Dharmaniyānatā & Idappaccayatā in SA 296

Some time ago, Coemgenu posted a question about SA 296.

I’ve been thinking about this quite a bit. We have three texts:

Pāli § SN 12:20,
Middle Chinese (MC) SA 296, ca 5th Century, and
Turfan Sanskrit (TS) 14, ca 13th C.

P is a Theravāda text, MC is Sarvāstivāda, and TS is probably also Sarvāstivāda (experience suggests that the Turfan Samyuktāgama texts are closer to MC than to P)


P. uppādā vā tathāgatānaṃ anuppādā vā tathāgatānaṃ, ṭhitāva sā dhātu dhammaṭṭhitatā dhammaniyāmatā idappaccayatā. (SN II.25)
“The arising of a tathāgata or the non-arising of a tathāgata, there is an enduring principle: the fact of dhammas being conditioned, the constraints on dhammas, and specific conditionality.”

MC. 若佛出世,若未出世,此法常住,法住法界 (T2.84.b17-18)
“Whether a Buddha arises in the world, or not, this is the unchangeable nature of dharma, the status of dharma, the element of dharma.” (Choong 2010: 45)

TS. ity utpādād vā tathāgatānām anutpādād vā sthitā eveyaṃ dharmatā dharmasthitaye dhātuḥ (TS 14.5)
The arising of a tathāgata or the non-arising of a tathāgata, this principle remains: naturalness and the stability of dharmas.


The Pāli seems to me to be the least corrupt text. The gist of what it says is that there is an enduring principle (ṭhitā dhātu) that has three characteristics. I think the context requires us to read these as applying to the items in the list of nidānas (dhammā) rather than to the teaching (Dhamma).

The Turfan Sanskrit manuscript has only two characteristics and one of them has lost a part, i.e. dharma…tā. This is a common scribal error. Also it has moved dhātu to the end. This is not a problem in terms of comprehension, in that sthitā must still be an adjective of dhātuḥ, but it might have introduced an ambiguity (see below). There is no apparent counterpart for idappaccayatā. (Note that in Sanskrit iyaṃ is a feminine pronoun)

MC also has three characteristics 此法常住, 法住, and 法界. Forget the Taishō punctuation, it’s misleading. A couple of things stand out. Guṇabhadra has ignored the context and translated as though dhamma is singular. Hence 此法常住 “The Dharma is eternally abiding”. We know this because later on when he wants to refer to dharmas plural he uses the form 此等諸法.

Now, if we look at this from a Theravāda point of view, then 此法常住 is problematic, because we think the text is talking about the arising of dhammas and they are supposed to be eternal. But from a Sarvāstivāda point of view they are. I’ve looked at this in an essay: Sarvāstivāda Approach to the Problem of Action at a Temporal Distance.

Next we notice that 法常住 and 法住 are essentially the same statement. One has “eternal” 常 and one does not. Why the repetition. The character 住 has more or less the same semantic field as √_sthā._ so there are both translations of an Indic term like dhammaṭṭhitatā. Why twice? I’ll come back to this.

The third Chinese characteristic is 法界 where 界 is the character commonly used to represent Indic dhātu. Ergo the obvious reading of 法界 is dharmadhātu/dhammadhātu. At first I thought this might also be an error based on the dhātu being out of place in the Sanskrit text. This assumes that MC and TS were translated from a common or similar original (probably in Gāndhārī). It has been assumed that 法界 = idappaccayatā merely on the basis of position, but this is a strange choice because neither character contributes anything semantic. A modern translation of idappaccayatā is 此緣性 where the Indic pronoun idaṃ is represented by the MC pronoun 此. This does not occur in the Digital Dictionary of Buddhism and nor have I been able to locate how DDB might translate it (C > E is easy; E > C is more or less impossible).

