How Were the Buddha’s Discourses Preserved?

A world in which the book was not the center of knowledge

It is difficult for a modern reader to imagine a culture in which the most important texts are kept not in a library but in human memory. In ancient India, oral transmission was neither unusual nor primitive. Long religious and philosophical texts were memorized, recited, and transmitted from teachers to students. The Buddhist community arose in an environment where such preservation of knowledge was already highly developed, and the early Buddhist suttas still bear the marks of that world: formulas recur, the same passages appear in several discourses, and teachings are organized into memorable lists such as the Four Noble Truths, the five aggregates, the six sense fields, the seven factors of awakening, and the Noble Eightfold Path.

A narrator often repeats almost the same structure several times, changing only one element. To a modern reader this can feel long-winded, but for someone who had to hear a text, memorize it, and recite it correctly, such a structure was extremely useful. Repetition, fixed formulas, and ordered lists were important aids to oral transmission.

“Thus have I heard” and the memory of the community

Many Buddhist sutras begin with the formula “Thus have I heard,” in Pāli evaṃ me sutaṃ. The text then often states where the Buddha was staying, who was present, and to whom he spoke. This is not an authorial signature in the modern sense, but a formula that places the text within a chain of transmission: the teaching was heard, preserved, and passed on. Later Buddhist tradition closely associated this formula with the Buddha’s disciple Ānanda, who is portrayed as an exceptional knower of his discourses.

This is also connected with the famous account of the first Buddhist council. According to traditional narratives, five hundred monks gathered soon after the Buddha’s death. The meeting was said to have been led by Mahākassapa, or Mahākāśyapa, and its purpose was to preserve the teaching and monastic discipline. Mahākassapa is said to have questioned Upāli about the Vinaya and Ānanda about the Buddha’s discourses; Ānanda then recited the teachings he had heard from the Buddha. Accounts of the first council are preserved in several Buddhist traditions, so the idea itself is very old, but traditional memory must be distinguished from historical reconstruction. We cannot demonstrate that five hundred monks at a single meeting recited the entire canon in the form known centuries later. A more cautious formulation is that after the Buddha’s death there was probably a very early need to organize, recite, and preserve his teaching and discipline, and the traditional account expresses that need in the image of a great first council.

Nor was the material preserved by a single person. The idea that one monk had to carry the entire Buddhist canon in memory is too simple. The early Buddhist community knew people associated with reciting, preserving, and explaining texts; later sources use the term bhāṇaka, often translated as “reciter.” Such people were not necessarily merely living recording devices, but also performers, storytellers, interpreters, and teachers of texts. Transmission was therefore a collective activity. If a text is known by a group of people and regularly recited together, an individual’s mistake need not become the mistake of the whole tradition, because others can correct it. The system is best imagined as a living library of people, not as a long chain of individuals whispering the same story to one another.

Stability without complete freezing

Oral transmission can be very stable, but that does not mean every discourse was frozen word for word from the first generation onward. Comparison of different surviving versions of early discourses shows both: much in common and real variation. Sometimes the order differs, sometimes an example, a place name, or a person; one version may be longer and another shorter, and a comment or explanation may gradually become part of the main text. The history of the early sutras is therefore probably not the history of an exact transcript, but of preserving stable teachings, structures, formulas, and narratives within which variants could emerge.

After the Buddha’s death, the Buddhist community also did not remain a single unchanged organization. Over time, different schools and transmission lineages arose. They inherited older material but organized their collections somewhat differently. The Theravāda tradition preserved the Nikāyas in Pāli, while other early Buddhist schools had related collections often called Āgamas. Much of these other Indian canons did not survive in their original languages, but many texts were preserved in Chinese translation, and some in Sanskrit fragments and Tibetan versions. At first this may seem like a problem, but for historians the branching of traditions is extremely valuable.

If two communities inherit common material, separate, and then transmit it independently for centuries, their later versions can be compared. Differences show us where transmission developed in distinct ways; similarities reveal older common roots. If the same or a very similar teaching appears in several early traditions, it is reasonable to infer that their shared core goes back to a period before the later development of the individual schools. Such similarity still does not prove that the historical Buddha uttered a specific sentence exactly as it stands, but it does show that a text or teaching probably belongs to an older common layer of Buddhist tradition.

When did the teachings begin to be written down?

With writing, too, we must distinguish between tradition and material evidence. Writing was known in India long before the oldest surviving Buddhist manuscripts were produced, but that does not mean writing immediately became the principal means of preserving the Buddhist canon. Theravāda tradition places the systematic writing down of the Pāli Canon in Sri Lanka around the 1st century BCE, often around 30 BCE. According to the traditional account, the move to writing took place during a period of political turmoil, famine, and danger to the monastic community, so writing served as an additional safeguard against the loss of the tradition.

But we do not possess this event in a manuscript from the 1st century BCE. It is a later traditional account of when the canon was supposedly committed to writing. It is therefore more accurate to say that Theravāda tradition places the writing of the canon in the 1st century BCE, rather than to claim that we know demonstrably that the entire present-day canon was written down at that time in its current form.

The material evidence takes us especially to Gandhāra, in the region of today’s northern Pakistan and eastern Afghanistan. Some of the oldest surviving Buddhist manuscripts come from this wider cultural zone. They are written on birch bark, in the Gāndhārī language, using the Kharoṣṭhī script. Important collections are dated roughly around the 1st century BCE and 1st century CE, though dates vary among individual collections. They contain discourses, verses, narratives, commentarial and other doctrinal material, and some texts have relatives in Pāli, Sanskrit, or Chinese translation. The oldest material glimpse into early Buddhist libraries therefore does not reveal a single unified “book of the Buddha,” but a branching multilingual and multi-school Buddhist world.

