The matter of which parts of the Pāli sutta-piṭaka are older is one that has a tragic past. The first scholar to look systematically at the issue was Caroline Augusta Foley Rhys Davids. As a student, Caroline A. Foley married her (much older) Pāli teacher, Thomas Rhys Davids and together the pair [pictured left] not only became the leading experts on Pāli, but also created a lasting organisation, The Pali Text Society (founded 1881), and made the Pāli suttas available to the masses, perhaps for the first time.
Caroline and Thomas had three children but their son Arthur was the apple of her eye. The Rhys Davids family archive (Cambridge University) contains no less than 262 letters from Arthur to his mother. Arthur is famous in his own right for being a highly decorated fighter pilot in WWI (one of the original "aces"). But he was tragically killed in action in 1917.
Caroline was heartbroken and, like many others of that time, she turned to spiritualism, seeking a sense of connection with Arthur. She was a very intelligent and successful woman and she did not start attending tawdry seances or consulting fraudulent "mediums". Rather, she took up the more private practice of "automatic writing". This involves taking both sides of a written conversation, but in a detached way that allows a stream of consciousness to flow. She filled many notebooks in this way and they are still held in the archives. This turn only intensified after the death of Thomas Rhys Davids in 1922.
This change in her circumstances forced Rhys Davids to confront the Buddhist view, which till then she had accepted, that there is no soul, nothing substantial that can pass from one life to the next. This would make spiritualism practically impossible. She began to comb through the suttas and eventually concluded that the Buddha had taught an ātman doctrine after all, but covertly, and thus she rescued spiritualism from Buddhism. Much to the disgust of her colleagues, I gather. But Rhys Davids was ambitious and talented, and her next move was to try to prove that the Buddha's ātman doctrine was older than the Buddhist anātman doctrine.
Rhys Davids invented the methods which we use to form conjectures regarding the relative dating of suttas. Still, as with so much else about early Buddhism, there is no external evidence with which we can corroborate or refute these conjectures. We do know that Pāli was a somewhat artificial language built on one or more Prakrit languages. Pāli was likely never anyone's first language, but was rather a "church language" that could be a lingua franca for Prakrit speakers. These days we might call it a "conlang".
We have to keep in mind that our evidence for Pāli in the ancient world is scarcely better than our evidence for the Buddha (which is nonexistent). The very oldest extant Pāli document is a small piece of gold foil from the sixth-century. The oldest complete Pāli Canon is no older than the 15th century. People say that the Pali canon was written down in the first century, but this is conjecture based on internal references in documents that post-date the suttas by several centuries. The whole history of Buddhism is based on such conjectures with little or no supporting evidence, or based on the naive use of religious documents for historical purposes. Scholars have, until recently, simply accepted the emic accounts of Buddhist history, adopted emic terminology and time periods, and generally been far too credulous with respect to tradition.
By way of contrast we have several very old physical manuscripts of Buddhist texts from Gandhāra that can be carbon-dated to the first or second century before the Common Era. These are, in fact, the oldest Buddhist documents of any kind. Moreover, a text like the Aṣṭasāhasrikā Prajñāpāramitā is known to have been written by the end of the first century CE because, again, we have a carbon-dated manuscript on birch bark. This is about 400 years earlier than the first physical evidence of Pāli texts.
Magadhisms
A few passages in Pāli contain evidence of case endings from a different Prakrit dialect than the one that mainly forms the basis of Pāli. For example, in Pāli the nominative form of the stem buddha is buddho. E.g. buddho dhammaṃ deseti "the Buddha teaches the Dhamma". In day-to-day use, the nominative singular is considered the most basic form of the verb. Traditional dictionaries, for example, use the nominative form. European dictionaries of Indic languages tend to use the "stem" form, a notional form that is rarely (if ever) encountered in practice that has no case information. The only place they regularly crop up is as the first member of a compound word.
In a few cases in Pāli we see a nominative form like buddhe. Same word, same case, different pronunciation. Think here about the Heart Sutra dhāraṇī: gate gate... scholars have long tried to shoehorn this into a Classical Sanskrit mould, but really it's Prakrit. "Gate" is not some convoluted feminine locative of the past participle gata or whatever. Rather gate is the nominative singular of gata, i.e. it's just the basic form of a word in practical use in that dialect.
In stories about relative dates, the stray occurrence of such forms as a nominative singular in -e is seen as evidence of antiquity. The idea here is that the case marker -e is archaic and older texts are more likely to have archaic forms.
Frankly, this makes no sense to me. Dialects are, generally speaking, regional. For example, people often remark on the Tibetan spelling of vajra, i.e. badzra. The substitution of /b/ for /v/ is normal for Eastern India. Tibetans got their Sanskrit terminology from Eastern India. The state of Bihar, for example, derives from the presence of many Buddhist vihara in the past. Indeed, we sometimes see this variation in Pāli: both byāpāda "malevolence" and (Skt vyāpāda). To the best of my knowledge, this substitution (or the similarly regional initial l/r substitution) are not seen as signs of antiquity.
