Collins - On The Very Idea Of The Pali Canon (1990)

On The Very Idea Of The Pali Canon

Steven Collins

In memory of I.B. Horner ¹

JPTS XV(4) pp. 89-126

In this paper I address the issue of the formation and role of the Pali Canon ² in Theravāda history and culture. My perspective is strictly that of an external observer wishing to make a contribution to historical scholarship, or at least to initiate an academic discussion of the issue: I mean to imply no evaluation whatsoever of any way in which the Canon has been or is seen by Theravāda Buddhists. From this perspective and for these purposes, I want to suggest that the role of the Canonical texts in Theravāda tradition has been misunderstood, and that the usual scholarly focus on the early period of Theravāda is misplaced. We must, I will suggest, reject the equation ‘the Pali Canon = Early Buddhism’, ³ and move away from an outmoded and quixotic concern with origins to what I would see as a properly focussed and realistic historical perspective. Rather than pre-existing the Theravāda school, as the textual basis from which it arose and which it sought to preserve, the Pali Canon - by which I mean the closed list of scriptures with a special and specific authority as the avowed historical record of the Buddha’s teaching should be seen as a product of that school, as part of a strategy of legitimation by the monks of the Mahāvihāra lineage in Ceylon in the early centuries of the first millenium A.D.

It seems to me useful to divide Theravāda Buddhist history into three periods, according to the different kinds of evidence which are available to us. ⁴ The first or ‘early’ period lasts from the time of the Buddha (whenever that was) to that of Asoka. We have no evidence of any kind which can be securely dated before Aśoka; to describe, speculatively, pre-Aśokan Buddhism, we must make inferences from his inscriptions, from the texts (whose extant form is due to the later period) and perhaps also from the material remains of later times. From the time of Aśoka onwards, in the second or ‘middle’ period, in addition to an increasing amount of textual materials we have inscriptions, coins, paintings, sculptures and other material remains to supplement and when necessary correct what the texts tell us. The third or ‘modern’ period refers to those recent centuries in which we have, in addition to material and textual primary sources, reports from western travellers, officials of imperial governments, anthropologists and others, as well as the modern records kept by indigenous rulers and bureaucracies. Much of the evidence for ‘early’, pre-Aśokan Buddhism is to be found in the Pali Canonical texts, or rather some of them; but in assessing the nature of this evidence we must be much more fully aware of their provenance in the ‘traditional’ Theravāda context than has hitherto been the case. ⁵ In the first part of the paper, I shall outline two senses of the word ‘canon’, and then look for comparable terms in Pali. In the second, I shall sketch in broad brushstrokes what I see as the context in which the Pali Canon emerged; and in conclusion I shall ask briefly what role has in fact been played by this Canon, and - more significantly - by the idea of such a Canon, in those religious cultures we denote by the short-hand term, ‘Theravāda’.

I

The word ‘canon’, in relation to textual materials, can usefully be taken in two ways ⁶ : first, in a general sense, as an equivalent to ‘scripture’ (oral or written). Used in this way, the term does not specify that the collection of texts so designated constitutes a closed list; it merely assigns a certain authority to them, without excluding the possibility that others could be, or may come to be included in the collection. In the second sense, however, the idea of a ‘canon’ contains precisely such an exclusivist specification that it is this closed list of texts, and no others, which are the ‘foundational documents’. The existence of some sort of scriptural or canonical materials in the nonspecific, inclusivist sense is surely a necessary condition for a religion to be or have what anthropologists used to call a ‘Great Tradition’. But the existence of a canon in the second, exclusivist sense is, on the contrary, a non-universal and contingent feature, dependent on the specific history of a given milieu which produces the selection and redaction of such a closed list. When compared with other extant collections of scriptures in Buddhism, I think the Pali Canon is unique in being an exclusive, closed list. Why did such a canon develop in traditional Theravāda Buddhism?

First, what Pali terms might correspond to ‘canon’? There are three main candidates: the word pāli itself, the notion of the tipiṭaka, ‘the three baskets’ of tradition, and most importantly, the concept of buddhavacana, ‘the Buddha’s Word(s)’.

(i) As is well-known, the word pāli was not originally the name of a language, but a term meaning firstly a line, bridge or causeway, and thence a ‘text’. [^7] It is often found in apposition to atṭhakathā, which is usually translated ‘commentary’, and so some scholars have taken pāli to mean ‘canon’. ⁷ I would not want to disagree with this, if the term is used in the general and inclusivist sense of ‘scripture’ outlined above. But the primary use of the distinction between pāli and atṭhakathā is not to classify documents into different categories (although it did come to have that function: e.g. Sp 549, Sv 581), and still less to denote explicitly a closed list of texts, as the terms ‘canon’ and ‘commentary’ might imply; rather, it was to distinguish between the precise wording of a text, in the text-critical sense, and the more flexible task of ‘saying what it means’, which is the literal translation of atṭhakathā.⁸ Pāli and attha are regularly applied to texts in this way (e.g. Mp IV 187, Th-a II 135-6 et freq.); these terms are often given in commentarial exegesis of the pair dhamma and attha (e.g. Pj II 333, 604, Ja II 351, VI 223; compare the ‘four-fold profundity’ at Sp 22 and Sv 20, the former using pāli, the latter tanti). Pāli can be used synonymously with pātha, ‘text’, in the sense of ‘reading’, often when discussing variants (e.g. Sv 49, Ud-a 105-6, Th-a II 203). ⁹ Quotations can be introduced by phrases such as tarrāyam pāli, ‘on this matter (there is) this text’, (e.g. Sp 13, 395, Spk I 200, Th-a III 105); the term pāli-vanṇanā, ‘text-commentary’, can be used in the same way as pada-vaṇnanā, ‘word-commentary’ (Sv 771, 982, Mp II 306), both of which are complementary to vinicchaya-kathā, ‘exegesis’ or attha-vaṇnanā, ‘explanation of the meaning’ (Vibh-a 291, Vism 16, Pj I 123 foll.). Pāli can refer to the text of a specific individual work, as Udāna-pāli (Ud-a 4) or Apadāna-pāli (Th-a II 201, III 204). The phrases pāliyam (an)āgata (or (an)ārūlha) are used to mean ‘(not) handed down in a/the text’, referring to textual passages, topics and names of people (e.g. Sp 466, 841, 1112, Sv 989, Mp I 272, IV 143, Th-a I 44, III 203); the term pālimuttaka, ‘not found in a (the) text(s) is used both of sermons by the Buddha not rehearsed at the Councils and thus not extant (Sv 539, Ud-a 419-20, cp. Sv 238, 636, Spk I 201) and of Disciplinary decisions and rulings in use by the monkhood but not found in the text of the Vinaya itself (Sp 294 et freq.). In none of these uses, however, does the term in itself imply that the texts so referred to are a closed list. ¹⁰

(ii) The term pitaka is usually taken to mean ‘basket’. ¹¹ If this is in fact the same word as pitaka meaning ‘basket’, ¹² then it is intriguing to speculate on what could be the metaphor underlying its use to mean ‘tradition’, given that one cannot literally put oral ‘texts’ in baskets: Trenckner (1908, pp. 119-121) held that just as in excavations or digging work in ancient India, baskets of earth were passed along a row of labourers, so the Buddhist tradition was passed along a line of transmission, in piṭakas, from teacher to pupil. Wintemitz (1933, pp. 8-9 note 3) suggested that the idea is of ‘receptacles in which gems, family treasures, were preserved from generation to generation’. In any case, we must agree, I think, with Rhys Davids (who accepted Trenckner’s view, (1894), p. 28) that the term tipiṭaka refers to ‘three bodies of oral tradition as handed down from teacher to pupil’. It is, perhaps, not necessary to see a metaphor underlying the term: just as the term āgama, in both Sanskrit and Pali, means colourlessly ‘something which has come down’, ‘a text’, and samhitā in Sanskrit means ‘a putting together, a sequence, a collection (of words, ideas, etc.)’ and hence ‘a text’, so pitaka can simply mean ‘a collection (of words, stories, etc.)’ and hence ‘a (part of a) tradition’. ¹³ The word is used in canonical texts to mean a ‘tradition’ or ‘customary form’ of religious teaching: but interestingly, in a pejorative sense, as a poor second-best to personal spiritual experience and knowledge. ¹⁴ The earliest extant uses of the word tipiṭaka date from inscriptions and texts of the 1st century A.D. ¹⁵ At this period, I think, it should be taken to denote not three closed lists of documents, but rather three different genres within the tradition; and to point to generic differences in style and content in the Disciplinary Rules (Vinaya-pitaka), the Discourses (Sutta-pitaka) and the ‘Further Teachings’ (Abhidhammapitaka). This tripartite division continues another, said in the canon to have existed during the Buddha’s lifetime: the division of labour between vinaya-, sutta-, and mätika-dhara-s, ‘those who bear (in memory) the disciplinary rules, the teachings and the mnemonic lists’. ¹⁶ Clearly during the Buddha’s lifetime, there can have been no closed canon ¹⁷ : and I agree with Lamotte (58, p. 164), when he says that ‘all that the classification of scripture into three baskets does is to attest to the existence within the religious community of three different specialisms, having for their objects the doctrine, the discipline and scholastic matters (la scolastique) respectively’. Eventually, of course, the term tipiṭaka did indeed come to have the sense of a closed and fixed Canon. ¹⁸

(iii) Originally, then, neither pāli nor tipiṭaka referred to a closed canon. This is true also of the third term buddha-vacana, ‘The Word of the Buddha’; but here we do begin to approach something like our ideas of a ‘canon’ and ‘canonical authority’. ¹⁹ The term, and other words and phrases referring to ‘what was said by the Buddha’ can be found in the Canonical texts. ²⁰ One of Aśoka’s inscriptions reads e keci bhamte bhagavatā budhena bhasite save se subhāsite vā, ‘everything which was said by the Blessed One, the Buddha, was well-said’. ²¹ The idea behind these terms can be, and has been taken in Buddhism in two crucially different ways. On the one hand it can be used, as it most commonly has been in the extant Mahāvihārin tradition of Theravāda, to mean the actual word(s) of the historical Buddha Gotama - despite the fact that it has always been evident that the the collection of texts so designated includes many which cannot have been actually spoken by him (those spoken by other monks before and after his death, for example). For this reason and others, on the other hand, there is also an historically unspecific sense of the term, which refers in general to the - eternal and eternally renewable - salvific content of Buddhist Teaching: to use a phrase ubiquitous in the Canon, it refers to the ‘spirit’ (attha) rather than the mere ‘letter’ (vyatjana) of the Buddha’s law (dhamma).

