Introduction
In many respects, the Chinese Ekottarikāgama 增壹阿含經 T125 has defied the attempts of modern scholarship to understand it. As I will discuss in greater detail below, in catalogues and successive iterations of the canon, the ascription of the translatorship of the collection has oscillated between Saṅghadeva 僧伽提婆 (fl. 383–398) and Zhu Fonian 竺佛念 (fl. 382–413). Modern scholarly opinion has been divided on the same question. External evidence, in documents like prefaces, biographies, early catalogues, and early excerpts or citations, suggests that the collection had a complex early history, and multiple versions may have existed. The nature of the Indic Vorlage(n) is unclear, and unlike the other Chinese Āgama translations, modern scholarship has come to no consensus about the sectarian (nikāya) affiliation of the collection. Numerous features of the content of the collection strike us as anomalous, by comparison with other Mainstream Āgamas.
In this study, I will principally aim to show that anomalous features of various types are unevenly distributed within T125, clustering in particular discourses and sections. I will argue that this distribution indicates that T125 is heterogeneous in nature. I will argue further that it is likely that some discourses or sections were modified substantively, probably in content as well as form and wording, sometime after the death of Dao'an 道安 in 385.
A second arm of my argument is more tentative and speculative. Various indications suggest that at least some of this probably later layer in the collection is due to Zhu Fonian. However, my evidence in support of this possibility is more circumstantial, and I present this second layer of my argument in the mode of a hypothesis, rather than a firm conclusion.
The study is divided into two main parts. First, I review anomalous features of T125 that have already been discussed in prior scholarship. Here, my contribution is to map the distribution of these features, to show that anomalies cluster. Second, I present new, internal (stylistic) evidence, discovered with computational tools,Footnote 1 which also suggests anomalies of different types, and show that this evidence, too, clusters. The paper is accompanied by a series of Appendices, presenting in tables the evidence supporting my arguments.
The many riddles of T125
Prior scholarship has already touched upon several respects in which the nature and textual history of T125 is complex and murky.Footnote 2 A first set of problems is adumbrated by a basic fact. As several scholars have noted, an unusually low proportion of T125 discourses is paralleled in the sister collection in Pāli, the Aṅguttara-nikāya, or even in other Āgamas and Vinayas more generally.Footnote 2 This raises tricky questions. Where did all the unparalleled material come from? Is it plausible to imagine an Indic Āgama collection with so little matching to other Āgamas and Nikāyas?
A different type of problem is presented by three discourses: EĀ 48.3 is duplicated more or less verbatim in T453, which is ascribed to *Dharmarakṣa 竺法護 (fl. ca. 266–308); EĀ 30.3 is likewise identical to T128b, ascribed to Zhi Qian 支謙 (ca. 193–252);Footnote 2 and EĀ 32.5 is virtually identical to T61, ascribed to Dharmarakṣa.Footnote 5 Uncertainty about the ascription of these materials has sometimes prompted the suggestion that T125 may have absorbed originally extrinsic material over time.Footnote 6
T125 has also proven to be a dark horse in terms of sectarian affiliation. Various theories on the sectarian affiliation of the collection have been proposed: usually Mahāsāṅghika (Akanuma, Mizuno, Takasaki, Waldschmidt,Footnote 7 Bareau,Footnote 8 Schmithausen,Footnote 9 Pāsādika,Footnote 10 KuanFootnote 11); less often Dharmaguptaka (Matsumoto Bunzaburō, Hirakawa).Footnote 12 Hiraoka has also pointed to a number of elements that show connections to Sarvāstivāda narrative traditions – in some cases, quite strong connections.Footnote 13 These disagreements suggest that no clear determination is possible about the sectarian affiliation of T125 as a whole, and this is indeed the conclusion drawn by scholars such as Mizuno, Shizutani, Enomoto, Anālayo, and Hiraoka.Footnote 14 I will return to this question later, when discussing Hiraoka's evidence of possible Sarvāstivāda connections. To anticipate, however, the very assumption that a single, unitary affiliation exists to be discovered may be misleading and unhelpful.
As scholars have noted time and again, Dao'an writes in his Preface that the collection has 472 sutras – exactly matching the count in our extant text (if we include EĀ 1) – but only forty-one fascicles, against the fifty-one fascicles of the extant T125.Footnote 15 This already suggests that material may have been added to the collection after it left Dao'an's handsFootnote 16 (though fascicle division can also be notoriously unstable in the course of transmission history).Footnote 17
Nor is this the only potential anomaly here. Dao'an writes, further, that the reciter, Dharmanandi(n)(?) 曇摩難提Footnote 18 (fl. 383–391), could remember the uddānasFootnote 19 for the first twenty-six fascicles (of the forty-one he reports), but for the remaining fifteen fascicles, the uddānas were “omitted” 失 (i.e. had been forgotten by the reciter?).Footnote 20 But as Su has shown, the situation with the uddānas in our extant T125 is more complex, and does not entirely match with Dao'an's report. In fact, some chapters in the first part of the extant text are missing uddānas, and some chapters in the last part of the text have them.Footnote 21 In addition, Su also studies cases in which the phraseology of the uddānas sometimes does not match the phraseology expressing the same meanings in the corresponding sutras; the sequence of sutras differs from the sequence of their keywords in the uddānas; a discourse indicated in an uddāna is missing from the collection; a discourse in the collection is not keyed in the uddānas; and the present uddānas point to discourses in two or more chapters.Footnote 22 This reinforces the impression that the collection might have been revised, edited, reshuffled, or otherwise modified after its initial translation by Dao'an's group.
Another possible structural anomaly is suggested by a report in the Fenbie gongde lun 分別功德論 T1507 (a commentary on EĀ, on which more below). This commentary reports a tradition that EĀ originally had sections for every number up to a hundred (i.e. a section each for Twelves, Thirteens … through to Ninety-Nines, Hundreds), but that most of this mass of text was forgotten through imbalances in the practice of the Saṅgha, until only ten sections were left. The extant T125, however, has eleven vargas, not ten.Footnote 23 Palumbo has related this tradition to apparent wrinkles in the Zhuanji sanzang ji zazang zhuan 撰集三藏及雜藏傳 T2026. This text presents a narrative of the compilation of the canon that gives pride of place to EĀ, and Palumbo has suggested that it was originally attached to a version of EĀ as a kind of preface or appendix.Footnote 24 Palumbo suggests further that the elements in T2026 relating to the Elevens may have been interpolated into an original text that only knew ten vargas.Footnote 25
The possibly anomalous status of the Elevens probably relates to further anomalies in the final fascicles of T125, which were pointed out by Mizuno.Footnote 26 According to T2026, the “Elevens” section is supposed to finish with a sutra on compassion 行慈,Footnote 27 which Mizuno identifies with EĀ 49.10. He points out further that of the twenty-nine sutras in Chs. 50–52, only three teach sets of eleven dharmas, as we should expect from their position; the others teach other sets, or have content that does not relate to numbered sets at all.Footnote 28 On this basis, Mizuno proposes that these chapters are a latter addition to the collection.
Mizuno also drew attention to cases in which citations in the Jing lü yi xiang 經律異相 T2121 (JLYX) do not match the extant T125, and the same problem was subsequently treated by scholars like Lin, Su, and Palumbo.Footnote 29 Most significantly, two discourses quoted from an EĀ cannot be found in the extant T125;Footnote 30 according to Palumbo's analysis, two other discourses match parallel independent sutras (speculatively identified by Mizuno as vestiges of an alternate translation of EĀ) better than T125.Footnote 31 Other discourses differ to a degree that “cannot be explained as the result of imprecise quotation, abridgement or periphrasis, but must ensue from a different underlying text.”Footnote 32 Su gives sensible cautions about the extent to which citations in JLYX might not be verbatim, but still concludes, “The quotations in [JLYX] … seem to have been extracted from a different Ekottarikāgama.”Footnote 33 In a similar vein, scholars have also noted that T1507 – clearly a commentary composed in Chinese, to the Chinese translation of EĀ – in places cites EĀ with wording not exactly matched in the extant T125.Footnote 34 Several complex questions are raised by these observations, but they need not detain us here. For present purposes, the important point is that they may show that multiple versions of “the Ekottarikāgama” were once in circulation – and raise the question of how the extant T125 is related to other, partially lost or hypothetical versions, and where it sits relative to them in textual history.
The JLYX citations are also central to the arguments of Lin Jia'an.Footnote 35 Among prior studies, Lin comes closest to anticipating the theses of the present study. Lin argues that the extant T125 was produced by Zhu Fonian, working alone, around 410, as a revision of the original Dao'an/Chang'an translation. Lin argues that the JLYX citations stem from the old Dao'an/Chang'an translation, and that the size of the collection reported in old primary sources, expressed in number of fascicles, indicates that the present T125 is larger than the original translation. However, as I will discuss later in this paper (n. 211), despite the superficial similarity of Lin's arguments to my own, her evidence and reasoning is quite different (and in some respects, obscure or questionable).
The question of possible alternate versions of the collection is complicated still further by conflicted reports in the bibliographic tradition about who translated T125. Dao'an's preface to EĀ is unambiguous (but see again discussion after Palumbo below of finer complications introduced by other documents): the text was recited by Dharmanandin from Tokhara 兜佉勒;Footnote 36 Zhu Fonian was the translator 譯傳, and Tansong 曇嵩 (d.u.) was the amanuensis; the work was underway from summer 384 until spring 385. The translation was then revised by Dao'an and Fahe 法和 (d.u., c. early fourth to early fifth century); and Senglüe 僧略 (344–416) and Sengmao 僧茂 (d.u.) assisted in proof-reading.Footnote 37 However, Daoci 道慈 (fl. 391–401), writing in the early fifth century, reported that Saṅghadeva and Fahe had revised the EĀ of the Dao'an group.Footnote 38 Over time, this led to the emergence of claims that Saṅghadeva had also “translated” a version of the collection, and eventually, to the adoption of the mistaken ascription to Saṅghadeva as canonical (as it remains today).Footnote 39
For present purposes, it is important to note that all this confusion about the ascription of T125 led to an influential false hypothesis about a supposedly lost or fragmented, “alternate” translation of EĀ by Zhu Fonian. Building upon these reports of two different translations, Mizuno championed a hypothesis that the “alternate” translation had not been entirely lost, but survived in scraps scattered through the canon as individual texts.Footnote 40 Further, and more problematically, because Mizuno adhered to the canonical ascription of T125 to Saṅghadeva, he supposed that this alternate EĀ was by Zhu Fonian and his collaborators. He also held that the same translators produced the group of texts he regarded as surviving vestiges of an alternate Madhyamāgama.Footnote 41 It would go beyond the confines of my study here to consider the group of texts Mizuno points to fully (if indeed they constitute a coherent group at all). For our purposes, we need not broach these issues, since the Zhu Fonian translation of EĀ has been found – it is the canonical T125, which was under our noses all the time.Footnote 42 Thus, the mistaken tradition that Saṅghadeva produced a “translation” of EĀ (rather than a mere revision), and the misascription of the extant T125 to Saṅghadeva, conjured up a ghost in the scholarly imagination haunting Zhu Fonian: the idea that his EĀ only partly survived, in the scattered shards identified by Mizuno. We risk confusion unless we firmly exorcise this ghost.
