The Structure of the Sagātha-Vagga of the Saṃyutta-Nikāya
Buddhist Studies Review 24(1) 2007, 7–34 ISSN (print): 0256–2897 doi: 10.1558/bsrv.v24i1.7 ISSN (online): 1747–9681
The Structure of the Sagātha-Vagga of the Saṃyutta-Nikāya
RodERIck S. BUckNELL
School of History, Philosophy, Religion, and classics, The University of Queensland rodbucknell@yahoo.com
ABSTRACT: The meaning of its title, ‘Section with Verses’, may appear sufficient to explain why the Sagātha-vagga was identified as a discrete entity within the Saṃyuttanikāya. However, this article looks beyond that simple explanation, to discover whether some other rationale may underlie this grouping of saṃyuttas. It examines evidence that the compiling of the Sagātha-vagga was probably based on a familiar, although doctrinally marginal, piece of Buddhist teaching, namely the ‘eight Assemblies’.
The Sagātha-vagga (Section with Verses), the first of the five vaggas of the Pāli Saṃyutta-nikāya (SN), comprises 271 short suttas grouped according to topic into eleven saṃyuttas. These eleven topics/saṃyuttas are: Devatās (gods), Devaputtas (sons of gods), Kosala (Pasenadi’s kingdom), Māra (the Evil one), Bhikkhunīs (nuns), Brahmās (higher gods, who tend to think of themselves as creators), Brāhmaṇas (priests), Vaṅgīsa (a senior monk), Vanas (forest spirits), Yakkhas (fierce spirits) and Sakka (Indra, king of the gods). The Devatā-saṃyutta is exceptionally large, with eighty-one suttas; the remaining saṃyuttas range in size from ten suttas to thirty.
Stylistically, the Sagātha-vagga is a very natural grouping; every one of its component suttas contains at least one gāthā (piece of verse), usually embedded within a prose framework.3 This stylistic uniformity may appear sufficient to explain why these eleven saṃyuttas were brought together as a discrete section within SN. Yet this seemingly reasonable explanation for the compiling of the Sagāthavagga is problematic. The first problem is that the sagāthā form (verses embedded in a prose framework) is common elsewhere in the Sutta-piṭaka. The 112 suttas of the Itivuttaka are all in sagāthā form; the Aṅguttara-nikāya contains no fewer than 174 suttas in sagāthā form, seventy of them conspicuously grouped at the beginning of the Fours (AN II 1–76); and within SN itself the Bhikkhu-saṃyutta, located not in the Sagātha-vagga but in the Nidāna-vagga (SN II 273–85), is predominantly (ten suttas out of twelve) in sagāthā form. conversely, there are saṃyuttas in the Sagātha-vagga whose presence there seems superfluous. All but two of the fifty-one gāthās in the Vaṅgīsa-saṃyutta are also present in the Theragāthā, and in precisely the same sequence. Also, the gāthās in the first eight of the ten suttas that make up the Bhikkhunī-saṃyutta replicate ones found in the Therīgāthā, although sometimes imperfectly and in a different sequence. It appears likely that at least these two saṃyuttas were created intentionally out of existing material by providing each of the selected verses with a narrative introduction-commentary. In short, a large amount of material in sagāthā form was excluded from the Sagātha-vagga, while other material in sagāthā form has the appearance of having been assembled into saṃyuttas specifically for inclusion in this vagga. Such facts indicate that this literary form does not, by itself, account fully for the identity of the Sagātha-vagga as a discrete entity within SN. Accordingly, this article has the aim of discovering whether some other rationale may underlie the grouping of these eleven saṃyuttas (Devatā to Sakka) into a recognized vagga. It will be argued, on the basis of a wide range of evidence, that the existing Sagātha-vagga derives from an earlier collection whose structure was based on a familiar, although doctrinally marginal, piece of Buddhist teaching.
Sources ANd METHod
The principal method employed here is comparison of the Pāli Sagātha-vagga with its two counterparts in the chinese Tripiṭaka, contained in Taishō no. 99 and no. 100 (hereafter written ‘T99’ and ‘T100’).
T99, titled 雜阿含經 (Zá āhán jīng) ‘diverse Āgama’, comprises 1362 suttas,10 of which about 1050 correspond to suttas in the Pāli Sutta-piṭaka, most of them in SN.11 T99 is recognized as a translation of a lost Sanskrit Saṃyuktāgama, belonging to either the Sarvāstivāda or the Mūlasarvāstivāda.12 A statement following the Chinese title attributes the translation to an Indian monk, Guṇabhadra, during the (Liú) Sòng dynasty (420–79 cE).13 Whereas the Pāli Sagātha-vagga is located at the beginning of SN, the corresponding section of T99, comprising 309 suttas, is located at the end of the Zá āhán jīng.14 T100, titled 別譯雜阿含經 (Bié yì Zá āhán jīng) ‘other Translation of the diverse Āgama’, was done by an unknown translator and its sectarian affinities are unclear.15 The text appears to be incomplete; it contains only 364 suttas, corresponding to about the last quarter of T99. The greater part of it (280 suttas) corresponds closely to the Sagātha-vagga of T99 and less closely to the Pāli Sagātha-vagga.16
remains (Mark Allon, personal communication). Translations into Western languages exist for only a few suttas or gāthās of T99 and T100, and for some Sanskrit fragments. See especially the English translation of the Bhikkhu-saṃyutta of T100 (scroll 1 = suttas 1–22) by Bingenheimer (2006), which is the first installment of a planned full translation of T100.
10. This figure (also that for T100, below) is as indicated by the numbering of the suttas in the Taishō edition. The actual number of suttas is unclear, as it is also for SN (see Saigusa 1978, 613–69). In the present article, glosses of chinese terms and section titles, if not in English, are usually given in Pāli; e.g. sutta. This is done for ease of comparison with the Pāli texts, and despite the fact that the language of the source text for each of the chinese versions is likely to have been some form of Sanskrit. Transcriptions of chinese are in hànyǔ pīnyīn.
11. Sutta correspondences between the Chinese and Pāli versions of Sagātha-vagga are set out in Anesaki (1905, 4–7); Akanuma (1929, 1–2, 62–4, 90–92, 94–100, 102–19, 172–91, 204); Taishō (1924–4, supp. vol. 1, 166–7, 170–71, 174–9); and Fóguāng (198, IV, 5–72). Only Taishō and Fóguāng use the Taishō sutta numbers. Only Anesaki (1905) has a table of Chinese-to-Pāli correspondences for T100: his ‘Text β’, which, however, follows the ‘Chinese arrangement’ discussed below. Only Akanuma gives Pāli-to-Chinese correspondences. And only Anesaki (1905) and Fóguāng recognize the restored scroll sequence for T99 discussed below.
12. For scholarly opinion on the likely background of the source text, see the summary provided by Glass (2006, 22–5). For the Sarvāstivāda attribution, see Mayeda (1985); for the MūlaSarvāstivāda, see Enomoto (1980; 1986, 2).
13. T II 1a2. Recent research dates the translation to 435–6 cE and raises questions about Guṇabhadra’s role in the translation team; see the review by Glass (2006, 7, 20–25).
14. In T99 the Sagātha-vagga comprises the suttas numbered 88–102, 576–603, 995–1022, 1062–120, 1145–63, 1178–240, 1267–362. This discontinuous distribution, seemingly at odds with the stated location of the vagga ‘at the end’, is discussed below.
15. A summary of the various published opinions is offered by Bingenheimer (2006, 22), who himself supports attribution to the (Mūla-)Sarvāstivāda.
16. In T100 the Sagātha-vagga comprises suttas 1–110, 132–42, 161–89, 214–329, 351–64. A few Sagātha-vagga suttas contained in T99 are not represented in T100, and vice versa. In the extreme case, of the twenty suttas making up the Māra-saṃyutta in T99 (nos. 1084–103) only the first ten have counterparts in T100 (nos. 2–2). Within short sections of text, T100 agrees very closely with T99 as regards the sutta sequence; see Bingenheimer (2006, 23–5, table). Another incomplete chinese translation of a Saṃyuktāgama, Taishō no. 101 at T II 49–8, contains only twenty-seven suttas, eight of them (nos. 1–5, 21, 25, 26) from the Sagātha-vagga, so cannot be made use of in this study.
Also useful as sources for this study are the Mūlasarvāstivāda-vinaya and the Yogācārabhūmi-śāstra. The Mūlasarvāstivāda-vinaya contains a list of the vaggas of the Saṃyuktāgama as known to the Mūlasarvāstivādin tradition (T XXIV 407b21–8). As for the Yogācārabhūmi-śāstra, one of its five divisions, the Vastusaṅgrahaṇī, is a partial commentary on a text that clearly resembled fairly closely the Saṃyuktāgama version preserved in chinese as T99. This commentary does not cover the Sagātha-vagga but is of value here in containing a further listing of the vaggas of the Saṃyuktāgama.17 T99, the complete Saṃyuktāgama translation by Guṇabhadra, contains only fragmentary indications of a division into vaggas, and none for saṃyuttas. Instead, it is divided mechanically into fifty equal-sized ‘scrolls’ (卷 juàn), a purely chinese development.18 It has long been known that the existing text is to some extent in disarray; some of the fifty scrolls must have been accidentally interchanged.19 In the restored ‘correct’ sequence, the section of T99 that contains the Sagāthavagga comprises the thirteen scrolls that now bear the numbers 38, 39, 40, 46, 42, 4, 44, 45, 36, 22, 48, 49, 50.20 Although most of the vagga titles (and subtitles) have been lost from T99, it is clear from those that remain, and from indications given in the Yogācārabhūmiśāstra and the Mūlasarvāstivāda-vinaya, that the text translated by Guṇabhadra was divided into the following seven vaggas:
1. Khandha-vagga (Section on the Aggregates)
2. Saḷāyatana-vagga (Section on the Sense-bases)
3. Nidāna-vagga (Section on causation)
4. Sāvakabhāsita-vagga (Section Spoken by disciples)
5. Magga-vagga (Section on the Path)
6. Buddhabhāsita-vagga (Section Spoken by the Buddha)
7. Sagātha-vagga (Section with Verses) 21
17. The chinese translation of Vastu-saṅgrahaṇī is at T XXX 772–882, the list being at 772c11–15; the Tibetan is Peking no. 5540 in vol. 111, 121–218, the list being at page 121, folio 143b1–5.