The trouble is that 此緣性 does not occur in the Āgama translations. So how do they translate idappaccayatā? It turns out that in the whole SN idappaccayatā only occurs in SN 12:20. So there is no way to see how Guṇabhadra translated it in other contexts. In fact this word is very uncommon. Across the Nikāyas it only occurs again in MN 26, and there only once. It will take me a while to dig it out.

In the meantime something occurred to me. SA 296 is a Sarvāstivāda text and eternal dharmas (i.e. dharmas that are always existent sarva-asti) is a doctrine peculiar to them. So what if of the three characteristics in this text–此法常住, 法住, and 法界–that 法常住 is the odd one out.

For the sake of argument let us assume that the original text had just two characteristics corresponding to P. dhammaṭṭhitatā and dhammaniyāmatā. Somewhere along the way a Theravāda editor inserted idappaccayatā here because it had become an important term in Theravāda. This happened in South India or Sri Lanka. Meanwhile in the far North where Sarvāstivāda was the major school for centuries, an editor decided to emphasis that dharmas are not just abiding, but eternally abiding.

How would this scenario change how we read the Chinese translation? If we take 法常住 as a late interpolation then we have two terms and they would match the first two Indic terms

dhammaṭṭhitatā = 法住
dhammaniyāmatā = 法界

But is this plausible? Since 住 is commonly used to represent Indic √_sthā_, dhammaṭṭhitatā = 法住 seems unproblematic. But 界 is most often dhātu, so how do we explain this? If we look up 界 in a Middle Chinese dictionary (I use Kroll) then we find it means “boundary, limit”. And what does niyāma mean? It means a “constraint, restriction”; or with a process, “the fixed course and/or outcome of that process.” Or in other words, a limit. So perhaps here, 界 is being used in its usual Middle Chinese sense of limit? At least there is considerably more overlap of semantic fields here than with 法住.

However, later in the text, where TS has dharmaniyāmatā has the Chinese text has 法空. Expert opinion suggests this must be wrong and suggests that 法定 was intended. This kind of mistake is easy to make in Chinese. Also Choong Mun-keat has shown that by this time Buddhist authors were beginning to insert the word śūnyatā or 空 in all sorts of places, and not always felicitously. But then if a character is definitely wrong, who is to say that the correct character is one thing and not another. All we know is that 空 is out of place - by accident or design.

As with idappaccayatā, dhammaniyāmatā only occurs in this sutta of the Samyuttanikāya. We have no record of how Guṇabhadra usually translated niyāmatā because it only occurs this once. So resolving any problems will be difficult. I’ll try to look into whether Guṇabhadra translated other texts that contained this word.

All this is conjecture and I am trying to think of ways of tying it down. It would be useful to get examples of how other Middle Chinese translators dealt with idappaccayatā, for example (perhaps a reader can help with this?). And I will endeavour to track it down in MN 26. Still I think this is a plausible, if complicated, alternative reading that also makes sense of the available information.

I should also note that this text contains other quite unique terminology that does not seem to be assimilated into mainstream Theravāda. And this makes the sutta very interesting for any historian of ideas.

Post scriptum

It turns out that the Chinese version of MN 26, i.e. MA 204 omits the passage that includes the word idappaccayatā. But Brahmā’s request to teach (Section 20 of MN 26) is recounted in EA 19.1. EA 24.5 recounts part of the story but also misses out the passage of interest.



In the Sanskrit, one should translate dharma-tā as “the essence (the nature) of the dharma” .
॰ता -tā forms suffixes of state & qualities.

Idappaccayatā also appears in SN 6.1.

But this generation delights in attachment, is excited by attachment, enjoys attachment. For a generation delighting in attachment, excited by attachment, enjoying attachment, this/that conditionality and dependent co-arising (idappaccayatāpaṭiccasamuppādo) are hard to see.

EA 19.1 (the parallel to SN6. 1,) does not address the issue of idappaccayatā.