A special problem is presented by EĀ 50.4. In content, this discourse doubles up with a narrative also presented, with significant differences, in EĀ 1 (the story of Ma[k]hādeva). Anālayo has argued on the basis of translation terminology that EĀ 50.4 is by a different translator than the remainder of T125 (i.e. not by Zhu Fonian).Footnote 43 He therefore suggests that it was added wholesale to T125 at some point in its transmission history in China.Footnote 44 A follow-up study by Hung, using computational, statistical, quantitative methods, found strong evidence from a number of perspectives supporting Anālayo's hypothesis that EĀ 50.4 is not by the same translator as the remainder of T125.Footnote 45
In a similar vein, Palumbo has argued that one of the building blocks in the construction of EĀ 48.2 was Chinese – an otherwise unknown translation of a prātimokṣa, possibly Sarvāstivādin,Footnote 46 represented by the Dunhuang manuscript Stein 797.Footnote 47 Palumbo is cautious about the conclusions we can draw from the parallel he points out, since he notes that it “does not necessarily imply a forgery” (viz. in EĀ 48.2) – the translator could have recognised a block of text in his Indic Vorlage, and used an existing translation known to him. But an alternate possibility is certainly that the Chinese represented by Stein 797 was in fact the only source for these materials in EĀ 48.2.Footnote 48
In another study, Anālayo has plausibly suggested that a passage has been corrupted in a manner that is best explicable if we posit that the change occurred in writing, and in Chinese.Footnote 49 He has also analysed cases where sudden shifts in protagonist, and other “continuity” problems, seem to betray the splicing together of originally separate materials.Footnote 50 Anālayo has also discussed an instance of a confusion between Jain and Buddhist doctrine, which he proposes would be unlikely to arise “in an Indian setting familiar with the contrast between the tenets” of the two groups.Footnote 51 In sum, various clues point to the likelihood that parts of the extant T125 were modified in China, after the point of translation.
Thus, to sum up so far: remarkably few sutras in T125 are paralleled in other versions or languages. T125 has also been exceptionally resistant to efforts to identify the sectarian affiliation of its Indic Vorlage. These factors already make the collection an outlier. Nor can we be confident that we have received it as it was originally translated. The received T125 appears to have considerably more fascicles than the EĀ documented by Dao'an's preface. Dao'an himself reports that the recitation of the Vorlage was attended by failures of memory for the uddānas, but the present state of T125 does not fit with Dao'an's description of the resulting situation. The Elevens section, in particular, may have been a late interloper even in the Indic Vorlage;Footnote 52 contains a large number of texts with no apparent relation to the numbering principle structuring the collection; and according to T2026, should not include Chs. 50–52. Later citations in JLYX, and a commentary on the Chinese translation, T1507, show varying degrees of mismatch with the extant T125. The bibliographic tradition eventually reported two separate Chinese “translations” of EĀ, and scholars down to the modern era have disputed which of these putative Doppelgänger survived. Indeed, a number of individual sutras still extant, scattered through the canon, may be leftover scraps of an alternate version. Further, at least one discourse (namely, EĀ 50.4) appears to have been translated by different hands than the remainder of T125, and the closest studies to date have suggested it may be a survival of such an alternate translation. One sutra overlaps verbatim with an early Sarvāstivāda prātimokṣa in Chinese, and Anālayo has suggested that another sutra contains an error that probably took place in writing, and in Chinese. These structural problems furnish ample reasons to suspect that T125 harbours anomalies in content.
Still further complications have been brought much more clearly into view by Palumbo's brilliant, intricate and sensitive analysis of a wide range of evidence. Consequently, we can see that the textual history of the Chinese EĀ was even more complex than previously thought. For our purposes, Palumbo's most important and convincing findings are that multiple versions of EĀ were produced by the Dao'an group;Footnote 53 and that EĀ then experienced a turbulent transmission history down to the sixth century, probably circulating in multiple versions.
Palumbo has plausibly argued that in fact, no less than four distinct recensions of EĀ were most probably produced in rather quick succession at the end of the fourth century, three among the Dao'an translation group in the course of little more than a year, and one later revision:
• A first recension by the Dao'an group in 384, in 46 juan.Footnote 54 Palumbo identifies this recension with Mizuno's above-mentioned “alternate” translation of EĀ.Footnote 55
• A second recension, also by the Dao'an group, later in 384 (completed by February 385). Dao'an states that this version of the text was forty-one juan long, divided into two parts of twenty-six and fifteen juan, with and without uddānas respectively. It comprised 472 sutras.Footnote 56
• A third recension, produced in a period of forty days immediately after the second by the revision work of Dao'an and Fahe, with the assistance of Senglüe and Sengmao. The resulting text still comprised 472 sutras, but to the original forty-one juan, the revisers added an extra fascicle of summaries (now lost). This version should have been complete by March/April 385.Footnote 57
• A fourth recension, resulting from the revisions of Saṅghadeva and Fahe, which Palumbo argues was probably produced in Luoyang between 390 and early 391.Footnote 58
As already mentioned, copious internal evidence shows that the extant T125 is far more closely associated with Zhu Fonian than with Saṅghadeva (as represented by MĀ T26).Footnote 59 For our present purposes, we can therefore set aside Palumbo's fourth recension, i.e. the Saṅghadeva version, as of limited relevance for our investigation into T125. However, Palumbo's theories about the other three recensions, and their afterlives, must be addressed, since they bear closely on the question of various apparent anomalies in the extant T125 and their origins.Footnote 60
In his preface to EĀ, Dao'an reports that in the course of recitation for the translation, some of the uddānas “went missing” 失 (seemingly meaning they were forgotten by the reciter). Palumbo's theories about the origins of the extant T125, including some of the anomalous elements we are discussing here, rest upon a speculative interpretation of this line. Palumbo opines that it is “none too credible” that Dharmanandin could have forgotten “the brief uddānas” for “well over one third of the entire collection,” and yet not have forgotten “the much longer sutras.”Footnote 61 On the strength of this doubt, he speculates that Dao'an “is glossing over a far more embarrassing situation”: that “Dharmananda may in fact have been unable to recite at least part of the sutras in the relevant vargas” as well.Footnote 62 “Other members of the group – Zhu Fonian, Dao'an, the other foreign masters” are then supposed to have “step[ped] in on occasion to supply the missing portions. Versions of individual sutras that were known within the group might even have been chosen to replace those that Dharmananda had initially recited.”Footnote 63 Palumbo thus speculates that the “large, ‘composite sutras’ [i.e. merger discourses – MR] that stand out in the received [T125]” were added between his hypothetical first and second or third recensions, as part of a strategy to redress the fallout from Dharmanandin's memory lapses.Footnote 64 He also conjectures further that the collection was reshuffled, still in the context of the Chang'an group, in order to sort the sutras into those for which uddānas were remembered, and those that were forgotten – in the process “decisively subvert[ing] the numerical progression of the series,” that is to say, the very eponymous principle upon which an Ekottarikāgama is founded.Footnote 65
This intricate construct is not convincing. It piles speculation on speculation – guessing that the true extent of Dharmanandin's amnesia was greater than reported, and then imagining a particular solution to this problem. It also has the considerable disadvantage that it requires us to imagine that Dao'an and several “foreign masters,” including Dharmanandin, would have intervened in buddhavacana to this extent.Footnote 66 To my mind, the anomalies in the extant T125 cannot be accounted for by Palumbo's hypotheses, and we are still in search of explanations.
As a counterpart to this theory about the creative interventions of Dao'an's Chang'an group in the text, Palumbo is also disinclined to believe that the content of EĀ was changed in any substantive manner after the production of the third recension in early 385 – which would mean that apart from sequence and arrangement,Footnote 67 the extant T125 represents that third recension.Footnote 68 A major plank of Palumbo's argument in this regard is the fact that T1507, which comments on EĀ, reflects a text very close to the extant T125. However, T1507 only comments on the first three and a half chapters of EĀ (less than three fascicles in T125, or about 5% of the whole). It cannot testify to anything in the extant T125 after Ch. 4.Footnote 69 Its evidence is thus not relevant for other portions of the extant T125 – including, especially, the portions in later parts of the text, which we see below are particularly problematic.
Moreover, for T1507 to serve as evidence that T125 had reached more or less its present contents by Dao'an's death, T1507 itself would also have to have assumed its present form by the same time. But to my mind, Palumbo does not clearly establish that T1507 could not have been modified after Dao'an's death.Footnote 70 The possibility therefore still remains open that T125 underwent significant changes – in content as well as organisation – after the death of Dao'an.
Palumbo also presents a range of evidence suggesting that “the” EĀ – possibly at times circulating in at least two different versions – had a somewhat volatile history for a time after its first production. This evidence mitigates against his theory that T125 was more or less closed in 385. Fascicle numbers given for words in EĀ glossed by the Fan Fanyu 翻梵語 T2130 suggest that the compiler was referring to a collection with a different order and smaller format.Footnote 71 As already mentioned, Palumbo also considers at length the above-mentioned problem of citations in JLYX, which are also incompatible with the received T125, as studied before him by Mizuno and Lin Jia'an; these citations also seem to reflect a collection with a different structure and order than T125.Footnote 72 Palumbo further points out that in CSZJJ, Sengyou lists a version of EĀ in thirty-three juan;Footnote 73 a long discussion in the Da Zhou kanding zhongjing mulu 大周刊定眾經目錄, which is presented (Palumbo believes plausibly) as a citation from Baochang's 寶唱 (fl. ca. 502–516) catalogue (roughly contemporaneous with Sengyou), confirms the one-time existence of this thirty-three fascicle version, and further specifies that it was produced on an imperial commission 明定.Footnote 74 The same Da Zhou kanding zhongjing mulu entry, having discussed two versions of EĀ, ascribed to Saṅghadeva and to Dharmanandin and Zhu Fonian respectively, further states explicitly that both are extant; as Palumbo points out, this is valuable evidence that two different versions were known at the same time at some point in history.Footnote 75 Against this backdrop of these numerous reports about various orderings and fascicle divisions of the collection, Palumbo's summary of the treatment of the collection, as cited in later works and listed in the catalogues, shows that the first evidence of a collection with fifty-one fascicles (or fifty), as the received T125, dates only to the Sui.Footnote 76
As already mentioned above, Palumbo himself aims to restrict the implications of this evidence to the structure and order of the collection, and contends that the content of T125 was more or less fixed by the time Dao'an died in 385. However, the chaotic state of EĀ traditions in the fifth to sixth centuries, which we glimpse here, also leaves ample room for the extant T125 to incorporate products of interventions into content also, effected later than the death of Dao'an.
In sum, Palumbo has shown that even at its birth in Chang'an in 384–385, the Chinese EĀ already had a singularly chequered beginning, and so far as we can discern through the haze of our evidence, at least down to the late sixth century, volatility prevailed. However, even Palumbo's imaginative and magisterial orchestration of a rich body of evidence, much of it external to the text, does not establish that all the anomalies in the received T125 were the products of Dao'an's group, and in place by 385. Indeed, the various puzzles besetting the text, as surveyed above, are still only a part of a broader picture, in which anomalies multiply all the more. To that picture we now turn.
Nine classes of possible anomaly in T125
When we read T125, it can seem that anomalous features of various types catch the eye at nearly every turn. In the following, I will give an overview of nine different classes of evidence that indicate possible anomalies. I will first synthesise results from prior research.Footnote 77 To date, treatments of five different types of evidence have been scattered over a fairly wide number of disparate, more focused studies.Footnote 78 Few attempts have been made to give an overview,Footnote 79 and no scholar has tried systematically to coordinate the evidence with individual T125 discourses. I will present such a survey. This exercise is already illuminating, since it reveals that the evidence clusters in particular parts of the collection. Following that, I will introduce four classes of internal (phraseological-stylistic) evidence, which likewise serve as possible indicators of anomalies. Finally, I will consider all these types of evidence in concert, and ask whether patterns in their distribution give us clues about the nature of some individual discourses, and the provenance, nature, and shape of the extant collection as a whole.
