Buddhist texts form a corpus almost endlessly interesting, and offer a fertile field for explorations both broad and deep.Footnote 1 At a conference held in December 2019, a number of scholars gathered on the campus of the Hebrew University in Jerusalem to discuss “The Idea of Text in Buddhism.” Happening to be in Israel at the time, thanks to the openness of the conveners, and especially to the gracious invitation of Eviatar Shulman, I was able to join the conference and to share some thoughts on a topic that has been at the center of my scholarship for at least three decades. Some of the papers offered during the conference are published here in this issue of the Journal of Indian Philosophy, and having been kindly offered the opportunity to reflect on them, what follows are some notions of my own, mixed with reactions to some of the published studies, although for the most part reference to the papers is tacit. To be clear, my small contribution is not intended as either a summary of those studies or a response to them as such.Footnote 2 In this light, much of what is proposed by my colleagues I pass over in silence, a silence which, however, has no necessary implications either of consent or dissent; on the whole, it points simply to differences in foci. I would also mention here at the outset that if in my partial use of the work of my colleagues I have in any way misrepresented their work or their intentions, this is an unintentional side effect of my pursuit of my own agenda, and nothing in the following should necessarily be construed, therefore, as an accurate portrayal of the position of those whose work is adduced.

In referring to “The Idea of Text in Buddhism,” the first parameter to catch my attention was the scope of the notion, and in particular the narrowness with which the idea of “text” has been engaged in the papers published here.Footnote 3 Leaving aside the paper of Morris, which deals with a very different domain, and with the exception of one contribution (by Wu) that deals with Vinaya materials and their relation with a Chinese travelogue, all offerings focus on scriptures as sūtras, whether those belong to the earlier phases of the Indian Buddhist tradition as preserved principally in Pāli, and also in Gāndhārī,Footnote 4 or are Mahāyāna sūtras, originally composed in either Middle Indic or Sanskrit and often preserved only in translations into, chiefly, Chinese and Tibetan.Footnote 5 Some readers might be misled into concluding that for the participants “texts” are commonsensically equivalent to scriptures, or even assume that this was the intention of the conference organizers. This, however, was clearly not the case for either, and while I cannot speak for others, such a restriction is not what the word “text” would mean to me, for surely minimally, restricting ourselves of course to Buddhism, and even to Indian Buddhism, we would want to include commentaries, śāstras, stotras and so on—that is, we would want to give the word a quite significantly broader signification. In view, however, of what we do have presented in the papers here, we might be led to ask a question: what do we—what might we—learn by restricting our focus in considering “text” (in “the idea of text”) to one particular genre of Buddhist literature, that attributed traditionally to the Buddha, to his “authorship,” even if we consider that attribution fictive? Could it be meaningful to consider results of inquiries limited to scriptures as directly relevant to the manifestly much broader “idea of text”? Pulling our focus back further still, or focusing from another angle, we might wonder: is there any way that we can, perhaps, imagine “text” as an emic category? In other words, might it be that a limitation to scripture is, or somehow could be, validated by an emic restriction of “text” to scripture.Footnote 6 I think not. I have, in fact, the strong impression that the division between works attributed to the Buddha and those considered to have human authorship was fairly clear in Indian Buddhist contexts, although certainly there was also some (perhaps considerable) porosity. In the opening section of Vasubandhu’s Abhidharmakośabhā⋅ya, for instance, he holds (or reports the Vaibhā⋅ikas as holding) the orthodoxy of the Abhidharma, the systematic reorganization of the doctrines propounded by the Buddha, as buddhavacana, authentic “Buddha-word,” since, as he tells us (ad I.3), the Abhidharma was created by Kātyāyanīputra and others (solely) from what the Buddha taught. This Vasubandhu directly compares to the manner in which Dharmatrāta compiled the Udānavarga out of utterances of the Buddha; Dharmatrāta (in this understanding) did no more than place the udānas into vargas, that is, he did not compose, did no more than organize what was authentically, again, buddhavacana (Ejima, 1989, pp. 3.10–14.). There are further examples of works considered as inspired (and thus scriptural) that we might classify differently, and these borderline cases are indeed opportunities to clarify our thinking on such definitions. But the fact remains that in so far as we can think that Indian Buddhists had an idea that might map roughly onto “text,” this idea was not coextensive with “scripture.”Footnote 7