Paṭiccasamuppāda is just about the stable part, (the fixity) of the nidanas’ chain - the “big” dhamma.
Paṭiccasamuppannā is just about the impermanence (and changing nature) of the “small” dhammas.
What is “eternal” is the former, (if we can say “eternal”).

此法常住 can actually still mean “these dharmāḥ eternally abide” because Chinese doesn’t slip into “singular=default” if plurality isn’t specified anymore than Japanese or Korean does. Plurality and singularity are often context sensitive. Afaik.

As in, it is possible that there is no instances of dharma as law and all instances of dharma are actually dharmāḥ here. Technically, afaik.

And I’m not sure about the character(s) you asked me about in PM. I’m looking it up.

The error here IMO was more likely

dharmaniyāmatā --> dharmanairātmyatā

Than a “śūnyatā” insertion.

I assume this was what Venerable Yin Shun hypothasized occurred (afaik he is the one that says this is a manuscript error).

法空 is a regular translation of dharmanairātmya

This is problematized by the hypothesis that 法定 is dhammaniyāmatā

1 Like

Maybe you can find more out of Ven Yin Shun’s work.

This is Ven Yin Shun’s footnote from his edition of the saṃyuktāgama & associated commentary, it is footnote 2, in volume 2, page 36:


I found the Chinese text a long time ago on the internet while searching for some stuff. I think there is a English translation of his saṃyuktāgama & associated commentary, but I am not sure.

It’s not very detailed, he just says he thinks its 定 instead. He may have written more on why he made the decision, though.

It turns out that the passage is also cited in 《大智度論》 T1509.

有佛無佛,是因緣法相續常在世間 (T1509.253c.1-2)

Ani Migme’s translation is, “Whether there is a Buddha or there is no Buddha, this causality (idaṃpratyayatā), this nature of things (dharmatā), is always present in the world.”

However, here 因緣 seems to corresponds not to idaṃpratyāya but to hetu-pratyāya “causes and conditions” , while 法相 corresponds to dharmatā. So I’m not in agreement with Ani Migme on translating 因緣.

I can get by in Middle Chinese, but I can’t read Mandarin at all.

Note that dharmanairātmya is not listed as a translation of 法空 in the Digital Dictionary of Buddhism. So I’d want to know more about where it is used in this way.

Hmm. This is interesting but I’m not sure of your reasoning. Why is nairātmyatā more likely? In Sanskrit scripts the two words look very different! It is quite noticeable in Devanāgarī for example नियामता vs नैरातम्यता. Also the two words mean very different things. So why would one replace the other. And note that in the only extant Sanskrit manuscript we have this has not happened.

I wondered if he was aware of the Sanskrit manuscript, which has dharmaniyāmatā.

1 Like

The Nan Tien Institute’s online dictionary.


Your relevant footnotes are:

(Ding ‘法空’; FGDB ‘法空’; SH ‘法空’)

Ding is Ding Fubao 丁福保 1922, Dictionary of Buddhist Studies 《佛學大辭典》.

FGDB is “Fo Guang Dictionary of Buddhism 佛光大辭典.”

SH is Soothill, WE & LA Hodous 1937, A Dictionary of Chinese Buddhist Terms. London: Kegan Paul, Trench, Trubner, and Co. Ltd.

The first two are in Chinese, I presume that last is in English.

You can view the FGDB online at it has the latin text “dharma-nairātmya” amidst the Chinese in the article.

It is possible that the error would have occurred as a result of transcribing a heard recitation, rather than copying a text directly.

These words are even more similar phonetically in Prākrits than in their artificial Sanskrit looks.

Imagine dharmanairātmyatā in Pāli or a related phonetically constrained Prākrit of the like.

I’m not saying that’s what happened. But I think I can come up with examples of errors of textual transmission in a manuscript based on hearing it read out wrongly.

1 Like

I’m definitely aware of texts being misheard - I note one in my article on the 100 Syllable Vajrasattva Mantra from a few years ago. But, any example of this particular error would be interesting.

