Are Chinese Agamas less reliable than Pali Nikayas?

I don’t know about the evidences you asked for, but I prefer the mention of first 3 angas in MN 122 is likely an abbreviation of 9 angas. The missing of remaining items is not neccesary a proof of later addition of remaining items.

Take another case. In Ariyapariyesana Sutta (MN 26) there is no mention of Four Noble Truths as first teaching of the Buddha and the parallel in MA 204 only mentioned the middle way. Many scholars think this mean Four Noble Truths are not original teaching of the Buddha, but a later development of doctrine.

But it seems to me it more likely that the first teaching of the Buddha is not included in Ariyapariyesana Sutta and it’s parallel because the text focus on how Gotama strive for enlighment so there no need to mention to the first teaching completely.

Besides that, there is an Ekottarika Agama version of the first discource (EA 19.2) which mentioned only the middle way, but another version of the first discource from the same collection (EA 24.5) mentioned Four Noble Truths. It is not that the shorter EA 19.2 is an earlier version and EA 24.5 details of Four Noble Truths is addition. It is more likely that the EA 19.2 is just an extract from complete version of the first discourse known by all early Buddhist schools.

So I think it’s the similar case of abbreaviated list of angas in MN 122 or AN 5.194.

1 Like

Also, Yinshun cites a passage in the Mahāvibhāṣā (T1545.659c-660a) that defines the vyākaraṇa aṅga in very specific terms: 1) when disciples ask a question, and the Tathāgata explains; 2) when the Tathāgata asks a question, and the disciple explains it; 3) a disciple asks a question, and another disciple explains it; and 4) conjured gods and the like ask questions, and it’s explained.

This brings in other scenarios for these sutras than the narrow one (a disciple asking a question) that I used as an example.

1 Like

Something about lists like 9 and 12 angas is that they do appear to be expanded when I compare them from different sources. The list after the first three items is kind of jumbled. The items are all the same, but the orders aren’t stable. So, it would seem like it wasn’t an original list that every sect preserved the same way (like the eightfold path or five powers, etc.). Just the fact that the Theravada seems like an outlier in having only 9 instead of 12 suggests the list grew in stages.


Just to note here for the record, I agree with Analayo that that particular data point is weak, although it is only one of several indications. Even taking other details into account, however, it’s just a hypothesis. I still think it holds up pretty well, though. It’s a simple hypothesis with a strong explanatory power.

I always wanted to revisit this point with a proper statistical analysis, which I suspect might be the only way to really discern the underlying Samyutta structure.


Ah, yes. He does cite that passage in CSA’s introduction. The impression I’m getting, however, is that it was in the later Sarvâstivāda tradition that SA rose to this prominent position. Earlier Indian accounts like in the DZDL and EA claim that EA is the first collection of sutras recited. Analayo is not wrong to bring up that point, but it only proves that Buddhists weren’t agreed on what exactly happened in the early days of the canon.

The list of the first 3 angas mentioned in both MN 122 and its corresponding MA 191. So, it is very unlikely they are “the similar case of abbreviated list of [all 9 or 12] angas in MN 122 [and MA191]”.

Yes, that’s the main argument proposed by Yinshun and supported by Sujato, but as shown by Analayo in long notes posted above, this is not conclusive because there are occurrences of 3 (or 4) angas in other texts because abbreviation, etc.

I think there are many possibility of this, either this is the original or just abbreviated or textual/transmission loss, but we cannot sure about it because all of these is lack of solid evidence and just a hypothesis. Personally, I prefer Analayo’s explanation.

Thank you :anjal:

Though rare, āgama (in the singular only) is actually evidenced in Pali sources long before the Dīpavaṃsa. You’ll find it, for example, in the Kusinarasutta (AN 10:44) and its parallel in the Vinaya’s Cūḷavagga.

Furthermore, a mendicant who wants to accuse another should check this: ‘Am I very learned, remembering and keeping what I’ve learned? These teachings are good in the beginning, good in the middle, and good in the end, meaningful and well-phrased, describing a spiritual practice that’s entirely full and pure. Am I very learned in such teachings, remembering them, reinforcing them by recitation, mentally scrutinizing them, and comprehending them theoretically? Is this thing found in me or not?’ If it’s not, there will be people who say: ‘Come on, venerable, memorize the scriptures first (‘iṅgha tāva āyasmā āgamaṃ pariyāpuṇassū’ti).’


