(Originally posted on April 27. Major revisions on June 3, 2022.)

In chapter 5 of A Buddha Land in This World, I wrote that

until the sūtras in the Pāli canon were written down they were recited in periodic meetings of monks, but we have no consistent evidence about the nature, form, and frequency of these meetings, nor about how reliable this process was.1

However, when I reread this, I wasn’t entirely happy with this sentence because it seems to suggest that I think that oral transmission is the biggest problem for the authenticity of the content of the Pāli Canon. Such skepticism about the reliability of oral transmission is indeed a common source for doubt in the reliability of the Pāli Canon, but I don’t think that this is the biggest problem. Furthermore, after further study, I have become considerably less skeptical about the reliability of the Pāli Canon recently.

According to traditional accounts, in the First Council, soon after the Buddha’s death, Ananda and Upala recited the texts that became the Pāli Canon. The historicity of the First Council has been contested – that is, it is doubtful that it ever took place – and it is unlikely that the whole of the extremely voluminous Pāli Canon was recited by just two people. Even if we ignore those problems, there is another problem. Whoever retold the events recorded in the sūtras did so from memory. Some sūtras record events that took place four or more decades earlier, and whoever recited those was almost certainly not even present at the events described in them. Furthermore, decades of research on memory has taught us that memory is nothing like a recording device, as often mistakenly assumed. Every time you remember something, you reconstruct it on the basis of a few memory “traces”. The rest you (unconsciously!) make up on the spot to assure a more or less consistent story (and because this is largely unconscious, the result tends to be colored or even distorted by other memories and knowledge). In other words, memory is very unreliable.

If it would indeed be true that all the sūtras were originally recited by just one person, Ananda, in one long meeting, then it is almost certain that very little of the content of the original versions of the sūtras would be authentic, just because of the unreliability of memory. However, because the First Council is probably a myth, and because it is much more likely that (many) others were involved in the creation of the original versions of the sūtras, they may actually have been closer to historical reality in some respects. While the memory of a single individual is unreliable, the larger the number of individuals involved, the less unreliable the collectively agreed upon historical reconstruction.

Most sūtras probably originated in a story or a collective retelling of a teaching or sermon by the Buddha or some high-ranking monk or some of some significant event. These stories could be considered some kind of proto-sūtras and the closer in time their retellings were to the original teaching or event and the more people who witnessed that teaching or event were involved in the retelling, the more accurate the proto-sūtra. Such proto-sūtras are still far removed from the sūtras we know from the Pāli Canon and other sources, however.

Sutras were “designed to be memorized and transmitted verbatim”, writes Mark Allon.2 Towards this end, they were highly standardized, extremely repetitive, and made extensive use of metric schemes and other tricks to aid memorization and ritual recitation. Or in Allon’s words:

They are not verbatim, or tape-recorder, records of the saying and discourses of the individuals concerned nor casual descriptions of their actions and related events. They are highly structured and stylized, extremely formulaic and repetitive, carefully crafted constructs, at least as we have them. And this is so at all levels. Further, the wording used to describe or depict a given event, concept, teaching, or practice is highly standardized across the corpus of such texts transmitted by a given monastic community. As such they do not reflect how a person would normally speak, preach, debated, and interact, or describe an event.3

This process of crafting and standardizing sūtras probably took many decades, if not several centuries, and may have started already during the Buddha’s lifetime (rather than after his death). Furthermore, in the following centuries, the texts continued to change, albeit often subtly. However,

despite these changes, the core of these texts, the ideas and teachings they promote, the general account they give of events, and possibly also much of the wording is likely to stem from the period immediately after the death of the Buddha based on material composed during the Buddha’s lifetime.4

After their creation, sūtras were preserved in collective recitation for several centuries. This is probably mostly a quite reliable process, as mistakes by a single monk are unlikely to have much effect. Except if that single monk has a position of authority within his community. There is significant divergence between sūtra versions preserved in the Pāli Canon, in Chinese, or found in Gandhāra or in Tibetan archives. Some of such differences seem to be due to translation, but there are also differences that seem to have an ideological or other background, and most of such changes must have been made intentionally by authorities within their sects or schools.5

According to the standard account, the sūtras in the Pāli Canon were written down in Sri Lanka between 29 and 17 BCE. This may have not been the first time they were written down, and certainly wasn’t the only, but only parts remains of the alternative Canons. We only have fragments of what may have been the Gandhāri Canon,6 Chinese translations of some alternative versions (belonging to different sects) of some sūtras, and some further variants found in Tibet (either in Sanskrit or translated into Tibetan). Whether the Pāli Canon was really written down around the time it traditionally is believed to have been is not known, moreover. The first reliable sources we have indicating that there was some kind of Pāli Canon (and thus, not even that Canon itself!) date to the fifth century. Hence, it seems that a version of what became the Pāli Canon was written down some time between the end of the first century BCE and the fifth century, but not necessarily the version we know now. Furthermore, as Steven Collins writes, “The Pali Canon, like most other religious Canons, was produced in a context of dispute, here sectarian monastic rivalries.”7