Be this as it may, it is perhaps a rather different question (or for us, challenge) how Buddhists living centuries after the death of Śākyamuni could have nevertheless believed that Mahāyāna sūtras were to be accorded the same status of buddhavacana as texts that (the tradition informed them) had been handed down since the days Gautama walked the earth. What is however clear is that, no matter whether—to refer to a possible counter-example—some held that key Yogācāra śāstric works were inspired, or even dictated, by the future Buddha Maitreya (see Delhey, 2019), the mainstream view and assumption was that śāstric works such as the Mūlamadhyamakakārikā were authored by humans. These texts belonged, therefore, to a different category than did scriptures. The closest term here for “authoritative text” in terms of category might be āgama, and this cannot help but remind us of the two bases upon which valid arguments may be offered: namely they may find their support in reasoning, yukti, or in authoritative text, āgama, distinct from any other form(s) of text, and this, if nothing else, makes clear that the latter is not equivalent to “text” as such.

A very important question of the reception of Buddhist scriptures concerns the relation of the audience to the presenter of the text. In the first place, whom should we imagine as present in an Indian audience of a scripture recitation? In what language was the scripture recited? Were a few/some/many audience members expected to understand (even basically, in a literal sense) what was being proclaimed aloud? Were the scriptures recited in the Sanskrit in which they (that is, as we have them, Mahāyāna sūtras on the whole) present themselves (when we have them in Indic form at all)? Did audience composition—in terms of linguistic competence or social or monastic/lay status—change? Was there one (kind of) audience in the days in which (some early) Mahāyāna scriptures were composed in Gāndhārī, and therefore at least theoretically understandable to persons for whom this was a daily language, and another in later periods in which the language of the texts (and perhaps one of the languages at least of the monastic community) was (whether grammatically classical or not) Sanskrit? Just how widespread was knowledge of Sanskrit within Buddhist communities? Aside from brahmin and perhaps some other converts, are we justified in thinking that ordinary monks (much less nuns, for that matter) could have understood a text recited in Sanskrit? Or should we rather think, as a default, as it were, that the recitation of a text in Sanskrit held a power somewhat closer to that conveyed by the recitation of the Hebrew Bible in a congregation that does not generally understand Hebrew, or the Latin Mass in the Catholic tradition?Footnote 8

Perhaps we might learn something suggestive from a (seemingly) comparable case. The way in which a Sanskrit text (in casu, the Bhāgavatapurāṇa) is recited in Sanskrit to an audience that does not understand the language is discussed by M. Taylor.Footnote 9 In his examinations of the practice of Bhāgavata-saptāha, the seven day ritual of “recitation” of the holy text, what is mostly offered orally is a type of translation-cum-rephrasing/glossing-cum-interpretation that I will discuss below in a different cultural matrix. We do not know whether similar practices existed in Buddhist India, but it would be naive to assume that they did not, if there were indeed public recitations or presentations of Buddhist texts which were performed with any goal other than that of exposing the audience to the sacred sounds.Footnote 10 Naturally, there is a sort of continuum evident in attitudes toward translation both within traditions and across them. Islamic tradition generally prohibits the translation of the Quran (at least for religious purposes),Footnote 11 Jewish tradition has complex attitudes which nevertheless primarily maintain the centrality and priority of the Hebrew text,Footnote 12 while the Buddhists freely translate most varieties of their literature (the case of dhāraṇīs is somewhat different in this regard, and will be left aside here).Footnote 13

