Sorting out the Saṅgāvara Sutta (MN 100)

The Saṅgāvara Sutta (MN 100) is one of the trickier texts in the canon. The bulk of it repeats the account of the Buddha’s pracyice before awakening, but the narrative framework, while fairly short, is full of challenges both in the language and interpretation.

Indeed, such are the textual difficulties that K.R. Norman wrote a whole essay (The Buddha’s View of devas, also in Collected Papers II) arguing that it suffers from major textual corruption, for which he proposed a reconstructed version. But Anālayo showed that the reconstructions are not supported by the Chinese parallels. Much ink has been spilled on these matters; Piya Tan discusses details. There have been previous discussions on this forum, to which I hope to contribute.

Ven Bodhi says:

The sequence of ideas in this passage is difficult to follow and it is likely that the text is corrupt. K.R. Norman has proposed a reconstruction of this portion of the dialogue, but it is hard to follow him in details.

I’ll spare you another go through all this mass of argumentation. I believe these illustrious former interpreters have missed the mark, and here I’ll explain my take.

who is Saṅgāvara

In MN 100 Saṅgāvara is presented as a mānava, a brahmanical student, of sixteen years of age, living at Caṇḍalakappa (otherwise unknown). He is one of those precocious teens who has learned the whole scripture and thinks he knows better than everyone else. He snaps at the brahmin lady Dhanañjānī when she expresses homage to the BUddha. The relationship between the two is not spelled out in the text, but I think she is the wife of his brahmin teacher.

Now, a “brahmin” Saṅgārava asks about the near shore and far shore (AN 10.117, AN 10.169), and about how hymns are remembered (SN 46.55, AN 5.193). It would appear to be the same Saṅgārava of Sāvatthī who practiced purification in water (AN 7.21) and who asked about sacrifice (AN 3.60).

Apart from the “far shore” discourses, in all of these he was said to go for refuge at the end, as does the Saṅgārava of the current discourse. This is historically unlikely, but not unusual in the suttas. In any case it indicates we should be cautious about drawing historical inferences here.

Whether that Saṅgārava, who lived at Sāvatthī and is always identified as “brahmin”, is the same person as the teenage “student” (māṇava) living at Caṇḍalakappa is unclear.

the difficult passage

Here’s my current translation. Saṅgārava, at the end of the long passage on the Buddha’s awakening, abruptly changes topic.

When he had spoken, Saṅgārava said to the Buddha, “The striving of Mister Gotama was indeed assiduous and that of a true person, since he is a perfected one, a fully awakened Buddha. But Mister Gotama, do gods survive?”

“I’ve understood about gods in terms of causes.”

“But Mister Gotama, when asked ‘Do gods survive?’ why did you say that you have understood about gods in terms of causes? That being so, is it not hollow and false?”

“When asked ‘Do gods survive’, whether you reply ‘Gods survive’ or ‘I’ve understood in terms of causes’ a sensible person would come to the categorical conclusion in this matter that gods survive.”

“But why didn’t you say that in the first place?”

“It is agreed by the eminent in the world that gods survive.”

Now, clearly the passage is a bit choppy and confusing. The key point, to me, is that the confusion is part of the text itself. Those who were speaking are well aware that there has been a confusion or difficulty. This suggests that what has been taken as corruption is simply the text representing naturalistic dialogue.

language details

There are few linguistic difficulties to sort out.

“indeed assiduous”

This represents the hapax aṭṭhitavataṁ. This, I believe, should be resolved with aṭṭhita as ā+ṭhita, “assiduous”; see Ja 242:2.1 in the same sense, and compare Sanskrit āsthita. Most dictionaries and translators read as if it were a+ṭhita, which I think is a mistake.

For the suffix, I follow the variant vata rather than vataṁ, as exclamations with repeated vata are idiomatic.