Mahāyāna features
Previous scholarship has noted that a number of T125 discourses include features usually associated, in varying degrees, with the Mahāyāna.Footnote 80
For the purposes of this analysis, I will treat the following elements as possibly (proto-)Mahāyāna (note that some discourses contain more than one of these elements, and therefore appear more than once in this list):Footnote 81
• “vehicles” (yāna, 乘) discourse,Footnote 82 including references to the “Mahāyāna” 大乘 in so many words (1, once referring to 方等大乘; 19.8),Footnote 83 to “hīnayāna” (26.9),Footnote 84 three vehicles (*triyāna 三乘) (1, 3.3, 12.6, 24.6, 26.9, 32.1, 43.2, 45.5, 48.3, 48.5),Footnote 85 the itemised set of *śrāvaka(yāna),Footnote 86 *pratyekabuddha(-yāna),Footnote 87 and *buddha(-yāna) (聲聞, 緣覺/辟支, 佛乘, etc.) (1, 23.1,Footnote 88 24.6,Footnote 89 28.5, possibly 32.5,Footnote 90 43.2, 45.5,Footnote 91 47.3, 48.5);
• Maitreya as a bodhisattva (20.6, 27.5, 42.6, 48.3);Footnote 92
• (possibly) the “concentration like adamant” (vajropamasamādhi), associated with magical powers, and appearing as part of longer lists of samādhi names (4.9, 42.4, 48.6);Footnote 93
• the bodhisattva path as a viable or desirable goal for adherents other than ŚākyamuniFootnote 94 (1, 10.5, 20.3, 20.7, 20.10, 24.5, 27.5, 35.2, 36.5, 38.7Footnote 95), including the ideal of achieving buddhahood and the thirty-two marks (10.3Footnote 96), or the ambition of “becoming a Buddha like Śākyamuni” (1Footnote 97);
• the Buddha preaching bodhisattva practices 菩薩行 to othersFootnote 98 (20.3, 48.5, 52.6Footnote 99);
• (possibly) the notion of (a “ground” that is) *avaivartika 不退轉(地)Footnote 100 (26.9,Footnote 101 42.8,Footnote 102 43.2Footnote 103);
• the notion of *bodhicitta (1, 10.5, 35.2, 52.6);Footnote 104
• mention of the six perfections in application to persons other than Śākyamuni or past Buddhas (1, 27.5);Footnote 105
• the concept of *mahākaruṇā, especially if it is recommended as a practice for persons other than the Buddha(s) (38.1,Footnote 106 49.4);
• the concept of ekajātipratibaddha 一生補處 (20.10, 42.3);Footnote 107
• the notion of “Buddha lands” 佛土,Footnote 108 including simultaneous remote lands in the present, and the idea of teleporting between them (26.9, 37.2,Footnote 109 38.7) – to which we should add other notions connected with an expanded cosmology,Footnote 110 such as the *trisāhasramahāsāhasralokadhātu Footnote 111 (三千大千世界, 三千大千剎土) (18.5, 24.3, 24.5, 36.5, 37.2, 37.3, 41.1, 42.3, 43.2, 48.3), or “other buddhakṣetras” 他方剎土 (41.1);
• the names of other little-known Buddhas:Footnote 112 奇光如來 (37.2), 寶藏如來 (43.2), 師子應如來, 柔順佛, 光焰佛, 無垢佛 (48.4);
• elements in EĀ 1 redolent of the “cult of the text”;Footnote 113
• the term bodhisattva-mahāsattva 菩薩摩訶薩 (20.7, 27.5);Footnote 114
• *anuttarasamyaksaṃbodhi (無上正真之道, 無上正真等正覺) as a generalised goal (rather than a special property of Śākyamuni and past and future Buddhas) (20.7,Footnote 115 26.5, 39.3, possibly 49.2Footnote 116);
• the “four fundamentals of the Dharma” 四法本(末), 四事之教, 四法之本 (26.8, 26.9, 31.4, 42.3);Footnote 117
• the presence of multiple bodhisattvas in the same scene (1, 10.8, 38.7, 48.3Footnote 118);
• the concept of vajrakāya 金剛之身 (3.1, 26.9, 42.3),Footnote 119 and the idea that the Buddha in his dharmakāya has an extremely long lifespan, that the dharmakāya endures when the fleshly body passes away, etc. (1, 48.2Footnote 120);Footnote 121
• descriptions of the Buddha issuing prophecies (*vyākaraṇa, 莂/別, 記) explicitly so named, specifically for buddhahood (24.5, 35.2, 36.5);Footnote 122
• Subhūti (of Prajñāpāramitā fame) expressing a doctrine redolent of “emptiness” (śūnyatā) (36.5).Footnote 123
The following forty-five discourses are presently known to feature one or more of the above-listed Mahāyāna elements:Footnote 124
EĀ 1, 3.1, 3.3, 4.9, 10.3, 10.5, 10.8, 12.6, 18.5, 19.8, 20.3, 20.6, 20.7, 20.10, 23.1, 24.3, 24.5, 24.6, 26.5, 26.9, 27.5, 28.5, 32.5, 35.2, 36.5, 37.2, 37.3, 38.1, 38.7, 39.3, 41.1, 42.3, 42.4, 42.6, 42.8, 43.2, 45.5, 47.3, 48.2, 48.3, 48.4, 48.5, 48.6, 49.4, 52.6.
Of course, it is not easy to judge the exact degree in which some of these elements really indicate “Mahāyāna,” nor what we should make of them even if they are such.
However, Nattier plausibly suggests that Zhu Fonian's “track record,” as a compiler of original sutras with Mahāyāna colouring, makes him a plausible candidate to have introduced at least some of the peculiarities we observe in the text – which naturally includes such Mahāyāna colouring.Footnote 125 On the other hand, Anālayo has also pointed out that the introduction of Mahāyāna elements to the collection might have equally naturally occurred in Central Asia, i.e. the source culture of the reciter, Dharmanandin.Footnote 126 Palumbo points to a number of ways that the flirtation with apparently Mahāyāna tropes and notions might be a more general feature of other texts associated with EĀ, including other apparently Sarvāstivāda texts, and other texts produced by the Dao'an group.Footnote 127 We must also consider the possibility that Mahāyāna elements, and/or discourses containing them, could have been added to our extant T125 by someone other than Zhu Fonian, at other points in the history of the text.
“Merged” discourses
Following the lead of a study of EĀ 48.6 by Lamotte,Footnote 128 Anālayo has also studied a number of instances in T125 of a phenomenon he calls “discourse merger.” Material paralleled in two or more different texts in other collections or branches of the tradition – often, material elsewhere found in different collections or genres of text – has here been combined to form a single discourse. In some cases, Anālayo's analysis shows that the merging of sources has been imperfectly achieved, and scars remain from the surgery involved, such as abrupt changes of protagonist, or choppy progression in the narrative. Anālayo also shows cases in which merged discourses are paralleled in suttas immediately succeeding one another in MN; and a case in which two different merged T125 discourses combine material paralleled in the same pair of separate MN/MĀ suttas.
At least the nineteen discourses mentioned in Table 1 are probable products of such mergers.Footnote 129
Table 1. ‘Merged’ discourses in EĀ.
1 Mizuno (Reference Mizuno1989), pp. 20–21; Anālayo (Reference Anālayo2014–2015), pp. 64–65.
2 I have added this discourse merger to the list I derived from Anālayo.
2Anālayo (Reference Anālayo2014–2015), pp. 65–66.
3 Anālayo (Reference Anālayo2014–2015), pp. 78–79.
4 Anālayo (Reference Anālayo2014–2015), pp. 79–80.
5 Anālayo (Reference Anālayo2014–2015), p. 68.
6 Anālayo (Reference Anālayo2014–2015), 66; and Anālayo (Reference Anālayo2015a), with further references.
7 Anālayo (Reference Anālayo2014–2015), pp. 75–78. Anālayo notes that different portions of the same MN/MĀ parallel sutras are merged in EĀ 50.8; 75.
8 Anālayo (Reference Anālayo2014–2015), p. 66.
9 Anālayo (Reference Anālayo2014–2015), p. 85 n. 6.
10 Anālayo (Reference Anālayo2014–2015), p. 85. n. 5.
11 Anālayo (Reference Anālayo2014–2015), pp. 66–67.
12 Mizuno (Reference Mizuno1989), pp. 21–23; Anālayo (Reference Anālayo2014–2015), p. 68.
13 Lamotte (Reference Lamotte1967); Anālayo (Reference Anālayo2014–2015), pp. 63–64.
14 Anālayo (Reference Anālayo2014–2015), p. 65; Reference Babain buddhakṣetrapariśodhana: a festschrift for paul harrison, edited by c. disimone and n. witkowski, 29–40. marburg: indica et tibetica verlag..
15 Like 50.8 and 49.7, 49.6 merges material from two sutras that appear in immediate succession in MN; Anālayo (Reference Anālayo2014–2015), p. 78.
16 Like 50.8 and 49.6, 49.7 merges material from two sutras that appear in immediate succession in MN; Anālayo (Reference Anālayo2014–2015), p. 78.
17 Anālayo (Reference Anālayo2014a 1); Anālayo (Reference Anālayo2014–2015), pp. 68–74. Anālayo notes that a final passage in the text is not matched in either of the MN/MĀ parallels, and suggests it is likely to be from a Vinaya source; 73 w. n. 74. He also notes that different portions of the same MN/MĀ parallel sutras are merged in EĀ 43.5; 75. Like 49.6 and 49.7, 50.8 also merges material from two sutras that appear in immediate succession in MN; 78.
18 Anālayo (Reference Anālayo2014–2015), p. 67.
Strikingly, the vast majority of known instances of such “merged” discourses are concentrated in the last ten chapters of T125.
Citing some of the evidence discussed above, showing that parts of T125 may betray modification in China and in Chinese, Anālayo argues that two cases of discourse merger must have occurred in China, and several other cases are most likely to have taken place in a written medium (and therefore, most likely in China, given that the originals reached China by oral transmission).Footnote 130
Discourses with late or post-canonical Pāli parallels
T125 also contains a number of discourses including material that in the Pāli tradition only appears in post-canonical or commentarial sources (Table 2).Footnote 131
Table 2. EĀ discourses with late/post-canonical Pāli parallels.
1 Anālayo (Reference Anālayo2014b), pp. 119–20 w. n. 33.
2 Kuan (Reference Kuan2012), p. 190.
2Kuan (Reference Kuan2012), p. 190.
3 Anālayo (Reference Anālayo and Meisig2010), p. 7.
4 Cf. Bareau (Reference Bareau, Lienhard and Piovano1997), p. 25; Anālayo (Reference Anālayo2016), pp. 306–7 n. 20.
5 Matsuda (Reference Matsuda2001), pp. 2–3.
6 Anālayo (Reference Anālayo and Meisig2010), pp. 7.
7 Anālayo (Reference Anālayo and Meisig2010), pp. 7; Anālayo (Reference Anālayo2015a).
8 Kuan (Reference Kuan2012), p. 195.
9 Hiraoka (Reference Hiraoka2013), pp. 83–84 w. n. 24.
10 Kuan (Reference Kuan2012), p. 195, Anālayo (Reference Anālayo2014a), p. 17 n. 25 and p. 20 n. 27.
12 Mizuno (Reference Mizuno1989); Anālayo (Reference Anālayo2015b).
13 Anālayo (Reference Anālayo2014b), p. 120 w. n. 34.
14 Kuan (Reference Kuan2012), p. 186, Pāsādika (Reference Pāsādika, Klaus and Hartmann2007).
Material of this type has usually been taken as a sign that the Indic Vorlage of T125 remained open relatively late, and some of its individual sutras may therefore also be late.