18. Although now printed in modern book format, the chinese canon continues to indicate the old division into scrolls (often called ‘fascicles’).
19. As demonstrated by Anesaki (1908); Lǚ (192); Mayeda (1964, 654–7); Yìnshùn (198, I i–lxxiv, esp. xli–liii); Mukai (1985); and others. For a summary, see Glass (2006, 25–30).
20. This explains the discontinuous distribution mentioned in n.14, above. According to the most recent research of Yìnshùn (1983, I xlv–xlix) and Mukai (1985, 18), the correct scroll sequence for the entire T99 is: 1, 10, 3, 2, 5–9, 43, 11, 13, 12, 14–21, [23], 31, 24, [25], 26–30, 41, 32–5, 47, 37–40, 46, 42, 4, 44, 45, 36, 22, 48–50. Scrolls 23 and 25 belong not to the Saṃyuktāgama but to the unrelated Aśokāvadāna, apparently having been used inappropriately to fill gaps created by accidental loss of two scrolls. The inferred original sequence of the scrolls is adopted in the Fóguāng edition of 198.
21. A heading ‘2. Saḷāyatana-vagga’ appears in the Yuán and Míng editions of the chinese Tripiṭaka, although not in the Taishō edition (see n.8 to T II 49b3). The Nidāna-vagga of the Saṃyuktāgama is identified by title at the beginnings of only the 4th and 5th of its five scrolls: ‘. Nidānavagga, Part 4’ and ‘3. Nidāna-vagga, Part 5’ (T II 108c27, 116c9). The Saṃyuktāgama counterpart
The Sagātha-vagga is the last of these seven. In the Yogācārabhūmi-śāstra it is actually called by a different title, discussed below. Four of the seven vagga titles in the above list, namely Khandha, Saḷāyatana, Nidāna, and Sagātha, match with titles in the Pāli SN. A fifth, Magga-vagga, clearly corresponds (on the basis of the vagga’s content) to the Pāli title, Mahā-vagga. The Pāli SN lacks counterparts for the two remaining titles, Sāvakabhāsita-vagga (Section Spoken by disciples) and Buddhabhāsita-vagga or Tathāgatabhāsita-vagga (Section Spoken by the Buddha).22 Most of the saṃyuttas that make up these two extra vaggas in T99 do exist in SN but are scattered throughout its second to fifth vaggas. The Sāvakabhāsita-vagga comprises six saṃyuttas whose component suttas are spoken by disciples (sāvaka) rather than by the Buddha, and which therefore form a natural group.23 The Buddh abhāsita-vagga comprises about nine saṃyuttas24 whose component suttas are spoken by the Buddha but are for some reason set apart from the remaining vaggas. In the case of T100, the incomplete ‘other Translation’ of the Saṃyuktāgama, an even more serious disarrangement of the text is known to have occurred. In this respect the Taishō edition of the Tripiṭaka (compiled by Japanese scholars in the 1920s), along with the thirteenth-century korean edition on which it is directly based, disagrees with some at least of the editions produced in china.25
of the Pāli Mahā-vagga is marked at its beginning: ‘5. Magga-vagga, Part 1’ (T II 170c27). Between Nidāna- and Magga-vaggas is a heading ‘4. Sāvakabhāsita-vagga’ (T II 126a3). (These five extant vagga headings are cited by Anesaki [1908, 70].) This scarcity of vagga titles helps explain how T99 could become disarranged: not one of the twelve transposed scrolls bears a title. The present scroll numbers on these twelve are taken to be secondary additions; the sutta numbers in T99 date only from the compiling of the Taishō edition (1924). There exist variations in the sequence of listing the seven vaggas. The Vinaya list, at T XXIV 407b21–8, has Buddhabhāsita before Magga rather than after it. The chinese Yogācārabhūmi-śāstra list, at T XXX 772c11–15, has Buddhabhāsita and Sāvakabhāsita at the beginning, ahead of Khandha; but the Tibetan, at Peking vol. 111, p. 121, f. 143b1–5, agrees with the Vinaya sequence.
22. Their distribution is shown in Mukai (1985, 18). A Sanskrit Saṃyuktāgama fragment described by Hosoda (1989, 541) has the word Buddhabhāṣita as a ‘running header’ on the folio. It is unclear whether there is any connection with the terminology of the Pāli Vinaya (Vin IV 15,9– 10), where the Dhamma is equated with four categories: Buddhabhāsita, Sāvakabhāsita, Isibhāsita, and Devatābhāsita.
23. The six are: Sāriputta, Moggallāna, Anuruddha, Mahākaccāna, Ānanda, Citta.
24. The saṃyuttas are Mahākassapa, gāmaṇī, Anamatagga, and others. The count of nine is approximate because of uncertainty regarding the boundaries of a few seeming saṃyuttas that lack evident counterparts in SN. The reason these saṃyuttas are grouped as a separate vagga is perhaps that they mostly do not deal with specific Dhamma topics. The title, Buddhabhāsita, ‘Spoken by the Buddha’, is odd, since it seems equally applicable to all vaggas other than Sāvakabhāsitavagga.
25. Just which editions have the Taishō arrangement and which the ‘Chinese’ arrangement is a question for future research. Editions having the chinese arrangement include those numbered 29, 35, 39, 43 in the list by Grönbold (1984, 24–5). (The Taishō edition and its Korean antecedent are Grönbold’s nos. 49 and 32, respectively.) How T100 corresponds with the chinese arrangement is shown in tabular form at Fójiào Dàzàngjīng XXV 867–8 (Grönbold’s no. 48). How the Sagātha-vagga section of T100 corresponds with that of the reconstituted T99, and with the It is evident that the Taishō arrangement (T100) has developed out of the Chinese arrangement through several accidental transpositions of textual material. The earlier (chinese) arrangement corresponds closely with the arrangement of the last two vaggas of the reconstituted T99, even to the sequence of nearly all the component suttas.26 There is, however, one substantial difference: in the antecedent of T100 (that is, the earlier chinese arrangement) the sequence of these two vaggas is Sagātha, Buddhabhāsita, while in the reconstituted T99 the sequence is Buddhabhāsita, Sagātha.27 In terms of vaggas, therefore, the arrangements found in our three sources are as shown in Table 1. Table 1. The vaggas of SN, T99, and T100
SN T99 T100 Sagātha Khandha Sagātha Nidāna Saḷāyatana Buddhabhāsita28 Khandha Nidāna Saḷāyatana Sāvakabhāsita Mahā
Buddhabhāsita
Sagātha
composition oF SAgathaa-Vagga
From the structure of SN, T99, and T100 in terms of vaggas, let us now turn to the structure of the Sagātha-vagga in terms of saṃyuttas. Neither in T99 nor in T100 are the saṃyuttas demarcated in any way. Nevertheless, most of the saṃyutta boundaries are easily discerned because many of the component suttas match, one to one, with those of the SN saṃyuttas. This can be illustrated in the following compari-
relevant scrolls of the chinese arrangement, is shown in Yìnshùn (1988, 669–72). The chinese arrangement (identified as ‘N. 546’, i.e. ‘Nanjio no. 546’) is the one that Anesaki refers to in his account of the four chinese Āgamas (1908, 70) and in his Sagātha-vagga correspondence table (1905, 31–7).
26. This is evident in the sequence of sutta numbers in Anesaki’s table (1905, 1–7). (‘Text α’ is T99; ‘Text β’ is the Chinese arrangement of T100.)
27. T100 preserves two vagga titles: ‘First Vagga’ and ‘Second Vagga’ (at T II 374a3 and 414a17). However, because the contents of T100 are seriously disarranged, the boundaries of these two vaggas have to be inferred from the distribution of the saṃyuttas in the version of the text preserved in chinese editions. In that version vagga titles are totally lacking, yet the vagga structure is clearly apparent. It reveals one anomaly: Vana-saṃyutta, properly the last saṃyutta of Sagātha-vagga, is actually located some distance away, after Buddhabhāsita-vagga. Probably Vana-saṃyutta was accidentally transposed from the end of First Vagga (originally containing the entire Sagātha-vagga) to the end of Second Vagga (originally containing only Buddhabhāsitavagga).
28. As regards content, T100 differs from T99 in lacking not only the first five of the seven vaggas (Khandha to Magga) but also the last four saṃyuttas of Buddhabhāsita-vagga. son based on the simplest and clearest case, that of the Bhikkhunī-saṃyutta. The ten suttas of T99 that correspond to the ten suttas of the Pāli Bhikkhunī-saṃyutta are those bearing the numbers 1198 to 1207; that is, the chinese counterpart suttas are similarly located together as a single block. The sequence is a little different; in terms of the Pāli numbering the sequence of the Chinese suttas is 1, 2, 3, 5, 10, 9, 4, 6, 7, 8. Exactly the same situation is found with the suttas of T100 that correspond to those in the Pāli Bhikkhunī-saṃyutta.29 That is to say, although the Bhikkhunī-saṃyutta in the two chinese Saṃyuktāgama texts is not demarcated, its boundaries can be readily discerned. The boundaries between saṃyuttas in T99 and T100 are not always so clear-cut. In some cases it is apparent that suttas properly belonging to a certain saṃyutta have somehow found their way into another. However, in only one case is such blurring of saṃyutta boundaries serious enough to obscure the essential structure. The case in question is that of the consecutive Devatā- and Devaputta-saṃyuttas. The cluster of eighty-two suttas that one would be inclined to identify as making up the Devatāsaṃyutta in T99 and T100 actually includes eleven suttas whose SN counterparts are in the Devaputta-saṃyutta instead.30 consequently, the boundary between these two saṃyuttas of T99 and T100 cannot be clearly located, a situation that perhaps originated in part out of difficulty in distinguishing devaputtas from devatās.31 despite such complications, there emerges one important fact concerning the structure of the two chinese versions: once allowance is made for the above-mentioned accidental transpositions of textual material in both T99 and T100, these two collections agree with each other completely as regards the sequence of the saṃyuttas within the Sagātha-vagga. Therefore, in matters pertaining to the internal structure of the Sagātha-vagga, T99 and T100 can be grouped together as a single case. Most saṃyuttas that are common to T99/T100 and SN are located in the same vagga in the two cases. The only exception that is significant here is the Bhikkhusaṃyutta; it is at the beginning of the Sagātha-vagga of T99/T100 but at end of the Nidāna-vagga of SN.32 This discrepancy means that the Sagātha-vagga of T99/T100 encompasses one saṃyutta more than the Sagātha-vagga of SN. The composition of the two sagāthā collections is as shown in Table 2.