Gandhāra also changes how we think about an “original language.” We sometimes speak of the “original Sanskrit” of a Buddhist text as if all ancient Buddhist texts had first been written in Classical Sanskrit. The history is much more complicated. Buddhist texts circulated in different Middle Indo-Aryan languages: Pāli represents one literary tradition, Gāndhārī another, and other Prakrits and regional forms also existed. Some texts were later Sanskritized, while others are known only from Chinese or Tibetan translations. The Gandhāran manuscripts show especially clearly that Buddhist textual history is not a simple path of Buddha → Sanskrit → all other languages. Some early Chinese translations may themselves have been based on Gāndhārī or other Middle Indo-Aryan exemplars. The search for one single “original Buddhist book” therefore quickly becomes problematic. Such a book probably never existed.

Writing also did not bring the oral tradition to an end. Manuscripts on palm leaf or birch bark are fragile; moisture, fire, insects, and war destroy them, so they had to be copied repeatedly. Recitation continued at the same time. The transition was therefore not first memory, then book. For many centuries both systems existed together, and a text could be copied, recited, memorized, explained, and ritually performed.

When preservation is joined by translation

As Buddhism spread beyond India, texts had not only to be preserved but also translated — into Chinese, Tibetan, the languages of Central Asia, and later many others. Translation is never completely neutral. What should be done with terms such as dharma, nirvāṇa, śūnyatā, or dukkha? Should they be translated, transliterated, or rendered with a local philosophical term that is only approximately equivalent? Every choice can change how a reader understands the text. For that reason, studying an important sutra often requires more than consulting a single modern translation; different old versions must be compared. Their differences can reveal how the text developed.

What, then, was preserved?

The answer depends on the question. If we ask whether we possess an audio recording of the Buddha’s speech or a perfectly verbatim transcript of his words, the answer is no. We have no such evidence. If we ask whether early texts changed, expanded, contracted, and were organized differently over the centuries, the answer is yes; comparison among traditions shows this clearly. But if we ask whether enough common material survives across different early traditions to identify very old layers of Buddhist teaching, the answer is considerably more positive.

It is precisely the existence of several surviving traditions that makes comparison possible. Again and again we encounter the Four Noble Truths, dependent arising, impermanence, non-self, the five aggregates, the sense fields, meditative training, ethical discipline, and liberation. Individual formulations may differ, but much of the conceptual world clearly reaches back to a very early period of Buddhist tradition.

Early Buddhist texts therefore present an unusual paradox. We do not possess an original, a manuscript written by the Buddha, a transcript, or written texts from the first generations. What we do have are several independently preserved descendants of an older tradition: Pāli suttas, Chinese Āgamas, Sanskrit fragments, Gandhāran manuscripts, Tibetan translations, and monastic-discipline texts from different schools. Where these sources overlap, their similarities and differences allow us to look toward an older common core. The loss of a single “original” has thus left us another kind of evidence — not one book, but a network of witnesses.

The question is therefore not only, “Did the Buddha really say this?” For historical research, it is often more useful to ask how early a teaching can be traced, in how many independent traditions it appears, what the versions share, where they begin to differ, whether a passage belongs to an older structure or a later expansion, and which concepts already occur in the earliest comparable layers. The history of early Buddhism is reconstructed not from faith in one perfectly unchanged text, but through comparison.

The evidence therefore points to a more complex picture than either extreme — “everything is authentic” or “nothing can be known.” The Buddha’s teachings lived in human memory; communities recited them; textual specialists preserved, performed, and explained them; and different groups organized them into collections. Buddhist schools diverged, texts developed distinct versions, some were written down, others were lost, and still others were translated into languages thousands of kilometers away. When we place those different branches beside one another today, we can still recognize common roots. The next question therefore follows naturally: What exactly are the Pāli Canon, the Āgamas, and the Mahāyāna sutras — and why are they not one and the same collection?

Sources and literature

Primary texts and reference editions

• SuttaCentral — early Buddhist texts, translations, and parallels; for Pāli texts, Bhikkhu Sujato as the principal digital reference translation.
• 84000: Translating the Words of the Buddha — a reference digital corpus for published Tibetan canonical Mahāyāna translations.

Research and verification literature

• Allon, Mark. Style and Function: A Study of the Dominant Stylistic Features of the Prose Portions of Pāli Canonical Sutta Texts and Their Mnemonic Function. Tokyo: International Institute for Buddhist Studies, 1997.
• Allon, Mark. The Composition and Transmission of Early Buddhist Texts with Specific Reference to Sutras. Hamburg Buddhist Studies 17. projektverlag, 2021.
• Anālayo, Bhikkhu. Early Buddhist Oral Tradition: Textual Formation and Transmission. Wisdom Publications, 2022. ISBN 978-1-61429-827-4.
• Gethin, Rupert. “Playing with Formulas.” Indo-Iranian Journal 68.1 (2025): 35–56. DOI 10.1163/15728536-06801004.

Note: verification sources are part of the editorial apparatus. Where the original edition used during reconstruction was not preserved, a standard reference edition is not presented as though it were the presumed original source of the article.