“gods survive”

The phrase atthi devā is established at MN 90:13.2, where it is explained as whether gods “survive”, i.e. whether they are reborn. The implicit contrast is with gods who exist eternally. There are a number of similar passages. I’ve written about this detail a number of times, most recently here, and here is the relevant portion:

Compare the following passage from MN 90 (see also MN 100). It uses the exact same phrasing, but instead of asking “does the self exist?” (atthi attā) it asks “do gods exist” (atthi devā). King Pasenadi asks the Buddha:

“But sir, do gods survive (/exist)?”
“But what exactly are you asking?”
“Whether those gods come back to this state of existence or not.”
“Those gods who are subject to affliction come back to this state of existence, but those free of affliction do not come back.”

I’ve translated atthi in such passages as “survive” rather than “exist”, and here you can see why. Pasenadi is not asking about whether gods exist in an abstract sense, but about whether they are reborn. But the phrasing is unclear, so even the Buddha requires clarification. This shows that the verb atthi, while often having the simple sense of “exist” can also carry a philosophically pregnant sense of “continue to exist”, “exist forever”, or more simply, “survive”.

We find a similar question posed to Yama, god of death, by the student Naciketa in Kaṭha Upaniṣad 1.20. Naciketa wonders what happens to a man after he dies, as “some say he survives, while others say he does not survive” (astīty eke nāyam astīti caike). Yama initially refuses to answer such a difficult question. Pressed, he reveals that those who insist there is only this world simply fall under his sway (2.6), but one who knows the subtle true Self realizes the ultimate goal.

This lets us make better sense of the conversation with Vacchagotta. He is asking whether a self survives after death (eternalism) or whether it does not survive after death (Annihilationism). Both cases are of the “do you beat your wife often?” variety, as they assume a self. It also shows that the problematic aspect of Vacchagotta’s question is not the notion of a “self” but the ambiguous verb atthi.

What is really striking here is the Buddha’s response in the three atthi passages under consideration:

  • SN 44.10: the Buddha stays silent, refusing to answer.
  • MN 90: the Buddha immediately asks for clarification of the question.
  • MN 100: the Buddha gives an oblique answer that confuses the questioner.

Clearly, atthi is a tricky verb. The exact implications were a niggling source of confusion in those days just as today. Given that it is, after all, just a verb “to be”, this only makes sense if it has multiple meanings, in particular, a specialized sense in such philosophical contexts.

Thus it seems that Saṅgārava is asking a question that can mean either, “Are gods real, do they exist?”, or “Do gods exist in an eternal state?” The Buddha’s response attempts to satisfy both these senses.

“about gods”

Multiple commentators have proceeded as if adhidevā meant “higher gods”. But this meaning appears to be entirely absent from the Pali. Both here and at AN 8.64:10.1 the sense is, rather, “about the gods”, drawing on a well-known distinction between things that are “about the gods”, those “about beings” (adhibhūta), and those “about the body/self” (adhyātmam, Bṛhadāraṇyaka Upaniṣad 3.7.14–15).

That this sense applies here is established at MN 90:17.8.

“in terms of causes”

Ṭhānaso sometimes means “immediately”, but here it means “in terms of cause” or “by way of cause”. It typically appears in the context of kamma and rebirth, followed by a verb of knowing (eg. MN 12:11.2, ṭhānaso hetuso vipākaṁ yathābhūtaṁ pajānāti, “understands results [of kamma] in terms of grounds and causes”).

The Buddha’s point is that the survival of gods is not a metaphysical absolute but depends on their kamma. They exist in a relative and conditioned sense, not in an absolute and eternal sense.

“by the eminent”

Uccena is rendered “widely by Ven Bodhi, which I earlier followed. But the sense of ucca is “high, lofty”, and here it is, I believe, a noun referring to those who are “eminent”. This, I think, is meant to include Saṅgārava’s teachers, in contrast with the less reputable annihilationists.