This non-historicist approach to scriptural authority, although not absent from Theravāda, is much more characterisic of Mahāyāna traditions, where the eternal truth of the Dharma may be revealed in texts of any and every historical provenance. The attitude is nicely captured in the phrase ‘whatever is well-spoken is spoken by the Buddha’. ²² A sutta from the Anguttara Nikāya (A IV 162-66), contains this phrase, and is worth looking at in more detail. ²³ It describes a conversation between the monk Uttara and the king of the gods, Sakka (Indra). Indra is impressed with a talk he has been told of, given by Uttara to some monks; he descends from heaven and asks Uttara whether what he said was own inspiration (sakaṃ paṭibhānaṃ) or the word of the Buddha (Bhagavato vacanaṃ). Uttara replies with a simile: ‘it is just as if there were a great heap of grain near some village or town, and people were to take grain from it in buckets or baskets (pitakehi), in their laps or hands. If one were to go up to these people and ask them “where are you bringing this grain from ?”, how would they properly explain themselves ?’ Indra replies that they would do so simply by saying that they got the grain from the heap. Uttara explains ‘in the same way, king of the gods, whatever is wellspoken is all the word of the Blessed One … Whenever I or others preach, what we say is derived from there’ (yaṃ kiñci subhāsitaṃ sabban taṃ tassa Bhagavato vacanaṃ … , tato upādāy’ upādāya mayañ c’ aññe ca bhanāma). (The choice of bhanati here is not accidental: bhāna and other derivatives are regularly used both for sermons and for the recitation of passages from the canonical texts.) Clearly the point of the remark here is simply that Uttara is saying that what he teaches comes from the Buddha; but grammatically there would be nothing wrong with interpreting his remark in the Mahāyānist sense. (In contrast, the inscription of Aśoka cited above is unambiguously not the Mahāyānist sentiment, since it serves as an introduction to his list of recommended texts (see below, and notes 22, 27): the logic of the edict is that ‘everything said by the Buddha was well-said, but these texts are especially good … ’.) Why then did what has become Theravāda ‘orthodoxy’ choose to emphasise an historicist and exclusivist idea of its ‘Canon’, ‘the Buddha’s Word(s)’ ?

II

For the sake of brevity, I will present my argument schematically. Before the 1st century B.C., all Buddhist texts are said to have been preserved orally ²⁴; there is a large amount of evidence from a wide variety of sources, mutually contradictory for the most part, which suggests that a series of meetings were held, usually called ‘Councils’ in English but more precisely ‘Communal Recitations’ (sañgiti), one of whose functions was for monks to recite together the scriptures, whatever they were. ²⁵ Apart from Aśoka’s inscription which mentions by name some texts still extant, ²⁶ however, we simply have no idea which texts in fact pre-date Aśoka, and which might have been thus recited. The traditional account has it that Pali texts were transmitted to Ceylon in the 3rd.century B.C., along with commentaries, and there again to have been preserved orally (the commentaries being translated into and elaborated in Sinhalese). Both texts and commentaries were then written down during the (second) reign of King Vattagāmanī, between 29 and 17 B.C. ²⁷ (see below). The following two statements, both written by staunchly orthodox modern Theravādins, make it clear that we cannot know the relation between ‘the canon’ as we now have it and the canon as it was being transmitted at this time; still less can we know that this canon was thought of in the closed, exclusivist sense. Malalasekara writes, in his standard work The Pali Literature of Ceylon (1928, p. 44), ‘how far the Tipiṭaka and its commentary reduced to writing at Ālu-vihāra resemble them as they have come down to us today no-one can say’. In fact, the earliest date to which we can assign the Canon in the specific and final form in which we now have it is the time of Buddhaghosa. As Walpola Rāhula observes in his History of Buddhism in Ceylon (1956, p. xix):

‘Although there is evidence to prove the growth of the Pali Scriptures during the early centuries of Buddhism in India and Ceylon, there is no reason to doubt that their growth was arrested and the text was finally fixed in the 5th century A.C. when the Sinhalese Commentaries on the Tipiṭaka were translated into Pali by Buddhaghosa’. ²⁸

The Pali Canon, like most other religious Canons, was produced in a context of dispute, here sectarian monastic rivalries. King Vattagämiṇī supported the rivals of the Mahāvihārin monks, those of the recently founded Abhayagiri monastery. (In the 4th century there arose a third sub-sect, the Jetavana group, but my focus here will be on the Mahāvihāra-Abhayagiri rivalry. ²⁹ ) Both groups existed throughout the first millenium, up until king Parakkamabāhu I suppressed the others in favour of the Mahāvihāra in the 12th century (the extant Mahāvihārin texts call this his ‘unification’ of the monkhood); and at certain periods Abhayagiri was clearly the more numerous and dominant. With some disputed exceptions, ³⁰ no Abhayagiri texts survive, although texts and commentaries are ascribed to them (directly or indirectly) in extant Mahāvihārin works. ³¹ We can trace, I think, a significant difference between Mahāvihārin texts written before Parakkamabāhu’s ‘reform’ and those written after: that is, in the direction of an increasingly triumphalist re-writing of earlier history.

One area where this change is particularly evident is in accounts of the writing down of the canon: the earliest versions are remarkably brief and restrained, giving little idea of the real reasons for this development, to us so significant. ³² The Dīpavamsa (XX 20-1) and Mahāvaṃsa (XXXIII 100-1) have exactly the same stanzas:

pitakattayapāliñ ca tassā atṭhakathaṃ pi ca mukhapāṭhena ānesuṃ pubbe bhikkhū mahāmatī; häniṃ disvāna sattānaṃ tadā bhikkhū samāgatā ciratṭhitathaṃ dhammassa potthakesu likhāpayuṃ.

‘Previously, intelligent monks (had) preserved the text of the three piṭakas and its commentary orally; but (now) when the monks saw the häni of beings they came together and had them ³³ written in books, in order that the Teaching should endure for a long time.’

The word hāni, which I have left untranslated, means ‘loss’, ‘decay’, ‘diminution’, ‘abandonment’, etc. The issue here is how to take it in context. The Dīpavaṃsa account places these stanzas in the midst of what is more or less a list of kings, with minimal narrative embellishment. It mentions Vattagämanī, but simply gives the bare details of his accessions to the throne (he was king twice), and the length of his reign. Oldenberg’s translation (1879, p. 211) has ‘decay’, Law’s (1959, p. 249) ‘loss’, neither of which attempts to interpret the term. The Mahāvamsa places the stanzas immediately after its account of the secession of the monk Mahātissa, and the subsequent split between the two monastic fraternities. Mindful of this perhaps, Geiger (1912, p. 237) translates hāni as ‘falling away (from religion)’. In modern secondary works, there has arisen a tendency to associate the writing of the texts most closely with conditions of war and famine, and so to translate hāni as ‘decrease (in numbers)’, or more generally ‘disastrous state’. ³⁴ This seems first to have been suggested by Adikaram (1946, Chap. 4); Rāhula’s account (1956, pp. 81-2, 157-8) is very frequently cited in other secondary works. These authors recount stories concerning war between Sinhalese and Tamil kings, and a famine associated with a brahmin turned bandit called Tissa. ³⁵ The Mahāvaṃsa mentions Tissa briefly earlier in the Chapter (XXXIII, 37-41), but not the famine.

Although it is quite plausible to connect the decision to commit the texts to writing with the troubled conditions of the time, it is worth noticing that this is not given as a reason in any of the primary sources, early or late. ³⁶ Adikaram himself suggests (pp. 115 foll.) that conditions in Rohana, in the south of the island, may not have been as bad as in the north; and as Gunawardana (1982) has shown, it is anachronistic to think of the island at this period as a single state centred at Anurādhapura. I suggest, not necessarily a replacement for their account but perhaps as a complement to it, that we follow the Mahāvamsa and associate the writing of the texts and commentaries with the contemporary rivalry between the Mahāvihāra and Abhayagiri monasteries; and I would argue that at least one of the motives for the decision was the fixation, through writing, of a definitive list of scriptures, at a time when the position of the Mahāvihāra as sole legitimate custodians of Buddhism was under threat. ³⁷ Certainly in the following centuries, one of the major themes in Mahāvihārin writing about its rivals concerns their use of ‘heterodox’ scriptures, in addition to the Pali texts shared by all three groups. It seems that at least from the 3rd century A.D., and perhaps before, the Abhayagiri monks used what we would now call Mahāyāna texts ³⁸; it is revealing that this is standardly referred to by their Mahāvihārin opponents as their embracing the vetulla-vāda. The term vetulla, Sanskrit vaitulya or vaipulya, meaning ‘extended’ or ‘enlarged’, refers to the great extent of certain Mahāyāna scriptures. ³⁹ Later triumphalist chronicles condemn with increasing vehemence the heresy of these unacceptable texts, and tell of repeated book-burnings by pro-Mahāvihārin kings. ⁴⁰

In the 5th century the great Indian monk Buddhaghosa spent some time in Ceylon at the Mahāvihāra, writing what are now the standard Pali commentarial works, on the basis of the earlier Sinhalese texts. ⁴¹ This also took place during the reign of a king who supported the Abhayagiri, Mahānāma (409-431). Thus Adikaram (1946, p. 94) aptly remarks:

‘It is worthy of notice that the two most important events, namely, the writing down of the Pali texts at Āloka-vihāra and the translation of the Commentaries into Pali, both took place during the reigns of kings who were not favourably disposed towards the Mahāvihāra and who actively helped the opposing camp, the Abhayagirivihāra’.

