According to the Sinhalese chronicles, the Pali canon was written down in the reign of King Vaṭṭagamiṇi (29-17 B.C.).[1] It has been generally accepted, therefore, that the canon contains information about the early history of Indian Buddhism, from the time of the Buddha (c.484-404 B.C.) until the end of the first century B.C. [2] That the canonical texts are a record of the period of Buddhism before they were written down in Sri Lanka seems to be confirmed by the fact that their language, Pali, is north Indian in origin. Thus the Pali canon shows 'no certain evidence for any substantial Sinhalese additions ... after its arrival in Ceylon.' [3] If the language of the Pali canon is north Indian in origin, and without substantial Sinhalese additions, it is likely that the canon was composed somewhere in north India before its introduction to Sri Lanka, and is therefore a source for the period of Buddhism in northern India before this. The Sinhalese chronicles state that the canon was brought to Sri Lanka by Mahinda during the reign of Aśoka, implying that it predates the middle of the third century B.C. [4] According to this history, the Pali canon, particularly the Vinaya and Sutta portions, is a reliable source for the early history of Indian Buddhism in the period before Aśoka. [5]

This version of events is not accepted by all, however. Gregory Schopen in particular has argued against the view that the canonical texts can be taken as accurate historical sources for the earliest period:

Scholars of Indian Buddhism have taken canonical monastic rules and formal literary descriptions of the monastic ideal preserved in very late manuscripts and treated them as if they were accurate reflections of the religious life and career of actual practising Buddhist monks in early India. [6]

This point of view has two aspects to it. On the one hand, normative religious literature must not be taken at face value, as if it contains evidence of real historical events. As Schopen puts it:

Even the most artless formal narrative text has a purpose, and that in "scriptural" texts, especially in India, that purpose is almost never "historical" in our sense of the term. [7] On the other hand, Schopen doubts that texts preserved in 'very late manuscripts' contain accurate historical evidence -- he wishes us to believe that the canonical texts cannot be taken as evidence for the period before the fifth century A.D.:

We know, and have known for some time, that the Pali canon as we have it -- and it is generally conceded to be our oldest source -- cannot be taken back further than the last quarter of the first century B.C.E, the date of the Alu-vihara redaction, the earliest redaction that we can have some knowledge of, and that -- for a critical history -- it can serve, at the very most, only as a source for the Buddhism of this period. But we also know that even this is problematic since, as Malalasekera has pointed out: "...how far the Tipiṭaka and its commentaries reduced to writing at Alu-vihara resembled them as they have come down to us now, no one can say." In fact, it is not until the time of the commentaries of Buddhaghosa, Dhammapala, and others -- that is to say, the fifth to sixth centuries C.E. -- that we can know anything definite about the actual contents of this canon. [8]

A central theme running through Scopen's work is his claim that we cannot know anything for sure about Indian Buddhism from its texts that were redacted in the fifth century A.D. (for the Pali canon), or the fourth century A.D. (approximately, for the canonical material of various sects preserved in Chinese translations). Consequently, Schopen believes that the only way we can find out anything about Buddhism before this time is through accurately dated epigraphical and archaeological material. It is clear from Schopen's work that this evidence has not been given the attention it deserves; it is vitally important to study the material remains, which tell us something concrete about what Buddhists were doing at particular places in particular times. But does this mean that we should concentrate exclusively on the material remains? Should we throw out the texts, or merely allow their evidence to be restricted and subordinated to the material evidence? The impression given by Schopen's work is that the study of early Buddhism can only progress by subordinating the literary evidence to the material evidence, an approach which seems to have become standard in some quarters. But before we consign ourselves to a radical reorientation in the study of early Buddhism, we should critically examine some of the presuppositions of this approach. There seem to be three questions of importance here:

1) How old are the canonical texts?

2) Are the canonical texts purely normative, or do they include descriptive material which can be used to reconstruct historical events?

3) And finally, how much importance is to be assigned to the epigraphical and archaeological evidence?

I radically disagree with Schopen's answers to each of these questions. In what follows, I hope to show why Schopen's views are untenable, and I will argue that the only way of knowing anything about early Buddhism is through its texts. I will begin with the last point first: it seems to me that the worth of the epigraphical and archaeological evidence has been overstated by Schopen. This is not to deny its great importance for the study of Indian Buddhism -- without it, the historian is fumbling in the dark, and his conclusions will lack verisimilitude. Be that as it may, the material evidence has its own limitations, and the fact is that it does not tell us that much about the thought and practices of Buddhists in ancient India. So although Schopen has used this evidence to draw attention to hitherto neglected aspects of Indian Buddhism (e.g. that monks and nuns probably instigated the cult of the image, or that monks and nuns were involved in the stupa cult from the earliest times), he does not acknowledge the fact that this does not tell us very much about Indian Buddhism as it was practised. It does not allow us to probe very far into the beliefs and practices of Buddhist monks and nuns in India; its content is limited, much more limited than the content of the early texts, which seem to me to contain a wealth of information on the diverse beliefs current in early Buddhism.

According to Schopen, the epigraphical material ' tells us what a fairly large number of Indian Buddhists actually did, as opposed to what -- according to our literary sources -- they might or should have done.'9 What exactly Schopen has in mind when he says 'a fairly large number of Indian Buddhists' is unclear, but certainly misleading: the relevant inscriptions number only a few thousand, which is evidence, surely, for the activity of a small minority of monks and nuns. They can hardly be taken as indicative of the activity of the Buddhist populace at large -- just over a couple of thousand inscriptions does not, to my mind, represent a large number of Indian Buddhists, considering that this must have been a tiny fraction of the number of Indian Buddhists from about 400 B.C. to 500 A.D.

In other words, there is a tendency in Schopen's work to make generalisations about Indian Buddhism based on a very small amount of evidence. Even if the generalisations were true in every respect, it would only reveal the historical reality of a tiny part of Indian Buddhism. Perhaps if there were epigraphical evidence representing every Buddhist who existed in ancient India, it would be similar to the evidence of the extant inscriptions. But we cannot presume what is not there. For all we know, the inscriptions might represent only a small minority of the ancient Indian Saṅgha, the minority who had personal wealth and who could endow Buddhist institutions in different ways.

In this situation, we should not underestimate the worth of the textual evidence, even if its antiquity cannot be established accurately. For example, Schopen records that two inscriptions at Mathura record the donations of monks who are called prahaṇika-s, 'practisers of meditation'.[10] But without consulting the evidence of the Pali canon for the word padhana or the Buddhist Sanskrit evidence for the word pradhana/prahaṇa (or variations on them), we would have absolutely no idea what the term signified for the two monks, and why they used it. The fact is that the texts are indispensable: the literary evidence, even if only normative, and even if it was periodically revised until the rather late redactions, is most certainly a useful record, not to be used as subsidiary to the material evidence, as Schopen believes, but in tandem with it, so that the two sorts of evidence are used equally. In short, if the inscriptions are to have any significance for the study of early Indian Buddhism, they must be considered alongside the canonical evidence, as has been argued by Hallisey:

It will only be after we have learned to combine our interest in "what really happened" with a sensitivity to the changing thought-worlds of the Theravada that we will begin to discern the historical reality behind the literary and archaeological traces of ancient Buddhist monasticism.[11]

It seems to me that Schopen's work is most convincing when he follows this method, and uses the literary, epigraphical and archaeological sources equally, [12] instead of just dismissing the literary evidence out of hand. [13] Unfortunately, in his eagerness to point out that the studies of previous generations of Buddhist scholars were one-sided, Schopen has created another one-sided version of history. What is needed is a balanced approach that gives both sets of evidence, the literary and material, their due worth. But what is the worth of the literary evidence? This brings me to two of the questions posed above, viz. the age and nature of the canonical texts. Schopen's position on these two points is quite clear, as we have seen, although it is strange that he does not give any evidence to support his view that the narrative Buddhist literature is 'almost never historical', as if this were a self-evident fact. As for his point that we cannot know if the canonical material is old, he attempts to demonstrate this by claiming that the general method of higher criticism -- the method which is often used to prove the antiquity of canonical texts -- is inapplicable. He sums up this method of higher criticism as follows: 'If all known sectarian versions of a text or passage agree, that text or passage must be very old; that is, it must come from a presectarian stage of the tradition.' [14]

The alternative explanation of the agreement of 'all known sectarian versions of a text or passage' is that the agreement was produced by the different sects sharing literature at a later date. It is this hypothesis which Schopen attempts to prove by showing that the similar versions of the story of the stupa of Kaśyapa at Toyika, found in Mahasaṅghika, Mahiśasaka, Dharmaguptaka and Theravadin texts, are later than versions found in the Mulasarvastivadin Vinaya and in the Divyavadana. [15] The former group of texts claim that the Buddha manifested a stupa momentarily, after which a stupa was built (by monks) or appeared. The version of the story in the Mulasarvastivadin Vinaya and in the Divyavadana, however, is described by Schopen as follows: 'Firstly, it has none of the various subplots found in the other versions -- a fairly sure sign of priority -- and, second, it knows absolutely nothing about a stupa at Toyika or its construction.' [16] Schopen's main argument then is that the story in the Mulasarvastivadin Vinaya and the Divyavadana is earlier because it does not mention a stupa: 'This version, in short, reflects a tradition -- apparently later revised -- that only knew a form of the relic cult in which the stupa did not yet have a part.' [17]

The first thing which I find odd about Schopen's assessment of this story is his claim that, on the basis of the evidence in the Mulasarvastivadin Vinaya and Divyavadana, there was a form of the relic cult that did not include the stupa. The narratives in these texts mention caitya-s, and although Schopen states that this term has nothing to do with stupa-s, this is not at all clear. In his article 'The Stupa Cult and the Extant Pali Vinaya', [18] he has in fact argued that in the Pali literature, the word cetiya is equivalent to stupa. [19] It could easily be the case that the word has the same meaning in the Mulasarvastivadin Vinaya and the Divyavadana. But even if not, are we to accept a form of relic worship without a stupa? If we take the canonical texts seriously, it is hard to imagine that this could ever have been the case. The Mahaparinibbana Sutta, for example, states that the Buddha's relics are to be contained in a stupa, [20] which suggests that the stupa goes back to the very beginning of Buddhism. The stupa was certainly a feature of Buddhism by the time of Aśoka, who records in his Nigali Sagar Pillar Edict that twenty years into his rule, he had the thuba of Konakamana doubled in size. [21] Moreover, Aśoka seems to have known a portion of the text found in the Sanskrit version of the Mahaparinirvaṇa Sutra -- in his Rummindei inscription, he records that he visited Lumbini and worshipped there saying 'Here the Blessed One was born', [22] which corresponds to the Sanskrit version of the Mahaparinirvaṇa Sutra (41.8: iha bhagavaṭ jataḥ). [23] This part of the text is close to the parts in the Pali and Sanskrit versions which mention stupa-s, and so it seems natural to conclude that stupa worship was not only a part of Buddhism at this date, but also that it was mentioned in canonical Buddhist texts at this point. This is an important point, for according to the most plausible theory of sect formation (the theory proposed by Frauwallner), some of the Sthavira sects formed as a result of the Aśokan missions in 250 B.C. (see below p.11ff). If the Aśokan evidence suggests that by about this time the stupa was a feature of Buddhism and its texts, a presectarian period that did not relate relic worship to the construction of stupa-s is hardly plausible. It seems that there are no obvious reasons for taking the story in the Mulasarvastivadin Vinaya and Divyavadana to be older than the versions in the Mahasaṅghika, Mahiśasaka, Dharmaguptaka and Theravadin texts.

However, even if Schopen has got it right and his argument is valid, it actually shows that the Pali canon was closed to material received from other sects. What Schopen fails to mention is that the method of higher criticism used to establish old strata in the Buddhist literature usually compares the canonical literature of different sects: he is reluctant to note that the Pali version of the story of the stupa of Kaśyapa at Toyika is found in the Dhammapada-aṭṭhakatha -- this information is conveniently confined to footnote 28. This means that if Schopen is correct, it seems then that whereas some of the other sects periodically shared literature and changed their canonical material in the sectarian period, the Theravadins of Sri Lanka did not: they confined the received material to non-canonical books. It seems that Schopen might have inadvertently proved that the Pali canon was relatively closed after its redaction at an early date. This depends of course on whether or not he has interpreted the different versions of the story about the stupa at Toyika correctly, and this is far from clear. A thorough study of the different versions of the story is surely necessary. However, it is worth taking a short digression to show that another inadvertent proof of the antiquity of at least the Suttapiṭaka is given by Schopen in the very same article ('Two Problems in the History of Indian Buddhism: The Layman/Monk Distinction and the Doctrines of the Transference of Merit').

In this article, Schopen shows that the belief in the transference of merit was widespread in India from the third century B.C. onwards (pp.34-42). Thus, a late Mauryan/early Śuṅga inscription from Pauni, a few inscriptions from third century B.C. Sri Lanka, a singular early inscription from Bharhut, as well as a significant number of later Hinayana inscriptions from various parts of India all record the idea. If the idea was a standard belief of Buddhists in early times, even in Sri Lanka, and if the Suttapiṭaka was not finally closed until the Alu-vihara recension in the fifth century A.D., then it is reasonable to expect that it should be well attested in the Suttapiṭaka. But this is not the case -- although much is said on the subject of meritorious activity, the idea of the transference of merit is found in only two separate occurrences in the four principle Nikaya-s. [24] How can we explain the fact that Theravada Buddhists of Sri Lanka did not compose more texts which included the idea of merit transference? There can only be one answer -- the texts were closed in an earlier period, when the belief was marginal in Buddhist circles. This is surely the only answer to the problem. Even if this does not definitely prove that the canon was closed at an earlier date, the fact that the ancient guardians of the Suttapiṭaka did not compose texts on the transference of merit shows that they must have had some idea of canonical orthodoxy, which means that the canon must have been relatively fixed. By attempting to show that the canonical texts are not reliably old, and that we must turn to the epigraphic evidence to gain any idea about the historical reality of ancient Indian Buddhism, Schopen has inadvertently shown that some collections of texts must indeed be old and contain evidence for the period before most of the inscriptions.

Exactly the same fact emerges from Schopen's article 'The Stupa Cult and the Extant Pali Vinaya'. He attempted to show that 'the total absence of rules regarding stupas in the Pali Vinaya would seem to make sense only if they had been systematically removed', [25] meaning that the Pali canon was altered 'at a comparatively recent date', after the supposed recensions made in the first century B.C. and the fifth century A.D.

This argument is based upon the fact that all the other extant Vinayas include rules concerning the construction and cult of the stupa, whereas the Pali Vinaya does not. There are two possible explanations for this fact. Either it is because the Pali Vinaya was closed before these rules were formulated, or it is because these sections were written out of the Pali Vinaya, accidentally or on purpose; Schopen chooses the latter option. But Gombrich and Hallisey have shown that this interpretation is based on a mistranslation of the twelfth century Sinhalese inscription, the Maha-Parakramabahu Katikavata. [26] It therefore seems likely that the other solution to the problem is correct -- the Pali Vinaya was closed before this section was composed and added to the other Vinayas. Gombrich notes: 'One does not have to posit that it received no further additions after the first century B.C., merely that the Pali tradition had left the mainstream and naturally failed to record later developments on the Indian mainland.' [27] But because we know that the Pali tradition remained in contact with the Indian mainstream (it received texts from north India after the first century B.C.), I think it more likely that no further additions were made after the first century B.C.

The points Schopen makes about the post-canonical sharing of literature, the transference of merit, and the Pali Vinaya's evidence on stupa-s, if correctly interpreted, suggest that the Pali canon was relatively fixed from at least the first century B.C. onwards. This is despite the fact that the Pali tradition remained in contact with other Buddhist sects in India, as has been noted already by scholars such as Oldenberg and Norman. According to Norman, 'some of the best known stories in Buddhism ... are known in the Theravadin tradition only in the commentaries, although they are found in texts which are regarded as canonical in other traditions.' [28] Such stories must have reached Sri Lanka before Buddhaghosa, for he includes them in his commentaries. But why were they not inserted into the canon? Norman thinks that it was because 'at least the Vinaya- and Sutta-piṭaka had been closed at an earlier date.' [29] Norman has also pointed out that certain Pali works for which a North Indian origin is supposed, such as the Milindapaṭha, the Peṭakopadesa and the Nettipakaraṇa, are highly respected by the commentators but are not given canonical status by them. They even contain 'a number of verses and other utterances ascribed to the Buddha and various eminent theras, which are not found in the canon ... There was no attempt made to add such verses to the canon, even though it would have been a simple matter to insert them into the Dhammapada or the Theragatha.' [30] The point that the Pali tradition received literature from other sects but excluded it from the canon had been made already by Oldenberg in 1879: 'These additions are by no means altogether unknown to the Singhalese church, but they have been there placed in the Aṭṭhakathas, so that the text of the Tipiṭaka, as preserved in Ceylon, has remained free from them.' [31] This suggests that they arrived in Sri Lanka 'after the closure of the Canon'.

If we remind ourselves of Norman's point that the Pali canon contains no definite evidence for a substantial amount of Sinhalese prakrit (see above p.1), it seems quite clear that after the Tipiṭaka was written down in the first century B.C., it was not substantially altered, at least in content, and as such, it must have been very similar to the extant Pali Canon. This means that the Suttapiṭaka in existence today can be taken as an accurate record of Buddhist thought from the time of the Buddha (c. 484-404 B.C.) until the first century B.C. at the latest . [32] This is significantly older than Schopen is willing to acknowledge, but the terminus ante quem can be pushed back even further; it depends upon the date when the Pali texts reached Sri Lanka, i.e. the date at which the sectarian period began.

According to Schopen, 'we do not actually know when the sectarian period began.' [33] To support this view he cites Bareau's work which points out that the Buddhist sects all give different dates for the schisms. [34] But he does not make any mention of what is probably the most convincing work on the subject. Erich Frauwallner, in The Earliest Vinaya and the Beginnings of Buddhist Literature, used a mixture of epigraphical and literary sources to argue that some of the Sthavira sects owed their origination to the missions said to have taken place in the reign of Aśoka, c.250 B.C. Firstly, there are records in the Sinhalese chronicles (and the Samantapasadika) of a series of Buddhist missions which went out to different parts of India and neighbouring kingdoms in the reign of Aśoka. Although these Pali accounts as we have them do not seem plausible -- they might have been embellished to trump up the monastic lineage from which the Pali texts issued [35] -- the missions are confirmed by the inscriptions found on a couple of reliquaries unearthed in the ancient Buddhist centre of Vidiśa. According to Frauwallner, these reliquaries contain the remains of the Hemavata masters Dudubhisara, Majjhima and Kassapagotta, names which he identified with the missionaries Durabhisara, Majjhima and Kassapagotta, all of whom travelled to the Himavanta according to the chronicles. [36] Willis has recently pointed out that Frauwallner misread this evidence slightly by mistaking the relics of Gotiputa, heir of Dudubhisara, for Dudubhisara himself, [37] but at the same time he has argued that all five names on the two different reliquaries correspond to the five names in the chronicles. [38] It strongly implies that the missionaries to the Himavanta hailed from Vidiśa and that some of their relics were returned there some time after their death. The chronicles also record that Mahinda's mother was from Vidiśa, and that he stayed there before journeying to Sri Lanka. [39] This is an impressive correspondence of epigraphical and literary evidence, and it makes it almost certain that the account of the missions in the Pali chronicles contains some historical truth.

