The Pali Canon is the scripture collection of Theravada Buddhism. It has a variety of names in both English and Pali. "Pali Canon" is the commonest name in the former; it comes from Pali, its language, which is similar to Sanskrit. The name "Tipiṭaka" is used in both languages, and is usually translated "three baskets" (see below).
The other forms of Buddhism at the present day group themselves under the heading of Mahayana, which tends to regard the Tipiṭaka as a sort of "Old Testament". Most scholars recognize the Canon as the oldest source for the Buddha's teachings, in a rough sense.
The earliest surviving detailed lists of the books included in the Canon, i.e. the canon in another sense, appear in three commentaries probably compiled in the 5th century AD. One of those commentaries gives some alternative lists, but it also includes the list appearing in the other two, and this became more or less standard: most but not all (secondary) sources before modern times that have been examined more or less follow it. However, in modern times, when we have primary sources, actual Pali Canons, rather than having to rely on secondary sources, we find more variation. The inscriptions approved by what came to be generally recognized retrospectively as the 5th Council in Mandalay in 1871 include three "additional" books, listed at the end of the Khuddakanikāya below. (These are not new works, but ancient ones, if not quite as ancient as the others.) The 6th Council, held in Rangoon from 1954 to 1956 by all five Theravada countries, approved an edition of the Canon that includes the same three. In practice, however, the council was Burmese dominated and, while the Burmese government does not (or did not as of 1968) allow any other editions to be printed, the other countries tend to pay only lip service to the council, and Cambodia, Sri Lanka and Thailand have their own editions. The Buddha Jayanti edition, which is the standard one in Sri Lanka, includes the first two of those books but not the third. A number of editions published in Thailand and the Khmer edition omit all three of those books, though in 2005 an organization sponsored by the Supreme Patriarch of Thailand published a transcript of the council edition including them. Scholars do not yet seem to have properly investigated the position regarding these books, or the concept of the canon generally, making a variety of apparently contradictory statements on the topic with only anecdotal evidence, if anything, apparently cited. The order of books also varies, with the following seemingly commonest.
- Vinayapiṭaka, on monastic discipline
- Suttavibhaṅga: commentary on Pātimokkha, a basic code of rules for monks (227 rules) and nuns (311, but many apply to monks too and are not commented on again), not itself apparently in any printed edition of the Canon except in so far as embedded here, though some secondary sources do list it separately, including the above-mentioned commentaries; the Suttavibhaṅga includes stories of the occasions for the Buddha's laying down of the rules as well as explanations and "judicial precedents"
- Khandhaka: arranged topically in 22 chapters; mainly organizational rules, with stories and explanations; the last 2 chapters give accounts of the first two councils
- Parivāra: further analysis in 19 chapters, most of which have no narrative frame; some verses at the end of some editions seem to identify the author as Dīpa or Dīpanāma
- Sutta- or Suttanta-piṭaka, discourses; divided into five nikayas (nikāya); the first four of these are in a fairly uniform style, mainly prose dialogues, sermons, etc., mostly featuring the Buddha
- Dīghanikāya: 34 long discourses
- Majjhimanikāya: 152 medium discourses
- Saṃyuttanikāya: thousands of short discourses arranged topically in 56 groups (saṃyuttas)
- Aṅguttaranikāya: thousands of short discourses arranged by numbers of items listed in them, from 1s to 11s
- Khuddakanikāya: a miscellaneous collection of books in prose and/or verse
- Khuddakapāṭha: 9 short texts in prose and/or verse
- Dhammapada: popular book of 423 verses in 26 chapters, topically
- Udāna: 80 "inspired utterances", mostly verse, with introductory narratives, featuring the Buddha
- Itivuttaka: 112 prose pieces followed by verse paraphrases or supplements; the frame formulae ascribe them to the Buddha
- Suttanipāta: basically poetry, but sometimes with prose frames that feature the Buddha