The account in the Cūlavamsa, written after Parakkamabāhu I and in part as a panegyric on him, tells us that when Buddhaghosa had produced his digest of Theravāda scholasticism, the Visuddhimagga, the Mahāvihārin elders exclaimed ‘assuredly, he is Metteyya (the future Buddha) (nissaṃsayaṃ sa Metteyyo); then when he had rendered their commentaries into Pali, they are said to have received them pālim viya, literally ‘just as (or ‘as if they were’) Canonical texts’, or more loosely ‘as the authoritative version’. ⁴² The parallelism is obvious: the Buddha Gotama produced the Texts (pāli) as buddha-vacana, ‘the Buddha Metteyya’ produces an authoritative redaction of the commentaries, pālim viya !44

Finally, I think we should see the writing and fixing of a closed canon in relation to the creation of historical chronicles in Ceylon: the vamsa tradition. ⁴³ The term vamsa (Sanskrit vamśa) was used in India for a variety of forms of historical text, primarily genealogies, from the time of the Brāhmanas. Another meaning of the term is ‘bamboo’, and I think we may see some significance in this. Bamboo grows by sending out one, and only one, shoot: unlike our concept of a genealogical tree, therefore, a vamśa genealogy allows only one legitimate successor at a time. Thus the term not only describes a line of transmission, but at the same time ascribes to the members of the vamśa a specific status and authority as legitimate heirs of that transmission. In the tradition of purāna writing, two of the traditional five characteristics (pañcalakṣana) alleged to be present in any such text are vamśa and vamśannucarita; the former term refers to a genealogy of gods, patriarchs, kings and great families, the latter to the deeds of such a vamśa. (How far these five characteristics actually do apply to the extant purānas is a complex issue.) The texts in question here are not only the great compendia of mythology, theology, etc., concerning various great gods such as Viṣnu and Śiva; they include also, amongst others, a little-studied genre of regional, caste purānas, about which Ludo Rocher says, in his recent book on the subject (1986, p. 72):

Even though this type of texts relate to single castes in limited areas of the subcontinent, they are again not fundamentally different from purāṇic literature generally … [then, quoting another writer:] The caste-purānas may be considered to be the extension of Vamśānucarita, in the sense that they devote themselves to the history of some Vamsa, in the broad sense’.

I suggest that we see the Pali chronicles in this perspective as a part of the literary genre of the purāna in the widest sense, listing the genealogy and deeds of the lineage of the Buddha and his heritage. In addition, both by their very existence and by such details of their content as the stories of visits by the Buddha to the different Theravāda lands, the vamsa texts produced in Ceylon and later in mainland Southeast Asia served the heilsgeschichtliche purpose of connecting these areas with India. More specifically, as Heinz Bechert has argued (1978), the early examples in Ceylon may have served the political purpose of enhancing and encouraging Sinhalese nationalism. It has long been recognised that the ideology of these vamsa texts is that of the dhammadipa, the island which the Buddha prophesied would be the historical vehicle of his saving truth. ⁴⁴

It has often been noted that the dominant Theravāda attitude to its scriptures, unlike other Buddhist groups, is an historicist one; but it has not been noticed, I think, that this development coincides with the production by Theravāda monks of what Bechert calls the only ‘historical literature in the strict sense of the word [in South Asia] prior to the period of the Muslim invasions’. ⁴⁵ The earlier Sinhalese commentarial materials, shared by both Mahāvihāra and Abhayagiri groups, contained vamsa sections, and there may have been at least one specifically Abhayagiri vamsa ⁴⁶; but a particular characteristic of the development of the Mahāvihāra tradition is its rich and varied collection of these texts, usually called ‘Chronicles’ in English. There were probably many different reasons for their being produced, and it is true that earlier Sanskrit and Pali works with vamsa sections were preserved orally. Nonetheless I suggest that a revealing perspective on the issue can be gained from the comparative historical and anthropological study of literacy, where it is widely recognised that one of the earliest functions of writing was the making of lists. ⁴⁷ I suggest that both the idea of a fixed and closed Canon and the vamsa genre may be seen together as members of the same class: the ‘list’. The vamsa genre is descended from name-lists (genealogies) and event-lists (annals); the closed ‘canon’ is also descended from name-lists and word-lists, but adds to the simple idea of a list of texts (a librarian’s concern, in itself) the crucial political element of closure: nothing can be added or taken away.

In brief, then, I argue that the following four developments in the Theravāda tradition, taking place over the first half of the first millenium A.D., are related, not only conceptually and historically, but also as connected parts of a strategy of self-definition and selflegitimation by the Mahāvihārin monks:

1. the writing down of the canon and commentaries;
2. the production of a closed and historically specific canon of scripture;
3. the standardisation of authoritative commentaries, and
4. the development of the historiographical tradition of vamsa texts.

(Incidentally, not only might we explain the creation of a fixed Canon by this historicism; it may be that this form of religious legitimation was one reason for the birth, or at least the first real flourishing of historiography in South Asian culture at this place and time.)

There have been, of course, other forms of legitimation in Theravāda, notably the possession and control of relics and images. ⁴⁸ But one of the most salient characteristics of the Mahāvihārin lineage has always been its conservative and/or reformist, text-oriented selfdefinition; this was significantly underlined and extended, both in Buddhism and in Buddhist scholarship, by the modern ‘scripturalism’ specific to the 19th and 20th centuries. ⁴⁹ It is well-known that Buddhism in South and Southeast Asia includes many more things than are described and prescribed in the Pali Canon; these are often seen as ‘later developments’, many of which are standardly but misleadingly referred to as ‘Mahāyāna elements’. Rather than see things in this way, I suggest, we should take this wider Buddhist culture as the contemporary context in which the move to an historicist ‘orthodoxy’ was made. We know that the Mahāvihārin lineage became ultimately dominant in Ceylon; and throughout its spread across mainland Southeast Asia as ‘Sinhala’ Buddhism, it seems to have been perceived precisely as a ‘reform’ movement, and to have been supported by kings with this rhetoric against already-existing forms of Buddhism. ⁵⁰ Within established Theravāda cultures, again, periodic reform movements have taken place, with the same rhetoric; and this is one important ingredient in Buddhist modernism: ‘back to the Canon !’ (Something like this seems to be happening in the Theravāda revival in contemporary Nepal. ⁵¹)

III

But what role did the actual Canon play in all this ? Did these and only these texts function as ‘scripture’, with no others having canonical authority in the first and more general sense I distinguished earlier? No.

We know that throughout Theravāda history, up to and including the modern world, many other texts, both written and in oral-ritual form, have been used. ⁵² The evidence suggests that both in so-called ‘popular’ practice and in the monastic world, even among virtuosos, only parts of the Canonical collection have ever been in wide currency, and that other texts have been known and used, sometimes very much more widely. ⁵³ Keyes writes (1983, p. 272): ‘The relevance of texts to religious dogma in the worldview of any people cannot be assumed simply because some set of texts have been recognized as belonging to a particular religious tradition. It is necessary, in every particular case, to identify those texts that can be shown to be the sources of dogmatic formulations that are being communicated to the people through some medium. There is no single integrated textual tradition based on a “canon” to the exclusion of all other texts … . The very size and complexity of a canon leads those who use it to give differential emphasis to its component texts. Moreover, even those for whom a defined set of scriptures exists will employ as sources of religious ideas many texts which do not belong to a canon. For example, the evidence from monastery libraries in Laos and Thailand … reveals that what constitutes the Theravādin dhamma for people in these areas includes only a small portion of the total Tipiṭaka, some semi-canonical commentaries such as Buddhaghosa’s Visuddhimagga, a large number of pseudo-jātaka and other pseudo-canonical works, histories of shrines and other sacred histories, liturgical works, and popular commentaries. Moreover, for any particular templemonastery in Thailand or Laos, the collection of texts available to the people in the associated community are not exactly the same as those found in another temple-monastery. In brief, the relevance of textual formulations to religious dogma in popular worldviews is problematic in each specific case’. ⁵⁴

It might well be that the content of most smaller monastery libraries is in effect a ‘ritual canon’; that is, it contains the texts, canonical or otherwise, which are in actual use in ritual life in the area concerned.⁵⁵ A monastic library with larger holdings may perhaps be compared to a modern academic library: for those few who happen to have access to it, it affords a seemingly obvious and straightforward resource, which provides and defines a cultural ‘world’; but one which gives a wildly misleading picture of the actual experience (literate, cultural, religious and otherwise) of those communities without such access.

If we wish to delineate the actual ‘canon’ or ‘canons’ of scripture (in the wider sense) in use at different times and places of the Theravāda world, we need empirical research into each individual case, not a simple deduction from the existence of the closed tipiṭaka produced by the Mahāvihāra. We need more research, for example, historical and ethnographic, on the actual possession and use of texts, in monastery libraries and elsewhere, and on the content of sermons and festival presentations to laity, to establish more clearly than we currently can just what role has been played by the works included in the canonical list. The hypothesis I have sketched out here suggests that the actual importance of what we know as the Pali Canon has not lain in the specific texts collected in that list, but rather in the idea of such a collection, the idea that one lineage has the definitive list of buddha-vacana. ⁵⁶ So the Pali Canon should be seen as just a ‘canon’ (in one sense of that word) in Pali, one amongst others.

