Establishment of the Text


In establishing this text I have followed certain principles which relate to choosing the correct reading, and how to record the variants. Wherever possible I have attempted to explain why I have taken certain readings in preference to the alternatives (or pointed it out when the choice necessarily is arbitrary). Some of the considerations I have used to establish the text include the rules of Pāḷi grammar and metre, and idiom; and other guides like context and consistency of reading and presentation. I give some examples below:


29: āgatā ca Arahanto vijite phāsuṁ (accusative, ChS: phāsu, nominative) vihareyyun?’-ti

161: “Ko nu kho Ānanda Pāṭaligāme nagaraṁ māpetī (singular)?” ti (BJT: māpentī, plural)

193: dukkhasamudayo (nominative) Ariyasaccaṁ anubuddhaṁ paṭividdhaṁ (BJT, Thai: dukkhasamudayaṁ, accusative, but nominative is needed with the past participle)

339: Ossaṭṭhe ca Bhagavatā (PTS: Bhagavato, which is the dative or genitive, when an instrumental is required, this is possibly a printing error) āyusaṅkhāre mahābhūmicālo ahosi


252: Bhagavā kira Vesāliṁ anuppatto (BJT, PTS: Vesāliyaṁ, locative, but the normal idiom is that anupatto takes the accusative rather than the locative)

401: Evaṁ vutte ahaṁ (BJT: vuttohaṁ = vutto ahaṁ. Normally the locative in used in these constructions, this may be a printing error) ... Māraṁ Pāpimantaṁ etad-avocaṁ


183: in a Vetālīya line: Kullaṁ hi jano pabandhati, (Thai: kullaṁ jano ca bandhati, probably an attempt to conform the metre to a Siloka as the last line can be scanned as either Vetālīya or Siloka)

195: in a Siloka line: saṁsitaṁ Thai: saṁsaritaṁ, an alternate form of the verb, but one that ruins the metre)

340: in a Opaccandasaka line: Bhavasaṅkhāram-avassajī Muni, (Thai, ChS: avassaji, which spoils the metre, PTS: Munī, but there is no reason for a long vowel here)


49: Text: ponobhavikāya, BJT, Thai, ChS: ponobbhavikāya, but there is no reason for gemination here.

149: Text: brahmacārino, BJT, ChS, PTS: brahmacārayo, but the correct form for plural masculines in -ī is -ino.

180: Text: sammiñjeyya, ChS: samiñjeyya, similarly throughout. Although there is no historical reason for the doubling of the consonant it appears this is the normal way the form is written in Pāḷi.

288: Text: Vihārapacchāyāyaṁ (Thai: Vihārappacchāyāyaṁ, but gemination is not warranted in this word)


52: bhikkhū sandississanti (future, which context demands, BJT, PTS: sandissanti, present)

138: Thai has the Buddha going to the rest house (āvasathāgāraṁ) pubbaṇhasamayaṁ, in the morning time, which from the context we can see is the wrong time; text has sāyanhasamayaṁ, in the evening time (PTS omits the time)

230: tasmiñ-ce kālakate (Thai: tasmiṁ tasmiṁ kho; ChS: tasmiṁ yeva; but we need a conditional statement here) Tathāgataṁ upasaṅkamitvā

263: Atha kho te Licchavī aṅgulī poṭhesuṁ (plural, BJT, ChS: aṅguliṁ poṭesuṁ, singular, here and below. It would seem a plural is required as one finger is hard to snap)

286: Atha kho Bhagavato ... kharo ābādho uppajji ... Bhagavā sato sampajāno adhivāsesi (aorist, BJT, PTS: adhivāseti, present tense)

290: (Ānanda speaking): diṭṭhaṁ (singular, Thai: diṭṭhā; PTS diṭṭhā me, plural, but a singular is required by the context) me ... Bhagavato khamanīyaṁ

301: Gaṇhāhi Ānanda nisīdanaṁ yena Cāpālaṁ Cetiyaṁ tenupasaṅkamissāma (plural, PTS: tenupasaṅkamissāmi, singular, however a plural is needed as they were going together)

352: Tathāgato anuttaraṁ Dhammacakkaṁ pavatteti (when the Realised One sets the unsurpassed wheel of the Teaching rolling, present tense, Thai: pavattesi, aorist form which doesn't fit here) ... tadāyaṁ Paṭhavī kampati


3: Ajātasattu Vedehiputto Vajjī (plural, Thai: Vajjiṁ singular, not wrong, but uses the plural form everywhere else) abhiyātukāmo hoti

10: BJT, Thai: vījiyamāno, but below BJT and Thai write vījamāno.