29. In T100 the Bhikkhunī suttas are nos. 214–2 (at T II 45b–456a). Their sequence is just as in T99; see Anesaki (1905, 34).
30. T99 suttas 999, 1001, 583, 585, 588, 593, 595–7, 1269, 1276; cf. Anesaki (1908, 125–30).
31. This difficulty in distinguishing the two classes of gods is noted by Bodhi (2000, 75–6). The table of saṃyuttas offered in the Fóguāng edition (198, IV 11) groups Devatā and Devaputta together as a single group, called 諸天 (zhū tiān), ‘All Devas’; similarly Yìnshùn (1983, III 219). Also cf. Yìnshùn (1988, 672–3, 681), where the two are combined under one heading 天子天女 (tiānzǐ tiānnǚ), ‘devaputtas and devatās’.
32. SN II 273–86 (suttas 1–12); T II 276a–284b (T99 suttas 1062–83); T II 374a–381a (T100 suttas 1–22). outside the Sagātha-vagga there are two further exceptions: Vedanā-saṃyutta is in Saḷāyatanavagga of SN but in Nidāna-vagga of T99; and Sacca-saṃyutta is in Mahā-vagga of SN but in Nidānavagga of T99; see choong (2000, 19). Table 2. Saṃyuttas of Sagātha-vagga in SN and T99/T100. SN T99/T100
It is evident that, as regards their overall composition, these two versions of the Sagātha-vagga are likely to be divergent derivatives of a single ancestral version. one must ask, therefore, whether that ancestral version would have included the Bhikkhu-saṃyutta. In T99/T100 the Bhikkhu-saṃyutta comprises twenty-two suttas, all of which are in sagāthā form. In SN the Bhikkhu-saṃyutta comprises twelve suttas, of which all except the first two are in sagāthā form; and those two exclusively prose suttas of the SN Bhikkhu-saṃyutta lack counterparts in the T99/T100 Bhikkhu-saṃyutta. Seven suttas are common to the SN and T99/T100 versions of the Bhikkhu-saṃyutta, and all of them are in sagāthā form. In this respect, therefore, the situation seen in T99/T100, where the Bhikkhu-saṃyutta is located within the Sagātha-vagga, is the more natural and logical. Also relevant here is that the Sagātha-vagga contains a Bhikkhunī-saṃyutta. Pairing of texts relating to bhikkhus and bhikkhunīs is not uncommon elsewhere in the Tipiṭaka, a good example being the collections called Theragāthā and Therīgāthā. The Bhikkhu-saṃyutta forms an equally natural pair with the Bhikkhunī-saṃyutta, and would therefore be expected to be located close to it, at least in the same vagga. Taken together, these considerations make it likely that the T99/T100 version reflects the ancestral situation; at some early stage in the development of the Pāli tradition the Bhikkhu-saṃyutta of SN was probably also located in the Sagātha-vagga. In that case, how did the Bhikkhu-saṃyutta become separated from the other sagātha saṃyuttas within the Pāli tradition? In SN the Bhikkhu-saṃyutta (saṃyutta no. 21) is located as shown in Table 3.
Table 3. Location of Bhikkhu-saṃyutta within Saṃyutta-nikāya
Volume & vagga component saṃyuttas
I. Sagātha-vagga: 1. Devatā-saṃyutta … 11. Sakka-saṃyutta
II. Nidāna-vagga: 12. Nidāna-saṃyutta … 21. Bhikkhu-saṃyutta
III. Khandha-vagga: 22. Khandha-saṃyutta …
In the PTS edition, the Bhikkhu-saṃyutta is at the end of volume II, while the Sagātha-vagga occupies the whole of volume I. As can be seen, if one were to experimentally interchange the positions of the Sagātha- and Nidāna-vaggas – in effect, putting volume II ahead of volume I – then the Bhikkhu-saṃyutta (no. 21) would come to be adjacent to the other sagātha saṃyuttas (nos. 1–11). That is, the Bhikkhu-saṃyutta would become the first in a continuous series of twelve saṃyuttas with verses, just as in T99. In that situation (that is, with Nidāna-vagga preceding Sagātha-vagga in the Pāli), the suggested transfer of the Bhikkhu-saṃyutta amounts simply to a shift of the boundary between these two consecutive vaggas.
Rather than an accidental switching of volumes I and II of the PTS edition of SN, one needs, of course, to imagine a corresponding switching of large blocks or bundles of palm-leaf manuscript, each containing one vagga. This could have resulted from a careless replacing of the relevant manuscript bundles after use, a process resembling the mis-shelving of books when they are returned to a library shelf. The likelihood that such a shift actually happened is enhanced by the variable location of the Sagātha-vagga in the versions under consideration. While the Sagātha is the first vagga in the present SN, it is the last vagga in T99, and the second to last in T100. This variation shows that the location of the Sagātha-vagga within the Saṃyutta-nikāya/ Saṃyuktāgama did change over time in some of the traditions that preserved it.
In the case of SN, the postulated earlier vagga sequence, Nidāna, Sagātha, …, offers a possible explanation for the apparent shift of the Bhikkhu-saṃyutta: the Bhikkhu-saṃyutta was formerly the first saṃyutta of the Sagātha-vagga, but subsequently came to be thought of as the last saṃyutta of the preceding Nidāna-vagga. Relevant to this proposition is the numerical distribution of saṃyuttas between these two vaggas. Before the postulated transfer of the Bhikkhu-saṃyutta there would have been nine saṃyuttas in the Nidāna-vagga (the present nos. 12–20) and twelve in the Sagātha (the present nos. 21 and 1–11; see Table 3, above). The transfer would have yielded a more even distribution: ten saṃyuttas in the Nidāna-vagga and eleven in the Sagātha (as at present). Furthermore, since ten is the standard number of items in a vagga (whether ten saṃyuttas or ten suttas), the transfer would have given the Nidāna-vagga the standard number, again an outcome that would have been deemed desirable. Such considerations indicate that the transfer of the Bhikkhu-saṃyutta may even have been intentional.40 Physically, the transfer would have been easy to achieve if each of the SN saṃyuttas was contained in a single scroll or a single tied set of palm-leaves – an inherently likely situation.41 In this way the Bhikkhu-saṃyutta could have become part of the Nidāna-vagga; and when that vagga subsequently moved from first place to second place in the vagga sequence, the saṃyutta’s move from Sagāthavagga to Nidāna-vagga would have been rendered irreversible.42 This interpretation is viable only if the postulated transpositions are supposed to have occurred well before the time of Buddhaghosa (fifth century CE).
40. Of the five vaggas of the present SN, only the second and fourth, Nidāna and Saḷāyatana, have the ‘standard’ complement of ten saṃyuttas each. The third, Khandha-vagga, has thirteen saṃyuttas (nos. 22–34); however, three of these, Okkantika, Uppāda, and Kilesa (nos. 25–7), are demonstrably secondary developments, which means that this vagga too formerly had just ten saṃyuttas. (SN saṃyuttas 25, 26, 27 are represented in T99 by three suttas: nos. 892, 899, 900 [Yìnshùn 1983, III 553, 557–9, nn.5, 14, 15]; contra Akanuma [1929, 214–15]. This, plus their internal structure, indicates that these three Pāli ‘saṃyuttas’ were formerly three suttas, each of which was subsequently split into its ten component parts.) The fifth vagga of SN, Mahā-vagga, has twelve saṃyuttas, an irregularity typical of the last vagga in a set. Therefore, it is likely that, following the transfer of the Bhikkhu-saṃyutta and the subsequent switching of the Sagātha- and Nidānavaggas, the numbers of saṃyuttas in the five SN vaggas were: 11, 10, 10, 10, 12.
41. Admittedly, in existing Pāli palm-leaf manuscripts the grouping of the leaves into tied bundles often does not match the grouping of the textual material into saṃyuttas (cf. the similar mismatch between scrolls and saṃyuttas in T99 and in T100). However, the observed variation in the sequence of saṃyuttas between SN and T99/T100 indicates that in earlier times each saṃyutta did occupy an entire bundle; otherwise the demonstrable transposing of entire saṃyuttas could not easily have occurred. Also to be acknowledged is that in existing Pāli manuscripts the leaves are often ‘numbered’ consecutively (e.g., with ka, kā, ki, kī, …) throughout an entire vagga. But similar reasoning suggests that in earlier times the practice was to begin the numbering afresh in each saṃyutta. compare the relatively recent introduction of simple continuous numbering of suttas in T99 and T100 (since 1924), also in Somaratne’s new edition of the Sagātha-vagga of SN (1998), as discussed by Bodhi (2000, 69). In earlier times, when each saṃyutta was contained in a separate physical manuscript unit and individually paginated, the sequence of the saṃyuttas within the vagga would have been controlled only by the final uddāna – a point discussed later in this article.
