuni conversing with past buddha Prabhūtaratna in the aforementioned Stūpasaṃdarśana chapter of the Saddharmapuṇḍarīka (Ishida 1997: 9; Karashima 2018: 476). This depiction resembles two similar representations of a stūpa accompanied by a pair of smaller stūpas (but without anthropomorphic 191 The Global ConneCTions of Gandhāran arT buddhas) on the rock painting in Shelter I in the group of undated stūpa rock-paintings near Gambhirpura (Sonawane 2013) and on the ivory seal excavated at Vadnagar, palaeographically dated to as early as the third to second century BC (Rawat 2011: 219, 221/fig.11.6), both from north Gujarat itself. After all, the identity of Devnīmorī buddha images, which have some iconographic features of the late Gandhāran preaching buddha such as the wavy hairstyle, the inner robe and the lotus throne, is unclear. This is owing especially to the dhyānamudrā shown instead of the Gandhāran dharmacakramudrā. The comparative analysis of Devnīmorī and Mīrpur Khās buddhas clearly concerns the invariable presence of the dhyānamudrā shown by them as a hand gesture of meditation. I have shown above the importance of the state of samādhi in Mahāyāna Buddhism with reference to the late Gandhāran preaching buddha and thus it is probable that deep attention was paid to this mudrā in association with the samādhi at regional levels in north Gujarat and Sindh. The dhyānamudrā including its variants was first shown in Swat, by Śākyamuni venerated by Brahmā and Indra (Huntington n.d.), depicted on some of the earliest Buddha cult images ever created in South Asia (Miyaji 2005). In Gandhāra proper, the dhyānamudrā was associated with meditation and superhuman powers, most significantly, in the narrative of Indra’s visit of Śākyamuni meditating inside Indrasālaguhā (Inrasālaguhā in Pāli) on the Vediya mountain. According to Miyaji, one group of frontal meditating images of the Buddha in a cave, in a landscape context with the Vediya mountain, depicts flames around the opening of Indrasālaguhā indicative of the flame-samādhi (huoyan zanmai 火焔三 味) of the Buddha described in one of the different versions of the same story in the Chang ahan jing 長阿含經 (the *Dīrghāgama) (T.0001:01.0562c12-13; Shichi 1987; Miyaji 2010). The dhyānamudrā is also shown in Gandhāran narratives by another flame-samādhi at Urubilvā (Miyaji 2010) or some individual cult images of buddhas and bodhisattvas (Filigenzi 2005: 108-109) including the emaciated Siddhārtha (Brown 1997). In India proper, by and large, the dhyānamudrā was shown by any tīrthaṅkaras in Kushan and post-Kushan Mathurā, and occasionally by buddhas in Buddhist art, for instance, some of the Seven Buddhas of the Past (Behrendt 2014: 31: fig. 3 a, b, c). The narrative of the Buddha meditating in the Indrasālaguhā was also depicted in Kushan and post-Kushan Mathurā occasionally with the dhyānamudrā (Huntington n.d.: 22). The dhyānamudrā is also shown by the fully-fledged Gupta Buddha images dating to the mid-fifth century AD placed in the four directions over the Sāñcī Stūpa 1, which were mentioned earlier. These Sāñcī Buddha images are considered to be the prototypes of the later jinas in tantric Buddhism at their formative stage, in which they were not yet distinguished by individual distinctive mudrās (Huntington 1985: 197-198), though this view remains highly speculative. Yet Radich also observes the further development of the vajrakāya in later tantric Buddhism (Radich 2012: 282), and thus analogies gained from later tantric traditions, may enhance understanding of earlier material. The dhyānamudrā is also shown by multiplied buddhas of the depiction of ‘the Buddha’s Great Miracle at Śrāvastī’ at Sārnāth from the late fifth century AD (Brown 1984: 83/fig. 6). This takes us to discussion of the prominence of both the dhyānamudrā and the dharmacakramudrā of the Gandhāra origin. In late Gandhāra, these two closely associated mudrās were depicted in the same relief as on the Muhammed Nari stele (Figure 26) (the central buddha with the dharmacakramudrā and a pair of buddhas on the top corners with the dhyānamudrā) or on the same stūpa as at the Jauliāñ monastery (Huntington Archive, Digital Database Collection, no. 9957). The semiotic meaning of the dhyānamudrā is samādhi while that of the dharmacakramudrā is the dharma: I have discussed above the close relationship between the samādhi and the dharma with reference to the visualization practice of pratyutpanna-samādhi. According to Harrison, ‘The principal fruit of this encounter (pratyutpanna-samādhi) is the hearing of the dharma preached by the Buddha, which a practitioner is urged to remember and preach to others after 192 ken ishikaWa: more Gandhāra Than maThurā emerging from the samādhi’ (Harrison 2003: 120). Therefore, it is clear that the Gandhāran preaching buddha first went into samādhi and then taught the dharma. However, although the late Gandhāran preaching buddha is supposed to be showing the samādhi state with the dhyānamudrā, he is actually teaching, exhibiting the dharmacakramudrā, which does not signify meditation. Therefore, there seems to be a discrepancy in the depiction of the Gandhāran preaching buddha, in that two different scenes of the samādhi experience and the following teaching of the dharma had to be incorporated in one image! Conversely, the dhyānamudrā shown by Devnīmorī buddhas does not negate their other aspect of teaching the dharma, which is the key Mahāyāna practice of not only attaining buddhahood but also helping other sentient beings to reach the same psychospiritual goal. The Devnīmorī buddhas also do not show triad compositions, unlike their late Gandhāran counterparts. However, the terracotta plaque mould of a Gandhāran-style stūpa with a juxtaposed buddha image on its façade flanked by a pair of what appear to be bodhisattvas wearing so-called Gupta crowns, excavated at Devnīmorī (Mehta & Chawdhary1966: pl. 18, B.), may evince a more advanced Mahāyānist idea of the embodiment of a buddha. Therefore, Devnīmorī buddha images may show some transitional stage in the development of buddha-bodies, and there seems to be a clear overlap between the dharmakāya and other new Mahāyānist forms of embodiments as well as identifications between Śākyamuni and other Mahāyānist buddhas. My arguments developed in this section, overall, support the ongoing scholarly consensus regarding Mahāyāna association with the late Gandhāran preaching buddha, as outlined earlier, but they further strengthen this general claim. My new view, that the late Gandhāran preaching buddha image may have a link with a specific visualization practice as a meditation object, as explained in the Pratyutpannasaṃmukhāvasthita-samādhi-sūtra, is particularly significant considering the recent identification of fragments of a Gāndhārī/Kharoṣṭhī manuscript of the Pratyutpanna-saṃmukhāvasthita-samādhi-sūtra (Harrison, Lenz & Salomon 2018). My findings fit well with the Buddhological discourse on the increasing recognition of the associations of Mahāyāna Buddhism in Gandhāra. There have been continuing discoveries/identifications of more Mahāyānist manuscripts, yet this is still a handful, among far too many of the mainstream Śrāvakayāna materials in the Mahāyāna trend of relating itself to mainstream Buddhism (Harrison, Lenz & Salomon 2018: 119). In turn, at Devnīmorī, Gandhāran Mahāyānist iconographic influence was subtly expressed and the possible overlap between old and new ideas of buddha-bodies, as argued in this section, also agrees with the general view of Mahāyāna Buddhism as uninstitutional and non-sectarian. The juxtaposition of the Middle Indic inscription (Hinüber 1985) with that in Sanskrit on the same reliquary at Devnīmorī (Srinivasan 1968) is also noteworthy. Hinüber explains this coexistence of the Middle Indic and Sanskrit as the difference between scriptural and administrative languages, respectively (Hinüber 1985: 198), and this may be the case. However, the contrast of the archaic idea of pratītyasamutpāda or ‘dependant origination’ in the Middle Indic inscription with specifically Mahāyānist technical terms such as śākyabhikṣu or saṃyaksaṃbuddha in Sanskrit, may also be indicative of a religious climate of inclusivity of Mahāyāna Buddhism within a conventional monastery. The fact that the relics of the mahāstūpa at Devnīmorī were established by a lay mason but the mahāstūpa itself was built by two Mahāyāna monks or śākyabhikṣu (Srinivasan 1968: 68) is also suggestive of a diverse Buddhist community at Devnīmorī. The further demonstration of the ongoing iconographic influence of the late Gandhāran preaching buddha in the Western Deccan (Fukuyama 2014: part 2) after Devnīmorī possibly demonstrates the further spread of Mahāyāna Buddhism from Gandhāra to the Western Deccan through Gujarat. 193 The Global ConneCTions of Gandhāran arT Discussion and conclusion In Gujarat, the process of the major pan-Indian transition from ‘aniconic’ to anthropomorphic Buddha images took place at Devnīmorī around AD 400, two to three hundred years later than at Mathurā, in Gandhāra, and in South India. Buddhist art in Gujarat starts with an ‘aniconic’ phase in the Western Kṣatrapa period, characterized by mostly plain Buddhist rock-cut caves, with anthropomorphic cult images of the Buddha being strictly absent (Nanavati & Dhaky 1969: 15). This contrasts strongly with the iconic representations found to the north and east, in the territory of the Kushans and the Sātavāhanas. However, image worship was not absent in Gujarat prior to Devnīmorī: a seated so-called Kapardin Buddha/Bodhisattva image in Sikri sandstone from Kushan Mathurā was found at the Buddhist site of Vadnagar in north Gujarat in 1992 (it is now in Vadnagar Museum; Hinüber & Skilling 2016) (Figure 27). The active importation of religious imagery at Vadnagar is further attested by the result of the recent excavations at the Buddhist monastic complex there, which have revealed votive square-platform stūpas, largely similar in structure to those of Devnīmorī and Mīrpur Khās (Rawat 2011). The archaeological finds include a fragment of a portable Buddhist narrative relief of the offering of the monkey in schist from Gandhāra (Rawat 2011: figs. 11, 19), as well as the small fragmentary image of the head of a Buddha/Bodhisattva in red Sikri sandstone from Kushan Mathurā (Rawat 2011: figs. 11, 18). The longdistance trade and circulation of religious imagery from Gandhāra and Mathurā under the Kushans is well known: for instance, a few sculptures in schist from Gandhāra were found in Mathurā, while several others in red Sikri sandstone from Mathurā were found in Gandhāra as well as in the Gangetic Valley (Van Lohuizen-De Leeuw 1972: 38-39), and this enhances our understanding of the direct influences of these two production centres on other regions. I have argued in this paper that different regions adopted one or more elements of Gandhāran Buddhist art. Consequently, distant Buddhist sites should have been connected remotely yet ‘closely’ by travelling monks, pilgrims, and merchants. Foucher, in his 1907 paper entitled ‘The Beginnings of Buddhist Art’, made an argument that the idea of, stūpa architecture spread all over South Asia because of pilgrims possibly carrying portable miniature stūpas while travelling from site to site (Foucher 1917: 11) though there is no archaeological support for his hypothesis. Yet it appears that Devnīmorī and Mīrpur Khās had much more direct contacts with Gandhāra than other regions. However, the wider picture looks much more inter-regional and pan-Indian: Czuma’s 1970 paper argues that for portable Buddhist bronzes, the Gandhāra prototype served as a substantial body for later Gupta-style bronzes in South Asia, and different influences from Mathurā, Sārnāth, and Gandhāra merged within individual sites and reached far beyond even Nepal (Czuma 1970). Buddhists or artists seem to have had positive attitudes towards a mix-and-match of different regional styles in experimental ways. Gandhāra was often especially influential in the rest of the subcontinent and Gandhāran ‘heritage’ appears in different regional formations over time. As is evident from the result of this paper, it is now clear that Buddhist material culture at Devnīmorī was achieved by the integration of different cultures through a complex network of interregional connections, in this case with Gandhāra and Mathurā. North Gujarat was a natural junction as well as a crossroads of trade routes that connected mainland Gujarat with the Kutch/Indus Delta to the south of Gandhāra, as well as with North India through Bairāt to Mathurā. As such, the case study of Devnīmorī provides an alternative explanation regarding the formation of a widespread material culture. Direct Gandhāran influence on Devnīmorī and Mīrpur Khās further gives a sense of the harmonious development of the Devnīmorī buddhas between Gandhāra and Mathurā. At the same time, the receptivity of foreign influences at Devnīmorī strongly shows highly creative and innovative features such as the almost exclusive use of terracotta and the completely new kind of buddha images in the Gupta style. 194 ken ishikaWa: more Gandhāra Than maThurā Overall evidence supports the view that the regional Devnīmorī buddha images are among the earliest examples of pan-Indian Gupta material culture that came into existence clearly under the influence of the two main cultural sources of Gandhāra and Mathurā. Gandhāran ‘influences’ may be the most recognizable in Mathurā in terms of the importation of sculptures and the assimilation of specific iconographic features such as the pleated garment. However, Devnīmorī buddhas seem to show more of direct Gandhāran influences not just iconographically but also stylistically. Meanwhile, Devnīmorī and subsequent Gupta Buddha/buddha sculptures represent a new historic phase in the development of Buddhist art, having incorporated features from Mathurā, where Gandhāran influences had been constantly assimilated. This formation process often involved the import of small-sized images from these regions, as attested by finds at Vadnagar as discussed above. Such a regional and inter-regional network of influences was often kept intact, as seen in the emergence of Devnīmorī under the influence of Gandhāra and Mathurā, and its subsequent by-product, Mīrpur Khās, under the influence of Gandhāra and Sārnāth (or panIndian). A strikingly similar convergence or melding of two traditions was noted at Sārnāth, where the then universal Buddha image type, especially in the bhadrāsana, was created out of the Buddha images of Gandhāra and Mathurā. Equally significant is the second wave of Gandhāran influence seen at Devnīmorī through the singularly most important iconographic type of the late Gandhāran Mahāyānist preaching buddha, which within a century eventually reached the Western Ghats (Fukuyama 2014) and Sārnāth and even after a century or so, locally persisted at Dhānk as well as at Mīrpur Khās in Sindh. The preaching buddha images of late Gandhāran origin appeared almost simultaneously at Dhānk, Ajaṇṭā, and Sārnāth in the late fifth to early sixth century AD. Such different regional/local manifestations characterize stylistic development around this period and belong to the pan-Indian tradition. Mīrpur Khās and Devnīmorī are sites precisely located at the articulation between the earlier and still partly contemporary art of northern Pakistan and eastern Afghanistan and the art of India proper and southern Pakistan. At the same time, Devnīmorī's Buddhist material culture may have remained influential even after the fall of the Western Kṣatrapa, as indicated by Maitraka coins found at the site in association with later repairs. Gujarat can be conceived of as having a long-established regional network of communication and interactions as a local heritage rather than the dominant widespread culture such as Gandhāra and Gupta influencing ‘provincial’ regions even if not parts of the Gupta empire. In this respect, the notorious debate on the origin of the Buddha images, disputed between Mathurā and Gandhāra (Linrothe 1998), may be missing the whole point. Among different theories I support Van Lohuizen’s initial emphasis on the model of cultural contacts between Gandhāra and Mathurā as a chief driver of the origin of the Buddha images in these two regions, in which she argues that the concepts of the Buddha, his iconographies, and artistic styles were exchanged in both directions between the two (Van LohuizenDe Leeuw 1972). I am inclined to elaborate that two different origins of Buddha images at Mathurā and in Gandhāra may not have been mutually exclusive and pre-Buddhist Graeco-Roman and Indic anthropomorphic images may well have been in circulation prior to the appearance of the first Buddha images between the two places. The case study of Devnīmorī, certainly reinforces her argument and has given us an alternative view on formations of regional and pan-Indian material cultures, although she had been known to advocate Mathurā as an independent origin of the Buddha image (Van Lohuizen-De Leeuw 1949; 1981). However, my inter-regional model as ‘transculturation’, argued through the emergence of Devnīmorī buddha images, does not provide any justification for the ultimate origin of the Buddha image in South Asia. However, it may contribute to the fresher direction of investigation into the emergence of Gandhāran art from the perspective of transculturation (Filigenzi 2012; Bhandare 2018; Karashima 2013; 195 The Global ConneCTions of Gandhāran arT Filigenzi 2019; Kellner 2019). It can be argued that so-called Gupta material culture can be explained by the case-study of Devnīmorī as a product of transculturation of Gandhāra and Mathurā. This viewpoint also stimulates the ongoing debate on the incorporation of Mahāyāna Buddhism, as evident from the transcultural transmission of the iconography of the late Gandhāran preaching buddha. Meanwhile there have been several recent, intensive surveys and excavations of newly discovered Buddhist sites in north Gujarat, in Taranga and beyond, showing some affinities with the Buddhist material culture at Devnīmorī, which await further contextualization. Acknowledgements My research was conducted in formal collaboration with M.S. Baroda University, Gujarat, and I am grateful to my collaborator, Prof K. Krishnan, as well as my mentors Prof Anup Dave and Yadubir Singh Rawat in Gujarat. I am also in debt to M.S. Baroda University and the Huntington Photographic Archive for permissions to use photographs for this publication. My gratitude also goes to Dr Claudine BautzePicron and Dr Chris Jones for their valuable suggestions. This paper is dedicated to Lance Cousins and Prof Stefano Zacchetti, who also greatly supported my research. I also credit Eiko Uehara for our initial collaboration. Last but not least, I am also grateful for my parents and my supervisors Dr Anke Hein and Dr Julia Shaw. References Agrawala R. 1959. Some Unpublished Sculptures from Southwestern Rajasthan. Lalit Kalā 6: 63-71. Asher F. 2008. Bodh Gayā. New York: Oxford University Press. Bautze-Picron C. 1990. The Nimbus in India upto the Gupta Period. Silk Road Art and Archaeology l: 81-97. Behrendt K. 2003. Buddhist Architecture of Gandhāra. Leiden: Brill. Behrendt K. 2014. Maitreya and the Past Buddhas: Interactions between Gandhāra and Northern India. Pages 29-40 in Changing Forms and Cultural Identity: Religious and Secular Iconographies, vol. 1. Turnhout, Belgium: Brepols Publishers. Bhandare S. 2006. Political Transition in Early 5th Century Gujarat: A Numismatic Reappraisal Based on Silver Issues of the Western Kṣatrapas, the Guptas and their contemporaries. Numismatic Digest 29-30: 69-107. Bhandare S. 2018. Numismatics of ‘The Other’: Investigating Coinage and ‘Greekness’ at Taxila. Pages 70-103 in H. P. Ray (ed), Buddhism and Gandhara: An Archaeology of 22 Museum Collections, New Delhi: Routledge. Bhandarkar D. 1920. Buddhist Stūpa at Saidpur in Sind. Pages 89-96 in J. Marshall (ed), Archaeological Survey of India Annual Report 1914-15. Calcutta: Superintendent Government Printing, India. Boucher D. 1991. The Pratītyasamutpādagāthā and its Role in the Medieval Cult of the Relics. Journal of International Association of Buddhist Studies 14/1: 1-27. Brown P. 1940. Indian Architecture (Buddhist and Hindu). London: Trübner & Co. Brown R. 1984. The Śrāvastī Miracles in the Art of India and Dvāravatī. Archives of Asian Art 37: 79-95. Brown R. 1997. The Emaciated Gandhāran Buddha Images: Asceticism, Health, and the Body. Pages 105115 in N. Eilenberg, M. Subhadradis Diskul, and R. Brown (eds), Living a Life in Accord with Dhamma: Papers in Honor of Professor Jean Boisselier on His Eightieth Birthday, Bangkok: Silpakorn. Brown R. 2004. Vākāṭaka-Period Hindu Sculpture. Pages 59-69 in H. Bakker (ed), The Vākāṭaka Heritage: Indian Culture at the Crossroads. Groningen: Egbert Forsten. Chappell D. 1983. Tʿien-tʿai Buddhism: an outline of The fourfold teachings (Tʿien-tʿai ssŭ chiao i). Tokyo: DaiichiShobō; Honolulu: The University Press of Hawaii.r Chatterjee I. 2015. Monastic ‘Governmentality’: Revisiting ‘Community’ and ‘Communalism’ in South Asia. History Compass 13/10: 497–511. 196 ken ishikaWa: more Gandhāra Than maThurā Chandra M. 1964. A Study in the Terracottas from Mirpurkhas. Bulletin of the Prince of Wales Museum 7: 1-22. Chawdhary S. 1962. A Bronze Statuette of Atlas from Shamalaji. Journal of Oriental Institute 11/3: 309–315. Chawdhary S. 1964. Buddhist Monuments of Devnimori – North Gujarat (a dissertation). Vadodara: M.S. Baroda University. Chowdhary S. 2010. Devnimori: Buddhist Monuments. Baroda: M.S. Baroda University/The Government of Gujarat. Cifuentes B. 2013. Bodh Gayā: A Study of the Site of the Buddha’s Enlightenment and the Related Collections in the Victoria and Albert and British Museum (a dissertation). Durham: Durham University. Clough B. 2012. The Cultivation of Yogic Powers in the Pāli Path Manuals of Theravāda Buddhism. Pages 77-95 in Jacobsen K. (ed), Yoga Powers: Extraordinary Capacities Attained Through Meditation and Concentration. Leiden and Boston: Brill. Coninghan et al. 2007. The State of Theocracy: Defining An Early Medieval Hinterland in Sri Lanka. Antiquity 81: 699-719. Conze E. (ed and trans.). 1962. The Gilgit Manuscript of the Aṣṭādaśasāhasrikāprajñāpāramitā: Chapters 70 to 82, Corresponding to the 6th, 7th and 8th Abhisamayas. Roma: Istituto italiano per il Medio ed Estremo Oriente. Cousens H. 1914. Buddhist Stūpa at Mīrpur Khās, Sindh. Pages 89-96 in J. Vogel (ed), Archaeological Survey of India: Annual Report 1909-10. Calcutta: Superintendent Government Printing, India. Cousens H. 1926. The Architectural Antiquities of Western India. London: The India Society. Cousins L. 2003. Sākiyabhikkhu/sakyabhikkhu/śākyabhikṣu: A Mistaken Link to the Mahāyāna? Nagoya Studies in Indian Culture and Buddhism: Sambhāsā 23: 1-27. Cowel E. and R. Neil (eds) 1970. The Divyāvadāna: A Collection of Early Buddhist Legends. Cambridge: University Press. Czuma S. 1970. A Gupta Style Bronze Buddha. The Bulletin of the Cleveland Museum of Art 57/2: 55-67. Damsteegt T. 1978. Epigraphical Hybrid Sanskrit. Leiden: Brill. Desai D. 2013. Bronzes from Sopara: Seven Buddhas and a Maitreya. Pages 124-133 in D. Desai, Art and Icon: Essays on Early Indian Art. New Delhi: Aryan Books International. Errington E. and H. Falk 2002. Numismatic Evidence for Dating the ‘Kaniṣka’ Reliquary. Silk Road Art and Archaeology 8: 101-119. Faccenna D. & Spagnesi P. 2015. Buddhist Architecture in the Swat Valley, Pakistan (2nd edn.). Rome and Bologna: ISMEO; Università di Bologna. Filigenzi A. 2005. Gestures and Things: The Buddha’s Robe in Gandharan Art. East and West 55/4: 103-116. Filigenzi A. 2012. Orientalised Hellenism versus Hellenised Orient: Reversing the Perspective on Gandharan Art. Civilizations from Scythia to Siberia 18: 111-141. Filigenzi A. 2019. Non-Buddhist Customs of Buddhist People: Visual and Archaeological Evidence from North-West Pakistan. Pages 53-84 in B. Kellner (ed), Buddhism and the Dynamics of Transculturality: New Approaches. Berlin/Boston: De Gruyter. Fiordalis D. 2012. The Wondrous Display of Superhuman Power in the Vimalakīrtinirdeśa: Miracle or Marvel? Pages 97-125 in K. Jacobsen (ed), Yoga Powers: Extraordinary Capacities Attained Through Meditation and Concentration. Leiden; Boston: Brill. Fogelin L. 2015. An Archaeological History of Indian Buddhism. New York: Oxford University Press. Foucher A. 1905. L’art gréco-bouddhique du Gandhâra: étude Sur les origines de l’influence classique dans l’art bouddhique de l’Inde et de l’Extreme-Orient. Paris: E. Leroux. Foucher A. 1917. The Beginnings of Buddhist Art, and Other Essays in Indian and Central-Asian Archaeology. Paris: P. Geuthner. Fox A. 2005. The Practice of Huayan Buddhism. Pages 259-286 in 漢傳佛教研究的過去現在未來 (Chinese Buddhism: Past, Present and Future). Center For Buddhist Studies Fo Guang University: I-lan hsien, Taiwan. 