Montréal Steven Collins

NOTES

BIBLIOGRAPHY

• Adikaram, E.W. (1946) Early History of Buddhism in Ceylon (Gunasena, Colombo)
• Barr, J. (1983) Holy Scripture: Canon, Authority, Criticism (Westminster, Philadelphia)
• Bechert, H. (1966) Buddhismus, Staat und Gesellschaft (Band XVII/1 der Schriften des Instituts für Asienkunde in Hamburg. Metzner, Frankfurt am Main)
• • (1969) ‘Zur Geschichte der buddhistischen Sekten in Indien und Ceylon’, La Nouvelle Clio, vol. 7-9, 1955-7
• • (1973) ‘Sangha, State, Society, “Nation”’, Daedalus Winter 1973
• • (1976) ‘Buddha-feld und Verdiensübertragung: Mahāyāna-Ideen im Theraväda-Buddhismus Ceylons’, Bulletins de la Classe des Letters et des Sciences Morales et Politiques, Académie Royale de Belgique, 5th series, vol. 62
• • (1977) ‘Mahāyāna Literature in Sri Lanka: the early phase’, in L. Lancaster (ed.) Prajñāpāramitā and related systems: studies in honour of Edward Conze (Berkeley Buddhist Studies Series 1, Berkeley)
• • (1978) ‘The Beginnings of Buddhist Historiography: Mahāvaṃsa and Political Thinking’, in B.L. Smith (ed.) Religion and Legitimation of Power in Ceylon (Anima, Chambersburg)
• • (1979) Buddhism and Society (Buddhist Publication Society, Wheel 265, Kandy)
• • (1984) ‘“To be Burmese is to be a Buddhist”: Buddhism in Burma’, in H. Bechert and R.F. Gombrich (eds.) The World of Buddhism (Thames and Hudson, London)
• • (1985) ‘Orthodoxy and Legitimation in the context of Early and Theraväda Buddhism’, in Buddhism and Its Relation to Other Religions. Essays in Honour of Dr. Shozen Kumoi on His Seventieth Birthday (Heirakuji Shoten, Tokyo)
• Bechert, H (ed.) (1978) Buddhism in Ceylon and Studies in Religious Syncretism in Buddhist Countries, (Abhandlung der Akademie der Wissenschaft in Göttingen, Philologische-Historische Klasse, Dritte Folge Nr. 108)
• Bechert, H. and Hartmann, J.-U. (1988) ‘Observations on the Reform of Buddhism in Nepal’, Journal of the Nepal Research Centre, vol. VIII
• Bizot, F. (1976) Le Figuier à cinq branches (Publications de l’école française de l’extrême-orient, vol. CVII, Paris)
• • (1979) ‘Le Grotte de la naissance’, in Bulletin de l’école française de l’extrême-orient, Tome LXVI
• • (1981) Le Don de soi-même (Publications de l’école française de l’extrême-orient, vol. CXXX, Paris)
• Bloch, J. (1950) Les Inscriptions d’A soka (Société d’édition ‘Les Belles Lettres’, Paris)
• Bode, M.H. (1909) The Pali Literature of Burma (Royal Asiatic Society, London)
• Bollée, W. (1969) ‘Some less well-known Burmese Pali texts’, in Pratidänam (Festchrift Kuiper) (Mouton, Den Haag)
• • (1970) Kunäla-Jätaka (SBB, vol. XXVI, Luzac, London)
• Bond, G.D. (1975) ‘Two Theraväda Traditions of the Meaning of “The Word of the Buddha”’, Mahä Bodhi Journal, vol. 83 no. 10-2.
• • (1980) ‘Theravāda Buddhism and the Aims of Buddhist Studies’, in A.K. Narain (ed.) Studies in History of Buddhism (B.R. Publishing, Delhi)
• • (1982) The Word of the Buddha (M.D. Gunasena, Colombo)
• • (1988) The Buddhist Revival in Sri Lanka (University of South Carolina Press)
• Brough, J. (1962) The Gändhärī Dharmapada (Oxford University Press)
• Clark, S. (1986) The Mysteries of Religion (Blackwell)
• Collins, S. (1990) ‘Introduction’, in M. Wijayaratna (translated by C.Grangier and S.Collins) Buddhist Monastic Life (Cambridge University Press)
• Cousins, L. (1972) ‘Dhammapāla and the Tīkā Literature’, in Religion, vol. 2
• Davidson, D. (1984) Inquiries into Truth and Interpretation (Oxford University Press)
• De Silva, L. (1970) Dighanikāyatṭhakathäṭikā, vol. 1 (Pali Text Society, London)
• De Silva, W.A. (1938) Catalogue of Palm Leaf Manuscripts in the Library of the Colombo Museum (Ceylon Government Press, Colombo)
• De Zoysa, L. (1875) Reports on the Inspection of Temple Libraries (G. Skeen, Colombo)
• • (1885) A Catalogue of Pali, Sinhalese and Sanskrit Manuscripts in the Temple Libraries of Ceylon (G. Skeen, Colombo)
• Denis, E. (1977) La Lokapañatti et les idées cosmologiques du Bouddhisme ancien (Atelier Reproduction des Thèses, Université de Lille III)
• Duroiselle, C. (1911) review of Bode (1909), Journal of the Burma Research Society, vol. 1.
• Eliade, M. (ed.) (1987) Encyclopaedia of Religion (Macmillan, New York)
• Fausbøll, V. (1896) ‘Catalogue of the Mandalay Mss. in the India Office Library’, Journal of the Pali Text Society, vol. IV
• Fernando, C.M. (1908) The Nikāya Sangrahawa (H.C. Cottle, Colombo)
• Filliozat, J. (1981) ‘La Valeur des connaissances Gréco-Romaines sur l’Inde’, Journal des savants, Avril-Juin 1981
• Finot, L. (1917) ‘Recherches sur la littérature laotienne’, Bulletin de l’école française d’extrême orient, Tome XVII
• Frauwallner, E. (1956) The Earliest Vinaya and the Beginnings of Buddhist Literature (ISMEO, Rome)
• Geertz, C. (1968) Islam Observed (University of Chicago Press)
• Geiger, W. (1908) The Dlpavamsa and Mahāvamsa and their historical development in Ceylon (Engl. transl. E.M. Coomaraswamy) (Cottle, Colombo)
• • (1912) The Mahāvamsa (Pali Text Society, London)
• Godakumbara, C.E. (1980) Catalogue of Sinhalese Manuscripts (The Royal Library, Copenhagen)
• Gombrich, R.F. (1988) Theraväda Buddhism: a social history from ancient Benares to modern Colombo (Routledge and Kegan Paul, London)
• Gomez, L.O. (1987a) ‘Buddhist Literature: Exegesis and Hermeneutics’, in Eliade (ed.) (1987)
• • (1987b) ‘Language: Buddhist Views of Language’, in Eliade (ed.) (1987)
• Goody, J. (1977) The Domestication of the Savage Mind (Cambridge University Press, London)
• Gunasena, H.M. (1901) Catalogue of Pali, Sinhalese and Sanskrit Manuscripts in the Colombo Museum Library (H.C. Cottle, Colombo)
• Gunawardana, R.L.H. (1979) Robe and Plough (University of Arizona Press, Tucson)
• • (1982) ‘Prelude to the State’, in Sri Lanka Journal of the Humanities, vol. 8 (published 1985)
• Harvey, G.E. (1925) History of Burma (Cass, London)
• Hinüber, O. von (1977) ‘Zur Geschichte des Sprachnamens Pāli’, in Beiträge zur Indienforschung: Ernst Waldschmidt zum 80. Geburtstag gewidmet, Veröffentlichungen des Museums für Indische Kunst Berlin, Band 4
• • (1978) ‘On the Tradition of Pāli Texts in India, Ceylon and Burma’, in H. Bechert (ed.) (1978)
• Horner I.B. (1963) Milinda’s Questions (Pali Text Society, London)
• • (1975) Minor Anthologies III (SBB. vol. XXXI, Pali Text Society, London)
• Htin Aung, Maung (1970) Burmese History before 1287: a Defence of the Chronicles (The Asoka Society, Oxford)
• Jayawickrama, N.J. (1968) Epochs of the Conqueror (Pali Text Society, London)
• Keyes, C.F. (1977) The Golden Peninsula (Macmillan, New York)
• • (1983) ‘Merit-Transference in the Kammic Theory of Popular Theravāda Buddhism’, in C.F. Keyes and E.V. Daniel (eds.) Karma: An Anthropological Inquiry (University of California Press, Berkeley)
• • (1987) Thailand (Westview, Boulder)
• Kloppenborg, R. (1977) ‘Theravāda Buddhism in Nepal’, Kailash vol. 5 no. 4
• Lafont, J (1962) ‘Les ecritures du pāli au Laos’, in Bulletin de l’école française de l’extrême-orient, vol. 50
• Lamotte, E. (1958) Histoire du Bouddhisme indien (Publications Universitaires, Louvain)
• Law, B.C. (1959) ‘The Chronicle of the Island of Ceylon’, in Ceylon Historical Journal, vol. VII, nos. 1-4
• Luce, G.H. (1969) Old Burma - Early Pagán, Artibus Asiae Supplementum 25
• McDermott, J. (1984) ‘Scripture as Word of the Buddha’, Numen vol. 31,1
• MacQueen, G. (1981) ‘Inspired Speech in Early Mahāyāna I’, Religion vol. 11
• Malalasekara, G. (1928) Pali Literature of Ceylon (Gunasena, Colombo)
• • (1935) Vamsathappakäsini (Pali Text Society, London
• • (1938) Dictionary of Pāli Proper Names (Pali Text Society, London)
• Malalgoda, K. (1970) ‘Millenialism in relation to Buddhism’, Comparative Studies in Society and History, vol. 12
• Mori, S. (1988) ‘Uttaravihāraṭthakathā and Sārasamāsa’, Journal of the Pali Text Society, vol. XII
• Mudiyanse, N. (1967) Mahāyāna Monuments in Ceylon (M.D. Gunasena, Colombo)
• N̄anamoli Bhikkhu (1962) The Guide (Pali Text Society, London)
• • (1975) The Path of Purification (3rd ed., Buddhist Publication Society, Kandy)
• Norman, K.R. (1978) ‘The Role of Pali in early Sinhalese Buddhism’, in Bechert (ed.) (1978)
• • (1983) Pāli Literature (Harrassowitz, Wiesbaden)
• • (1985) review of Stache-Rosen (1984), Journal of the Royal Asiatic Society
• Oldenberg, H. (1882) ‘List of Mss. in the India Office Library’, Journal of the Pali Text Society, vol. 1
• • (1879) The Dipavamsa: an ancient Buddhist historical record (Williams and Norgate, London)
• Olivelle, J.P. (unpublished ms.) ‘Scripture and Canon: Hindu Perspectives’
• Paranavitana, S. (1928) ‘Mahāyānism in Ceylon’, Ceylon Journal of Science, Section 6 - Archaeology, Ethnology, etc., vol. II Part 1
• Pathak, V.S. (1966) Ancient Historians of India (Asia Publishing House, New York)
• Philips, C.H (ed.) (1961) Historians of India, Pakistan and Ceylon (Oxford University Press, London)
• Prebish, C. (1987) ‘Councils: Buddhist Councils’, in Eliade (ed.) (1987)
• Rāhula, W. (1956) History of Buddhism in Ceylon (Gunasena, Colombo)
• Reynolds, F.E. and M. (1982) The Three Worlds According to Phra Ruang (Berkeley Buddhist Studies Series 4, University of California Press)
• Rhys Davids, T.W. (1894) The Questions of King Milinda, vol. 1 (SBE, vol. 35, Oxford University Press, London)
• Rhys Davids, T.W. and Oldenberg, H. (1885) Vinaya Texts, vol. 1 (SBE, vol. 13, Oxford University Press, London)
• Rocher, L. (1986) The Purānas (Harrassowitz, Wiesbaden)
• Saddhātissa, H. (1965) Upāsakajanālañkāra (Pali Text Society, London)
• Schopen, G. (1982) ‘The Text on the “Dhāraṇī Stones from Abhayagiriya”: A Minor Contribution to the Study of Mahāyāna Literature in Ceylon’, in Journal of the International Association of Buddhist Studies, vol. 5
• • (1984) ‘Filial Piety and the Monk in the Practice of Indian Buddhism’, in T’oung Pao vol. LXX
• • (1985) ‘Two Problems in the History of Indian Buddhism’, in Studien zur Indologie und Iranistik, Heft 10
• • (1989) ‘The Stūpa Cult and the extant Pali Vinaya’, in the Journal of the Pali Text Society, vol. XIII
• Sheppard, G. (1987) ‘Canon’, in Eliade (ed.) (1987)
• Smith, J.Z. (1982) Imagining Religion (Chicago University Press, Chicago)
• Somadasa, K.D. (1987) Catalogue of the Hugh Nevill Collection of Sinhalese Manuscripts in the British Library, vol. 1 (Pali Text Society and British Library)
• — - (1989) Catalogue of the Hugh Nevill Collection of Sinhalese Manuscripts in the British Library, vol. 2 (Pali Text Society and British Library)
• Stache-Rosen, V. (1984) Upāli-pariprcchāsūtra (Abhandlung der Akademie der Wissenschaften in Göttingen, PhilologischeHistorische Klasse, Dritte Folge, Nr. 140)
• Tambiah, S.J. (1968) ‘Literacy in a Buddhist Village in North-East Thailand’, in J.Goody (ed.) Literacy in Traditional Societies (Cambridge University Press, Cambridge)
• • (1970) Buddhism and the Spirit Cults of Northeast Thailand (Cambridge University Press)
• • (1976) World Conqueror World Renouncer (Cambridge University Press)
• Tedesco, P. (1952) ‘Sanskrit pitaka - basket’, in G.C.Miles (ed.) Archaeologica Orientalia in memoriam Ernst Herzfeld (Augustin, New York)
• Tiwari, R.C. (1983) ‘Socio-Cultural Aspects of Theravāda Buddhism in Nepal, Journal of the International Association of Buddhist Studies, vol. 6 no. 2
• Trenckner, V. (1908) ‘Notes to the Milinda-pañha’ (ed. Andersen), Journal of the Pali Text Society, vol. VI
• Upham, E. (1833) The Mahāvansi, the Rājāratnacari, and the Rājāvali, forming the Sacred and Historical Works of Ceylon, 3 vols. (Parbury, Allen and Co., London)
• Warder, A.K. (1972) An Introduction to Indian Historiography (Monographs of the Department of Sanskrit and Indian Studies, University of Toronto; Bombay, Popular Prakashan)
• Wickremasinghe, M. de Z. (1900) Catalogue of the Sinhalese Manuscripts in the British Museum (British Museum, London) Winternitz, M. (1933) History of Indian Literature, vol. II (Calcutta)