Leaving aside questions of language for a moment (I will return to them below), what of the reciter? N. Gummer sees the dharmabhāṇaka as an empowered individual, a virtual buddha in the flesh in a post-Buddha world, capable through his recitation of the Buddha’s words of making buddhahood present. While there is much that remains unclear to me in Gummer’s vision, and I may therefore not correctly understand all of her points, I believe that we have no evidence that scriptures were received in this fashion. Let us try to exercise our imaginations: Should—even, can—we imagine that the dharmabhāṇaka was seen as an embodied buddha if the audience meant to perform this act of conception consisted of the fellow monks of this reciter? I mean this as a human question: Let us stipulate for the moment that, in view of the question of language addressed above, as well as the highly technical content of many of the texts in question, the intended audience of the scripture recitation was monastic. I suggest this because, most especially, the content of a great many (though certainly not all) Mahāyāna sūtras includes quite complex and encoded ideas which would almost certainly have been little more than gibberish to anyone not very familiar indeed with Buddhist doctrine. Now, at least some monks in the imagined monastic audience might well have entered the community as novices together with, or even known as a boy, the individual who had become the dharmabhāṇaka. Given this, would—even could—they have subsequently conceived of him as a virtually supramundane entity in his role as reciter? This is, quite obviously, nothing more than a doubt of mine based on the limitations of my own imagination, something I would not dream of denying. But while I do understand that, for example, it is possible to see a person known to one, even known well, as divinely inspired—when he is in trance, for instance—it seems difficult (to me!) to imagine that the recitation of a text, the author of which was felt (or: known) to be the Buddha, would confer onto the reciter of that text the same holy and transcendent charisma and status. I am not suggesting that this was not an aspiration of the authors of the text, which I understand to be Gummer’s position; I remain agnostic about this. My doubt is rather in the way of a “sociological” speculation, and does not venture into the territory of what the sūtra expects.

But let us briefly unpack this last expression: the sūtra itself expects nothing. The expression, if meaningful at all, must be a shorthand for “the authors of sūtras,” and refer thus to their intentions. This obviously brings up the question of what the authors may have thought, but I fear that we are ill-equipped to answer such a question. In this regard, we might recall the useful articulation offered by Sheldon Pollock (2014). For Pollock, one may seek to understand the meaning of a text to, first, its author, second to (elements of) the tradition which receives the text, and finally one may—or must—consider one’s own subjective position. These three “planes” are for Pollock the historicist, the traditionist (sic), and the presentist modes of interpretation, as he denotes them. My own feelings about this breakdown, as someone almost exclusively interested in the history of the tradition (rather than, for instance, living innovations: I cannot and do not want to speak as an insider) is that the third mode is one entirely foreign to me. I am not, I hope needless to say, naively claiming that my own subjectivity plays no role in my reading, for it quite obviously does; this is unavoidable. Rather, I am saying that there is no particular reason for me to feel, as a historian of Buddhism, that my individual understanding of a text should be of interest to anyone. Neither do I imagine that anyone else thinks that either. Each reader will be entitled to their subjective reactions to a text, but I do not see an etic stance that could value that subjectivity in a scholarly framework: only insiders amongst themselves will be interested in such contemporary readings, and again, I am not an insider.Footnote 14 For entirely different reasons, I do not think that we can by any stretch of the imagination recover the meaning(s) of any Buddhist scripture intended by its author(s). We know nothing, or next to nothing, whatsoever of where, when and by whom any Buddhist scripture was composed (as long as we, again, limit ourselves to India). This means, I believe, that we are entirely and inevitably cut off from access to some putative originary (or historicist, in Pollock’s vocabulary) meaning. This leaves us with what is actually our (my) only object of interest in any event, the ways—always multiple—that a text has been interpreted traditionally, which is to say by Buddhists (and by this I mean almost exclusively Buddhist communities, rather than individuals) in the past. This is what we must, and all that we can, aim to recover.

I think that in one way this leads us back to our question: to whom were the texts recited (and were they recited? When? Where? On specific occasions?)?Footnote 15 The recitation of Vinaya texts appears to have been prohibited to laypersons (or at least the Prātimok⋅a, minimally); what then of all of the narrative content of the Vinaya literature? Was this not shared with the lay community, or was this done only when the stories were extracted, as evidenced by a text like the Divyāvadāna? Is the limitation meant to pertain only to the rules embedded (in the case of the Mūlasarvāstivada Vinaya, very impressively) in mountains of narrative? The plain answer is the one that we are forced to offer repeatedly: we do not know. I will return to the dharmabhāṇaka in a moment, but first I will consider briefly a contribution that led me to what I feel could be a productive examination of thematically or phenomenologically similar or even parallel phenomena.