Frauwallner equated this epigraphic and literary evidence with further epigraphic evidence from Aśoka's thirteenth Rock Inscription: on pp.15-17 of The Earliest Vinaya, he noted that the areas mentioned in this edict, to which he despatched emissaries, correspond to the areas of missionary activity mentioned in the Pali chronicles. Both sources, according to him, mention the North-West, West and South but omit the East, and he comments 'This is certainly no freak chance.' Lamotte's table (p.302) shows at least a superficial agreement between the places mentioned in both sources, but Gombrich is probably correct in commenting: 'The geographical identifications are too uncertain to help us'. [40] With the geographical identifications uncertain, Lamotte was sceptical of the notion that there was one concerted missionary effort in Aśokan times. He argued that the Buddhists were natural missionaries and would have spread Buddhism throughout India from the beginning. [41] Thus he concluded his study of the early Buddhist missions by stating 'Whatever might have been said, Aśoka was not directly involved in Buddhist propaganda.' [42] Gombrich, on the other hand, agrees with Frauwallner, and notes:

While Lamotte is right to point out that some of the areas visited, notably Kashmir, had Buddhists already, that does not disprove that missions could not be sent there. The chroniclers, as so often happens, had no interest in recording a gradual and undramatic process, and allowed history to crystallize into clear-cut episodes which could be endowed with edifying overtones; but this over-simplification does not prove that clear-cut events never occurred.[43]

The notion that there was a clear-cut missionary episode in the spread of Buddhism across India seems to be confirmed by the epigraphical record. L. S. Cousins has surveyed the references to the sects in inscriptions (pp.148-51), and noted that the related Vibhajjavadin sects (the Vibhajjavadin-s made up a subset of the ancient Sthaviras) were most widespread of all Buddhist sects in the first few centuries C.E. On the other hand, the other sects were distributed randomly across India. This is exactly what is to be expected if there was a gradual diffusion of Buddhism throughout India, as well as a concerted missionary effort by one ancient monastic community, which thereafter separated into separate sects due to the geographical dispersal. Cousins comments on the tradition of the Buddhist missions as follows:

It seems clear that whatever the traditions about these missions may or may not tell us about events in the third or second century BCE, they do certainly correspond to what we know of the geographical spread of the schools early in the first millennium CE. They must then have some historical basis. Vibhajjavadins really were the school predominant in Ceylon and Gandhara at an early date, as well as being present, if not predominant, in other parts of Central Asia, China, South India and South-East Asia by around the turn of the third century CE at the latest. No other school has a comparable spread at this date.[44]

It seems then that there is no reason to doubt that there was some sort of mission in the third century B.C. which set out from Vidiśa to the far North-West, West and South of India. Frauwallner thought that this missionary activity founded the Sarvastivadin sect in the North-West, as a result of Majjhantika's mission to Kaśmir and Gandhara,[45] whereas Cousins considers only the Vibhajjavada sects in the North-West and South. Was the Sarvastivadin sect of the North-West produced by a missionary effort that otherwise seems to have produced only Vibhajjavadin sects? This is certainly possible. Frauwallner made it quite clear that the formation of distinct communities ought to be distinguished from schools of thought: 'from the first we have stressed the principle that the foundation of communities and the rise of dogmatic schools are two quite separate things.' [46] This led him to conclude that the dogmatic affiliation of the Mulasarvastivadin and Sarvastivadin sects came later than the original foundation of these two as monastic communities.[47] It is possible then that different dogmatic affiliations could have been produced in the sects founded by missionary activity, and that the dogmatic affiliation to sarvastivada ideas by the community that came to be known as the 'Sarvastivadin' sect came about later. This seems to be shown by the fact that the literature of the Sarvastivadin sect is in many regards similar to the literature of the other Vibhajjavadin sects.

In the beginning of The Earliest Vinaya, Frauwallner notes that the Skandhaka section of the Sarvastivadin, Dharmaguptaka, Mahiśasaka and Pali Vinaya-s are 'strikingly close'; the Skandhaka of the Kaśyapiya school is not considered because it has not survived. [48] According to Frauwallner then, the Sarvastivadin Skandhaka is closer to the Skandhaka-s of sects known to be Vibhajjavadin in affiliation than it is to the Mulasarvastivadin Skandhaka. Elsewhere, Frauwallner has noted that the Sarvastivadin Abhidharma contains much that 'was held in common with the Pali school.' [49] And the Śariputrabhidharma, which according to Frauwallner is a Dharmaguptaka text, is also a development of the same material inherited by the Sarvastivadin and Pali schools. [50] This is again in contrast with the Abhidharma of the Mulasarvastivadins, which according to Frauwallner 'possessed only one Matṛka.' [51] The canonical literature of the sect in the North-West that came to be known as the Sarvastivadin-s is therefore closer to the Vibhajjavadin sects, particularly the literature of the Pali tradition.

It seems likely that all these sects share a common antecedent, which we can think of as the ancient Sthavira community of Vidiśa. Nevertheless, it is striking that only one of the sects produced by the missions adopted sarvastivada ideology, whereas the others seem to have been affiliated to the vibhajjavada. I think the best explanation of the evidence is that this sarvastivadin development must have occurred later on within the community founded by Majjhantika in Kaśmir-Gandhara. Originally, the Abhidharma literature of the missionary community was less fixed than its Sutra and Vinaya sections. It allowed considerable room for development. In Aśokan times the dogmatic outlook of the missionary community was vibhajjavada, but at a later date, sarvastivada ideology came to dominate in Kaśmir. Incidentally, the Mulasarvastivadin school, originally from Mathura according to Frauwallner, came to exist in the North-West and claimed that it originated from the mission of Madhyantika.[52] It is plausible to think that this claim -- almost certainly an interpolation into the Mulasarvastivadin Vinaya, as Frauwallner has shown -- was taken from the old Sarvastivadin-s of the North-West, and used as part of the Mulasarvastivadin strategy of claiming supremacy in the North-West. This claim would only have been borrowed if it was thought to be true, so it seems that yet another piece of evidence supports the accuracy of the Theravadin tradition of the missions.

The evidence for a Sthavira mission taking place in the third century B.C., probably from a school that was vibhajjavadin in the dogmatic sense, is very good. But was this mission related to Aśoka? Despite Lamotte's doubts, I think that the Aśokan inscriptions show that this must have been the case. The confusion on this point seems to have been caused by a failure of previous scholars, especially Lamotte, to distinguish Aśoka's references to his 'Dharma-ministers' (dhamma-mahamatta) from the evidence contained in the thirteenth Rock Edict. Lamotte's table on the sources of the missions (p.302) sums up the evidence, presented on the previous page, of the second Rock Edict, the fifth Rock Edict and the thirteenth Rock Edict. But RE II has nothing to do with missionary activity -- nor does RE V, which mentions the dhamma-mahamatta. In fact every mention of the dhamma-mahamatta limits them to Aśoka's Kingdom, and so they should have been, for according to the inscriptions they were involved in all sorts of activities which might be called 'social welfare', and which cannot have been carried out in other kingdoms.[53] Socially beneficial activity in other kingdoms is mentioned in RE II, which states that Aśoka provided medicines and medical herbs for men and cattle, and had wells dug, [54] but this falls well short of the activities of the dhamma-mahamatta. Whether or not Aśoka really carried out such beneficial acts in neighbouring kingdoms -- it might have been an exaggeration of his own righteous endeavours -- this activity can hardly have been carried out by his dhamma-mahamatta. In the same way, any missionary activity initiated by Aśoka cannot have been undertaken by his dhammamahamatta. This brings us to RE XIII: Aśoka tells us that he has achieved a 'Dharmavictory' (dhamma-vijaya) in his own kingdom as well as others, and even in places where his 'envoys' (duta-s) have not gone. [55] From this we can see that this victory must have been achieved by his envoys -- duta-s, not dhamma-mahamatta. Who were these envoys then, sent by Aśoka to the border areas of this kingdom, as well as to neighbouring kingdoms, through which he attained his victory of dhamma?

The obvious answer is that they were the people responsible for taking medicines and medicinal herbs to other Kingdoms, and for having wells dug there, mentioned in RE II. Supporting this idea is the fact that the areas outside Aśoka's Kingdom mentioned in RE XIII and RE II are almost identical: in RE XIII the kingdoms mentioned are those of Aṃtiyoka, the Yonaraja, and his four neighbours in the North-West, as well as the Choḍa-s, Paṃda-s and Taṃbapaṇṇiya-s in the South; in RE II, the Satiyaputra-s and Keraḍaputra-s are added to the list of southern kingdoms, and the neighbours of Aṃtiyoka are not named. This is almost an identical correspondence. Nevertheless, it is problematic think that the duta-s mentioned in RE XIII were merely carriers of medicines and supervisors of well-digging. After all, Aśoka says that through them he has achieved his dhamma-vijaya: can a 'Dharma-victory' have been achieved by the implementation of some social policies? In other words, would socially beneficial acts undertaken by Aśoka in his kingdom and elsewhere have induced him to claim that he had achieved a Dharma-victory? One might think that this is possible, because the sort of dhamma promoted by the Dharma-ministers was exactly this sort of socially beneficial action. But against this idea is the fact that in PE VII at Delhi-Topra, the same sort of socially beneficial deeds are declared to be 'of little consequence'. For with various comforts have the people been blessed both by former kings and by myself.' [56] After this declaration, Aśoka states that he has merely provided material needs so that the people will conform to the practice of dhamma. [57] We have then a distinction between Aśoka's ideal of dhamma and socially beneficial policies which might be called dhamma. Which of the two meanings of dhamma did Aśoka mean by the word in the compound dhammavijaya? I find it unlikely that Aśoka would have proclaimed a dhamma-vijaya because of his social policies, which in PE VII he claims are of little value. It is much more likely that when Aśoka spoke of his dhamma-vijaya, he had in mind a victory of dhamma in its higher meaning of a set of ethical practices and attitudes. In the Delhi-Topra edict, this ideal of dhamma is outlined as follows:

- Thereby they have made progress and will (be made to) progress in obedience to mother and father, in obedience to elders, in courtesy to the aged, in obedience to Brahmaṇas and Śramaṇas, to the poor and distressed, (and) even to slaves and servants. [58]

This ideal of the dhamma is outlined in more or less the same fashion by Aśoka in RE III, IV, IX, XI, and crucially, in RE XIII, the edict in which he claims his dhammavijaya. [59] This 'victory' was the spread of ideals such as respect to śramaṇa-s and brahmaṇa-s, obedience to mother and father, courtesy to slaves and servants etc. If the envoys who took these ideals to the distant corners of Aśoka's kingdom and beyond were not Dharma-ministers, who were they? Are we to believe that Aśoka had a class of officials who went out and taught what are essentially religious ideals? This is hardly likely. The more plausible answer is that the duta-s included the professional religious men and women to whom Aśoka was partial, i.e. the Buddhists. In other words, it is likely that the envoys who spread Aśoka's ideals included Buddhist monks and nuns. [60] There is even some indication in the Sinhalese chronicles that is indeed what happened. In chapter XI, the Mahavaṃsa describes how envoys were sent by Aśoka to King Devanaṃpiyatissa of Laṅka:

- The Lord of Men [Aśoka], having given a palm-leaf message (paṇṇakaraṃ) at the appropriate time for his friend [Devanaṃpiyatissa], sent envoys (dute) and this palm-leaf message concerning the true doctrine (saddhammapaṇṇakaraṃ), [which said:]

- Saying: "Carry out the consecration of my friend once more," having honoured his friend's ministers, he despatched them. [61]

There is no mentions of Buddhist monks and nuns in the imperial embassy of duta-s, but the implication is that if there were contacts such as this between Aśoka and his neighbouring kings, then Buddhists must have been involved, or would have followed soon afterwards. This is also indicated by the Dipavaṃsa which, although including a standard account of the missions sent by Moggaliputta (at VIII.4-13), also includes three accounts of the duta-s sent to Laṅka by Aśoka, each account describing how Mahinda arrived in Laṅka soon after the envoys, without any mention of Moggaliputta. These accounts imply that the envoys paved the way for the Buddhist monks who followed. In the most elaborate account, Dip XII.8ff, after describing how Aśoka sent gifts and a request that Devanaṃpiya of Laṅka should have faith in the triple jewel, it says that the thera-s of the Asokarama requested that Mahinda establish the faith in Laṅka. [62] And at Dip XI.41 and Dip XVII.91-92, it says that Mahinda arrived in Laṅka one month after the Aśokan envoys, without any mention of Moggaliputta. [63] It seems that the author of the Sinhalese chronicles, as well as Buddhaghosa, had various sources available to them recording different versions of the mission to Sri Lanka. [64]

The version that eventually became the orthodox account was of course the one that had Moggaliputta as the organiser of the missions. But the accounts in the Dipavaṃsa that do not mention Moggaliputta seem much more plausible in the light of the evidence from RE XIII. It is clear that some of the information in the chronicles is accurate: the name of the missionary monks, for instance, as the evidence at Saṭchi indicates. But the Sthavira tradition from which the missions came could hardly have made the monks merely part of Aśoka's ministerial envoys, or even following in the wake of these envoys -- they probably felt that they had to exaggerate the prominence of their tradition with the idea that Moggaliputta sent them. [65] We can conclude that the imperial envoys (duta-s) of Aśoka, which for him had effected a dhamma-vijaya, probably did include Buddhist monks. By welcoming these envoys, and heeding Aśoka's written requests that they take refuge in the triple jewel, the neighbouring kings maintained good relations with the mighty Indian emperor, and Aśoka himself propagated dhamma.[66]

The result of this long digression into the evidence for the Aśokan missions is that it appears that Frauwallner was correct to relate the evidence of the Sinhalese chronicles and inscriptions of Vidiśa to Aśoka's RE XIII. We can therefore date the arrival of the canonical texts in Sri Lanka to the middle of the third century B.C. Frauwallner has shown, however, that the lower limit of the early Buddhist literature can be pushed back even further. He noted that the Mahasaṅghikas had a version of the old Skandhaka, meaning that it must have been composed before the schism between themselves and the Sthaviras, which certainly occurred before the Aśokan missions. The Mahasaṅghika Vinaya also includes the account of the second council at Vaiśali: the old Skandhaka must have been composed, or at least redacted, after this council, and before the schism between the Sthaviras and Mahasaṅghikas.[67] Frauwallner thought that the old Skandhaka was probably composed shortly before or after this council, [68] which Gombrich reckons that it took place about 60 years after the Buddha's death, c. 345 B.C. [69] If this is correct, it means that very sophisticated literary tracts were being composed little more than half a century after the Buddha's death.

There is great significance in these investigations for the date of texts contained in the Suttapiṭaka. According to Frauwallner:

At the time of the compilation of the old Skandhaka work ...[70] the Buddhist tradition had already reached an advanced stage of development. A collection of sacred scriptures, including Dharma and Vinaya, was already in existence. The Vinaya included the Praṭimokṣa, narratives of the type of the Vibhaṅga and much material on the monastic rules, which the Buddha was said to have communicated to his disciples. The collection of Sutra, which existed on its side, was handed down by a regular machinery of transmission, and we can ascertain a number of texts which belonged to it already in that period. [71]

One Sutra text which Frauwallner singles out is the Aṭṭhakavagga (Sn IV), for the same story mentioning it is preserved in all the extant Vinayas; it probably belonged to the old Sutra collection. It is therefore possible that much of what is found in the Suttapiṭaka is earlier than c.250 B.C., perhaps even more than 100 years older than this. If some of the material is so old, it might be possible to establish what texts go back to the very beginning of Buddhism, texts which perhaps include the substance of the Buddha's teaching, and in some cases, maybe even his words. I have no intention of going into the important but complex question of what the Buddha did or did not teach. In the following, I will address the two questions posed on p.3, but I will at least attempt to show that some of the details of the Buddha's biography, namely those which record some of his activities as a Bodhisatta, have recorded accurate historical information about events that happened in the fifth century B.C. This will show that a careful use of textual sources is the only way to know anything about Buddhism in the pre-Aśokan period, and will lead to the conclusions that, contrary to what Schopen thinks, some material in the Suttapiṭaka is historically accurate and extremely old.

Various Suttas describe the Buddha's visits to the sages Aḷara Kalama and Uddaka Ramaputta, although the source for the account is probably the Ariyapariyesana Sutta (APS, M no.26). [72] Andre Bareau has translated a Chinese Sutra that corresponds to the APS as well as an account found in the Chinese version of the Dharmaguptaka Vinaya.[73] There are also versions of the narrative in the Mahasaṅghika Mahavastu [74] and the Mulasarvastivadin Saṅghabhedavastu.[75] It seems that the account of the training under the two teachers was embedded in the pre-sectarian Buddhist tradition, that is, if one accepts the idea that corresponding parts of the sectarian literature are likely to be pre-sectarian. There is also material on the two teachers scattered throughout the Suttapiṭaka. Scholars have generally accepted Bareau's opinion that the tradition of the two teachers' instruction to the Bodhisatta was a fabrication,[76] but more recently, Zafiropulo has shown that Bareau's arguments are fallacious.[77] If we are to take the tradition seriously, as we must do in the light of Zafiropulo's comments, we must also take into consideration the fragmentary information about the two teachers that is scattered throughout the Suttapiṭaka. I hope to show that a re-evaluation of the data on the two teachers makes two things quite clear. Firstly, some of the information on the two teachers cannot have been shared at a later date -- it must reflect a presectarian tradition. And secondly, a peculiar detail in the account of the Bodhisatta's training under the teachers shows that the two men must have existed. They must have been teachers of some repute in the fifth century B.C. in northern India, teachers of meditation who probably taught the Bodhisatta.

To show the former point, I will consider the information found in various sources concerning the location of Uddaka Ramaputta. Hsṃan tsang mentions some legendary evidence that relates Udraka Ramaputta to Rajagṛha; it seems that this represents the local tradition of the Buddhists living in the area of Rajagṛha.[78] This tradition is confirmed by the account of the Bodhisatta's training in the Mahavastu, which also places Udraka Ramaputra in Rajagṛha.[79] The coincidence between these two sources might have been reached in the sectarian period. There is, however, similar evidence in the Suttapiṭaka which makes it almost certain that the tradition must be presectarian. In the Vassakara Sutta, the Brahmin Vassakara, chief minister of Magadha, is said to visit the Buddha in Rajagaha and tell him that the raja Eḷeyya has faith in the samaṇa Ramaputta; the commentary names him as Uddaka Ramaputta.[80] Vassakara also appears in the Mahaparinibbana Sutta as the chief minister of King Ajatasattu of Magadha.[81] Vassakara's connection with Rajagaha and Magadha suggest that the raja Eḷeyya was a local chieftain in Magadha, probably situated somewhere near to Rajagaha. If so, it is likely that Uddaka Ramaputta was situated in the vicinity of Rajagaha.