(According to Gregory Schopen a process of redaction lead to convergence of texts, such that later versions of the same texts are all very similar to each other, but sometimes deviate significantly from earlier versions,8 but Alexander Wynne and Bikkhu Anālayo have shown this idea to be false: convergence between texts from different traditions strongly suggests that those texts (or parts of texts) are old.9)

(It should also be taken into account that in this this phase of the history of the Pāli Canon oral transmission no longer could perform the checks it had performed before, because authority gradually shifted from memory to texts – that is, the textual versions gradually became authoritative. This probably wasn’t even a conscious or intentional process, but just a byproduct of the availability of texts. If monks still memorized sūtras during this period, then they probably memorized those from texts, for example.)

While we know that there was some kind of Pāli Canon in the fifth century, and thus that something was written down and edited in the centuries before that, this isn’t the Canon we have now. Sectarian struggles – and sectarian editing to support the sects’ views – didn’t end in the fifth century. There is another phase in the Canon’s history, which starts with the gradual emergence of some kind of Canon (i.e. probably in the fifth century) and ends in the twelfth century when king Parakkamabāhu I decided that the Canon of the Mahāvihāra sect was authoritative and suppressed the other sects and their Canons (in Sri Lanka). Significantly, Helmer Smith has shown that the language in which the Pāli Canon is written also dates to the twelfth century.10 It is worth emphasizing once again, that even though the Pāli Canon was supposedly written down in the end of the first century BCE, the version we actually have dates to the twelfth century, well over a millennium later.

So these are the rough stages of the history of the Pāli Canon:

  1. Creation of the original versions of the sūtras etcetera on the basis of memory traces by early monks in the years (or decades) after the Buddha’s death.
  2. A period of oral transmission lasting at least five centuries.
  3. An initial period during which the sūtras etcetera were written down and edited in a context of inter-sectarian rivalry lasting until the emergence of some kind of Canon in or before the fifth century.
  4. A period of continuing sectarian rivalry ending with the royal selection of the Canon of one particular sect and freezing thereof in the twelfth century.

The issue here is not this history itself, but what all of this means for the historical authenticity of the Pāli Canon. The first phase is obviously problematic. Even the original versions of the sūtras constructed in this phase did in many cases not literally record what the Buddha had said or done, but a standardized and ritualized version of what his followers thought they remembered that the Buddha had said and done, and memory is not as reliable as we would like it to be. Nevertheless, it is unlikely that the earliest sūtras deviated much from the Buddha’s teachings in spirit, because there were probably multiple witnesses involved in the creation of the sūtras. They may not have been literal, word-for-word recordings, but they were probably reasonably accurate when it comes to the Buddha’s teachings and ideas.

The second phase is probably less problematic than the first. As Mark Allon’s work shows, significant changes were made in this period, leading to a divergence of the canons of different sects, but it seems unlikely that this lead to major changes in the basic ideas recorded in the sūtras.11 I used to think that the third phase is the most problematic phase. The popular belief is that Buddhist philosophical ideas were based on sūtras, but in reality it was probably also the other way around: sūtras were adapted to the philosophical and other doctrines of a sect (and in case of the Mahāyāna, entirely new sūtras were composed for that purpose). It must be noted, however, that this often doesn’t require very major changes. Apparently small changes can significantly change the meaning and/or philosophical implications of a text, and as sectarian rivalries can concern the tiniest of details, sometimes a difference in a single word (or even the order of two words) can support the doctrine of one sect versus another. It is almost certain that changes were made in this third phase of the history of the Pāli Canon, but they were almost certainly very minor. The lack of Sinhalese influence on the language in the Pāli Canon strongly suggests a lack of major changes,12 but as mentioned, radically different philosophical views sometimes require only the tiniest and subtlest alterations. Nevertheless, I now think that it is likely that more significant changes were made in the sūtras before they arrived in Sri Lanka than after, which would imply that the first phase and early part of the second phase had a much bigger impact than the third. Probably, the extent of editing in the fourth phase was (even) smaller than in the third, but it is hard to be sure as the only version of the Canon we have is the twelfth century version of the sect that won these last seven (!) centuries of sectarian struggle.