T. Walker in his paper discusses, as he has elsewhere (e.g. Walker, 2020), what he calls bitexts, bilingual texts. Walker examines the environment and textual sources for materials prepared for the presentation of Pāli texts in a cultural matrix in which Pāli was unknown to all, aside from (some) monks. The consequent necessity for translation—if the texts were to be understood “literally” at all, by which I mean not simply consumed for their sound value, for instance—is a situation not at all uncommon also in other Buddhist realms, perhaps especially in Central Asia.Footnote 16 Walker himself refers to East Asian situations, in which for instance Japanese glossed Chinese texts with kakikudashi readings (though Walker does not mention it, the Uighurs did something very similar as well),Footnote 17 but there are a number of other intriguing Central Asian materials which also speak to this challenge of understanding.Footnote 18 Walker mentions the ways in which the nature of his bilingual documents offers opportunities for an examination of the written and the oral, the nature of a receptive audience, and interpretive avenues. This immediately reminded me of the Jewish tradition of the Targums, Aramaic translations of the Bible text, the complexity of which has been concisely characterized as follows: “A Targum is a translation that combines a highly literal rendering of the original text with material added into the translation in a seamless manner.”Footnote 19 The similarities between the experience described by Walker in the Pāli-Khmer context and the Hebrew-Aramaic interface are striking. Even the typologies are similar. Since the reader can turn to Walker’s paper here, I will not summarize it, but instead consider what interpretive inspirations might be obtained by a brief look at the situation of the Targums.

It is a commonplace that (Stec, 2000, p. 184), “The Targums probably arose in a bilingual setting, and were intended to be used in conjunction with the Hebrew original, as either an exegetical aid or an educational aid to introduce Aramaic speakers to the world and language of Biblical Hebrew.” This description is useful precisely because it is misleading in several ways, the first of which is that it is much better to say “multilingual” in place of “bilingual.”Footnote 20 What is more, the aim of the Targums is itself not simple; there is a continuum from what we are probably compelled to call a “literal” rendering (fully aware of the highly problematic nature of this notion) to the exegesis mentioned in the quotation. Emanuel Tov clearly articulates the issue, first acknowledging its complexity, and clarifying that, “Of course, it cannot be defined objectively what constitutes the plain sense or what qualifies as a deviation. What one scholar (or translator) considers a move away from the original meaning of the text may be defined by another as a reflection of its true sense.” He then states (Tov 2017, 40n12 and 41):

In spite of these problems of definition, scholars usually agree on the general profile of a given version's style and techniques. All translations include a layer of information beyond the original content of Scripture. The scholarly approach to translations considers this layer an added value that was not a necessary part of the translation enterprise. Translations could be produced with only a minimal amount of exegesis, involving merely linguistic exegesis. All elements beyond the linguistic exegesis may be considered content exegesis visible in various forms. Phrased differently, translations involving merely linguistic exegesis are typically described as literal, and those involving content exegesis are often understood as free, with various gradations in between. However, the notion of freedom in the translation process is complex. If it was the translator's intention to transfer to the target language the spirit of the source text, small changes, omissions, and additions possibly should not be considered exponents of freedom. Therefore, the definition of what is a free rendering needs to be analyzed in greater detail. Both translation styles were acceptable in antiquity. A faithful (i.e., literal) approach to the act of translating was considered respectful to the word of God, but there was also room for contextual, free, and paraphrastic renderings. Such translations were also conceived of as presenting the spirit of the word of God, even if from a formal equivalence point of view free renderings deviated from the plain sense of Scripture.