As far as I can tell Pāli only uses anattā as a synonym for nairātmya, Edgerton doesn’t list a Prakrit form in his BHSD. And there is no entry for it in the Gāndhārī dictionary.

The Pāḷi counterpart would be *dhammanerāttaniya. And vs dhammaniyāmatā I’m not seeing any great ambiguity.

1 Like


नैरात्म्य nairātmya (vṛddhi of nir-ātman-ya), the non-Atman/atman doctrine, has nothing to do with emptiness; but with no- continuity.

Per extension, dharmanairātmya, that means “the essence of no atman/Atman in the dharma”, might have the underlying meaning that dharmas are empty of atman/Atman - but certainly not that the nature of the dharmas themselves, is inherently empty.

  • There is no water in this glass !
  • Yes indeed, the glass is empty.
  • Ok, but the inherent nature of the glass itself, is not to be empty.

That’s a fair point, but I don’t think you were prepared for the sheer vagueness of my claim, which is really only a speculation.

Pāli is only one Prākrit. It deletes the M and geminates the T, but Prākrits have a wealth of options on terms of ways they can change Sanskrit, and, depending on the time and geography of a speaker, their Sanskrit is likely to be pronounced in the style of a Prākrit.

In the 2nd century BC, for instance, with added literary attestation into the 10th, in Northern India, we have people pronouncing “atma” as “appa” (the Hindi ap being a further descendant in this same phonetic direction) in Shauraseni Prākrit.

Its quite possible that one of these would elect to geminate the m instead of the T.

IMO “appa” likely came from an “amma” before. Consider the phonetic location of the P vs an M as opposed to a T mutating into a P.

Starting on page 281 of this German dictionary of various Prākrits (Grammatik der Prakrit-Sprachen - Richard Pischel - Google Books) there are a bunch of options for ātma.

Some Prākrits made it into addha, some appa, some aya, some aga…

No amma, though.

It looks like the only realistic options to choose from aside from the Pāli are ātpa, ātva, appā, ādā, & aẏā.

Ha. I’ve come around. I think you are right that 法空 is dharma-nairātmyatā, in this context at least. I looked at the other occurrences of it in Samuktāgama and it makes sense.

BTW AMg will be Ardhamāgadhī and it is unlikely to be the source language of a Buddhist text. It was used extensively used by Jains, but not to the best of my knowledge by Buddhists.

Newton was suffering from a very common cognitive bias called, simplicity bias.

Simplicity bias is a cognitive bias towards holding views which can be explained by a simple narrative, as opposed to views which take more complex details into account and are generally more accurate, realistic, and balanced.

It seems likely that this bias is primarily caused by limitations of individual memory: it is much easier to remember details that fit neatly into a unidirectional conclusion.

1 Like

Does this refer to specifically āẏā?

Perhaps this didn’t deserve such a quick responce (he said, in jest), after all, it was a question asked in a way that seemed like a “test question” or something of the like.

I asked because there is an attestation in that German text of āẏā as Maghadi Prakrit for ātma. I wanted your opinion without knowing that āẏā was actually substantiated. I should have been more honest with my intentions.

It is likely that there are other forms than āẏā, in my thinking at least, suggested or presented in Maghadi Prakrit for ātma.

But Maghadi Prakrit, I presume, is likely highly varied in its manuscript presentation, and most likely is substantially reconstructed rather than observed.

I think you should revisit this.

Because, at least from a viewpoint informed only by the Chinese text, no parallels, IMO, it is quite possible to read this as a discourse in which 此法 is more or less always “these dharmāḥ”, referring to the list given in the text later.

This would be in-line with a tendency towards “labelling” or “naming” things as dharmāḥ, explicitly, in Chinese texts.

Consider SA 890 parallel SN 43.11 & SN 43.1 etc, asaṅkhata, is 無為法, “[the] uncreated dharma”.

No comment.