42. compare the similar transposition of Vana-saṃyutta in T100 (n.27, above), which may have been associated with the switching of the Buddhabhāsita- and Sagātha-vaggas. This again indicates that each saṃyutta was written on an individual scroll. Buddhaghosa’s commentary on SN, Sārattha-ppakāsinī, deals with the five SN vaggas in the canonical sequence, Sagātha, Nidāna, Khandha, Saḷāyatana, Mahā; and it treats the Bhikkhu-saṃyutta as the last saṃyutta of the Nidāna-vagga, just as in the present SN text (SN-a I 2.22–24; II 233–48). Therefore, the present arrangement of SN certainly goes back to Buddhaghosa’s time, and may well go even further back to the old Sinhala commentaries, of which Buddhaghosa’s are essentially Pāli translations.
In this connection it is relevant to consider the following feature of Buddhaghosa’s commentaries on the four main Nikāyas. Each of the four commentaries includes a sutta-nikkhepa or ‘laying down of the sutta’, a section (amounting to about a page in the PTS editions) which explains four reasons why suttas were delivered. The wording is essentially identical in all four commentaries, although with some adaptation to the different textual contexts. In the commentaries on the Dīgha-, Majjhima-, and Aṅguttara-nikāyas, the fully-worded general statement of the sutta-nikkhepa is found within the section dealing with the first sutta of the Nikāya; however, in the commentary on the Saṃyutta-nikāya it is instead located within the section dealing with the first sutta of the second vagga, the Nidāna-vagga.
Since the general statement of the sutta-nikkhepa is applicable to all suttas in the Nikāya in question, the natural place for it is near the beginning of the entire commentary; that is, its location in the Dīgha, Majjhima, and Aṅguttara commentaries is as expected, but its location in the SN commentary is aberrant. This odd feature of the SN commentary is consistent with the proposition that, at some earlier time (earlier than Buddhaghosa), the section dealing with the Nidāna-vagga occupied first place in the (Sinhala) commentary rather than second place.
The facts just noted support the hypothesis that in SN itself the sequence of the vaggas was formerly Nidāna, Sagātha, … rather than Sagātha, Nidāna, …. When the Nidāna- and Sagātha-vaggas of SN switched places, the corresponding sections of the commentary were similarly switched. How such a switching of the first two SN vaggas could have occurred is not difficult to see, when one considers the traditional form of Pāli manuscripts. A historically later instance of just such a rearrangement in the Sārattha-ppakāsinī itself is documented in a study by Tseng. on examining twenty-two palm-leaf manuscripts of this commentary held in Sri Lankan monasteries, Tseng found that only nine agreed with the canonical Pāli vagga sequence: Sagātha, Nidāna, Khandha, Saḷāyatana, Mahā; the remaining thirteen interchanged the positions of the Nidāna- and Saḷāyatana-vaggas. This demonstrates that the main component sections of large Pāli manuscripts did sometimes get moved around relative to one another. It thereby supports the proposition that the vagga sequence of SN itself underwent a comparable change at some much earlier time.
The above reasoning suggests that within SN the Bhikkhu-saṃyutta formerly occupied first place in the Sagātha-vagga. It does not necessarily follow that this earlier form of the Pāli Sagātha-vagga can be reconstituted simply by attaching the Bhikkhu-saṃyutta at the beginning of the present collection (that is, ahead of Devatā-saṃyutta). This is because of the possibility that, following the transfer of the Bhikkhu-saṃyutta out of the Sagātha-vagga, the remaining eleven saṃyuttas may have undergone subsequent rearrangement. However, as regards content, rather than arrangement, it can be affirmed that the Pāli Sagātha-vagga formerly had the same twelve saṃyuttas as its counterparts in T99 and T100.
LINk WITH THE EIGHT ASSEMBLIES
Let us now turn to one of the additional sources of data mentioned earlier, the Saṃyuktāgama commentary contained in the Vastu-saṅgrahaṇī, the last section of the Yogācārabhūmi-śāstra. Detailed studies of this commentary, by Lǚ (192; 1992), Yìnshùn (1983), and Mukai (1985), have revealed that the text that the commentator had before him was broadly similar to T99.
Although the Sagātha-vagga is not actually dealt with in the commentary, it does receive a mention in its introduction, where the contents of the Saṃyuktāgama are summarized. The section in question reads as follows:
The Saṃyuktāgama is the one in which the teachings given out by the World-Honoured one are grouped thus: saṃyuttas spoken by the Tathāgata and by his disciples; saṃyuttas on the Aggregates, the Elements, and the Sense Fields; saṃyuttas on conditioned Arising, Nutriments, and the Truths; saṃyuttas on the Foundations of Mindfulness, the Right Efforts,
the Bases of Supernormal Power, the Faculties, the Powers, the Factors of Awakening, the Factors of the Path, Mindfulness of In- and out-breathing, the Trainees, and Serene Trust; and, furthermore, saṃyuttas that speak of the Assemblies according to the eight Assemblies.
(T XXX 772c14–15 ~ Peking vol. 111, 121, f. 143b1–5)
Apart from a difference in sequence, this summary of the contents of the Saṃyuktāgama corresponds fairly closely to the set of seven vaggas that can be identified in T99 and that are listed in the Sarvāstivāda-vinaya. There is, however, one conspicuous exception: instead of the expected section with verses (Sagāthavagga) the quoted summary speaks of a section based on the ‘eight Assemblies’ – chinese: 八眾 (bā zhòng); Tibetan: ’khor brgyad. The significance of this discrepancy is to some extent clarified by the subsequent discussion. There the listed sections of the Saṃyuktāgama are grouped into three categories: (1) ‘who speaks’, (2) ‘what is spoken’, and (3) ‘for whom it is spoken’; and these three are then explained thus:
(1) who speaks: ‘the sections spoken by the Tathāgata and by his disciples’;
(2) what is spoken: ‘sections comprising saṃyuttas on the Five Aggregates (as objects of) Clinging, the Six Sense Fields, and Conditioned Arising, together with the Path Section’;
(3) for whom it is spoken: ‘monks, devas, Māra, and so on, as in the chanted Section’. (T XXX 772c16–23 ~ Peking vol. 111, 121, f. 143b5–144a1)
What was earlier referred to as the section on the eight Assemblies is here called the ‘chanted Section’ – chinese: 結集品 (jié jí pǐn); Tibetan: brjod pa’i sde tshan – presumably representing Sanskrit Saṃgīta-nipāta. This, together with the mention of ‘monks, devas, Māra, and so on’, appears to confirm that the reference is indeed to the Section with Verses.
Further clarification can be found in the Mūlasarvāstivāda-vinaya. Whereas the Vastu-saṅgrahaṇī account of the Saṃyuktāgama describes the seventh vagga as ‘saṃyuttas that speak of the Assemblies according to the eight Assemblies’ (as quoted above), the otherwise very similar account in the Mūlasarvāstivādavinaya describes the seventh vagga as comprising ‘discourses connected with gāthās’. This correspondence confirms again that the ‘saṃyuttas that speak of the Assemblies according to the eight Assemblies’ are the saṃyuttas that make up the Sagātha-vagga. Also relevant here is a very similar list contained in the conclusion to an earlier section of the Yogācārabhūmi-śāstra. This is a list of nine topics (vastu) of the Buddha’s teaching, and although it does not specifically mention the Saṃyuktāgama, it amounts to a slightly rearranged version of the very same list. Its importance derives partly from the fact that, unlike the sections cited above from the Vastu-saṅgrahaṇī, this one happens to have survived in the Sanskrit. From the composition of the total list of topics it is clear that the last item corresponds to the Sagātha-vagga, yet this last item is named as Pariṣad-vastu, ‘Assemblies topic’. The meaning of this is then clarified with the words aṣṭau pariṣadaḥ, ‘eight Assemblies’, followed by a full list of the eight. This information confirms that the reference is to the set of eight classes of human and divine beings known in the Pāli suttas as the aṭṭha parisā, namely the following:
1. Khattiya-parisā (Assembly of Warriors)
2. Brāhmaṇa-parisā (Assembly of Priests)
3. gahapati-parisā (Assembly of Householders)
4. Samaṇa-parisā (Assembly of Renunciants)
5. Cātummahārājika-parisā (Assembly of the Gods of the Four Great kings)
6. Tāvatiṃsa-parisā (Assembly of the Thirty-three Gods)
7. Māra-parisā (Assembly of Tempter Gods)
8. Brahma-parisā (Assembly of Higher Gods)
The above facts indicate that, within the tradition represented by the Yogācārabhūmi-śāstra, there existed a belief that the Sagātha-vagga was closely connected with the eight Assemblies – a point highlighted by Yìnshùn (1983, I xxxi–xxxii). The remainder of the present article is therefore devoted to exploring the nature of this connection, to see what it may tell us about the structure of the Sagātha-vagga.