197 The Global ConneCTions of Gandhāran arT Fukuyama Y. 2014. アジャンタ－後期壁画の研究 / Ajaṇṭā-kōki-hekiga no Kenkyū (Studies on the Ajaṇṭā Paintings in the Later Phase). Tokyo: Chūō Kōron Bijutsu Shuppan. Fujichika K. 1995. 初期大乗経典における転法輪 / Shoki-bukkyō-kyōten ni okeru tenbōrin (Dharmacakra-pravartana in Early Mahāyāna Sūtras). 印度學佛教學研究 / Indo-gaku Bukkyō-gaku Kenkyū (Journal of Indian and Buddhist Studies) 43/2: 178-183. Fussman G. 1987. Numismatic and Epigraphic Evidence for the Chronology of Early Gandharan Art. Pages 67-88 in M. Yaldiz and W. Lobo (eds), Investigating Indian Art (Veröffentlichungen des Museums für Indische Kunst, vol. 8). Berlin: Museum für Indische Kunst. Fussman G. 1999. La place des Sukhāvatī-vyūha dans le bouddhisme indien. Journal asiatique 287/2: 523-586. Ghosh A. 1967. A Note on the Relic Casket from Devnimori. Journal of M.S. University of Baroda 16: 135. Giès J. and Cohen M. (eds) 1996. Sérinde, Terre de Bouddha: dix siècles d’art sur la Route de la Soie. Paris: Éditions de la Réunion des musées nationaux. Gilliland K. 2013. The Dry Tank: Development and Disuse of Water Management Infrastructure in the Anarudhapura Hinterland, Sri. Journal of Archaeological Science 40: 1012-1028. Gombrich R. 1980. The Significance of the Former Buddhas in the Theravādin Tradition. Pages 62-72 in S. Bālasūriya and W. Rāhula (eds), Buddhist Studies in Honor of Walpola Rahula. London: Gordon Fraser and Sri Lanka: Vimansa. Gorakshkar S. 1991. Sculptural Activity of the Gupta Period in Western India: Mīrpur Khās, Devnimori, Kanheri. Pages 81-84 in K. Khandalavala (ed), The Golden Age: Empire, Province and Influence, Bombay: Marg Publications. Goshima K. 2015. 文殊菩薩と「３種の奇蹟（pratiharya）」/ Monjubosatsu to Sanshu no Kiseki [Bodhisattva Manjusri and the Three Kinds of Miracle (Pratiharya)]. 佛教大学仏教学会紀要 The Bulletin of the Association of Buddhist Studies Bukkyō University 20: 1-38. Griffiths P. 1981. Concentration or Insight: The Problematic of Theravāda Buddhist Meditation-Theory. Journal of the American Academy of Religion 49/4: 605-624. Griffiths P. 1990. Omniscience in the Mahāyānasūtrālankāra and its Commentaries Source. Indo-Iranian Journal 33/2: 85-120. Griswold A. 1963. Prolegomena to the Study of the Buddha’s Dress in Chinese Sculpture: With Particular Reference to the Rietberg Museum’s Collection. Artibus Asiae 26: 85-131. Hakamaya N. 1986. チベットにおけるマイトレーヤの五法の軌跡 Chibetto ni okeru Maitreya no gohō no kiseki (The Tibetan Tradition of the Five Treatises Attributed to Maitreya). Pages 235-268 in Z. Yamaguchi (ed), チベットの仏教と社会 / Chibetto no Bukkyō to Shakai (Buddhism and Society in Tibet). Tokyo: Shunjūsha. Halkias G. 2014. When the Greeks Converted the Buddha: Asymmetrical Transfers of Knowledge in IndoGreek Cultures. Pages 43-73 in P. Wick and V. Rabens (eds), Religions and Trade: Religious Formation, Transformation, and Cross-Cultural Exchange between East and West. Leiden: Brill. Harle J. 1974. Gupta Sculpture: Indian Sculpture of the Fourth to the Sixth Centuries A.D. Oxford: Clarendon Press. Harrison P. 1978. Buddhanusmrti in the Pratyutpanna-Buddha-Sammukhavasthita-Samadhi-Sutra. Journal of Indian Philosophy 6/1: 35-57. Harrison P. 1998. The Pratyutpanna Samādhi Sutra. Berkeley: Numata Center for Buddhist Translation and Research. Harrison P. 2003. Mediums and Messages: Reflections on the Production of Mahāyāna Sūtras. The Eastern Buddhist n.s. 35/1-2: 115-151. Harrison P. and Luczanits C. 2011. New Light on (and from) the Muhammad Nari Stele. Pages 69-127 in BARC (ed), Special International Symposium on Pure Land Buddhism. Kyoto: BARC, Ōtani University. Harrison P., Timothy L. and Salomon R. 2018. Fragments of a Gandhārī Manuscript of the Pratyutpannasaṃmukhāvasthitasamādhisūtra. Journal of International Association of Buddhist Studies 41: 117-143. 198 ken ishikaWa: more Gandhāra Than maThurā Hauptmann, H. 2008. Felsbildkunst am Oberen Indus. Pages 352-357 in C. Luczanits (ed), Gandhara: Das buddhistische Erbe Pakistans. Legenden, Klöster und Paradiese. Bonn: Kunst- und Ausstellungshalle der Bundesrepublik Deutschland; Mainz: Philipp von Zabern. Hinüber O. von 1985. Epigraphical Varieties of Continental Pāli from Devnimori and Ratnagiri. Pages 185-200 in Kumoi Shōzen (ed), 仏教と異宗教: 雲井昭善博士古稀記念 / Bukkyō to Ishūkyō: Kumoi Shōzen Hakushi Koki-kinen (Buddhism in Relation to Other Religions: Essays in Honour of Dr. Shōzen Kumoi on His Seventieth Birthday). Kyoto: Heirakuji Shoten. Hinüber O. von & Skilling P. 2016. An Inscribed Kuṣāṇa Bodhisatva from Vadnagar. ARIRIAB (Annual Report of The International Research Institute for Advanced Buddhology) 19: 21-28. Howard A. 1986. The Imagery of the Cosmological Buddha. Leiden: Brill. Huntington J. 1980. A Gandhāran Image of Amitāyus’ Sukhāvatī. Annali dell’Istituto Universitario Orientale di Napoli 40: 651-672. Huntington S. 1985. The Art of Ancient India: Buddhist, Hindu, Jain. New York: Weatherhill. Huntington J. 2000. The Dharmacakramudrā Variant at Ajaṇṭā: Iconological Studies. Ars Orientalis 30: 33-39. Huntington J. n.d. ‘Mudra In Pan-Asian Buddhism Part 1: Primary Mudras of the Major Buddhas’ (PowerPoint presentation). Huntington Archive. Indian Archaeology - A Review 1960-61. 1961. New Delhi: Archaeological Survey of India. Ishikawa K. and E. Uehara. 2014. ‘Devnimori and Mīrpur Khās: Buddhist Material Culture on the Western Peripheries of South Asia’ (an unpublished joint presentation at EASAA [The European Association for South Asian Archaeology and Art] 2014 held in Stockholm, Sweden, from the 30th of June to 4th July, 2014). Ishida T. 1997. 法華経の神変覚え書き: 光明と化仏 / Hokkekyō no shinpen oboegaki: kōmei to kebutsu (Notes on Divine Miracles in the Saddharmapuṇḍarīka; Light and Emanations). 桂林学叢Keirin-gakusō 16: 1-22. Jayawickarama N. 1995. Buddhavaṃsa and Cariyāpiṭaka. Oxford: The Pāli Text Society. Jha A. and D. Rajgor 1992. Studies in the Coinage of the Western Ksatrapas. Nasik: Indian Institute of Research in Numismatic Studies. Jones C. 2016. Beings, Non-Beings, and Buddhas: Contrasting Notions of tathāgatagarbha in the Anūnatvāpūrṇatvanirdeśaparivarta and Mahābherī Sūtra. Journal of the Oxford Centre for Buddhist Studies 5: 53-84. Jongeward D., Errington E., Salomon R. and S. Baums 2012. Gandharan Buddhist Reliquaries. Seattle: Early Buddhist Manuscripts Project. Kato N., Yatani S. and M. Masui 2017. 中央ガンダーラ・タキシラ・スワートにおける仏塔基壇の編年的 考察 / Chūō Gandāra, Takisira, Suwāto ni okeru buttō no hennenteki-kōsatsu (Chronological Study of Plinth Forms of Stupas in Central Gandhara, Taxila and Swat”): ガンダーラ仏教寺院における仏 塔に関する研究その 4/ Gandāra bukkyō-jiin ni okeru buttō ni kansuru kenkyū sono yon (Study of Forms of Stupas in the Gandhara Buddhist Temples, part 4), 日本建築学会計画系論文集 / Nihonkenchiku-gakkai-kei-ronbun-shū (Journal of Architecture and Planning) 82/741: 2979-2987. Katsusaki, Y. 2015. 小品系・大品系般若経の成立と小部般若経典 / Shōbon-kei/Daibon-kei-Hannyakyō no seiritsu to Shōbu-Hannya-kyōten (The Formation of smaller and larger Prajñāpāramitā-sūtras and the smaller collection of Prajñāpāramitā-sūtras/the Mahāprajñāpāramitā-sūtra). Pages in 27-49 in 般若経大全 / Hannya-kyō-taizen (A Collection of Papers on Prajñāpāramitā -sūtra). Tokyo: Shunjyū-sha. Kajiyama, Yuichi. 2000. The Saddharmapuṇḍarīka and Śūnyatā Thought. Journal of Oriental Studies, 10: 72–96. Karashima S. 2013. Was the Aṣṭasāhasrikā Prajñāpāramitā Compiled in Gandhāra in Gāndhārī. ARIRIAB (Annual Report of The International Research Institute for Advanced Buddhology) 16: 171-188. Karashima S. 2015. Vehicle (yāna) and Wisdom (jñāna) in the Lotus Sutra: the Origin of the Notion of yāna in Mahāyāna Buddhism. ARIRIAB (Annual Report of The International Research Institute for Advanced Buddhology) 18: 163-196. 199 The Global ConneCTions of Gandhāran arT Karashima S. 2018. A Gandhāran Stūpa as Depicted in the Lotus Sutra. ARIRIAB (Annual Report of The International Research Institute for Advanced Buddhology) 21: 471-478. Kellner B. 2019. Introduction. Pages 1-13 in B. Kellner (ed), Buddhism and the Dynamics of Transculturality: New Approaches. Berlin/Boston: De Gruyter. Koezuka T. 1984. マトゥラーにおけるグプタ古典様式の展開 / Maturā ni okeru Guputa-koten-yōshiki no tenkai (Development of Classic-Style Sculptures of Gupta Dynasty at Mathurā). 佛教藝術 / Bukkyō Geijutsu (Ars Buddhica) 156: 80-96. Krishna A. 1981. An Exceptional Group of Painted Buddha Figures at Ajanṭā. The Journall of the International Association of Buddhist Studies 4/1: 96-100. Kurita I. 2003. Gandhāran Art/ガンダーラ美術/Gandāra Bijutsu. Tōkyō; Nigensha. Lahiri N. 2011. Revisiting the Cultural Landscape of Junagadh in the time of the Mauryas. Purātattva 41: 115-130. Le T. 1992. Buddhist Monuments and Antiquities of Gujarat (a masters dissertation). Vadodara: M.S. Baroda University. Lewis-Williams D. 2002. The Mind in the Cave: Consciousness and the Origins of Art. London: Thames & Hudson. Linrothe R. 1998. Inquiries Into the Origins of the Buddha Image: A Review. East and West 43: 1-4. Lohuizen-de Leeuw J. van 1949. The Scythian Period. Leiden: Brill. Lohuizen-de Leeuw J. van 1972. Gandhāra and Mathura: Their Cultural Relationship. Pages 27-43 in P. Pal (ed), Aspects of Indian Art: Papers Presented in a Symposium at the Los Angeles County Museum of Art, October, 1970. Leiden: Brill. Lohuizen-de Leeuw J. van 1979. The Pre-Muslim Antiquities of Sindh. Pages 151-174 in J. van Lohuizende Leeuw (ed), South Asian Archaeology 1975. Leiden: Brill. Lohuizen-de Leeuw J. van 1981. New Evidence with Regard to the Origin of the Buddha Image. Pages 377-400 in H. Härtel (ed), South Asian Archaeology 1979: Papers from the 5th Associations of South Asian Archaeologists in Western Europe, Berlin: Dietrich: 377-400. Luczanits C. 2008. Buddhism in a Cosmopolitan Environment: The Art of Gandhara. Orientations 39/7: 46-52. Lyons I. and H. Ingholt 1971. Gandharan Art in Pakistan. Hamden: The Connecticut Academy of Arts and Sciences. Maisey F. 1847-1854. Relic Series. Plate.12. Relic Caskets from No.2 Sthupa at Satdhara (an original drawing in the British Library: WD546/4a). Marshall J. 1931. Mohenjo-daro and the Indus Civilization: Being an Official Account of Archaeological Excavations at Mohenjo-daro Carried Out by The Government of India between the Years 1922 and 1927. London: A. Probsthain. Mehta R.N. 1963. Ancient Bunds in Sabarkantha District, Gujarat. Journal of Oriental Institute 10/4: 359-365. Mehta R.N. 1965. Chronology of the Buddhist Stūpa at Deva-ni-mori. Journal of The Oriental Institute (M.S. University of Baroda, Baroda) 14/3-4: 410-413. Mehta R. and Chowdary S. 1966. Excavation at Devnimori: A Report of the Excavation Conducted from 1960 to 1963. Baroda: Department of Archaeology and Ancient History, Faculty of Arts, M.S. University of Baroda. Mirashi V. 1965. Devni Mori Casket Inscription of the Reign of Rudrasena, Year 127. Vishveshvaranand Indological Journal 3: 101-104. Mitra D. 1963. The Varāha Cave of Udayagiri: An Iconographic Study. Journal of The Asiatic Society (Calcutta) 5/3-4: 99-106. Mitra D. 1971. Buddhist Monuments. Calcutta: Sahitya Samsad. Miyaji A. 1980. インド彫刻史におけるグプタ様式の生成－マトゥラ彫刻を中心に / Indo-chōkokushi ni okeru Guputa-yōshiki no keisei (Formation of the Gupta Style in the History of Indian Sculpture: Centring Around the Mathurā Sculptures). 佛教藝術 / Bukkyō Geijutsu (Ars Buddhica) 130: 11-33. 200 ken ishikaWa: more Gandhāra Than maThurā Miyaji A. 1985. ガンダーラ三尊形式の両脇侍菩薩像について / Gandāra-sanzon-keishiki no ryōwakijbosatsuzō ni tsuite’ (Iconography of Two Flanking Bodhisattvas in Gandhāran Buddhist Triads). Pages 11-30 in A. Miyaji (ed), インド・パキスタンの 仏教図像調査 / Indo, Pakistan no Bukkō-chōsa (化研報告書 / Kaken-hōkoku-sho) (Iconographical Study of Buddhist Art in India and Pakistan [Kaken Report)]). Hirosaki: Hirosaki University. Miyaji A. 1993. 宇宙主としての釈迦仏: インドから中央アジア・中国へ/ Uchūnushi to shiteno Shaka-butsu: Indo kara Chūō-Ajia, Chūgoku he (Śākyamuni as the Cosmological Buddha). Pages 235269 in Musashi Tachikawa (ed), 曼荼羅と輪廻/Mandara to Rinne (Maṇḍala and Saṃsāra): その思想と 美術 / Sono Shisō to Bijutsu. Tokyo: Kōsei Shuppan. Miyaji A. 1995. インドの大日如来の現存作例について / Indo no Dainichinyorai-zō no genzon-sakurei ni tsuite / Images of Vairocana in Indian Art. Mikkyō Zuzō 14, 1995: 1-30. Miyaji A. 2002. 舎衛城の神变と大乗仏教美術の起源 / Shaei-jō no shinpen to daijō-bukkyō no kigen (The Miracle at Śrāvastī and the Origin of Mahāyāna Buddhist Art). 美学美術史研究論集 / Bigakubijutsu-shi-kenkyū-ronshū (Studies in Aesthetics and Art History) 20: 1-27. Miyaji A. 2006. 「火を発する仏陀」と説話表現 / Hi wo hassuru Budda to setsuwa hyōgen (‘Buddha Radiating Fire’ and Its Narrative Representations). Pages 21-38 in T. Tamaoka, T. Koezuka & H. Sobukawa (eds), 芸術学フォーラム 4: 東洋の美術 / Jeijutsu-gaku fōramu 4: Tōyō no bijutsu (Forum on the Study of Art 4: Asian Art). Tokyo: Keisō Shobō. Miyaji A. 2005. ガンダーラにおける最初期の仏像について / Gandāra ni okeru saishoki no butsuzō ni tsuite (Aspects of the earliest Buddha images in Gandhāra). Pages 5-26 in 真鍋俊照博士還暦記念 論集 仏教美術と歴史文化 / Manabe Shunshō Hakushi Kanreki-kinen-ronshū: Bukkyō-bijutsu to Rekishibunka (Essays in Commemoration of the Sixty-First Birthday of Dr. Manabe Toshiaki: Buddhist Art, Buddhism and Culture). Kyoyo: Hōzōkan. Miyaji A. 2011. Response to ‘New Light on (and from) the Mohammand Nari Stele’ by Paul Harrison & Christian Luczanits. Pages 128-130 in BARC (Research Center for Buddhist Cultures in Asia) (ed), Special International Symposium on Pure Land Buddhism. Kyoto: BARC, Ōtani University. Mori M. 2007. (11) バローダ MS 大学博物館 / Barōda-daigaku-hakubutsukan (MS Baroda University). Pages 284-291 in A. Miyaji (ed), 古代インドにおける宗教的造形の諸相 / Kodai-Indo ni okeru Syūkyō-teki- zōkei no Syosō (Aspects of Religious Art in Ancient India). Nagoya: School of Letters, Nagoya University. Mukherjee S. 2008. The Reconstruction of a Buddhist Stupa at Mirpurkhas. Pages 63-73 in P. Pal (ed), Sindh: Past Glory, Present Nostalgia. Mumbai: Marg Publications. Nakanishi M. 2013. Kanganhari no ‘shotenbō-rin’-zu nit suite / カンガンハリの「初転法輪」図について (An Iconographic Study of the First Preaching in the Kanganhalli Relief). Bulletin of the Institute of Shin Buddhist Culture / 真宗文化: 真宗文化研究所年報 22: 1-25. Nanavati J. and M. Dhaky. The Maitraka and the Saindhava Temples of Gujarat. Artibus Asiae, Supplementum 26. Nattier J. 2008. A Guide to the Earliest Chinese Buddhist Translations. Tokyo: International Research Institute for Advanced Buddhology.Odani N. 2003. ガンダーラ仏教における廻向儀礼 / Gandāra ni okeru ekō-girei (Transfer of Merit in Gandhāran Buddhism: An Archaeological and Epigraphic Consideration), 紀要論文 / Kiyō-ronbun (Departmental Bulletin Paper) 38: 45-71. Neelis, J. 2011. Early Buddhist Transmission and Trade Networks: Mobility and Exchange within and beyond the Northwestern Borderlands of South Asia. Leiden and Boston: Brill. Nyanatiloka B. 1980 [first published in 1952]. Buddhist Dictionary: Manual of Buddhist Terms and Doctrines (the fourth revised edition). Kandy: Buddhist Publication Society. Okada Y. 2017『法華経』と「大品般若」における仏の神力・神変 / “Hokkekyō” to ‘Daibonhannya’ ni okeru hotoke no jinriki/shinpen (The Supernatural Powers of the Buddha in “the Lotus Sūtra” and ‘the Pañcaviṃśatisāhasrikā’. 印度学仏教学研究 / Journal of Indian and Buddhist Studies 143: 96-102. Ogasawara A. 2019.『観仏三昧海経』と華厳経類の関係について:「雑華」の文言を中心として / Kanbutsu-zanmai-kaikyō to Kegon-kyō-rui no kankei ni tsuite: Zōke no bungen wo chūshin to shite 201 The Global ConneCTions of Gandhāran arT (The Guan Fo Sanmei Hai Jing and the Avataṃsaka-sūtra and its Related Texts: Focussing on the Wording in the Gaṇḍavyūha-sūtra). Journal of Indian and Buddhist Studies 67/2: 714-717. Ominami R. 1975. 三昧経典における見仏と観仏 / Sanmai-kyōten ni okeru kenbutsu to kanbutsu (Seeing and Visualisation of a Buddha in samādhi-sūtras). Journal of Indian and Buddhist Studies 23/2: 732-735. Osto D. 2004. The Gaṇḍavyūha-sūtra: A Study of Wealth, Gender and Power in an Indian Buddhist Narrative. PhD dissertation: SOAS, the University of London. Osto D. 2018. Altered States and the Origins of the Mahāyāna. Pages 177-205 in P. Harrison (ed), Setting Out the Great Way. Sheffield/Bristol: Equinox. Owen L. 2001. Constructing Another Perspective for Ajanta’s Fifth-Century Excavations. Journal of International Association of Buddhist Studies 24/1: 27-59. Pal P. 1986. Indian Sculpture, Vol I: circa 500 B.C.-A.D. 700. Los Angeles & London: University of California Press, Berkeley. Palumbo A. 2013. An Early Chinese Commentary on the Ekottarika-Āgama: The Fenbie Gongde Lun 分別功德 論 and the History of the Translation of the Zengyi ahan jing 增一阿含經. Taipei: Dharma Drum Buddhist College. Pinto-Orton N. 1991. Red Polished Ware in Gujarat: A Catalogue of Twelve Sites. Pages 46-81 in V. Begley and R. de Puma (eds), Roma and India: The Ancient Sea Trade. Madison, WI: University of Wisconsin Press. Pinto-Orton N. 2013. Red Polished Ware in Gujarat: Surface Collection from Inland Sites. Pages 195222 in S.A. Abraham, P. Gullapalli, T.P. Paczek and U.Z. Raizvi (eds), Connections and Complexity: New Approaches to the Archaeology of South Asia. Walnut Creek: Left Coast Press. Quagliotti A. 1996. Another Look at the Mohammed Nari Stele with the So-Called ‘Miracle of Śrāvastī’. Annali dell’Istituto Universitario Orientale di Napoli 56: 274-289. Radich M. 2012. Immortal Buddhas and their Indestructible Embodiments: The Advent of the Concept of Vajrakāya. Journal of the International Association of Buddhist Studies 34/1-2: 227-290. Radich M. 2015. The Mahāparinirvāṇa-mahāsūtra and the Emergence of Tathāgatagarbha Doctrine. Hamburg: Hamburg University Press. Rawat Y. 2009. Hill Fort of Anarta: Discovery of a unique Early Historical Fort with Cave: dwellings, Buddhist Idols and Remains at Taranga in North Gujarat. Purātattva 39: 20-25, 96-106. Rawat Y. 2011. Recently Found Monastery and Other Buddhist Remains at Vadnagar and Taranga in North Gujarat, India. Pages 210-242 in Department of National Heritage, Ministry of Information, Communications and Culture (ed), Bujang Valley and Early Civilisations in Southeast Asia, Malaysia. Kuala Lumpur: Department of National Heritage, Ministry of Information, Communications and Culture. Ray H. 1989. Early Historical Trade: An overview. Indian Economic & Social History Review 26: 437–457. Ray H. 2004. Beginnings: The Artisan and the Merchant in Early Gujarat, Sixth-Eleventh Centuries. Ars Orientalis 34: 39-61. Revire N. 2016. The Enthroned Buddha in Majesty: An Iconological Study. 2 vols. PhD Dissertation. University of Sorbonne Nouvelle. Paris. Rhi J. 1994. Gandhāran Images of the Śrāvastī Miracle: An Iconographic Reassessment. PhD dissertation, University of California, Berkeley. Rhi J. 1991. From Bodhisattva to Buddha: The Beginning of Iconic Representation in Buddhist Art. Artibus Asiae 54: 207-225. Rhi J. 2003. Early Mahāyāna and Gandhāran Buddhism: An Assessment of the Visual Evidence. The Eastern Buddhist 35/1-2: 152-190. Rhi J. 2006. Bodhisattvas in Gandhāran Art: An Aspect of Mahāyāna in Gandhāran Buddhism. Pages 151182 in P. Brancaccio and K. Behrendt (eds), Gandhāran Buddhism: Archaeology, Art, Texts. Vancouver: UBC Press. Rhi J. 2008. Complex Steles: Great Miracle, Paradise, or Theophany? Pages 254-259 in P. Adriana (ed), Gandhara - The Buddhist Heritage of Pakistan. Legends, Monasteries and Paradise, Mainz: Zabern and Bonn: Kunst- und Ausstellungshalle der BRD. 202 ken ishikaWa: more Gandhāra Than maThurā Rhi J. 2008. Identifying Several Visual Types in Gandhāran Buddha Images. Archives of Asian Art 58: 43-85. Rhi J. 2011. Wondrous Visions: The Mohammad Nari Stele from Gandhara. Orientations 42/2: 112-115. Rhi J. 2018. Looking for Mahāyāna Bodhisattvas. Pages 243-273 in P. Harrison (ed), Setting out on the Great Way: Essays on Early Mahāyāna Buddhism, Sheffield/Bristol: Equinox. Rosenfield J. 1963. On the Dated Carvings of Sārnāth. Artibus Asiae 26/1: 10-26. Rosenfield J. 1967. The Dynastic Arts of the Kuṣāṇas. Berkeley & Los Angeles: University of California Press. Salomon R. 1998. Indian Epigraphy. Oxford: Oxford University Press. Schastok S. 1985. The Śāmalājī Sculptures and 6th Century Art in Western India. Leiden: Brill. Schopen G. 1977. Sukhāvatī as a Generalized Religious Goal in Sanskrit Mahāyāna Sūtra Literature. IndoIranian Journal 19 3/4: 177-210. Schopen G. 1979. Mahāyāna in Indian Inscriptions, Indo-Iranian Journal 21: 1-19. Schopen G. 1987. The Inscription on the Kuṣān Image of Amitābha and the Character of the Early Mahāyāna in India. Journal of the International Association of Buddhist Studies 10/2: 99-136. Schopen G. 1994. Doing Business for the Lord: Lending on Interest and Written Loan Contracts in the Mūlasarvāstivāda-Vinaya. Journal of the American Oriental Society 114/4: 527-554. Shah U. 1960. Sculptures from Śāmalājī and Roḍā (North Gujarat) in the Baroda Museum. Baroda: Baroda Museum and Picture Gallery. Shah U. 1972. Western Indian Sculpture and the So-called Gupta Influence. Pages 44-48 in P. Pratapaditya (ed), Aspects of Indian Art. Leiden: Brill. Shaw J. 2007. Buddhist Landscapes in Central India: Sanchi Hill and Archaeologies of Religious and Social Change, c. Third Century BC to Fifth Century. London: British Association for South Asian Studies; British Academy. Shaw J. 2011. Monasteries, Monasticism, and Patronage in Ancient India: Mawasa, a Recently Documented Hilltop Buddhist Complex in the Sanchi Area of Madhya Pradesh. South Asian Studies 27/2: 111-130. Shaw J. 2016. Religion, ‘Nature’ and Environmental Ethics in Ancient India: Archaeologies of Human: Non-human Suffering and Well-being in Early Buddhist and Hindu Contexts. World Archaeology 48/4: 517-543. Shichi T. 1987. 帝釈窟説法経典の成立について: 仏陀観 / Taishakuten-seppō-kyōten-no-seiritsuni-tsuite (About the formation of the sūtras on Indrasālaguhā). The Journal of the Japanese Buddhist Research Association 53: 35-49. Shikii S. 1965. 観仏三昧海経と観無量寿経 / Kanbutsu-zaimai-kai-kyō and Muryōju-kyō (The Guan Fo Sanmei Hai Jing and the Amitāyurdhyāna-sūtra). Journal of Indian and Buddhist Studies 25: 227-230. Shizutani M. 1953. インド佛教銘文に見出されるŚākyabhikṣu (釋種比丘) なるタイトルについて/ Indo-bukkyō-meibun ni miidasareru śākyabhikṣu (shakushu-biku) naru taitoru ni tsuite (On the Śākyabhikṣu as Found in Indian Buddhistic Inscriptions). Journal of Indian and Buddhist Studies 1/2: 362-363. Sircar D.C. 1965. The Reckoning of the Kathika Kings. Journal of The Oriental Institute (M.S. University of Baroda, Baroda) 14/3-4: 336-339. Skilling P. 2016. Caitya, Mahācaitya, Tathāgatacaitya: Question of Terminology in the Age of Amaravati. Pages 23-36 in A. Shimada and M. Willis (eds), Amaravati: The Art of an Early Buddhist Monuments in Context. London: British Museum Press. Sompura K. 1969. Buddhist Monuments and Sculptures in Gujarat: A Historical Survey. Hoshiarpur: Vishveshvaranand Institute. Sonawane V. 2013. Rock Paintings Depicting Stupas in Gujarat, India. Bulletin of the Asia Institute, n.s. 27: 91-97. Sounders E. 1960. Mudrā; A Study of Symbolic Gestures in Japanese Buddhist Sculpture. New York: Bollingen Foundation. Spink W. 2009. Ajaṇṭā: History and Development: Painting, Sculpture, Architecture—Year by Year. Leiden: Brill. Srinivasan P. 1968. Devni-mori Relic Casket Inscription of Rudrasena, Kathika Year 127. Epigraphia Indica 37: 67-69. 