What is the difference between the Pali text of the VRI and that of the Mahāsaṅgīti?

SC uses the Mahasangiti (MS) text of the Pali canon. This is a revised and edited version of the text originally created by the Vipassana Research Institute (VRI). While the MS text is clearly a cleaner and more nicely edited edition, to what extent are these changes merely cosmetic, and to what extent do they reflect meaningful differences in the text?

This question has been bugging me for a while in my translation work. I frequently come across cases where the MS text has quirks which I would not expect in a thoroughly edited text. For example, the same sentence might be repeated with different punctuation, say by substituting a comma for a semicolon. This becomes very noticeable in the interface I use, which breaks the text on “major” punctuation, but not on commas.

There are other trivial details, for example the anunāsika might be spelled as ṅg in one sentence and ṃg in the next. Nothing terrible, but again I would expect a fully standardized text to resolve these details: it certainly helps when searching.

Sometimes the errors are more obvious. Consider the following case. AN 6.34 ends with a stock phrase. However, the phrase is presented with major punctuation errors. In the VRI text we have:

Atha kho āyasmā mahāmoggallāno tissassa brahmuno bhāsitaṃ abhinanditvā anumoditvā – ‘‘seyyathāpi nāma balavā puriso samiñjitaṃ vā bāhaṃ pasāreyya, pasāritaṃ vā bāhaṃ samiñjeyya, evamevaṃ – ‘brahmaloke antarahito jetavane pāturahosī’’’ti.

Whereas the Mahasangiti has:

Atha kho āyasmā mahāmoggallāno tissassa brahmuno bhāsitaṃ abhinanditvā anumoditvā— “seyyathāpi nāma balavā puriso samiñjitaṃ vā bāhaṃ pasāreyya, pasāritaṃ vā bāhaṃ samiñjeyya; evamevaṃ— ‘brahmaloke antarahito jetavane pāturahosī’”ti.

The problem seems to stem from the fact that the phrase, which appears at the end of the sutta, closes with pāturahosīti, which the Pali texts treat as direct speech. But it’s not, it’s a narrative portion. The Suttas usually indicate the end of the text with a closing -ti, for example, Idamavoca bhagavā. Attamanā te bhikkhū bhagavato bhāsitaṃ abhinandunti. However in this case the Pali texts try to punctuate it as if it were direct speech. It should be corrected to something like:

Atha kho āyasmā mahāmoggallāno tissassa brahmuno bhāsitaṃ abhinanditvā anumoditvā—seyyathāpi nāma balavā puriso samiñjitaṃ vā bāhaṃ pasāreyya, pasāritaṃ vā bāhaṃ samiñjeyya—evamevaṃ brahmaloke antarahito jetavane pāturahosīti.

So apart from mechanically correcting the punctuation to proper Unicode glyphs, there’s no difference between VRI and MS, even though it is obviously just a mistake. Even a quick scan reveals this, so I wonder how it escaped the editing process of the Mahasangiti?

Given that I am relying on this text for my translation, under the assumption that it’s the most reliable version available, I realized I needed to look more closely at the differences. I have not done a thorough study, but made a quick check of three texts: MN 2, MN 102, and Pvr 1.1. Here’s a summary of what I found.

I used the Meld diff tool to investigate the differences. I took the plain text from SC and from the VRI site. For simplicity, I removed the variant readings.

The most immediately apparent difference is in the punctuation conventions. The MS has used a better formed and nicer set of conventions. For example, the VRI edition uses two single apostrophes (’’), which the MS changes to double (”); and it changes en-dash to em-dash, eliminates extra spaces, and so on. These choices are well-considered and make for a nicer text, but they don’t affect the meaning.

The MS text frequently changes anunāsika from ṅ to ṃ, however this is not entirely consistent. For example the VRI has āraññikaṃgaṃ piṇḍapātikaṃgaṃ paṃsukūlikaṃgaṃ where the MS has āraññikaṅgaṃ piṇḍapātikaṅgaṃ paṃsukūlikaṅgaṃ. I haven’t been able to discern any pattern here, but perhaps the MS is simply sticking more closely to the original printed edition of the Sixth Council.

The position of the apostrophe marking end quote is corrected, eg. mūla’nti becomes mūlan’ti.

MS tends to keep long words together that are separated in VRI. Eg Dassanā pahātabbāsavā becomes Dassanāpahātabbāsavā.

Somewhat more significantly, as can also be seen in the above example, VRI keeps all headings in the nominative, while MS uses the stem forms. Presumably the sources for these texts did not use headings, and they are adopted from the endings of the texts, or else the uddānas. VRI has kep the same form as the original, creating a more, if you like, “native” text. MS, by using the stem form which is not found in manuscripts, calls attention to the fact that the headings are a modern convention. I like this way of doing things. We do the same thing on SC by making the headings a different color and font than the body text.

Leaving aside these small matters of style, I found few substantive differences. In fact in the two MN suttas there was no difference.

I chose to also compare a chapter of the Parivara, since I guessed that this obscure text would have received less editorial care. And I found that this was indeed the case, although the differences are still minor.

There’s only one case where the MS has added punctuation:

VRI: Kā paṭipattīti? Na evarūpaṃ karissāmīti
MS: Kā paṭipattīti? “Na evarūpaṃ karissāmī”ti

However there are a number of spelling corrections:

VRI: Katamena uddesena uddesaṃ āgacchatīti? Tatiye uddesena uddesaṃ āgacchati.
MS: Katamena uddesena uddesaṃ āgacchatīti? Tatiyena uddesena uddesaṃ āgacchati.