Footnotes

1. In 1981, when I had the honour to be invited to serve on the Council of the Pali Text Society, my first task was to prepare for publication Miss I.B. Horner’s last work, an unfinished translation of fifty stories originating from Chiang Mai in Thailand in the fifteenth century, and very closely modelled on the canonical Jātaka tales. She was working from the draft of the edition made by P.S. Jaini, which was subsequently published by the PTS as PaMāsa Jātaka (vol. 1, 1981; vol. 2, 1983). Professor Jaini also completed the translation. In choosing a title for the translation volumes, we followed a suggestion found in Miss Horner’s notes for the work, where she referred to it as ‘Apocryphal Birth Stories’; the volumes were published thus in 1985 (vol. 1) and 1986 (vol. 2). At that time Professor Jaini and I discussed, without coming to a clear conclusion, the issue of what is really meant in a Buddhist context by the opposition between ‘canonical’ and ‘apocryphal’ texts; at his instigation, I included in the brief preface to Volume 1 some notes on the background in Christian usage of the term ‘apocryphal’. This paper is a preliminary result of the research inspired by those initial discussions. It was first given, under the present title, as the Second I.B. Horner Memorial Lecture for the PTS in London, September 1987. I am glad to be able to publish it here in memory of Miss Horner, whose contribution both to Pali studies in general and to the PTS in particular has been so great. My title is adapted from the philosophical paper by Donald Davison, ‘On the Very Idea of a Conceptual Scheme’ (reprinted in Davison 1984). ↩

2. References to Pāli texts use the abbreviations of the Critical Pāli Dictionary. ↩

3. The general tenor of the re-evaluation I am recommending here is very much in line with the work being produced by Gregory Schopen, who has shown that for so many things either not found or not emphasised in the Canon, and usually seen as ‘later’ developments, there is in fact extensive evidence in the earliest archaeological and epigraphical remains: see, for example Schopen 1984, 1985 and 1989 . ↩

4. I have discussed this further in Collins (1990). The first two of my three periods are similar to those identified by Heinz Bechert (e.g. 1966, 1973, 1979, 1985) as ‘early’ and ‘traditional’; but his criterion for division and designation is the relation of the monastic community to society, and my third, ‘modern’ period does not correspond to his third, ‘modernist’ one. (I am grateful to Prof. Bechert for clarifying this issue, in correspondence.) ↩

5. I agree wholeheartedly with the suggestions made about the value of the commentaries in this regard by Bond (1980). Certain arguments from the content of the Canon do, I think, have force. For example, apart from a few Suttas which deal with the ‘mythical’ figure of the Universal Emperor, the cakkavatii, the texts do not betray any knowledge of large-scale political units such as that of Aśoka. (I use the word ‘mythical’ here in the same way as Gombrich (1988, p. 82); cf. also pp. 20-21 on this subject.) Anachronism of various sorts is not usually a problem in Buddhist literature; and so it would seem likely that these texts, in general, do indeed come from pre-Atokan times. But this kind of argumentation is very complex, and of course we cannot know that because something is not in the texts, it did not exist: the history of Hindu literature furnishes many counter-examples. (See further note 25 below.) ↩

6. In the argument of this paragraph I have profited from articles by Sheppard (1987) and, especially, Olivelle (unpubl. ms.). Sheppard writes that ‘on the one hand, [the term “canon”] can be used to refer to a rule, standard, ideal, norm, or authoritative office or literature, whether oral or written. On the other hand, it can signify a temporary or perpetual fixation, standardization, enumeration, listing, chronology, register, or catalog of exemplary or normative persons, places, or things [and, in our case, texts]. The former dimension emphasizes internal signs of an elevated status. The latter puts stress on the precise boundary, limits, or measure of what … belongs within or falls outside of a specific “canon”’.
   In proposing a closely related distinction, Olivelle argues that ‘a canon, like an orthodoxy, may be exclusive or inclusive. An exclusive canon both lists the documents included in the scripture and implicitly or explicitly excludes all other documents; the canon is a closed list. An inclusive canon also has a list of documents contained in the scriptures. But it makes no claim to be exhaustive. The list merely has a positive function and it does not intend to exclude documents outside the list. In cases such as the [Indian] Veda, the tradition explicitly admits the possibility that there may exist other documents belonging to the Veda. Other traditions, such as most oral ones, may simply ignore the issue. In all cases of inclusive canons, however, the traditions do not feel the need to precisely demarcate the canonical boundaries’. McDermott (1984, p. 32) remarks aptly that ‘the Mahāyāna Sūtras in India fit into a more Sanskritized concept of scripture and canon (or lack thereof) than does the Theravāda Tipäaka’. [^7] The metaphor here, as in other words for texts meaning ‘line’, ‘thread’, etc. (e.g. gantha, tanti, and sutta, if this is indeed equivalent to Sanskrit sūtra), seems rarely if ever to remain alive in the use of the term. One use of the term in parts of the Manoratha-pūranī may preserve a sense of ‘line’ or ‘list’. The Añguttara text names a series of monks, nuns, laymen and laywomen, each of whom is said to be ‘pre-eminent’ in some sphere. At the end of each commentarial section, the text states therapālivannanā niṭhitā (Mp I 337), (and similarly) theripāli- (381), theripāli- (381), upāsakapāli- (401), upāsikāpāli- (458). (There are variant readings therapāliyā, theripāliyā, and upāsikāpāliyā (sic) vannanā in the first three places.) This may be translated, taking the first example, ‘the commentary on the list of elders is completed’, instead of simply ‘the commentary on the text of (or about) elders … .’ At the beginning of the commentaries on the last three ‘lists’, the text states theripāliyam pathame (337), upāsakapāliyam (482 pathame must have been accidentally omitted here; there is a v.l. upāsakapālivannanāya pathame), and upāsikāpāliyam pathame (401). Pathame cannot agree with -pāliyam (or -vannanāya); there must be some appropriate masculine noun implied (such as sutta: see A I 23 note 3), so that we may translate ‘in the first sutta in the list of (or text about) nuns (laymen, laywomen)’. The v.l. at 337, theripāliyā, which could be genitive, makes this rendering easier, ‘in the first sutta of the list (text) of nuns’. (Cp. e.g. Mp II 34 catutthavaggassa pathame.) At Mp I 29 there is rāpapāli, at II 1 aṭhānapāliyam (v.l. -pāliyā); at II 18 aṭhānapāli vannanā niṭhitā and, beginning the next section, ekadhammapāliyam.
   Filliozat proposed that in the compounds pāli-bhāsā and its equivalent tanti-bhāsā (Sanskrit tanra) both first terms should be understood as referring literally to ‘lines’, i.e. lines of the text in manuscripts (1981, p. 108). This would be extremely important if it could be shown to be true; it would, for example, render problematic the whole tradition which says that both pāli and aṭṭhakathā were transmitted orally before the 1 st century B.C. But I know of no evidence to support the hypothesis: Filliozat’s brief discussion, ibid. note 21 , is simply an argument from analogy. At one place in the Jātaka, VI 353, the term pāli is used of what is clearly an oral (and non-religious) ‘text’ (cf. von Hinüber (1977, p. 244)). ↩