203 The Global ConneCTions of Gandhāran arT Stone E. 1994. The Buddhist Art of Nāgarjunakoṇḍa. Delhi: Motilal Banarsidass Publishers. Strenski I. 1983. On Generalized Exchange and the Domestication of the Sangha. Man n.s. 18/3: 463-477. Strong J. S. 2004. Relics of the Buddha. Princeton; Oxford: Princeton University Press. Sutcliffe J., Shaw J. and E. Brown 2011. Historical Water Resources in South Asia: The Hydrological Background. Hydrological Sciences Journal 56/5: 775-788. Takata O. 1967. 仏像の起源 / Butsuzō no Kigen (The Origin of the Buddha Image). Tokyo: Iwanami Shuppan. Tanabe K. 1982. Dhammapada-atthakathāにみられるUdena王物語 / Dhammapada-atthakathā ni mirareru Udena-ō-monogatari (he Tale of King Udena in the Dhammapada-atthakathā). 印度學佛教 學研究 / Indo-gaku Bukkyō-gaku Kenkyū (Journal of Indian and Buddhist Studies) 30/2: 971-964 (71-78). Tucci G. 1930. On Some Aspects of the Doctrines of Maitreya[ndtha] and Asatiga. Calcutta: University of Calcutta. Vogel J. 1930. La Sculpture de Mathurā. Paris; Bruxelles: Les Éditions G. van Oest. Vaidya P. 1960. Saddharmapuṇḍarīkasūtram. Darbhanga: Mithila Institute of Post-Graduate Studies and Research in Sanskrit Learning. Williams G. 1982. The Art of Gupta India: Empires and Province. Princeton: Princeton University Press. Williams P. 2009. Mahāyāna Buddhism: The Doctrinal Foundations. London: Routledge. Willis M. 2014. Archaeology of Hindu Ritual. New York: Cambridge University Press. Woodburn. A. 1897. Note on Brick Figures found in a Buddhist Tower in Kahu, near Mirpur Khás. Journal of the Bombay Branch of the Royal Asiatic Society 19: 44-46. Yamabe N. 2019. Nine Similes of Tathāgatagarbha in Tathāgatagarbha-sūtra and the Six Similes of Buddhānusmṛti in Guanfo sanmei hai jing. Pages 397-419 in L. Shravak and R. Supriya (eds), Investigating Principles: International Aspects of Buddhist Culture: Essays in Honour of Professor Charles Willemen. Hong Kong: The Buddha-Dharma Centre of Hong Kong. Zacchetti S. (ed) 2005. In Praise of the Light: A Critical Synoptic Edition with an AnnotatedTranslation of Chapters 1-3 of Dharmarakṣa’s Guang zan jing 光讚經, being the earliest Chinese translation of the Larger Prajñāpāramitā. Tokyo: The International Research Institute for Advanced Buddhology, Soka University. Zwalf W. 1996. A Catalogue of the Gandhāra Sculpture in the British Museum. London: British Museum Press. 204 Part 4 Gandhāra and China (207-233): DOI: 10.32028/9781789696950-11 Cross-cultural Buddhist monastery ruins on the Silk Road and beyond: the layout and function of Buddhist monasteries reconsidered Joy Yi Lidu Introduction New archaeological finds sometimes push and even force scholars to revisit established theories, long accepted traditions as well as previous archaeological excavations. The new excavations in Yungang above the caves in the western area are such, and they have high academic value. The excavations not only shed new light on the configurations of the monasteries in Pingcheng (modern Datong), but they also made clear how each section in the Yungang complex functioned. Furthermore, they proved that the freestanding monasteries above the caves were essential components of the Yungang complex as an entity. In addition, the excavations provided new opportunities to re-examine the configurations of freestanding Buddhist monasteries in the process of developments from west to east. This will help us understand the evolution of Buddhist dissemination along the Silk Road in the west and all the way to Pingcheng in the east in the context of Buddhist architecture and art. The dissemination of Buddhism is not just limited to the teachings of Buddhism; the configurations of Buddhist monasteries and Buddhist images are also an important part of the content of the propagation of Buddhism. Finally, the new excavations of the monastery will allow us to reconsider the associations and influences between the monasteries in Central Plain China, especially in the capitals (Pingcheng, Luoyang and Ye) of the Northern Dynasties (AD 386-534), and those in the Greater Gandhāran area, in particular Taxila and the Termez area in southern Uzbekistan, in Central Asia. Consequently, through the analysis, using new archaeological finds, recent and previous research, and literary sources, of the monasteries and the links between them, it is hoped that this study will delineate the evolution and features of the Buddhist monastery configurations in these regions. The author proposes that the layout of the monastery in Yungang under discussion was directly associated with that of the monasteries in Taxila, and that the monastery configurations in the capitals of the Northern Dynasties were directly related to those in old Termez in Central Asia. The direct Buddhist influence on central China may have come from the Greater Gandhāran area, instead of India where the religion itself originated. It will be seen that the dominant configuration of a main stūpa in front and Buddha hall in rear (hereafter qianta houdian 前塔后殿) after the fourth century, in fact, first came from central Asia, not from China itself. And it eventually exerted a strong impact on the layout of Buddhist monasteries in Baekje and the Silla Kingdoms in ancient Korea. Archaeological excavations of Buddhist monasteries in Yungang The earliest literary record of Yungang is by the Northern Wei (AD 386–534) geographer and essayist Li Daoyuan (d. 527) who described the grandeur of the complex (Li 2007: 316): Stones were chiseled and the mountain was hewn according to the structure of the cliff surface. The images are realistic and grandiose. They are rare by the standards of this time. The [Buddha] Halls on the mountain and over water, and the smoke [of the incense]-filled temples, look toward each other. The grove and pond are like a bright mirror. Looking into the distance, a new vista dazzles your eyes… At the time, Yungang was called the Lingyan cave-monastery (Li 2007). Later, in the Weishu (History of the Wei), Yungang was called Mount Wuzhou Buddhist cave-monastery (Wei 1974: VI:130; VII:151). By 207 The Global ConneCTions of Gandhāran arT Figure 1. The Yungang cave complex on Mount the Tang dynasty (618-907), a large cave in Yungang could be higher than 60 meters and hold some 3,000 people (Junjirō & Kaigyoku 1924-32: T50:2060:425c26 and T50:2060:427c23).1 The eminent monk Daoxuan (AD 596-667) observed that the carving of the images was fantastic, the beauty of the ornamentation was unparalleled, and each cave was unique (Junjirō & Kaigyoku 1924-32: T50:2060:427c27). The name Yungang was not used until the Ming dynasty (AD 1368–1644), when it occurs for the first time in an inscription recording the repairs to the Yungang Fortress. The Yungang cave complex derived its name from the sacred peak of Mount Wuzhou in which all the caves were excavated. Yungang means ‘cloud ridges’. The Yungang complex is located south of Mount Wuzhou and north of the Wuzhou River (modern Shili River) and is 6 kilometers west of Datong city in Shanxi Province in north China. The 45 major caves are divided into three sections – east (caves 1 to 4), middle (caves 5 to 20), and west (caves 21 to 45). They were hewn from the mountain cliff surface and stretch out from east to west for more than half a mile (Figure 1). The excavation of the rock-cut caves was initiated by the imperative to carry on the Buddha Dharma infinitely and the wish to pray for blessings for the Northern Wei imperial family who commissioned the caves. In 1902, the Japanese architect Itō Chūta (AD 1867–1954) ‘rediscovered’ Yungang accidently and published two articles introducing it to the world (Itō 1906: nos. 197/198).2 In 1907, the French sinologist Émmanuel-Édouard Chavannes (AD 1865–1918) investigated Yungang and other caves, recording them with his lens. His Mission archéologique dans la Chine septentrionale (Chavannes 1909-15) contains seventyeight valuable photographs of Yungang. After this, the study of Yungang entered a new era of visual images (Chavannes 1909-15). These early expeditions to China at the beginning of the twentieth century opened a new chapter in the scholarship of Yungang. Ground-breaking research was made possible when Su Bai came across the Jin stele inscription, on the basis of which he was able to shed fresh light on periodization and chronological sequence of the caves and provided a new dating scheme, which, due to lack of clear evidence, had long puzzled scholars. What really advanced the study of the Yungang complex was the archaeological findings of the freestanding monasteries above the rock-cut caves in 2010, i.e. the monastery remains in the vast terrace between T refers to Taishō Shinshū Daizōkyō (Junjirō & Kaigyoku 1924-32). See also Dongfang zazhi, 1919: 16, nos. 2/3. This article was translated into Chinese and collected together with Chen Yuan’s article (Chen 1980: 398-409). Chūta Itō was the first to give numbers to the caves (current caves 5 to 13). 1 2 208 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond Wuzhou. (Photo: Yungang Research Academy.) caves 39 and 45 (Figure 2).3 For the first time, we learned that the freestanding monasteries were an inseparable component of the entire cave complex and the residential area was, in fact, above the rockcut caves. In addition, the finds shed significant light on the configuration of a Buddhist monastery in the Northern Wei capital, of which we had no previous indication. Most importantly, the light they threw on cross-cultural influences led scholars further to understand the direct association between Taxila and Yungang, of which we had no clear, hard evidence before. The fully excavated remains of the monastery under discussion sit to the north and face south. The north part of the monastery is well preserved, but the south side is badly damaged without many ruins. The remains are 60-62 metres wide from west to east, and 44 meters from south to north. Judging from the extant ruins, archaeologists believe that the configuration of the monastery is primarily composed of a stūpa and living cells surrounding it (Figure 3). A row of cells (F20-F22) bounds both the north and east sides of the monastery’s quadrangle. Another row (the middle cells, F18 and F19) separates the quadrangle into east and west courts, the former being slightly larger. In front of all of the cells was a cloister, the plinths of which (Z1-Z12) are positioned 1.8-2.1 m away from the cells. In the centre of the east court, slightly towards the south, is the base of a stūpa (Figure 4). The dimensions of the square stūpa, measuring 15 m wide and 0.5 m in height, are not very large. Around the stūpa, glazed flat and semicircular tiles were unearthed. This means that the glazed tiles were used for the stūpa and that the stūpa is a multi-storied wooden structure with glazed tiles for the eaves. This can be attested to by the stūpas carved in the caves in imitation of the structure of wooden stūpas. The stūpa also faces south like the monastery itself and is made of rammed earth. The stūpa base, on top of which are forty column holes, is one of the earliest unearthed thus far. The north side of the base is 12 m from the bases of the columns in front of the north cells. The cells contain single and double rooms (cells F6, F7, F14, F18, F21 are double) with rammed-earth walls, the inner side of which are of plaster mixed with grass, while the exterior is covered with white lime. All the doorways face south as well. Only two north-south cells (F18 and F19) survived in the middle of the court. Their interior walls show traces of plaster mixed with grass and white lime. The exterior 3 Since Japanese archaeologists first started the excavations in the 1940s, many archaeological excavations have been conducted in the Yungang complex. Among them, the excavations conducted in the 1990s in front of caves 9 and 10, and the two in 2010 and 2011 above cave 39, and caves 5 and 6 are especially important. All of the excavations were primarily in front of and above the caves: four areas above the caves (east of caves 1, 3, 5 and 6, as well as in the area between caves 39 and 45 in the western section), and four in front of the caves (caves 3, 8, 9–13, and the five Tanyao caves), among which the excavation in front of caves 9 and 10 is of particular significance, see Mizuno & Toshio 1951, VII: 57–68, 123–9; XV: 91–9, 185–90). For the excavations above cave 39, see Zhang, Li & Jiang 2011: 127–130. See also Zhang 2016: 533-562. 209 The Global ConneCTions of Gandhāran arT 210 Figure 2. Yungang monastery remains above the area between caves 39 and 45, 2010. (Photo: Zhang Qingjie). Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond Figure 3. Yungang monastery with single and double room cells. (Photo: modified from Zhang Qingjie, Kaogu xuebao 4 [2016], pl. 2.) walls show only plaster mixed with grass. Cell 18 is a double room cell of rectangular shape. The main wall is made of rammed earth. Cell 14, rectangular in shape, is also a double room cell. On the front wall is a cooking stove half sunken into the wall. The top of the stove is round. Two column plinths (Z11 and Z12) are found in front of the cell. The distance between them is 3.65 m. Cell 7 is also a double room cell, rectangular in shape. The doorway is in the south-east corner. Two column plinths (Z5 and Z6) are found. Cell 6 is a double room cell as well, but square in shape. The main wall is made of rammed earth. The two plinths in the front are 4.9 m apart. The inner room cell reveals vermilion paint in grass-mixed plaster. Vermilion paint on walls has only, so far, been found on the walls in the ruins of the royal palace of the Northern Wei in Figure 4. Yungang stūpa Base, 2010. (Photo: modified from Zhang Qingjie, Kaogu xuebao, 4 [2016], pl. 13.) 211 The Global ConneCTions of Gandhāran arT Caochangcheng 操場城 in Datong and in the ruins of the Yonggu imperial mausoleum. This indicates that the decorations in these cells were above standard and they were the living quarters for high-rank, eminent monks. Cells 19, 20, 22, 16, and 17 are single rooms. Three cells (F20-22) are on the east side of the monastery. The interior and exterior walls show traces of grass-mixed plaster, but the interior wall is also covered with white lime. It should be mentioned that in front of cell 22, there is a short wall made of semicircular tiles with column plinths in front of it. It is worth noting that cells 16 and 17 are juxtaposed east-west, side by side, instead of north-south, as they should have been. The interior wall is covered with grass-mixed plaster and white lime. Cell 16 is almost square. The east, west, and north walls are made of rammed earth, but the south wall is made of piled-up stone slabs and tiles. The doorway faces south in the middle. Cell 17 is rectangular in shape. The west and north walls are made of rammed earth. The doorway is in the south-west corner of the cell facing south. Up to now, these findings are the first scientifically excavated and relatively intact freestanding monastery ruins in China. Archaeologists have unearthed the ruins of a stūpa base, twenty cells (thirteen in the north [F3-15], two in the middle [F18-19], three in the east [F20-22], and two in the south-west [F16-17]), two pottery kilns (Y501 and Y502) in the south-west section of the monastery, many tiles and tools. Some tiles are glazed, and others are engraved with characters of xiku (west cave) or chuanzuo wuqiong 傳祚無窮 (support the imperial rulers and extend prosperity infinitely). The xiku tile indicates that at the time the cave complex was divided into at least east and west sections. It is unclear whether there was a middle section as we divide the caves now. They also found tiles with lotus-born figures, a stone stele (Figure 5) with a pointed niche and acanthus pattern on the face of it, and other images. The stūpa was the physical centre and main object of veneration. It was primarily for the liturgical purposes of worshipping and circumambulation. It is worth mentioning that archaeologists also discovered earth beds, tops of kitchen ranges, and flues in the cells. These important clues led scholars to believe that the cells were the living quarters for daily use. The significance of the findings of the monastery cannot be emphasized enough. They not only enriched our knowledge about the essential components of the Yungang complex, but also provided concrete evidence about where Buddhist monks lived and translated the Buddhist canonical texts mentioned in the literary records.4 Moreover, the excavations explain why the Yungang caves themselves are all shrine chapels with images and stūpas, and used primarily for worshipping, repentance, making offerings, chanting, and possibly jiangjing (sūtra lecture), changdao (vernacular sūtra singing and preaching), or merely for merit and virtue accumulation in some caves. More importantly, they explain why there were no vihāra (monks’ residence) caves. That, for a long time, was a mystery to scholars. Several observations need to be specially pointed out here: 1. The date of the monastery is essential. In the remains of the monastery, flat-glazed tiles were used but there were almost no polished black tiles or eaves-tiles found used in the Northern Wei Palace in Caochangcheng and the Yonggu Mausoleum in Fangshan. Many of them are grey ceramic tiles thinner than the polished black tiles and eaves-tiles. A large number of glazed tiles appeared. Previously, only a In the 1147 Jin stele inscription, Cao Yan mentions the ten temples in Yungang and the ‘stone chambers’ in which monks translated the sūtras in the Yungang complex, but for a long time, we had no hard evidence to verify the authenticity of the inscription before these new excavations revealed the full picture of the site. According to the Jin stele, there were ten temples in the great rock-cut cave-temple complex, and they were constructed during the Northern Wei. Mention was also made of several stone chambers above the cave temples where the Indian monks translated sūtras. For the record of the stone chambers above the caves in the Jin stele inscription, see Su 1966: 52-75. 4 212 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond Figure 5. Stone stele from Yungang monastery. (Photo: modified from Zhang Qingjie, Kaogu xuebao 4 [2016], pl. 14). few glazed bricks appeared in the ruins of the well in Figure 6. Drawing of the image on Figure 5. (Image: modified from Zhang Qingjie, Kaogu xuebao 4 [2016], pl. 19.) the Northern Wei Palace. Very few varieties of tiles were found and many have characters. Animal tiles often used during the Taihe era (AD 477-499) of the Northern Wei were not seen. The reliquary pit and objects unearthed in the base of the stūpa in Dingzhou, Hebei Province, dated the fifth year of Taihe (AD 481), were not found in the stūpa base here.5 Taking everything into consideration and on the basis of what was unearthed in the ruins, archaeologists believed that the initial date of the monastery was earlier than that of the Northern Wei monastery in Dingzhou (Zhang 2016: 533-62). It should have been before the Taihe era, and could have been as early as Emperor Wencheng’s era (reigned AD 452-465). Now both literary sources and hard evidence confirm that the monastery was started as early as the Wencheng era. This is the earliest monastery thus far excavated. Most importantly, the configuration revealed has significant academic value in figuring out the evolution of the monasteries from Greater Gandhāra through Central Asia eastwards to Pingcheng, the Northern Wei capital. 2. According to the archaeological report, the two cells (F16 and 17) in the south-west corner in the west court are aligned east-west, not north-south, as they should have been, and as the middle or east cells are. In addition, as discussed above, the material used for cells 16 and 17 is different from that used for the north cells. Furthermore, the ruins of aprons (south of plinth Z1-Z4) in front of the northern cells extend all the way south to where they are crushed beneath the east wall of cell 17. All the evidence implies that cells 16 and 17 were added at a later time after the aprons were constructed. In addition, and more importantly, the apron ruins also suggest that there must have been earlier structures on the west side of the west court to the north of cells 16 and 17. The traces of two columns and the large space suggest that it is highly likely the structures at the north of cells 16 and 17 were intended to be shrines 5 See Hebeisheng Wenhuaju Wenwu Gongzuodui 1966: 252. 213 The Global ConneCTions of Gandhāran arT for liturgical purposes. No traces of vihāra cells in the west side of the west court were seen. The wide distance between the two columns also provides a clue that the northernmost structure in this space was bigger than the north-side cells. It is more like a shrine with larger space. 3. In addition, the above-mentioned narrative stone stele which was discovered on site can also be helpful to support our proposal about the supposed shrines on the west of the monastery, since the figures in the narrative scene are holding ritual implements (Figure 6). This may illustrate the ritual practices of the time at the site. The stele contains a pointed niche, inside of which in the centre are a cat-like animal on a ritual implement and a bird at the bottom. The animal between the cat and bird is illegible. Outside the niche on the right, above a lion is a kneeling figure holding the ritual implement. On the left, above a dog is a standing figure who is wearing a V-neck top and loose pants of the Northern Wei holding a Heaven pestle ritual implement. On the top, above the niche front, are two animals and a kneeling figure who is holding a moon-shaped symbol in his right hand and a ritual implement with a handle in his left hand. The ritual implements and animals shown here in this stele are worth further examination, but at least they display some sort of ritual being conducted. Further concrete evidence will also be needed to answer our question about the surmised shrines with full confidence. The observation is made according to the analysis of the excavated ruins and the sizes of the two structures to the north of cells 16 and 17, and based on the configuration of the monasteries in Taxila, Greater Gandhāra, which had a direct connection, as I shall argue, with the monastery in Yungang. 4. The excavations shed significant light not only on the monastery configuration in Yungang itself, but also in Pingcheng (Siyuan monastery 思远寺), and on those constructed during the Pingcheng era (Siyan monastery 思燕寺). These are all single-court monasteries, but the Siyuan monastery developed to the configuration of qianta houdian. The configuration of the Siyan Monastery in Chaoyang city, Liaoning Province, commissioned also by Empress Dowager Wenming (d. 490) during the Pingcheng era, is similar to the Yungang plan but it is a single court monastery.6 The layout of the Yungang monastery, with the stūpa in the centre surrounded by cells, had not been found before in any of the excavated monasteries. The Yungang monastery, though differentiating itself from those excavated thus far, shares striking similarities in configuration with some early monasteries in Taxila, especially with Pippala and Khāder Mohrā near Dharmarājikā. At the same time, each monastery displays its own cultural characteristics, with local artisans’ own innovation to fit religious ritual purposes, spaces, and aesthetic tastes. An analysis of these monasteries in a comparative manner will delineate the dissemination of Buddhism and changes of the configuration of Buddhist monasteries in the process. Monasteries in Taxila in Greater Gandhāra Greater Gandhāra includes the Swat Valley to the north, the western Punjab including the ancient metropolis of Taxila to the east, eastern Afghanistan to the west, northern Afghanistan, southern Uzbekistan, and even parts of the region around the Tarim Basin in today’s Xinjiang Uyghur Autonomous Region of China (Salomon 2018: 11). All of these regions came under the cultural influence of Gandhāra in the early centuries of the Common Era. Gandhāra provides one of the most fascinating chapters in ancient history, a vital crossroads of diverse cultural and political traditions that thrived for several centuries, with a predominately Buddhist orientation. Trade routes facilitated the movement of artistic ideas and techniques that entered Gandhāra from four directions, linking the Mediterranean, the Indian subcontinent, Central Asia, Persia, and China (Jongeward et al. 2012: 8). 