We know, from the distribution of Asoka inscriptions, that eastern dialects of Prakrit prefered the -e ending. And the -e ending found in Pāli is sometimes called a "Magadhism" to reflect the usage in the language used in the Asoka inscriptions around Patna, the capital of Magadha. It's possible that what we call Māgadhī was the mother language of modern Bihari.
This is not to say that dialects did not change over time. Pāli is a Middle Indic language from a pre-classical form of Old Indic (not necessarily the same one as gave rise to Classical Sanskrit). If we accept the conjecture that Pāli was written down circa first century BCE, and that this fixed the forms at that point, though later editing is clearly evidenced, then we really have to wonder how an archaic form survived for several centuries. I can tell you that when you stumble across one of these forms in practice, it can be very confusing because buddhe is something in Pāli also. It is the locative singular (the locative is mainly used to indicate the location of the action of a verb in a sentence), e.g. buddho gahe dhammaṃ deseti "the Buddha taught Dhamma in the house". This is to say, that these odd case endings stand out; one stumbles over them. How does something like that survive in an oral literature for centuries when, every time one encounters it, one is struck by the cognitive dissonance.
On the other hand, some Magadhisms are ubiquitous, such as the honorific bhante (vocative singular) or the term yebhuyyena "generally" which corresponds to Sanskrit yadbhūyasā. Following regular patterns of sound change, we expect the Pāli to be yad-bhūyena or yad-bhiyyena. Ye is Māgadhī for yad. And note also that we have some Sanskrit loanwords like brāhmaṇa for which we expect the Pāli to be bāmaṇa (see my discussion of this: A Pāli Pun).
Had we not been looking for evidence of chronology we might have concluded those texts that preserve so-called Magadhisms were preserved in a Māgadhī-speaking region where they recognised the forms. In other words, these Magadhisms in Pāli need not be evidence of change over time, they may reflect a text compiled in a different region. The presence of alternative case endings reflects contemporary regional differences in pronunciation. Not that this conjecture is any more solid than the change over time conjecture. Once again, we simply don't know. A chronological explanation is not the most obvious one to me and I think some kind of geographical explanation is probably better.
Systematic
Another argument for the antiquity of some texts is that they are "less systematic" (with reference to the standardised Buddhism of modern Theravāda) and thus older. This is a form of the teleological fallacy. The idea here is that ideas become more sophisticated and more organised over time. The presumption here is also that the Pāli texts are otherwise homogeneous and forms a static backdrop against which change can be discerned. I would argue that Theravāda Buddhism, as we meet it in the twenty-first century, is a simplified, less sophisticated form of the pluralistic Buddhism we find in early Buddhist texts.
Some of the texts use very different terminology. It is true that one of these is in the Suttanipāta (Sn 862-877) which experts say makes it old, based on the methods we are exploring now. But the sequence in DN 21 is just as odd. Moreover, it partially reverses the order of causality found in, say, MN 18. A more sophisticated variant of the standard 12 nidānas is also found in Suttanipāta (Sn 722-765) which in the standard view makes is later than most other nidāna texts.
The idea of using structural features like how "systematic" a doctrine is to determine relative age is starting to look quite doubtful. The logic of it does not account for which variations of the nidānas that we find here. Again, the standardisation on 12 nidānas, ignoring variants, is considerably less sophisticated than we see in Pāli. In order to interpret such differences in terms of chronology we have to presume that differences are caused by passing time (that is to say time passing is what causes variations to arise). But again, we could have chosen to see these as contemporary sectarian differences, for example. It's only when we ignore the obvious sectarianism in Pāli that we see anything like an "underlying unity".
Under this heading we may also discuss the fact that no one claims that the whole Suttanipāta is old. Only parts of it. And yet there is no evidence that the Suttanipāta circulated in fragments. It is true that a few parts of Suttanipāta turn up in other places. The Sela Sutta (Sn 3.7) is identical to MN 92, while the Vāseṭṭha Sutta (Sn 3.9) also occurs at MN 98. But this just tells us that the compiler of the Suttanipāta had access to the same sources as the compiler of the Majjhimanikāya.
Metre.
I know little about meter, though I have dabbled in analysing Pāli meter from time to time. The argument in this case is that the use of certain metres in, say, Suttanipāta reflects antiquity. These are referred to by Roy Norman as "old metres". I've never been clear how anything can be considered a priori "old". We have no evidence of Buddhist literature before the written texts. We cannot judge the antiquity of a metre in a Buddhist text from the types of metres used in non-Buddhist texts. And I cannot think how else this chronological distinction could be made.
Metred verses are common enough in Pāli and some texts show a preference for one metrical scheme over another. But in a Buddhist context, what constitutes "old"? Old in comparison to what? We have nothing to base a chronology of metre on in a Buddhist context. We may be able to say that non-Buddhist texts show an evolution of the use of metres, but showing that the same evolution happened in Buddhist texts is not possible.
Self-referential.