A principal aim of the present study is to assess the likelihood that changes were made to T125 after the death of Dao'an. In this light, a methodological caution is called for. We have no simple way of judging when or where elements like those in Table 2 might have been added to the collection, and must keep clearly in view the possibility that they entered the collection during the development of the Indic Vorlage.Footnote 132 We can appreciate the need for such caution by reference to work by Baba, who has studied a series of doctrinal elements found in Northern Āgamas, including T125, which only appear in post-canonical works in Pāli (e.g. paracanonical works, commentaries):Footnote 133 for example, a number of instances of taking refuge in the Buddha alone with the transcribed formula *namo buddhāya 南無佛. This element appears in EĀ 48.3 and 49.9 (twice). However, Baba shows that *namo buddhāya also occurs in SĀ, a number of Chinese Vinayas (including Zhu Fonian's T1428), and the Divyāvadāna. Thus, this element is no monopoly of T125 alone.Footnote 134 Other examples of such elements include the *Nagara-sūtra explanation of the pratītyasamutpāda, or mention of “emptiness in the ultimate sense” (*paramārthaśūnyatā), or the addition of “emptiness” (śūnyatā) to the older set of anitya, duḥkha, and anātman (the “three characteristics” of all dharmas). However, Baba's examples again are drawn from several Āgamas, not only T125. Baba also points out that such elements in the Chinese Āgama translations are matched in multiple Sanskrit manuscripts, and so cannot be an addition of the translators (or compilers) of the Chinese texts.Footnote 135 For our present purposes, the main lesson of Baba's examples is that we cannot conclude that everything in T125 that looks unusual, against the benchmark of the Pāli tradition in particular, necessarily entered the text at the point of translation into Chinese, or subsequent further development of the collection.
Discourses without parallels (“sole exemplars”), and with
Another key item of information that we must track for each discourse is whether or not the discourse is paralleled in other texts, collections, or languages.
In the first instance, the main significance of such evidence (especially where the parallels are in languages other than Chinese, and no earlier Chinese parallels exist) is that parallels indicate that a discourse (or the paralleled part of a discourse) cannot have been invented from whole cloth in China. In addition, if we suspect that a discourse might not originate from the Indic Vorlage of the first translation of the collection as a whole, the existence of parallels (unless they can be shown to be derived from T125 itself) forces us to ask where the compiler of the discourse in question could have got their information about the content of the discourse.Footnote 136
Conversely, when a discourse is unparalleled (a “sole exemplar”), the chance that it is an addition or independent invention is somewhat greater, though the lack of a parallel is no guarantee that a sutra is anomalous, and we also cannot be sure when or where it might have been added to the collection.
In Appendix 5, I have therefore listed parallels where I am aware of them.Footnote 137 However, we should note that various closer studies of individual discourses regularly reveal parallels, sometimes quite numerous, not included in such listings. We should also note that supposed “parallels” listed in various sources correspond in varying degrees to their T125 counterparts, and sometimes only match a part of a longer, complex T125 discourse (see once more “merged discourses” above).
Possible Sarvāstivāda elements
As mentioned above, Hiraoka has pointed to a number of elements in T125 that show connections to Sarvāstivāda narrative traditions – in some cases, quite strong connections.Footnote 138
To be clear, a connection between Sarvāstivāda and T125 does not necessarily count as anomalous. The main thing that might make it appear anomalous is the fact that, as reviewed above, the hypothesis of a Mahāsāṅghika affiliation for the Indic Vorlage of EĀ has tended to dominate the uncertain scholarly treatment of the problem of the sectarian affiliation of the collection. The strong association of Sarvāstivāda and Abhidharma in the modern scholarly imagination might also lead us to perceive a certain possible tension between (quasi-)Mahāyāna features of the collection, and any Sarvāstivāda colouring we can find. Neither of these considerations is certain of interpretation. Nonetheless, the possibility that Sarvāstivāda elements might be heterogeneous with some other features of the collection means, to my mind, that it is worth tracking the distribution of those elements, in case they correlate with other phenomena.
To anticipate somewhat, Sarvāstivāda content may also be worth tracking for other reasons. First, in what we might call his “middle period,” Zhu Fonian first collaborated with Dharmanandin, around 391, on the production of the *Dharmavivardhana-sūtra T2045, which has close associations with the Divyāvadāna – a collection which, in turn, has close associations with Sarvāstivāda. Zhu Fonian then also collaborated with *Saṅghabhūti 僧伽跋澄/僧伽䟦澄 (fl. 381–399?) on the production of the Udānavarga T212 in 397–399 – and T212 also has Sarvāstivāda associations.Footnote 139 Further, Dharmanandin himself was also the reciter of the initial Chinese translation of the Madhyamāgama, which has also been identified as Sarvāstivāda. In conjunction with his work on T2045, this suggests that Dharmanandin had Sarvāstivāda training or affiliations. Saṅghabhūti also had Sarvāstivāda associations – he brought the *Vibhāṣā to China, and is credited with its earliest translation (鞞婆沙論 T1547). We will consider below the hypothesis that Zhu Fonian could be responsible for some of the heterogeneous content in T125. His ongoing association with Dharmanandin, his association with Saṅghabhūti, and his translation of these two Sarvāstivāda texts all indicate that he could have been a vector for the introduction of Sarvāstivāda elements into EĀ.
Finally, and again to anticipate, I will also discuss below internal stylistic evidence that some EĀ discourses might have been composed after the production of the Chinese Sarvāstivāda Vinaya T1435, and in knowledge of its content. This factor, too, suggests that Sarvāstivāda content could be worth tracking.
Hiraoka organises his Sarvāstivāda evidence into three categories, depending upon his degree of confidence that it shows a genuine and exclusive relation between Sarvāstivāda and T125. My impression of the evidential value of some points differs somewhat from Hiraoka's own. In my view, the strongest items of evidence that he adduces are these: details in the description of the Buddha's miraculous smile in connection with the issue of a vyākaraṇa (EĀ 43.2);Footnote 140 details of the Buddha's prediction of eventual attainment of pratyekabuddha status for Devadatta (EĀ 49.9);Footnote 141 details in a story about an offering of 2,500 parasols (EĀ 38.11)Footnote 142; details of a past life story of Dharmaruci in the context of the Dīpaṃkara story (EĀ 20.3);Footnote 143 details of the jātaka relating the escape of a caravan-leader from the island of the rākṣasīs (EĀ 45.1);Footnote 144 and a tradition holding the period of human gestation to be eight or nine months (EĀ 31.10, 33.2, 34.1, 38.6, 38.11, 51.3).Footnote 145 With the exceptions of EĀ 31.10 and 34.1, all the discourses in this list feature a concentration of anomalous features in other categories (see further below).
In earlier work, I myself have also pointed to possible Sarvāstivāda connections in details of the treatment of the story of Ajātaśatru (EĀ 43.7).Footnote 146 Kuan notes that Sarvāstivāda sutras also seem to share the predilection he detects in T125 for settings in Magadha.Footnote 147 Palumbo suggests Sarvāstivāda connections for the idea that the four great rivers issue from Lake Anavatapta (EĀ 29.9, 48.5), which is also found in the *Mahāvibhāṣā; and a parallel to one discourse (EĀ 29.3), which is cited and discussed in all three Vibhāṣās.Footnote 148 Palumbo has also proposed a number of circumstantial reasons that a Sarvāstivāda connection, no less than a Mahāsāṅghika one, might be compatible with the Mahāyāna colouring that we observe in some T125 discourses.Footnote 149 Palumbo further discusses at length narrative traditions in T1507 and T2026 that appear to connect the Vorlage(n) of one or more Chinese EĀ(s) with the Sarvāstivāda.Footnote 150 However, we also should note a tradition, communicated in T1507, that the Sarvāstivādins had an EĀ in ten vargas, rather than eleven. This tradition has sometimes been taken to rule out a Sarvāstivādin affiliation for the present T125 as a whole, since it goes up to the Elevens.Footnote 151
Hiraoka has advanced the stimulating suggestion that T125 may in fact be “a patchwork” from the perspective of sectarian affiliation.Footnote 152 Such an approach might indeed offer a way out from the apparent impasse of scholarship on the question of sectarian affiliation. If the collection indeed combines materials from more than one different transmission lineage, trying to determine “the” affiliation of the collection as a whole may be asking the wrong question. Nonetheless, I have judged that it may also be worth tracking possible Sarvāstivāda-like materials, and I have therefore included this evidence as a separate category in Appendix 5.
Phraseology regular in Dharmarakṣa, but never in Zhu Fonian outside T125
Thus far, I have been surveying evidence of types already discussed in prior scholarship. In the sections to follow, I turn to newly discovered internal, phraseological evidence of several types, which also suggests that T125 discourses containing it may be anomalous in some way.
I was initially spurred to undertake the work leading to this study by the observation, in the course of other work,Footnote 153 that T125 discourses sometimes feature words or turns of phrase, otherwise rare, that are not found in Zhu Fonian's corpus outside T125, but do recur in several works by Dharmarakṣa. It was natural to ask whether such phraseology indicated something about the nature of the discourses in which it featured, and whether discourses featuring such phraseology were still otherwise stylistically representative of the Zhu Fonian idiom.
We search for evidence with which we can address these questions by a simple method. We first search for markers that distinguish Zhu Fonian against Dharmarakṣa. Here, we naturally bracket out T125 itself from the Zhu Fonian corpus, since it is the target under investigation. We then look for the subset of the evidence so found that also occurs in T125.Footnote 154 We also test for the contrary hypothesis: we reverse the procedure, and see whether we can discover, conversely, markers that distinguish Dharmarakṣa specifically from Zhu Fonian. We then see whether there is any difference in the distribution of each type of marker across the various T125 discourses.
To anticipate, it is vital to note immediately that these tests discovered much more phraseology associating T125 with Zhu Fonian, than with Dharmarakṣa. Thus, the first result of this set of tests was to affirm the basic ascription of all discourses to Zhu Fonian. I discuss this aspect of the results in more detail below (p. 26 ff.).
Nonetheless, the tests also discovered a substantial body of phraseology in T125 suggesting close connections to Dharmarakṣa/the W. Jin. In Appendix 2, I list 337 items of phraseology shared by T125 and multiple Dharmarakṣa texts, but never otherwise found in Zhu Fonian. It is striking that it was possible to find so much evidence of this type. Apparently, parts of T125 sport features that are more strongly “Dharmarakṣa-like,” or “W. Jin-like,” than they are typical for Zhu Fonian. At the same time, as I have just mentioned, we cannot doubt the basic ascription of all of T125 to Zhu Fonian, including discourses with a concentration of this Dharmarakṣa-like diction. I will return to the discussion of this pattern later, after examining some other classes of anomalous-looking phraseology in T125.
Phraseology typical of Zhu Fonian's original works (the “Mahāyāna quartet”), but not his translations
Zhu Fonian's corpus includes four Mahāyāna sutras: the Shi zhu duan jie jing 十住斷結經 T309, the Pusa chu tai jing 菩薩處胎經 T384, the Zhongyin jing 中陰經 T385, and the Pusa yingluo jing 菩薩瓔珞經 T656. Nattier showed that T309 is in fact a Chinese composition, not a translation – but that the ascription is correct, that is to say, it is indeed by Zhu Fonian.Footnote 155 In a follow-up study, Lin Qian and I showed that the same is true of T656, and suggested more circumstantially that it is probably also true of T384 and T385.Footnote 156 At the present state of research, then, the best-grounded hypothesis is that all four of these texts are original Zhu Fonian compositions. I will refer to these texts as the “Mahāyāna quartet.”
In Appendix 3, I list 222 items of phraseology that are found in these four texts and in T125, but not in “real translations” by Zhu Fonian.Footnote 157 We might consider various reasons that these items would appear in the “Mahāyāna quartet,” and not in Zhu Fonian's other works, and consequently, for the presence of the same items in T125.
Some items have a clearly discernible link to Mahāyāna content, e.g. 一生補處 (#O004), 三世諸佛 (#O008), 佛國 (#O043). By contrast to the “Mahāyāna quartet,” none of the genuine translations in Zhu Fonian's corpus have any discernible connections to the Mahāyāna, and such phraseology may therefore distinguish the “Mahāyāna quartet” because of the differing content that typifies those texts. The presence of the same items in T125 might then overlap considerably with the presence of various types of (tendentially) Mahāyāna colouring, already discussed above. However, this consideration certainly cannot explain all the items on our list.