6 See Liaoningsheng & Chaoyangshi 2007. 214 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond Figure 7. Map of Taxila (modified from Marshall 1951: vol. 3, pl. 1). The crisscross cultures and influences are mirrored and valued in the development of Buddhist dissemination. The configuration of the Buddhist monastery is a direct reflection of the interlaced cultures. Taxila in Greater Gandhāra, strategically situated at the junction of the great trade routes from eastern India, western Asia, Kashmir, and Central Asia, became a religious heartland with Buddhist monuments throughout the valley. Buddhist archaeological sites at Taxila include the Dharmarājikā complex, the four groups of Chir Tope remains (A, B, C, D), the Kālawān grouping, the Giri monasteries, the Jaṇḍiāl complex, the Mohṛā Morādu monasteries, the Pippala monastery, the Jauliāñ complex and many other remains (Figure 7).7 Among them, the Khādeṛ Mohṛā remains (Chir Tope D) and the Pippala monastery demonstrate architectural resemblance with the counterpart in Yungang, and are our focus for discussion. The numbers and names of the monasteries and stūpas used here were first used by John Marshall in his account of archaeological excavations. 7 215 The Global ConneCTions of Gandhāran arT Figure 8. Layout of the Khādeṛ Mohṛā D2 monastery. (Plan modified from Marshall 1951: vol. 3, pl. 69a). The four Chir Tope monasteries are situated to the south-east of the Dharmarājikā. They are of great interest ‘because they date from the period (c. AD 40-150) when the diaper type of masonry was in vogue, and show us how the quadrangular monastery was being evolved under the early Kushan kings. They also furnish examples of several varieties of masonry not met on other sites’ (Marshall 1951: vol. 1, 315). Most significantly, they seem to have survived only to bear witness to the shared features with the newly excavated monasteries in the Yungang Buddhist cave complex. They demonstrated that by the end of the first century the living cells were in a more private enclosure and the old type of monastery with its disordered planning was gone. The stūpa is still of primary importance, but ‘the tendency is to separate it from the living quarters of the monks, which are now securely enclosed in a walled-in quadrangle’ (Marshall 1951: vol. 1, 320). Clearly, the old monastery layout initially had the stūpa and the living quarters together in one court, as shown in the Yungang monastery, and as we also see in Khādeṛ Mohṛā D2 (Figure 8) and Pippala (Figure 9), as well as in Dharmarajika M5. Only later on was the stūpa separated from the living cells. It is worth noting that the Khāder Mohrā complex includes two sets of monastery complexes, D1 and D2, which together include three courts, not two or one, as is usual, and that of the four Chir Tope (A, B, C, D) sites, Khāder Mohrā D2 is the only one with stūpa and living cells in one space. Here the stūpa is in the 216 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond Figure 9. Layout of the Pippala monastery. (Plan modified from Marshall 1951: vol. 3, pl. 98a). centre of the quadrangle with cells on three sides and larger chambers on the fourth side, surrounding it within one court. This suggests that D2 should be the earliest monastery of all four, and the rest were newer versions from when the living cells were disposed in a separate and more securely enclosed space. D2 does not seem to belong to the complex and should be considered a separate monastery 217 The Global ConneCTions of Gandhāran arT altogether. It should have been the first constructed in the group with the complex later extended to D1. What John Marshall refers to in his discussion of the evolution of the monasteries is D1, which does show further development in plan, with two separate courts. However, D2 is quite different, as Marshall (1951: 321) admitted himself: In the other group (D2) the plan is quite different. Here the stūpa stood in the midst of a large court, with rows of cells on three sides and what appear to have been several larger apartments, including no doubt an assembly hall, on the fourth side, though only a few fragments of the latter have survived. The plan is thus generally similar to that of the small monastery M5 at the Dharmarājikā, which is also referable to the second century A.D., though to a somewhat later date than this one. Only later in the new layout were the stūpa and the living quarters segregated. The stūpa initially was simply left outside without any enclosure, as seen in Chir Topes A and C (Figures 10 and 11). In the new living court, only three sides of the monastery have cells. The fourth side is either left bare or occupied by a small stūpa chapel for the private use of the monks. ‘On these two sites the hall of assembly and other indispensable adjuncts were in all probability outside the monastery, and may have been built of perishable materials, as they had been in the earlier saṅghārāmas.’ (Marshall 1951: vol. 1, 320.) Further developments and changes are best exemplified in Chir Tope B (Akhauri) (Figure 12). Here only two sides of the monastery have living cells; the east side contains a private chapel (F25) in the middle, and the north side comprises an assembly hall, a common-room, and a stūpa-chapel (D1). The main stūpa still remains left in the open, facing the entrance to the monastery, as it was on sites A and C. The small subsidiary stūpas (A1-A5), the row of five chapels (B1-B5), and the smaller monastic court E are all believed to be later accretions. Khāder Mohrā is the largest of the four Chir Tope monasteries. Monastery D1 (Figure 13), the westernmost section of Khāder Mohrā, now seems to be the final stage of the four in the group since it is the most Figure 10. Plan of Chir Tope A. (Modified from Marshall 1951: vol. 3, pl. 67a.) 218 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond Figure 11. Plan of Chir Tope C. (Modified from Marshall 1951: vol. 3, pl. 68a.) Figure 12. Plan of Chir Tope B (Akhauri). (Modified from Marshall 1951: vol. 3, pl. 67b.) 219 The Global ConneCTions of Gandhāran arT Figure 13. Plan of Khāder Mohrā monastery D1. (Modified from Marshall 1951: vol. 3, pl. 68b.) complicated. It contains a stūpa and court with living cells situated in a separate space. The stūpa is unique. Instead of having the flight of steps facing the entrance to the monastery, as usual, it faces north. What this could suggest is unclear now. Further investigation is needed. Does this imply that the initial plan only included the stūpa court as with D2? Both stūpas face north. The walls surrounding the stūpa of D1 have mostly disappeared. The stūpa therefore now looks like it is standing in the court on its own with some surviving cells. Initially the stūpa court should have been centred with the stūpa surrounded by living cells. This is similar to the configuration of its counterpart in Yungang under discussion, with the stūpa in the centre of the cells. The monastery court at the west end of Khāder Mohrā not only bears cells on three sides of the quadrangle, it also has an assembly hall at the south-east corner, as well as a complex of several small chambers alongside. On the north side of the quadrangle court, a rectangular hall, possibly a refectory, was added to the monastery later. Unfortunately, its counterpart in Yungang does not provide us with a clear picture of 220 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond the configuration of its western court. The clues we have are cells 16 and 17 to the south, as discussed earlier, the aprons sunken beneath the cells, and the traces of the two columns to the west of the aprons. There must have been some structures with the aprons in front. Nevertheless, both the Yungang monastery and Khāder Mohrā D1 have two courts, and both eastern courts have the stūpa as its physical centre of veneration. They both also share similarities with Khāder Mohrā D2, the earliest monastery configuration in the group. Another monastery that shares similar configuration is the Pippala complex (Figure 9). There might be a direct connection (Li 2014: 288). Situated in the north-east of the Buddhist hub of Dharmarājikā at the foot of the hills in Taxila between Mohṛā Morādu and Jauliāñ, the Pippala complex is composed of a main stūpa courtyard in the east section, and a quadrangle monastery of later addition to its west, as well as a stūpa enclosed in a court to the north. The main stūpa (A) is placed in the centre of the courtyard which is to the east of the quadrangle monastery. In addition, four small stūpas (B, C, D, and E) are put around the main stūpa and one is placed outside of the courtyard (K). This is unusual. In general, only one main stūpa is placed in the centre of the courtyard. Here we see five stūpas in the stūpa courtyard to help to relieve crowds of worshippers to the main stūpa. The cells originally all faced the main stūpa. It must be emphasized that the remains of the monastery are found to be from two periods. The stūpa courtyard of the monastery to the east, which comprises an open quadrangle in the centre and ranges of cells on its four sides, dates from late Parthian or early Kushan times and fell into ruins before the fourth and fifth centuries since ‘at that time a second monastery was erected over the western side of it, completely hiding beneath its foundations all that remained of the old cells and veranda on this side’ (Marshall 1951: 365). The rest of the early monastery was converted into a stūpa court because everything was levelled to the ground and dismantled except the stūpas in the open quadrangle and the back wall of the cells, which is the enclosure wall of the new courtyard (Marshall 1951: 365). Thus, the original cells surrounding the main stūpa were removed, enlarging the stūpa court. A later stūpa (G) was built in the north-east corner of the courtyard, partially atop the foundations of cells 4, 5 and 6. The western section, the later monastery, is built of heavy semi-ashlar masonry and is well preserved, but it is smaller than the stūpa court monastery on the right. The plan is similar to that of the monasteries at Jauliāñ and Mohṛā Morādu. It consists of a court of cells on the north, with a hall of assembly, kitchen, and refectory on the south, and the converted stūpa-court on the east. Worth noting is that the hall of assembly, kitchen, and refectory resemble the corresponding chambers of Mohṛā Morādu and Jauliāñ. This suggests that the further development of the monasteries began to show certain evolving patterns. The court of cells was two storeys tall and consisted of an open quadrangle with cells on its four sides and a broad-pillared veranda. In the centre of the court was a small rectangular depression about 30 cm in depth and paved with stone, which received the rain-water from the roof and directed it out through a drain on the western side passing under cell 23. Inside cell 31 is a well preserved stūpa. The floor level of the cell in which the stūpa stands is 75 cm below that of the rest of the monastery, and this circumstance as well as the character of the stūpa itself, which is of diaper masonry, led John Marshall to believe that the stūpa was built originally in one of the cells, or possibly in a chapel, of the earlier monastery and then incorporated into the later monastery, when the latter was erected on the ruins of its predecessor (Marshall 1951: vol. 1, 366). From the analysis of these monasteries, it can be seen that the four Chir Tope monasteries throw fresh light on the evolution and types of the early monastery configuration in Taxila. The simplest early ones have only one court. This early type of configuration is represented by the Khāder Mohrā D2, possibly D1, and Pippala monasteries. Further development evolved into two courts. The living cells are not only in a private space, but some courts also contain private chapels, as seen in Chir Tope B (Akhauri). Khāder Mohrā D1 represents the later developments and reflects some generality with certain shared features 221 The Global ConneCTions of Gandhāran arT Figure 14. Plan of Jauliāñ monastery. (Modified from Marshall 1951: vol. 3, pl. 101.) in many other later monasteries. It should be mentioned that the monasteries in the late Parthian or early Kushan periods do not contain any image-chapels or multiple courts. They are, in general, simple in configuration, whether a single or double court. The focus of the structure is primarily on the stūpa, with the living cells and sometimes private chapels surrounding it. The final steps in the evolution of the Buddhist saṇghārāma remained to be taken in the fourth and fifth centuries, when the image-chapel had become as constant and ubiquitous a feature as the stūpa itself. In the saṇghārāma of Jauliāñ (Figure 14), for example, we shall see how, as time went on, the living-quarters had come to be completely separated from the courts of public worship; and how the chief cult stūpa was placed in a quadrangle of its own, with ranges of image-chapels, in place of the older living cells, on all four sides of it; how, apart from a single small private chapel and some cult images, the monastic quadrangle was reserved exclusively for the living quarters of the monks; and how, finally, the hall of assembly, refectory and kitchen came to be grouped together outside this quadrangle in a position where they would be least likely to interfere either with the meditations of the monks or the devotions of lay-worshippers in the public courts (Marshall 1951: vol. 1, 321). What we can get from the above is that, a) the image-chapel became as important as the stūpa itself, but in earlier times, the stūpa court was more prominent (either it was in the centre of the quadrangle court or separated from the living cells); b) the living quarters were later completely separated from the stūpa. This again suggests that the stūpa and living quarters were together initially and supports what was discussed above; c) now the stūpa is in the middle of the quadrangle, and the image-chapels replaced the living cells in the court, in other words, both the stūpa and the image-chapels have an equal liturgical function, i.e. they both became the objects of worship; d) more importantly, the old quadrangle with cells on three sides and a single private chapel on the fourth side now has cells on all four sides. The quadrangle became exclusively a living quarter now for the monks; and e) finally, the additional rooms for the practical functions of the living quadrangle, i.e. the hall of assembly, refectory, and kitchen, are 222 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond put together in a group outside the living quarters so that they will not interfere with meditation and lay-worshippers’ practice. It can be said that up until the fourth and fifth centuries, Buddhist monasteries in the Greater Gandhāra area went through several configural transformations. Not only did the stūpa and living quarter courts get separated, but the living quarters were now in a more secured enclosure, and the image-chapels gradually became as important as the stūpa itself for satisfying the increasing needs of the devotees for worshipping and other liturgical purposes. The configuration of Chinese monasteries did not entirely emulate that of the early monasteries in Taxila such as Chir Tope B or Jauliāñ, but only followed the layout of Khāder Mohrā D2 and Pippala, with the stūpa in the centre of the monastery and the cells for monks surrounding it. However, the architectural developments of the monasteries, in general, share similarities with those in Taxila. In the beginning, the stūpa in the centre was the main characteristic of all the early monasteries, be it in China, Gandhāra, or India. The monasteries mentioned above (Yungang, Siyan and Siyuan) all had the stūpa in the centre of the space and it was the primary object for worshipping although the layout of the Siyuan monastery in Pingcheng, the capital, changed with the stūpa in the front aligned with Buddha halls behind, but the focus was still on the stūpa. Later on, when the capital moved south to Luoyang, the Yongning monastery there adopted the configuration of the Siyuan monastery and had the stūpa and the Buddha halls aligned in the configuration of qianta houdian. At the time of the Eastern Wei-Northern Qi (534-577) dynasties in the fifth and sixth centuries, the monasteries were transformed in configuration again, and had multiple courts. This can be best seen from the newly excavated Zhaopengcheng 趙 彭城 monastery in Linzhang (modern Handan), Hebei Province.8 This development in configuration had happened earlier in the Greater Gandhāra area. The direct connection was unclear until the new excavations, as we shall discuss below. Qianta Houdian monasteries in Central Plain China and their origin The abovementioned excavations can provide us with the evolution of the monasteries in Central Plain China during the fifth and sixth centuries and eventually trace their origins. During the Pingcheng era of the Northern Wei Dynasty, Buddhism flourished. According to the historic record, there were more than one hundred monasteries in the capital and more than two thousand monks and nuns (Wei 1974, CXIV114: 3025). The two thus far excavated monasteries of the time, the Siyuan (Figure 15) and Siyan (Figure 16) monasteries, differ from each other in configuration, and from the Yungang plan, even though both of them were patronized by Empress Dowager Wenming. Situated on the southern side of Mount Fangshan, north of Pingcheng and south of Empress Dowager Wenming’s Yonggu Mausoleum, the Siyuan Monastery, facing south, was constructed in the third year of Taihe (479) (Wei 1974, VII: 147). At the centre of the remains is a square platform in the rectangular courtyard (57 m east-west by 88 m north-south). In the courtyard, the entrance gate, stūpa, and Buddha halls are on the north-south axis. The base of the wooden stūpa is square (12 m in size). To the north of the stūpa is a Buddha hall of width 21 m, east to west (with seven bays), and depth 6 metres, south to north (two bays). It can be seen that the monastery has a qianta houdian configuration. In the north-west corner of the hall, remains were found of bed holes made of mudbricks.9 Archaeologists believed these were residential cells. Around the stūpa base is a square cloister, five bays wide, in whose interior wall (i.e. the exterior wall of the stūpa) were unearthed a small number of fragments of Buddha and bodhisattva images. This indicates that the cloister was enshrined with Buddha images along stūpa walls, so that the devotees were able to circumambulate the stūpa and worship the Buddha images. This is very much the 8 9 See Zhongguo shehui kexueyuan kaogu yanjiusuo & Hebeisheng wenwu yanjiusuo yecheng gongzuodui 2003: 3-6. For the detailed report of the excavation, see Datongshi bowaguan 2007. 223 The Global ConneCTions of Gandhāran arT Figure 15. Siyuan Monastery, Fangshan, Datong. (Plan: modified from Qian Guoxiang, Zhongyuan wenwu 4 [2017], pl. 1.) 224 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond Figure 16. Siyuan Monastery, Fangshan, Datong. (Plan: modified from Qian Guoxiang, Zhongyuan wenwu 4 [2017], pl. 2.) same as what is seen on the votive stūpa walls and shrines in the monasteries of the Greater Gandhāra area. After the Siyuan monastery was abandoned, it was never reused or reconstructed. Therefore, the monastery remains the original structure of the Northern Wei. Another monastery, the Siyan fotu (monastery) in Chaoyang city, Liaoning province, was also constructed, as the Beishi (History of Northern Dynasties) informs us, under the patronage of the Empress Dowager Wenming (Li 1974: 13: 496).10 The monastery is square in shape (49 m in size), at the centre of which is the base of a wooden stūpa. The plinths of the columns are arranged in four concentric squares. The outside square initially had twenty-eight plinths, outside of which there seems to have been a fifth 10 Li Yuqun suggested that the monastery was built during the Taihe era (AD 477-490). See Li 2009: 310. 225 The Global ConneCTions of Gandhāran arT square. Between the fourth and fifth squares there probably was a cloister around the monastery. Many images of Buddhas, bodhisattvas, disciples, and flying Apsaras were unearthed in the ruins of the stūpa, which indicates that images were on the stūpa walls, as in the Siyuan monastery, and many others in Greater Gandhāra. It can be seen that the Siyuan fotu and Siyan fotu monasteries differ in configuration. The latter has the early monastery configuration, centred on the stūpa, which is surrounded by a cloister or cells on its four sides. The Siyan monastery is more like the Yungang monastery, the earliest thus far excavated. All three of these monasteries were constructed during the Pingcheng era, but differ from one another in configuration. At the same time, they share one feature, i.e. they all have a wooden stūpa as the primary structure and the main object. The Siyan monastery had only a stūpa in the centre with a cloister surrounding it, which was the layout of a Buddhist monastery and traditional Chinese courtyard before the fourth century. It is interesting to note that both the Siyan and Siyuan monasteries were patronized Figure 17. Yongning Monastery, Luoyang. (Plan modified from Qian Guoxiang, Zhongyuan wenwu 4 [2017], pl. 3.) 226 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond by the Empress Dowager Wenming, but their layouts are different. This implies that when the new architectural layout of qianta houdian emerged in the fifth century, the traditional early monastery configuration did not immediately disappear. Caves 5 and 6, and caves 11 and 13 in Yungang also have qianta houdian configurations. This is the same as the configuration of free-standing monasteries in Central Plain China. The excavations of these monasteries led scholars to a better understanding of the cave composition. We were not previously clear why these caves were paired in this manner. The qianta houdian configuration in Pingcheng was followed by the Yongning monastery (Figure 17) in Luoyang. In A Record of Buddhist Monasteries in Luoyang (Yang 1978: 2; Wang 2014: 13 and 15-18), we are informed: The Yongning Monastery was constructed in the 1st year of the Xiping period (516) by decree of Empress Dowager Ling… Within the precincts [of the monastery] was a nine-storied wooden stūpa. Rising nine hundred Chinese feet above the ground, it formed the base for a mast that extended for another one hundred Chinese feet; thus together they soared one thousand Chinese feet above the ground, and could be seen as far away from the capital as one hundred li. In the course of excavating for the construction of the monastery, thirty golden statues were found deep underground; this was interpreted as an auspicious reward for the Empress Dowager’s conversion to Buddhism… On top of the mast was a golden jar inlaid with precious stones. It had the capacity of twenty-five piculs. Underneath the bejeweled jar were thirty tiers of golden plates to receive the dew. Golden bells hung from each of the plates. In addition, chains linked the mast with each of the four corners of the stūpa. Golden bells, each about the size of a stone jar, were also suspended from the linkworks… The stūpa has four sides, each having three doors and six windows. Painted in vermillion, each door had five rows of golden nails... North of the stūpa was a Buddhist hall, which was shaped like the Palace of the Great Ultimate... In the Hall was a golden statue of the Buddha eighteen Chinese feet high, along with ten medium-sized images – three of sewn pearls, five of woven golden threads, and two of jade. The superb artistry was matchless, unparalleled in its day. The monastery had over one thousand cloisters for the monks, both single cloisters and multilevel ones, decorated with carved beams and painted walls. The doors, painted in blue designs, had carved windows… The walls of the monastery were all covered with short rafters beneath the tiles in the same style as our contemporary palace walls... Under the archway were images of four guardians and four lions, adorned with gold, silver, pearls, and rare stones… The East and West Gates resembled the South Gate, except that the towers had only two stories… Travelers in the capital city often took shelter there. The monastery was burned down in the third year of Yongxi (AD 534). In 1979, the Institute of Archaeology at the China Academy of Social Sciences conducted an archaeological excavation at the monastery (Zhongguo shehui kexueyuan kaogu yanjiusuo 1996). According to their report, the monastery faced south and was rectangular in shape. The ruins of the stūpa base, square in shape, are in the centre but slightly towards the south. The stairs in the middle of each side of the stūpa base create a cruciform, a reminiscence of the Rawak stūpa discovered by Aurel Stein in 1901 (Stein 1907). Exquisite images were excavated around the base of the stūpa.11 To the north of the stūpa is a large rammed earth Buddha hall. Clearly, the Yongning monastery continued the Siyuan monastery layout of qianta houdian with the stūpa in front and the Buddha in the rear. The transformation in configuration, as stated earlier, took place in the south of Yecheng during the Eastern Wei-Northern Qi. The monastery began to have multiple courts as shown in the After several excavations, more than 2000 colored images of Buddha, bodhisattva and disciples were excavated. These images should be from the niches on the stūpa and cloister walls. See Qian 2007. 