The coincidence of this different evidence from the Theravadin and Mahasaṅghika sources, as well as the information of Hsṃan tsang, is not to be overlooked. It is inconceivable that this correspondence was produced by a later leveling of texts, for it is entirely coincidental -- different source materials, not corresponding Suttas, state or imply the same thing. It is hardly likely that a Mahasaṅghika monk or nun gained knowledge of obscure Pali Suttas, from which he deduced that Uddaka Ramaputta must have been based in Rajagaha, and after which he managed to insert this piece of information into the biographical account in the Mahavastu. And it is even more unlikely that a Theravadin Buddhist, in the early centuries A.D., studied the Mahasaṅghika Vinaya, from which he learnt that Udraka Ramaputra was based in Rajagṛha, following which he fabricated Suttas which contained circumstantial evidence which indirectly related Ramaputta to Rajagaha. Anyone who believes this version of textual history is living in cloud-cuckooland. It is clear that the information on the geographical situation of Uddaka Ramaputta must precede the Aśokan missions, and even the schism between Sthavira-s and Mahasaṅghika-s. This implies that the biographical tradition of the training under the two teachers goes back to the very beginning of Buddhism. It surely means that accurate historical information has been preserved, and suggests that Uddaka Ramaputta was based in Rajagaha, no doubt as a famous sage of Magadha. Incidentally, it is clear that this material has no normative value whatsoever, and so rebuts Schopen's claim that 'even the most artless formal narrative text' has a normative agenda.

Another detail, found in almost all the sectarian accounts of the training under the two teachers, can hardly have been produced by a later leveling of the Buddhist literature; it occurs in the account of the training under Uddaka Ramaputta. This account is identical in almost all regards to the description of the training under Aḷara Kalama. It tells us that the Bodhisatta first of all mastered the teaching, i.e. he gained an intellectual understanding of it,[82] after which he attained the direct realisation of the sphere of 'neither perception nor non-perception' through understanding (abhiṭṭa).[83] But the account of the training under Uddaka Ramaputta makes it clear that it was not Uddaka Ramaputta who had attained the sphere of neither perception nor non-perception, but Rama, the father or spiritual teacher of Uddaka.[84] This is seen in the following exchange. The Bodhisatta is said to have contemplated that Rama (not Ramaputta) did not proclaim (pavedesi) his attainment through mere faith, but because he dwelt (vihasi) knowing and seeing himself.[85] The corresponding passage in the account of the training under Aḷara uses the same verbs in the present tense (pavedeti, viharati), indicating that Aḷara was living and Rama was dead, and that Ramaputta had not attained and realised the dhamma he taught.

The same phenomenon is found in the rest of the passage. Thus the Bodhisatta is said to have asked Ramaputta: 'The venerable Rama proclaimed (pavedesi) [his attainment], having himself realised this dhamma to what extent (kittavata)?' [86] The reply, of course, is as far as nevasaṭṭanasaṭṭayatana. The Bodhisatta is then said to have contemplated that not only did Rama have faith, energy, mindfulness, concentration and insight, but that he too possesses these virtues. And at the end of the episode, Uddaka Ramaputta is reported to have said: 'Thus the dhamma that Rama knew (aṭṭasi), that dhamma you [the Bodhisatta] know; the dhamma you know, that dhamma Rama knew.' [87] This is different from the corresponding speech that Aḷara is reported to have made to the Bodhisatta: 'Thus the dhamma I know (janami), that dhamma you know; the dhamma you know, that I know.' [88] And whereas Aḷara is willing to establish the Bodhisatta as an equal to him (samasamaṃ), so that they can lead the ascetic group together (imaṃ gaṇaṃ pariharama ti),[89] Uddaka acknowledges that the Bodhisatta is equal to Rama, not himself (iti yadiso ramo ahosi tadiso tuvaṃ), and asks the Buddha to lead the community alone (imaṃ gaṇaṃ parihara ti).[90]

The distinction between Uddaka Ramaputta and Rama is also found in the Sarvastivadin, Dharmaguptaka, and Mahasaṅghika accounts of the Bodhisattva's training.[91] Although the Saṅghabhedavastu (plus parallel Tibetan translations) and the Lalitavistara fail to distinguish Ramaputta from Rama,[92] this must be because of a later obfuscation of the tradition. Exactly the same mistake has been made by I. B. Horner, the PTS translator of the Majjhima Nikaya, who has been duped, by the repetitive oral style, into believing that the accounts of the training under Aḷara and Uddaka must be the same apart from the difference between the names of the two men and their meditative attainments.[93]

It hardly needs to be pointed out that there is no need to trouble over these details in an oral tradition where adjacent passages are often composed in exactly the same way, one passage frequently being a verbatim repetition of the previous one with a minor change of one or two words. The tendency for reciters of this autobiographical episode would have been to make the two accounts identical bar the substitution of Uddaka's name for Aḷara's. A conscious effort has been made to distinguish Uddaka Ramaputta from Rama, and not to let the repetitive oral style interfere with this. This effort must surely go back to the beginning of the pre-sectarian tradition of composing biographical Suttas, and the distinction can only be explained if Rama and Ramaputta were two different people. Otherwise, it is part of an elaborate hoax, and there is no reason for such a hoax.

Bareau maintained that the correspondence between the two descriptions of the training under each of the teachers proved their artificial (i.e. unhistorical) nature.[94] But repetition is normal in Pali oral literature. And it seems that the two parallel accounts, having preserved the important distinction between Ramaputta and Rama, rather than leaving an impression of 'contrivance', have preserved valuable historical information. The conclusion is that the three men were real.[95] It is hardly likely that Buddhists got together a few hundred years after the Buddha's death and decided to make up the idea that Rama and not Ramaputta had attained the state of neither perception nor non- perception, and then had such an influence that the idea found its way into recensions of texts being made in regions as far apart as central Asia and Sri Lanka. The idea must have been in the Buddhist tradition from the beginning, and can only be explained as an attempt to remember an historical fact. There is no other sensible explanation. It is also worth pointing out that if this biographical material is so old and really does represent an attempt to record historical facts, then it means that this portion of the Bodhisatta's biography is most likely to be true. It is likely that the Bodhisatta really was taught by Aḷara Kalama and Uddaka Ramaputta.

And finally, I attempted to show that some of the information preserved in the literature of the various Buddhist sects shows that historical information about events occurring in the fifth century B.C. has been accurately preserved. The corresponding pieces of textual material found in the canons of the different sects (especially the literature of the Pali school, which was more isolated than the others) probably go back to pre-sectarian times. It is unlikely that these correspondences could have been produced by an endeavour undertaken in the sectarian period, for such an endeavour would have required organisation on a scale which was simply inconceivable in the ancient world. We can conclude that a careful examination of the early Buddhist literature can reveal aspects of the pre-Aśokan history of Indian Buddhism. The claim that we cannot know anything about early Indian Buddhism because all the manuscripts are late is vacuous, and made, I assume, by those who have not studied the textual material properly.

-ooOoo-

Étienne Lamotte, however, proposed a different solution: he dismissed Frauwallner's theory about the difference between Sarvastivadin and Mulasarvastivadin, by claiming, without presenting any corroborating evidence, that the Mulasarvastivadin Vinaya was simply the version of the Sarvastivadin Vinaya completed at a later date in Kaśmir (p.178). This explanation, however, leaves too many questions unanswered. For example, if the Mulasarvastivadin Vinaya was a later recension of an earlier Sarvastivadin Vinaya, then why is it so different? And why did the sect who revised the work change their name from Sarvastivadin to 'Mulasarvastivadin'? The only answer to the last objection to Lamotte's thesis is that the name was changed to 'Mulasarvastivadin' by the reforming sect who expanded the old Sarvastivadin Vinaya, and who thus used the prefix mula in order to differentiate themselves from the old Sarvastivadin-s and create a new sect, i.e. an explanation which corresponds to the second of my proposed solutions above. But there is no clear evidence for the theory that the Mulasarvastivadin Vinaya is an expansion of the Sarvastivadin Vinaya and that the Mulasarvastivadin-s were an offshoot of the Sarvastivadin-s.[96]

We must therefore look for a different answer to the problem, an answer along the lines of my proposed first solution above, i.e. that two Sarvastivadin groups came into contact and had a dispute. Such an explanation would simply be a reworking of Frauwallner's hypothesis -- he proposed two different communitites, one from Kaśmir and one from Mathura, both of whom came to accept sarvastivadin ideas, but he did not state that they had a dispute.[97] I am proposing that the Mathura school moved to Kaśmir and disputed with the existing community there, and I think we can detect such a dispute in the Chinese works mentioned by Lamotte pp.174- 75. First of all, however, I will show that these works cited by Lamotte support Frauwallner's theory.

Lamotte did not offer any explanation of the peculiar fact that Kumarajiva, the fifth century translator of the Mahaprajṭaparamitopadeśa, knew of the existence of two different Vinaya-s, one from Mathura and one from Kaśmir. According to Kumarajiva, the Vinaya of Kaśmir contained 10 sections, and we can deduce that this was the Sarvastivadin Vinaya, for he was himself from Kaśmir and translated the Sarvastivadin Vinaya into Chinese in 404-05 A.D. (Lamotte p.168). Moreover, the Sarvastivadin Vinaya was also known as the 'Vinaya in ten sections' (Daśadhyaya, Lamotte p.168). This means that it is likely that what he calls the Vinaya of Mathura in 80 sections was the Mulasarvastivadin Vinaya, that is, if the information of Kumarajiva and the two other Chinese authors mentioned by Lamotte (p.175) definitely concerns the Sarvastivadin and Mulasarvastivadin traditions of Kaśmir in the fifth century A.D. All three Chinese works mentioned by Lamotte (pp.174-75) knew that the original Vinaya (of Upali/Kaśyapa) consisted of 80 sections. Sêng yu and Hui Chao related this Vinaya to the patriarchal lineage ending in Upagupta, the fifth patriarch of the Buddhist community of Mathura.[98] Kumarajiva went further and stated that the Vinaya in 80 sections was that of the community in Mathura. So we have good reasons to suppose that two Vinaya-s -- and therefore two monastic communities, in some way similar -- were known in Kaśmir, and that one had come from Mathura. Lamotte's theory simply brushes over this fact, whereas Frauwallner adduces good evidence to show that the Mulasarvastivadin sect's connection with Kaśmir is late, and written onto an earlier church history of Mathura (1956 pp.26-36).

The dispute between the adherents of these two Vinaya-s is just about detectable in the words of Kumarajiva and Sêng yu. Kumarajiva, taught in the tradition of the Sarvastivadin Vinaya, tells us that the Vinaya of Kaśmir (i.e. the Sarvastivadin Vinaya) had only ten sections, but that it also had a Vibhaùa consisting of 80 sections. Why did he do this? Why did he state that the Vinaya has a commentary consisting of 80 sections? It might be a redundant statement, but I think that the evidence suggests otherwise. From his words (Lamotte p.174), it seems that he was aware of the claim that the original Vinaya consisted of 80 sections. He was also aware of the fact that the Vinaya of Mathura, probably the Vinaya of a very old Buddhist community, consisted of 80 sections. Surely his statement is that of an apologist, forced into making it because there were others who criticised the Sarvastivadin Vinaya of Kaśmir for lacking the full 80 sections.

We can see the nature of such a critique in Sêng yu's Ch'u san tsang chi chi: he states that the Vinaya with 10 sections was a reduction of the Vinaya in 80 sections, undertaken by Upagupta for the sake of those with 'weak faculties'. Sêng yu is blatantly polemical, arguing against the worth of the Sarvastivadin tradition. Thus, Kumarajiva stated that there is nothing amiss with the fact that one Vinaya (his Vinaya) had only 10 sections, whereas Sêng yu said that there was something deficient in it -- it was an offshoot of the Vinaya in 80 sections for those with a weaker disposition. Kumarajiva was the apologist, asserting the antiquity of his Kaśmirian Vinaya in the face of a rival, whereas Sêng yu was the inclusivist, attempting to include the Vinaya of Kaśmir and its tradition within the tradition of Mathura. It seems that in fifth century Kaśmir, there was some quarrel between the adherents of two different communities -- one original to Kaśmir, which defended its position, and the other more recently arrived community which arguing that the Kaśmir tradition was an offshoot of itself.

Why would two groups have clashed in this way? Surely there would have been no need for one community, when moving to another area, to attempt to usurp the position of the resident community. I suggest that everything makes sense if we accept that the disputed issue concerned the ownership of the sarvastivada idea: one group accepting sarvastivada ideas had moved from Mathura to Kaśmir, and there encountered another community which at some point adopted a similar sarvastivada ideology. In response to this, and considering itself to be the original source of the sarvastivada, it labelled itself the 'Mulasarvastivada'. If this argument, which I claim can be detected in the words of Kumarajiva and Sêng yu, was still fresh in the fifth century A.D., then it seems that the dispute broke out some time after the two groups had co-existed in the same area: Frauwallner noted that interpolations into the Mulasarvastivadin Bhaisajyavastu (indicating a relocation from Mathura to the North West) were probably made between c.150-300 A.D.[99] Therefore, we can posit a period in which sarvastivada ideas circulated between the two groups. But it is more likely that the sect to whom the idea belonged at the beginning of this contact was the sect from Mathura: the mission that led to the origination of Majjhantika's community in Kaśmir/Gandhara was probably vibhajjavadin in the early period, and we can guess, from the name of the sect itself, as well as from Sêng yu's aggressive stance and Kumarajiva's seemingly defensive position, that the Mulasarvastivadin community was more irked by the dispute. All this is of course highly speculative, but if we are to explain the Mulasarvastivadin/Savastivadin distinction as the result of sectarian dispute, then some explanation must be found.

It seems to me as if the Chinese works mentioned by Lamotte support Frauwallner's theory and the reworking of it I suggest here. At the least, they show that there was a problem in the Buddhist traditions of Kaśmir concerning two different Vinaya-s, whereas Lamotte failed even to notice that this is a historical problem; Frauwallner's theory seems to be the best explanation of the evidence. More recently, Enomoto has recently argued that the Sarvastivadin sect was no different from the Mulasarvastivadin -- he argues that the two words mean the same thing, but this does not explain the odd facts: two different Vinaya-s, two similar names, and two explanations of their relationship in fifth century Chinese works. However, Enomoto's argument is flawed: it begins with the late and unreliable evidence of the fanciful etymology of the word mulasarvastivadin in Śakyaprabha's Prabhavati (c.8th century), as well as equally unreliable evidence in the colophons of this work and others by Śakyaprabha -- all are inconsistent in the use of the prefix mula-.[100] He then attempts to show that Yi-jing used the words Sarvastivada and Mulasarvastivada interchangeably.[101] But it seems to me that the section of Yi-jing translated by him does not support such a view. The important section reads:

What are treated in this work [102] mostly resemble the Shi-song-lu. The three different sects divided from the Sarvastivada sect -- 1. Dharmaguptaka; 2. Mahiśasaka; 3. Kaśyapiya -- are not prevalent in the five parts of India... However, the Shi-song-lu does not belong to the 'Mula-sarvastivada' sect, either.[103]

The proximity of the title Shi-song-lu and the Sarvastivada sect in sentences one and two suggest that the former is the work of the latter, and from the last sentence, we know that this was not a work of the Mulasarvastivada sect. The translation of Enomoto certainly does not say that "the 'Mulasarvastivada' sect was divided into four sects: Dharmaguptaka, Mahiśasaka, Kaśyapiya and 'Mulasarvastivada' itself", or that "what is here called the 'Sarvastivada' sect is the same as the Mulasarvastivada sect."[104] On the contrary, it relates sects which, I have argued, originated from the same missionary endeavour, and distinguishes them from the Mulasarvastivadin-s. Enomoto's theory does not make sense and it does not explain the difficulties. It seems then that Frauwallner's explanation of the difference between Sarvastivadin and Mulasarvastivadin explains most of the facts while leaving fewer unresolved problems. Some of the remaining problems I have attempted to solve by showing that the name 'Mulasarvastivadin' originated in the course of sectarian debate, and this seems to offer the best explanation of the various facts.

-----*-----

(All Pali citations are from Pali Text Society editions; citations from the Aśokan edicts are from Hultzsch's edition).

Bareau, André. 1963, Recherches sur la biographie du Buddha dans les Sutrapitaka et les Vinayapitaka anciens I: De la quête de l'éveil à la conversion de Sariputra et de Maudgalyayana. Paris: École française d'Extrême-Orient.

Beal, Samuel. 1981, Si-Yu-Ki: Buddhist Records of The Western World. Delhi: Motilal Banarsidass. First Published: London (1906).

Bechert, Heinz. 1973, 'Notes on the Formation of Buddhist Sects and the Origins of Mahayana'. In German Scholars on India (Contributions to Indian Studies Vol. I), pp.6- 18. Varanasi: The Chowkhambha Sanskrit Series Office (1973).

Bronkhorst, Johannes. 1993, The Two Traditions of Meditation in Ancient India. New Delhi: Motilal Banarsidass. First published, Stuttgart: Steiner Verlag (1986).

Collins, Steven. 1990, 'On the very idea of the Pali Canon'. Journal of the Pali Text Society XV, pp.89-126.

Cousins, L. S. 2001, 'On the Vibhajjavadins. The Mahiṃsasaka, Dhammaguttaka, Kassapiya and Tambapaṇṇiya branches of the ancient Theriyas'. Buddhist Studies Review, vol. 18 no.2 pp.131-82.

Frauwallner, Erich. 1956, The Earliest Vinaya and the Beginnings of Buddhist Literature. Rome.

Frauwallner, Erich. 1995, Studies in Abhidharma Literature and the Origins of Buddhist Philosophical Systems. New York: State University of New York Press.

Gombrich, Richard. 1971, Precept and practice: traditional Buddhism in the rural highlands of Ceylon. Oxford: Clarendon Press.

Gombrich, Richard. 1988, Theravada Buddhism: A social history from ancient Benares to modern Colombo. London and New York: Routledge.

Gombrich, Richard. 1990, 'Making Mountains Without Molehills: The Case of the Missing Stupa'. Journal of the Pali Text Society, vol. XV pp.141-43.

Gombrich, Richard. 1992, 'Dating the Historical Buddha: A Red Herring Revealed'. In The Dating of the Historical Buddha Part 2, ed. Heinz Bechert, pp.237-59. Göttingen: Vandenhoeck & Ruprecht.

Hallisey, Charles. 1990, 'Apropos the Pali Vinaya as a Historical Document: A Reply to Gregory Schopen'. Journal of the Pali Text Society, vol. XV pp.197-208.