- Vimānavatthu: 85 poems giving descriptions of heavenly "mansions" and the karma leading to them; mostly in dialogue form between the deity concerned and, most often, the Buddha's disciple Moggallāna
- Petavatthu: an obverse, 51 poems describing sufferings of ghosts and the karma leading to them
- Theragāthā: 264 poems ascribed in headings and colophons to senior monks
- Therīgāthā: 73 poems similarly ascribed to senior nuns
- Jātaka: 547 poems; the poetry is often more or less unintelligible through lack of context; the Niddesa says that the Buddha taught them about the past of himself and others; some can be identified from what is said in the Canon as being about his past lives; according to the late Professor Warder, the most popular book of the Canon
- Niddesa: commentary on parts of Suttanipāta
- Paṭisambhidāmagga: 30 treatises on various topics
- Apadāna: about 600 poems, mostly in the names of monks or nuns telling how meritorious deeds in past lives led to good karmic results and eventual nirvana
- Buddhavaṃsa: verse book mainly on previous Buddhas and "our" Buddha's meritorious acts towards them in his past lives, told in the first person with an introductory narrative
- Cariyāpiṭaka: 35 poems about previous lives of the Buddha, told in the first person
- Netti(ppakaraṇa): treatise on methods of interpretation, in the name of the Buddha's disciple Kaccāna
- Peṭakopadesa: similar and overlapping
- Milindapañha: dialogue between King Menander of Bactria (c. 150 BC) and a monk called Nāgasena
- Abhidhammapiṭaka, higher or special teaching, more formal and analytical than the discourses; according to the Parivāra, taught by the Buddha himself
- Dhammasaṅgaṇi: enumeration and classification of mental and physical phenomena
- Vibhaṅga: 18 chapters analysing different topics using, among other things, ideas and material from the previous book
- Dhātukathā: analysis of interrelations among various ideas, mostly from the previous two books
- Puggalapaññatti: classifications of persons
- Kathāvatthu: over 200 debates on doctrinal points; does not identify the disputants
- Yamaka: converse pairs of questions, with answers, in 10 chapters on different topics
- Paṭṭhāna: analysis of 24 types of causal conditionality
According to tradition, the Canon is "The Word of the Buddha" (Buddhavacana), and was compiled by the First Council immediately after the Buddha's death, which it dates around 544 BC. Neither statement is intended literally, the Canon in fact including teachings by followers and accounts of events after the First Council (tradition says these latter were added by later councils). Being actually said by the historical Buddha is not a necessary requirement for counting as Buddhavacana. Nevertheless, tradition regards most of the Canon as recording the actual words of the Buddha.
Modern scholars are not prepared to accept this postition, for a variety of reasons. The most obvious is that the Canon is not unique. We have most of another early canon, that of the Sarvastivada, in Chinese translation, together with greater and lesser amounts of various canons in various languages, along with information from secondary sources about the contents of some. These canons are more or less different from the Pali and each other. Scholars reject as implausible Theravada claims that other schools, deliberately or not, added, subtracted and altered while Theravada preserved the exact original teachings.
On the basis particularly of such comparisons, scholars generally divide the Canon, with some disagreement on detail, into earlier and later halves, assuming similar material likely to be earlier than distinctive.
- earlier: the main body of the Vinaya, the first four nikayas, and some of the more or less poetic books of the Khuddakanikaya (there is some disagreement on which ones); versions of these books seem to have been in existence in all schools, and the surviving versions are more or less similar
- later: the Parivara, other books of the Khuddakanikaya, and the Abhidhamma; these books seem to be either absent altogether in other schools or else quite different from their versions
This division is only rough and ready: "early" texts may include later additions, while "late" texts may include early elements. (The statement that the Canon is the earliest source is subject to similar qualifications.)