Alternatively, we have substantial primary external evidence showing that when working on real translations, Zhu Fonian worked in teams.Footnote 158 By contrast, interestingly enough, for the “Mahāyāna quartet,” we have very little evidence in the primary documents of our external evidence. We consequently do not know much about the circumstances under which these texts were produced, nor even their dates (though we have a relatively clear idea about the dates of almost all other texts in Zhu Fonian's corpus). Given the different nature of the work, however, we can surmise that Zhu Fonian most likely worked alone on the “Mahāyāna quartet.”
Now, it is logical to expect that when he was working in groups, Zhu Fonian's wording was subject to various types of “control” or modulation. The wording he originally suggested in his (presumably oral) translation of the Indic text might have been modified through discussion, for instance; or other members of the group might have subsequently edited or polished the text in various ways (as we know was the case for the original translation of EĀ itself in the Dao'an groupFootnote 159). Indeed, when working in a group, Zhu Fonian himself might have made a conscious effort to conform to certain decisions of group policy in choice of translation terminology, at least as concerned such salient aspects of the text as technical terms, proper names, and formulae. By contrast, when he was working alone, we might expect that Zhu Fonian was able to relax back into his own voice, so to speak. This dynamic might explain some differences between “his” idiom when he was really translating, and his style in original compositions. The presence of the same traits in T125 discourses, in turn, might hint that they, too, are more of the nature of original compositions.
Alternatively again, if Zhu Fonian did have a collaborator (or collaborators) when working on his original compositions, some differences in phraseology might reflect the habitual usages of those unknown collaborators.
Whatever the reasons might be, Appendix 3 shows clearly that a number of fine differences can be observed between recurring features of Zhu Fonian's distinctive “original composition” idiom, and the idiom characteristic of his real translations.Footnote 160 It is noteworthy that so many of these turns of phrase should be shared by discourses in T125. This suggests that there may be places in T125 itself where we also catch snatches of Zhu Fonian's “composition” voice. For that reason, it seems worthwhile tracking this evidence, too, as part of our larger treatment of possible anomalies in T125, and seeing whether it clusters in particular parts of the collection.
Phraseology typical of later Zhu Fonian translations
As shown by Palumbo's close analysis of the primary documents, so far as we can tell from our external evidence, work on any version of EĀ that we might associate with Zhu Fonian's name – meaning, in this case, the work of Dao'an's Chang'an group – should have come to an end with the death of Dao'an in 385.
In Zhu Fonian's corpus, more broadly, we are fortunate, in that – with the exception of the “Mahāyāna quartet,” as just discussed – we have unusually solid external evidence, in primary documents, for the dates of translation of most of his other works.Footnote 161 This gives us the opportunity to detect differences between two widely separated phases in Zhu Fonian's translation career: first, the very earliest phase of his work, when he collaborated with Dao'an's group in Chang'an; and second, the very tail end of his career (once more, we bracket out T125, since we want our test to be agnostic about its status).Footnote 162
• T1505 (382 CE), T1464 (383), T1543 (383), T194 (384), T1549 (384);
• T1428 (410–412), DĀ T1 (413).
Appendix 4 presents 149 items of wording shared by T125 with the two large, late translations (DĀ and T1428), but never found in the earliest translations. As can be seen from scrutiny of that Appendix, many of these items also appear in texts known to have been produced in the middle period of Zhu Fonian's career, after Chang'an, but before the very latest texts: T2045 (translated in 391), T212 (399), and the “Mahāyāna quartet” (dates unknown). The point I wish to make with this phraseology, then, is not that it necessarily means that texts featuring it are as late as T1428 and DĀ, but that such diction is atypical for the Chang'an period.
According to the picture of the production of Zhu Fonian's EĀ, reconstructed in such fine detail by Palumbo on the basis of external evidence, then, such phraseology should not feature in T125, because Palumbo holds that the text was in essence complete in content (if not arrangement) by Dao'an's death. In some cases (especially that of some Vinaya terms, like 針筒#L139), we can be confident that the phrasing in question simply did not exist in the 380s. Even if some of these usages did already exist at that time, our test demonstrates that they were not (otherwise?, i.e. outside T125) used by Zhu Fonian (or his group) in that early period. This suggests that discourses in which this sort of phraseology concentrates may have been altered or even added to the collection later than the original production of the first version of the collection, i.e. after 385. Such discourses should therefore be targeted for further investigation.
Phraseology first appearing after the year 398
Finally, we turn to one more type of phraseological evidence that indicates possible anomalies in T125. The application of TACL allows us to discover about eighty items of phraseology that never appear in any reliably ascribed and dated textFootnote 163 before the Madhyamāgama (trans. 398) or the time of Kumārajīva (who arrived in China around 402).Footnote 164
~中有比丘尼, 一切智人, 三昧力故, 不種善根, 乞食, 食已…, 五受陰, 人不應與~, 今云何言, 今當更~, 付囑, 何因緣故, 何況餘人,Footnote 165 作妄語, 作房舍, 偈中, 先生,Footnote 166 共和合,Footnote 167 再三教, 冷病, 到佛所頭面禮足, 受其語, 唾器, 因緣、果報, 因緣譬喻, 好守護, 婆羅門、居士 (also #L062), 實定, 實爾世尊, 小因緣, 居士婦, 巷中, 弟子應~, 從何處來, 必應, 惡不善法有覺有觀, 應作是言, 打揵椎/揵稚/揵推/揵搥, 拘尸城, 按摩, 提婆達多,Footnote 168 放牛人, 斷食, 是故應~, 更無異~, 最上最妙/最妙最上, 未犯, 正位,Footnote 169 比丘住處, 比丘少欲知足, 汝等實~, 生悔心, 生、老、病、死、憂、悲、苦、惱, 異論, (為, etc.) 病人說法, 看病人, 瞻波,Footnote 170 破魔, 神力故, …者善, 若不~, 聲聞乘、辟支佛乘、佛乘, 菩薩摩訶薩成就~, 見餘比丘尼, 語居士, 語餘比丘, 諸佛神力, 諸寶物, 諸比丘不應~, 講堂門, 財施法施, 貧賤人,Footnote 171 迴向,Footnote 172 遊行到~, 邊地人, 針筒/針筩, 間錯, 阿那律、難提、金毘羅,Footnote 173 飢儉,Footnote 174 首陀羅, 默然受已.
On the received chronology, the presence of such phraseology in T125 is anachronistic. Generally speaking, then, all this phraseology suggests that contrary to Palumbo's hypothesis about its history, our extant T125 incorporates modifications to the content of the texts dating from after the time of Dao'an.
Some items on this list are very strongly associated with the Kumārajīva corpus as a whole, e.g. 偈中 (T1509, T1521, T1564, T1646). Curiously, other items are concentrated in MPPU/DZDL T1509, e.g. 今云何言, 今當更~, 何況餘人, 因緣、果報.
A considerable number of the items listed here are concentrated in the Sarvāstivāda Vinaya translated by Kumārajīva's group, 十誦律 T1435 (and in some cases, other Vinayas following): e.g. ~中有比丘尼, 人不應與~, 作房舍, 共和合, 作妄語, 再三教, 好守護, 打揵椎/揵稚/揵推/揵搥. In fact, even items without obvious Vinaya connotations can reveal such a connection on closer examination. For example, …乞食, 食已… (“begged for food, and having eaten it…”) features in a distinctive Vinaya rendering of a formula describing a monastic rising at dawn to go on the almsround; 共和合 (“mutual harmony”) almost always features in contexts speaking of (and urging) concord in the Saṅgha; 唾器 “spittoon” appears to have been coined because of the need to translate regulations about such utensils (were they exotic at the time?);Footnote 175 and even 實爾世尊 (“Indeed it is so, O Bhagavat”) is heavily concentrated in Vinaya texts.
This pattern hints that some T125 discourses are distinguished by interest in topics, narratives, and tropes more at home in Vinaya contexts. Admittedly, even in Dao'an's time, EĀ is already reported to have had a distinctly Vinaya colouring.Footnote 176 Nonetheless, the use of this specific phraseology suggests that someone who contributed to those EĀ discourses, in their present form, was working after the production of T1435 in 405–406.Footnote 177 In light of the fact that T1435 is a Sarvāstivāda Vinaya, this evidence may also cast fresh light on the apparent Sarvāstivāda connections of some narrative material in T125 (discussed above, p. 19 ff.), and also, on the verbatim match with the possibly Sarvāstivādin prātimokṣa of Stein 797 studied by Palumbo (see p. 7). This aspect of T125 certainly merits further investigation.
Generally speaking, it stands to reason that most of these items are entirely atypical of the Zhu Fonian corpus. My test targeted items that never appear before the year 398, and that period includes most of the firmly datable works in the Zhu Fonian corpus. All items in those texts were therefore excluded by the test conditions. However, we should note that some of the phraseology here appears in Zhu Fonian's latest translations, DĀ T1 and T1428 (which were produced after Kumārajīva's time, around 410–413), e.g. 五受陰, 今當更~ (T1); … 乞食, 食已…, 今云何言, 今當更~ (T1428).
Incidentally, some of these items also appear in Zhu Fonian's “Mahāyāna quartet.” Examples are 放牛人, 飢儉, 菩薩摩訶薩成就~ (T309, T384); 神力故 (T309, T384, T385); 打揵椎/揵稚/揵推/揵搥 (T384); 先生, 破魔 (T384, T385); and 生、老、病、死、憂、悲、苦、惱 (T656). This may hint that the texts in the “quartet” indeed post-date Kumārajīva's arrival, as Nattier suggested.Footnote 178
To anticipate the discussion below, these items are by no means evenly distributed in T125. Almost none appear in the first twenty fascicles of the collection (with the notable exception of the “Preface,” EĀ 1). Thereafter, they appear in almost every chapter of the collection, with some increase in frequency in the last ten fascicles or so.Footnote 179
Confirming the ascription of all T125 discourses to Zhu Fonian
With all the evidence of anomalies in prior scholarship surveyed above, it is natural to ask whether T125 might include some discourses that have nothing to do with Zhu Fonian (at least, I found myself asking that question). However, as I will now demonstrate, the basic ascription of all T125 discourses to Zhu Fonian, including even those sporting numerous anomalies, can hardly be called into question.
Phraseology shared by T125 and Zhu Fonian, but not Dharmarakṣa
As mentioned above, prior work led me to discover some diction in T125 that is more characteristic of Dharmarakṣa than of Zhu Fonian. Spurred by that discovery, as already described above (p. 21), I undertook a set of tests in which I set Dharmarakṣa and Zhu Fonian “head to head” – that is to say, I looked explicitly for phraseology associating T125 with Dharmarakṣa and not with Zhu Fonian, and vice versa.
These tests easily discovered a substantial body of evidence clearly corroborating the basic ascription of all discourses in EĀ to Zhu Fonian, against Dharmarakṣa. Over 600 items of phraseology are listed in Appendix 1. Each occurs (a) in T125; and (b) multiple times in at least one other Zhu Fonian text (usually more than one text – often quite a number); but (c) never in Dharmarakṣa (with a very few select exceptions).Footnote 180 This evidence comprises a wide range of parts of speech and turns of phrase, many very typical for Buddhist texts. Between them, these markers comprise a rich and convincing signature of translation style. In all discourses in T125, such phraseology far outweighs phraseology associated with Dharmarakṣa against Zhu Fonian.Footnote 181 Thus, even discourses with unusually “Dharmarakṣa-like” diction are genuine products of Zhu Fonian, or a group that he belonged to.
The basic ascription of all T125 discourses to Zhu Fonian can also be corroborated by an even larger body of more rigorously discovered evidence. However, presentation of that evidence in full is a complex matter, and must be left to a future study.Footnote 182
Rare phraseology shared with Zhu Fonian's “Preface” to the *Dharmavivardhana-sūtra
An additional class of evidence associates some discourses in T125 even more tightly with Zhu Fonian as an individual – not just with the corporate entities that produced the translations we associate with his name.