7. E.g. Norman (1983, p. 1), von Hinüber (1977, p. 243). ↩

8. In this connexion, Frauwallner’s speculations on the oral nature of the early tradition are suggestive (1956, pp. 172-177, 189). Although he does not mention this, it seems to me highly probable that the structure he describes, of fixed (though not yet written) ‘memorial sentences’ fleshed out with freely composed ‘oral explanations … given not in Pāli but in the local language’ was what lay behind the distinction between pāli and aṭṭhakathā. (We have evidence for this structure in the modern period also: see Finot (1917, p. 41); Somadasa (1987, p. ix); Tambiah (1970, p. 166). This might also have helped to bring about the confusion between pāli as a word for ‘text’ and as the name of a language. (As I hope to show elsewhere, however, I remain quite unconvinced by the overall hypotheses of Frauwallner’s work, not least because in the main body of the text he seems quite to forget the oral nature of the early tradition, in arguing for a single text grandly and precisely conceived and organised by ‘the author of the Skandhaka’.) ↩

9. von Hintuber, (1978, p. 52), gives an example where alternative readings of a word are cited in different manuscripts of a text, one of which calls the alternative reading a pātha, the other a pāli. In two versions of the same commentarial exegesis discussing variant readings, one (Th-a III 201) reads pāli, the other pātho (Pj II 350). ↩

10. Of course, by the time of Buddhaghosa the list of texts had come to be fixed, though not without disagreements (see Norman (1983, p. 9)), and thence de facto the term pāli was restricted to that list, at least in Ceylon, just as the term aṭhakañā came only to be used of commentaries on pāli texts, others being tikā. A number of texts are sometimes said to have been added to the Canon in Burma: The Sutta-sangaha, Netti-pakarana, Petakopadesa, Milindapatha (see Oldenberg (1882, p. 61); Bode (1909, p. 5); Duroiselle (1911, p. 121), who disagreed with Bode; Nānamoli (1962, p. xii); and Bollée (1969, p. 494), who says that King Mindon’s stone edition of the tipiṭaka contains the last three of these texts, as does the modern Chathasangāyana edition). The word pāli is used of the Suttasanigaha in Burmese manuscripts (Oldenberg (op. cit., p. 80); Fausbøll (1896, p. 31)). The Netti-pakarana, which itself claims to have been composed by Mahākaccāna, praised by the Buddha and recited at the first Council (Nett 193), is called by its commentary a pāli (Nett, Intro. p. XI; see also Nānamoli, op. cit., p. xi); and the commentary is classed as an aṭhakañā by the Gandhavamsa (p. 60). For the use of pāli in relation to the complex issue of the ‘canonical’ verses of the Jātaka, in opposition to the non-canonical and commentarial prose passages, see, for example, the references given by Fausbøll in Ja VII p. III, and the comments of Bollée (1970) Preface. In the commentary to the Nidāna-kathā, a prose section is referred to as a pāli, and an account of its atha is given (Ja I 7). ↩

11. One philosopher of religion has recently referred to the (‘Eastem’) ‘Religions of the Baskets’, in opposition to the (‘Western’) ‘Religions of the Book’: see Clark (1986), p. 16, etc. ↩

12. Tedesco, (1952, p. 209), suggests that it might not be. ↩

13. At Sp 20-21 Buddhaghosa explains the term as meaning either ‘learning’ (pariyatti) or ‘a container’ (bhājana), and says that the two senses are to be taken together in understanding, e.g. the term Vinaya-pitaka. For remarks on the use of pitaka in the title of the (canonical but probably post-Aśokan) Cariyāpitaka, see Horner (1975) Cp Preface pp. iii foll. ↩

14. Pitaka-sampadā and -sampadāna, both meaning ‘expertise in a tradition’ are used in this way of the tradition of learning Vedic mantras (M II 169) and in a general sense, as in the famous Kālāma Sutta (A I 189 foll.) and elsewhere (e.g. M I 520; A II 191 foll.). ↩

15. For inscriptions, see Lamotte (1958, pp. 163-64, 347-50), where the chronology is not clearly described (see Schopen (1985) pp. 10-11); the word tipetaki occurs in the Pariyāra (Vin V 3), an ‘appendix’ to the Vinaya included in the canon but usually taken to have been produced in Ceylon in the 1st century A.D. The same date is often given for the occurrence of tepitakam buddhavacanam and tepitako in the Milinda-patha (pp. 18, 90), although the dating of this text is far from easy: see Horner (1963, pp. xxi foll), Norman (1983, pp. 110-11). ↩

16. See Norman (1983, pp. 96-97). Individuals could, of course, become expert in all three branches. ↩

17. This is perhaps an appropriate place to deal with a well-known, but very problematic text, the passage of the Mahāparinibbāna Sutta (D II 123 foll., found also as a separate sutta at A I 167 foll.), dealing with the ‘Four Appeals to Authority’ (cattāro mahāpadesā) Here the Buddha is made to say that if a monk claims to have ‘heard’ (sutam) and ‘received’ (patiggahitam) from himself, the Sangha, a group of monks or a single monk, that ‘this is dhamma, this is vinaya, this is the Teacher’s Doctrine’ (sattha sāsanam), then what he says (tāni padavyatjianāni) is to be compared with the Sutta and Vinaya. It is true that, coming at the end of his life, we might be expected to assume that most of these two bodies of Teaching had by then been given; but it strains credulity to imagine that what is in question here is a straightforward checking of one ‘text’ against a known and fixed body of such texts, collected as the Sutta- and Vinaya-piṭakas. There would be a logical problem here of self-reference: according to its own criterion, this text itself could not be accepted, since at the time of its utterance it could not yet have been included in such fixed piṭakas, as could not all the other texts, including the Mahāparinibbāna itself, said to have been composed after the Buddha’s death. Perhaps more seriously, it is quite unclear, to me at least, exactly what is the force of the terms I have paraphrased as ‘to be compared’: otāretabhāni and sandassetabhāni. Perhaps the most obvious way to take them is in the sense of a general conceptual and practical agreement (in ‘spirit’ as opposed to ‘letter’). This is the way the Netti-pakarana (pp. 21-22) interprets the Sutta. As the Buddha says elsewhere, ‘those things (‘doctrines’, ‘states of mind’, dhamme) which you know lead to … nibbāna you may preserve (dhāreyyāsi) as the dhamma, the vinaya, the Teacher’s Doctrine’ (sattha sāsana) (A IV 143). (See MacQueen (1981, pp. 314-15) on these texts.) But this leads one immediately to a non-specific, non-historicist interpretation of what dhamma and vinaya are, which would argue very much against either the existence or the desirability of a fixed collection of texts. (See further text below, and notes 22 24, discussing Ałoka’s edict and A IV 162-66.) ↩

18. For example, in Buddhaghosa’s introduction to the Samantapasādikā; but note that he also says here that the Vinaya-pitaka contains material not recited at the First Council (pahamasaangitiyam sangitah ca asangitah ca (Sp 18; cp. Sv 17); see also note 11 above). I suspect that the adjective tipitakin, when used in commentarial narratives not directly on the subject of the scriptures, often does not refer to those (presumably fairly rare) monks who had actually themselves memorised the entire corpus, but rather to that part of the Order whose allegiance was explicitly to the Mahāvihārin orthodoxy of the Tipiṭaka, as opposed both to those who used other texts, and to those ascetics and holy men in the yellow robe whose religious practice, and hence popular appeal, tended not to rely on books and the institutions which housed them, but on broader, less predictable and hence less controllable spiritual achievements. Arguing for this, however, must await another occasion. ↩

19. In writing of this term and its meaning, I have learned most from George Bond’s rich and sympathetic treatments (e.g. 1975, 1982), and from MacQueen (1981) and McDermott (1984). ↩

20. Examples: buddhavacana at Vin IV 54, Th 403 (these seem to be the earliest uses; cf. also Mil 17); bhagavato vacana at A IV 163, 164; buddhabhāsita at Vin IV 15; buddhassa sāsana at Thī 202 et freq., Th 639; buddhasāsana at Dh 368, 381; satthu-sāsana at Vin I 12, D I 110, etc.; tathāgata-bhāsita at S II 267, A I 72 . ↩

21. The Bhabrā inscription, cited from Bloch (1950, p. 154). ↩

22. The quotation is from the Adhyādayasamcodana Sūtra, cited in Śāstideva’s Ś ikṣāsamuccaya (I 15): yatkimcinmaiteya subhāṣitam sarvam tadbuddhabhāṣitam. Gomez ( 87 a, see also 87 b) provides a lucid overview of the different Buddhist attitudes to ‘the Buddha’s word’, making reference a number of times to the issue of historicist and non-historicist hermeneutical strategies. ↩