Horner, I. B. 1954, The Collection of Middle Length Sayings Vol.I: the first fifty discourses. London: Luzac and Company Ltd.

Hultzsch, E. 1991, Corpus Inscriptionum Indicarum Vol. I: Inscriptions of Aśoka. New Delhi: The Director General Archaeological Survey of India. First published, Oxford: Clarendon Press for the Government of India (1925).

Jones, J. J. 1949-56, The Mahavastu: translated from the Buddhist Sanskrit (Sacred Books of the Buddhists volumes 16, 18, 19). London: Luzac and Company Ltd.

Bhikkhu Ñaṇamoli and Bhikkhu Bodhi. 1995, The Middle Length Discourses of the Buddha: A Translation of the Majjhima Nikaya. Boston: Wisdom Publications.

Norman, K.R. 1978, 'The role of Pali in early Sinhalese Buddhism', in Collected Papers II, pp.30-51. First published in Heinz Bechert ed., Buddhism in Ceylon and Studies on Religious Syncretism, Göttingen, pp.28-47.

Norman, K.R. 1982, Aśokan sila-thaṃbha-s and dhamma-thaṃbha-s. In Acarya-vandana (D.R.Bhandarkar Birth Centenary Volume), Calcutta 1982, pp.311-18. Reprinted in Collected Papers volume II, pp.224-33.

Norman, K.R. 1983, Pali literature: including the canonical literature in Prakrit and Sanskrit of all the Hinayana schools of Buddhism. Wiesbaden: Harrassowitz

Norman, K.R. 1997, A Philological Approach To Buddhism: the Bukkyo Dendo Kyokai Lectures, London: School of Oriental and African Studies.

Norman, K.R. 1990-2001, Collected Papers I-VII. Oxford: Pali Text Society.

Schopen, Gregory. 1997, Bones, Stones and Buddhist Monks. Collected Papers on the Archaeology, Epigraphy, and Texts of Monastic Buddhism in India. University of Hawaii Press.

Skilling, Peter. 1981-82a, 'Uddaka Ramaputta and Rama'. Pali Buddhist Review 6.2, pp.99-105.

Thomas, E. J. 1993, The Life of Buddha as Legend and History. Delhi: Motilal Banarsidass. First published, London: Kegan Paul (1927).

Vetter, Tilmann. 1988, The Ideas and Meditative Practices of Early Buddhism. Leiden: E.J.Brill.

Waldschmidt, Ernst. 1950-51, Das Mahaparinirvanasutra. Text in Sanskrit und Tibetisch, verglichen mit dem Pali nebst einerṃbersetzung der chinesischen Entsprechung im Vinaya der Mulasarvastivadins auf Grund von Turfan-Handschriften. Berlin: Akademie- Verlag.

Willis, Michael. 2001, 'Buddhist Saints in Ancient Vedisa'. Journal of the Royal Asiatic Society, Series 3, II, 2, pp.219-28.

Zafiropulo, G. 1993, L'illumination du Buddha: de la quête à l'annonce de l'Éveil: essais de chronologie relative et de stratigraphie textuelle. Innsbruck : Institut fur Sprachwissenschaft der Universität Innsbruck.

-----*-----

[1] Dip XX.20-21, Mhv XXXIII.100-01; See Collins p.97.

[2] Accepting Richard Gombrich's dates; see below p.11 n.32.

[3] Norman 1978 p.36.

[4] On this evidence, see below pp.19-20.

[5] It is unlikely that the Abhidharma works of various schools were fixed at this date. See below p.15.

[6] Schopen p.3.

[7] Schopen p.3.

[8] Schopen pp.23-24.

[9] Schopen p.56. See also Schopen's comments, p.71 n.50: 'We do know, however, that from the very beginning of our actual epigrahical evidence (Bharhut, Saṭci, etc.), a large number of monks were doing

[10] Schopen p.31.

[11] Hallisey p.208.

[12] See in particular his article 'Monks and the Relic Cult in the Mahaparinibbana-sutta: An Old Misunderstanding in Regard to Monastic Buddhism' (= Schopen 1997 pp.99-113).

[13] See in particular his article 'Two Problems in the History of Indian Buddhism: The Layman/Monk Distinction and the Doctrines of the Transference of Merit' (= Schopen 1997 pp.23-55), parts of which I will consider below.

[14] Schopen pp.25-26.

[15] Schopen pp.28-29.

[16] Schopen p.29.

[17] Schopen p.29.

[18] Schopen pp.86-98.

[19] Schopen pp.89-91.

[20] See D II.142.5ff: ...catummahapathe raṭṭo cakkavattissa thupaṃ karonti. evaṃ kho Ananda raṭṭo cakkavattissa sarire paṭipajjanti. yatha kho Ananda raṭṭo cakkavattissa sarire paṭipajjanti evaṃ tathagatassa sarire paṭipajjitabbaṃ. catummahapathe tathagatassa thupo katabbo; and D II.164.28: aham pi arahami bhagavato sariranaṃ bhagaṃ, aham pi bhagavato sariranaṃ thupaṭ ca mahaṭ ca karissami. The Sanskrit Mahaparinirvaņa Sutra edited by Waldschmidt also mentions śarirastupa-s in portions of text which correspond to these Pali references: 36.7 corresponds to D II.142.5, 50.5 corresponds to D II.142.5. The compound śarirastupa also appears at 46.7, 50.16, 50.20, 51.9, 51.22.

[21] Hultzsch p.165: (A) devanaṃpiyena piyadasina lajina chodasavasabhisitena, budhasa konakamanasa thube dutiyaṃ vaḍhite.

[22] Hultzsch p.164: (A) ...atana agacha mahiyite hida budhe jate sakyamuni ti.

[23] Waldschmidt p.388. The Pali version is only slightly different: idha tathagato jato ti (D II 140.20 = A II 120.24).

[24] D II 88.28ff = Ud 89.20 = Vin I 229.35; A V.269-73. On these passages see Gombrich 1971 p.267 and p.272.

[25] Schopen p.91.

[26] Gombrich 1990 pp.141-142, Hallisey pp.205-206. It seems to me that Hallisey has made it clear that: 'Buddhaghosa, Sariputta, and the otherṭikacariyas did not include the observances concerning stupas and bodhi trees among the observances specified in the Vinaya itself' (p.205). This does not explain the passage in the Visuddhimagga quoted by Schopen p.88, which still presents difficulties about the exact meaning of the compound khandhakavattani, although Hallisey notes: 'Perhaps it grouped a range of practices according to their family resemblances, rather than by their common origin in specific parts of the Vinaya.' (p.206).

[27] Gombrich 1990 p.143.

[28] Norman 1997 p. 140.

[29] Norman 1997 p.140.

[30] Norman 1997 p.140.

[31] Oldenberg 1879, p.xlviii.

[32] Accepting Richard Gombrich's dating of the Buddha: '...the Buddha died 136 years before Aśoka's inauguration, which means in 404 B.C.' (1992 p.246). Gombrich estimates the margin of error to be 7 years before to 5 years after this date, i.e. 411-399 B.C. (p.244). But he also notes that uncertainty about the date of Aśoka widens the margin of error, making the upper limit 422 B.C. K.R. Norman comments: 'If we take an average, then the date is c.411 ± 11 B.C.E.' (Norman 1999 p.467).

[33] Schopen p.26.

[34] Schopen p.26 on A. Bareau, Les sectes bouddhiques du petit véhicule (Paris, 1955).

[35] See below pp.19-20 on the notion that the Thera Moggaliputta sent the missions.

[36] Frauwallner 1956 pp.13-14.

[37] Willis p.226 n.26.

[38] Willis pp.222-23.

[39] Dip XII.35ff, Mhv XIII.18-20.

[40] Gombrich 1988 p.135.

[41] Lamotte p.297.

[42] Lamotte p.308.

[43] Gombrich 1988 p.135.

[44] Cousins p.169.

[45] Frauwallner 1956 p.22: 'The mission of Kassapagotta, Majjhima and Dundubhissara gave origin to the Haimavata and Kaśyapiya. The mission of Majjhantika led to the rise of the Sarvastivadin. The Dharmaguptaka school is perhaps issued from the mission of Yonaka-Dhammarakkhita.'

[46] Frauwallner 1956 p.38.

[47] He says this on p.38 in his discussion about the difference between Sarvastivadin and Mulasarvastivadin, but it applies in general to his thought on the dogmatic affiliation of all the sects resulting from the missions.

[48] Frauwallner 1956 p.2.

[49] Frauwallner 1995 p.37.

[50] See the chapter on the Śariputrabhidharma in Frauwallner 1995.

[51] Frauwallner 1956 p.39.

[52] Frauwallner 1956, pp.26-31, especially p.31: 'We come thus to the conclusion that the episode of Madhyantika and of the conversion of Kaśmir represents a late interpolation in the Vinaya of the Mulasarvastivadin.' In the light of Frauwallner's work, the Sarvastivada inscriptions of Mathura mentioned by Lamotte, p.523, might be those of the school that in the North-West came to be known as the Mulasarvastivadin-s. See the appendix for a further consideration of the Sarvastivadin/Mulasarvastivadin issue.

[53] They are mentioned in RE V, RE XII and PE VII; it would have been beyond the jurisdiction of a visitor to another kingdom to carry out some of these duties. ś

[54] RE II (Shahbazgarhi, Hultzsch p.51): (A) ...duvi 2 chikisa kriṭa manuśa-chikisa... paśu-chikisa cha (B) oshaḍhani manuśopakani cha paśopakani cha yatra yatra nasti savatra harapita cha vuta cha (C) kupa cha khanapita pratibhogaye paśu-manuśanaṃ.

[55] RE XIII (Shahbazgarhi, Hultzsch p.68/211): (S) yatra pi Devanaṃpriyasa duta na vrachaṃti...

[56] Hultzsch's translation, p.135, of PE VII (p.132): (U) la... esa paṭibhoge nama (V) vividhaya hi sukhayanaya pulimehi pi lajihi mamaya cha sukhayite loke.

[57] PE VII, Hultzsch p.132: (W) imaṃ chu dhammanupaṭipati anupaṭipajaṃtu ti etadatha me esa kaṭe.

[58] Hultzsch's translation, p.136, of PE VII (p.133): (FF) Devanaṃpiye P...s. laja hevaṃ aha (GG) yani hi kanichi mamiya sadhavani kaṭani taṃ loke anupaṭipaṃne taṃ cha anuvidhiyaṃti (HH) tena vaḍhita cha vaḍhisaṃti cha mata-pitisu sususaya gulusu sususaya vayo-mahalakanaṃ anupaṭipatiya babhanasamanesu kapana-valakesu ava dasa-bhaṭakesu saṃpaṭipatiya.

[59] In RE XIII, the crucial passage outlining his dhamma is found in section G, Hultzsch p.67/208.

[60] Erich Frauwallner related the Buddhist missions to Aśoka precisely because of the reference to duta-s in RE XIII (1956 p.15 n.1). He did not mention the evidence in the chronicles for the duta-s of Aśoka, however.

[61] Mhv XI.33-36: datva kale sahayassa paņṇakaraṃ narissaro, dute pahesi saddhammapaņṇakaram imam pi ca (33). "ahaṃ buddhaṭ ca dhammaṭ ca saṅghaṭ ca saraņaṃ gato, upasakattaṃ desesiṃ sakyaputtassa sasane (34), tvam pi'mani ratanani uttamani naruttama, cittam pasadayitvana saddhaya

[62] Dip XII.8: asokarame pavare bahu thera mahiddhika, laṅkatalanukampaya mahidaṃ etad abravuṃ (8).

Dip XVII.91cd-92ab: dutiyabhiseke tassatikkanta tiṃsarattiyo, mahidogaņa pamokkho jambudipa idhagato.

[64] As Norman points out (1983 p.118).

[65] Although there is every possibility that Mogalliputta, thera of the Asokarama, aided Aśoka in organising the missions.

[66] On the idea that Moggaliputta sent out the missions, Frauwallner states: '...we must remember that the data of the Sinhalese chronicles are uncertain on this point.' (1956 p.17). He concludes: 'The mother community tried apparently to enhance the glory of its patriarch by putting on his merit the sending out of the missions.' (1956 p.18).

[67] Frauwallner 1956 p.54.

[68] Frauwallner 1956 p.67: 'It must have been composed shortly before or after the second council'.

[69] Gombrich 1992 p. 258: 'We may thus date the Second Council round 60 A.B. or round 345 B.C.; the dates are very approximate and the precise margin of error incalculable...'.

[70] At this point Frauwallner dates the old Skandhaka according to older views about the date of the second council, c.100 years after the Buddha's death. More recent research has modified this date somewhat; I follow Gombrich's date of c.345 B.C. for the second council.

[71] Frauwallner 1956 p.153.

[72] The Suttas including this account are the Maha-Saccaka Sutta (M no.36), the Bodhi-Rajakumara Sutta (M no.85) and the Saṅgarava Sutta (M no.100).

[73] Bareau pp.14-16.

[74] Mvu II.118.1ff.

[75] SBhV I.97.4ff; Skilling points out that there is a Tibetan translation of this SBhV account, as well as a 'virtually identical' Mulasarvastivadin version, preserved in the Tibetan translation of the Abhini ṣkramaņa-Sutra (Skilling p.101).

[76] Vetter p.xxii, Bronkhorst p.86; Bareau sums up his view as follows: 'Personnages absents, morts même avant que leurs noms ne soient cités, ils sont probablement fictifs. Plus tard, on s'interrogea sur ces deux mystérieux personnages et l'on en déduisit aisément qu'ils n'avaient pu être que les maîtres auprès desquels le jeune Bodhisattva avait étudié.' (pp.20-21).

[77] Zafiropulo pp.22-29.

[78] Si-Yu-Ki (Beal , Part II p.139ff).

[79] Mvu II.119.8.

[80] Mp III.164.23: samaņe ramaputte ti uddake ramaputte.

[81] D II.72.9ff = A IV.17.11ff (Sattakanipata, anusayavagga, XX). He also appears in the Gopakamogallana Sutta (M III.7ff), which is set in Rajagaha. At Vin I 228 (= D II 86.31ff, Ud 87), he and Sunidha are in charge of the construction of Paṭaligama's defences.

[82] M I.165.22ff: so kho ahaṃ bhikkhave nacirass' eva khippam eva taṃ dhammaṃ pariyapuņiṃ. so kho ahaṃ bhikkhave tavataken' eva oṭṭhapahatamattena lapitalapanamattenaṭaņavadaṭ ca vadami theravadaṭ ca, janami passamiti ca paṭijanami ahaṭ c' eva aṭṭe ca.

[83] M I.166.4ff: ...yan nunahaṃ yaṃ dhammaṃ Ramo sayaṃ abhiṭṭa sacchikatva upasampajja viharami ti pavedeti, tassa dhammassa sacchikiriyaya padaheyyan ti? so kho ahaṃ bhikkhave nacirass' eva khippam eva taṃ dhammaṃ sayaṃ abhiṭṭa sacchikatva vihasiṃ.

[84] Skilling discusses this in detail; the point had been made earlier by Thomas p.63 andṭaṇamoli and Bodhi p.258 n.303.

[85] M I.165.27ff: na kho ramo imaṃ dhammaṃ kevalaṃ saddhamattakena sayaṃ abhiṭṭa sacchikatva upasampajja viharami ti pavedesi, addha ramo imaṃ dhammaṃ janaṃ passaṃ vihasi ti.

[86] M I.165.32ff: kittavata no avuso ramo imaṃ dhammaṃ sayaṃ abhiṭṭa sacchikatva upasampajja [VRI: viharamiti] pavedesi ti?

[87] M.I.166.22ff: iti yaṃ dhammaṃ ramo aṭṭasi, taṃ tvaṃ dhammaṃ janasi; yaṃ tvaṃ dhammaṃ janasi, taṃ dhammaṃ ramo aṭṭasi.

I leave dhammaṃ untranslated here because it indicates the meditative sphere attained by both Rama and the Buddha. Before this, the Buddha is said to have mastered the dhamma intellectually (165.24 = 164.4-5; see n.68), which can hardly mean a meditative attainment and must refer to an intellectual understanding.

[88] M.I.165.3ff: iti yahaṃ dhammaṃ janami, taṃ tvaṃ dhammaṃ janasi; yaṃ tvaṃ dhammaṃ janasi, tam ahaṃ dhammaṃ janami.

[89] M I.165.5ff: iti yadiso ahaṃ tadiso tuvaṃ, yadiso tuvaṃ tadiso ahaṃ. ehi dani avuso ubho va santa imaṃ gaņaṃ pariharama ti. iti kho bhikkhave aḷaro kalamo acariyo me samano antevasiṃ maṃ samanaṃ attano samasamaṃṭhapesi, uḷaraya ca maṃ pujaya pujesi.

[90] M I.166.24ff: iti y diso ramo ahosi tadiso tuvaṃ, yadiso tuvaṃ tadiso ramo ahosi. ehi dani avuso tvaṃ imaṃ gaņaṃ parihara ti. iti kho bhikkhave udako ramaputto sabrahmacari me samano acariyaṭṭhane ca maṃṭhapesi, uḷaraya ca maṃ pujaya pujesi.

[91] See Skilling, pp.100-102.

[92] Skilling p.101.

[93] Horner pp.209-10. Jones (p.117), translator of the Mahavastu, preserves the distinction between Rama and Ramaputra, but fails to notice that in the Mahavastu, Ramaputra does not establish the Bodhisattva as an equal to him: it says that he established the Bodhisattva as the teacher (Mhv II 120.15: acaryasthane sthapaye). Jones translates: 'Udraka Ramaputra ... would make me a teacher on an equal footing with himself' (p.117).

[94] Bareau p.20: 'Mais le parallélisme avec l'épisode suivant, l'ordre trop logique et le choix trop rationnel des points de doctrine d'Aḷara Kalama et d'Udraka Ramaputra nous laissent un arrière-goût d'artifice qui nous rend ces récits suspects.'

[95] Zafiropulo (p.25) does not point out the difference between Rama and Ramaputta, but on the stereotyped description of the training under the two teachers he comments: 'Justement cela nous semblerait plutôt un signe d'ancienneté, caractéristique de la transmission orale primitive par récitations psalmodiées'.

[96] According to Frauwallner, 'the Vinaya of the Sarvastivadin largely agrees with the Vinaya of the other missionary schools and forms with them a close group, while the Vinaya of the Mulasarvastivadin shows considerable differences.' (1956 p.38).

[97] Frauwallner 1956 p.40: 'They were at first two independent communities of different origin... Later on both communities grew into one school throught their accepting of the theories of the philosophicaldogmatic school; but they never completlely lost their individualities.'

[98] Frauwallner 1956 pp.28-31.

[99] Frauwallner 1956 p.36.

[100] Enomoto pp.240-42.

[101] Enomoto p.243.

[102] This work being the Mulasarvastivadin Vinaya.

[103] As translated by Enomoto pp.242-43.

[104] Enomoto p.243.