11 227 The Global ConneCTions of Gandhāran arT Figure 18. Zhaopengcheng Monastery, Linzhang. (Plan modified from Zhongguo shehui kexueyuan kaogusuo, yecheng kaogudui, Kaogu 7 [2013], fig. 2.) Zhaopengcheng monastery (Figure 18). The archaeological report shows that the monastery, square in shape, larger in scale than the Yongning monastery, faces south (Zhongguo shehui kexueyuan kaogu yanjiusuo & Hebeisheng wenwu yanjiusuo yecheng kaogudui 2010).12 Judging from its location and scale, archaeologists believed that this was a royal monastery. In the centre of the central court, slightly towards the south is the base of the wooden stūpa (Figure 19). At the northernmost end of the central court is a large structure, seven bays wide. It should be noted that the distance between the large structure and the ruins of the wooden stūpa is far. It is unclear 12 See also Zhongguo shehui kexueyuan kaogu yanjiusuo & Hebeisheng wenwu yanjiusuo yecheng kaogudui 2013a; 2013b. 228 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond whether there were structures in between the two; further excavations are needed. In the southeast and southwest corners of the monastery, the ruins of two square cloisters, 110 m in size, were found on each side. This indicates that these were two large courts. At the centre, slightly towards north, of the southeast court, the ruins of a Buddha hall, seven bays wide and five bays deep, were excavated. On each side of the Buddha hall is a winged cloister which connects with the outer cloister on the east and west sides. In addition, ruins of a large Buddha hall were unearthed in the southwest court as well. The scale is roughly the same. From what archaeologists unearthed and the configuration of the Buddha halls, it is clear that the monastery is centred on the stūpa with multiple courts. From the single court with stūpa and living Figure 19. Zhaopengcheng stupa, Linzhang. (Plan modified from Qian Guoxiang, Zhongyuan wenwu 4 [2017], pl. 5.) cells together, to the dominant qianta houdian configuration, and then to multiple courts, monasteries in Central Plain China underwent several transformations. One question that arises is: where did the dominant configuration of qianta houdian influence come from? As is widely accepted, early Chinese Buddhism was influenced by Central Asian Buddhism, but not directly influenced by Indian Buddhism. Central Asia’s contribution to the history of Buddhism lies largely in its role as an intermediary in the spread of the dharma to East Asia (Robinson et al 1996: 166). In the first century AD, a group of nomadic Indo-Scythians swept down from the north and gained control of northern India, Afghanistan, and a large part of central Asia from the Aral Sea east to the border of China, founding the Kushan dynasty (c. AD 32-375) (Robinson et al 1996: 167). The ideal location of the Kushan empire allowed it to control part of the Silk Road and to open it to cultural influences from all directions. The Kushans developed a synthesis of Graeco-Roman, Persian (Sasanian), and Indian styles in what appear to be among the first sculptures of the Buddha in human form. They also seem to be responsible for introducing the towering form of the Buddhist stūpa to India, topped by a tall, tapering spire, replacing the earlier hemispherical form. The Kushan taste in stūpa architecture continued to influence stūpa design throughout Asia, from the tall spires of Thai and Burmese cetiyas (caitya stūpas) to the multi-storey pagodas of China, Korea, and Japan (Robinson et al 1996: 167). During the Kushan period, Buddhism and its associated architecture spread from Afghanistan north-east into Central Asia and, ultimately, eastward to China and Japan, rather than directly from its north-eastern Indian origin (Ball 2008: 106). From the descriptions of the nine-storey pagoda at the Yongning monastery above, it can be said that the initial influence of the pagoda was from Central Asia. Central Asians were not only on the receiving end of outside influences during this period. They also were active exporting Buddhist ideas to other areas, most notably to China (Ball 2008). This is reflected not only by the Buddhist teachings themselves, but also from Buddhist image-making and architecture. The layout of Buddhist monasteries in central 229 The Global ConneCTions of Gandhāran arT Asia, through Xinjiang, eventually exerted great influence on that of Central Plain China where it became predominant. This is closely related to the active interaction between Buddhism in Central Plain China and Kucha. As Tang Yongtong pointed out, in the many kingdoms of the Western Regions, Kashmir, Khotan, and Kucha are three important strategic posts and Buddhism there had a great impact on Buddhism in China (Tang 1997: 265). The earliest qianta houdian plan, in fact, first emerged in Central Asia, as represented in Kara-Tepe and Fayaz-Tepe (Stavisky 1988).13 Kara-Tepe is a major Buddhist monastery site occupying a small hill in the north-west corner of old Termez, within the city walls. Fayaz-tepe, the site of another Buddhist monastery, lies about one kilometre north-west, outside the city walls (Rhie 1999). As Nancy Steinhardt suggested, third- to fifth-century cave-temples near Termez in Uzbekistan, also formerly part of the Kushan empire, similarly comprised a stūpa and Buddha hall at the focus of the courtyards (Steinhardt 2014: 121). Remains near Tumshuk in western Xinjiang between Kuche and Uzbekistan are another example of this kind of temple complex (Hambis et al 1961-1964). Also, it was suggested that the qianta houdian configuration of the Wushituer 烏什吐尔 and Xiahetuer 夏合吐尔 monasteries in Xinjiang were also influenced by that of Kara-tepe in central Asia (Lin 2018: 42). It can be seen that the roots of the qianta houdian configuration in Central Plain China lie in the monasteries of thirdcentury in Xinjiang, but the deeper roots lie in southern Uzbekistan of central Asia. The deepest roots lie in Greater Gandhāra. The earliest monastery discovered in Yungang traced its roots all the way to Taxila. Likewise, the multiple court plan shown in the Zhaopengcheng monastery traces its roots to the Greater Gandhāra area as well. Concluding remarks The above analysis not only helped us further understand these monasteries themselves, but also, more importantly, made clear the connections to one another in the process of developments in Buddhism dissemination from west to east. In the early times, there was no such configuration of Buddhist monasteries as a self-contained monastery with a main stūpa and vihāra within the high-walled space. There was neither stūpa, nor the monastery cells, let alone Buddha or lecture halls. Buddhist monks spent much of their time in the Bamboo garden near Rājagṛiha, in the Jetavana near Śrāvastī, in the Mango Grove near Vaishālī and in the Deer Park near Benares. The vināya texts also did not mention either the stūpa or the vihāra. We do not have evidence of a stūpa earlier than the reign of Aśoka in the middle of the third century BC, and we cannot find an example of a walled, self-contained monastery until the first or second century AD (Marshall 1951: vol. 1, 232). The stūpa did not become an object of veneration until the time of Aśoka who was one of the most famous kings in the history of India, and was portrayed as a great devotee and supporter of the Buddhist sangha. He was a builder of stūpas. Soon, the stūpa, with or without relics, began to be regarded as the most outstanding and ubiquitous emblem of Buddhism and worthy of worship for its own sake. After that a stūpa almost became the symbol of the faith. To erect a stūpa of any shape or form is to build religious merit. This is the same in China. As Ge Hong (AD 283-343) pointed out, a stūpa is equal to a monastery. Consequently, building a stūpa is the same as building a monastery. In early times, the main structure in a Buddhist monastery was the stūpa. And the monasteries were often named after the number of storeys of a stūpa, such as the Five-Storey Monastery where Dao’an (AD 312-385) lived and the Five-Storey Grand Monastery in Pingcheng. When Buddhism developed further and flourished, the single stūpa was far from enough for liturgical functions. The monasteries were not just for worshipping, they became the venue for lectures and Dharma teachings as well. Image chapels and lecture halls therefore emerged in the fourth century. According to the record of Weishu, in the first 13 See also Stavisky & Mkrtychev 1996. 230 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond year of Tianxing (AD 398) the Five-Storey stūpa, the Mount Grdhrakuta (Vulture Peak) and Mount Sumeru Halls were built. In addition, lecture and meditation halls and monks’ cells were constructed (Wei 1974, CXIV114: 3030). It can be seen that the stūpa was built first and then Buddha halls, halls for lectures and meditation, as well as the living cells being added accordingly to the main stūpa. Evidently, the monastery plan was transformed from a single stūpa structure to that of stūpa, Buddha hall, and lecture hall on the north-south axis, a qianta houdian configuration. In addition to the stūpa in the centre, image chapels and lecture halls are now equally important. Image worshipping became an important part of liturgy for Buddhist devotees and image halls became an important component of Buddhist monasteries. The prevailing configuration of qianta houdian was eventually exported east to the Baekje kingdom. The Buddhist monastery remains such as Chõngnimsa monastery and others during the Baekje (Paekche) period (18 BC-AD 660) in ancient Korea show the configuration of the monasteries is that of stūpa, Buddha hall, and lecture halls on the north-south axis, and this was probably influenced by the configuration of Buddhist monasteries in Central Plain China (Su 2011: 243). Our final question to be answered is why the Yungang plan came directly from Taxila? As is widely known, Emperor Taiwu (r. 424-52) of the Northern Wei had frequent contacts with Buddhist kingdoms from the Western Regions such as Shanshan, Yanqi, Kucha, Khotan, Sogdiana, and Kashmir, where the construction of Buddhist monasteries thrived (Su 1990: 123-25). Tanyao, the chief administrator śramana, was the main architect in charge of the excavation of the rock-cut caves of Yungang. Naturally he was involved in the construction of monasteries above the caves as well. He was responsible for the entire Yungang complex including the translation of the sūtras. One of the sūtras, the Sūtra of the Miscellaneous Treasures (Zabaozang jing), translated by Kikkāya and Tanyao, is an important source to answer our questions about why the earliest monastery thus far excavated is directly connected with Taxila, but not with the traditional Chinese structure, central Asia, or even with other Greater Gandhāra areas outside Taxila. First of all, many stories narrated in the sūtra happened in Greater Gandhāra. Secondly, and more importantly, the translator Kikkāya is actually believed to have been from there as well (Willemen 1992: 507-15). Furthermore, I have argued elsewhere that the sūtras translated by Tanyao and others played an essential role in image-making in the excavation of the caves (Yi 2018: 11, 46). The subjects of the caves are primarily from these sūtras. For instance, the stories from the Zabaozang jing are visually portrayed in caves 9 and 10. Evidently, words and images are closely associated in Yungang and the primary inspiration for builders was from the sūtras translated by Kikkāya and others. If imagemaking was directly associated with Greater Gandhāra, it is not difficult to imagine that monastery building could be from there as well. It is therefore not unreasonable to believe that there was a direct connection between Yungang and Taxila, and Buddhist monks played a pivotal role in the dissemination of Buddhism in many aspects. Tanyao and Kikkāya, both of whom came from the regions that connected and transmitted different cultures, were key figures in the construction of the Yungang complex. Chinese Buddhist monasteries were not only directly influenced by those in Taxila, but also by those in Central Asia, in the old Termez area, which we tended to neglect in the past. Now new archaeological findings offer us opportunities to revisit the literary sources and hard evidence. The ‘[Buddha] halls on the mountain and over water, and the smoke [of the incense]-filled temples’ mentioned in the first literary record of Yungang had never been comprehended or corroborated before the new excavation (Li 2007: 316). It is just that John Marshall did not live to see that Gandhāran influence could have been spread as far east as Pingcheng, the capital of the Northern Wei dynasty in China. 231 The Global ConneCTions of Gandhāran arT References Ball W. 2008. The Monuments of Afghanistan: History, Archaeology and Architecture. London: I.B. Tauris. Chavannes E. 1909-15. Mission archéologique dans la Chine septentrionale (Archaeological Expeditions in Northern China). 13 Volumes. Paris: E. Leroux. Chen Yuan. 1980. Ji Datong Wuzhoushan shikusi. Pages 398-409 in Chen Yuan. Chen Yuan xueshu lunwenji, vol. 1. Beijing: Zhonghua shuju. Datongshi bowuguan. 2007. Datong Beiwei Fangshan Siyuan fosi fajue baogao. Kaogu 4: 4-26. Hambis L. et al. 1961-1964. Toumchouq. 2 volumes. Paris: Libraire Adrien-Maisonneuve. Hebeisheng Wenhuaju Wenwu Gongzuodui. 1966. Hebei Dingxian chutu beiwei shihan. Kaogu 5: 252-59. Itō C. 1906. Shina sansei Unkō no sekkutsuji. Kokka 17, no. 197: 431-35 and no. 198: 483-94. Jongeward D., Errington E., Salomon R., and Baums S. 2012. Gandharan Buddhist Reliquaries. Seattle: Early Buddhist Manuscripts Project. Junjirō T. and Kaigyoku W. (eds). 1924-32. Taishō Shinshū Daizōkyō. 85 volumes. Tokyo: Taishō issaikyō kankōkai. Li Chongfeng. 2014. Buddhist Archaeology from India to China. 2 vols. Shanghai: Shanghai Chinese Classics Publishing House. Li Daoyuan. 2007. Shuijing zhu jiaozheng, annotated by Chen Qiaoyi. Beijing: Zhonghua shuju. Li Yanshou. 1974. Beishi. Beijing: Zhonghua shuju. Li Yuqun. 2009. Sui Tang yiqian zhongguo fojiao siyuan de kongjian buju ji yanbian in Bianjiang minzu kaogu yu minzu kaoguxue jikan. Beijing: Wenwu chubanshe. Liaoningsheng Wenwu Kaogu Yanjiusuo and Chaoyangshi Bowuguan (eds). 2007. Chaoyang Beita – Kaogu fajue yu weixiu gongcheng baogao. Beijing: Wenwu chubanshe. Lin Li. 2018. Xiyu gufosi. Beijing: Kexue chubanshe. Marshall J. 1951. Taxila: An Illustrated Account of Archaeological Excavations Carried out at Taxila under the Orders of the Government of India between the Years 1913 and 1934. Cambridge: Cambridge University Press. Mizuno S. and Toshio N. 1951. Unkō sekkutsu: seireki goseiki ni okeru Chūgoku hokubu Bukkyō kutsuin no kōkogakuteki chōsa hōkoku: Tōhō Bunka Kenkyūjo chōsa (Yungang: The Buddhist Cave-Temples of the Fifth Century A.D. in North China: detailed report of the archaeological survey carried out by the Mission of the Tōhōbunka kenkyūsho 1938–45). 16 vols. Kyoto: Kyōto Daigaku Jinbun Kagaku Kenkyūjo. Qian Guoxiang. 2007. Beiwei Luoyang yongningsi suxiang de chubu yanjiu. Pages 446-59 in Hanwei Luoyangcheng yizhi de yanjiu. Beijing: Kexue chubanshe. Rhie M. 1999. Early Buddhist Art of China and Central Asia. Leiden: Brill. Robinson R., Willard J., Wawrytko S. and Ṭhānissaro Bhikkhu. 1996. The Buddhist Religion: A Historical Introduction. Religious Life in History Series. Belmont, CA: Wadsworth Publishing Company. Salomon R. 2018. The Buddhism Literature of Ancient Gandhara. Somerville, MA: Wisdom Publications. Stavisky B.J.1988. Kara-Tepe in Old Termez (Southern Uzbekistan): Summary of the Work Done in 19781982. Pages 1391-1405 in G. Gnoli and L. Lanciotti (eds), Orientalia Iosephi Tucci Memoriae Dicata. Serie Orientale Roma, LVI, 3 vols. Rome. Stavisky B.J. and Mkrtychev T. 1996. Qara-Tepe in Old Termez: On the History of the Monument. Bulletin of the Asia Institute 10: 219-32. Stein A. 1907. Ancient Khotan. 3 Volumes. Oxford: Clarendon Press (reprint New Delhi, 1981). Steinhardt N. 2014. Chinese Architecture in an Age of Turmoil, 200-600. Honolulu: University of Hawaiʻi Press. Su Bai. 1990. Pingcheng shili de jiju he Yungang moshi de xingcheng yu fazhan in Su Bai Zhongguo shikusi yanjiu. Beijing: Wenwu chubanshe. Su Bai. 1996. Dajin xijing Wuzhoushan chongxiu dashikusi bei jiaozhu. Pages 52-75 in Su Bai, Zhongguo shikusi yanjiu. Beijing: Wenwu chubanshe. Su Bai. 2011. Donghan Weijin Nanweichao fosi buju chutan in Su Bai, Weijin Nanbeichao Tang Song kaogu wengao jicong. Beijing: Wenwu chubanshe. 232 Joy yi lidu: Cross-CulTural buddhisT monasTery ruins on The silk road and beyond Tang Yongtong. 1997. Weijing Nanbeichao fojiao shi. Beijing: Beijing daxue chubanshe. Wang Yi-tung (trans.). 2014. Record of Buddhist Monasteries in Lo-Yang. Princeton, New Jersey: Princeton University Press. Wei Shou. 1974. Weishu. Beijing: Zhonghua shuju. Willemen C. 1992. A Chinese Kṣudrakapiṭaka (T. IV. 203). Asiatische Studien - Études Asiatiques 46/1: 507-15. Yang Xuanzhi. 1978. Luoyang qielan ji. Shanghai: Guji chubanshe. Yi J.L. 2018. Yungang: Art, History, Archaeology and Liturgy. New York: Routledge. Zhang Qingjie. 2016. Yungang shiku kuding xiqu beiwei fojiao siyuan yizhi. Kaogu xuebao 4: 533–562. Zhang Qingjie, Li Baijun and Jiang Weiwei. 2011. Shanxi Yungang shiku kuding beiwei fosi yizhi. Pages 127-30 in Guojia wenwuju (ed.) 2010 nian Zhongguo zhongyao kaogu faxian. Beijing: Wenwu chubanshe. Zhongguo shehui kexueyuan kaogu yanjiusuo. 1996. Beiwei Luoyang yongninsi. Beijing: Kexue chubanshe. Zhongguo shehui kexueyuan kaogu yanjiusuo and Hebeisheng wenwu yanjiusuo yecheng gongzuodui. 2003. Hebei Linzhangxian yecheng dongwei beiqi fosi taji yiji de faxian yu yanjiu. Kaogu 10: 3-6. Zhongguo shehui kexueyuan kaogu yanjiusuo and Hebeisheng wenwu yanjiusuo yecheng kaogudui. 2010. Yecheng yizhi zhaopengcheng dongwei beiqi fosi yiji kantan yu fajue. Kaogu 7: 31-42. Zhongguo shehui kexueyuan kaogu yanjiusuo and Hebeisheng wenwu yanjiusuo yecheng kaogudui. 2013a. Hebei Yecheng yizhi zhaopengcheng beichao fosi yu beiwuzhuang fojiao zaoxiang maicangkang. Kaogu 7: 49-68. Zhongguo shehui kexueyuan kaogu yanjiusuo and Hebeisheng wenwu yanjiusuo yecheng kaogudui. 2013b. Hebei Linzhangxiang yecheng yizhi zhaopengcheng beichao fosi yiji 2011-2012 nian de fajue. Kaogu 12: 25-35. 233 (234-247): DOI: 10.32028/9781789696950-9 The sinicization and secularization of some Graeco-Buddhist gods in China1 Juping Yang Introduction It is generally admitted that the sinicization or ‘Chinization’ of Buddhism was a long process. However, there is still much discussion in Chinese academia about when the process began and was largely realized, and how and to what extent it involved adaption to and fusion with Chinese traditional culture. I am well aware that I have insufficient expertise to take part in such specialist discussion. The purpose of this paper is therefore to offer my macroscopic observations on the evolution of Graeco-Buddhist figures in China, and to try to analyze the background and the context of these changes. As we know, there are different opinions about the foreign elements in the development of Gandhāran art. The hypotheses of Greek and Roman origins predominate. Even among these there are disagreements about which tradition is more important (see, for example, the contributions by Stoye and Stewart in this volume). However, no matter which is given priority, the fact that Roman culture was in many respects the continuation and development of Greek culture is undeniable. So the essence of Gandhāran art is a fusion or combination of Indian Buddhism with the plastic art and religious ideas of the GraecoRoman world. It is therefore not surprising that some deities from Graeco-Roman mythology became the prototypes for Buddhist figures in Gandhāran art, such as Apollo-Buddha, Herakles-Vajrapāṇi, NikeApsara, Atlas, Helios-Surya, Tyche-Hariti, and so on. Nevertheless, their iconography was not static. With the spread of Gandhāran art out of its homeland in the northwest of the Indian subcontinent, the Buddhist figures had to be adapted to the new cultural environment. In China the change appears more conspicuous and fundamental. For example, the Buddha styled on Apollo provided the imagery for Chinese emperors of the North Wei Dynasty (北魏, AD 386-534), and even the personification of Empress Wu Zetian (武则天) of the Tang Dynasty (AD 618-907). The ‘Atlas’-style man appeared in the tombs of non-Buddhists. The Nike-Apsara first of all turned into flying deities in Heaven, Feitian (飞天), and eventually went so far as to be assimilated by the foreign Nestorian Christians in the Yuan Dynasty (AD 1271-1368). However, the most characteristic example is Herakles-Vajrapāṇi who not only lost his vajra and the position of bodyguard of the Buddha, but also unfortunately degenerated into a warrior guarding the tombs of non-Buddhists in Tang Dynasty. This paper will consider just a few more similar cases in order to illustrate the particular multi-cultural syncretism that occurred in China. From Herakles-Vajrapāṇi to the Worrier Figurine2 Herakles is a demigod and hero in Greek mythology. Why would he be combined with the Vajrapāṇi of Buddhism? One answer seems to point to Alexander the Great and his expedition to India. It is pointed out that he and his father claimed Herakles and Zeus as the ancestors and progenitors of their family.3 This paper is one of the preliminary outputs of research for the key project sponsored by the National Foundation for Social Science of China: Hellenistic Civilization and the Silk Road (15ZDB059). Many thanks to Dr Peter Stewart for inviting me to participate the conference in March 2019 and for his hard work on correcting my paper. I am also indebted to Professor Daniel Waugh for his precious directions about its modification. I am also grateful to those who offered me their generous helps in various ways, such as Professor Osmund Bopearachchi and Mr Joe Cribb. 