23. This is discussed by both MacQueen (1981, p. 314) and McDermott (1984, pp. 28-30). ↩

24. The argument first put forward by Rhys Davids and Oldenberg (1885, pp. xxxii-xxxvi) must, I think, still stand: the Vinaya texts give minutely detailed accounts of the daily life of the monkhood, but although writing is certainly known in them, we never read, even obliquely, of monks writing scriptures or reading manuscripts. It is true that, as Gregory Schopen showed in the last volume of this journal (Vol. XIII, 1989), we cannot be sure that because something is not in the Pāli Vinaya, it did not exist. All other extant Vinayas apart from the Pāli contain rules concerning stūpas; but his close reading of passages from the Vinaya itself, as well as from later Pāli and Sinhalese texts, suggests the strong possibility that in fact it did originally contain such rules. In the case of writing, however, none of the extant Vinayas describes monks as writing the scriptures, and so despite the fact that the argument is one from silence, and although it was originally based on the Pāli Vinaya alone, it has been supported by the discovery of other traditions. Brough (1962, pp. 28-29, 218 foll.) argues for the likelihood of a manuscript tradition of the verses now known as the Dharmapada (Dhammapada) earlier than the redaction of the Pāli version; although individually the examples of textual relationships he cites to prove ‘a very early written transmission’ seem to me less than compelling, common sense would suggest that the transition from oral to written would be gradual and piecemeal, rather than sudden and dramatic as the Chronicles’ accounts tell us. ↩

25. The most recent brief account is Prebish (1987), with bibliography. ↩

26. The Bhabrā inscription cited above mentions seven texts, of which some have been identified with sections of the last two vaggas of the Sutta-Nipāta. See Lamotte (1958, pp. 256-59). ↩

27. Norman (1983, pp. 7-11) is a succinct survey; for a lengthier consideration of the evidence see Norman (1978). ↩

28. This fact renders futile, in my opinion, the work of those scholars who imagine that anything found in the Canon must be grist for the mill of ‘early Buddhism’, while anything in the commentaries is ‘later’ and therefore to be ignored in our search for the ‘original Buddhism’. The fact is that the same tradition, at the same time and in the same place, has simultaneously preserved for us both the canon as we have it and the commentaries. No doubt, as said earlier (note 5), some judgements of relative chronology can be made on the basis of the internal evidence of these texts; but such judgements are always risky and piecemeal. ↩

29. See Gunawardana (1979, pp. 7-37). ↩

30. Three extant texts have been claimed to be Abhayagiri productions: the Upālipariprcchā-sūtra, which is said to have replaced the Parivāra of the Mahāvihārin Vinaya (see Stache-Rosen (1984), pp. 28 foll, with Bechert’s Introduction pp. 11 foll., and Norman’s review (1985)); and two later texts, the Vimuttimagga (see Norman (1983, pp. 113-14)) and the Saddhammopāyana (see Saddhātissa (1965, pp. 32-33, 59-64); Bechert (1976, p. 29 note 2); Norman (1983, pp. 159-60)). ↩

31. With the exception of a reference to an Uttaravihāra-mahāvamsa at Mhv-1 134 (and assuming the Uttara-vihāra and the Abhayagiri-vihāra are identical), no texts are attributed directly to the Abhayagiri group in the commentaries. Other works, including a vetulla-pitaka (variously spelt: see text and note 40 below) are named in commentaries and said to be abuddha-vacana: at Sv 566 and Mp III 160 the Gulha-vessantara, Gulha-ummagga, Gulha-vinaya, and vedalla-pitaka are to be rejected since ‘they do not conform with the Suttas’ (na sute otaranti, a phrase in the Mahäpadesa Sutta, here being commented on in both places). Sp 742 and Spk II 201-202 (for the tikā on this passage see Cousins (1972, p. 160)) add to these names the Vanna-pitaka, A ngulimäla-pitaka, Ratthapäla-gajjita, and A lavaka-gajjita. The Nikāya-sangraha (Fernando (1908, pp. 9-10)) lists these texts and others, assigns their composition to various schools in India, and says that only some came to Ceylon; these included the vaitulya pitaka which it later says was adopted by the Abhayagirivihāra-vāsins. Adikaram, (1946, pp. 98100), discusses these texts, and attempts to find versions in Chinese. It may be, as Rāhula suggests (1956, p. 90), that in the later period the term vaitulya came to be used in a general way to refer to any ‘dissenting views and new interpretations not acceptable to the Mahāvihāra’. The commentary on the Mahāvamsa mentions an Uttaravihāra-atṭhakahā several times: see Geiger (1908, pp. 47 foll.); Malalasekera (1935, vol. 1 pp. lxv-lxvii). The commentaries often discuss alternative views and interpretations, which may have been those of the Abhayagiri commentaries: see De Silva (1970, vol 1 p. lxvii foll.); Mori (1988). ↩

32. The change can be clearly seen by comparing the accounts in the Dipavamsa and Mahāvamsa, written in the 4 th and 5 th centuries, with those of the Nikāyasangraha (in Fernando (1908, p. 10-11)) and Saddhamma-samgaha (Chapter 6, JPTS (1890) pp. 46-50), written in the 14 th. ↩

33. It seems natural to take both pitakattayapālim and atthakaham as governed by likhäpayum as well as ânesum; and so we have both ‘Canon’ and Commentary written down for the first time together. ↩

34. Gombrich (1988, p. 152). The commentary to the Mahāvamsa (Mhv-t 623) rather surprisingly glosses hānim as ‘the decline in mindfulness and wisdom of beings whose length of life is diminished in the Kali-age’ (or perhaps simply ‘(that) unlucky time’) kalikāle parihināyukasatānam sati-buddhiparihānim). ↩

35. The main texts used are Mv XXXIII 37 foll., Mp I 92-93, Vibh-a 445 foll.; the account at Mp I 92-93 uses the name Caṇ̣āalatissa but seems to be the same story. (See Malalasekara (1938) s.vv. Caṇ̣āalatissa-mahābhaya and Brahmanatissa-cora.) ↩

36. Both Adikaram and Rāhula give as an example of the threat posed ‘during this period’ by the famine the statement that only one monk was alive who knew the Mahāniddesa. The version of this story in the PTS edition of the Samantapasādikā (695-96) indicates the time of the tale simply by saying mahāb:haye. I do not see why this has to be read as ‘in the Great Famine’, referring specifically to this period; it could just mean ‘in a famine’ or more simply ‘in (a time of) great danger’. ↩

37. The earlier accounts do not mention the place of the writing down of the texts; from the 13th and 14th centuries onward, in the Päjāvaliya and Nikāyasamgraha (see Norman (1983, p. 11)) and the Sāra- or Sāratha-sangaha (see Jayawickrama (1968, pp. 82-83) and Norman (1983, 173)) arises the tradition, so often found in modern secondary works, that this took place far from the capital at Alu- or Āloka-vihāra near modern Matale in central Ceylon. If this was so, Adikaram (1946, p. 79) may be right to suggest that the location, and the fact that it took place under the patronage of a local chieftain rather than the king, afford further evidence that the development is to be seen in the light of Vattagāmini’s patronage of the Abhayagiri monks. This idea is supported by the fact that the Saddhamma-samgaha, which re-writes the tale by giving the king a leading role in the story, has the ‘Council’, as it is there called, take place in a hall which he had built specially for the occasion in the Mahāvihāra itself at Anurādhapura (Saddhamma-s Chapter 6 p. 48). ↩

38. The Nikāyasangraha (Fernando (1908, pp. 12-13)) tells us that in the reign of king Vohārikatissa (269-291) the Abhayagiri monks ‘adopted the Vaitulyan Pitaka’ (on this term see text below), and that the king subsequently ‘suppressed [this] heresy’. Bechert (1976, pp. 43 foll. and 1977, p. 364) has argued that Mahāyāna literature was written before this time, the only extant example being the Buddhäpadāna, written in the 1st or 2nd century and now included in the Pāli canonical text called the Āpadāna; he does not suggest that this was specifically an Abhayagiri text, however. As was mentioned above (note 32), the Nikayasangraha describes vaitulya texts as coming to Ceylon long before the 3rd century. ↩

39. In his Abhidharmasamuccaya Asanga says that the terms vaipulya, vaidalya and vaitulya refer to the same thing, which he also calls the Bodhisattva-pitaka (p. 79, cited in Rāhula (1956, p. 89)). (On this term see also Winternitz (1933, pp. 283, 316)). It is unlikely, and unnecessary, that these terms, a number of variants of which occur in the Pāli sources, should have had any more precise denotation than does the general term ‘Mahāyāna’, which refers not to one or more specific Nikāyas in the Buddhist legal sense, but to a general tendency in Buddhist religion. The classic discussion of ‘Mahāyānism in Ceylon’ is Paranavitana’s article with that title (1928); for recent discussion see Rāhula (1956, pp. 89-90), Norman (1978, pp. 40-41), Bechert (1976) and (1977). ↩

40. This is perhaps most evident in the Nikāya-sangraha. ↩

41. The best survey of the evidence for Buddhaghosa and his activity is Nānamoli (1975, pp. xv-xxvii). ↩

42. Chapter 37 verses 215-46. Buddhaghosa’s own Visuddhimagga (p. 96) provides a remarkable story expressing the attitudes he encountered at the Mahāvihāra: a monk called Tipiṭaka-Cūlāhhaya, who had not learnt the commentaries (aṭhakaṭham anuggahervā) announced that he would give a public discourse on the scriptures (pañcanikāyamandale tini pitakāni parivatessāmi; later he says pariyatim parivatessāmi - it is not clear to me whether this refers simply to a recitation of texts or to commentarial discourses on them, or both). The monks tell him that unless he does so according to the understanding of their own teachers (attano ācariyuggaham) they will not let him speak. He then goes to his Preceptor, who asks for an example: ‘how do the teachers say (or ‘explain’) this passage ?’ (idam padam katham vadanti). Although the monk then gives the passage correctly, his Preceptor simply grunts (hun ti); he then gives it twice more, each time differently (aṭ̂̂ena añtiena pariẏāyena), but his Preceptor merely grunts again, and then explains: ‘your first version follows the way of the Teachers, but because you have not learnt it from them in person, you could not establish that it is their version’ (tayā pathamam kaṭhito yeva ācariyamaggo, ācariyamukhato pana anuggahitattā evam ācariyā vadantī ti saṃthātum nāsakkhi). ↩