The eight Assemblies appear in the suttas as eight different classes of listeners to whom the Buddha teaches the Dhamma. In a stereotype passage, the Buddha speaks of his ability to adopt the language and manners of whichever assembly he happens to be instructing (dN II 109–10; MN I 72; AN IV 307). In a less specific reference, it is frequently said that the Tathāgata ‘roars his lion’s roar in the Assemblies and sets rolling the Brahma-wheel’ (e.g. MN I 69–71; SN II 27,25–7; AN II 7–8; III 417–19; V 33–8). Regarding the composition and sequence of the list of the eight Assemblies, the various sources generally agree. Exceptionally, the chinese translation of the Yogācārabhūmi-śāstra, attributed to 玄奘 Xuánzàng, gives the seventh assembly as Yāma where the corresponding Sanskrit version has the expected Māra.56
As Yìnshùn observes (198, I, xxxi), the first four of the eight Assemblies are types of human being (manussa), while the last four are types of divine being (deva). The four human Assemblies – Khattiya, Brāhmaṇa, gahapati, Samaṇa (Warrior, Priest, Householder, Renunciant) – appear in several suttas as an independent set.57 This set resembles another well-attested foursome: Khattiya, Brāhmaṇa, Vessa, Sudda – Warrior, Priest, Merchant-artisan, Worker.58 This is the Buddhist listing of the four Brahmanical ‘castes’; the Brahmanical listing differs in putting the Priest before the Warrior.59 The human section of the list of Assemblies differs from the Buddhist list of the four castes as follows: the third caste, Merchant-artisan (Vessa), is replaced in the Assemblies list by Householder (gahapati);60 and the fourth caste, Worker (Sudda), is absent and in its place is found Renunciant (Samaṇa). As regards the four deva Assemblies, the listing resembles closely the series of heavenly realms recognized in the Buddhist cosmology, especially given the interchanging of Māra with Yāma in the chinese Yogācārabhūmi-śāstra.61 In the cosmology, the Gods of the Four Great kings (Cātummahārājika) are located halfway up Mount Sineru, the Thirty-three Gods (Tāvatiṃsa) dwell on Sineru’s summit, the Yāma gods (Māras in the Assemblies list) are one level above Sineru, and
56. T XXX 294b1, where the term appears in chinese transcription as 焰摩 (yànmó). The corresponding note states that several earlier Chinese editions agree, although using a different transcription, 夜摩 (yèmó), but that the Tempyō manuscripts have just 魔 (mó), the usual (abbreviated) transcription for māra. Here yāma is certainly an error for māra, perhaps involving confusion with Yama, the god of death. A comparable discrepancy is found in two chinese parallels to dN 16 (Mahāparinibbāna-sutta). They identify the sixth assembly with the Tusita gods rather than the Tāvatiṃsa gods (T I 16b20, T I 192a8). 57. For example, AN IV 114,31–3 ~ T I 421b24–6 ~ T I 810a25–6 ~ T II 728c27–729a1; AN III 39,16–17 ~ T II 681a1; SN III 6,16–17 ~ T II 33c20. 58. For example, dN III 82,6–7 ~ T1:37b19–21 ~ T I 217a24–5; MN I 429,7 ~ T I 804c29. 59. As at MN II 177 ~ T I 661a7–8; here Brāhmaṇa precedes Khattiya, presumably because the speaker is himself a Brāhmaṇa. 60. Perhaps connected with this is an occasional obscuring of the distinction between the Vessa and the gahapati in the chinese Āgamas. Whereas the chinese term 長者 (zhǎngzhě) consistently corresponds to gahapati, the term 居士 (jūshì) corresponds sometimes to vessa (e.g. T I 477a26; T I 661a8) and sometimes to gahapati (e.g. T I 474b26–7; T I 16b19). The combination 長者居士 (zhǎngzhě jūshì) similarly corresponds sometimes to vessa (T II 120c10 ~ SN IV 219) and sometimes to gahapati (T I 192a7 ~ dN II 109,7).
61. For the series of heavens or gods see, for example, MN I 210,24–7 ~ T I 731a14–16, or SN V 423,17– 31 ~ T II 104a19–22. the Brahmā gods are at a higher level again.62 That is to say, Assemblies 5 to 8 are listed in the same sequence as the corresponding heavenly realms. The full list of eight Assemblies, therefore, brings together a recognized set of four human categories – 1. Khattiya, 2. Brāhmaṇa, 3. gahapati, 4. Samaṇa – and an approximation to a recognized set of divine categories – 5. Cātummahārājika,
6. Tāvatiṃsa, 7. Māra, 8. Brahmā.63
Let us now follow up the clue provided in the Yogācārabhūmi-śāstra and highlighted by Yìnshùn, by comparing the eight Assemblies with the twelve saṃyutta topics of the Sagātha-vagga, as represented in T99/T100 and as inferred for SN. Three categories are common to the two sets, namely Brāhmaṇa, Māra, and Brahmā. A further three categories are easily correlated: the Bhikkhu and Bhikkhunī are renunciants and therefore belong to the Samaṇa assembly; Sakka is the leader of the Tāvatiṃsa gods; and the subject of the Kosala-saṃyutta, king Pasenadi of kosala, is a Khattiya. Relevant to the last-mentioned case is the fact that the uddāna or ‘table of contents’ of the Pāli Sagātha-vagga names the saṃyutta in question as Rājā, ‘king’, rather than Kosala (SN I 240,22), thereby recognizing that the subject of the saṃyutta is the king himself, rather than his kingdom. Four of the five remaining saṃyuttas relate to divine beings, namely Devatā, Devaputta, Vana, and Yakkha, while one of the two remaining Assemblies refers to a four-membered set of divine beings, the Cātummahārājikā Devā or ‘Devas of the Four Great kings’. The latter set of devas is discussed in the Mahāsamaya- and Āṭānāṭiya-suttas,64 along with the kings of the Four directions who rule over them. There the Devas of the Four Great kings are named as gandhabbas, Kumbhaṇḍas, Nāgas, and Yakkhas. Thus, the four saṃyuttas in question (Devatā, Devaputta, Vana, Yakkha) and the set of devas associated with the four Great kings (Gandhabbas, Kumbhaṇḍas, Nāgas, Yakkhas) have only the Yakkha in common. In this case, therefore, the correspondence between saṃyutta topics and Assemblies is loose but nevertheless discernible.
This leaves one saṃyutta topic, Vaṅgīsa, and one Assembly, the gahapati (Householder). Vaṅgīsa is a senior monk (thera), who is an outstandingly gifted poet and (according to the last sutta in the saṃyutta) an Arahant.65 He is no householder, and therefore does not belong to the one remaining Assembly. Despite this one exception, a significant degree of correspondence has been
62. In the cosmology three more classes of gods (Tusita, Nimmānarati, Paranimmitavasavatti) come between the Yāma and Brahmā realms; e.g. SN V 423,29–30 ~ T II 104a19–21.
63. This combination broadly resembles a frequently attested five-membered set exemplified in the following two passages from the Dhamma-cakka-ppavatana-sutta (SN 56.11): sadevake loke samārake sabrahmake sassamaṇa-brāhmaṇiyā pajāya (SN V 423,1–2); samaṇena vā brāhmaṇena vā devena vā mārena vā brahmuṇā vā (SN V 423,20–21). These bring three divine categories (deva, māra, brahmā) together with two human categories (samaṇa, brāhmaṇa) in two different ways. 64. For Mahāsamaya: dN II 257,7–23 ~ T I 79c29–80a7 ~ T I 258b29–c9; Sanskrit version in Waldschmidt (1980, 155–6, verses 13–16); and, with parallel translation of the chinese versions, in Waldschmidt (1932, 364–7). For Āṭānāṭiya: dN III 197,1–8, 198,1–8, 198,37–199,2, 202,7–14; cf. T XXI 217a29–b29; Sanskrit version in Hoffmann (1939, 41–9); repr. Sander (1987, 57–65).
65. Bodhi (2000, 84) rates the quality of Vaṅgīsa’s poetry as by far the best in the entire Sagāthavagga. found between the saṃyuttas of the Sagātha-vagga and the eight Assemblies. Eight of the twelve saṃyuttas match up closely and a further three match up more loosely. This is shown in Table 4, in which the saṃyuttas are set out according to the Assemblies sequence, and the unmatched items are labelled #. Table 4. The eight Assemblies and corresponding saṃyuttas of Sagātha-vagga
Assemblies Saṃyuttas
5. Cātummahārājika Devatā, Devaputta, Vana, Yakkha
This correspondence, incomplete though it is, indicates that the Yogācārabhūmiśāstra’s linking of the Sagātha-vagga with the eight Assemblies is a statement of the principle underlying the vagga’s composition. A partial answer is thereby provided to the question raised at the beginning of this article. It can now be asserted, provisionally at least, that the twelve topics covered in the Sagāthavagga correspond – with one unexplained exception – to the twelve classes of being represented in the eight Assemblies.
SEQUENCE OF THE SAṂYUTTAS
In the above comparison with the Assemblies, the sequence of the saṃyuttas differs more or less from the sequences found in the Pāli and Chinese versions of the Sagātha-vagga. Just how these two versions of the vagga match up with the Assemblies sequence is shown in Table 5. Here the saṃyuttas of each version of the Sagātha-vagga are set out in their actual order, but labelled with the numbers of their corresponding Assemblies. A glance at these numbers reveals that the sequence of the T99/T100 saṃyuttas resembles the Assemblies sequence more closely than does the sequence of the SN saṃyuttas. Particularly striking in the T99/T100 version is that the four saṃyuttas that we have identified with Assembly no. 5 (the cātummahārājika devas) are all together in one block. This distributional feature can hardly be fortuitous. It suggests that the set of eight Assemblies may have influenced not only the content of the Sagātha-vagga but also its arrangement, at least in the case of the T99/T100 version, which we therefore now examine more closely.
The suggestion that the saṃyutta sequence in the Sagātha-vagga of T99/T100 could be based on the Assemblies list may appear to be counter-indicated by the case of the Bhikkhu- and Bhikkhunī-saṃyuttas: these saṃyuttas both correspond
Table 5. The saṃyuttas of the Sagātha-vagga in SN and in T99/T100, labelled with the numbers of their corresponding Assemblies
SN T99/T100
4. Bhikkhu
to Assembly no. 4, the Samaṇa-parisā, yet they are widely separated in T99/T100. This separation could, however, be a secondary development. As noted earlier, the saṃyutta sequence in both T99 and T100 is known to have been disrupted through accidental transposition of some of the scrolls. It is relevant, therefore, to consider whether the present separation of the Bhikkhu- and Bhikkhunī-saṃyuttas could be due to a similar, but historically much earlier, process of accidental transposition.