Zhu Fonian potentially represents an exceptionally rich object of study, because he is one of the rare cases in which we have both translations, and works from his own brush.Footnote 183 In fact, in his case, we have original works in two quite different genres. As already discussed, we have the “quartet” of Mahāyāna sutras, which are demonstrably his own original compositions (T309, T656), or probably so (T384, T385). In addition, however, Chu sanzang ji ji has transmitted a “Preface to the Sutra of the Tale of How Prince *Dharmavivardhana Lost His Eyes” 王子法益壞目因緣經序, i.e. the preface to Zhu Fonian's translation of the *Dharmavivardhana-sūtra T2045.Footnote 184 This preface bears a byline ascribing it to Zhu Fonian, and we will see immediately below that this ascription is borne out by copious internal evidence.
We would ordinarily assume, by default, that a preface of this type was authored by a single figure, unassisted. Zhu Fonian's preface thus gives us an exceptional opportunity to sample his individual voice. In addition, the preface is not constrained by the dynamics of translation, where word choice is influenced by the effort to reflect a Vorlage with many characteristics alien to the writer's native idiom. The genre of the preface certainly has equally ornate and artificial constraints of its own, but within those limits, the preface also gives us an opportunity to see how Zhu Fonian words himself when he is not stretching to meet an Indic text halfway.
Using TACL, we can discover the following rare phrasing shared by Zhu Fonian's Preface, and his translation works, including T125 discourses:
受困 (also T212); ~之累 (also T212, T309); ~之深(~) (also T212, T309); 神寺 (also T212); 玄鑒 (also T212, T309, T384, T656); 難計 (also T212, T656); (設有)毫氂/毫釐 (also T212, T309, T384, T656); 著翅~Footnote 185 (also T385); 有所由 (also T212, T309); 出斯 (also T212, T309);Footnote 186 天竺 (also T212, T1464, T1505, T1543);Footnote 187 在王宮 (also T212); 將來之~ (also T212); 殃舋 (also T212); 耳目 (also T656); 來變 (not 如來變) (also T212); 萌兆 (also T212, T309, T656); 不朽 (also T212, T384, T656, T1464, T1549, T2045 body); ~貌殊特 (also T212, T1428); 萬民 (also DĀ T1, T212, T2045 body); 登位 (also DĀ T1, T309, T384, T1428); 師宗 (also T384, T656, T2045 body); 喪目 (also T1549); 引入 (also T309, T384, T1464).
For reasons of space, I have refrained from showing the full distribution of these items in canonical translations; interested readers can easily check for themselves using the usual digital search tools. It should be emphasised, however, that especially if we restrict our view to the period before Sengyou, all of these items are rare in translations outside the Zhu Fonian corpus. In particular, 玄鑒 is a “smoking gun,” never found in any translator before Sengyou except for Zhu Fonian. Another item, 毫氂/毫釐, is apparently a real favourite, occurring more than fifty times in eight Zhu Fonian translations.
One clear implication of this test is that there can be no doubt (if there ever was any) that the Preface to T2045 is genuinely by Zhu Fonian. It is remarkable how many such items can be found, given the brevity of the Preface (492 characters). This short text is like a dense showcase of characteristic Zhu Fonian turns of phrase.
Outside the Zhu Fonian corpus, some rare items in the Preface appear most frequently in the Dharmarakṣa corpus, e.g. ~之深, ~之室, 神寺, 不寤, 難計, 殃舋, 將來之~, 罪福之~.Footnote 188 This fact further affirms the proximity of Zhu Fonian's idiom to that of Dharmarakṣa. Indeed, one item, 殃舋, is a “smoking gun” for Zhu Fonian, if we only exclude Dharmarakṣa (and vice versa); another, ~之室, is likewise a “smoking gun” for the two combined in the period before Sengyou; and 神寺 is also nearly exclusive to these two translators. Because these same items also appear in the T2045 Preface, they serve as particularly strong evidence that Zhu Fonian's idiom, even when he was not translating, was indeed particularly close to Dharmarakṣa.
It is striking to note how many of these items are found in Zhu Fonian's original “Mahāyāna quartet,” and especially, in T212.Footnote 189 This finding dovetails with the evidence in Appendix 3, and may reinforce the suggestion that when he was not translating, Zhu Fonian gravitated towards a slightly different idiom. Once more, the presence of items characteristic of that sub-idiom in some T125 discourses may be a sign that those discourses were (in part?) produced by processes more analogous to those operative for the “Mahāyāna quartet,” or for this preface, than for Zhu Fonian's “true translations.”
Synthesis and analysis: the uneven distribution of these types of evidence in T125
To recapitulate: a number of features suggest that T125 is the most peculiar Āgama. An unusually large number of its sutras are unparalleled; near-verbatim Doppelgänger of three discourses float free in other parts of the Taishō; the question of the “sectarian affiliation” of T125 has thus far defeated scholarship; the size of the present collection, and the distribution of the uddānas, does not appear to match Dao'an's account at the moment of production; initial production of the collection was hampered by some forgetfulness on the part of the reciter; the Elevens may have been a late afterthought on the part of the Indic tradition behind the collection; also in the Elevens, the evidence of T2026 suggests that the collection should end at Ch. 49, so that the last three chapters (50–52) may be a late addition; and tradition and modern scholarship have struggled to determine who translated our received text. The collection also has a complex and chequered history. EĀ 50.4 doubles up with narrative material in EĀ 1, and appears to be by a different hand; verses in EĀ 48.2 match a Sarvāstivādin prātimokṣa preserved in a Dunhuang manuscript; and one passage may show signs of an error committed in writing, in Chinese. As Palumbo has shown, moreover, the Dao'an group probably produced three recensions of the text in quick succession, and the collection appears to have then circulated in various arrangements, and possibly two different contemporaneous versions, down to the sixth century.
We then surveyed five types of evidence which might indicate possible anomalies in individual sutras in T125, by comparison with what is usually more typical for Mainstream Āgama collections. These elements have already been studied in prior scholarship: (proto- or quasi-) Mahāyāna-like elements; “merged” discourses, combining material found in two or more separate sutras in other transmission traditions; discourses which are only paralleled in late (post-canonical or commentarial) materials in other transmission lineages (especially Pāli); and material with probable Sarvāstivāda associations, which may be anomalous, against the backdrop of common judgements that the collection as a whole was most probably not basically Sarvāstivāda. A further indication of possible room for anomalies, though only a weak and circumstantial one, is the absence of known parallels for an individual sutra (i.e. sutras that are “sole exemplars”); conversely, sutras that do have known parallels are thereby less likely to be simple inventions out of whole cloth in China.
In addition, we surveyed four classes of phraseological evidence which likewise indicate possible anomalies in T125: wording typical of Dharmarakṣa (or the W. Jin), but not of Zhu Fonian; wording typical of Zhu Fonian's original composition sutras, but not his true, unadulterated translations; wording typical of Zhu Fonian's later translations, rather than the earliest ones produced in the Chang'an years under Dao'an; and wording that first appears after the year 398, in MĀ or Kumārajīva.
What, then, does all this evidence show us when we consider it in concert? Appendix 5 is the heart of the analysis upon which my argument here is built. It organises the evidence in all the various classes we have discussed, as it features in individual discourses in T125. Perusal of Appendix 5 shows that the evidence in all classes is very unevenly distributed, and the distribution of various anomalies is correlated. In short, certain discourses and sections in T125 are hotspots for anomalies of multiple types.
In this section, I will walk the reader through the most important aspects of this evidence, to highlight several patterns that emerge. On that basis, I will advance two arguments. First, portions of T125 must have been altered after the initial translation of the collection by Dao'an's group in 384–385 – probably after the year 400. Second, and more speculatively, we should hypothesise that these alterations were introduced to the collection, at least in part, by Zhu Fonian himself.
I will begin with the clearest cases. As can be seen in Appendix 5, a small number of individual discourses leap out particularly, because they display densely concentrated evidence of possible anomalies: discourses like EĀ 20.3, 23.1, 24.5, 24.6, 24.8, 26.9, 27.5, 28.1, 30.3, 31.1, 31.8, 32.5, 33.2, 34.2, 34.5, 36.5, 37.2, 37.3, 38.6, 38.11, 42.3, 43.2, 43.7, 44.7, 48.2, 48.3, 48.5, 49.6, 49.9, 51.3, 51.8, 52.1, and 52.2. (Noticeably, apart from EĀ 1, we find no such discourses in the first nineteen chapters; see further below.) For the sake of argument, I will focus first on these discourses, which I will provisionally refer to as the “weirdest thirty-three.” Anomalies are so heavily concentrated in these discourses that something is undeniably afoot.
Let us take EĀ 36.5 as a first example.Footnote 190 This discourse tells the story of the Buddha returning from the Heaven of the Thirty-Three, where he has spent the rains retreat preaching to his mother, Māyā. He descends on the central staircase of three that have miraculously appeared for the occasion. On the flanking staircases, Indra and Brahmā wait upon him sedulously, with the royal yak-tail fan and parasol. During his absence, King Udayana has had the first ever Buddha image made. Upon the return of the Buddha in the flesh, a throng rushes to see him – note the overtones of what we might call “Buddhist darśan.”Footnote 191 Subhūti, however, decides not to join the darśan-seekers. He reflects that the Buddha cannot truly be seen in the faculties of his fleshly body, because “all dharmas are empty and tranquil” 一切諸法皆悉空寂. There are obvious echoes here of the famous Vakkali-sutta trope, “He who sees the dharma, sees the Buddha.”Footnote 192 Subhūti further reflects, in verse, “Anyone who wants to pay obeisance to the Buddha … should contemplate empty dharmas” 若欲禮佛者….當觀於空法. He reiterates to himself that all dharmas are empty and tranquil 諸法皆悉空寂, and decides to take refuge in “the collectivity of true dharma(s)” 我今歸命真法之聚. Then he goes back calmly to mending his robe.Footnote 193 Subhūti's attitude is contrasted with that of the nun Utpalavarṇā, an enthusiastic Buddha-viewer, who transforms into a cakravartirāja to win prime position in the crowd.
We need not quibble about the exact extent to which Subhūti, emptiness, or the Buddha image count as “Mahāyāna.” The content of EĀ 36.5 is already remarkable. The evidence collected in the entry for EĀ 36.5 in Appendix 5 shows that the discourse also has an unusual concentration of other remarkable features. It sports elements elsewhere found only in commentarial or post-canonical texts: the story of the nun Utpalavarṇā becoming a cakravartin, and the “miracle at Saṃkaśya” (of the Buddha's descent). In these traditions, the discourse merges material found elsewhere in separate texts. It features a number of Mahāyāna-leaning notions: the bodhisatva path as a generalised goal, reference to the *trisāhasramahāsāhasralokadhātu, and the Buddha giving vyākaraṇa to others, as well as the figure of Subhūti and his reflections on “emptiness.”
In phraseological terms, meanwhile, EĀ 36.5 also features unusually large quantities of Dharmarakṣa-like diction, wording characteristic of Zhu Fonian's original compositions, and phrasing typical for Zhu Fonian's later translations; and it even features one item heavily characteristic of Kumārajīva's Prajñāpāramitā texts (爾時釋提桓因白佛言, in T223, T227, MPPU/DZDL T1509 – though this item also appears in *Mokṣala's T221). Thus, the indications of content and style converge. A discourse featuring several outlier traits, which have already sounded alarm bells for prior scholars, also has a heavy concentration of stylistic features atypical for the regular Zhu Fonian corpus, the period of the original translation of EĀ, the genre of Āgama translations, and T125 more generally.