According to the Sinhalese chronicles, the Pali canon was written down in the reign of King Vaṭṭagamiṇi (29-17 B.C.).[1] It has been generally accepted, therefore, that the canon contains information about the early history of Indian Buddhism, from the time of the Buddha (c.484-404 B.C.) until the end of the first century B.C. [2] That the canonical texts are a record of the period of Buddhism before they were written down in Sri Lanka seems to be confirmed by the fact that their language, Pali, is north Indian in origin. Thus the Pali canon shows 'no certain evidence for any substantial Sinhalese additions ... after its arrival in Ceylon.' [3] If the language of the Pali canon is north Indian in origin, and without substantial Sinhalese additions, it is likely that the canon was composed somewhere in north India before its introduction to Sri Lanka, and is therefore a source for the period of Buddhism in northern India before this. The Sinhalese chronicles state that the canon was brought to Sri Lanka by Mahinda during the reign of Aśoka, implying that it predates the middle of the third century B.C. [4] According to this history, the Pali canon, particularly the Vinaya and Sutta portions, is a reliable source for the early history of Indian Buddhism in the period before Aśoka. [5]

This version of events is not accepted by all, however. Gregory Schopen in particular has argued against the view that the canonical texts can be taken as accurate historical sources for the earliest period:

Scholars of Indian Buddhism have taken canonical monastic rules and formal literary descriptions of the monastic ideal preserved in very late manuscripts and treated them as if they were accurate reflections of the religious life and career of actual practising Buddhist monks in early India. [6]

This point of view has two aspects to it. On the one hand, normative religious literature must not be taken at face value, as if it contains evidence of real historical events. As Schopen puts it:

Even the most artless formal narrative text has a purpose, and that in "scriptural" texts, especially in India, that purpose is almost never "historical" in our sense of the term. [7] On the other hand, Schopen doubts that texts preserved in 'very late manuscripts' contain accurate historical evidence -- he wishes us to believe that the canonical texts cannot be taken as evidence for the period before the fifth century A.D.:

We know, and have known for some time, that the Pali canon as we have it -- and it is generally conceded to be our oldest source -- cannot be taken back further than the last quarter of the first century B.C.E, the date of the Alu-vihara redaction, the earliest redaction that we can have some knowledge of, and that -- for a critical history -- it can serve, at the very most, only as a source for the Buddhism of this period. But we also know that even this is problematic since, as Malalasekera has pointed out: "...how far the Tipiṭaka and its commentaries reduced to writing at Alu-vihara resembled them as they have come down to us now, no one can say." In fact, it is not until the time of the commentaries of Buddhaghosa, Dhammapala, and others -- that is to say, the fifth to sixth centuries C.E. -- that we can know anything definite about the actual contents of this canon. [8]

A central theme running through Scopen's work is his claim that we cannot know anything for sure about Indian Buddhism from its texts that were redacted in the fifth century A.D. (for the Pali canon), or the fourth century A.D. (approximately, for the canonical material of various sects preserved in Chinese translations). Consequently, Schopen believes that the only way we can find out anything about Buddhism before this time is through accurately dated epigraphical and archaeological material. It is clear from Schopen's work that this evidence has not been given the attention it deserves; it is vitally important to study the material remains, which tell us something concrete about what Buddhists were doing at particular places in particular times. But does this mean that we should concentrate exclusively on the material remains? Should we throw out the texts, or merely allow their evidence to be restricted and subordinated to the material evidence? The impression given by Schopen's work is that the study of early Buddhism can only progress by subordinating the literary evidence to the material evidence, an approach which seems to have become standard in some quarters. But before we consign ourselves to a radical reorientation in the study of early Buddhism, we should critically examine some of the presuppositions of this approach. There seem to be three questions of importance here:

1) How old are the canonical texts?

2) Are the canonical texts purely normative, or do they include descriptive material which can be used to reconstruct historical events?

3) And finally, how much importance is to be assigned to the epigraphical and archaeological evidence?

I radically disagree with Schopen's answers to each of these questions. In what follows, I hope to show why Schopen's views are untenable, and I will argue that the only way of knowing anything about early Buddhism is through its texts. I will begin with the last point first: it seems to me that the worth of the epigraphical and archaeological evidence has been overstated by Schopen. This is not to deny its great importance for the study of Indian Buddhism -- without it, the historian is fumbling in the dark, and his conclusions will lack verisimilitude. Be that as it may, the material evidence has its own limitations, and the fact is that it does not tell us that much about the thought and practices of Buddhists in ancient India. So although Schopen has used this evidence to draw attention to hitherto neglected aspects of Indian Buddhism (e.g. that monks and nuns probably instigated the cult of the image, or that monks and nuns were involved in the stupa cult from the earliest times), he does not acknowledge the fact that this does not tell us very much about Indian Buddhism as it was practised. It does not allow us to probe very far into the beliefs and practices of Buddhist monks and nuns in India; its content is limited, much more limited than the content of the early texts, which seem to me to contain a wealth of information on the diverse beliefs current in early Buddhism.

According to Schopen, the epigraphical material ' tells us what a fairly large number of Indian Buddhists actually did, as opposed to what -- according to our literary sources -- they might or should have done.'9 What exactly Schopen has in mind when he says 'a fairly large number of Indian Buddhists' is unclear, but certainly misleading: the relevant inscriptions number only a few thousand, which is evidence, surely, for the activity of a small minority of monks and nuns. They can hardly be taken as indicative of the activity of the Buddhist populace at large -- just over a couple of thousand inscriptions does not, to my mind, represent a large number of Indian Buddhists, considering that this must have been a tiny fraction of the number of Indian Buddhists from about 400 B.C. to 500 A.D.

In other words, there is a tendency in Schopen's work to make generalisations about Indian Buddhism based on a very small amount of evidence. Even if the generalisations were true in every respect, it would only reveal the historical reality of a tiny part of Indian Buddhism. Perhaps if there were epigraphical evidence representing every Buddhist who existed in ancient India, it would be similar to the evidence of the extant inscriptions. But we cannot presume what is not there. For all we know, the inscriptions might represent only a small minority of the ancient Indian Saṅgha, the minority who had personal wealth and who could endow Buddhist institutions in different ways.

In this situation, we should not underestimate the worth of the textual evidence, even if its antiquity cannot be established accurately. For example, Schopen records that two inscriptions at Mathura record the donations of monks who are called prahaṇika-s, 'practisers of meditation'.[10] But without consulting the evidence of the Pali canon for the word padhana or the Buddhist Sanskrit evidence for the word pradhana/prahaṇa (or variations on them), we would have absolutely no idea what the term signified for the two monks, and why they used it. The fact is that the texts are indispensable: the literary evidence, even if only normative, and even if it was periodically revised until the rather late redactions, is most certainly a useful record, not to be used as subsidiary to the material evidence, as Schopen believes, but in tandem with it, so that the two sorts of evidence are used equally. In short, if the inscriptions are to have any significance for the study of early Indian Buddhism, they must be considered alongside the canonical evidence, as has been argued by Hallisey:

It will only be after we have learned to combine our interest in "what really happened" with a sensitivity to the changing thought-worlds of the Theravada that we will begin to discern the historical reality behind the literary and archaeological traces of ancient Buddhist monasticism.[11]

It seems to me that Schopen's work is most convincing when he follows this method, and uses the literary, epigraphical and archaeological sources equally, [12] instead of just dismissing the literary evidence out of hand. [13] Unfortunately, in his eagerness to point out that the studies of previous generations of Buddhist scholars were one-sided, Schopen has created another one-sided version of history. What is needed is a balanced approach that gives both sets of evidence, the literary and material, their due worth. But what is the worth of the literary evidence? This brings me to two of the questions posed above, viz. the age and nature of the canonical texts. Schopen's position on these two points is quite clear, as we have seen, although it is strange that he does not give any evidence to support his view that the narrative Buddhist literature is 'almost never historical', as if this were a self-evident fact. As for his point that we cannot know if the canonical material is old, he attempts to demonstrate this by claiming that the general method of higher criticism -- the method which is often used to prove the antiquity of canonical texts -- is inapplicable. He sums up this method of higher criticism as follows: 'If all known sectarian versions of a text or passage agree, that text or passage must be very old; that is, it must come from a presectarian stage of the tradition.' [14]

The alternative explanation of the agreement of 'all known sectarian versions of a text or passage' is that the agreement was produced by the different sects sharing literature at a later date. It is this hypothesis which Schopen attempts to prove by showing that the similar versions of the story of the stupa of Kaśyapa at Toyika, found in Mahasaṅghika, Mahiśasaka, Dharmaguptaka and Theravadin texts, are later than versions found in the Mulasarvastivadin Vinaya and in the Divyavadana. [15] The former group of texts claim that the Buddha manifested a stupa momentarily, after which a stupa was built (by monks) or appeared. The version of the story in the Mulasarvastivadin Vinaya and in the Divyavadana, however, is described by Schopen as follows: 'Firstly, it has none of the various subplots found in the other versions -- a fairly sure sign of priority -- and, second, it knows absolutely nothing about a stupa at Toyika or its construction.' [16] Schopen's main argument then is that the story in the Mulasarvastivadin Vinaya and the Divyavadana is earlier because it does not mention a stupa: 'This version, in short, reflects a tradition -- apparently later revised -- that only knew a form of the relic cult in which the stupa did not yet have a part.' [17]

The first thing which I find odd about Schopen's assessment of this story is his claim that, on the basis of the evidence in the Mulasarvastivadin Vinaya and Divyavadana, there was a form of the relic cult that did not include the stupa. The narratives in these texts mention caitya-s, and although Schopen states that this term has nothing to do with stupa-s, this is not at all clear. In his article 'The Stupa Cult and the Extant Pali Vinaya', [18] he has in fact argued that in the Pali literature, the word cetiya is equivalent to stupa. [19] It could easily be the case that the word has the same meaning in the Mulasarvastivadin Vinaya and the Divyavadana. But even if not, are we to accept a form of relic worship without a stupa? If we take the canonical texts seriously, it is hard to imagine that this could ever have been the case. The Mahaparinibbana Sutta, for example, states that the Buddha's relics are to be contained in a stupa, [20] which suggests that the stupa goes back to the very beginning of Buddhism. The stupa was certainly a feature of Buddhism by the time of Aśoka, who records in his Nigali Sagar Pillar Edict that twenty years into his rule, he had the thuba of Konakamana doubled in size. [21] Moreover, Aśoka seems to have known a portion of the text found in the Sanskrit version of the Mahaparinirvaṇa Sutra -- in his Rummindei inscription, he records that he visited Lumbini and worshipped there saying 'Here the Blessed One was born', [22] which corresponds to the Sanskrit version of the Mahaparinirvaṇa Sutra (41.8: iha bhagavaṭ jataḥ). [23] This part of the text is close to the parts in the Pali and Sanskrit versions which mention stupa-s, and so it seems natural to conclude that stupa worship was not only a part of Buddhism at this date, but also that it was mentioned in canonical Buddhist texts at this point. This is an important point, for according to the most plausible theory of sect formation (the theory proposed by Frauwallner), some of the Sthavira sects formed as a result of the Aśokan missions in 250 B.C. (see below p.11ff). If the Aśokan evidence suggests that by about this time the stupa was a feature of Buddhism and its texts, a presectarian period that did not relate relic worship to the construction of stupa-s is hardly plausible. It seems that there are no obvious reasons for taking the story in the Mulasarvastivadin Vinaya and Divyavadana to be older than the versions in the Mahasaṅghika, Mahiśasaka, Dharmaguptaka and Theravadin texts.

However, even if Schopen has got it right and his argument is valid, it actually shows that the Pali canon was closed to material received from other sects. What Schopen fails to mention is that the method of higher criticism used to establish old strata in the Buddhist literature usually compares the canonical literature of different sects: he is reluctant to note that the Pali version of the story of the stupa of Kaśyapa at Toyika is found in the Dhammapada-aṭṭhakatha -- this information is conveniently confined to footnote 28. This means that if Schopen is correct, it seems then that whereas some of the other sects periodically shared literature and changed their canonical material in the sectarian period, the Theravadins of Sri Lanka did not: they confined the received material to non-canonical books. It seems that Schopen might have inadvertently proved that the Pali canon was relatively closed after its redaction at an early date. This depends of course on whether or not he has interpreted the different versions of the story about the stupa at Toyika correctly, and this is far from clear. A thorough study of the different versions of the story is surely necessary. However, it is worth taking a short digression to show that another inadvertent proof of the antiquity of at least the Suttapiṭaka is given by Schopen in the very same article ('Two Problems in the History of Indian Buddhism: The Layman/Monk Distinction and the Doctrines of the Transference of Merit').

In this article, Schopen shows that the belief in the transference of merit was widespread in India from the third century B.C. onwards (pp.34-42). Thus, a late Mauryan/early Śuṅga inscription from Pauni, a few inscriptions from third century B.C. Sri Lanka, a singular early inscription from Bharhut, as well as a significant number of later Hinayana inscriptions from various parts of India all record the idea. If the idea was a standard belief of Buddhists in early times, even in Sri Lanka, and if the Suttapiṭaka was not finally closed until the Alu-vihara recension in the fifth century A.D., then it is reasonable to expect that it should be well attested in the Suttapiṭaka. But this is not the case -- although much is said on the subject of meritorious activity, the idea of the transference of merit is found in only two separate occurrences in the four principle Nikaya-s. [24] How can we explain the fact that Theravada Buddhists of Sri Lanka did not compose more texts which included the idea of merit transference? There can only be one answer -- the texts were closed in an earlier period, when the belief was marginal in Buddhist circles. This is surely the only answer to the problem. Even if this does not definitely prove that the canon was closed at an earlier date, the fact that the ancient guardians of the Suttapiṭaka did not compose texts on the transference of merit shows that they must have had some idea of canonical orthodoxy, which means that the canon must have been relatively fixed. By attempting to show that the canonical texts are not reliably old, and that we must turn to the epigraphic evidence to gain any idea about the historical reality of ancient Indian Buddhism, Schopen has inadvertently shown that some collections of texts must indeed be old and contain evidence for the period before most of the inscriptions.

Exactly the same fact emerges from Schopen's article 'The Stupa Cult and the Extant Pali Vinaya'. He attempted to show that 'the total absence of rules regarding stupas in the Pali Vinaya would seem to make sense only if they had been systematically removed', [25] meaning that the Pali canon was altered 'at a comparatively recent date', after the supposed recensions made in the first century B.C. and the fifth century A.D.

This argument is based upon the fact that all the other extant Vinayas include rules concerning the construction and cult of the stupa, whereas the Pali Vinaya does not. There are two possible explanations for this fact. Either it is because the Pali Vinaya was closed before these rules were formulated, or it is because these sections were written out of the Pali Vinaya, accidentally or on purpose; Schopen chooses the latter option. But Gombrich and Hallisey have shown that this interpretation is based on a mistranslation of the twelfth century Sinhalese inscription, the Maha-Parakramabahu Katikavata. [26] It therefore seems likely that the other solution to the problem is correct -- the Pali Vinaya was closed before this section was composed and added to the other Vinayas. Gombrich notes: 'One does not have to posit that it received no further additions after the first century B.C., merely that the Pali tradition had left the mainstream and naturally failed to record later developments on the Indian mainland.' [27] But because we know that the Pali tradition remained in contact with the Indian mainstream (it received texts from north India after the first century B.C.), I think it more likely that no further additions were made after the first century B.C.

The points Schopen makes about the post-canonical sharing of literature, the transference of merit, and the Pali Vinaya's evidence on stupa-s, if correctly interpreted, suggest that the Pali canon was relatively fixed from at least the first century B.C. onwards. This is despite the fact that the Pali tradition remained in contact with other Buddhist sects in India, as has been noted already by scholars such as Oldenberg and Norman. According to Norman, 'some of the best known stories in Buddhism ... are known in the Theravadin tradition only in the commentaries, although they are found in texts which are regarded as canonical in other traditions.' [28] Such stories must have reached Sri Lanka before Buddhaghosa, for he includes them in his commentaries. But why were they not inserted into the canon? Norman thinks that it was because 'at least the Vinaya- and Sutta-piṭaka had been closed at an earlier date.' [29] Norman has also pointed out that certain Pali works for which a North Indian origin is supposed, such as the Milindapaṭha, the Peṭakopadesa and the Nettipakaraṇa, are highly respected by the commentators but are not given canonical status by them. They even contain 'a number of verses and other utterances ascribed to the Buddha and various eminent theras, which are not found in the canon ... There was no attempt made to add such verses to the canon, even though it would have been a simple matter to insert them into the Dhammapada or the Theragatha.' [30] The point that the Pali tradition received literature from other sects but excluded it from the canon had been made already by Oldenberg in 1879: 'These additions are by no means altogether unknown to the Singhalese church, but they have been there placed in the Aṭṭhakathas, so that the text of the Tipiṭaka, as preserved in Ceylon, has remained free from them.' [31] This suggests that they arrived in Sri Lanka 'after the closure of the Canon'.

If we remind ourselves of Norman's point that the Pali canon contains no definite evidence for a substantial amount of Sinhalese prakrit (see above p.1), it seems quite clear that after the Tipiṭaka was written down in the first century B.C., it was not substantially altered, at least in content, and as such, it must have been very similar to the extant Pali Canon. This means that the Suttapiṭaka in existence today can be taken as an accurate record of Buddhist thought from the time of the Buddha (c. 484-404 B.C.) until the first century B.C. at the latest . [32] This is significantly older than Schopen is willing to acknowledge, but the terminus ante quem can be pushed back even further; it depends upon the date when the Pali texts reached Sri Lanka, i.e. the date at which the sectarian period began.

According to Schopen, 'we do not actually know when the sectarian period began.' [33] To support this view he cites Bareau's work which points out that the Buddhist sects all give different dates for the schisms. [34] But he does not make any mention of what is probably the most convincing work on the subject. Erich Frauwallner, in The Earliest Vinaya and the Beginnings of Buddhist Literature, used a mixture of epigraphical and literary sources to argue that some of the Sthavira sects owed their origination to the missions said to have taken place in the reign of Aśoka, c.250 B.C. Firstly, there are records in the Sinhalese chronicles (and the Samantapasadika) of a series of Buddhist missions which went out to different parts of India and neighbouring kingdoms in the reign of Aśoka. Although these Pali accounts as we have them do not seem plausible -- they might have been embellished to trump up the monastic lineage from which the Pali texts issued [35] -- the missions are confirmed by the inscriptions found on a couple of reliquaries unearthed in the ancient Buddhist centre of Vidiśa. According to Frauwallner, these reliquaries contain the remains of the Hemavata masters Dudubhisara, Majjhima and Kassapagotta, names which he identified with the missionaries Durabhisara, Majjhima and Kassapagotta, all of whom travelled to the Himavanta according to the chronicles. [36] Willis has recently pointed out that Frauwallner misread this evidence slightly by mistaking the relics of Gotiputa, heir of Dudubhisara, for Dudubhisara himself, [37] but at the same time he has argued that all five names on the two different reliquaries correspond to the five names in the chronicles. [38] It strongly implies that the missionaries to the Himavanta hailed from Vidiśa and that some of their relics were returned there some time after their death. The chronicles also record that Mahinda's mother was from Vidiśa, and that he stayed there before journeying to Sri Lanka. [39] This is an impressive correspondence of epigraphical and literary evidence, and it makes it almost certain that the account of the missions in the Pali chronicles contains some historical truth.