2 The Chinese scholars Xing Yitian (I-Tien Hsing) and Xie Mingliang have made in-depth studies on this subject. See Hsing 2005: 103-154; Xie 1997: 32-53. 3 Arrian, Anabasis of Alexander, 4.11.6-7; Kallisthenes says that Alexander is ‘a son of Philip, a descendant of Herakles and of Aiakos, whose forefathers came from Argos to Macedonia, and have continued to rule the Macedonians not by force but in accordance with custom’. Aiakos in Greek mythology is also the son of Zeus. Alexander’s mother Olympias claimed to be a descendant of the Aeacidae. So Alexander also established descent from Zeus through his mother. Plutarch also states: ‘As for the lineage of Alexander, on his father’s side he was a descendant of Herakles through Karanos, and on his mother’s side a descendant of Aiakos 1 234 JupinG yanG: The siniCizaTion and seCularizaTion of some GraeCo-buddhisT Gods in China The images of Herakles and Zeus appeared earlier on the coins of Amyntas III (392390 BC), Perdikkas (364-359 BC), and Philip II (359-336 BC) respectively, before the conquests of Alexander the Great. On the way to the east, Alexander minted his own coins, the Herakles/Zeus type (Figure 1). On the obverse is the image of Herakles’s head wearing the lion’s scalp helmet and on the reverse is Zeus seated on his throne. Although he died suddenly in 323 BC, the Figure 1. Silver tetradrachm of Alexander the Great, minted at Babylon, heritage of his empire did not vanish and 325-323 BC. (From the Andritsaena Hoard, Elis.) Obverse: head of his coins still circulated in the Hellenistic Herakles. Reverse: seated Zeus. New York, American Numismatic Society, 1944.100.80601 (Photo: copyright ANS: http://numismatics.org/ world created by him, and even beyond. collection/1944.100.80601.) Seleukos, one of the Diadokhoi, inherited the largest legacy in Asia. Although Seleukos abandoned India soon, in c. 305 BC, and the two satrapies in the upper provinces declared their independence from the Seleukid Kingdom around the middle of the third century BC, one of the new kingdoms, the Graeco-Bactrian kingdom became the centre of the civilization of the Hellenistic Far East. In the early second century BC, the Greeks in Bactria crossed the Hindu Kush and entered north-west India, and remained there for nearly two centuries. These so-called Indo-Greeks tried to retain their Greekness, but at the same time they were obliged to accept local culture, including Buddhism and local deities. Some of them apparently even converted to Buddhism and become Buddhists, notably the famous Indo-Greek King Menander.4 Herakles appeared in various forms on the coins of some Indo-Greek kings, for example Euthydemos I and his son Demetrios.5 Patron deities generally appear on the reverses of Hellenistic coins. Therefore, besides Herakles other Greek deities such as Zeus, Athena, Apollo, the Dioskouroi, Poseidon, Artemis, Nike, Helios, Hermes, etc. also appeared on the coins issued by these Greek kings in Bactria and India. The worship and popularity of these deities not only promoted their fusion and identification with Buddhist figures, but also contributed to the development of Gandhāran iconography. The identification of Herakles with Vajrapāṇi is a typical example. He was known for his valor, his fearlessness in the face of hardship and danger, and his supernatural strength displayed in the twelve labors. So it is natural that he was accepted by Buddhists and was identified with Vajrapāṇi (the bodhisattva with ‘Vajra in [his] hand’) who became the bodyguard and guide of the Buddha in the early art of Gandhāra. This radical change can be seen in the sculptural group of the Buddha and his attendants at Tapa-e Shotor at Haḍḍa. Herakles still wears his lion-skin draped on his left shoulder, passing over his back and reappearing over his lap, while his club had become Vajrapāṇi’s thunderbolt or diamond pestle (Figure 2).6 However, the similarity in the sitting position between the Vajrapāṇi and the Herakles on the coin of Euthydemos is apparent (Figure 3). through Neoptolemos; this is accepted without any question’ (Plutarch, Alexander, 2.1). According to Greek mythological tradition, Karanos, the descendant of Herakles, was the first king of Macedonia. Isokrates exhorted Philip II to conquer Persia like his ancestor Herakles who had destroyed Troy. (To Philip, 109-115). All these sources attest to contemporary Greeks’ acceptance of the story that Philip and Alexander were the descendants of Zeus and Herakles. It is therefore understandable that the images of Zeus and Herakles appeared on coins of kings of Macedonia. (All citations of English translations from the Loeb Classical Library.) 4 The tradition was that he once had a long discussion on the Buddhist doctrine with a master of Buddhism, Nagasena, and finally he was determined to abandon his throne and become a lay Buddhist. For the details of his story in the Milindapañha and the arguments around him, see Yang 2016: 111-22. 5 See Bopearachchi 1991: pls. 2-3 (Euthydème I, 1-16, 35); pl. 4-5 (Demetrios I, 1-3). 6 See Tarzi 1976: 394-396, figs. 9, 10 and 11; Boardman 2015: fig. 122; F. Tissot, ‘Afghanistan ix. Pre-Islamic Art’, in Encyclopædia Iranica, I/5: 544-547, fig. 3/4; an updated version is available online at <http://www.iranicaonline.org/articles/afghanistan-ixpreislamic-art> (last accessed 16th January 2020). 235 The Global ConneCTions of Gandhāran arT Figure 2. Clay sculpture of Herakles-Vajrapāṇi in niche V2 at Tapa-e Shotor, Haḍḍa, Afghanistan (now destroyed). Second to fifth century AD. (Photo: Francine Tissot, courtesy of Professor Osmund Bopearachchi.) Figure 3. Reverse of silver tetradrachm of Euthydemos, showing Herakles sitting on a rock covered by the lion-skin and with a club resting on his right leg; c. 230-190 BC. New York, American Numismatic Society, 1995.51.28. (Photo: copyright ANS, http:// numismatics.org/collection/1995.51.28.) Similar figures of Herakles-Vajrapāṇi could be seen in the early period of Gandhāran Buddhist art (Figures 4 and 5). From these images, we can see that Herakles had become the Buddha’s guardian. Sometimes he wears Herakles’s ‘helmet’ of of the lion-scalp; sometimes he only has the vajra pestle, converted from Herakles’s club; and sometimes he is nude or half-naked. It is this kind of Herakles-Vajrapāṇi that was spread into China and was transformed gradually into a Chinese folk deity. Two changes are worth noticing. One is his position and role in the Buddhist temples and grottoes. He rarely stood side by side with the Buddha as he used to do in earlier Gandhāran art. Moreover, his symbol, the vajra, was sometimes lost or become a sword. The pelt helmet generally remained but it is hardly recognizable whether it is of lion or tiger. He seems to become a Lokapāla (heavenly king) and often stands at the entrances of temples. We can see such a change clearly from Cave no. 205 of Dunhuang Grottoes and Maijishan Grottoes (Figures 6 and 7).7 Why might this change have taken place? The reason seems to be explaind by his original role. Although he was a guardian deity (Dharmapāla) of the Buddha and the Dharma, nevertheless, with the deepening and expanding of sinicization of Mahāyāna Buddhism, his rank became lower and lower, so that his position was replaced by the disciples of the Buddha Śakyamuni and the other Buddhas and bodhisattvas, such as Dīpaṅkara and Maitreya. Another factor is his secularization. His clay figurines appeared as protectors in the tombs of nonBuddhists in the Tang Dynasty (AD 618-907) (Figures 8, 9, and 10). This means he was accepted by Chinese as a secular deity as early as this period. Even so, these figurines still remain some features of Herakles-Vajrapāṇi, such as the lion/tiger-scalp helmet, the paws tied on chest, and even the club that reappeared in his hand in one tomb. There is some debate among in Chinese researchers as to the identity of the helmets, namely, whether they are scalps of lions or tigers. Some scholars think that the headgear of the warrior figures originated from warriors wearing the pelts of beasts in Tubo (modern Chinese Tibet, and parts of Qinghai and Gansu) and Nan Zhao (modern Yunnan). Even if this hypothesis sounds somewhat rational, the images of the warrior from Tubo might also have been influenced by Central Asian and Indian Buddhism, that is to say, by the Herakles-Vajrapāṇi of Gandhāran art, as Hsing has suggested (Hsing 2005: 139-145). 7 For the details of the clay statue, see Hsing 2005: 124-126, fig. 31. 236 JupinG yanG: The siniCizaTion and seCularizaTion of some GraeCo-buddhisT Gods in China Figure 4. Herakles-Vajrapāṇi with the Buddha, from a Gandhāran scene of the Buddha’s First Sermon; c. second century AD. New York, Metropolitan Museum of Art, inv. 1980.527.4. (Photo: Metropolitan Museum of Art, CC Zero Licence.) Figure 6. Clay sculpture of Vajrapāṇi, Maijishan Grottos, Gansu, China. Bei Zhou (北周, AD 557-581). (Photo: author, August 2007.) Figure 5. Herakles-Vajrapāṇi with the Buddha, in a Gandhāran scene of the Dīpaṅkara Jātaka from Jamalgarhī; c. second century AD. Kolkata, India Museum. (Photo: courtesy of the Warburg Institute, London.) Figure 7. Vajrapāṇi (right) wearing the pelt of a tiger in Dunhuang Grottos, Cave no. 205, early Tang Dynasty (AD 618-907). (Photo: after Liu 2003, 129, pl. 112.) 237 The Global ConneCTions of Gandhāran arT Figure 8. Tang Dynasty (AD 618-907) tricolour warrior figurine, excavated in the village of Hongqing, Baqiao district, Xi’an. Xi’an Museum. (Photo: author, July 2013.) Figure 9. Pottery warrior figurines from a tomb of the Tang dynasty, Changzhi, Shanxi Province. (Photo: after Shen 1962, pl. 8.1-2.) Figure 10. Pottery warrior figurine from a tomb of the Tang dynasty, Xian County, Hebei Province. (Photo: after Wang 1990, pl. 4.1.) Why should the Buddhist Vajrapāṇi become a warrior figurine guarding the tombs of lay people? The answer must been sought in the Chinese cultural tradition, especially in funerary culture. Strictly speaking, in China, there were no discrete religions like Buddhism, Judaism, Christianity, and Islam. Confucianism was the soul or core of Chinese traditional culture, but it has never been a religion. It was a kind of secular ideology, or an ethical philosophy that provided directions for the administration of the state and personal morality and social behavior; in another words, it was a way of life for Chinese people. Confucius was a man not a god. He was a teacher, a master, not a founder or a preacher of one religion. His ideas were favored by the emperors of China in the Han Dynasty and gradually became the principles of Chinese political and social life. So when Buddhism spread into China it faced a vast challenge to make itself adaptable to the needs of Chinese society. It was not until the Tang Dynasty that the sinicization of Buddhism was realized. It is not a coincidence that Vajrapāṇi become a warrior guarding the tomb at this time. Since Vajrapāṇi had the duty and power to protect the Buddha, it was entirely feasible that he could protect the dead in the nether world. From Nike-Apsara to Feitian (飞天) The Apsara was originally a female sprite of water and cloud in Indian mythology. Its earliest extant sculptural representation is probably that of the stūpa at Bharhut in central India, dated to the second century BC (Figure 11). At that date the anthropomorphic image of the Buddha had not yet been created, so the two winged Apsaras fly face to face above a stūpa, one of the symbols of the Buddha, holding a garland and a 238 JupinG yanG: The siniCizaTion and seCularizaTion of some GraeCo-buddhisT Gods in China Figure 11. The winged, flying Feitian (Apsaras) holding garland and palm, on a relief Bharhut, India, second century BC. Freer Gallery of Art and Arthur M. Sackler Gallery, Washington. (Photo: author, March 2014). palm branch respectively in their hands. It seems possible to discern some similarities between them – and particularly the various apsaras in Gandhāran art – and Nike, the Greek goddess of Victory. Can we assume that there are links between them? Some evidences seem to support that possibility.8 At first, we should notice the basic features of Nike’s image in classical art. She has two wings, and often holds a wreath with one or both hands when she dedicates it to other Greek gods and goddess, such as Zeus or Demeter. On the coins of Hellenistic kings, she frequently stands on the extended hand of Zeus and flies to crown him with a wreath. We can also see such examples on the ‘Poros’ medallion issued by Alexander, and the coins of Seleukos I and some Indo-Greek kings (Figure 12).9 It is noteworthy that the Apsara in Gandhāran art played similar roles. Generally, she flies above the head of the Buddha and has two wings, as we see both in Gandhāra and Central Asia (Figure 13), and in the temple of Tumshuk in Xinjiang, China (relief preserved in the Musée Guimet in Paris). However, with her spread into China, her image was changed. For example, she became a dancing-girl or a musician with an instrument such as flute, pipa, or sheng, instead of the garland, palm branch, lotus branch, or similar. Her Figure 12. Nike and Zeus on the reverse of a silver tetradrachm of Seleukos I, minted at Sardis, 282-281 BC. New York, American Numismatic Society, 1944.100.78148. (Copyright: ANS, http:// numismatics.org/collection/1944.100.78148). John Boardman has also noticed the similarity between the Apsaras and the classical Victory (Nike), and attributes it to influence from the west. See Boardman 2015: 132-133, fig. 70. 9 On the coins of some Graeco-Bactrian and Indo-Greek kings, Nike stands holding wreath. See Bopearachchi 1991: 379, figs. 10 (Antimaque I, 5), 14-15 (Antimaque II, 1), 22 (Eucratide I, 22-23), 31-33 (Menandre I, 22, 27, 31-34), 37 (Straton I, 29-30), 44 (Philoxene 12), 48 (Epandre, 2-3), 49 (Menandre II, 1-2), 50 (Artemidore, 5-6), 51 (Archebios, 11). Some coins show Nike standing on the right hand of Zeus and crowning him. See Bopearachchi 1991: 380 and figs. 26 (Heliocles, 3.4), 49 (Menandre II, 4), 39 -40 (Antialcidas, 1-5, 9, 12, 13). 8 239 The Global ConneCTions of Gandhāran arT feet, even her legs, were gradually lost and became a long flying skirt fluttering in the wind (Figures 1417).10 Although there needs to be further discussion about whether Nike actually influenced the evolution of the image of the Apsara in Gandhāran art, it is a fact that the Apsara iconography originated in India and finally become the Feitian in China, a flying dancing-girl or a musician in the Buddhist paradise. In the Yuan Dynasty (AD 12711368), the Buddhist Feitian even reappeared on the funerary stelai of Nestorian Christians in the city of Quanzhou, in Fujian Province in the south of China (Figures 18 and 19). Figure 13. Gandhāran relief showing the Buddha with Brahma, Indra, and Apsaras; c. second century AD. Art Institute of Chicago, inv. 1995.263. (Photo: Art Institute of Chicago, CC0 Licence.) Figure 14. Painting of Apsaras in Cave no. 7, the Western Thousand-Buddha Grottoes, Anxi Yulin Grottoes. North Wei Period (AD 386-534). (Photo: after Dunhuang Academy 1997; fig. 204). In the evolution of Nike-Apsara to Feitian, we can see three evident features. The first concerns her identity. Generally, she preserved her female gender. Of course, there are some Apsaras who look male or masculine in Chinese Grottos, yet they are only exceptional cases.11 Figure 15. Painting of an Apsara in Cave no. 285, Dunhuang Mogao Grottoes. Western Wei period (AD 535-556). (Photo: after Dunhuang Academy 1982a, fig. 138). 10 Note that such Apsaras appeared firstly in Dunhuang, Yulin Grottoes in the North Wei (AD 386-534), then spread into the central China. Also see Sizer 1925: plates (‘stone votive stele’, left); Priest 1930. 11 Such Apsaras mainly appeared in Dunhuang Grottos and Yungang Grottoes. See Zhao 2008: 81, 125. 240 JupinG yanG: The siniCizaTion and seCularizaTion of some GraeCo-buddhisT Gods in China Figure 16. Sandstone relief of flying Apsara from Tianlongshan Grottoes, Taiyuan, Shanxi Province. Northern Qi Dynasty, c. AD 570. Kansas City, Nelson-Atkins Museum of Art (Photo: Nelson-Atkins Museum of Art, https://art.nelson-atkins.org/ objects/22804). Figure 17. Earthenware tile with image of Apsara/ Feitian, from Xiuding-si pagoda, Henan Pronvince. Tang Dynasty, seventh century AD. New York, Metropolitan Museum of Art, inv. 65.178.2 (Photo: Metropolitan Museum, CC0 licence). Figures 18 and 19. Flying Feitian (Asparas) on the funerary stelai of Nestorians, Yuan Dynasty (AD 1271-1368). Quanzhou Maritime Museum, China. (Photos: author, December 2015). The second is her intimate relation with the Buddha. Although she ceases to be the flying sprite (the celestial maiden) who offers a wreath or garland to the Buddha, she still appears above or around him. Her role seems changed. She becomes a musician, or a dancer whose duty is to celebrate the happiness and joy of Sukhāvatī, the Pure Land or Western Paradise. The third is her posture of flying in the sky. Actually, her flying posture did not change even if her legs and wings were gradually lost in China. Apart from the two more characteristic cases above, we could identify some other, similar examples of iconographical transformation, such as the Atlantean figures of Gandhāra art and the god of sun, Surya. Although the manner of their transformation and adaption in China cannot be detected clearly, it seems certain that they have a direct relationship with some Greek deities and Buddhist figures. Their evolution in China also reflects the sinicization and secularization of Gandhāra art. 241 The Global ConneCTions of Gandhāran arT From Atlas to Atlantean Figure (力士) Atlas once supported the Titans in their war with the Olympians. After their defeat, Zeus condemned him to hold up the sky (Ouranos) on his shoulders at the western edge of the earth (Gaia) and prevent the two from embracing. So in the classical art, he often carries a globe or a celestial sphere on his back. Many similar images appear in Gandhāran Buddhist sculptures (see also Stewart’s contribution to the present volume). Generally, the ‘Atlas’-figure squats supporting the stūpa on his shoulders or with two hands upwards, and sometimes he has two wings on his back (Figure 20; cf. Figure 3 in Stewart’s contribution to the present volume). Later his image was brought to China, and he became a strong Buddhist figure who could support the very high Buddhist tower (塔, Chinesestyle stūpa) (Figure 21). His image was also borrowed for the tombs of lay people to support the stone bed on which the coffin was laid. By this time, his role had become that of a strongman (力士, atlantean figure), Figure 20. Gandhāran ‘Atlas’ figure in schist; c. second century AD. Los with no wings, with his two hands held up to Angeles County Museum of Art, inv. M.71.73.136 (Photo: LACMA Public Domain image.) support the stone bed (Figure 22). From Helios-Sūrya to Kingly Figure a b Some Chinese scholars have noted the similarities between the murals of the Sun god (日天) (Figure 23) and the moon god ( 月天) (Figure 24) in China and the relief of Sūrya (the deva of the Sun) discovered in Bodh Gaya in India (Figure 25). They further identify them with the Greek sun god Helios, who rides a chariot drawn by four horses every day from dawn to sunset (Figure 26) (Zhang 2009: 38-40). Similar figures have also been discovered on some pieces of brocade of the Sui and Tang Dynasties in the north-west of China. The most representative one is from Dulanreshui region of Qinghai Province. Figure 21 a and b. Atlantean figures supporting stūpas with their A kingly figure wearing a crown sits on hands. From Cave no. 10, Yungang Grottos. Northern Wei period, AD 471-494. (Photo: author, September 2017.) a chariot drawn by six horses, three on 242 JupinG yanG: The siniCizaTion and seCularizaTion of some GraeCo-buddhisT Gods in China either side, pointing in opposite directions (Figure 27). It is noted, however, that such figures had fused the Greek Helios with other cultural elements from Iran, India, and China (Zhao 1995: 179-183, fig. 4.3). This link seems rational and logical, but we need more evidence to verify this hypothesis because the differences between the images of Sūrya and Helios on the Chariot is too distinct. Nevertheless, transitional evidence might illustrate the mystery of this transformation. On the coins of some Indo-Greek Kings, we find very similar images of Helios who drives the chariot drawn by four horses (Figure 28) (Bopearachchi 1991: fig. 23 [Plato, 1-3]). Most probably this is the image that served as model for the creation of sun god (Sūrya) on a chariot in Gandhāra art (Boardman 2015: 99). Similar examples could also found in Mathurā and in Gandhāra during the first to third centuries (Figure 29). The appearance of similar images in the murals of Dunhuang caves and the brocade of Dulanreshui are presumably the result of this acculturation. a b c d Conclusion The secularization or sinicization of some Graeco-Buddhist gods of Gandhāran art was a natural evolution. In fact, it also mirrored the sinicization of Buddhism itself. It is Figure 22 a, b, c, and d. Naked strongmen supporting the stone funerary bed astonishing that Buddhism, an Indian from the tomb of Sima Jinlong; AD 484? Datong Museum. (Photo: author, September 2017.) religion, finally became a localized religion and one of the three main streams of the Chinese cultural tradition that influenced Chinese history for nearly two thousand years. Generally in history, Buddhism was welcomed by the ruling classes, represented by the emperors of all dynasties, except a few of them who persecuted Buddhism in specific periods, mainly because of economic conflicts. Consequently, the doctrine of Buddhism largely accorded with the ideology of the monarchs, and was helpful and useful for imperial rule. Some Confucianists and Taoists initiated debates rejecting Buddhism on the basis of specific problems: for example, which system emerged earliest in China – Buddhism, Confucianism and Taoism – and which one should occupy the highest position in Chinese society? And should Buddhists worship the emperors or not? But the final outcome was that 243 The Global ConneCTions of Gandhāran arT a b Figure 23 a and b. The Sun God in his chariot, Cave no. 285, Dunhuang Grottoes. (Photo: after Dunhuang Academy 1982a, fig. 116.) a b Figure 24 a and b. The Moon Goddess in her chariot, Cave no. 285, Dunhuang Grottoes. (Photo: after Dunhuang Academy 1982a, fig. 117.) Figure 25. Relief of Sūrya from the stūpa at Bodh Gaya, India; 2nd-1st century BC. (Photo: after R.R. Mitra, Buddha Gayá: The Hermitage of Śákyamuni [Calcutta, 1878], 406, via British Library Flickr Commons.) 244 JupinG yanG: The siniCizaTion and seCularizaTion of some GraeCo-buddhisT Gods in China Figure 26. Marble relief of Helios from the Temple of Athena, Troy; early 3rd century BC. Berlin, Altes Museum. (Image: after Meyers Großes KonversationsLexikon, 6th edn., vol. 9 [Leipzig, 1907], 143; scan by NN via <https://commons. wikimedia.org/wiki/File:Helios-Metope,_Troja,_Athena-Tempel.jpg>.) a b Figure 27 a and b. ‘Helios’ on two pieces of brocade, Northern Dynasty and Sui Dynasty, discovered in Dulanreshui, Qinghai Province. (Image: after Zhao Feng 1995: fig. 1.5, 6.) Figure 28. Silver coin of the Indo-Greek king Plato, second century BC. Reverse with image of Helios standing on a chariot drawn by four horses. (Photo: courtesy of Professor Osmund Bopearachchi.) Figure 29. Sūrya, the sun god, winged, with nimbus, From Mathurā. First century AD (?). (Photo: after Coomaraswamy 1927: fig. 44.) 245 The Global ConneCTions of Gandhāran arT Buddhism adapted itself thoroughly to Chinese tradition and became one of the main components of Chinese civilization.12 With the deeper and wider spread of Buddhism in China, its mysticism lessened progressively and its sinicization and secularization become swifter. Some Buddhist figures gradually became members of the Chinese pantheon and were worshipped as folk deities by ordinary Chinese people. So it is understandable that Herakles-Vajrapāṇi, the Nike-Apsara, Atlas, and Helios-Sūrya could be transformed into Chinese-style figures and accepted willingly by Chinese, even become the funerary motifs such as Vajrapāṇi and Atlantean figures. References An Yang Museum (安阳市博物馆). 