The most important difference that Buddhist scholars will need to take into account in (what I hope will be future) efforts to learn from studies of the Targums is that the Hebrew Bible text (the Masoretic text, MT) had long been (virtually) completely static, and thus there is little, and often even no, question of the precise shape of the Vorlage of a given translation.Footnote 21 This is most manifestly not the case with Indian Buddhist literature, and thus de facto we cannot ask some of the most fundamental questions asked by scholars of the Targums, namely whether a given rendering is “literal” or “interpretive.” Given the vibrant fluidity of Indian Buddhist textual composition and transmission, we have no touchstone against which to test the fidelity of any translator or translation team to its source text.Footnote 22

Thinking more about the short passage cited at the outset of this discussion, above, a further question is whether the Targums were genuinely intended “to introduce Aramaic speakers to the world and language of Biblical Hebrew,” to which I think the answer is rather that their purpose was to make accessible the meaning of the Bible text, but not to make the Hebrew itself more intelligible to those audience members (often the vast majority) ignorant of the language. The parallels here to Buddhist cases are also clear: I know of few cases in which Buddhist translations (in premodern times) were intended as guides to language acquisition or improved understanding of the original being glossed or rendered. (This is not to say that they could not more often have been used in this way, which of course is quite possible; I speak here only of basic intent.Footnote 23) It is clear that Rabbinic traditions recognized both the existence of communities in which knowledge of Hebrew was not common, or even absolutely non-existent, and the felt need for comprehension of the text (see inter alia Smelik, 2007). All of this raises, of course, the issue of “Church language” (though we should perhaps rather prefer “Synagogue language” in this context!). Attention has of course been given to questions of translation in Buddhist traditions, but to the best of my knowledge, for further inspiration and to help us pose better questions no advantage has yet been taken of the quite substantial Jewish reflection on scriptural language, translation, community practices and authority as found both in primary sources (including in the first place the Talmuds and related literature) and in modern scholarship. It is striking that in contrast to Buddhist traditions (even in Tibet and to a lesser extent in China), Jewish sources devote considerable attention to theorizing translation and its dynamic relation with interpretation. It is possible that, as noted above, this difference can be attributed to the fact that the Vorlagen of Jewish translations (referring here to books of the Bible) were both settled in their wording and also, and perhaps equally as important, textually limited. It may be that alongside its textual fluidity the very vastness of the Buddhist canons worked to discourage theoretical reflection on issues of translation and interpretation. What is also important, however, is that the bi- or multi-lingualism characteristic of Jewish communities was rarely a feature of Buddhist communities, at least of non-monastic communities. Leaving aside communities of Southeast Asia, such as those engaged with by Walker, with which I am not very familiar, I know of no evidence for any significant multilingualism among possible audiences in China or Tibet.Footnote 24 It is worthwhile noting here that, most generally, the basic question prompting translation is the (perceived or imagined) desire of an audience to understand the text. However, it is very clear that at least outside the Indian Buddhist world—and let it be stressed that for India we have close to no evidence, and thus no idea of whether comprehension was valued or not—understanding of texts was not always a requirement. Many text practices from Tibet (where Kanjurs can be publicly recited, but in a cacophony with multiple volumes of texts being read out simultaneously) to Japan (where in the practice of tendoku 転読 texts are not even vocalized, but the volumes merely flipped through in a pantomime of rapid reading) illustrate the lack of interest in the content of texts which are, nevertheless, publicly performed.

Returning for a moment to the dharmabhāṇaka and his role, on the one hand, and the intermediary interpreters in, for instance, Southeast Asia, on the other, it is most interesting to consider Jewish views, especially as these relate to the meturgemanim. These were the interpreters whose role it was in the synagogue service to translate from Hebrew into Aramaic the bible passages being read out from the Torah scroll, this translation ultimately constituting the Targums. This is outlined by S. Fraade, who wrote (1992 p. 266):

The practice of Targum as it mediates the reading of Scripture in the synagogue is compared in several texts to what is rabbinically understood to have occurred at Mt. Sinai, of which, in a sense, it is a performative reenactment. Just as God’s word was ‘given in reverence and fear,’ and just as it was ‘given by way of a middleman (sirsur),’ being Moses, so, too, the public reading of Scripture must be reverently mediated by a meturgeman.…A similar tradition relates that just as God’s voice and Moses’ voice complemented each other at Sinai, so, too, the voices of Torah-reader and translator must accommodate one another so that neither rises above the other. Thus, both at Sinai and in the synagogue it is by a dialogical combination of voices that revelatory communication is effected.