There is no direct evidence that such a transposition ever occurred. Let us, nevertheless, test the speculative proposition that it did occur by attempting, experimentally, to reverse the process. There are several possible ways of doing this. The model provided by the well-researched case of T99 suggests an interchanging of equal-sized blocks of text. In its simplest form, this could mean exchanging the Bhikkhunī-saṃyutta with the Māra-saṃyutta, the one that is currently located immediately after the Bhikkhu-saṃyutta. This exchange, besides achieving the intended effect of placing Bhikkhunī next to Bhikkhu, would have the incidental consequence of placing Māra next to Brahmā, which is an ‘acceptable’ outcome, given that the Assemblies corresponding to these two saṃyuttas are also adjacent in the Assemblies list: 7. Māra-parisā, 8. Brahmā-parisā. There is, however, an alternative possibility that deserves consideration. Instead of simply exchanging Bhikkhunī with Māra, one could exchange the pair of consecutive saṃyuttas, Bhikkhunī and Vaṅgīsa, with the likewise consecutive pair, Māra and Sakka, in the manner shown in Table 6. The saṃyutta sequence that results from making this simple exchange is shown on the right. It possesses some noteworthy characteristics. Not only is Bhikkhunī next to Bhikkhu as intended; in addition, each of the two broad categories, the human and the
divine, is now intact, as in the Assemblies list. Furthermore, the sequence of the corresponding Assemblies numbers is remarkably regular. When read upwards from the bottom, it runs: 5, 5, 5, 5, 6, 7, 8 (divine category); 2, 1, 3, 4, 4 (human category). With just one exception, the numerical sequence within each of the two broad categories is simply the reverse of that in the Assemblies list. The exception (1, 2, instead of 2, 1) involves the Kosala- and Brāhmaṇa-saṃyuttas, corresponding to the Khattiya and Brāhmaṇa Assemblies respectively; and even this finds a partial correlation in the Assemblies, given the above-noted variability in the listing of the corresponding castes: Khattiya, Brāhmaṇa versus Brāhmaṇa, Khattiya.
The saṃyutta sequence shown in the right-hand column in Table 6 has resulted from an experimental interchanging of two blocks of text in the T99/T100 Sagātha-vagga, with the intention of bringing together the Bhikkhu- and Bhikkhunīsaṃyuttas, as implied in the Assemblies list. Yet, as just seen, this transposition results in a much more far-reaching regularity and a surprisingly high degree of correspondence with the Assemblies list. This outcome is a strong indication that the experimental transposition reflects a historical reality. It implies that the transposition depicted in Table 6 reverses a switching of textual materials that did actually occur. In other words, the arrangement shown on the right (under ‘outcome’) actually existed at some early time; and it was subsequently converted into the arrangement shown on the left (the present T99/T100), when the Bhikkhunī-Vaṅgīsa and Māra-Sakka pairs were accidentally interchanged through a movement contrary to that indicated by the arrows.
The natural final step in this line of reasoning is to infer that the reconstructed earlier saṃyutta sequence (Table 6, right column) was itself derived from the Assemblies sequence. This still earlier development will have involved reversing the sequence of the Assemblies within both the human and the divine categories.
A possible motive for this reversing of the sequence is not hard to find. The intention may have been to produce a descending series that would put the Samaṇa in first place and the Khattiya ahead of the Brāhmaṇa, in accordance with the Buddhist evaluation of these human classes. In view of its subsequent development, as preserved in T99/T100, this series (Table 6, right column) might appropriately be termed the ‘pre-Sarvāstivādin sequence’. It has been shown, then, that the sequence of the saṃyuttas in the present T99 and T100 can be derived in a simple manner from the sequence of the eight Assemblies. Here let it be recalled that, even after restoration of the earlier scroll sequence, T99 and T100 are not entirely identical in content and arrangement; for example, the Sagātha-vagga comes after the Buddhabhāsita-vagga in T99 but before it in T100; also, some suttas present in T99 are lacking in T100, and vice versa. It follows that these two chinese translations are likely to represent two slightly different Sanskrit Saṃyuktāgama texts. Yet T99 and T100 agree as regards the sequence of the saṃyuttas within the Sagātha-vagga. Therefore, that sequence was common to the two Sanskrit textual traditions witnessed in these two chinese versions. All of the movements of sagātha material identified above will, therefore, have occurred before the divergence that yielded those two Indian textual traditions.67
Having accounted for the saṃyutta sequence of the Sagātha-vagga preserved in T99/T100 (Table 5, right column), we turn now to its SN counterpart (Table 5, left column). As noted earlier on the basis of Table 5, the saṃyutta sequence of the Sagātha-vagga in SN resembles the Assemblies sequence less closely than does its counterpart in T99/T100. In the T99/T100 version the four saṃyuttas corresponding to Assembly no. 5 – Devatā, Devaputta, Yakkha, Vana – are all together as a single block; but in SN these four are in two widely separated pairs: Devatā–Devaputta at the beginning of the list, and Vana–Yakkha near the end. Furthermore, these two pairs are the only features of the SN sequence that hint at a possible connection with the Assemblies sequence.68 Also worth looking out for are any features of the SN sequence that might link it with the pre-Sarvāstivādin sequence, the immediate precursor to the present T99/T100 saṃyutta sequence (Table 6, right column). The Sarvāstivādin and Vibhajjavādin traditions, which T99/T100 and SN respectively represent, appear to have diverged at some time between the Second and Third councils, perhaps a century and a half after the death of the Buddha. The pre-Sarvāstivādin sequence may have already existed before this divergence, in which case it could be a precursor of the SN sequence as well. With a view to exploring this possibility, the two sequences are compared in Table 7.
67. And therefore well before the similarly accidental movements that produced the disrupted scroll sequence seen in T99 and T100; cf. previous note.
68. Besides the two pairs of 5, the consecutive numbers, 2– and 5–6, are superficially suggestive; however, experimentation reveals that the total SN sequence cannot be derived in any plausible way from the Assemblies sequence. cf. following note. Table 7. The saṃyuttas of the SN Sagātha-vagga, numbered according to the Assemblies sequence and compared with the pre-Sarvāstivādin sequence from Table 6.
SN sequence Pre-Sarvāstivādin sequence
4. Bhikkhu
Here the shared pairs, Devatā Devaputta and Vana Yakkha, corresponding to Assembly no. 5, are again in evidence; and in addition there is a third pair shared in common: Brahmā Brāhmaṇa / Brāhmaṇa Brahmā. That is, the saṃyutta sequence in the present SN Sagātha-vagga (Table 7, left column) resembles the preSarvāstivādin sequence (right column) rather more closely than it resembles the Assemblies sequence. Even so, the gap between the two sequences shown in Table
7 is difficult to bridge. Experimentation reveals that it is impossible to derive the SN sequence from the pre-Sarvāstivādin sequence in a simple manner comparable to that identified above for T99/T100 (Table 6).
Having found that the SN Sagātha-vagga sequence (Table 7, left column) cannot be readily derived either from the Assemblies sequence or from its pre-Sarvāstivādin descendant, we now explore a rather different avenue. The sequence of the eleven saṃyuttas that currently make up the Sagātha-vagga of SN is recorded in the uddāna that appears at the very end of the vagga. This versified table of contents (a feature not matched in T99/T100) reads as follows: Devatā Devaputto ca Rājā Māro ca Bhikkhunī Brahmā Brāhmaṇa Vaṅgīso Vana-Yakkhena Vāsavo. Here the expected Kosala in third place is replaced by Rājā (king), as mentioned earlier; and at the end Sakka is represented by Vāsava, a common epithet for this leader of the Tāvatiṃsa gods.71 These name substitutions are likely to have been prompted by ‘poetical’ considerations, such as metrical constraints. Also attributable to such considerations are the two occurrences of ca, ‘and’, the uniting of Vana and Yakkha to form one compound noun, and the addition to this of the singular instrumental suffix –ena, ‘with’. With the aid of these devices, which are common features of uddānas in general, the list of saṃyutta titles has been preserved as an easily memorized piece of verse.
Besides satisfying the basic requirements of the metre (the very common vatta metre), the quoted uddāna possesses some less essential poetical features. There are three instances of alliteration: Devatā Devaputto, Brahmā Brāhmaṇa, and Vaṅgīso Vana-Yakkhena Vāsavo. of these, Devatā Devaputto and Brahmā Brāhmaṇa are very effectively balanced against each other through being located at the beginning of the first and second lines respectively. In addition, each of them exemplifies the Pāli literary device known as ‘waxing syllables’; the second word has one syllable more than the first, the syllable count being , 4 for Devatā Devaputto, and 2, 3 for Brahmā Brāhmaṇa.72 Unlike these two pairs, the third instance of alliteration, Vaṅgīso Vana-Yakkhena Vāsavo, is rather artificial and forced. It has been achieved by suppressing Yakkha through suffixing it to Vana, and replacing Sakka with the alternative name Vāsava. Most Pāli uddānas barely satisfy even the minimal metrical requirements and therefore well deserve the description ‘doggerel’.73 The existence of the abovenoted literary embellishments (alliteration, waxing syllables) in the uddāna of the Sagātha-vagga is unusual, which raises some significant questions. Given the techniques available to the redactors (for example, ad lib. insertion of ca), it is not at all remarkable that the given list of eleven saṃyutta titles (Devatā to Sakka) has been successfully fitted to the relevant metrical pattern. What is remarkable is that this list of saṃyutta titles should have also, so conveniently, yielded the other,
71. See Malalasekera (1974, II 857–8). In the Sakka-saṃyutta itself, Sakka is called Vāsava seven times, in each case evidently for metrical reasons; e.g. SN I 221,32 and 223,17. While the uddāna to Sagātha-vagga calls the eleventh and last saṃyutta Vāsava, that saṃyutta itself concludes with the words, Sakka-saṃyuttaṃ samattaṃ, ‘The Sakka-saṃyutta is concluded’ (SN I 240,20–21). Similarly, Kosala-saṃyuttaṃ samattaṃ at the end of the saṃyutta which the uddāna calls Rājā (SN I 102,30).
72. On waxing syllables in Pāli literature, see von Hinüber (1994, 16–0); and Allon (1997, e.g. 48). In a series employing this principle, each word has more syllables than its predecessor (or, within such a series, the same number of syllables); e.g. soko, socanā, socitattaṃ (‘sorrow, sorrowing, sorrowfulness’ at MN III 249), with 2, 3 and 4 syllables respectively. The words in question need not be alliterated and they need not be synonyms; but usually they do share some phonetic element and do have a semantic or doctrinal affinity. 73. Rhys davids and Stede 1925, 135, s.v. uddāna.
non-essential literary features. Particularly surprising are the cases of Devatā Devaputto and Brahmā Brāhmaṇa, with their apparently ready-made alliteration and waxing syllables. It is hard to believe that the composers of the uddāna could have been so fortunate as to find these two felicitous collocations already existing in the given list of saṃyutta titles and, what is more, so placed within it that they fell naturally at the beginning of the uddāna’s first and second lines.