1982. 唐杨偘墓清理简报 (A Brief Report on the Excavation of Yang Kan Tomb of Tang Dynasty). 文物资料丛刊 (The Collection of Cultural Relics) 6: 130-133. Boardman J. 2015. The Greeks in Asia. London: Thames and Hudson. Bopearachchi O. 1991. Monnaies gréco-bactriennes et indo-grecques: catalogue raisonné. Paris: Bibliothèque Nationale. Coomaraswamy A.K. 1927. The Origin of the Buddha Image. The Art Bulletin 4: 287-329. Dunhuang Academy (敦煌研究院). 1982a. 中国石窟：敦煌莫高窟 (China Grottoes: Dunhuang Mogao Grottoes), vol. 1. Beijing. Dunhuang Academy (敦煌研究院). 1982b. 中国石窟：敦煌莫高窟 (China Grottoes: Dunhuang Mogao Grottoes), vol. 3. Beijing. Dunhuang Academy (敦煌研究院). 1997. 中国石窟：安西榆林窟 (China Grottoes: Anxi Yulin Grottoes). Beijing. Institute of Cultural relics of Hebei Province (河北文物研究所), et al. 2001. 河北省安国市梨园唐墓发 掘简报 (A Brief Report on the Excavation of a Tomb of Tang Dynasty in Liyuan, Anguo City, Hebei Provence). 文物春秋（History of Cultural Relics) 3: 27-35. Hsing I-Tien. 2005. Herakles in the East: The Diffusion and Transformation of His Image in the Arts of Central Asia, India, and Medieval China (trans. W.G. Crowell). Asia Major, 3rd Series, 18/2: 103-154. Li Zhenqi (李振奇). 1993. 河北南和东贾郭唐墓 (A Tomb of Dongjiaguo, Nanhe County, Hebei Province). 文物 (Cultural Relics) 6: 28-33. Liu Jiahe (刘家和) and Liu Linhai (刘林海). 2019. 3-6世纪中西历史及文明发展比较研究（A Comparative Study Between the Roman World and China on the History and Civilization from the 3rd to 6th Century），北京师范大学学报（Journal of Beijing Normal University) 5: 73-99. Liu Yongzeng (刘永曾) (ed). 2003. 敦煌石窟全集: 塑像卷（Complete Works of Dunhuang Grottoes: Volume of Statues). Hong Kong. Priest A. 1930. Chinese Stele. The Metropolitan Museum of Art Bulletin 25, no. 11 (Nov., 1930): 229 and 234239. Shen Zhenzhong (沈振中). 1962. 山西省文物管理委员会/山西省考古研究所). 山西长治北槽唐墓 (A Tomb of Tang Dynasty in Beicao, Changzhi, Shanxi). 考古(Archaeology) 2: 163-68. Sizer T.T. 1925. A Note on Two Pieces of Chinese Sculpture in the Worcester R. Warner Collection. The Bulletin of the Cleveland Museum of Art 12/2: 23-26. Tarzi Z. 1976. Haḍḍa à la lumière des trois dernières campagnes de fouilles de Tapa-é-Shotor (19741976). Compte-rendu de l’Académie des Inscriptions et Belles Lettres 120/3: 381-410. Wang Minzhi (王敏之等) et al. 1990. 河北献县唐墓清理简报 (A Brief Report on the Excavation of a Tomb of Tang Dynasty in Xian County, Hebei Province). 文物 (Cultural Relics) 5: 28-33, 53, 101. Xie Mingliang (谢明良). 1997. 希腊美术的东渐？—从河北献县唐墓出土陶武士俑谈起 (The Spread Eastward of Greek art？To Begin with the Pottery Warrior Tomb-Figure Unearthed from the Tang Dynasty Tomb). 故宫文物月刊. (The National Palace Museum Monthly of Chinese Art) 7: 32-53. 12 Liu & Liu 2019, 93-95 246 JupinG yanG: The siniCizaTion and seCularizaTion of some GraeCo-buddhisT Gods in China Xin Lixing (信力行). 1982. 定县南关唐墓发掘报告 (Report on the Excavation of a Tomb of Tang Dynasty, Nan Guan District, Ding County). 文物资料丛刊. (The Collection of Cultural Relics) 6: 110-116. Xin Mingwei and Li Zhenqi (辛明伟、李振奇). 1993. 河北南和唐代郭祥墓 (Guo Xiang Tomb of Tang Dynasty, Nanhe County, Hebei Province). 文物 (Cultural Relics) 6: 20-27, 64. Yang Juping (杨巨平). 2016.《弥兰王还是米南德？—<那先比丘经>中的希腊化历史信息考》（King Milinda or Menander: A Note on the Hellenistic Information in Milindapanha. 世界历史 (World History) 5: 111-122. Yun Xizheng (云希正). 1963. 天津军粮城发现的唐代墓葬 (The Tomb of the Tang Dynasty Discovered in Junliang Cheng, Tianjin). 考古 (Archaeology) 3: 147-148. Zhang Yuanlin 張元林. 2009. Dialogue Among the Civilizations: The Origin of the Three Guardian Deities’ Images in Cave 285, Mogao Grottoes. The Silk Road 2: 33-48. Zhao Feng (赵丰). 1995. 魏唐织锦中的异域神祇 (The Foreign Deities on the Brocades between the Wei and Tang Dynasties). 考古 (Archaeology) 2: 179-183. Zhao Shengliang (赵声良). 2008. 飞天艺术：从印度到中国 (The Art of Apsara: From India to China). Nanjing. 247 Part 5 Epilogue (251-264): DOI: 10.32028/9781789696950-10 De-fragmenting Gandhāran art: advancing analysis through digital imaging and visualization1 Ian Haynes, Iwan Peverett, Wannaporn Rienjang with contributions by Luca M. Olivieri Students of Gandhāran art are necessarily acutely conscious of the impact of fragmentation upon their field. The very attributes that have drawn scholars to Gandhāra’s visual culture have also, historically, made it vulnerable to patterns of collection that would invite universal condemnation today. Sculpture has been hewn from the structures for which it was originally created, often leaving it without a documented provenance, almost always without a specific archaeological context.2 Not infrequently, sculptural groups have ended up at diverse locations, in different collections, circumstances that present their own challenges for researchers, even where the relationship between dispersed pieces is recognized. This paper considers two approaches that may both help mitigate these difficulties and offer new avenues of research. The first approach addresses the value of object scanning, focusing particularly on issues relating to the high-relief schist sculptures that characterize Gandhāran art. The second uses the remains of Saidu Sharif in Pakistan, an exceptionally well-excavated and documented site, as the basis from which to explore the potential for digital visualization of sites. It argues that the generation of such visualizations, or ‘provocations’, has its own value as an analytical method, testing hypotheses about the interplay of art, architecture, and agency. Imaging Gandhāran sculpture The use of laser scanning, white light/video scanning (WLS), structured light scanning, and Structure from Motion (SfM) photogrammetry to generate 3D digital models of sculpture is now widespread.3 The range of systems and the rapidity of development of these methods can appear bewildering. The speed of technical innovation, both in terms of data capture and data processing, means that detailed comment on a single device in a publication such as this can become redundant even before it reaches a larger audience. There are nevertheless several general observations that are likely to have a longerterm significance, these are rehearsed here. There are several reasons why 3D modelling of sculpture has grown in popularity. First, and arguably most importantly, it allows researchers to look at objects afresh. The levels of precision – sub-mm levels of accuracy are readily achieved – can generate unwieldy data sets, but properly managed they are a major asset. The digital output not only allows for a closer examination of abraded and eroded detail, it also allows for enhancement and manipulation at no risk to the original object. Faces of the digital model can also be studied in higher resolution on a computer while being illuminated by raking light from a range of angles, furthermore images of the object can be produced without shadow more easily than a photograph can. But the advantage goes beyond this. Precisely because the models are 3D, they record detail which a photographer, even a specialist photographer, might not consider significant at the time Ian Haynes gratefully acknowledges the insightful comments of Alex Turner on white light scanning undertaken in this project. 2 The term ‘archaeological context’ is used here in the specific, technical sense, of location of use within a place. 3 Structure from Motion imaging essentially draws upon overlapping images taken from multiple angles to create a 3D model. A good quality digital camera, with a fixed focal length lens, can now be used to produce these images, but the key to a successful SfM transformation is the software used. An obvious attraction of SfM is that the data capture process can be less expensive than the use of a purpose-built scanning unit, but a key consideration remains the greater precision that scanners can usually achieve when recording an object. Historic England (2018) offers a valuable introduction, with case studies, to the effective use of laser scanning to document a range of archaeological material, including sculpture. 1 251 The Global ConneCTions of Gandhāran arT the object is documented. In Gandhāran art, for example, damage done to the back of objects when they were removed is seldom photographed, but it can be significant in determining the relationship of one dispersed fragment to another, and potentially also determining the structure from which it was removed. The value of recording such information, particularly when a fragment is to be wall mounted in display and thus less readily examinable, can be considerable. Research is in turn greatly facilitated by digital models. While close examination of the original object must always be the aspiration of every specialist, the very fact that these precise models can be made available over distance and to multiple users simultaneously and at minimal cost enhances the scope for informed scholarly exchange. Details can be studied, conjoins identified and, with the exciting developments in the study of ancient colour now well advanced, the implications of different colour schemata can be modelled onto the 3D surface at no risk to the original object. Not only can various imaging systems now record with precision the colour and texture of the surface as it appears at the time of recording, but colours can be added, adapted and variously digitally manipulated. With the development of 3D printers, the outputs are not restricted to the digital realm; tangible full-size or scaled versions can be generated. As an instrument for taking innovative research straight into the public realm, the force of such models is considerable. The threat to Gandhāran art from illicit trade, looting, and iconoclasm is not just a facet of recent history.4 At a time when both random and calculated attacks on cultural heritage are sadly familiar, consideration must be given to both its safeguarding and, in the event of loss or destruction, its ‘reconstruction’.5 While a literal replacement may not be attainable, and the profound sense of loss will remain, the presence of digital models can go a long way to alleviating the injury, capable as they are of preserving detail down to the signature style of the original sculptor. In a test to compare the effectiveness of different forms of imaging when recording Gandhāran sculpture, colleagues from the Ashmolean Museum, the Classical Art Research Centre at Oxford, and Newcastle University undertook a programme to scan objects in the Ashmolean’s collections.6 While the extensive range of successful imaging of worked stone left no doubt that results could be achieved, the aim of this exercise was to take a deeper look at the challenges posed by the particular qualities of Gandhāran art. How would structured light scanning and SfM imaging manage the challenge of recording the deep relief of the sculpture? How would the schist respond to white light scanning? Would the higher resolution photography of SfM produce a more detailed representation of the surface texture of the sculpture than video-based structured light scanner? How time-consuming would the post-processing of structured light scanning data prove when registering the different outputs of consecutive scans to one another? L.M. Olivieri also notes the threat posed by the number of fakes appearing not only in the trade in antiquities, but also in museum collections. Some of these fakes are now being mistakenly included in scholarly publications. 5 The case of the Bamiyan Buddhas is not only one of the best known instances of this threat, it is also one where a wide range of imaging and scanning techniques have been brought in to attempt virtually to recover what has been lost (Jansen et al. 2008; Toubekis et al. 2011). Impressive though this work is, it can only hope to recover a fraction of the information that scanning before destruction would have recorded. L.M. Olivieri notes the amazing work undertaken in 2013-2016 by Giuseppe Salemi, Fabio Colombo, Livia Alberti, and others (ISMEO-ACT) in 2013-2016 for the restoration of the Buddha of Jahanabad in Swat. The face and other elements had been destroyed by insurgents in 2007. The team made extensive use of 3D scanning technology to aid in the restoration. See Olivieri 2014, for the first phase of restoration work. A further report was delivered at the 2018 EASAA Conference in Naples. 6 The team acknowledge with gratitude the support of Dr Mallica Kumbera Landrus, Keeper of the Eastern Art Department and Senior Curator of Indian, Himalayan and Southeast Asian Art at the Ashmolean for granting access to the collection. The scanning team consisted of Alex Turner, David Heslop, and Ian Haynes. SfM recording was undertaken using a Canon 750D SLR with 24.2 megapixel resolution. The software used to process the images was Photoscan, augmented by 3D modelling software (a combination of Meshmixer, Meshlab, Blender, and Autodesk Recap Pro). Two WLS scanners were employed, the ArtecEVA, capable of scanning with a resolution of up to 0.5mm, and the Artec Spider, with a resolution of 0.1mm. 4 252 ian haynes, iWan peVereTT, Wannaporn rienJanG: de-fraGmenTinG Gandhāran arT Figures 1a and 1b. Scanning of a grey schist standing figure of the Buddha c. second to third century AD (EAOS.26; height 0.95m) in the Ashmolean Museum. The scanner in use is the Artec EVA. The images of sculpture are both derived from rotations of a scanned digital model. (Model by Alex Turner, Newcastle University; copyright Ashmolean Museum, Oxford.) Figures 2a, 2b, 2c. Different views of a digital model of the schist sculpture of Hārītī, c. second century AD (EA1997.3; 19.5 x 13 x 7.5 cm). Derived from scans by the Artec Space Spider. The images of sculpture are derived from rotations of a scanned digital model. (Model by Alex Turner, Newcastle University; copyright Ashmolean Museum, Oxford.) 253 The Global ConneCTions of Gandhāran arT While the evolution of recording systems means that the test’s conclusions should be revisited whenever further imaging is envisioned, a clear result was that structured light scanning yielded the most satisfactory results in terms of precision and speed (Figures 1 and 2). Outputs from the test are shown here in 2D form.7 The latter allows a basic zoom and rotate facility. The full scan data, suitable for manipulation and research, is available with permission from the Ashmolean Museum. Visualizing Gandhāran Art and Architecture At its inception Gandhāran sculpture formed an integral part of a wider visual landscape. Its position in that landscape affected the way it was encountered by the viewer. Understanding the built environment of which it was a part is therefore integral to its analysis, but this remains intensely difficult for the reasons noted above. One approach to this problem is to exploit the potential of digital visualization. Like imaging, visualisation has now become a familiar instrument in the study of the past. Yet, it presents intellectual and ethical challenges too. The capacity of visualization software to generate photo-realistic images of partially or wholly hypothetical reconstructions of structures and spaces now partially or wholly destroyed is not unproblematic. Even though the advanced capabilities of architectural software can check the generation of hypothetical reconstructions that look fine on a page, but could never have stood – a notable problem in an earlier age – there is still a clear gap between an image of what is, and one of what is thought to have been. Concern about the implications of this gap are not new. Such is the power of images that the professionally rendered visualization may not only mislead the interested amateur but can also subtly impact on the understanding of experts too. To address this concern a group of specialists in historical visualization launched the London Charter.8 Integral to the Charter is the expectation that a detailed justification for each element of the visualization should be available. This expectation not only reflects the fact that expert visualizations are research outputs in their own right, but it also enables the viewer to distinguish between what is known, what is surmised, what is proposed based on analogy, and what is mere conjecture (Denard 2009: 7–8). While these criteria are seldom met in practice, the goal they embody remains an important one and crucially, it has the capacity to lead to further advances in scholarship. Attempting to visualize a structure in three dimensions and its entirety demands the resolution of debates that are otherwise unnecessary in text and can be avoided even in line-drawings and architectural plans. The quest to resolve key points frequently requires input from a range of specialists. Each stage of visualization in turn provokes new reactions and responses. This latter point is essential, and it is why the term provocation is to be preferred to visualization. The London Charter inspires us to document these so that the process of knowledge-building is rendered more transparent. A visualization/provocation is thus a vehicle for refining, question-setting, debate, rather than a depiction of what was, but it can crucially also help ascertain what was not. The capacity to test sight lines within such a 3D model, for example, is a powerful instrument when seeking to understand the role that, say, figural art played within a wider complex. Were two points inter-visible? How much detail could a viewer actually see? In order to explore this approach further, the authors generated a series of visualizations/provocations of the site of Saidu Sharif. The site was chosen because it offers an exceptionally well documented body of material for both its art and architecture. So precise was the excavation and so good the preservation, that Francesco Martore who worked with Domenico Faccenna, was able to generate The test objects were initially also made available through Sketchfab via the Gandhāra Connections Resources page. They are now archived at the Classical Art Research Centre and the Ashmolean Museum in Oxford. 8 See <http://www.londoncharter.org> last consulted 5th June 2020. 7 254 ian haynes, iWan peVereTT, Wannaporn rienJanG: de-fraGmenTinG Gandhāran arT fine plans, elevations and even a museum model (of the Main Stūpa).9 Nevertheless, the application of digital methods allowed the team to explore further points of debate about the structure’s original appearance. The Buddhist site of Saidu Sharif The Buddhist site of Saidu Sharif stands in the Saidu Valley, at the foot of the Shararai mountain, in Swat. The site is located on the side of an alluvial plateau formed by three rivers – the Swat, Jambil, and Saidu rivers – at the outskirts of an ancient urban area (today Mingora) (Faccenna 1995: fig. 1). At the junction between Saidu Jambil and Swat Rivers are other Buddhist sites of Swat, notable amongst which are Butkara I, Butkara III and Pānṛ I.10 The site of Saidu Sharif was built over a graveyard (Noci, Macchiarelli & Faccenna 1997). The chronology and stratigraphic relation of the graveyard to the monastery and sacred area has been recently reassessed (Olivieri 2016; 2019b). From the geographical point of view, although the mountainous Swat valley is a cul-de-sac (Olivieri, forthcoming), for centuries it was deliberately chosen as a ‘short-cut’ on the route that connects Tibet and western China with Gandhāra (Kuwayama 1991). It was used by the fifth-century Chinese pilgrim Faxian who travelled from China to India in search of Buddhist texts (Legge 1991). Today one can drive from Swat to Gandhāra in the Peshawar valley in less than four hours. It is well known that Gandhāra is one of the nodes of the ancient northern route, uttarāpatha, that connects the north-west of the Indian subcontinent with northern India, Bactria, and Central Asia (Neelis 2010). The Swat valley is one of the richest double-crop pocket zones in the north of the subcontinent (Olivieri, forthcoming).11 It is also home for mineral sources including semi-precious stones such as emerald (Kazmi & Donoghue 1990), as well as schist and steatite which are soapstones commonly used for Gandhāran sculptures (Law 2011; Shah 1997; Kempe 1986; Rafiq et al 1983; Faccenna et al. 1993; Di Florio et al. 1993; 1995). The location and physical geography accommodated the growth of the Buddhist sites in the Swat valley during the time when Buddhism flourished in the north-west Indian subcontinent. Two principal buildings constitute the structure of the Buddhist site of Saidu Sharif – the Main Stūpa and the monastery – and several minor structures, including subsidiary stūpas and shrines (Figure 3). The excavation of the site began in the twentieth century with the Istituto Italiano per il Medio ed Estremo Oriente (IsMEO) under the direction of Domenico Faccenna and Pierfrancesco Callieri (who directed the excavation of the monastery). Results from the excavations were published in four large reports (Callieri 1989, Faccenna 1995 [2 vols.], Noci, Macchiarelli, & Faccenna 1997, Faccenna 2001 [in Italian]). The first phase of excavation took place between 1963 and 1968 and focused on the stūpa terrace (Faccenna 1995). The second phase of excavation continued a decade later, lasting for five years between 1977-82, during which the monastery and more of the stūpa area was unearthed (Callieri 1989). Most recently, between 2011 and 2015, excavation and conservation at the site were completed by the new ISMEO under the direction of L.M. Olivieri (Olivieri 2014; 2016; Filigenzi et al. 2016). Thanks to the meticulous work of the Italian team, the Buddhist site of Saidu Sharif is one of the very few Gandhāran sites that can offer archaeological and architectural information with high precision and accuracy. The Main Stūpa is located on the lower terrace and is surrounded with several subsidiary stūpas, shrines, and columns built over periods. The monastery, built on a rectangular plan, stands on the upper terrace to the east of the Main Stūpa. The construction of the buildings at the Buddhist site of Saidu Sharif can be divided into three main building periods (Faccenna 1995; Callieri 1989). This paper presents the visualization of the beginning of the construction, i.e. the first phase of the first period of the For the new Swat Museum, Saidu Sharif (Gallery 4; 2012). See description in Olivieri 2019a. 11 See Olivieri et al 2006; Olivieri and Filigenzi 2018, on the implications for the Buddhist communities. 9 10 255 The Global ConneCTions of Gandhāran arT Figure 3. General view of the stūpa area and the monastery of the Saidu Sharif Buddhist site. (Photo: after Callieri 1989, fig. 3, courtesy ISMEO Italian Archaeological Mission in Pakistan.) Figure 4. Site plan of the Saidu Sharif stūpa terrace during the first phase of the first period. (Plan: after Faccenna 1995, fig. 9, courtesy ISMEO Italian Archaeological Mission in Pakistan.) construction (Figure 4), which is the period corresponding, according to Faccenna (1995), to the first half of the first century AD.12 12 The dating of these periods is based on a combination of evidence including numismatic and epigraphic evidence, and comparisons of the typology of structural techniques with the neighbouring Buddhist sites including Butkara I and Pānṛ. The majority of coin finds from Saidu Sharif I belong to the Kushan kings (from Soter Megas to Vasudeva I and the Later Kushans). But the coin of the earliest date is that which appears to be an issue of Azes II (Callieri 1989: 231, 232 fig. 161a) which was found in the monastery area. In contrast to a second coin (see Callieri 1989: 120), the ‘Azes II’ issue is firmly associated with the second floor [layer (3a)] of the Monastery’s courtyard in Period I B. The latter certainly corresponds to the last phases of Period I on the Stupa Terrace (last half of the first century AD; Faccenna 1995: 144, pl. XX) (Olivieri, pers. comm.). In fact, according to Joe Cribb (pers. comm.), the coin is most likely a posthumous issue of Azes II, which was probably issued during the latter half of the first century AD. Kharoṣṭhī inscriptions on pottery that were found in the monastery area are dated by Fussman (1989) to the period between mid-first century BC and second century AD or later. 