Dear Bhante,

I’m eagerly to see you revisiting this and updating your GIST (Grand Integrated Sutta Theory) in A History of Mindfulness. Your GIST didn’t much compare and analyse structure and content of AN dan EA because lack of data, but now we have paper who have done it here:



These passages would be more convincing if there were more of them. Having one or two aberrant lists here and there doesn’t give us enough evidence to come to a conclusion, it’s too easy to argue it’s a textual corruption because corruptions are not that uncommon in these texts.

Ultimately, I think these debates over individual passages are a distraction compared to analyzing the canons themselves. For example, as I mentioned earlier, the geya division of SA/SN is the most obvious and easiest to demonstrate. However, a geya sutra has a specific definition in sources like the Mahāvibhāṣā: It’s a sutra with a concluding verse (T1545.659c21).

When we look at the Pali Sagatha Vagga of SN, we see close parallels to SA’s Eight Assemblies that contain the same verse material. But the Pali suttas aren’t geya sutras according to the definition in the Mahāvibhāṣā. In fact, they don’t have the basic format of sutras according to the Buddhist world in ancient times. Only a select few have the requisite conclusion of a sutra, and many lack introductions beyond a simple place name. In fact, there are “suttas” in the Devata Samyutta that are just verses like the Dhammapada (e.g. SN 1.14). Other suttas in this division have conclusions in prose rather than verse, which also makes calling them geya more difficult (e.g. SN 2.20).

So, I would say that the Theravada was at some point in the past not nearly so orthodox and conservative as they present themselves today. Someone had the temerity to completely redact their canon to remove repetitive material. To think their canon never had the same basic format as other canons is just not likely to me. There’s nothing wrong with this redaction in my opinion, as I have a progressive mindset myself, but I point it out because it’s significant when we turn to the parallels in SA. The stripping of basic sutra formatting from SN suttas makes tracing their older forms more difficult. It’s no wonder Pali scholars scratch their heads regarding Yinshun’s ideas.

Turning to SA as a contrast, the sutras in the parallel samyuktas to SN 1-11 are in the format of geya sutras. There’s a minimal format that sutras retain if they don’t have a unique story backing them. Here’s an example using SA 25.1:

1. Thus I have heard. One time, the Buddha was staying at Anāthapiṇḍada’s Park in Jeta’s Grove of Śrāvastī.

2. There was then a devata of a marvelous appearance that visited the Buddha late at night, bowed at the Buddha’s feet, and withdrew to sit to one side. The radiance of its body illuminated the entirety of Anāthapiṇḍada’s Park and Jeta’s Grove.

3. That devata then spoke to the Buddha in verse:

"Not dwelling in the Nanda Grove,
You’ll never obtain [its] delight!
In the palace of the Thirty-Three Gods,
You’ll get the title ‘Lord of the Gods.’"

4. The Bhagavān then replied in verse:

"Foolish child, what do you know
About what the Arhat teaches?
All conditioned things are impermanent:
That’s the law of arising and perishing.
What arises will perish again;
Happiness is their extinguishment."

5. That devata spoke again in verse:

“Long have I watched this priest
Who has won parinirvāṇa.
He has gone beyond all fear,
Transcending forever worldly affection.”

6. When that devata heard what the Buddha taught, he rejoiced and was delighted. He bowed to the Buddha’s feet and promptly disappeared.

The part of this generic format that’s unique between sutras is the bolded verses. The rest is duplicated repeatedly so that they have the geya format.

So, now, I have the opposite problem: These texts have been consciously formatted in a standardized way. It suggests that the verses have either been created outside of the canon or extracted from other texts, but either way it’s the opposite problem of the Theravada canon: Artificial sutra formatting. One thing it makes me wonder is: How old is the Dharmapada and did its verses actually get brought in and converted into these sutras? One thing is for sure though: In the Sarvastivada canon, it was required that these texts be geya sutras.

All of this makes us scratch our heads. It’ll require a very close study and comparison of the two canons to draw out what we may have started with before the sectarian canons took the form that were given by their maintainers.


In Dhammajoti’s introduction to his Chinese Version of Dharmapada, he explored Yinshun hypothesis on how Dhammapada dan others Khuddhaka/Ksudraka texts formed. Based on a passage from Mahasanghika Vinaya which defined Ksudraka Pitaka as a collection of verses (from Sravakas and Pratyekabuddhas), Yinshun concluded that Khuddhaka/Ksudraka texts originated from verses of unknown authorship but gained popularity in early Buddhist community outside the four Nikayas/Agamas. These are considered as gatha and udana angas.