I think the point the Buddha is making here is that they share common ground, since respectable teachers agree that gods exist. So in his answer, the Buddha was taking this for granted, and assuming that Saṅgārava wanted to know about the manner in which gods exist.

the overall coherence

With this linguistic details sorted out, it seems to me that the passage makes clear sense.

Saṅgārava asks about the “existence” of the gods, and the Buddha replies in a way that implicitly assumes their existence, but mainly focuses on asserting that the manner of their existence was conditioned, dependent on kamma.

Saṅgārava misses the point and accuses the Buddha of lying. This passage would normally include a reference to the Buddha’s “speech” being false and hollow, and this is indeed supplied by some translators. But it is absent from the text, perhaps due to corruption, or, as I think, to represent Saṅgārava’s confused and reactive speech patterns.

The underlying assumption of those who argue the passage is corrupt is, I think, that the suttas should be coherent. Now, obviously we can expect that the Buddha’s speech is presented as coherent. But natural conversation is very often not. Normally the suttas, it is true, present a stylized and smoothed out form of dialogue. But there are numerous occasions where they are more jagged, more clumsy, more naturalistic, and I think this is a good example.

Thus the incoherence stems partly from the trickiness of the subject matter and terminology, and partly from Saṅgārava’s personality.

Right from the start, Saṅgārava exhibits a sharp reaction to Dhanañjānī’s behavior, evidencing a prickly and somewhat fundamentalist mindset. Such attitudes make it hard to have a decent conversation. He doesn’t follow the Buddha’s logic, and feeling confused and trapped, lashes out rather than asking for clarification.

Notably, the Saṅgārava of AN 3.60 is evasive and a little rude, reminiscent of the prickly and somewhat chaotic personality on display in the current sutta. But as noted above, the connection between Saṅgāravas of different suttas is unclear.

Given that Saṅgārava’s conversion here seems abrupt and unconvincing, I suspect that the ending was tacked on, and he was not converted until later.


I think the bulk of the problems of the text can be explained by a correct understanding of the terms involved, as well as naturalistic dialogue depciting the prickly and chaotic character of Saṅgārava. This doesn’t rule out the possibility of textual corruption, of course. But as far as I can see, the overall meaning of the passage is clear enough.


Is there an English translation for this part of the Chinese Canon? What message does the Chinese Canon convey?


If you go to the link for MN 100 on SuttaCentral, and click the “Parallels” button, you’ll see a list of parallels. In this case, there is no full parallel in Chinese. There are fragments of a full parallel in Sanskrit, and some parallels to parts of the discourse in Chinese, but they don’t cover the points raised in the text in much detail.

One significant difference in the Sanskrit fragments is that they place this discussion before the lengthy treatment of the bodhisatta’s practices, which seems plausible.

Well this is a big question, but very briefly, the Chinese canon contains collections of suttas known as the Agamas, which are essentially similar to the Pali suttas, with minor variations.

It also contains a lot of later texts, the Mahayana sutras and other treatises, which were composed about half a millenium after the Buddha (or more) and which represent the Buddhism in India of the period in which they were written.

1 Like

Thank you :slightly_smiling_face:

The “parallels” functionality is very interesting

To the best of my understanding and knowledge your transition & interpretation makes sense, if when Devas die their karma doesn’t send them to a lower (or higher) realm, then their rebirth happens in a deva realm (they still are impermanent beings though, which are reborn).

Setting aside the Mahayana Sutras, and therefore setting aside core doctrinal differences, in terms of coverage of the Agamas, do the Agamas cover subset of what the Pali Suttas cover, are they more expansive? How do they compare in terms of coverage?

I’m aware of the existence of the Agamas but never found a translation with quality commentary like it is possible for the Pali Suttas.
One item I always wanted to see is the Agama version of Anapanasati and see if the differences regarding it’s practice between Theravada and Zen ( where the awareness is placed at the Hara ) come from there, but never got down to do it.