43. Surveys of early historiography in India and Ceylon are found in chapters by Majumdar, Perera, Warder and Godakumbara in Philips (ed.) (1961), Pathak (1966) Chapter 1, Bechert (1969) and Warder (1972, Chapters 3-5). ↩

44. See Perera (op. cit. in previous note). Malalgoda (1970, pp. 431-32) has usefully compared this attitude to that of ancient Israel; while there are of course many disanalogies, I might add that this attitude has often been connected with the growth of an historical consciousness in Israel. ↩

45. Bechert (1978, p. 1). ↩

46. See Geiger (1908, Chapter 2), Norman (1983, pp. 114-18); and note 32 above. ↩

47. I am drawing specifically on Goody (1977) Chapter 4, ‘What’s in a list ?’, and especially Smith (1982) Chapter 3, ‘Sacred Persistence: Toward a Redescription of Canon’. For interesting and relevant remarks on the Judaic and Christian ‘canons’ see Barr (1983), esp. Chapter 3, ‘The concept of canon and its modern adventures’. ↩

48. It is not surprising that there are also a number of vamsa texts devoted wholly or in part to recounting the history of relics and their possession: e.g. the Dāthavamṣa, Thūpavamṣa, Cha-kesa-dhātu-vamsa, Jina-kāla-mālī. ↩

49. The term ‘scripturalism’ was first used in this way by Clifford Geertz (1968), and has been applied to Theravāda by Tambiah (1976) and Bond (1988). I think that this application is very fruitful, but less so when it is generalised to refer to the pre-modern period, as both Tambiah and Bond do. In Theravāda countries, as in the Islam of Indonesia and Morocco described by Geertz, it is most helpful to use the term to refer to a religious attitude arising as a reaction to a wide range of phenomena in the experience of colonialism and modernity: the downgrading of localised supernaturalism, the cultural prestige and practical power of western science, the centralization and bureaucratisation of power, the establishment of a ‘secular’ educational system, printing presses, and the resulting value placed on literacy. The search for indigenous resources to combat foreign dominance led, amongst other things, to an emphasis on the noble ideals of the early texts: their teachings are abstract and universal as opposed to localised, ‘rational’ and ‘ethical’ as opposed to magical, and fit better with the placing of cultural and political authority in the institutions of bureaucracy and education than do the personalised spiritual interactions of localism. This concatenation of phenomena is, of course, specific to the modern world; and the comparative insight which can be gained from using Geertz’s term to describe the Buddhist case seems to me to be lost when it is generalised to become an overall category applicable to all historical periods. ↩

50. Hence the recurring notion of the need for ‘purification’ of the Samgha by kings. For the influence of Ceylonese Theravāda, in its post-Parakkamabāhu ‘unified’ form, on mainland Southeast Asia see Keyes (1977, pp. 80-81; 1987, pp. 32-33). One example of the relevance, at least at the level of legend and ideological legitimation, of the possession of the Canon can be found in the story of the introduction of Theravāda to his kingdom by the Burmese King Anuruddha (1044-77). (This is, of course, before Parakkamabāhu I.) As Luce says (1969, pp. 18-19), although the Chronicles ‘at first seem hopelessly confused’, ‘all are agreed that he was a champion of Buddhism, whose main purpose was to secure copies of the Tipiṭaka and Relics of the Buddha’. In the various versions of the story recounted by the Sāsana-vamsa (pp. 56-65), for example, the legitimatory knowledge and possession of the Buddha’s ‘true’ teaching, as embodied in the canonical texts, is a central theme, and is opposed to the practices of ‘false ascetics’. (This is probably a reference to the practices and influence of the Ari.) Thus the texts, and certain relics, become emblems of orthodoxy, as Bechert’s recent summary of the story has it (1984, p. 148): ‘The Burmese chronicles report that Anuruddha was converted by a Mon monk called Shin Arahan, but that there were no copies of the holy scriptures and no relics in Pagan. The Mon king refused the Burmese king’s request for a copy of the holy scriptures and some relics. It is unlikely that this was the real reason for war as the texts claim; Anuruddha at any rate conquered Thaton in 1057, took the Mon king captive, and brought him, his family and many monks and skilled workmen to his capital Pagan, together with manuscripts of the sacred scriptures of Theravāda Buddhism. With them Mon culture and Theravāda Buddhism reached the Burmese. The supremacy of the Tantric monks was now broken, and though their doctrine survived for a time, particularly in the border territories of Burma, their influence diminished steadily while orthodox thought soon prevailed in all parts of the country’. The Sāsana-vamsa informs us (p. 63) that the king had the relics installed in a jewelled basket and the texts kept in a jewelled palace. There has, naturally, been much discussion of the historical validity of the Chronicles’ accounts: See Harvey (1925, pp. 23-34), Luce (1969, Chapter 2), Htin Aung (1970, Chapter 6). It is certain, however, that the Theravāda tradition gradually replaced what we now call ‘Mahāyānist’ forms of Buddhism: see, for example, Luce (1969, Chapter 10). ↩

51. See Bechert and Hartmann (1988), Kloppenborg (1977), Tiwari (1983). ↩

52. Much of this literature is called ‘Mahāyānist’, although again I doubt the usefulness of the term. To the references given in note 40 for the early phase, add also Mudiyanse (1967, Chapter 2) and Schopen (1982). J.S. Strong’s forthcoming work on Upagupta will detail the extensive presence in Southeast Asian ritual and indigenous literature (and at least one text in Pāli: see Denis (1977)) of this figure derived from the Sanskrit Sarvāstivāda tradition. F. Bizot’s striking reports from the ‘unreformed’ Mahānikay monasteries of Cambodia show texts and practices which can without much hesitation be called tantric: see Bizot (1976,1979,1981). ↩

53. Evidence for this in early 19th century Ceylon can be found in Upham (1833, vol. 3 pp. 167-215, 267), for early 20th century Laos in Finot (1917) (cf. Lafont (1962, p. 395 note 1)), and recently for Thailand by Tambiah (1968). Evidence from catalogues of manuscripts from Ceylon suggests that the contents of the tiptaka have circulated in the same way as, and alongside, a great deal of other literature; both canonical and non-canonical materials, for example, have often been written in the same manuscript. (See de Zoysa (1875, 1885), Wickremasinghe (1900), Gunasena (1901), de Silva (1938), Godakumbara (1980) Somadasa (1987, 1989)).
    Evidence for earlier historical periods may be difficult to collect. But as an example of the kind of evidence we need, I cite a list of four kinds of text mentioned in the commentaries (Ps II 264, Mp V 96-97, identical passages commenting on the same sutta). It is said that when young monks do not show special respect for their elders, they do not receive help from them, either materially, by not being provided with robes, bowl, etc., and not being nursed when weak or ill, or in relation to dhamma: the latter is explained as their not being taught pālim vā atṭhakatham vā dhammakathāhandham vā gulhagantham vā. It is not certain what either of the latter two terms refers to. Adikaram (1946, p. 98) remarks of the former that ‘perhaps it included books that formed the basis of the later tikās [sub-commentaries] or [narrative] works like the Rasavāhini’. It might also refer to books containing texts used in preaching, as in the modern Sinhalese bana books. If so, then like the latter, such compilations would have included canonical and non-canonical material (some of the most famous stories in the Buddhist world, such as that of Kisā-gotamī, being found in commentarial literature). Gulhagantha seems to mean ‘secret books’; not surprisingly, perhaps, it is not clear what they were. The lists of ‘heretical’, Vaitulya works cited earlier (note 32) contain titles with gulha- as a prefix; but I think it is unlikely that in the contexts here being discussed, we are dealing with an ‘esoteric’ literature in the Tantric sense. In the later Pali tradition we find works with gulha in the title, and they seem be elucidations of difficult passages in the Vinaya and Abhidhamma (see Malalasekera (1938, vol. 1 p. 781, vol. 2 p. 883); Bode (1909, pp. 18, 56)). The Visuddhimagga (pp. 115-16) contains a very similar passage, but does not mention dhammakathāhandha; the commentary (cited in Ñānamoli (1975, p. 119 note 35)) explains gulhagantha as ‘meditation-subject books dealing with the truths, the dependent origination, etc., which are profound and associated with voidness’. So it would seem that gulhagantha in this case refers to a class of sophisticated and technical literature on specialist topics. ↩

54. Writing of ‘traditional Buddhist culture’ in Thailand, Keyes (1987, p. 179) has said that ‘three texts - or, more properly, several versions of three texts define for most Thai Buddhists today, as in traditional Siam, the basic parameters of a Theravadin view of the world’: they are the ‘Three Worlds according to Phra Ruang’ (see Reynolds (1982)), the Phra Mali (a 15th century composition based on a Ceylonese story called the Maleyya-Sutta), and the VessantaraJäaka. Only the last of these has a canonical version. This generalisation, he says (p. 181), applies to both popular and elite traditions. ↩

55. Interestingly, one of the reasons for the frequent appearance of Abhidhamma texts in monasteries in Laos and Cambodia, where the Vinaya- and especially the Sutta-piṭakas are comparatively infrequent, is the fact that these texts are used for funeral recitation: the seven texts of the Abhidhamma collection correspond to the seven days of the week (J.S. Strong, personal communication; cf Bizot (1981, pp. 10 foll.)). ↩

56. Thus I think that what Bizot says of Cambodia is true of the whole Theraväda world: ‘the term [tipiṭaka] refers less to a collection of texts than to an ideological concept’ (1976, p. 21). ↩