All in all, the whole history from the first creation of the sūtras until the finalization of the version of the Pāli Canon we have now lasted about seventeen (!) centuries.13 And all we have is one version of the Canon of one particular sect that dates to the very end of those seventeen centuries of crafting, editing, ritual recitation, writing, re-writing, and so forth – all against a background of sectarian struggle. I used to think that this leads to an obvious conclusion: the Pāli Canon is best regarded as a work of historical fiction. However, I recently changed my mind. I now believe that while the Pāli Canon cannot be considered authentic or historically accurate in some strict sense of those terms, its source material probably was. We might not be able to read what exactly the Buddha said on some occasions, but careful and critical reading can probably tell us a lot about what he taught.

But even if it couldn’t, even if the Pāli Canon would be entirely fictional, would that matter? Personally, I don’t think it would. It would matter if you would believe that the Buddha was some kind of oracle and that everything worth knowing was taught by the Buddha, but that would be an absurd belief. Inoue Enryō, the father of Japanese Buddhist modernism and secular Buddhism wrote in 1887 that:

Although there is much talk among Christians that the original texts of Buddhism are Indian, that Mahāyāna is not the Buddha’s teaching, that Śākyamuni [i.e., the Buddha] really did no exist, and so forth, this doesn’t even concern me a little bit. That person’s biography may not be detailed/accurate and the origin of those teachings may be unclear, but I would never be so blind and ignorant to believe those teachings based on biography or origin. I will only believe it if it agrees with today’s philosophical reasoning, and I will reject it if does not.14

It seems to me that this is a much more fruitful – and much more defensible! – attitude. It doesn’t really matter whether the Pāli Canon is authentic,15 or even whether Mahāyāna sūtras are authentic; what matters is whether these texts and the philosophical tradition associated with them can provide us with ideas that stand up to scrutiny and that get us closer to truth and universal well-being.

If you found this article and/or other articles in this blog useful or valuable, please consider making a small financial contribution to support this blog, 𝐹=𝑚𝑎, and its author. You can find 𝐹=𝑚𝑎’s Patreon page here.


  1. Lajos Brons, A Buddha Land in This World: Philosophy, Utopia, and Radical Buddhism (Punctum, 2022), p. 141.
  2. Mark Allon (2021), The Composition and Transmission of Early Buddhist Texts with Specific Reference to Sutras (Bochum: Project Verlag), p. 1.
  3. Ibid., p. 10.
  4. Ibid., pp. 3-4.
  5. For a detailed account of the various kinds of changes in the Pāli Canon, see: Allon, The Composition and Transmission of Early Buddhist Texts.
  6. Richard Salomon, The Buddhist Literature of Ancient Gandhāra: An Introduction with Selected Translations (Boston: Wisdom, 2018).
  7. Steven Collins, “On the Very Idea of the Pali Canon” (1990), in Buddhism: Critical Concepts in Religious Studies, Volume I: Buddhist Origins and the Early History of Buddhism in South and Southeast Asia, ed. Paul Williams (London: Routledge, 2005), 72–95, p. 77.
  8. Gregory Schopen, Bones, Stones, and Buddhist Monks: Collected Papers on the Archaeology, Epigraphy, and Texts of Monastic Buddhism in India (Honolulu: University of Hawai‘i Press, 1997).
  9. Alexander Wynne (2005), “The Historical Authenticity of Early Buddhist Literature: A Critical Evaluation”, Wiener Zeitschrift für die Kunde Südasiense 49: 35–70. Anālayo (2012), “The Historical Value of the Pāli Discourses”, Indo-Iranian Journal 55: 223–53.
  10. Helmer Smith, Saddanīti: La grammaire palie d’Aggavaṃsa, Vol. 1 (Lund: Gleerup, 1928), vi.
  11. Allon, The Composition and Transmission of Early Buddhist Texts.
  12. K.R. Norman (1997), A Philological Approach to Buddhism (London: School of Oriental and African Studies).
  13. It is worth emphasizing that there is twice as much time between the (finalized) Pāli Canon and the lifetime of the Buddha as there is between the Canon and us.
  14. 故にヤソ教者中、インドに仏教の原書なし、大乗は仏説にあらず、釈迦は真に存するものにあらず等と喋々するものあるも、余がすこしも関せざるところなり。その人の伝記つまびらかならず、その教の由来明らかならざるも、余は決して伝記由来をもって、その教を信ずるがごとき無見無識のものにあらず。ただ余がこれを信ずるは、その今日に存するもの哲学の道理に合するにより、これを排するは哲理に合せざるによるのみ。 — Inoue Enryō, 『仏教活論序論』 (1887), in 『井上円了選集』, Vol. 3 (2003): 327–93, at 327–28.
  15. Except from a historical point of view, of course!