A further notion is also of interest. After explaining the tradition that the initial revelation of the Torah on Mt. Sinai took place in four languages, which later sources expand to (all the) seventy languages (of the world), Fraade observes (1992, p. 268), “Thus, to translate a text of Scripture into one of these languages may be thought of not so much as a distancing from Sinai as a return to it.” The thematic parallelism to the role of the dharmabhāṇaka is intriguing.Footnote 25 As I said above, while I am reluctant to accept (what I understand to be) Gummer’s portrayal of the way the dharmabhāṇaka would have been received in communities, it could well be fruitful in this process of imagination, despite all the many cultural and contextual differences between the situations, to think as a point of comparison about how the meturgemanim and their role were understood.

These are not the only possible reflections on roles. For M. Allon, there were two main functions of early scriptures (he does not share with us how he knows this): to record the teachings of the Buddha, and to attract converts and supporters. Allon further seems, at least rhetorically, to accept that the ultimate author (or perhaps we had better write “author”) of the sermons recorded in the scriptures was the Buddha. He suggests that the teaching career of the Buddha necessitated editing to eliminate repetitions. I believe that we can imagine, hypothetically taking the tradition’s narratives as historically factual for the moment, that as the Buddha wandered from place to place he indeed, as Allon suggests, returned time and again to the same theme. For me, this would imply, first, that there were multiple authoritative and authentic versions of a given sermon or on a given theme, and second that there was inherent variability in the corpus. But I do not see that it follows that this called for any authoritative culling and editing, any felt need for which would imply that the editors had at their disposal multiple records of the variant sermons. It further would require that those editors acknowledged these multiple versions—variant texts—as witnesses, out of which—precisely like a modern editor, assuming a stemma codicum and with a Lachmanian bent of mind—they sought to establish the Ur-text. Is this plausible? Is it not rather more likely that from the beginning there were multiple sermons and that only later with the attempt to compile an authoritative collection (when and where did this take place?) those who created what was to become the Pāli canon selected—for reasons at this remove entirely unknowable—some transmitted version, which they “canonized.”Footnote 26 It is of course entirely plausible that in some cases—again, for reasons unknowable—they preserved as “separate” sermons what may appear (to us) as (mere) parallels or variant versions. The picture sketched by Allon sees the versions preserved elsewhere (and paid little attention by Allon, or by Shulman), for instance in Chinese, or in Sanskrit (fragmentarily) or Tibetan (for instance as preserved in the translation of the Abhidharmakośopāyika of Śamathadeva) as of lesser validity than the sources in Pāli; and if this is indeed their view, I do not understand it. If our goal as philologists is to uncover as much of the textual history of Buddhist literature as possible, I think that full consideration of all preserved evidence is essential, and in this regard the Sanskrit, Chinese and Tibetan sources must be fully utilized. Furthermore, unless and until some rigorous and convincing evidence can be provided that one version is more original than all others, I remain unwilling to grant the priority to Pāli sources that at least Allon seems to assume.Footnote 27

In fact, I would go further: we know precisely nothing of the original or initial context of any Buddhist scripture in India, and this has an undeniable impact on what we can say about intertextuality and even reception history. Likewise, we know precious little more about the reception history—at least the earlier such history—of any Indian Buddhist scriptural text. What we do know about is the modern reception history. But I would maintain that the study of the latter belongs to a domain other than that of what we—what I—term Buddhist Studies.Footnote 28 We must therefore carefully examine how our assumptions frame our thinking about Buddhist texts in India, most particularly in light of the fact—and I think it is a fact—that we cannot test those assumptions against any touchstone of historical data. One thing I think we can safely assume—but it is an assumption—is that in India (as elsewhere) traditionally texts were read (whether that means through writing or through aural reception), when they were read for understanding, backwards, as it were, through commentaries, whether those commentaries for their part were written or presented by a teacher. Now, it is a very curious fact that the only Buddhist tradition for which we have evidence of a comprehensive (written) commentarial tradition is that of the so-called Theravādins. While it is true that the textual traditions of other lineages are preserved in a fashion much less complete than that of the canon preserved in Pāli, there is no evidence among the remains of what once were the canons of the Sarvāstivādins, Dharmaguptakas or others of the existence of the sort of exhaustive commentarial literature found in the Pāli tradition. In addition, while commentaries on Mahāyāna sūtras do exist, they are relatively few (in India; I am again not considering the productivity of the Chinese tradition, since it clearly belongs to an entirely different sphere). For the Mahāyāna sūtra literature, the role of commentary appears to have been usurped in two different ways: first, the tradition produced an extensive śāstric literature, which to some extent did fill a role close to that of commentary (and in this respect some śāstras perhaps less closely parallel the Abhidharma tradition, to which the Mahāyāna śāstras are normally compared). But the second is that Mahāyāna sūtras sometimes give the impression of serving as commentaries themselves on other Mahāyāna sūtras, a hypothesis yet to be properly explored.