The likelihood of this happening fortuitously is surely remote, which suggests that those who composed this uddāna may have followed a procedure rather different from the usual. In the vast majority of cases it is probably safe to assume that an uddāna was composed according to the sequence of the saṃyuttas (or suttas) in the collection it refers to. However, in the case of this Sagātha-vagga uddāna the evidence suggests it was the other way round: the sequence of the saṃyuttas was determined by the uddāna. The observed facts indicate that the monks responsible for preserving the Pāli Sagātha-vagga set out to compose an uddāna that would have maximum poetic-mnemonic effect,74 and then rearranged the saṃyutta sequence to match it.
This interpretation generates two further questions: what could have motivated the creation of an uddāna that would, so atypically, entail changing the sequence of the saṃyuttas listed within it? And what was the sequence of the saṃyuttas before the creation of this uddāna caused them to be rearranged? The Sagātha-vagga uddāna does not include Bhikkhu in its list of saṃyutta titles, so it must have been composed after the Bhikkhu-saṃyutta was transferred from the Sagātha-vagga to the Nidāna-vagga. That the Bhikkhu-saṃyutta could be transferred between vaggas in this way suggests that, at the time in question, no uddāna existed for the Sagātha-vagga (and similarly for the Nidāna-vagga).75 In any case, it is self-evident that the new distribution of saṃyuttas that resulted from this transfer would have provided a good reason for the creation of a corresponding uddāna – or if one already existed, for the creation of a new one to replace it. It was argued above that the structure of this newly created uddāna was guided not by the then existing saṃyutta sequence but rather by poetical considerations. The reason behind this unusual approach may lie in the nature of the collection in question, the Section with Verses (Sagātha-vagga). The collection itself contained some highly valued pieces of poetry, including works by the esteemed monk-poet, Vaṅgīsa. Perhaps this made the editors feel obliged to strive for some degree of poetic merit in the corresponding uddāna.
More difficult to answer is the second question: what was the sequence of the twelve saṃyuttas that made up the Sagātha-vagga of SN just prior to these changes? It has already been found that the existing saṃyutta sequence shows
74. In that case, why didn’t they do a better job with the third instance of alliteration, for example by putting Vaṅgīsa (rather than Yakkha) in the fourth pāda with Vana and Vāsava? The answer may lie in the combination Vana-Yakkhena. This is a fairly natural grouping, given the association of Yakkhas with trees and forests (Rhys davids and Stede 1925, 545 s.v. yakkha). However, the intention may have been to portray this as a single item in the list and thereby give the superficial impression that the Sagātha-vagga had just ten saṃyuttas (cf. n.40, above).
75. As mentioned in n.43, above. no clear sign of being derived from the pre-Sarvāstivādin sequence, the immediate precursor to the T99/T100 sequence (Table 7). This lack of evident continuity with the Sarvāstivādin line of development has now been accounted for: the existing SN sequence resulted from a rearrangement of the eleven saṃyuttas that remained following the transfer of the Bhikkhu-saṃyutta out of the Sagātha-vagga. The available data do not allow us to infer with any certainty the sequence of the twelve saṃyuttas prior to these changes. clearly, that sequence is inherently likely to have been the pre-Sarvāstivādin sequence or something very like it. However, tangible evidence that this was the case is meagre. All that we have is a single correlation: the Bhikkhu-saṃyutta is located at the beginning of the pre-Sarvāstivādin sequence; and the Bhikkhu-saṃyutta was probably also at the beginning of the SN Sagātha-vagga sequence prior to being transferred to the Nidāna-vagga (see the earlier discussion of the likely mechanism of this transfer, pp. 15–18). our examination of the saṃyutta sequence in the two versions of the Sagātha-
vagga (T99/T100 and SN) vis-à-vis the Assemblies sequence has yielded two very different outcomes. In the case of the T99/T100 version we have found that the saṃyutta sequence is readily accounted for as having developed out of the Assemblies sequence in two simple steps. The first step, which clearly was intentional, consisted in reversing the sequence of listing the Assemblies in each of the two broad categories, the human and the divine, yielding what we have termed the ‘pre-Sarvāstivādin sequence’. The second step, which probably was accidental, consisted in interchanging two pairs of adjacent saṃyuttas: Bhikkunī-Vaṅgīsa switched places with Māra-Sakka.
In the case of the SN version, however, no evidence has been found that the sequence of the present eleven saṃyuttas might have developed out of either the original Assemblies sequence or the pre-Sarvāstivādin sequence: an unexpected outcome, given the inherent likelihood of a common origin for the SN and T99/ T100 sequences. This situation is explained by the finding that the sequence of the eleven SN saṃyuttas is likely to be of more recent origin, having been determined by the newly created uddāna following the transfer of the Bhikkhu-saṃyutta.
Conclusion
This search for a principle underlying the structure of the Sagātha-vagga has confirmed a strong link with the eight Assemblies. The link has been demonstrated in two areas: content and sequence.
As regards content, it has been shown that the topics of the twelve saṃyuttas of the ancestral Sagātha-vagga correspond closely with the twelve classes of beings represented in the eight Assemblies. There is just one substantial exception: where the Assemblies list has the Householder (gahapati-parisā), the Sagātha-vagga has the monk-poet, Vaṅgīsa (the Vaṅgīsa-saṃyutta).
As regards sequence, it has been argued that the sequence of the twelve saṃyuttas in the T99/T100 version of the Sagātha-vagga is likely to have developed out of the Assemblies sequence in two simple steps, of which the first was intentional and the second accidental. In this case, therefore, the findings based on sequence support the findings based on content. Such support is lacking in the case of the SN version: the sequence of the eleven SN saṃyuttas has no evident connection with the sequence of the Assemblies. This lack of connection has been shown to be attributable to editorial rearrangement of the SN Sagāthavagga following the transfer of the Bhikkhu-saṃyutta; it does not, therefore, count against the argument, based on content, that this version of the Sagātha-vagga also derives ultimately from the eight Assemblies.
The above summary highlights again the discontinuity occasioned, in the Pāli trajectory, by the transfer of the Bhikkhu-saṃyutta out of the Sagātha-vagga and the subsequent rearrangement of the eleven remaining saṃyuttas. That such changes were made suggests that the people responsible within the Pāli tradition had no sense of a rational principle underlying the Sagātha-vagga. Indeed, nothing in the Pāli Nikāyas hints at a surviving awareness of an ultimate connection between the Sagātha-vagga and the set of eight Assemblies. In contrast, the Sarvāstivādin Āgamas do contain evidence of such an awareness. The saṃyutta sequence of the Sagātha-vagga preserved in T99/T100 departs only minimally from the preSarvāstivādin sequence, suggesting that the latter sequence survived over a long period. A similar contrast is evident in the commentarial literature. The Yogācārabhūmi-śāstra, commenting on a text very similar to T99, explicitly links the Sagātha-vagga with the eight Assemblies; the Sārattha-ppakāsinī, commenting on the Pāli SN, makes no mention of such a connection. It appears, therefore, that the Sagātha-vagga’s connection with the eight Assemblies was still recognized within certain Buddhist schools on the Indian mainland long after it had been forgotten within the Pāli tradition.
The findings arrived at through this analysis go a long way toward answering the questions posed at the outset. It can now be affirmed that the compiling of the Sagātha-vagga was indeed intended to do more than bring together pieces of the memorized Dhamma that happened to contain gāthās. The intention was to create a coherent collection with a definite underlying structure. This explains why much material in sagāthā form (for example, many suttas contained in the Aṅguttara-nikāya) was not included in the Sagātha-vagga: this material could not be made to fit any of the Assemblies categories. These answers in turn raise further questions, of which the most obvious is: why should the eight Assemblies, a very marginal piece of doctrine, have been chosen for this purpose? They also raise wider issues regarding the intentions of those involved in compiling the Saṃyutta-nikāya/Saṃyuktāgama. This points to the desirability of undertaking comparable research into the structure of the entire Saṃyutta-nikāya/Saṃyuktāgama along the lines already sketched out by Yìnshùn. For the present, however, I hope that it suffices to have demonstrated the likely existence, within one section of this collection, of a rational structure that had long remained largely obscured.
I am gratefully indebted to Mark Allon, Ven. Anālayo, Choong Mun-keat, and Primož Pecenko for their constructive comments on a draft of this article.
ABBREVIATIoNS
AN Aṅguttara-nikāya PTS Pāli Text Society
AN-a Aṅguttara-nikāya commentary SN Saṃyutta-nikāya
dN Dīgha-nikāya SN-a Saṃyutta-nikāya commentary
dN-a Dīgha-nikāya commentary T Taishō Shinshū Daizōkyō; see Bibliography
Fóguāng Fóguāng edition of the Saṃyuktāgama Vin Vinaya
in chinese; see Bibliography Vin-a Vinaya commentary
MN Majjhima-nikāya ~ ‘corresponding to’ (in references to Pāli/
MN-a Majjhima-nikāya commentary chinese parallel discourses)
Peking Tibetan Tripiṭaka, Peking edition; s
ee Bibliography
References to Pāli texts are to PTS editions.
BIBLIoGRAPHY
Akanuma, chizen. 1929. The Comparative Catalogue of Chinese Āgamas & Pāli Nikāyas. Nagoya: Hajinkaku Shobō. (Reprinted, Delhi: Sri Satguru, 1990 [[[Bibliotheca Indo-Buddhica]] no. 74].)
Allon, Mark. 1997. ‘The oral composition and Transmission of Early Buddhist Texts’. In Indian Insights: Buddhism, Brahmanism, and Bhakti, ed. Peter connolly and Sue Hamilton, 39–61. London: Luzac oriental.
Anesaki, Masaharu. 1905. ‘Le Sagātha-vagga du Saṃyutta-nikāya et ses versions chinoises’. Le Muséon (nouvelle série) 24: 23–37.