A number of suttas quote other suttas, sometimes verbatim, sometimes by name. An example I have encountered is the Channa Sutta (SN 22.90) in which Ānanda recalls hearing the Kaccānagotta Sutta (SN 12:15) being preached. The Channa Sutta repeats the Kaccānagotta Sutta in its entirety. And this, so the argument goes, proves that the Channa Sutta presumes the prior existence of the Kaccānagotta Sutta. But does it? What if the Channa Sutta is the original context for this text and the Kaccānagotta Sutta is simply a cut-down version of the story?
This alternative possibility is given credence when we study the Pāli version alongside the Āgama version which exists in a Chinese translation (from a Prakrit other than Pāli) and a Sanskrit translation that more or less corresponds to the Chinese. In my (to date) unpublished study of the three versions side by side I note:
[The Pāli] as well has having an abbreviated opening, has no end. It just finishes abruptly, and this reinforces other hints that it is a somewhat fragmented memory of the text.
It is clear that the Pāli record of the Discourse to Kātyāyana is not the ur-text. It's a fragment, with a slightly different selection and arrangement of sentences than the Chinese or Sanskrit versions.
Interpolations.
One of the most striking examples of interpolation I know of comes in the Mahāparinibbāṇa Sutta (DN 16; DN II.141). In the middle of a discussion between Ānanda and the Buddha about the Buddha's funeral arrangements, we suddenly find Ānanda asking "How should we behave towards women?" (Kathaṃ mayaṃ, bhante, mātugāme paṭipajjāma). The word for "women" here is a colloquialism made from mātā "mother" (in the genitive case mātu) and gāma "village". The Buddha tells Ānanda to ignore them (adassanaṃ; literally "don't look"). After a few more lines of this misogyny, we go back to discussing the Buddha's funeral arrangements. The change of subject is quite disorientating.
In this one case, I agree that there is an obvious reason to consider the passage on how to behave towards women has been inserted into the text at some point after the text was initially composed. It was done so badly that we cannot help but be struck by the incompetence of the editor. Still, it was done early enough to be considered canonical. On the other hand, this interpolation does not occur in any other surviving version of the Mahāparinibbāna Sutta. So we can conjecture that it was a Theravādin monk who did the interpolating. The negative attitude towards women is typical of Theravāda monasticism.
On the other hand, the sharing of passages between texts is so common as to constitute a major feature of Pāli texts. Suttas have a kind of modular structure with a framework of common tropes and expressions (aka pericopes). Shared stock passages are the norm.
Compare the Pāli, Chinese, and Sanskrit texts of the Kaccānagotta Sutta, for example. They are all closely related, but in some cases whole phrases are present or missing in one. And this in a very short text. Quotations in Mahāyāna texts make it seem that the core of this text is a statement against applying the duality of existence/nonexistence (astitā/nāstitā) to the world (loka). The framing details of this important statement vary considerable. Notably the nidāna is different in all three, and the name of the main protagonist also takes three different, though closely related, forms, i.e. P. Kaccāna, Skt. Sandhākātyāyana, C. Shāntuó Jiāzhānyán 𨅖陀迦旃延. Moreover, Sandhākātyāyana is almost certainly a mistake for Saddhā Kātyāyana, which means something like "Faithful youth of the Kātya clan."
Conclusion
Having briefly surveyed the main kinds of evidence that are used to try to establish a relative chronology within the Pāli suttas, I find that, except in the case of obvious interpolations, I can always think of a plausible alternative reading that does not result in any chronological speculation. That which is presented as evidence of chronology could just as well be evidence of regionalism or sectarianism.
The idea that we can discern any systematic chronology within Pāli suttas seems quite fanciful. It certainly appeals to Buddhist theologians, but it's a house of cards. The foundations are essentially religious beliefs that are not open to discussion. The most striking of these is the religious conviction that the Buddha was a real person. This is axiomatic, for example, for bhikkhus Sujato and Brahmali who have put a great deal of effort into arguments for the "authenticity" of the Pāli canon. But their definition of authenticity is itself incoherent. In their accounts of authenticity they assume that both the Buddha and Ānanda were real people who were just as described in the literature. There are no external criteria because there is no external evidence. Thus the whole rests on religious commitments rather than historical facts or events.
Moreover, the kind of relative chronology that is produced by these speculations offers little in the way of explanatory power. It is self-contained with very few exceptions, the most notable being that the cities in which the stories are largely set are real cities, although none of the characters in the stories can be considered historical characters.
As someone who likes to state clear conclusions, I sympathise with the historians who scrabble around trying to put things in chronological order. But the very aim of the project—to produce a chronology—determines what kind of outcome we get, i.e. a chronology. Other explanations for the same facts are never even considered as far as I can see. And despite all the efforts that go into this project we still cannot explain anything of importance using this artificially constructed chronology. This is partly because the relative chronology is not anchored to history at any point. Again, the lack of external evidence of any kind is telling.
As far as I can see the only "real" thing in the Pāli suttas are the cities. The stories are set in cities that we know from archaeology. I've walked among the ruins of Sāvatthī, for example. One can see it on Google Maps. The evidence that comes from analysing religious texts is something else again. And this may be part of the problem. Historians of Buddhism seem to forget that Buddhist texts are religious texts. And I'm not the only one pointing out the problems with this.
~~oOo~~