However, there remains much disagreement on absolute, as against relative, dates, and on further stratification. Professor Gombrich, for example, holds that most of the content of the first four nikayas goes back to the Buddha himself, though not usually in exact words. He himself admits that very few scholars go so far. He also holds that the Canon was much like its present form by about 250 BC, with perhaps some Khuddakanikaya books as the only substantial later additions. Perhaps at the other extreme within the mainstream point of view is the late Professor Nakamura, who held that only parts of the Suttanipata go back to the Buddha's lifetime, with Buddhism changing substantially by the time the first four nikayas were compiled some time after c. 230 BC, and that some of the Canon was at least as late as the 2nd century AD.
A few scholars are outside this mainstream view. Professor Schopen rejects the argument that similar material is likely to be earlier than distinctive, arguing to the contrary that it is likely to result from later harmonization; he holds inscriptions to be a better source for early Buddhism. And Professor Samuel holds that a wide range of teachings were in circulation in early Buddhism, and that it was only some centuries after the Buddha's time that some schools started rejecting some; so that the Canon was created by subtraction rather than addition, largely by the 5th century commentators.
Like Hinduism, Islam and Judaism, and unlike Christianity and Mahayana Buddhism, Theravada emphasizes the scriptural language. Study and recitation are usually in Pali. Traditionally it has been regarded as the "root language", the "language of reality", the language of gods, ghosts, talking animals and wolf-children.
Scholars classify it as a form of Middle Indo-Aryan, within the family of Indo-European languages. Like other literary languages, it is not a pure vernacular dialect, though it is thought to be based mainly on one such, not independently attested, with some admixture of others, including Sanskrit. It is considered to be closely related to, but not identical with, the dialect(s) spoken by the Buddha himself.
It cannot, however, be assumed that the Canon was originally composed in Pali rather than transposed from some other dialect(s). Much could depend on definitions: how much does a dialect have to change before it counts as a different dialect?
According to a tradition generally regarded favourably by scholars, the Canon was written down from oral tradition in the last century BC in Ceylon (Sri Lanka). (Other schools seem to have written down their versions a little later.) The tradition is not unquestioned among scholars, with some suggesting the process was less straightforward. It is, however, unquestioned that the Buddha's teaching was originally transmitted orally.
The climate of Theravada countries is not conducive to the survival of manuscripts. Apart from brief quotations in inscriptions, the oldest known manuscript is a two-page fragment from the 8th or 9th century found in Nepal, other manuscripts begin in the late 15th century, and probably most are from the 18th and 19th. Thus the manuscripts available are the result of multiple copying, with the inevitable errors accumulated. This is compounded by transcription between scripts, as Pali has none of its own, each country generally using its own. This problem is exacerbated by more than one occasion on which some texts were lost in one country and had to be reimported from another. Despite this, manuscripts tend to follow different national traditions, though with some interaction. The tradition itself has been aware of these problems, and at various times groups have gathered to re-edit texts after comparing a variety of manuscripts.
The East Asian anticipation of Western printing did not spread to Southeast Asia, so Pali texts were not printed before the era of Western penetration. Attempts to print the whole Canon started in the late 19th century, and at least one edition seems to have been completed by 1910. The great bulk of the Canon has made it impractical to carry out thorough studies of printed editions, but such sample studies as have been carried out tend to suggest that those published in Theravada countries tend to follow their own national manuscript traditions to a greater or lesser extent. Such editions have been published in Myanmar, Thailand, Cambodia and Sri Lanka. Indian editions are based on Burmese ones. In the West, the Pali Text Society's edition follows scholarly methods of text criticism, though without a consistent approach; on the whole, it tends to prefer the Sinhalese tradition. Most of it was produced around a century ago, using the very limited source material available then. All editions have their faults, and modern scholars try to compare them, which is made easier by the existence of electronic versions of increasing numbers of editions. At first, these were transcribed by hand, but more recently photographic reproductions of a number of editions have appeared online.
The Canon seems not to have survived absolutely complete: the main Burmese, Sinhalese and Thai editions include notes saying that a sutta in the Aṅguttaranikāya is mentioned in the commentary but not found in the manuscripts; a mediaeval list of jātakas includes 550, not 547 as in the surviving text and commentary; and the Cariyāpiṭaka, which gives verse stories illustrating "perfections", omits some of them.