Frauwallner equated this epigraphic and literary evidence with further epigraphic evidence from Aśoka's thirteenth Rock Inscription: on pp.15-17 of The Earliest Vinaya, he noted that the areas mentioned in this edict, to which he despatched emissaries, correspond to the areas of missionary activity mentioned in the Pali chronicles. Both sources, according to him, mention the North-West, West and South but omit the East, and he comments 'This is certainly no freak chance.' Lamotte's table (p.302) shows at least a superficial agreement between the places mentioned in both sources, but Gombrich is probably correct in commenting: 'The geographical identifications are too uncertain to help us'. [40] With the geographical identifications uncertain, Lamotte was sceptical of the notion that there was one concerted missionary effort in Aśokan times. He argued that the Buddhists were natural missionaries and would have spread Buddhism throughout India from the beginning. [41] Thus he concluded his study of the early Buddhist missions by stating 'Whatever might have been said, Aśoka was not directly involved in Buddhist propaganda.' [42] Gombrich, on the other hand, agrees with Frauwallner, and notes:

While Lamotte is right to point out that some of the areas visited, notably Kashmir, had Buddhists already, that does not disprove that missions could not be sent there. The chroniclers, as so often happens, had no interest in recording a gradual and undramatic process, and allowed history to crystallize into clear-cut episodes which could be endowed with edifying overtones; but this over-simplification does not prove that clear-cut events never occurred.[43]

The notion that there was a clear-cut missionary episode in the spread of Buddhism across India seems to be confirmed by the epigraphical record. L. S. Cousins has surveyed the references to the sects in inscriptions (pp.148-51), and noted that the related Vibhajjavadin sects (the Vibhajjavadin-s made up a subset of the ancient Sthaviras) were most widespread of all Buddhist sects in the first few centuries C.E. On the other hand, the other sects were distributed randomly across India. This is exactly what is to be expected if there was a gradual diffusion of Buddhism throughout India, as well as a concerted missionary effort by one ancient monastic community, which thereafter separated into separate sects due to the geographical dispersal. Cousins comments on the tradition of the Buddhist missions as follows:

It seems clear that whatever the traditions about these missions may or may not tell us about events in the third or second century BCE, they do certainly correspond to what we know of the geographical spread of the schools early in the first millennium CE. They must then have some historical basis. Vibhajjavadins really were the school predominant in Ceylon and Gandhara at an early date, as well as being present, if not predominant, in other parts of Central Asia, China, South India and South-East Asia by around the turn of the third century CE at the latest. No other school has a comparable spread at this date.[44]

It seems then that there is no reason to doubt that there was some sort of mission in the third century B.C. which set out from Vidiśa to the far North-West, West and South of India. Frauwallner thought that this missionary activity founded the Sarvastivadin sect in the North-West, as a result of Majjhantika's mission to Kaśmir and Gandhara,[45] whereas Cousins considers only the Vibhajjavada sects in the North-West and South. Was the Sarvastivadin sect of the North-West produced by a missionary effort that otherwise seems to have produced only Vibhajjavadin sects? This is certainly possible. Frauwallner made it quite clear that the formation of distinct communities ought to be distinguished from schools of thought: 'from the first we have stressed the principle that the foundation of communities and the rise of dogmatic schools are two quite separate things.' [46] This led him to conclude that the dogmatic affiliation of the Mulasarvastivadin and Sarvastivadin sects came later than the original foundation of these two as monastic communities.[47] It is possible then that different dogmatic affiliations could have been produced in the sects founded by missionary activity, and that the dogmatic affiliation to sarvastivada ideas by the community that came to be known as the 'Sarvastivadin' sect came about later. This seems to be shown by the fact that the literature of the Sarvastivadin sect is in many regards similar to the literature of the other Vibhajjavadin sects.

In the beginning of The Earliest Vinaya, Frauwallner notes that the Skandhaka section of the Sarvastivadin, Dharmaguptaka, Mahiśasaka and Pali Vinaya-s are 'strikingly close'; the Skandhaka of the Kaśyapiya school is not considered because it has not survived. [48] According to Frauwallner then, the Sarvastivadin Skandhaka is closer to the Skandhaka-s of sects known to be Vibhajjavadin in affiliation than it is to the Mulasarvastivadin Skandhaka. Elsewhere, Frauwallner has noted that the Sarvastivadin Abhidharma contains much that 'was held in common with the Pali school.' [49] And the Śariputrabhidharma, which according to Frauwallner is a Dharmaguptaka text, is also a development of the same material inherited by the Sarvastivadin and Pali schools. [50] This is again in contrast with the Abhidharma of the Mulasarvastivadins, which according to Frauwallner 'possessed only one Matṛka.' [51] The canonical literature of the sect in the North-West that came to be known as the Sarvastivadin-s is therefore closer to the Vibhajjavadin sects, particularly the literature of the Pali tradition.

It seems likely that all these sects share a common antecedent, which we can think of as the ancient Sthavira community of Vidiśa. Nevertheless, it is striking that only one of the sects produced by the missions adopted sarvastivada ideology, whereas the others seem to have been affiliated to the vibhajjavada. I think the best explanation of the evidence is that this sarvastivadin development must have occurred later on within the community founded by Majjhantika in Kaśmir-Gandhara. Originally, the Abhidharma literature of the missionary community was less fixed than its Sutra and Vinaya sections. It allowed considerable room for development. In Aśokan times the dogmatic outlook of the missionary community was vibhajjavada, but at a later date, sarvastivada ideology came to dominate in Kaśmir. Incidentally, the Mulasarvastivadin school, originally from Mathura according to Frauwallner, came to exist in the North-West and claimed that it originated from the mission of Madhyantika.[52] It is plausible to think that this claim -- almost certainly an interpolation into the Mulasarvastivadin Vinaya, as Frauwallner has shown -- was taken from the old Sarvastivadin-s of the North-West, and used as part of the Mulasarvastivadin strategy of claiming supremacy in the North-West. This claim would only have been borrowed if it was thought to be true, so it seems that yet another piece of evidence supports the accuracy of the Theravadin tradition of the missions.

The evidence for a Sthavira mission taking place in the third century B.C., probably from a school that was vibhajjavadin in the dogmatic sense, is very good. But was this mission related to Aśoka? Despite Lamotte's doubts, I think that the Aśokan inscriptions show that this must have been the case. The confusion on this point seems to have been caused by a failure of previous scholars, especially Lamotte, to distinguish Aśoka's references to his 'Dharma-ministers' (dhamma-mahamatta) from the evidence contained in the thirteenth Rock Edict. Lamotte's table on the sources of the missions (p.302) sums up the evidence, presented on the previous page, of the second Rock Edict, the fifth Rock Edict and the thirteenth Rock Edict. But RE II has nothing to do with missionary activity -- nor does RE V, which mentions the dhamma-mahamatta. In fact every mention of the dhamma-mahamatta limits them to Aśoka's Kingdom, and so they should have been, for according to the inscriptions they were involved in all sorts of activities which might be called 'social welfare', and which cannot have been carried out in other kingdoms.[53] Socially beneficial activity in other kingdoms is mentioned in RE II, which states that Aśoka provided medicines and medical herbs for men and cattle, and had wells dug, [54] but this falls well short of the activities of the dhamma-mahamatta. Whether or not Aśoka really carried out such beneficial acts in neighbouring kingdoms -- it might have been an exaggeration of his own righteous endeavours -- this activity can hardly have been carried out by his dhamma-mahamatta. In the same way, any missionary activity initiated by Aśoka cannot have been undertaken by his dhammamahamatta. This brings us to RE XIII: Aśoka tells us that he has achieved a 'Dharmavictory' (dhamma-vijaya) in his own kingdom as well as others, and even in places where his 'envoys' (duta-s) have not gone. [55] From this we can see that this victory must have been achieved by his envoys -- duta-s, not dhamma-mahamatta. Who were these envoys then, sent by Aśoka to the border areas of this kingdom, as well as to neighbouring kingdoms, through which he attained his victory of dhamma?

The obvious answer is that they were the people responsible for taking medicines and medicinal herbs to other Kingdoms, and for having wells dug there, mentioned in RE II. Supporting this idea is the fact that the areas outside Aśoka's Kingdom mentioned in RE XIII and RE II are almost identical: in RE XIII the kingdoms mentioned are those of Aṃtiyoka, the Yonaraja, and his four neighbours in the North-West, as well as the Choḍa-s, Paṃda-s and Taṃbapaṇṇiya-s in the South; in RE II, the Satiyaputra-s and Keraḍaputra-s are added to the list of southern kingdoms, and the neighbours of Aṃtiyoka are not named. This is almost an identical correspondence. Nevertheless, it is problematic think that the duta-s mentioned in RE XIII were merely carriers of medicines and supervisors of well-digging. After all, Aśoka says that through them he has achieved his dhamma-vijaya: can a 'Dharma-victory' have been achieved by the implementation of some social policies? In other words, would socially beneficial acts undertaken by Aśoka in his kingdom and elsewhere have induced him to claim that he had achieved a Dharma-victory? One might think that this is possible, because the sort of dhamma promoted by the Dharma-ministers was exactly this sort of socially beneficial action. But against this idea is the fact that in PE VII at Delhi-Topra, the same sort of socially beneficial deeds are declared to be 'of little consequence'. For with various comforts have the people been blessed both by former kings and by myself.' [56] After this declaration, Aśoka states that he has merely provided material needs so that the people will conform to the practice of dhamma. [57] We have then a distinction between Aśoka's ideal of dhamma and socially beneficial policies which might be called dhamma. Which of the two meanings of dhamma did Aśoka mean by the word in the compound dhammavijaya? I find it unlikely that Aśoka would have proclaimed a dhamma-vijaya because of his social policies, which in PE VII he claims are of little value. It is much more likely that when Aśoka spoke of his dhamma-vijaya, he had in mind a victory of dhamma in its higher meaning of a set of ethical practices and attitudes. In the Delhi-Topra edict, this ideal of dhamma is outlined as follows:

- Thereby they have made progress and will (be made to) progress in obedience to mother and father, in obedience to elders, in courtesy to the aged, in obedience to Brahmaṇas and Śramaṇas, to the poor and distressed, (and) even to slaves and servants. [58]

This ideal of the dhamma is outlined in more or less the same fashion by Aśoka in RE III, IV, IX, XI, and crucially, in RE XIII, the edict in which he claims his dhammavijaya. [59] This 'victory' was the spread of ideals such as respect to śramaṇa-s and brahmaṇa-s, obedience to mother and father, courtesy to slaves and servants etc. If the envoys who took these ideals to the distant corners of Aśoka's kingdom and beyond were not Dharma-ministers, who were they? Are we to believe that Aśoka had a class of officials who went out and taught what are essentially religious ideals? This is hardly likely. The more plausible answer is that the duta-s included the professional religious men and women to whom Aśoka was partial, i.e. the Buddhists. In other words, it is likely that the envoys who spread Aśoka's ideals included Buddhist monks and nuns. [60] There is even some indication in the Sinhalese chronicles that is indeed what happened. In chapter XI, the Mahavaṃsa describes how envoys were sent by Aśoka to King Devanaṃpiyatissa of Laṅka:

- The Lord of Men [Aśoka], having given a palm-leaf message (paṇṇakaraṃ) at the appropriate time for his friend [Devanaṃpiyatissa], sent envoys (dute) and this palm-leaf message concerning the true doctrine (saddhammapaṇṇakaraṃ), [which said:]

- Saying: "Carry out the consecration of my friend once more," having honoured his friend's ministers, he despatched them. [61]

There is no mentions of Buddhist monks and nuns in the imperial embassy of duta-s, but the implication is that if there were contacts such as this between Aśoka and his neighbouring kings, then Buddhists must have been involved, or would have followed soon afterwards. This is also indicated by the Dipavaṃsa which, although including a standard account of the missions sent by Moggaliputta (at VIII.4-13), also includes three accounts of the duta-s sent to Laṅka by Aśoka, each account describing how Mahinda arrived in Laṅka soon after the envoys, without any mention of Moggaliputta. These accounts imply that the envoys paved the way for the Buddhist monks who followed. In the most elaborate account, Dip XII.8ff, after describing how Aśoka sent gifts and a request that Devanaṃpiya of Laṅka should have faith in the triple jewel, it says that the thera-s of the Asokarama requested that Mahinda establish the faith in Laṅka. [62] And at Dip XI.41 and Dip XVII.91-92, it says that Mahinda arrived in Laṅka one month after the Aśokan envoys, without any mention of Moggaliputta. [63] It seems that the author of the Sinhalese chronicles, as well as Buddhaghosa, had various sources available to them recording different versions of the mission to Sri Lanka. [64]

The version that eventually became the orthodox account was of course the one that had Moggaliputta as the organiser of the missions. But the accounts in the Dipavaṃsa that do not mention Moggaliputta seem much more plausible in the light of the evidence from RE XIII. It is clear that some of the information in the chronicles is accurate: the name of the missionary monks, for instance, as the evidence at Saṭchi indicates. But the Sthavira tradition from which the missions came could hardly have made the monks merely part of Aśoka's ministerial envoys, or even following in the wake of these envoys -- they probably felt that they had to exaggerate the prominence of their tradition with the idea that Moggaliputta sent them. [65] We can conclude that the imperial envoys (duta-s) of Aśoka, which for him had effected a dhamma-vijaya, probably did include Buddhist monks. By welcoming these envoys, and heeding Aśoka's written requests that they take refuge in the triple jewel, the neighbouring kings maintained good relations with the mighty Indian emperor, and Aśoka himself propagated dhamma.[66]

The result of this long digression into the evidence for the Aśokan missions is that it appears that Frauwallner was correct to relate the evidence of the Sinhalese chronicles and inscriptions of Vidiśa to Aśoka's RE XIII. We can therefore date the arrival of the canonical texts in Sri Lanka to the middle of the third century B.C. Frauwallner has shown, however, that the lower limit of the early Buddhist literature can be pushed back even further. He noted that the Mahasaṅghikas had a version of the old Skandhaka, meaning that it must have been composed before the schism between themselves and the Sthaviras, which certainly occurred before the Aśokan missions. The Mahasaṅghika Vinaya also includes the account of the second council at Vaiśali: the old Skandhaka must have been composed, or at least redacted, after this council, and before the schism between the Sthaviras and Mahasaṅghikas.[67] Frauwallner thought that the old Skandhaka was probably composed shortly before or after this council, [68] which Gombrich reckons that it took place about 60 years after the Buddha's death, c. 345 B.C. [69] If this is correct, it means that very sophisticated literary tracts were being composed little more than half a century after the Buddha's death.

There is great significance in these investigations for the date of texts contained in the Suttapiṭaka. According to Frauwallner:

At the time of the compilation of the old Skandhaka work ...[70] the Buddhist tradition had already reached an advanced stage of development. A collection of sacred scriptures, including Dharma and Vinaya, was already in existence. The Vinaya included the Praṭimokṣa, narratives of the type of the Vibhaṅga and much material on the monastic rules, which the Buddha was said to have communicated to his disciples. The collection of Sutra, which existed on its side, was handed down by a regular machinery of transmission, and we can ascertain a number of texts which belonged to it already in that period. [71]

One Sutra text which Frauwallner singles out is the Aṭṭhakavagga (Sn IV), for the same story mentioning it is preserved in all the extant Vinayas; it probably belonged to the old Sutra collection. It is therefore possible that much of what is found in the Suttapiṭaka is earlier than c.250 B.C., perhaps even more than 100 years older than this. If some of the material is so old, it might be possible to establish what texts go back to the very beginning of Buddhism, texts which perhaps include the substance of the Buddha's teaching, and in some cases, maybe even his words. I have no intention of going into the important but complex question of what the Buddha did or did not teach. In the following, I will address the two questions posed on p.3, but I will at least attempt to show that some of the details of the Buddha's biography, namely those which record some of his activities as a Bodhisatta, have recorded accurate historical information about events that happened in the fifth century B.C. This will show that a careful use of textual sources is the only way to know anything about Buddhism in the pre-Aśokan period, and will lead to the conclusions that, contrary to what Schopen thinks, some material in the Suttapiṭaka is historically accurate and extremely old.

Various Suttas describe the Buddha's visits to the sages Aḷara Kalama and Uddaka Ramaputta, although the source for the account is probably the Ariyapariyesana Sutta (APS, M no.26). [72] Andre Bareau has translated a Chinese Sutra that corresponds to the APS as well as an account found in the Chinese version of the Dharmaguptaka Vinaya.[73] There are also versions of the narrative in the Mahasaṅghika Mahavastu [74] and the Mulasarvastivadin Saṅghabhedavastu.[75] It seems that the account of the training under the two teachers was embedded in the pre-sectarian Buddhist tradition, that is, if one accepts the idea that corresponding parts of the sectarian literature are likely to be pre-sectarian. There is also material on the two teachers scattered throughout the Suttapiṭaka. Scholars have generally accepted Bareau's opinion that the tradition of the two teachers' instruction to the Bodhisatta was a fabrication,[76] but more recently, Zafiropulo has shown that Bareau's arguments are fallacious.[77] If we are to take the tradition seriously, as we must do in the light of Zafiropulo's comments, we must also take into consideration the fragmentary information about the two teachers that is scattered throughout the Suttapiṭaka. I hope to show that a re-evaluation of the data on the two teachers makes two things quite clear. Firstly, some of the information on the two teachers cannot have been shared at a later date -- it must reflect a presectarian tradition. And secondly, a peculiar detail in the account of the Bodhisatta's training under the teachers shows that the two men must have existed. They must have been teachers of some repute in the fifth century B.C. in northern India, teachers of meditation who probably taught the Bodhisatta.

To show the former point, I will consider the information found in various sources concerning the location of Uddaka Ramaputta. Hsṃan tsang mentions some legendary evidence that relates Udraka Ramaputta to Rajagṛha; it seems that this represents the local tradition of the Buddhists living in the area of Rajagṛha.[78] This tradition is confirmed by the account of the Bodhisatta's training in the Mahavastu, which also places Udraka Ramaputra in Rajagṛha.[79] The coincidence between these two sources might have been reached in the sectarian period. There is, however, similar evidence in the Suttapiṭaka which makes it almost certain that the tradition must be presectarian. In the Vassakara Sutta, the Brahmin Vassakara, chief minister of Magadha, is said to visit the Buddha in Rajagaha and tell him that the raja Eḷeyya has faith in the samaṇa Ramaputta; the commentary names him as Uddaka Ramaputta.[80] Vassakara also appears in the Mahaparinibbana Sutta as the chief minister of King Ajatasattu of Magadha.[81] Vassakara's connection with Rajagaha and Magadha suggest that the raja Eḷeyya was a local chieftain in Magadha, probably situated somewhere near to Rajagaha. If so, it is likely that Uddaka Ramaputta was situated in the vicinity of Rajagaha.