The references to these texts (dhammapadani, parayanavagga, and atthakavagga) in four main collections indicated that the verses material are old. But these texts were canonize in later time as fifth collection (as in Theravada canon) or another pitaka (as in other sects called it Ksudraka Pitaka) because originally conservative Sangha reluctantly included them in the canon. The reason is apart from the unknown autorship, these text were transmitted outside Sangha and outside main early Buddhist region (madhyadesa). As time went on, the collection is growing with later material addition (Cariyapitaka, Patisambhidhamagga, Petakoppadesa, etc) forming what we know now as Khuddhaka Nikaya.

Just FYI :anjal:


If the list of the first 3 angas is mentioned only in MN 122, but not in its corresponding MA 191, then, one can easily argue it is a textual corruption, such as a textual/transmission loss, or an abbreviated list of 9 or 12 angas in MN 122 = MA191. However, the list of the first 3 angas is found in both the Pali MN 122 and its corresponding Chinese MA 191. One cannot just simply ignore or downplay the textual fact!

Thanks for this information indeed. But I am not sure why the Pali tradition prefers to use Nikaya instead of Agama.

I don’t know. Perhaps it was out of a wish to put a distance between themselves and those non-Buddhists who also used āgama; e.g., the Jains who used it (in the plural) as the name for their canonical texts and the brahmins who used it (in the singular) interchangeably with śruti. By contrast, no non-Buddhists were using nikāya in this sense.

Or perhaps they wished for a greater degree of specificity than was offered by āgama, which in its earliest usage (in the singular) was rather vague and might, for all we know, have included the vinaya too.

The Theravādins, by the way, were not the only school that used nikāya. Though their texts are no longer extant, there’s epigraphic evidence that the Aparaśailas/Aparaseliyas (a southern Indian Mahāsaṅghika school) used it too.


This topic is near and dear to my heart. To have a text for myself doesn’t mean a whole lot, if others cannot also have it too. CC0 is the best way to go to get texts into the public domain.

Oh, the bad old days… When I translated a text from the agamas ~10 years ago, some of the first comments were from someone criticizing it for being different from the Pali text. That type of response made me hesitant to translate more from the agamas. It seemed like people might not bother to read them at all, or might view them as inherently deficient. SuttaCentral in particular has helped to normalize the agamas.


I hope you don’t bother what others said about the Agamas. We need more translation from Chinese Agamas for understanding more about early Buddhism teaching and only a few person like you and @cdpatton can do it. Perhaps, you can collaborate with Mr. Patton to create a better translation of Chinese Agamas into English :grin:


Thanks, a lot has changed in the last decade. When starting out, my presumption was that the main people who might be interested in the agamas would be Mahayanists who were interested in reading early texts, and especially meditation instructions (e.g. anapana). But over time, it seems that there is also quite a bit of interest now in the EBT’s in general, inclusive of both the Pali and Chinese texts.


Another essential issue/interest currently in the studies of EBTs is about ven. Yinshun’s works (1971, 1983) on the formation of EBTs. This issue is particularly presented by Choong Mun-keat, e.g. his recent publication (2020) “Ācāriya Buddhaghosa and Master Yinshun 印順 on the Three-aṅga Structure of Early Buddhist Texts”, pp. 886, 903, note 24, 911. So, it will be very useful if the books (1971, 1983) by ven. Yinshun on the formation of EBTs are reviewed, and also completely translated into English.


It would take someone really erudite in Buddhist texts to translate Yinshun’s work - the challenge is that the English reading audience would probably want to have the accurate Sanskrit/English terms of what Yinshun referenced in Chinese. And most of those texts do not have readily available English translations.

Other than the Nikaya structure, Master Yinshun also defined Dhammapada as Udana itself in terms of Navanga Sasana. Allow me to restate what the navana sasana are according to master YS:

  1. Sutta (Nidana, khanda, salayatana, magga, bojjhanga, bala, indriya, iddhipada, satipatthana, anapana, sammappadhana, sotopatti)

  2. Geyya

  3. Vyakarana
    => The first three are collectively known as the “Samyutta/Samyukta”

  4. Gatha (Muni gatha, Rhino sutta, Attakavagga, Parayanavagga as collected in Sutta Nipata)

  5. Udana (Dhammapada, not the “udana” in KN)

  6. Itivuttaka

  7. Jataka

  8. Abbhutadhamma

  9. Vedalla