VRI: kāyasaṃsaggaṃ samāpajjantassa saṃghādiseso
MS: kāyasaṃsaggaṃ samāpajjantassa saṃghādiseso

VRI: Saññācikā kuṭiṃ kārāpentassa
MS: Saññācikāya kuṭiṃ kārāpentassa

VRI: Dubbacassa bhikkhuno yāvatatiyaṃ samanubhāsanā
MS: Dubbacassa bhikkhuno yāvatatiyaṃ samanubhāsanāya

VRI: paṭhamo aniyato kattha paññatto
MS: paṭhamo aniyato kattha paññatto

VRI: Eke samuṭṭhānena samuṭṭhāti
MS: Ekena samuṭṭhānena samuṭṭhāti

VRI: Bhikkhussa duṭṭhullāpattiṃ
MS: Bhikkhussa duṭṭhulla āpattiṃ

VRI: Bhūtagāmapātabyatā pācittiyaṃ kattha paññattanti
MS: Bhūtagāmapātabyatāya pācittiyaṃ kattha paññattanti

VRI: Samagge saṃghena cīvaraṃ datvā
MS: Samaggena saṃghena cīvaraṃ datvā

VRI: Channaṃ āpattisamuṭṭhānānaṃ ekena samuṭṭhāne samuṭṭhāti
MS: Channaṃ āpattisamuṭṭhānānaṃ ekena samuṭṭhānena samuṭṭhāti

In each of these cases the correction seems well-considered.

This was for me a heartening result. There’s about 9500 words in this text. So the error rate even in the VRI text is about 0.1%. This is a much better error rate than in most texts, so we shouldn’t think the VRI text is poorly done at all, especially bearing in mind that this text was chosen specifically because of its obscurity. And while a closer examination might reveal more, I didn’t find any errors at all in the MS text. So kudos to the efforts of both the VRI team and the MS team!

An unexpected finding was that, while a few of the errors were the usual random typos you might expect (saṃghādiseso, aniyato), most of the errors were of a specific sort. Namely, dropping the na or ya from the end of an oblique case, usually instrumental, occasionally locative. I don’t know why this is so, but it can’t be mere chance. However, it’s not that the VRI actually thinks these are correct readings. For example, the phrase samaggena saṃghena cīvaraṃ datvā is repeated many times in the text, and only once is it misspelled. Perhaps it’s just a laziness in typing, or perhaps in the original script these indications were swallowed for some reason.

So in the end, while I still have some questions about it, it seems clear that the VRI text is pretty good, and the MS text is even better.


Thanks for this, bhante. Vigilance is always the best policy. I too have come across a few places where the MS reading is not satisfactory, and I’ve had to use the PTS reading instead. The problem with this is that not all errors are immediately obvious, and as such the translation may not always reflect the best Pali reading. But I am not sure what can be done with this.

To give you some idea of the sort of errors we are dealing with, here are some examples:

Bhikkhu pārājika 4:

MS: Vatthu (basis)
PTS: Vattu (to say)

This is a heading in the permutation series and as such not very important. The context makes it clear, however, that the PTS reading must be correct.

Bhikkhu saṅghādisesa 1:

MS: Kaddamusseko (I don’t even know what this is supposed to mean.)
PTS: Kaddamudaka (mud and water)

This is found in the Vinīta­vatthu­uddāna­gāthā. The MS reading is rather mysterious, since usseka does not even seem to be a word.

Bhikkhu pācittiya 39:

MS: Maccho nāma udako vuccati. (Fish: water is what is meant.)
PTS: Maccho nāma udakacaro vuccati. (Fish: what lives in water is what is meant.)

This is a rather weird mistake. It shows that almost anything can happen.

Bhikkhu pācittiya 44:

MS: tiracchānagatāya vā manussaviggahitthiyā vā (an animal or a human woman)
PTS: tiracchānagatamanussaviggahitthiyā vā (or an animal in human form)

Only the PTS makes sense in the context. The MS reading implies that you cannot sit in private on a concealed seat with an animal!

Bhikkhu pācittiya 59:

MS: appaccuddhāraṇasaññī (perceiving it as not relinquished)
PTS: paccuddhāraṇasaññī (perceiving it as relinquished)

Just a basic editing glitch.

In addition to this there are number of minor editing mistakes that do not really affect the meaning.

My conclusion is that there is no way of avoiding consulting alternative readings.


Thanks for these. Yes, I have encountered many such examples. The fish one is particularly weird! I can understand if they stuck with their source in ambiguous cases, but I would have expected them to fix obvious mistakes.

In translating, I use the MS text as the basis, but like most translators, I just use whatever reading seems best on a case by case basis. I’ll also check the commentary, Tika, and parallels when necessary. Ven Bodhi’s choices are of course well considered and in most cases I follow him.

In AN 6.64, for example, I’ve adopted BB’s reading from the Sinhalese edition. The passage has a variant; either ghaṭessati or upaṭṭhahissati is added, apparently unnecessarily. A curious detail, not noted by BB, is that the Comm. says Samādhimhā cāvessatīti samādhito uṭṭhāpessati. So possibly upaṭṭhahissati was read back from the commentary and found its way in an altered form into the MS edition.

Someone should make a list of all these. Just kidding!