The coincidence of this different evidence from the Theravadin and Mahasaṅghika sources, as well as the information of Hsṃan tsang, is not to be overlooked. It is inconceivable that this correspondence was produced by a later leveling of texts, for it is entirely coincidental -- different source materials, not corresponding Suttas, state or imply the same thing. It is hardly likely that a Mahasaṅghika monk or nun gained knowledge of obscure Pali Suttas, from which he deduced that Uddaka Ramaputta must have been based in Rajagaha, and after which he managed to insert this piece of information into the biographical account in the Mahavastu. And it is even more unlikely that a Theravadin Buddhist, in the early centuries A.D., studied the Mahasaṅghika Vinaya, from which he learnt that Udraka Ramaputra was based in Rajagṛha, following which he fabricated Suttas which contained circumstantial evidence which indirectly related Ramaputta to Rajagaha. Anyone who believes this version of textual history is living in cloud-cuckooland. It is clear that the information on the geographical situation of Uddaka Ramaputta must precede the Aśokan missions, and even the schism between Sthavira-s and Mahasaṅghika-s. This implies that the biographical tradition of the training under the two teachers goes back to the very beginning of Buddhism. It surely means that accurate historical information has been preserved, and suggests that Uddaka Ramaputta was based in Rajagaha, no doubt as a famous sage of Magadha. Incidentally, it is clear that this material has no normative value whatsoever, and so rebuts Schopen's claim that 'even the most artless formal narrative text' has a normative agenda.

Another detail, found in almost all the sectarian accounts of the training under the two teachers, can hardly have been produced by a later leveling of the Buddhist literature; it occurs in the account of the training under Uddaka Ramaputta. This account is identical in almost all regards to the description of the training under Aḷara Kalama. It tells us that the Bodhisatta first of all mastered the teaching, i.e. he gained an intellectual understanding of it,[82] after which he attained the direct realisation of the sphere of 'neither perception nor non-perception' through understanding (abhiṭṭa).[83] But the account of the training under Uddaka Ramaputta makes it clear that it was not Uddaka Ramaputta who had attained the sphere of neither perception nor non-perception, but Rama, the father or spiritual teacher of Uddaka.[84] This is seen in the following exchange. The Bodhisatta is said to have contemplated that Rama (not Ramaputta) did not proclaim (pavedesi) his attainment through mere faith, but because he dwelt (vihasi) knowing and seeing himself.[85] The corresponding passage in the account of the training under Aḷara uses the same verbs in the present tense (pavedeti, viharati), indicating that Aḷara was living and Rama was dead, and that Ramaputta had not attained and realised the dhamma he taught.

The same phenomenon is found in the rest of the passage. Thus the Bodhisatta is said to have asked Ramaputta: 'The venerable Rama proclaimed (pavedesi) [his attainment], having himself realised this dhamma to what extent (kittavata)?' [86] The reply, of course, is as far as nevasaṭṭanasaṭṭayatana. The Bodhisatta is then said to have contemplated that not only did Rama have faith, energy, mindfulness, concentration and insight, but that he too possesses these virtues. And at the end of the episode, Uddaka Ramaputta is reported to have said: 'Thus the dhamma that Rama knew (aṭṭasi), that dhamma you [the Bodhisatta] know; the dhamma you know, that dhamma Rama knew.' [87] This is different from the corresponding speech that Aḷara is reported to have made to the Bodhisatta: 'Thus the dhamma I know (janami), that dhamma you know; the dhamma you know, that I know.' [88] And whereas Aḷara is willing to establish the Bodhisatta as an equal to him (samasamaṃ), so that they can lead the ascetic group together (imaṃ gaṇaṃ pariharama ti),[89] Uddaka acknowledges that the Bodhisatta is equal to Rama, not himself (iti yadiso ramo ahosi tadiso tuvaṃ), and asks the Buddha to lead the community alone (imaṃ gaṇaṃ parihara ti).[90]

The distinction between Uddaka Ramaputta and Rama is also found in the Sarvastivadin, Dharmaguptaka, and Mahasaṅghika accounts of the Bodhisattva's training.[91] Although the Saṅghabhedavastu (plus parallel Tibetan translations) and the Lalitavistara fail to distinguish Ramaputta from Rama,[92] this must be because of a later obfuscation of the tradition. Exactly the same mistake has been made by I. B. Horner, the PTS translator of the Majjhima Nikaya, who has been duped, by the repetitive oral style, into believing that the accounts of the training under Aḷara and Uddaka must be the same apart from the difference between the names of the two men and their meditative attainments.[93]

It hardly needs to be pointed out that there is no need to trouble over these details in an oral tradition where adjacent passages are often composed in exactly the same way, one passage frequently being a verbatim repetition of the previous one with a minor change of one or two words. The tendency for reciters of this autobiographical episode would have been to make the two accounts identical bar the substitution of Uddaka's name for Aḷara's. A conscious effort has been made to distinguish Uddaka Ramaputta from Rama, and not to let the repetitive oral style interfere with this. This effort must surely go back to the beginning of the pre-sectarian tradition of composing biographical Suttas, and the distinction can only be explained if Rama and Ramaputta were two different people. Otherwise, it is part of an elaborate hoax, and there is no reason for such a hoax.

Bareau maintained that the correspondence between the two descriptions of the training under each of the teachers proved their artificial (i.e. unhistorical) nature.[94] But repetition is normal in Pali oral literature. And it seems that the two parallel accounts, having preserved the important distinction between Ramaputta and Rama, rather than leaving an impression of 'contrivance', have preserved valuable historical information. The conclusion is that the three men were real.[95] It is hardly likely that Buddhists got together a few hundred years after the Buddha's death and decided to make up the idea that Rama and not Ramaputta had attained the state of neither perception nor non- perception, and then had such an influence that the idea found its way into recensions of texts being made in regions as far apart as central Asia and Sri Lanka. The idea must have been in the Buddhist tradition from the beginning, and can only be explained as an attempt to remember an historical fact. There is no other sensible explanation. It is also worth pointing out that if this biographical material is so old and really does represent an attempt to record historical facts, then it means that this portion of the Bodhisatta's biography is most likely to be true. It is likely that the Bodhisatta really was taught by Aḷara Kalama and Uddaka Ramaputta.

And finally, I attempted to show that some of the information preserved in the literature of the various Buddhist sects shows that historical information about events occurring in the fifth century B.C. has been accurately preserved. The corresponding pieces of textual material found in the canons of the different sects (especially the literature of the Pali school, which was more isolated than the others) probably go back to pre-sectarian times. It is unlikely that these correspondences could have been produced by an endeavour undertaken in the sectarian period, for such an endeavour would have required organisation on a scale which was simply inconceivable in the ancient world. We can conclude that a careful examination of the early Buddhist literature can reveal aspects of the pre-Aśokan history of Indian Buddhism. The claim that we cannot know anything about early Indian Buddhism because all the manuscripts are late is vacuous, and made, I assume, by those who have not studied the textual material properly.

-ooOoo-

Étienne Lamotte, however, proposed a different solution: he dismissed Frauwallner's theory about the difference between Sarvastivadin and Mulasarvastivadin, by claiming, without presenting any corroborating evidence, that the Mulasarvastivadin Vinaya was simply the version of the Sarvastivadin Vinaya completed at a later date in Kaśmir (p.178). This explanation, however, leaves too many questions unanswered. For example, if the Mulasarvastivadin Vinaya was a later recension of an earlier Sarvastivadin Vinaya, then why is it so different? And why did the sect who revised the work change their name from Sarvastivadin to 'Mulasarvastivadin'? The only answer to the last objection to Lamotte's thesis is that the name was changed to 'Mulasarvastivadin' by the reforming sect who expanded the old Sarvastivadin Vinaya, and who thus used the prefix mula in order to differentiate themselves from the old Sarvastivadin-s and create a new sect, i.e. an explanation which corresponds to the second of my proposed solutions above. But there is no clear evidence for the theory that the Mulasarvastivadin Vinaya is an expansion of the Sarvastivadin Vinaya and that the Mulasarvastivadin-s were an offshoot of the Sarvastivadin-s.[96]

We must therefore look for a different answer to the problem, an answer along the lines of my proposed first solution above, i.e. that two Sarvastivadin groups came into contact and had a dispute. Such an explanation would simply be a reworking of Frauwallner's hypothesis -- he proposed two different communitites, one from Kaśmir and one from Mathura, both of whom came to accept sarvastivadin ideas, but he did not state that they had a dispute.[97] I am proposing that the Mathura school moved to Kaśmir and disputed with the existing community there, and I think we can detect such a dispute in the Chinese works mentioned by Lamotte pp.174- 75. First of all, however, I will show that these works cited by Lamotte support Frauwallner's theory.

Lamotte did not offer any explanation of the peculiar fact that Kumarajiva, the fifth century translator of the Mahaprajṭaparamitopadeśa, knew of the existence of two different Vinaya-s, one from Mathura and one from Kaśmir. According to Kumarajiva, the Vinaya of Kaśmir contained 10 sections, and we can deduce that this was the Sarvastivadin Vinaya, for he was himself from Kaśmir and translated the Sarvastivadin Vinaya into Chinese in 404-05 A.D. (Lamotte p.168). Moreover, the Sarvastivadin Vinaya was also known as the 'Vinaya in ten sections' (Daśadhyaya, Lamotte p.168). This means that it is likely that what he calls the Vinaya of Mathura in 80 sections was the Mulasarvastivadin Vinaya, that is, if the information of Kumarajiva and the two other Chinese authors mentioned by Lamotte (p.175) definitely concerns the Sarvastivadin and Mulasarvastivadin traditions of Kaśmir in the fifth century A.D. All three Chinese works mentioned by Lamotte (pp.174-75) knew that the original Vinaya (of Upali/Kaśyapa) consisted of 80 sections. Sêng yu and Hui Chao related this Vinaya to the patriarchal lineage ending in Upagupta, the fifth patriarch of the Buddhist community of Mathura.[98] Kumarajiva went further and stated that the Vinaya in 80 sections was that of the community in Mathura. So we have good reasons to suppose that two Vinaya-s -- and therefore two monastic communities, in some way similar -- were known in Kaśmir, and that one had come from Mathura. Lamotte's theory simply brushes over this fact, whereas Frauwallner adduces good evidence to show that the Mulasarvastivadin sect's connection with Kaśmir is late, and written onto an earlier church history of Mathura (1956 pp.26-36).

The dispute between the adherents of these two Vinaya-s is just about detectable in the words of Kumarajiva and Sêng yu. Kumarajiva, taught in the tradition of the Sarvastivadin Vinaya, tells us that the Vinaya of Kaśmir (i.e. the Sarvastivadin Vinaya) had only ten sections, but that it also had a Vibhaùa consisting of 80 sections. Why did he do this? Why did he state that the Vinaya has a commentary consisting of 80 sections? It might be a redundant statement, but I think that the evidence suggests otherwise. From his words (Lamotte p.174), it seems that he was aware of the claim that the original Vinaya consisted of 80 sections. He was also aware of the fact that the Vinaya of Mathura, probably the Vinaya of a very old Buddhist community, consisted of 80 sections. Surely his statement is that of an apologist, forced into making it because there were others who criticised the Sarvastivadin Vinaya of Kaśmir for lacking the full 80 sections.

We can see the nature of such a critique in Sêng yu's Ch'u san tsang chi chi: he states that the Vinaya with 10 sections was a reduction of the Vinaya in 80 sections, undertaken by Upagupta for the sake of those with 'weak faculties'. Sêng yu is blatantly polemical, arguing against the worth of the Sarvastivadin tradition. Thus, Kumarajiva stated that there is nothing amiss with the fact that one Vinaya (his Vinaya) had only 10 sections, whereas Sêng yu said that there was something deficient in it -- it was an offshoot of the Vinaya in 80 sections for those with a weaker disposition. Kumarajiva was the apologist, asserting the antiquity of his Kaśmirian Vinaya in the face of a rival, whereas Sêng yu was the inclusivist, attempting to include the Vinaya of Kaśmir and its tradition within the tradition of Mathura. It seems that in fifth century Kaśmir, there was some quarrel between the adherents of two different communities -- one original to Kaśmir, which defended its position, and the other more recently arrived community which arguing that the Kaśmir tradition was an offshoot of itself.

Why would two groups have clashed in this way? Surely there would have been no need for one community, when moving to another area, to attempt to usurp the position of the resident community. I suggest that everything makes sense if we accept that the disputed issue concerned the ownership of the sarvastivada idea: one group accepting sarvastivada ideas had moved from Mathura to Kaśmir, and there encountered another community which at some point adopted a similar sarvastivada ideology. In response to this, and considering itself to be the original source of the sarvastivada, it labelled itself the 'Mulasarvastivada'. If this argument, which I claim can be detected in the words of Kumarajiva and Sêng yu, was still fresh in the fifth century A.D., then it seems that the dispute broke out some time after the two groups had co-existed in the same area: Frauwallner noted that interpolations into the Mulasarvastivadin Bhaisajyavastu (indicating a relocation from Mathura to the North West) were probably made between c.150-300 A.D.[99] Therefore, we can posit a period in which sarvastivada ideas circulated between the two groups. But it is more likely that the sect to whom the idea belonged at the beginning of this contact was the sect from Mathura: the mission that led to the origination of Majjhantika's community in Kaśmir/Gandhara was probably vibhajjavadin in the early period, and we can guess, from the name of the sect itself, as well as from Sêng yu's aggressive stance and Kumarajiva's seemingly defensive position, that the Mulasarvastivadin community was more irked by the dispute. All this is of course highly speculative, but if we are to explain the Mulasarvastivadin/Savastivadin distinction as the result of sectarian dispute, then some explanation must be found.

It seems to me as if the Chinese works mentioned by Lamotte support Frauwallner's theory and the reworking of it I suggest here. At the least, they show that there was a problem in the Buddhist traditions of Kaśmir concerning two different Vinaya-s, whereas Lamotte failed even to notice that this is a historical problem; Frauwallner's theory seems to be the best explanation of the evidence. More recently, Enomoto has recently argued that the Sarvastivadin sect was no different from the Mulasarvastivadin -- he argues that the two words mean the same thing, but this does not explain the odd facts: two different Vinaya-s, two similar names, and two explanations of their relationship in fifth century Chinese works. However, Enomoto's argument is flawed: it begins with the late and unreliable evidence of the fanciful etymology of the word mulasarvastivadin in Śakyaprabha's Prabhavati (c.8th century), as well as equally unreliable evidence in the colophons of this work and others by Śakyaprabha -- all are inconsistent in the use of the prefix mula-.[100] He then attempts to show that Yi-jing used the words Sarvastivada and Mulasarvastivada interchangeably.[101] But it seems to me that the section of Yi-jing translated by him does not support such a view. The important section reads:

What are treated in this work [102] mostly resemble the Shi-song-lu. The three different sects divided from the Sarvastivada sect -- 1. Dharmaguptaka; 2. Mahiśasaka; 3. Kaśyapiya -- are not prevalent in the five parts of India... However, the Shi-song-lu does not belong to the 'Mula-sarvastivada' sect, either.[103]

The proximity of the title Shi-song-lu and the Sarvastivada sect in sentences one and two suggest that the former is the work of the latter, and from the last sentence, we know that this was not a work of the Mulasarvastivada sect. The translation of Enomoto certainly does not say that "the 'Mulasarvastivada' sect was divided into four sects: Dharmaguptaka, Mahiśasaka, Kaśyapiya and 'Mulasarvastivada' itself", or that "what is here called the 'Sarvastivada' sect is the same as the Mulasarvastivada sect."[104] On the contrary, it relates sects which, I have argued, originated from the same missionary endeavour, and distinguishes them from the Mulasarvastivadin-s. Enomoto's theory does not make sense and it does not explain the difficulties. It seems then that Frauwallner's explanation of the difference between Sarvastivadin and Mulasarvastivadin explains most of the facts while leaving fewer unresolved problems. Some of the remaining problems I have attempted to solve by showing that the name 'Mulasarvastivadin' originated in the course of sectarian debate, and this seems to offer the best explanation of the various facts.

-----*-----

(All Pali citations are from Pali Text Society editions; citations from the Aśokan edicts are from Hultzsch's edition).

Bareau, André. 1963, Recherches sur la biographie du Buddha dans les Sutrapitaka et les Vinayapitaka anciens I: De la quête de l'éveil à la conversion de Sariputra et de Maudgalyayana. Paris: École française d'Extrême-Orient.

Beal, Samuel. 1981, Si-Yu-Ki: Buddhist Records of The Western World. Delhi: Motilal Banarsidass. First Published: London (1906).

Bechert, Heinz. 1973, 'Notes on the Formation of Buddhist Sects and the Origins of Mahayana'. In German Scholars on India (Contributions to Indian Studies Vol. I), pp.6- 18. Varanasi: The Chowkhambha Sanskrit Series Office (1973).

Bronkhorst, Johannes. 1993, The Two Traditions of Meditation in Ancient India. New Delhi: Motilal Banarsidass. First published, Stuttgart: Steiner Verlag (1986).

Collins, Steven. 1990, 'On the very idea of the Pali Canon'. Journal of the Pali Text Society XV, pp.89-126.

Cousins, L. S. 2001, 'On the Vibhajjavadins. The Mahiṃsasaka, Dhammaguttaka, Kassapiya and Tambapaṇṇiya branches of the ancient Theriyas'. Buddhist Studies Review, vol. 18 no.2 pp.131-82.

Frauwallner, Erich. 1956, The Earliest Vinaya and the Beginnings of Buddhist Literature. Rome.

Frauwallner, Erich. 1995, Studies in Abhidharma Literature and the Origins of Buddhist Philosophical Systems. New York: State University of New York Press.

Gombrich, Richard. 1971, Precept and practice: traditional Buddhism in the rural highlands of Ceylon. Oxford: Clarendon Press.

Gombrich, Richard. 1988, Theravada Buddhism: A social history from ancient Benares to modern Colombo. London and New York: Routledge.

Gombrich, Richard. 1990, 'Making Mountains Without Molehills: The Case of the Missing Stupa'. Journal of the Pali Text Society, vol. XV pp.141-43.

Gombrich, Richard. 1992, 'Dating the Historical Buddha: A Red Herring Revealed'. In The Dating of the Historical Buddha Part 2, ed. Heinz Bechert, pp.237-59. Göttingen: Vandenhoeck & Ruprecht.

Hallisey, Charles. 1990, 'Apropos the Pali Vinaya as a Historical Document: A Reply to Gregory Schopen'. Journal of the Pali Text Society, vol. XV pp.197-208.

