A Note on the Ekottarika Āgama: The Motif "Therefore ... Thus, you should train yourself."

For the past five or six months now, I’ve been translating the Ekottarika Āgama from cover to cover, and I quickly noticed something. More than half of its sutras feature a standard conclusion, which is usually fairly simple. The Buddha ends a discourse by saying, “Therefore, [insert basic moral of the discourse]. Thus, you should train yourself.” Occasionally, the moral that follows “therefore” can become a paragraph, but usually it’s a single sentence.

In some chapters of EĀ, all the sutras share this common format, or there are only one or two that do not. And the sutras that don’t have this conclusion often look out of place in other ways. One example of this is Chapter 10. Only EĀ 10.4 and EĀ 10.6 lack this motif. EĀ 10.4 is a sutra about Anāthapiṇḍika and EĀ 10.6 appears to come from an Itivuttaka collection. The other eight sutras have the conclusion, though EĀ 10.3 has had gāthās appended afterward.

I recognized this motif from Pali suttas early on. After translating several of the chapters in the Book of Ones of EĀ, though, I realized that there’s something significant about this. So, I sat down and documented all the sutras in EĀ and in the Pali Nikāyas that share this motif that I could find. I only quickly scanned these collections visually, so I’m sure I missed some cases of it, but the results are still interesting.

In the Ekottarika Āgama, 324 out of 450 sutras (excluding the introduction and the chapters on disciples) feature this motif. Even sutras that are ostensibly geyas in format (with verse conclusions) often have this type of prose conclusion.

In Pali, there aren’t nearly as many suttas with this type of ending, but they aren’t insignificant, either.

In SN, the most significant cluster of suttas with the motif are in SN 17 and SN 20. All but three of the 43 suttas in SN 17 include it, and all 12 suttas on SN 20 have this format. Outside of SN 17 and 20, it’s fairly rare: In occurs in SN 12.22, SN 41.10, and SN 47.20.

Naturally, I was curious about AN including this type of sutta format, since it should have direct parallels to EĀ. It does have around 35 suttas with this type of conclusion: AN 1.77, AN 1.79, AN 1.347, AN 2.1, AN 2.2, AN 2.5, AN 3.1-8, AN 3.15, AN 3.81-83, AN 3.92, AN 3.99, AN 4.73, AN 4.157, AN 5.1-2, AN 5.12-13, AN 5.56, AN 5.180, AN 6.19, AN 6.54, AN 7.73, AN 8.7, AN 8.73, AN 10.51, AN 10.59. In some cases, the format has been changed a bit, or it occurs in the body of the sutta instead at the end.

In MN, the motif appears to become a refrain in discourses like MN 21 and MN 39. Whether it’s directly related to this concluding motif or not is unclear, but it may have at least provided inspiration for a new kind of sutta format.

A difficulty that we face with this type of parallel in Pali suttas is that, for better or worse, Theravada redactors have stripped most individual intros and outros from the short suttas collections. This might have included stripping away some of these conclusions if they seemed extraneous. If that were the case, there might have been more parallels at one time.

Interestingly, this motif sometimes appears to indicate later redactions, where a commentary appears to be attached to a smaller discourse. For example, if we assume this motif marked the end of a discourse, it calls into question how SN 22.1 formed. Is the bulk of it a lengthy commentary that interprets the Buddha’s advice in the frame of the five aggregates? That Sāriputta provides the commentary fits a pattern that we see in other contexts.

AN 10.59 is another sutta that appears redacted when looked at in this light. It would appear that the conclusion of a sutta has been detached and had a list inserted into it. Then, that passage became the body of a sutta with a commentary attached as a conclusion.

I haven’t looked at the other Āgamas systematically yet, but when I’ve checked parallels in SĀ, I’ve noticed that often this concluding motif is missing from its parallels. An example of this can be seen by comparing SA 1084 with SĀ2 23. In this case, a short sutra appears to have had some gāthās involving Māra appended to it. The motif of concluding with “Thus, monks, you should train yourselves” in SĀ2 is missing from the SĀ version.

Overall, given that EĀ is much less developed than AN, it would seem that at the very least there was a recitation tradition in early Buddhism that made of habit of concluding sutras in this way. Over the centuries, these sutras have been scattered haphazardly or diluted by sutras with other formats. In the course of scanning the Pali suttas for these types of conclusions, it was easy to notice other standardized formats in both SN and AN. We often focus on philosophical keywords when analyzing patterns in early Buddhist texts because of the academic focus on doctrinal controversies, but I think this example highlights another way of identifying the strata of early Buddhist texts: The standardized formats used to frame the discourses.


It would be interesting to see how the philosophical keyword distributions map to your findings. Did you do any comparisons along those lines?

In the discourses (such as SN/SA) the motif of both introductory and concluding words is likely to have been added later.

Only a little bit while looking through the texts, though that’s an obvious next step to this kind of study. I did notice a strong stylistic format to the suttas of SN 22 as scanned through them, which emphasize knowing and seeing == liberation quite a bit compared to other suttas. I’ve always been struck by the absence of the idea that meditative accomplishment leads to liberation in those suttas. They present it as though one only needs to realize the nature of the five aggregates to escape rebirth. So, they would seem to be the product of a particular line of thought. It was pretty clear that they stand out as their own group of texts just based on elements like the statement of liberation.

I don’t think this makes sense. If the sutra collections were being memorized and recited, the intros and outros are needed to signal the beginning and end of a sutra for the audience that’s listening. When people are reading texts in writing, those intros and outros aren’t needed anymore. So, my conclusion is the opposite. It’s a late redaction from a time when oral recitation wasn’t required anymore. That the collections were still being memorized and recited in China in 400 AD tells me that the Pali redaction happened at a later time than that, if we assume the adoption of reading and writing happened at the same pace in South India as it did elsewhere in the Buddhist world.


The main issue is the extant collections were not entirely being recited. The texts were also artificial compilations.

The meaning in the suttas (not as an ending) is confined to beginner stages. This is shown in the Anapanasati sutta (Majjhima Nikaya 118) where it is used extensively, that text particularly covering such stages. In Majjhima Nikaya 39 it is shown to be an early stage of development separate from the ‘contemplation’ stage. This is connected with Majjhima Nikaya 97 in using ‘still more to be done .’ Samyutta Nikaya 20 includes three suttas with references to metta, further consolidating the level with the Majjhima Nikaya 97 theme.

"“But why, Sariputta — when there was still more to be done, having established Dhanañjanin the brahman in the inferior Brahma world — did you get up from your seat and leave?”