Mostly the same, except we don’t have a full set of Agamas from any one school. So there is say a Digha from one school, a Majjhima from another, and so on. The total number of texts is greater, because they sometimes duplicate things (there is one full Samyukta, for example, and two partial). But there are also things missing, for example there is no parallel to the Thera and Theri gatha, even those these are mentioned, the texts themselves are missing.

Yes, that’s a problem we are trying to address! Keep an eye on Chrales Patton’s translations, he is working on it!

There has been quite a lot of work in this specific area, and you can probably find what you want. I’d suggest starting a new thread and ask people to share this material. Good luck!

1 Like

This seems to be an aside only. When I look at (version up to, say 2010) I find a translation to german of KE Neumann, but which is much shorter than the current from Mettiko. In the current (Mettiko) version there is the whole bunch of recalling the buddhas exercises towards awakening, which in the Neumann version is completely missing. But also the translations of the part which is common to both, I find significant difference in the german presentation - which makes it difficult to get an opinion about a possible corruptness of the text of the sutta at all.

I’ve just tried to make something out via the “parallels”-option in sutta-central but besides looking at Mettiko’s translation I seem to be too dumb (technically?) to get any idea how I could increase my understanding what might be the reason for the vast difference in the pure sizes of the suttas nor I have an indication why Neumann comes to some different presentation about the matter of a question like “are gods existent?” and its appropriate answer compared to Mettiko’s.

Is something known about this detail of different text-bases of the german translations? And in which way this might (perhaps simply) be related to that, what Bhante Sujato discusses here?

I can see your confusion, let me try to help.

Early Buddhist texts texts are full of repetitions. These are frequently abbreviated to save space, or to spare the reader.

The differences in the length of translations is due to whether translators choose to translate the repeated sections, which make up the bulk of the sutta. This is the portion that deals in detail with the Bodhisattva’s past practices. Thus in English, my translation is 4100 words, while Ven Bodhi’s is 1,100.

This difference is found in the Pali editions. In the edition I am using, the Mahasangiti, the text is spelled out mostly fully, as it is in the Sinhala BJT edition, whereas Ven Bodhi’s appears to follow the PTS edition, which abbreviates.

This kind of difference is extremely common in Pali texts and all editors and translators are familiar with it. The PTS edition contains instructions from the English editor on how to expand the text:

&c. as from line 28 of Vol. I page 163 to page 167 line 8, substituting Bhāradvāja for bhikkhave

Sometimes such instructions are part of the Pali manuscript itself, whereas sometimes, as here, they are added by modern editors. Other times they are left implicit.

Translators work from different Pali sources, and in addition, we choose how much of such repetition to follow, which is why sometimes we might expand or abbreviate differently than the Pali text. The aim is always to ensure comprehensibility while not being overly repetitious. Sometimes people want a text that has all the repetitions spelled out; but trust me, you really don’t!

Textual corruption can only be studied by a close reading of the original language. It is a meaningless concept in translations. At best you might find a footnote or discussion by the translator.

In this case, the passage considered corrupt is quite separate from the long passage that is sometimes abbreviated. The difficult portion is, in fact, precisely that portion that is not abbreviated, because it is not found elsewhere.

You won’t find anything about this from the parallels. The parallels in this case are mostly the shared passages that are abbreviated, and in addition there are some Sanskrit fragments, but these are not published online. For a discussion of the parallels, you can check Analayo’s Comparative Study of the Majjhima Nikaya.


In this case, Neumann has a note in the printed edition to the effect that the bulk of the text of MN 100 is a repetition of a part found in MN 85 where the reader is obviously referred to. It seems doesn’t show these notes.


Thanks! Helpful as always.

1 Like

:pray:Thank you @sabbamitta and @sujato - I mostly read the *.chm-file :whale: which Wolfgang Geiger shared using his Simply the remark on the repetition-block in MN85 is missing.

1 Like