As a second example, we might refer more briefly to EĀ 48.3 (doubled in T453). This text treats narratives surrounding the coming of the future Buddha Maitreya. It places in the mouth of “our” Buddha, Śākyamuni, traditions concerning Maitreya himself, and his relations with Kāśyapa, against the backdrop of a Land-of-Cockaygne-like depiction of the future world in which these events transpire. For our analysis, it is significant that this discourse features concepts redolent of the Mahāyāna: *triyāna 三乘, Maitreya as a bodhisattva, the *trisāhasramahāsāhasralokadhātu, and the appearance of multiple bodhisattvas in the same scene. In addition, however, it features ten items of diction associated with Dharmarakṣa rather than Zhu Fonian, five items usually found only in original compositions in the Zhu Fonian corpus, and five items otherwise only found in later translations in the Zhu Fonian corpus. Once more, then, apparent anomalies in content are associated with an unusual concentration of anomalies in phraseology. EĀ 48.3 also features the above-mentioned “smoking gun” for Zhu Fonian, which appears in his Preface to T2045: 玄鑒; and one other item of rare diction also found in the Preface: 神寺. These features mean that at least parts of the text are probably closely associated with Zhu Fonian as an individual, not just with the groups he translated in.
A third example is significant for a different reason. Kuan has shown that an episode in EĀ 37.2, in which Mahāmaudgalyāyana journeys to a remote Buddha-field, is probably based upon a similar episode in Dharmarakṣa's *Tathāgataguhyaka-sūtra 密迹金剛力士會 T310(3).Footnote 194 On the basis of comparison with citations of a similar EĀ discourse in JLYX, Kuan argues further that EĀ 37.2 was the product of revision to an earlier version of EĀ produced in Chang'an by the Dao'an group, and that Maudgalyāyana's cosmic voyage was added as part of that revision.Footnote 195 On the basis of two phraseological clues,Footnote 196 and reference to aspects of Zhu Fonian's career and known practices (as also discussed elsewhere in the present paper), Kuan argues that Zhu Fonian was most likely the author of these revisions. This discovery is potentially quite significant.Footnote 197 If T310(3) was indeed the source of this passage in EĀ 37.2,Footnote 198 it would represent the first time, to my knowledge, that a specific earlier Chinese source has been found for material in T125 (excepting EĀ 1). For our purposes, then, it is potentially significant that EĀ 37.2 is, again, one of the “weirdest thirty-three.” As we see in Appendix 5, it features Mahāyāna-ish elements like “Buddha lands” 佛土, rare names of other Buddhas, and the *trisāhasramahāsāhasralokadhātu; it is (as a whole) unparalleled; and it features an unusual concentration of diction more typical of Dharmarakṣa than Zhu Fonian, of Zhu Fonian's original works rather than his translations, and of his later works rather than his early work under Dao'an. Finally, it is also suggestive that it features one word, 難計, which is otherwise rare, but found in Zhu Fonian's Preface to T2045.Footnote 199
If we zoom out from these examples, and consider the “weirdest thirty-three” as a group, we note that quite a number of these discourses feature possible Mahāyāna-like material (fifteen discourses, or about half: EĀ 20.3, 23.1, 24.5, 24.6, 26.9, 27.5, 32.5, 36.5, 37.2, 37.3, 42.3, 43.2, 48.2, 48.3, 48.5). A number also feature elements with possible Sarvāstivāda connections (seven discourses: 20.3, 33.2, 38.6, 38.11, 43.7, 49.9, 51.3). It is fairly common for these discourses to feature content found in post-canonical materials in Pāli (six discourses: 34.2, 36.5, 38.11, 43.2, 44.7, 52.1). “Discourse merger” is also quite common (eight discourses: 24.8, 31.8, 36.5, 38.6?, 42.3, 43.2, 48.2, 49.6).
Now, as mentioned above, it is in principle possible that discourses could have acquired such potential anomalies in content prior to translation into Chinese. It is also possible that some anomalies in content found their way into the collection at the time of its original translation under Dao'an in Chang'an in 384–385. However, we see clearly in this sample that anomalous features of content are strongly associated with unusual concentrations of phraseological anomalies. Some of the phraseological anomalies, moreover, strongly suggest that at least some layer in these discourses is later than the original EĀ produced in 384–385, and possibly later than the arrival on the scene of some Kumārajīva texts. Twenty of these discourses (about two thirds) feature phrasing associated with the period after the year 398 (23.1, 24.5, 24.8, 26.9, 27.5, 28.1, 31.8, 33.2, 34.2, 34.5, 38.6, 38.11, 42.3, 43.7, 44.7, 48.5, 49.6, 51.3, 52.1, 52.2).Footnote 200 All thirty-three discourses feature some diction characteristic of later Zhu Fonian, in contrast to his earliest translations produced under Dao'an – often in notable quantities. These phraseological anomalies can best be explained by two hypotheses: first, that parts of T125 were altered or added later than the initial translation, probably after the year 400; and second, more circumstantially, that Zhu Fonian may have had a hand in such changes. In this light, it is certainly suggestive that possible anomalies in content also cluster in the same discourses. At least some anomalies in content may be coeval with anomalies in phraseology.
How far, then, does this pattern reach? For the sake of argument, I began by focusing on the “weirdest thirty-three” – a restricted subset of discourses in which the clustering of anomalies is particularly clear. Notoriously, however, absence of evidence is not evidence of absence. Other discourses might also have undergone similar reworking, without sporting the evidence that signals the problem in the same degree. These thirty-three discourses might be merely the tip of an iceberg. Appendix 5 thus presents us with a “slippery slope” problem – how much evidence do we need before we declare that a discourse is clearly anomalous?
I would certainly not want to make too much of this potential “evidence” for all discourses in which it appears. I initially set the threshold for inclusion of a discourse in Appendix 5 at two or more items, in order to capture as much of the picture as possible. However, this threshold really is rather low. For example, with a few exceptions, very few discourses in the first twenty chapters (excepting EĀ 1) feature more than four items of my phraseological evidence (exceptions are 8.1, 10.8, 11.7, 11.8, 11.10, 12.1, 13.3, 13.5, 13.7, 16.1, 17.1, 18.4, 18.7, 19.1, 19.3, 19.11, and 20.3); and almost none feature post-398 diction (but see 3.9, 11.7, 11.8, 12.1, 13.3). We also see that the first nineteen chapters include no discourses with anomalies at the levels seen in the “weirdest thirty-three” (in order of appearance, the first discourse in the collection on my list is 20.3). Generally speaking, then, the first twenty chapters or so of T125 appear to be largely free of unusual concentrations of our potential evidence. Perhaps this indicates that a certain low-level presence of “possible anomalies” in at least some discourses is explicable by chance, or other dynamics, and we should only get suspicious when that concentration is exceeded.Footnote 201
In fact, if we zoom out to the level of T125 as a whole, we see that 208 discourses (a little under half) feature either no evidence in the various classes listed in Appendix 5, or only one such item.Footnote 202 To save space, I have excluded such untouched discourses from Appendix 5 entirely. It is instructive to list those discourses. Large stretches of T125 appear unscathed by the pattern we are tracing, in particular towards the beginning of the collection:Footnote 203
2.1, 2.2, 2.3, 2.4, 2.5, 2.6, 2.7, 2.8, 2.9, 2.10
3.2, 3.4, 3.6, 3.10
4.3, 4.4, 4.7
5.2, 5.4 (defective decade, five discourses)
6.1, 6.2, 6.3, 6.4 (defective decade, four discourses)
8.2, 8.3, 8.4, 8.5, 8.7, 8.9, 8.10
9.3, 9.4, 9.5, 9.6, 9.8, 9.9, 9.10
10.1, 10.2, 10.4, 10.7, 10.9, 10.10
11.1, 11.2, 11.3, 11.4, 11.5, 11.6, 11.9
12.2, 12.4, 12.5, 12.7, 12.8, 12.9, 12.10
13.2 (defective decade, seven discourses)
14.1, 14.2, 14.3, 14.4, 14.5, 14.6, 14.7, 14.8, 14.10
15.1, 15.2, 15.3, 15.4, 15.5, 15.6, 15.7, 15.8, 15.9, 15.10
16.2, 16.3, 16.5, 16.6, 16.7, 16.8, 16.10
17.2, 17.3, 17.5, 17.6, 17.8 (supernumerary decade, eleven discourses)
18.1, 18.6, 18.8, 18.9, 18.10
19.4, 19.5, 19.7, 19.9 (supernumerary decade, eleven discourses)
20.1, 20.4, 20.5, 20.9, 20.11 (supernumerary decade, thirteen discourses)
21.1, 21.2, 21.4, 21.6, 21.7, 21.8, 21.10
22.1, 22.2, 22.3, 22.4, 22.5, 22.6, 22.7, 22.8, 22.9, 22.10
23.7, 23.8, 23.9, 23.10
24.7, 24.10
25.1, 25.2, 25.3, 25.4, 25.5, 25.8, 25.9, 25.10
26.1, 26.2, 26.3, 26.4, 26.8
27.4, 27.6, 27.7, 27.8, 27.9
28.2, 28.3 (defective decade, seven discourses)
29.2, 29.4, 29.8, 29.10
31.3, 31.10, 31.11 (supernumerary decade, eleven discourses)
32.1, 32.2, 32.8, 32.11 (supernumerary decade, twelve discourses)
33.3, 33.5, 33.6, 33.7, 33.8, 33.9
34.4, 34.6, 34.7, 34.8, 34.9
35.1, 35.3, 35.4, 35.6, 35.8
36.1, 36.2, 36.3 (defective decade, five discourses)
37.4
38.2, 38.3 (supernumerary decade, twelve discourses)
39.4, 39.5, 39.6, 39.7
40.3, 40.8
41.2 (defective decade, five discourses)
42.9
43.8
44.1, 44.2, 44.4, 44.5, 44.8 (supernumerary decade, eleven discourses)
46.1, 46.2, 46.5, 46.9
47.2, 47.9
49.2, 49.10
50.2, 50.7
51.1, 51.2
52.3, 52.4, 52.5 (defective decade, nine discourses)
Large parts of T125 thus appear to be untouched by any phenomenon (or phenomena) our markers might indicate. Indeed, by the standard of the above list, some whole decades are completely clear of possible anomalies (e.g. Chs. 2, 15, 22). Moreover, through about the first half of the collection, most texts in many decades are largely unaffected.
Conversely, other stretches of T125 are hotspots. For example, in Chs. 45 (a defective decade with seven discourses) and 48 (also defective, six discourses), every discourse is at or over our threshold, and most discourses sport considerably more than two of our markers.
As I already mentioned, relatively little of our evidence occurs in the first thirty-odd chapters. Of course, there are exceptions. The highly anomalous “Preface” EĀ 1 is (unsurprisingly) riddled with evidence of all types, and must be treated separately. Another exception to the overall pattern is Ch. 13, where all discourses but one (of seven) are over the two-item threshold. Further, even in these portions of the collection, some individual discourses also have a particularly heavy dose of our evidence. On the whole, however, discourses in this first long part of the text, to Ch. 30 or so, rarely have more than a few items of possible evidence in total.
From about Ch. 30, by contrast, the density of our evidence increases, and discourses free of it taper off. The increase comes at first in patches. Then, most strikingly, in the last portion of the collection, beginning around Ch. 36, our evidence becomes more consistently thick on the ground. Thus, if our evidence indicates that someone tampered with T125, a first hypothesis might be that they did not lavish their attention evenly over the collection as a whole. Rather, it looks as if they concentrated their efforts in Ch. 1, and portions of the text after Ch. 20, perhaps working unevenly on particular stretches even within that range; and a particular focus was about the last third of the collection.
In sum, it seems likely that the pattern of problems we are tracing extends somewhat further than the “weirdest thirty-three” discourses, in which it is most strikingly evident. At the same time, it is difficult to know how much evidence suffices to show that an individual discourse is problematic, and we cannot assume that all later alterations or additions to the collection will have left visible traces. Closer investigation of individual discourses may allow us to make further progress with such questions (and our findings highlight the requirement for such work). Meanwhile, however, despite these difficulties, an overall imbalance in the distribution of the anomalies seems rather clear, and indicates that later interventions were more frequent in approximately the second half of the collection, than in the first.
In attempting to interpret this evidence, we should keep several factors in mind.