The role of the commentary in Indian Buddhism is not fully appreciated for other reasons as well. Modern scholars generally seek a mode of understanding that approaches what the original meaning is imagined to have been, despite the recognition that in fact we know nothing of the original/initial state or context of any given scripture (Pollock’s first dimension, as above). Without knowledge of such circumstances, one may naturally turn to contemporary materials in an effort to (re)create a context. But, for ancient India we have no idea what was contemporary or co-local, so we cannot imagine an originary context. Scholars (such as Shulman, if I read him correctly) therefore take the canon as transmitted as the context. A question we should then ask ourselves is whether we labor under a tyranny of received corpora. The Pāli canon may well represent—almost certainly does represent—an “official” corpus (though how old this status may be seems to remain unclear), but we know that our picture of other Āgama, Vinaya and Mahāyāna materials from India is radically partial and fragmentary. Nevertheless, we attempt to paint murals of the scriptural landscape on this basis, rarely acknowledging the poverty of our sources (and this leaves aside the sad fact that all too many scholars pay attention only to materials with preexisting modern translations; few venture into the oceans of the Kanjur or Dazangjing or even the Pāli commentarial literature on their own). Yet we know from all sorts of sources that the organized materials we have, that is, the collections, are not necessarily representative of anything. We know that various Tripiṭakas of Indian sects, in so far as we can recover them, had differing structures and contexts. The Sthaviravāda and Mahāsāṁghika Vinayas differ radically in structure, and so on. We have, moreover, Sammelhandschriften the existence of which demonstrates that the combinations of the Mahāyāna scriptures we see now in, for instance, the 49 sūtras of the Mahāratnakūṭa or the 17 of the Mahāsaṁnipāta were far from the only collections in circulation (and we do not have evidence for the existence of either of these collections in India in any case!).Footnote 29 If we acknowledge that we do not know the reasons for the preservation of the materials available to us, some of which we certainly know to have been randomly preserved (for instance the texts recovered from the sands of Central Asia), should not this temper our confidence in our reconstructions of the intellectual environment of ancient Indian Buddhism?

This also brings to my mind one issue raised on occasion, that of a putative distinction between local sources and, presumably, translocal sources. I do not understand this distinction, or at least, not in the sense in which it seems to be intended. I cannot imagine a scenario in which the creation of a text is anything other than local; the distinction between local and translocal texts, then, can be nothing other than a difference along a spectrum of popularity. Texts which were more popular, and which found audiences outside the regions of their creation, transitioned from local to translocal. However—and this is a point we should never forget when studying ancient India—our available evidence is drastically partial. Absence of evidence, in the old adage, is not evidence of absence. That we have no evidence for something means nothing more than that: we do not have evidence. Now, our lives, no doubt, revolve around ourselves, and what is available to us may well claim the lion’s share of our attentions. But as long as we play-act as historians, we must force ourselves to remain aware of our limitations.Footnote 30 In fact, in a related point, we often speak of texts as “obscure,” but we must certainly acknowledge that no text was obscure to its author(s) or to his/their community. It is a repudiation of our scholarly responsibility not to judge to relegate some works into categories such as “obscure” or “minor,” unless the latter is used to indicate size.Footnote 31

The few considerations offered above certainly constitute no comprehensive consideration of the multiplicity of vital and fascinating issues raised by the quest to think carefully about Indian Buddhist texts: they are in fact little more than ramblings. I do, however, perhaps not without some hope, believe that at least some of what I have wondered about here—much in the manner of thinking aloud—can be of use to others as they continue to engage with these absorbing and provocative sources.