256 ian haynes, iWan peVereTT, Wannaporn rienJanG: de-fraGmenTinG Gandhāran arT The construction during the first period began with the Main Stūpa and monastery. There are also remains of nine square bases on the stūpa terrace, to the north of the Main Stūpa, that were also built during the first period. According to the analysis by Faccenna (1995), these are bases for small stūpas (nos. 21, 31, 32, 57), columns (nos. 24, 29, 69, 75), and an unidentified structure (no. 80) possibly a pillar or an image (Faccenna 1995: 53, 197-98). It may be worth noting that no chapels to house images (vihāra) were built during the first period. The Main Stūpa, built with soapstone in an ashlar masonry, stands on an irregularly square platform.13 The stūpa is preserved up to the second storey, i.e. the first storey of the drum which rests upon the platform. Inside the platform, at the top and slightly off-centre, is a reliquary recess with its relic deposit in-situ (Faccenna 1995: 441, pls. 43, 44, fig. 188).14 There is no record of enlargement of the Main Stūpa. The architectural reconstruction of the Main Stūpa by Faccenna (1995: fig. 283) shows that the stūpa is accessed by a stairway to the north which leads to the top of the platform.15 An upper stairway leads up to the pradakṣiṇapātha, encircling the stūpa drum.16 Remains of a base plinth at a corner of the Main Stūpa in combination with remains of large fallen columns, with parts of shaft and capital, on the terrace floor next to the Main Stūpa on the south, east, and north suggest that four columns stood in each corner of the Main Stūpa base (Faccenna 1995: 481; Faccenna & Callieri 2003: 311). Part of one column (Column C), excavated between 2011-2012, shows that the column was fashioned with the Gandhāran Persepolitan capital.17 There are also remains of a harmika, which was decorated with a motif of rosette-type within filleted lozenges, as well as elements of the railing and cornice. Evidence of elements of chatrāvalī for smaller stūpas was also near the wall of the Main Stūpa (Faccenna 1995: 598).18 The dimensionsof this square platform are 21.14m x 21.09 m x 20.34m x 20.22m (Faccenna 1995). L.M. Olivieri is currently completing a study on the building processes related to the Main Stūpa foundation and relic deposition at Saidu Sharif I, and their chronology (pers. comm.). On relic deposits and Saidu Sharif I see also Provenzali 2019. 15 The reconstruction is based on excavated structures and associated materials at Saidu Sharif, in combination with comparisons of these elements with contemporary Buddhist sites in Swat including Butkara I, Panr I, Tokar-dara and Gumbatuna (Faccenna 1995). The elevation of the Main Stūpa at Saidu Sharif is based mainly on metrological analysis of the excavated structures and materials at Saidu Sharif with those of the stūpa of Tokar-dara (Faccenna 1995: 514-525). 16 Faccenna (1995: 502) based his reconstruction of the pradakṣiṇapātha’s position on the comparison with stupas of Tokar-dara and Gumbatuna. 17 Faccenna’s reconstruction of the four columns, each topped by a seated lion, is based on the comparison with sculptural fragments of lion documented at Butkara I and Panr I. A sculptural fragment in the form of a lion was found next to a column at Panr I, and on the pradakṣiṇapātha of the Great Stūpa of Butkara I (Faccenna, Khan & Nadeem 1993: appendix A; Faccenna 1980-81: 644, pls. 207b, 208a). There were also sculptural fragments of a lion at Saidu Sharif but their original context could not be determined at that time (Faccenna 1995: 497, note 1). During the excavation of Column C in 2012, fragments of the associated seated lion was found and inventoried in the Swat Museum as SS I 29, 30 and 31 (three fragments: front and head, back, and one front leg) (Olivieri, pers. comm.). 18 Two small square bases [69] and [75] are located close to the corner sides of the Main Stupa’s staircase, respectively to the E and W. In the model presented here, they are completed as two columns surmounted by a disc, following the arrangement suggested by Faccenna 1995: 565, fig. 283. Both monuments are coeval to the Main Stūpa (Period I, phase a), certainly part of the monumental gate of the latter, and part of the overall monumental project of the Main Stūpa. With regard to their structure and elevation, [69] and [75] can be interpreted either as columns, or as pilasters, in both cases freestanding. They were topped by a finial which in the first case might have been a sitting lion, in the second case, a stone disc (cakra) (see Faccenna 1984: 321-322). The hypothesis was formulated by Faccenna on the basis of his study on monuments of Butkara I nos. 65 and 68 (columns) and no. 135 (pillar) (Faccenna 1984: 321-322, 325-327). Pillar 135 at Butkara I was surmounted by a four-sided, square, GandharanCorinthian capital surmounted by a cakra whose fragments were brilliantly recontructed, also with the help of sculptural representations from Loriyān Tangai and Shnesha (Faccenna 1984). The surviving architecture of [69] and [75] leaves no space for a detailed interpretation of the section of the elevation (square or circular). However, excavations at the site (including those directed by L.M. Olivieri) yielded several fragments of statues of sitting lions in soapstone, including (in 2012) fragments pertaining to the lion topping Column C (collapsed from the NE corner of the first storey of the Main Stūpa). Amongst the various fragments, there are some which certainly belong to the Main Stūpa’s Columns A-D (total H. c. 1.5 m), others (referring to at least two complete statues, total H. c.1 m), which can be associated with columns [24] and [29] (to the left and right of the Main Stupa staircase), and another one even smaller (SSI 225; total H. c. 0.70 m). The dimensions of the latter would fit the reconstructed height of both [69] and [75], if they were supporting columns. The visual effect of the two lions topping columns [24] and [29] before the entrance, the two smaller ones guarding the sides of the staircase ([69] and [75]), would certainly have been redundant, with the four bigger lions dominating from the tops of the Columns A-D. However, such redundancy would not 13 14 257 The Global ConneCTions of Gandhāran arT The dominant feature of the Main Stūpa is perhaps the main frieze19 depicting the life of the Buddha20 that decorates the stūpa drum (Faccenna 1995; Faccenna & Callieri 2003; Filigenzi 2006). One hundred and twenty-two fragments, most of which were found near the Main Stūpa, and which came from approximately seventy panels, have survived (Faccenna 1995: 526 n. 1, 528). Based on material, styles, and technical characteristics, these panels are most likely to have been built during the first period and under the supervision of one sculptural master (Faccenna 1995: 525-540; Faccenna & Callieri 2003; Filigenzi 2006). The frieze is made of green schist but the indirect evidence of the documented sculptural material suggests that they may have been gilded (Faccenna et al. 1993: 133, fn. 3; Faccenna 2001; Pannuzi 2015; Zaminga et al 2019; Ramaso 2019). The visualization of the two square bases on the left and right of the stairway as bases of columns each crowned by a chakra is based on evidence of such columns found at Butkara I and their depictions on sculptures both in Swat and the Peshawar basin (Faccenna 1980-81; 1984). The monastery, also built in the first period, comprises ten rooms of the same size on each of the east and west sides, and twelve cells of different sizes on the south, with a square courtyard in the middle (Callieri 1989: fig. 5). Along the fourth side of the courtyard is a porch with fourteen pillars on each side. The monastery is accessed through a staircase on the west side. One of the main concerns for the visualization is the issue of colour. There was no trace of any colour other than white for the body of the Main Stūpa, however, traces of red colour were found on the plaster coat of a column shaft of the Main Stūpa (Faccenna 1995: appendix A, 488 n. 2, 492). As mentioned, the main frieze was possibly gilded.21 In any case, onlookers would have hardly escaped the amazing impression given by the contrast between the whitish soapstone of the railing of the first staircase and plinth, the sage green or gold of the main frieze, and the white plaster of the dome, all framed by the red shafts of the columns, and topped by the probable red details of the chatrāvalī. Visualization of the Saidu Sharif stūpa complex As part of the process of developing visualizations of the Saidu Sharif site, the team used concept (shape only) and rendered models (3D models where colours, textures and the fall of light are incorporated/ simulated). The figures in this paper represent 2D shots taken at static points within the models. A preliminary stage was to generate a basic model placing the Main Stūpa in relation to the monastery building and drawing on the excellent reports of Callieri (1989) and Faccenna (1995) (Figure 5). The desire to encompass the two major structures in one model, something not previously attempted, was have been so unusual for the onlookers in mid-first century AD. One would immediately think of the clusters of sitting lions at the entrance of the stūpas in the representations from the frieze of Amarāvati in the British Museum. [Note by L.M. Olivieri.] 19 According to the reconstruction by Faccenna (1995: 525), there are two superimposed sculptural registers on the Main Stūpa drum: register A with a figural motif and register B with a pseudo-railing motif. Faccenna (1995; Faccenna & Callieri 2003) places the figural motif (register A) below the pseudo-railing motif (register B). The visualization in this paper follows a reconstruction by L.M. Olivieri and F. Martore. Basing their argument on the dimensional data of the materials, Olivieri and Martore suggest that the figural panels may have been positioned above the pseudo-railings. According to Olivieri, this hypothesis was formulated in consideration of (a) the better visibility of the panels during the ritual circumambulation, and (b) the perspective of the figures which were sculpted so as to be seen at eye level (Olivieri, pers. comm.). This hypothesis was also considered by D. Faccenna amongst others (Olivieri and Martore pers. comm.; Faccenna 1995: 529 ff.). However, Faccenna’s final reconstruction was guided by the comparison with surviving sculptural representations and miniature stūpas (Faccenna 1995: 528-540; 2001; 2002: 127; Faccenna & Callieri 2003: 319). An updated study of the frieze, on the basis of the new elements discovered during the 2011-2015 excavation seasons, is currently in progress (a study directed by A. Filigenzi with A. Amato), and it will be preliminarily presented at the forthcoming EASAA Conference in Barcelona (Olivieri, pers. comm.). 20 Identified scenes include the dream of Māyā, the birth of Siddhārtha, the wrestling match, the meeting with hunters, the farewell of Kanthaka, the quarrel over the Relics, and the transportation of the relics (Faccenna 1995: 526 note 1, 528). 21 Elsewhere at Butkara I, traces of colour, including red, golden-ochre, blue, and black, were found on the drum of the Main Stūpa and smaller stūpas during different periods (Faccenna 1980-81: pls. E.a-b, F.a-b, G.a-b, M.a-b). 258 ian haynes, iWan peVereTT, Wannaporn rienJanG: de-fraGmenTinG Gandhāran arT Figure 5. Basic model showing the relationship of the monastery and the Main Stūpa. (Iwan Peverett/New Visions Heritage.) Figure 6. Work-in-progress model showing the view of the Main Stūpa from the Monastery entrance. (Iwan Peverett/New Visions Heritage with Ian Haynes.) partially driven by the wish to understand better lines of sight between the two. This allows us to consider, for example, the sight that greeted a monk as he passed through the monastery courtyard and out of the building’s main entrance (Figure 6). There is a challenge in visualizing the monastery elevation on current evidence, so we have offered a minimalist model, which draws upon the report (Callieri 1989) but does not elaborate the façade (Figure 7). 259 The Global ConneCTions of Gandhāran arT 260 Figure 7. Work-in-progress model of the Sacred Area. (Iwan Peverett/New Visions Heritage with Ian Haynes.) ian haynes, iWan peVereTT, Wannaporn rienJanG: de-fraGmenTinG Gandhāran arT Figure 8. Close up of the Main Stūpa, incorporating models derived from figures in Callieri and Filigenzi (2002, pl. XI, cat. 126 (s 1137) and fig. 47, S. 1112). (Iwan Peverett/New Visions Heritage with Ian Haynes.) The image places the surviving fragments in what is believed to be their correct location. Details of the two cornices are indicative. The colour (rendered) models were developed in part in order to visualize the impact of different reconstructed colour schemes. The process threw up challenging questions from the start. Was the practice of gilding the sculptures, well attested in later Gandhāran art (Pannuzi & Talarico 2018), a feature of the earliest phase at Saidu Sharif, for example? If it was, might that gilding have been partial, or might it have covered the entire surface? In our models, we have visualized the sculpture both with and without. This is one example of the challenge of incorporating material from analogies elsewhere. Another is to ask how far even the decoration of contemporary sites in the same region employed similar colour schemes. Butkara I (Faccenna 1980-81) for example, has yielded many examples of coloured material, but in the absence of directly comparable evidence from Saidu Sharif would it be appropriate to use it in the visualization? Having the capacity to move around the model also raises that question of how those making a circumambulation of the stūpa would have seen the sculpted frieze. Faccenna (1995: fig. 282) envisioned the frieze below the false railing, but as mentioned above,22 this alternative restoration follows the arguments of Luca M. Olivieri, that the frieze was sculpted to be visible at eye-level and that it would have been more visible if it was positioned above the false railing. Olivieri further notes that positioning the false railing below achieves what was probably the desired perspective, that the railing was in the foreground, the frieze behind (Olivieri, pers. comm.). Our model confirms at least that this configuration would place the frieze on a more natural eye line for the average visitor. Models such as this can also be used, with due and explicit caution, to consider how pieces without a precise archaeological context may once have been displayed. 22 See n. 15 above. 261 The Global ConneCTions of Gandhāran arT It will be evident to readers, that these visualizations are very much work in progress, that certain points of detail have been omitted.23 We are in the process of inserting these elements but note that the generation of appropriate 3D digital models for insertion even from the excellent technical photos in Callieri and Filigenzi’s (2002) study of the Saidu Sharif sculpture is a difficult and time-consuming process (Figure 8). The scanning of original fragments as outlined in the first section of this paper must always be the preferred approach for accuracy and efficiency. What we do have here though is a provocation, in what we hope is the best sense of the word, a model that raises its own further and important questions about how we understand Gandhāran art and how it operated in its original setting. References Callieri P. 1989. Saidu Sharif I (Swat, Pakistan). 1. The Buddhist Sacred Area: The Monastery. (IsMEO Reports and Memoirs 23.1.) Rome. Callieri P. and Filigenzi A. (eds) 2002. Il Maestro di Saidu Sharif: alle origini dell’arte del Gandhara. Rome: IsIAO Denard H. 2009. The London Charter for the Computer-Based Visualisation of Cultural Heritage (Edition 2.1). London. Di Florio M.R., Lorenzoni S., Zanettin Lorenzoni E., and Olivieri L.M. 1993. Evidence of Ancient Stone Quarrying in Lower Swat, NWFP, Pakistan: A Geo-archaeological Study. Science and Technology for Cultural Heritage 2: 63-74. Di Florio, M.R., Faccenna C., Lorenzoni S., and Zanettin Lorenzoni E. 1995. Lithological Analysis of Material Used in the Buildings of the Sacred Area of Saidu Sharif I (Swat Valley, Northern Pakistan) and Their Origins. Pages 605-623 in D. Faccenna, Saidu Sharif I (Swat, Pakistan), vol. 2, The Buddhist Sacred Area: The Stūpa Terrace. Text. (IsMEO Reports and Memoirs 23.2.) Rome. Faccenna D. 1980-81. Butkara I. (Swāt, Pakistan). 1956-1962. Rome: IsMEO. Faccenna D. 1984. Note Gandhariche – 1. Riconstruzione di un pilastro con cakra nell’area sacra di Butkara I (Swāt, Pakistan). Annali 44/2: 319-338. Faccenna D. 1995. Saidu Sharif I (Swat, Pakistan), vol. 2, The Buddhist Sacred Area: The Stūpa Terrace. Text. (IsMEO Reports and Memoirs 23.2.) Rome Faccenna D. 2001. Il fregio figurato dello Stūpa Principale dell’area sacra buddhista di Saidu Sharif I (Swat, Pakistan). (IsIAO Reports and Memoirs 23.) Roma. Faccenna D. and Callieri P. 2003. At the Origin of Gandharan Art. The Contribution of the IsIAO Italian Archaeological Mission in the Swat Valley Pakistan: The Buddhist Sacred Area of Saidu Sharif I. Ancient Civilizations from Scythia to Siberia 9/3-4: 307-318. Faccenna D., Khan A.N., and Nadeem I.H. 1993. Pānṛ I (Swat, Pakistan). (IsMEO Reports and Memoirs 25.1.) Rome. Filigenzi A. 2006. From Mind to Eye: Two Dimensional Illusions and Pictorial Suggestions at Saidu Sharif I. Pages 17-40 in P. Callieri (ed), Architetti, capromastri, artigiani: L’organizzazione dei cantieri e della produzione artistica nell’Asia ellenistica. Studi offerti a Domenico Faccenna nel suo ottantesimo complenno. (Serie Orientale Roma C.) Rome: IsIAO. Filigenzi A., Olivieri L.M., with a contribution by Terrasi F., Marzaioli F., Passariello I. and Capano C. 2016. Buddhist Settlements and Proto-Historic Graveyards: New Chronometric and Archaeological Data from Saidu Sharif I (Swat, Pakistan). Newsletter di Archeologia CISA (“L’Orientale”Università degli Studi di Napoli) 7: 23-33. Fussman G. 1989. Kharoṣṭhī Inscriptions. Pages 225-230 in P. Callieri. Saidu Sharif I (Swat, Pakistan). 1. The Buddhist Sacred Area: The Monastery. (IsMEO Reports and Memoirs 23.1.) Rome. 23 The bases of the columns should, for example, be scotia-type on a high moulded pedestal. We have also inserted a south boundary wall for the stūpa terrace in Figure 3. Though the latter was hypothesized by Faccenna (1995: 56) it was not documented in the excavation (Olivieri pers. comm.). 262 ian haynes, iWan peVereTT, Wannaporn rienJanG: de-fraGmenTinG Gandhāran arT Historic England. 2018. 3D Laser Scanning for Heritage: Advice and Guidance on the Use of Laser Scanning in Archaeology and Architecture. Swindon. Historic England. Jansen M., Toubekis G., Walther A., Döring-Williams M. and Mayer I. 2008. Laser Scan Measurement of the Niche and Virtual 3D Representation of the Small Buddha in Bamiyan. Pages 83-90 in A. Posluschny, K. Lambers, and I. Herzog (eds), CAA 2007 - Layers of Perception. Proceedings of the 35th International Conference on Computer Applications and Quantitative Methods in Archaeology. Verlag Rudolf Habelt: Bonn. Kazmi A.H. and Donoghue M. 1990. Gemstones of Pakistan: Geology and Gemmology. Peshawar: Gemstone Corporation of Pakistan. Kempe D.R.C. 1986. Gandhara Sculptural Schist: Proposed Source. Journal of Archaeological Science 13: 79-87. Kuwayama S. 1991. Across the Hindukush of the First Millennium: A Collection of the Papers. (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.) Kyoto: Institute for Research in Humanities, Kyoto University. Law R. 2011. Inter-Regional Interaction and Urbanism in the Ancient Indus Valley: A Geologic Provenience Study of Harappa’s Rock and Mineral Assemblage. Kyoto: Indus Project, Research Institute for Humanity and Nature. Legge J. 1991. A Record of Buddhistic Kingdoms. New Delhi: Munshiram Manoharlal. Neelis J. 2010. Early Buddhist Transmission and Trade Networks: Mobility and Exchange within and beyond the Northwestern Borderlands of South Asia. Leiden; Boston: Brill. Noci F., Macchiarelli R. and Faccenna D. 1997. Saidu Sharif I (Swat, Pakistan). 3. The Graveyard. IsMEO Reports and Memoirs XXIII, 3. Rome. Olivieri L.M. 2014. Restauro conservativo e mobilizzazione sociale in siti archeologici della valle dello Swat (Pakistan). Restauro archeologico 2014/1: 57- 77. Olivieri L.M. with a contribution by Terrasi F., Marzaioli F., Passariello I., Capano C. and Aatif Iqbal. 2016. The Graveyard and the Buddhist Shrine at Saidu Sharif I (Swat, Pakistan): Fresh Chronological and Stratigraphic Evidence. Journal of Ancient History 76/3: 559-578. Olivieri L.M. 2019a. Saidu Sharif Buddhist Monastery, Early Historic Graves in the Swat Valley, Pakistan. Science 365: 165-168. Olivieri L.M. 2019b. The Early-Historic Funerary Monuments of Butkara IV. New Evidence on a Forgotten Excavation in Outer Gandhara. Rivista degli Studi Orientali 1/2: 231-257. Olivieri L.M. Forthcoming. Gandhāra and North-Western India. In R. Mairs (ed), The Graeco-Bactrian World. Taylor and Francis. Olivieri L.M. and Filigenzi A. 2018. On Gandhāra Sculptural Production from Swat: Recent Archaeological and Chronological data. Pages 71-92 in W. Rienjang and P. Stewart (eds), Problems of Chronology in Gandhāran Art. Proceedings of the First International Workshop of the Gandhāra Connections Project, University of Oxford, 23rd-24th March, 2017. Archaeopress: Oxford. Olivieri, L.M. and M. Vidale, with contributions by Nasir A., Saeed T., Colliva L., Garbini R., Langella L., Micheli R. and Morigi E. 2006. Archaeology and Settlement History in a Test-Area of the Swat Valley. Preliminary Report on the AMSV Project (1st Phase). Pages 73-150 in L.M. Olivieri (ed), East and West, 56, 1-3 (Special Issue for the 50th Anniversary of the IsIAO Italian Archaeological Mission in Pakistan). IsIAO: Rome. Pannuzi S. (ed). 2015. Ambito, metodologia e finalità del progetto di ricerca sulle sculture gandhariche in pietra e in stucco del Museo Nazionale d’Arte Orientale ‘Giuseppe Tucci’. Pages 9-15 in S. Pannuzi (ed), Gandhara: tecnologia, produzione e conservatione. Rome: Gangemi. Pannuzi S. and Talarico F. 2018. Pigments and Gold in Gandharan Stone and Stucco Sculptures. Pages 183-192 in S. Bracci, G. Giachi, P. Liverani, P. Pallecci and F. Paulucci, F. (eds), Polychromy in Ancient Sculpture and Architecture. Livorno: Sillabe Editrice. Provenzali, A. 2019. Semi fecondi. Reliquie e reliquiari nel Gandhāra. Pages 43-65 in G. Boccali and M. Angiolillo (eds.), Arte e Letteratura nelle società in Asia, Biblioteca Ambrosiana, Bulzoni Editore: Milano. 263 The Global ConneCTions of Gandhāran arT Rafiq M., Ahmad I. and Tahirkheli T. 1983. Geological Map of the Surroundings of the Peshawar Plain. Geological Bulletin 16: 189. Ramasso C. 2019. The Gilded Bracket at the MAO Museo d’Arte Orientale di Torino (Museum of Oriental Art Turin). Restauro archeologico 2019/1: 164-173. Shah S.M.I. 1997. Stratigraphy of Pakistan. (Memoirs of the Geological Survey of Pakistan 12.) Quetta: Geological Survey of Pakistan. Toubekis G., Mayer I., Döring-Williams M., Maeda K., Yamauchi K., Taniguchi Y., Morimoto S., Petzet M., Jarke M. and Jansen M. 2011. Preservation and Management of the UNESCO World Heritage Site of Bamiyan: Laser Scan Documentation and Virtual Reconstruction of the Destroyed Buddha Figures and the Archaeological Remains. In E. Stylianidis, P. Patias and M. Santana (eds), CIPA Heritage Documentation Best Practices and Applications, Series I, 2007 2009 2011, International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 38, Part 5/C19. Zaminga M., Angelini I., Vidale M., Olivieri, L.M., Guida, G. 2019. Investigating Gilding Techniques on Gandharan Stone Sculptures and Architectural Components: A Preliminary Note. Restauro Archeologico 2019/1: 90-99. 264