In theory, the Canon is the highest authority for the teaching. In practice, its great bulk (editions without extra material, such as translations, seem to vary from 38 to 45 volumes) means few are familiar with it as a whole. Therefore there is a tendency to specialize. The Vinaya Pitaka mentions vinaya and sutta specialists. The Milindapanha mentions specialists in each of the five nikayas. The commentaries mention abhidhamma specialists. In modern times, those wishing to be ordained as monks in Sri Lanka have had to memorize the Dhammapada. In Myanmar one can earn the title Teacher of Religion (Dhammācariya) by passing an examination where the set texts are the first volume of each pitaka. In practice, at a local level, the important texts may comprise some canonical and some non-canonical texts, varying with place. Before modern times at least, many monasteries did not have a complete set. Professor Collins suggests that the importance of the Canon lies in the idea of it, not in its actual contents.
The Canon was composed, or evolved, for the most part orally, and is adapted to that medium, and so to memorization. There are rare cases of monks who know the whole Canon by heart, and many know substantial parts. Even lay people usually know a few short passages.
Many commentaries have been written on books of the Canon, including subcommentaries, i.e. commentaries on commentaries, and further layers. The vast majority of this literature is connected with the names of just four authors. Buddhaghosa was a North Indian monk working in Ceylon in the 4th or, more likely, the 5th century. The authenticity of some of the works traditionally ascribed to him has been questioned or rejected by recent scholarship, though they may be of his "school". The works ascribed to him are largely based on earlier commentaries in Sinhalese, now lost apart from a few quotations in mediaeval Sinhalese literature. Dhammapāla was a monk of South India, for whom various dates have been suggested from the 5th century to the 10th. Most scholars, though with some major exceptions, now believe two different commentators have been conflated. Sāriputta worked in Ceylon in the 12th century; in addition to his own works there are many by his pupils. Ñāṇakitti worked in Thailand around 1500.
Commentaries comprise three main types of material: linguistic analysis, explanation of the teachings, and stories, particularly giving background context for the teachings, but also sometimes just illustrative. Ñāṇakitti focuses on the first of these, which may explain why his works have had much less currency and influence than the other writers mentioned here.
Collected editions of the primary commentaries (by Buddhaghosa, Dhammapāla and others) published in Theravada countries, like those of the Canon itself, vary somewhat in contents. The Thai edition catalogued by Dr Skilling comprises exactly one commentary on each book in the standard Thai editions of the Canon (as above). The Burmese and Sinhalese editions also include a commentary on the Netti, and a few other works (not all the same ones). When it comes to the Burmese collected edition of subcommentaries (including some by Sāriputta and others attributed to Dhammapāla), this practice of uniqueness no longer applies, with for example three subcommentaries on the Vinaya included.
Versions of the Vinaya and most of the Sutta exist in Chinese. These are inherited from other schools of ancient Indian Buddhism and differ somewhat from the Pali versions. Similarly, there is a version of the Vinaya in Tibetan. There are also numerous bits and pieces of alternative versions in various languages. All of these can sometimes shed light on the readings and interpretations of the Canon, though few Pali scholars can make direct use of texts surviving only in non-Indic languages.
Little study has gone into this. Some of the poetry has been spoken of favourably. The Milindapañha, included in some editions, was described by Rhys Davids as the greatest work of classical Indian prose, but Winternitz said this was true only of the earlier portions.
There is no generally accepted system of referencing passages in the Canon. Western scholars generally refer to the Pali Text Society editions, using volume/page/line for most books.
See the /Addendum for a more detailed account of the Canon.
- There are various typographical variants: Pali/Pāli/Pāḷi Canon/canon
- varying between editions
- The exact number varies depending on how the material is divided.
- Catukkanipāta, Valāhakavagga, number 6
- Journal of the Pali Text Society, volume XXVI, page 134
- Journal of the Pali Text Society, volume XXVII, pages 55-58