Horner, I. B. 1954, The Collection of Middle Length Sayings Vol.I: the first fifty discourses. London: Luzac and Company Ltd.

Hultzsch, E. 1991, Corpus Inscriptionum Indicarum Vol. I: Inscriptions of Aśoka. New Delhi: The Director General Archaeological Survey of India. First published, Oxford: Clarendon Press for the Government of India (1925).

Jones, J. J. 1949-56, The Mahavastu: translated from the Buddhist Sanskrit (Sacred Books of the Buddhists volumes 16, 18, 19). London: Luzac and Company Ltd.

Bhikkhu Ñaṇamoli and Bhikkhu Bodhi. 1995, The Middle Length Discourses of the Buddha: A Translation of the Majjhima Nikaya. Boston: Wisdom Publications.

Norman, K.R. 1978, 'The role of Pali in early Sinhalese Buddhism', in Collected Papers II, pp.30-51. First published in Heinz Bechert ed., Buddhism in Ceylon and Studies on Religious Syncretism, Göttingen, pp.28-47.

Norman, K.R. 1982, Aśokan sila-thaṃbha-s and dhamma-thaṃbha-s. In Acarya-vandana (D.R.Bhandarkar Birth Centenary Volume), Calcutta 1982, pp.311-18. Reprinted in Collected Papers volume II, pp.224-33.

Norman, K.R. 1983, Pali literature: including the canonical literature in Prakrit and Sanskrit of all the Hinayana schools of Buddhism. Wiesbaden: Harrassowitz

Norman, K.R. 1997, A Philological Approach To Buddhism: the Bukkyo Dendo Kyokai Lectures, London: School of Oriental and African Studies.

Norman, K.R. 1990-2001, Collected Papers I-VII. Oxford: Pali Text Society.

Schopen, Gregory. 1997, Bones, Stones and Buddhist Monks. Collected Papers on the Archaeology, Epigraphy, and Texts of Monastic Buddhism in India. University of Hawaii Press.

Skilling, Peter. 1981-82a, 'Uddaka Ramaputta and Rama'. Pali Buddhist Review 6.2, pp.99-105.

Thomas, E. J. 1993, The Life of Buddha as Legend and History. Delhi: Motilal Banarsidass. First published, London: Kegan Paul (1927).

Vetter, Tilmann. 1988, The Ideas and Meditative Practices of Early Buddhism. Leiden: E.J.Brill.

Waldschmidt, Ernst. 1950-51, Das Mahaparinirvanasutra. Text in Sanskrit und Tibetisch, verglichen mit dem Pali nebst einerṃbersetzung der chinesischen Entsprechung im Vinaya der Mulasarvastivadins auf Grund von Turfan-Handschriften. Berlin: Akademie- Verlag.

Willis, Michael. 2001, 'Buddhist Saints in Ancient Vedisa'. Journal of the Royal Asiatic Society, Series 3, II, 2, pp.219-28.

Zafiropulo, G. 1993, L'illumination du Buddha: de la quête à l'annonce de l'Éveil: essais de chronologie relative et de stratigraphie textuelle. Innsbruck : Institut fur Sprachwissenschaft der Universität Innsbruck.

-----*-----

[1] Dip XX.20-21, Mhv XXXIII.100-01; See Collins p.97.

[2] Accepting Richard Gombrich's dates; see below p.11 n.32.

[3] Norman 1978 p.36.

[4] On this evidence, see below pp.19-20.

[5] It is unlikely that the Abhidharma works of various schools were fixed at this date. See below p.15.

[6] Schopen p.3.

[7] Schopen p.3.

[8] Schopen pp.23-24.

[9] Schopen p.56. See also Schopen's comments, p.71 n.50: 'We do know, however, that from the very beginning of our actual epigrahical evidence (Bharhut, Saṭci, etc.), a large number of monks were doing

[10] Schopen p.31.

[11] Hallisey p.208.

[12] See in particular his article 'Monks and the Relic Cult in the Mahaparinibbana-sutta: An Old Misunderstanding in Regard to Monastic Buddhism' (= Schopen 1997 pp.99-113).

[13] See in particular his article 'Two Problems in the History of Indian Buddhism: The Layman/Monk Distinction and the Doctrines of the Transference of Merit' (= Schopen 1997 pp.23-55), parts of which I will consider below.

[14] Schopen pp.25-26.

[15] Schopen pp.28-29.

[16] Schopen p.29.

[17] Schopen p.29.

[18] Schopen pp.86-98.

[19] Schopen pp.89-91.

[20] See D II.142.5ff: ...catummahapathe raṭṭo cakkavattissa thupaṃ karonti. evaṃ kho Ananda raṭṭo cakkavattissa sarire paṭipajjanti. yatha kho Ananda raṭṭo cakkavattissa sarire paṭipajjanti evaṃ tathagatassa sarire paṭipajjitabbaṃ. catummahapathe tathagatassa thupo katabbo; and D II.164.28: aham pi arahami bhagavato sariranaṃ bhagaṃ, aham pi bhagavato sariranaṃ thupaṭ ca mahaṭ ca karissami. The Sanskrit Mahaparinirvaņa Sutra edited by Waldschmidt also mentions śarirastupa-s in portions of text which correspond to these Pali references: 36.7 corresponds to D II.142.5, 50.5 corresponds to D II.142.5. The compound śarirastupa also appears at 46.7, 50.16, 50.20, 51.9, 51.22.

[21] Hultzsch p.165: (A) devanaṃpiyena piyadasina lajina chodasavasabhisitena, budhasa konakamanasa thube dutiyaṃ vaḍhite.

[22] Hultzsch p.164: (A) ...atana agacha mahiyite hida budhe jate sakyamuni ti.

[23] Waldschmidt p.388. The Pali version is only slightly different: idha tathagato jato ti (D II 140.20 = A II 120.24).

[24] D II 88.28ff = Ud 89.20 = Vin I 229.35; A V.269-73. On these passages see Gombrich 1971 p.267 and p.272.

[25] Schopen p.91.

[26] Gombrich 1990 pp.141-142, Hallisey pp.205-206. It seems to me that Hallisey has made it clear that: 'Buddhaghosa, Sariputta, and the otherṭikacariyas did not include the observances concerning stupas and bodhi trees among the observances specified in the Vinaya itself' (p.205). This does not explain the passage in the Visuddhimagga quoted by Schopen p.88, which still presents difficulties about the exact meaning of the compound khandhakavattani, although Hallisey notes: 'Perhaps it grouped a range of practices according to their family resemblances, rather than by their common origin in specific parts of the Vinaya.' (p.206).

[27] Gombrich 1990 p.143.

[28] Norman 1997 p. 140.

[29] Norman 1997 p.140.

[30] Norman 1997 p.140.

[31] Oldenberg 1879, p.xlviii.

[32] Accepting Richard Gombrich's dating of the Buddha: '...the Buddha died 136 years before Aśoka's inauguration, which means in 404 B.C.' (1992 p.246). Gombrich estimates the margin of error to be 7 years before to 5 years after this date, i.e. 411-399 B.C. (p.244). But he also notes that uncertainty about the date of Aśoka widens the margin of error, making the upper limit 422 B.C. K.R. Norman comments: 'If we take an average, then the date is c.411 ± 11 B.C.E.' (Norman 1999 p.467).

[33] Schopen p.26.

[34] Schopen p.26 on A. Bareau, Les sectes bouddhiques du petit véhicule (Paris, 1955).

[35] See below pp.19-20 on the notion that the Thera Moggaliputta sent the missions.

[36] Frauwallner 1956 pp.13-14.

[37] Willis p.226 n.26.

[38] Willis pp.222-23.

[39] Dip XII.35ff, Mhv XIII.18-20.

[40] Gombrich 1988 p.135.

[41] Lamotte p.297.

[42] Lamotte p.308.

[43] Gombrich 1988 p.135.

[44] Cousins p.169.

[45] Frauwallner 1956 p.22: 'The mission of Kassapagotta, Majjhima and Dundubhissara gave origin to the Haimavata and Kaśyapiya. The mission of Majjhantika led to the rise of the Sarvastivadin. The Dharmaguptaka school is perhaps issued from the mission of Yonaka-Dhammarakkhita.'

[46] Frauwallner 1956 p.38.

[47] He says this on p.38 in his discussion about the difference between Sarvastivadin and Mulasarvastivadin, but it applies in general to his thought on the dogmatic affiliation of all the sects resulting from the missions.

[48] Frauwallner 1956 p.2.

[49] Frauwallner 1995 p.37.

[50] See the chapter on the Śariputrabhidharma in Frauwallner 1995.

[51] Frauwallner 1956 p.39.

[52] Frauwallner 1956, pp.26-31, especially p.31: 'We come thus to the conclusion that the episode of Madhyantika and of the conversion of Kaśmir represents a late interpolation in the Vinaya of the Mulasarvastivadin.' In the light of Frauwallner's work, the Sarvastivada inscriptions of Mathura mentioned by Lamotte, p.523, might be those of the school that in the North-West came to be known as the Mulasarvastivadin-s. See the appendix for a further consideration of the Sarvastivadin/Mulasarvastivadin issue.

[53] They are mentioned in RE V, RE XII and PE VII; it would have been beyond the jurisdiction of a visitor to another kingdom to carry out some of these duties. ś

[54] RE II (Shahbazgarhi, Hultzsch p.51): (A) ...duvi 2 chikisa kriṭa manuśa-chikisa... paśu-chikisa cha (B) oshaḍhani manuśopakani cha paśopakani cha yatra yatra nasti savatra harapita cha vuta cha (C) kupa cha khanapita pratibhogaye paśu-manuśanaṃ.

[55] RE XIII (Shahbazgarhi, Hultzsch p.68/211): (S) yatra pi Devanaṃpriyasa duta na vrachaṃti...

[56] Hultzsch's translation, p.135, of PE VII (p.132): (U) la... esa paṭibhoge nama (V) vividhaya hi sukhayanaya pulimehi pi lajihi mamaya cha sukhayite loke.

[57] PE VII, Hultzsch p.132: (W) imaṃ chu dhammanupaṭipati anupaṭipajaṃtu ti etadatha me esa kaṭe.

[58] Hultzsch's translation, p.136, of PE VII (p.133): (FF) Devanaṃpiye P...s. laja hevaṃ aha (GG) yani hi kanichi mamiya sadhavani kaṭani taṃ loke anupaṭipaṃne taṃ cha anuvidhiyaṃti (HH) tena vaḍhita cha vaḍhisaṃti cha mata-pitisu sususaya gulusu sususaya vayo-mahalakanaṃ anupaṭipatiya babhanasamanesu kapana-valakesu ava dasa-bhaṭakesu saṃpaṭipatiya.

[59] In RE XIII, the crucial passage outlining his dhamma is found in section G, Hultzsch p.67/208.

[60] Erich Frauwallner related the Buddhist missions to Aśoka precisely because of the reference to duta-s in RE XIII (1956 p.15 n.1). He did not mention the evidence in the chronicles for the duta-s of Aśoka, however.

[61] Mhv XI.33-36: datva kale sahayassa paņṇakaraṃ narissaro, dute pahesi saddhammapaņṇakaram imam pi ca (33). "ahaṃ buddhaṭ ca dhammaṭ ca saṅghaṭ ca saraņaṃ gato, upasakattaṃ desesiṃ sakyaputtassa sasane (34), tvam pi'mani ratanani uttamani naruttama, cittam pasadayitvana saddhaya

[62] Dip XII.8: asokarame pavare bahu thera mahiddhika, laṅkatalanukampaya mahidaṃ etad abravuṃ (8).

Dip XVII.91cd-92ab: dutiyabhiseke tassatikkanta tiṃsarattiyo, mahidogaņa pamokkho jambudipa idhagato.

[64] As Norman points out (1983 p.118).

[65] Although there is every possibility that Mogalliputta, thera of the Asokarama, aided Aśoka in organising the missions.

[66] On the idea that Moggaliputta sent out the missions, Frauwallner states: '...we must remember that the data of the Sinhalese chronicles are uncertain on this point.' (1956 p.17). He concludes: 'The mother community tried apparently to enhance the glory of its patriarch by putting on his merit the sending out of the missions.' (1956 p.18).

[67] Frauwallner 1956 p.54.

[68] Frauwallner 1956 p.67: 'It must have been composed shortly before or after the second council'.

[69] Gombrich 1992 p. 258: 'We may thus date the Second Council round 60 A.B. or round 345 B.C.; the dates are very approximate and the precise margin of error incalculable...'.

[70] At this point Frauwallner dates the old Skandhaka according to older views about the date of the second council, c.100 years after the Buddha's death. More recent research has modified this date somewhat; I follow Gombrich's date of c.345 B.C. for the second council.

[71] Frauwallner 1956 p.153.

[72] The Suttas including this account are the Maha-Saccaka Sutta (M no.36), the Bodhi-Rajakumara Sutta (M no.85) and the Saṅgarava Sutta (M no.100).

[73] Bareau pp.14-16.

[74] Mvu II.118.1ff.

[75] SBhV I.97.4ff; Skilling points out that there is a Tibetan translation of this SBhV account, as well as a 'virtually identical' Mulasarvastivadin version, preserved in the Tibetan translation of the Abhini ṣkramaņa-Sutra (Skilling p.101).

[76] Vetter p.xxii, Bronkhorst p.86; Bareau sums up his view as follows: 'Personnages absents, morts même avant que leurs noms ne soient cités, ils sont probablement fictifs. Plus tard, on s'interrogea sur ces deux mystérieux personnages et l'on en déduisit aisément qu'ils n'avaient pu être que les maîtres auprès desquels le jeune Bodhisattva avait étudié.' (pp.20-21).

[77] Zafiropulo pp.22-29.

[78] Si-Yu-Ki (Beal , Part II p.139ff).

[79] Mvu II.119.8.

[80] Mp III.164.23: samaņe ramaputte ti uddake ramaputte.

[81] D II.72.9ff = A IV.17.11ff (Sattakanipata, anusayavagga, XX). He also appears in the Gopakamogallana Sutta (M III.7ff), which is set in Rajagaha. At Vin I 228 (= D II 86.31ff, Ud 87), he and Sunidha are in charge of the construction of Paṭaligama's defences.

[82] M I.165.22ff: so kho ahaṃ bhikkhave nacirass' eva khippam eva taṃ dhammaṃ pariyapuņiṃ. so kho ahaṃ bhikkhave tavataken' eva oṭṭhapahatamattena lapitalapanamattenaṭaņavadaṭ ca vadami theravadaṭ ca, janami passamiti ca paṭijanami ahaṭ c' eva aṭṭe ca.

[83] M I.166.4ff: ...yan nunahaṃ yaṃ dhammaṃ Ramo sayaṃ abhiṭṭa sacchikatva upasampajja viharami ti pavedeti, tassa dhammassa sacchikiriyaya padaheyyan ti? so kho ahaṃ bhikkhave nacirass' eva khippam eva taṃ dhammaṃ sayaṃ abhiṭṭa sacchikatva vihasiṃ.

[84] Skilling discusses this in detail; the point had been made earlier by Thomas p.63 andṭaṇamoli and Bodhi p.258 n.303.

[85] M I.165.27ff: na kho ramo imaṃ dhammaṃ kevalaṃ saddhamattakena sayaṃ abhiṭṭa sacchikatva upasampajja viharami ti pavedesi, addha ramo imaṃ dhammaṃ janaṃ passaṃ vihasi ti.

[86] M I.165.32ff: kittavata no avuso ramo imaṃ dhammaṃ sayaṃ abhiṭṭa sacchikatva upasampajja [VRI: viharamiti] pavedesi ti?

[87] M.I.166.22ff: iti yaṃ dhammaṃ ramo aṭṭasi, taṃ tvaṃ dhammaṃ janasi; yaṃ tvaṃ dhammaṃ janasi, taṃ dhammaṃ ramo aṭṭasi.

I leave dhammaṃ untranslated here because it indicates the meditative sphere attained by both Rama and the Buddha. Before this, the Buddha is said to have mastered the dhamma intellectually (165.24 = 164.4-5; see n.68), which can hardly mean a meditative attainment and must refer to an intellectual understanding.

[88] M.I.165.3ff: iti yahaṃ dhammaṃ janami, taṃ tvaṃ dhammaṃ janasi; yaṃ tvaṃ dhammaṃ janasi, tam ahaṃ dhammaṃ janami.

[89] M I.165.5ff: iti yadiso ahaṃ tadiso tuvaṃ, yadiso tuvaṃ tadiso ahaṃ. ehi dani avuso ubho va santa imaṃ gaņaṃ pariharama ti. iti kho bhikkhave aḷaro kalamo acariyo me samano antevasiṃ maṃ samanaṃ attano samasamaṃṭhapesi, uḷaraya ca maṃ pujaya pujesi.

[90] M I.166.24ff: iti y diso ramo ahosi tadiso tuvaṃ, yadiso tuvaṃ tadiso ramo ahosi. ehi dani avuso tvaṃ imaṃ gaņaṃ parihara ti. iti kho bhikkhave udako ramaputto sabrahmacari me samano acariyaṭṭhane ca maṃṭhapesi, uḷaraya ca maṃ pujaya pujesi.

[91] See Skilling, pp.100-102.

[92] Skilling p.101.

[93] Horner pp.209-10. Jones (p.117), translator of the Mahavastu, preserves the distinction between Rama and Ramaputra, but fails to notice that in the Mahavastu, Ramaputra does not establish the Bodhisattva as an equal to him: it says that he established the Bodhisattva as the teacher (Mhv II 120.15: acaryasthane sthapaye). Jones translates: 'Udraka Ramaputra ... would make me a teacher on an equal footing with himself' (p.117).

[94] Bareau p.20: 'Mais le parallélisme avec l'épisode suivant, l'ordre trop logique et le choix trop rationnel des points de doctrine d'Aḷara Kalama et d'Udraka Ramaputra nous laissent un arrière-goût d'artifice qui nous rend ces récits suspects.'

[95] Zafiropulo (p.25) does not point out the difference between Rama and Ramaputta, but on the stereotyped description of the training under the two teachers he comments: 'Justement cela nous semblerait plutôt un signe d'ancienneté, caractéristique de la transmission orale primitive par récitations psalmodiées'.

[96] According to Frauwallner, 'the Vinaya of the Sarvastivadin largely agrees with the Vinaya of the other missionary schools and forms with them a close group, while the Vinaya of the Mulasarvastivadin shows considerable differences.' (1956 p.38).

[97] Frauwallner 1956 p.40: 'They were at first two independent communities of different origin... Later on both communities grew into one school throught their accepting of the theories of the philosophicaldogmatic school; but they never completlely lost their individualities.'

[98] Frauwallner 1956 pp.28-31.

[99] Frauwallner 1956 p.36.

[100] Enomoto pp.240-42.

[101] Enomoto p.243.

[102] This work being the Mulasarvastivadin Vinaya.

[103] As translated by Enomoto pp.242-43.

[104] Enomoto p.243.