First, to reiterate, there can be little doubt that all discourses in T125 – even the “weirdest thirty-three,” for all their anomalies – are strongly associated with Zhu Fonian in some fashion. As discussed above (p. 26), all discourses in T125 strongly evince distinctive signs of the Zhu Fonian translation idiom. Many also feature items among the set of rare wordings I identified in Zhu Fonian's T2045 preface – in fact, even in the “weirdest thirty-three,” an unusually high proportion of discourses feature such diction (sixteen discourses of thirty-three, or about half: 23.1, 24.6, 24.8, 31.8, 33.2, 34.2, 36.5, 37.2, 38.6, 38.11, 42.3, 43.7, 48.3, 51.3, 51.8, 52.2). Of course, this evidence is not unequivocal. Zhu Fonian was also the main translator of EĀ in 384–385, and as I have already mentioned, ample stylistic evidence demonstrates that T125 as a whole is most closely associated with his style and no other. We might imagine, then, that evidence of Zhu Fonian's stylistic fingerprint belongs entirely to the initial stratum of the texts, produced in the Dao'an group in 384–385; by contrast, evidence of later diction, anachronistic against the work of the Dao'an group, could be due to the later intervention of another hand.
However, this hypothetical possibility is rendered rather unlikely by the fact that characteristic Zhu Fonian phraseology is found through all parts of T125, and further, that this phraseology is consistent not only with Zhu Fonian's translations, but also (perhaps all the more) with his probably original Mahāyāna sutras. Moreover, my evidence includes stylistic traits in several classes associated specifically with Zhu Fonian, but with parts of his corpus outside the context in which EĀ was originally translated, and outside its genre – his original works, his later works, and the T2045 preface – and these traits, too, concentrate particularly in anomalous discourses. It remains most likely, then, that on the whole, anomalous phraseological traits, too, stem from Zhu Fonian.
Next, the discourses in our restricted sample of the “weirdest thirty-three” discourses, for example, do include some “sole exemplars” – texts without known parallels. However, it is perhaps more important to note that for most of them, parallels are in fact known. Taking our “weirdest thirty-three” again as an example, in that group, twenty-seven texts are paralleled (more than four fifths): 23.1, 24.5, 24.6, 24.8, 26.9, 28.1, 30.3, 31.1, 31.8, 32.5, 34.2, 34.5, 36.5, 37.2, 37.3, 38.6, 38.11, 42.3, 43.2, 43.7, 44.7, 48.2, 49.6, 49.9, 51.8, 52.1, and 52.2. Thus, only six texts in this group are “sole exemplars.” This holds true more broadly of discourses featuring anomalous evidence – no particular correlation can be seen between concentrations of possible anomalies and “sole exemplar” discourses. Thus, it cannot be the case that all anomalous discourses were invented out of whole cloth in China. Any hypothesis we form about the origin of discrepancies in T125 must include a way of accounting for knowledge of the source materials indicated by the parallels (possibly in earlier Chinese versions of EĀ itself, which may have been reworked in some manner to produce extant discourses sporting anomalies).
Additionally, we should recognise that even if, hypothetically speaking, all our different classes of possible evidence do in fact indicate the presence of anomalies in the text, those anomalies may not necessarily all have been introduced to the text at the same stage, or by the same processes or actors. This holds especially for anomalies at the level of content. Merged discourses, for instance, or elements paralleled in Pāli post-canonical and commentarial sources, or possible Sarvāstivāda elements – or at least some such elements – might have been introduced to a Vorlage in India or Central Asia.Footnote 204 Some anomalies in content may also have been introduced at the time of initial translation under Dao'an in Chang'an. In fact, for the “Preface” chapter (EĀ 1), some oddities must at least have pre-existed the commentary, T1507, which comments extensively on that chapter.Footnote 205 Finally, we also cannot exclude a priori the possibility that the collection was modified in China, after its initial production by the Dao'an group, by two or more persons or groups at different times, so that phraseological evidence apparently indicating a layer (or layers) heterogeneous with the expected style of the Chang'an group and period might not all point to the same interventions.
Even admitting these complications, however, we can draw some cautious inferences from the evidence of Appendix 5.
Evidence of possible anomalies is very unevenly distributed in T125. This already suggests that the collection, in its present form, must be a blend of material heterogeneous in nature, content, and provenance. A considerable portion of this unevenly distributed evidence of possible anomalies is phraseological in nature, suggesting that some considerable portion of this heterogeneity does not only inhere at the level of an Indic Vorlage, but is also at least in part a reflection of vicissitudes undergone by the text in its post-translation, Chinese guise. These various types of evidence, unevenly distributed as they are, roughly correlate – where we find anomalies of content, we are more likely also to find anomalies of phraseology, and vice versa. This suggests that at least some of the anomalous features of content were introduced in association with the same processes that gave rise to the discrepancies in style. Further, some of the anomalous phraseology is associated with a period later than the initial production of EĀ by the Dao'an group in Chang'an in 384–385, and some, more specifically, with a period after the year 398. This suggests revision or modification later than the initial translation process – that is, after the production of all three Chang'an recensions revealed by Palumbo's research. Some later revision of this sort would have the additional advantage of possibly accounting for the apparent discrepancy in the present size of T125, the mystery of the extra chapters among the Elevens (Chs. 50–52), and potentially, some of the recensional chaos that prevailed down to the sixth century.
Therefore, we have strong reason to believe that some discourses in T125 underwent significant modifications or interventions after 385. Further, it is likely that those interventions went beyond the rearrangements in structure and order envisioned by Palumbo for the post-Dao'an life of the text. The interventions are reflected rather globally in the phraseology of some discourses; and the uneven distribution of that phraseology, and its correlation with anomalies in content, strongly suggests that at least some anomalies in content, too, in some discourses, are products of those post-385 interventions.
Next, I turn to the more speculative second arm of my argument. Even if all apparent anomalies in T125 may not have the same origin, it is natural to favour economical hypotheses. In this light, we should recall several salient facts. Zhu Fonian was already closely associated with EĀ through his participation in the original translation project. He is known to have produced at least two substantial sutras of his own (T309, T656), and possibly two more (T384, T385); he may also have added some layers of narrative material to his translation of the Udānavarga T212.Footnote 206 He thus had a demonstrated “track record” of interest in producing new scripture, and the capacity to do so. Zhu Fonian also had a long-time association with Dharmanandin, the original reciter of EĀ (they worked together at least until the translation of T2045 in 391). Through Dharmanandin, he had a possible source of information about genuine Indic textual traditions, including Sarvāstivāda traditions, of the type that might be reflected by the evidence of parallels.Footnote 207 (However, we should not forget that Zhu Fonian himself is also supposed to have travelled widely when young,Footnote 208 and may thus have had his own body of such knowledge.) Finally, Zhu Fonian was from Liangzhou,Footnote 209 and it is therefore plausible to assume that he was formed in a culture of Buddhist textual learning steeped in the texts of the W. Jin and their associated scriptural idiom. In this light, it is also little wonder that generally speaking, his own idiom is demonstrably close to that of Dharmarakṣa, and it is plausible that when freed from the constraints of collaboration and the guardrails of an exact Vorlage, his diction might have drifted even further in that direction.Footnote 210 In sum, Zhu Fonian provides us with a point of convergence where many of the apparently anomalous features in T125 could have been produced. Consequently, we should consider the hypothesis that Zhu Fonian himself revised the collection in some manner, probably after the year 398, and even after the translation of the Sarvāstivāda Vinaya around 405–406.Footnote 211
For the present, I advance this hypothesis as just that – a hypothesis. We might hope that future work will discover finer clues to the exact period of later interventions into T125, and for or against the contention that the author of those interventions was Zhu Fonian. It is also possible, however, that the exact timing and agent of such interventions will forever lie outside the reach of clearer proofs.
In closing, I must also underline the point that in my view, the evidence treated here allows us only limited insight into the exact nature of the interventions that T125 later underwent. Some things can be inferred, but others remain unclear. As noted above, many of the most anomalous discourses, including the “weirdest thirty-three,” include elements paralleled in other scriptural sources. That means that later interventions most likely did not invent whole discourses out of thin air; it is more likely that they represented some sort of reworking (rearrangement, rewriting, expansion, etc.) of existing material. On the other hand, I have already stated that some anomalies could have been introduced before translation, into the Vorlage; or at the time of initial translation under Dao'an in 384–385; as well as after translation, in the later interventions for which I here argue. This possibility makes it difficult for us to discern exactly which anomalies in content might be products of later intervention.
In this connection, it is important to recall once more that the evidence I treat is so unevenly distributed in T125. For that reason, whatever their nature may have been, later interventions did not affect the whole collection evenly. Most probably, they did not touch many discourses in the collection at all. Thus, the later interventions certainly did not produce an entirely new work, and should not be imagined in the key of categories like “retranslation.” Consequently, even if T125 does, as I contend, include substantial material dating from after 385, at the same time, it must still be heavily based upon the work of the translation group in Chang'an before Dao'an's death. My argument presents a picture of the extant T125 as a complex, mixed, or layered textual artefact – not as a post-Dao'an work, root and branch.
With these caveats, my central conclusion, that T125 was modified in some manner probably impinging upon content after the initial translation of EĀ in 384–385, implies some consequences for our treatment of the collection in scholarly analysis. The collection sports an impressive array of anomalies, and the anomalies cluster in some sutras more than others. Some of this evidence points to alteration of the collection in China at a point later than the initial translation. These facts suffice to urge considerable caution in using the collection as evidence for anything in early or Mainstream Buddhism, especially in India – even more so, in using sutras in the later parts of the collection; and even more so again, sutras with a high concentration of the anomalous features we have surveyed. These same facts also suggest that the quest for “the” sectarian affiliation of T125, as we have received it, may be doomed, and that the quest to identify a sectarian affiliation for a hypothetical Indic Vorlage may first need to confront the difficult task of disentangling the Vorlage from subsequent accretions.
Abbreviations
- AN
Aṅguttara-nikāya
- ARIRIAB
Annual Report of The International Research Institute for Advanced Buddhology at Soka University
- BSR
Buddhist Studies Review
- Chn
Chinese
- CSZJJ
Chu sanzang ji ji 出三藏記集 T2145
- DĀ
Dīrghāgama 長阿含經 T1
- Dhr
*Dharmarakṣa 竺法護
- Divy
Divyāvadāna
- DN
Dīgha-nikāya
- DZDL
see MPPU
- EĀ
Ekottarikāgama (to be distinguished from T125)
- IBK
Indogaku Bukkyōgaku kenkyū 印度學佛教學研究
- Ja
Jātaka
- JIABS
Journal of the International Association of Buddhist Studies
- JLYX
Jing lü yi xiang 經律異相 T2121
- MĀ
Madhyamāgama 中阿含經 T26
- MN
Majjhima-nikāya
- MPPU/DZDL
*Mahāprajñāpāramitopadeśa 大智度論 T1509
- MSV
Mūlasarvāstivāda Vinaya
- Mv
Mahāvastu
- SĀ
Saṃyuktāgama (usu. 雜阿含經 T99; where noted, 別譯雜阿含經 T100)
- Skt
Sanskrit
- SN
Saṃyutta-nikāya
- T
Taishō shinshū daizōkyō 大正新脩大藏經
- Th
Theragāthā
- Tib
Tibetan
- Ud
Udānavarga 出曜經 T212
- ZFn
Zhu Fonian 竺佛念
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479591424000366.
Acknowledgements
I (Radich) here record my gratitude for helpful feedback on earlier drafts of this work to Venerable Anālayo, Norihisa Baba, Sharon Chi, Rafal Felbur, Ikuma Hiromitsu, Sangyop Lee, Antonello Palumbo, Yamabe Nobuyoshi, and an anonymous reviewer for the International Journal of Asian Studies. This research was conducted during a one-year stay as Visiting Professor at the Institute for Advanced Studies on Asia at Tokyo University, and I thank the Institute for their generous support.
Open access