Anesaki, Masaharu. 1908. ‘The Four Buddhist Āgamas in chinese’. Transactions of the Asiatic Society of Japan 35(3): 68–76.
Bechert, Heinz. 1980. Die Sprache der ältesten buddhistischen Überlieferung [The language of the earliest Buddhist tradition] (Symposien zur Buddhismusforschung II). Göttingen: Vandenhoeck & Ruprecht.
Bhattacharya, Vidhushekhara, ed. 1957. The Yogācārabhūmi of Ācārya Asaṅga: The Sanskrit Text Compared with the Tibetan Version, Part I. calcutta: University of calcutta Press.
Bingenheimer, Marcus. 2006. ‘The Shorter chinese Saṃyukta Āgama–Preliminary Findings and Translation of Fascicle 1 of the Bieyi za ahan jing 別譯雜阿含經 (T.100)’. Buddhist Studies Review 23(1): 21–60.
Bodhi, Bhikkhu, trans. 2000. The Connected Discourses of the Buddha, 2 vols. Boston: Wisdom.
Bucknell, Rod. 2006. ‘Saṃyuktāgama’. Encyclopaedia of Buddhism, vol. VII (fascicle 4): 684–7. choong, Mun-keat. 2000. The Fundamental Teachings of Early Buddhism: A Comparative Study Based on the Sūtrāṅga Portion of the Pāli Saṃyutta-Nikāya and the Chinese Saṃyuktāgama (Beiträge zur Indologie 32). Wiesbaden: Harrassowitz.
Choong, Mun-keat. 2006. ‘A Comparison of the Pāli and Chinese Versions of the Bhikkhu Saṃyutta, a collection of Early Buddhist discourses on Monks’. Buddhist Studies Review 23(1): 61–70.
Enomoto, Fumio 榎本文雄. 1980. ‘Udānavarga shohon to Zō-agongyō, Betsuyaku Zō-agongyō, Chūagongyō no buha kizoku’ Udānavarga 諸本と雜阿含經, 別譯雜阿含經, 中阿含經の部派歸屬 [on the recensions of Udānavarga and the sectarian affinities of the Chinese Saṃyuktāgama (T99 & T100) and Madhyamāgama]. Indogaku Bukkyōgaku Kenkyū 18(2): [55]–[57].
Enomoto, Fumio. 1986. ‘On the Formation of the Original Texts of the Chinese Āgamas’. Buddhist Studies Review 3(1): 9–30.
Enomoto, Fumio. 1994. A Comprehensive Study of the Chinese Saṃyuktāgama: Indic Texts Corresponding to the Chinese Saṃyuktāgama as Found in the Sarvāstivādin-Mūlasarvāstivādin Literature, Part 1: *Saṃgītanipāta.
kyoto: kacho Junior college.
Fóguāng Dàzàngjīng: Āhán zàng: Zá āhán jīng 佛光大藏經: 阿含藏: 雜阿含經 [Fóguāng Tripiṭaka: Āgama: Saṃyuktāgama] 198. ed. Fóguāng Dàzàngjīng Biānxiū Wěiyuánhuì. Kao-hsiung: Fóguāng Chūbǎnshè.
Fójiào Dàzàngjīng 佛教大藏經 [[[Wikipedia:Chinese|Chinese]] Tripiṭaka]. 1978. Taipei: Fójiào Chūbǎnshè (Repr. of the Pínjiā Vihāra Tripiṭaka of 1909–14).
Glass, Andrew. 2006. Connected Discourses in gandhāra: A Study, Edition, and Translation of Four SaṃyuktāgamaType Sūtras from the Senior Collection. Ph.d. dissertation, University of Washington. www.andrewglass. org/phd.php (summary); www.andrewglass.org/downloads/Glass_2006.pdf (full text), accessed 24 February 2007.
Grönbold, Günter. 1984. Der buddhistische Kanon: eine Bibliographie. Wiesbaden: Harrassowitz.
Hoffmann, Helmut. 199. Bruchstücke des Āṭānāṭikasūtra aus dem zentralasiatischen Sanskritkanon der Buddhisten (kleinere Sanskrit-texte, Heft V). Leipzig: Brockhaus. Reprinted as Nachträge zu ‘Kleinere Sanskrit-texte, hefte III-V’, ed. Lore Sander (Stuttgart: Franz Steiner, 1987).
Hosoda, Noriaki細田典明. 1989. ‘Torufan shōrai mokuhansatsu Bonbun Zōagongyō danken: R. Pischel kōhyō Bonbun danken ni tsuite’ トルファン將來木版刷梵文雜阿含經斷簡: R. ピシェル 公表梵文斷簡について [Fragmentary woodblock manuscripts of the Sanskrit Saṃyuktāgama from Turfan: on Sanskrit fragments published by R. Pischel]. Indogaku Bukkyōgaku Kenkyū 37(2): 540–46.
Ishigami, Zennō 石上善應. 1966. ‘Sagātha-vagga no keishō no kōsei’ Sagātha-vagga の偈頌の構成 [Verse structure of the Sagātha-vagga]. Sankō Bunka Kenkyūjo Nempō 1: 185–263.
Lǚ, Chéng 呂澂. 192. ‘Zá āhán jīng kāndìng jì’ 雜阿含經刊定記’ [Revision of the chinese Saṃyuktāgama]. Nèi Xué 內學 1: 104–25. Reprinted in Fóguāng Dàzàngjīng, Āhán zàng, Fùlù [Fóguāng Tripiṭaka: Āgama: Appendix], 659–79 (1988).
Lǚ, Chéng 呂澂. 1992. Biāodiǎn běn Zá āhán jīng lùn 標點本雜阿含經論 [Punctuated edition of Saṃyuktāgama and its commentary]. Taipei: Xīnwénfēng.
Malalasekera, George Peiris. 1974. Dictionary of Pāli Proper Names. London: Pāli Text Society.
Mayeda, Egaku 前田惠學. 1964. genshi Bukkyō seiten no seiritsushi kenkyū 原始佛教聖典の成立史研究 [A history of the formation of early Buddhist texts]. Tokyo: Sankibō Busshorin.
Mayeda, Egaku. 1985. ‘Japanese Studies on the Schools of chinese Āgamas’. In Zur Schulzugehörigkeit von
Werken der hīnayāna-Literatur (Symposien zur Buddhismusforschung III), vol. I, ed. Heinz Bechert, 94–103. Göttingen: Vandenhoeck & Ruprecht.
Mukai, Akira 向井亮. 1985. ‘Yugashichiron Shashibun to Zō-agongyō’ 瑜伽師地論攝事分と雜阿含經 [The Vastusaṅgrahaṇī of the Yogācārabhūmi and the Saṃyuktāgama]. hokkaidō Daigaku Bungakubu Kiyō 32(2): 1–41.
Nanjio, Bunyiu. 1980. A Catalogue of the Buddhist Tripitaka, with Additions and Corrections by Lokesh Chandra. delhi: Lokesh chandra (updated reprint of 1883 ed.).
Norman, k. R. 1983. Pāli Literature, Including the Canonical Literature in Prakrit and Sanskrit of All the hīnayāna Schools of Buddhism (History of Indian Literature VII 2). Wiesbaden: Harrassowitz.
Pecenko, Primož. 2002. ‘Līnatthapakāsinī and Sāratthamañjūsā: The Purāṇaṭīkās and the Ṭīkās on the Four Nikāyas’. Journal of the Pāli Text Society 27: 61–113.
Rhys davids, T. W. and Stede, William, eds. 1925. Pāli–English Dictionary. London: Pāli Text Society. (Reprinted, New delhi: oriental Books Reprint corporation, 1975).
Saigusa, Mitsuyoshi 三枝充德. 1978. Shoki Bukkyō no shisō 初期佛教の思想 [[[Principal]] thoughts of early Buddhism]. Tokyo: Tōyō Tetsugaku Kenkyūjo.
Skilling, Peter. 2001. ‘Nuns, Laywomen, donors, Goddesses: Female Roles in Early Indian Buddhism’. Journal of the International Association of Buddhist Studies 24(2): 241–74.
Taishō Shinshū Daizōkyō 大正新修大藏經 [[[Taishō Tripiṭaka]] revised edition], 1924–4, eds Junjirō Takakusu and K. Watanabe. Tokyo: Taishō Issaikyō Kankokai. (Reprinted, Taipei: Xīnwénfēng, 1983).
Tibetan Tripiṭaka, Peking edition, 1955–58. otani reprint, ed. daisetz T. Suzuki. Tokyo and kyoto: Tibetan Tripiṭaka Research Institute.
Tseng, H. Vinita. 2001. ‘The Nidānavagga of Sāratthappakāsinī’. d.Phil. thesis, University of oxford.
von Hinüber, oskar. 1994. Untersuchungen zur Mündlichkeit früher mittelindischer Texte der Buddhisten. Stuttgart: Franz Steiner.
von Hinüber, oskar. 1996. A handbook of Pāli Literature. Berlin and New York: de Gruyter.
Waldschmidt, Ernst. 1932. Bruchstücke buddhistischer Sūtras aus dem zentralasiatischen Sanskritkanon (kleinere Sanskrit-texte, IV). Leipzig: Brockhaus. (Reprinted, Wiesbaden: Franz Steiner, 1979.) Waldschmidt, Ernst. 1980. ‘central Asian Sūtra Fragments and Their Relation to the chinese Āgamas’. In Die Sprache der ältesten buddhistischen Überlieferung (The language of the earliest Buddhist tradition) (Symposien zur Buddhismusforschung II), ed. Heinz Bechert, 155–6. Göttingen: Vandenhoeck & Ruprecht.
Yìnshùn 印順. 1983. Zá āhán jīng lùn huìbiān 雜阿含經論會編 [[[Saṃyuktāgama]] sūtra and commentary], vols. Taipei: Zhèngwén.
Yìnshùn 印順. 1988. Yuánshǐ Fójiào shèngdiǎn zhī jíchéng 原始佛教聖殿之集成 [The compilation of the scriptures of Early Buddhism], 2nd edn. Taipei: Zhèngwén.