14: ChS, PTS: Vajjī- in Vajjīkaraṇīyāni here, and PTS below, but not 3rd time.

45: PTS omits bhikkhave here but has it elsewhere in this position.

49: bhikkhū uppannāya taṇhāya ponobhavikāya na vasaṁ gacchissanti, (will not come under the influence of craving which has arisen for continued existence, hypothetical future, consistent with others in the list, BJT, PTS: gacchanti, present tense)

141: BJT: sīlaṁ vipattiyā, parsed form here sīlavipattiyā eslewhere.

There are three variants which completely reverse the meaning of the text, and so are worth pointing out here:

Other differences occur, such as PTS always writes kāmāsavā bhavāsavā diṭṭhāsavā avijjāsavā whereas the sequence appears as kāmāsavā bhavāsavā avijjāsavā here. Both sequences are known to the texts, but none of the other editions include it, and I have therefore not included it in the text here.

Another class of variants are those of words that show gemination, where many times there appears to be no rule we can call upon to establish which is the correct reading. Examples include Text: upaṭṭhitassatī, where BJT, PTS prefer upaṭṭhitasatī; Text: supatiṭṭhitacittā where BJT has suppatiṭṭhitacittā.


As the text is, in places, very repetitious, when recording variants I have preferred to summarise them rather than list them individually, so that they are normally noted by saying ‘here and below’, ‘throughout’, ‘always’, etc. If a variant occurs only at a certain place I mark it as ‘here’, or ‘here, but below (otherwise)’.

When I have gone against the reading taken in PTS I have many times shown that PED was also in disagreement, and presumably therefore the Editor had changed his mind between preparing the text and the Dictionary. Even a glance at the text established here will show that there are many hundreds of variant readings recorded. This despite the fact that I have summarised the variants rather than spelt them out one by one. However, the vast amount of variants recorded are, in fact, trivial.

They mainly consist of either variant spellings, such as paññāpenti/paññapenti, acchiddāni/ āchiddāni; or variant forms, such as disvā/disvāna (alternate forms of the absolutive) mama/mamaṁ (alternate forms of the genitive). We many times see that words that appear in the text in parsed form are in compound in the variants, e.g. pacchimaṁ bhittiṁ/pacchimabhittiṁ; or, conversely, are compound forms of the variant reading: garukaronti/garuṁ karonti.

Unfortunately many others are simply printing errors, like: BJT writing parihāpenti ti; or PTS writing Koṭhigāme (elsewhere correctly written Koṭigām-). BJT In one class of error in BJT, that of the u/ū vowel sign written under certain letters in Sinhala, I have simply had to ignore the variants as they would have amounted to hundreds, none of which seem to be a sure guide to what the editors intended. and PTS are very bad in this regard, but ChS and Thai also have their fair share of errors.

Other variants include spelling variations where we again have no real way of establishing what the correct reading should be, such as the name of the river that the Buddha drank from and bathed in on the last leg of his tour, should it be Kakutthā (as in the Text and PTS), Kukutthā as in BJT, or Kakudhā as in Thai and ChS? Is it Cāpāla as Text, BJT, ChS and PTS have it, or Pāvāla as Thai records? In fact, we have no real way of knowing, and can only examine the manuscripts to find out what the bhāṇakas thought the word was.

Very often the editions show assimilation of the nasal to the following consonant, so that what is written here as evaṁ me, sutaṁ metaṁ, evaṁ pasanno, etc. may appear in the printed editions as evam-me, sutam-metaṁ, evam-pasanno. Many times I feel these simply represent scribal preferences, and not wishing to multiply the variants unnecessarily, I have not recorded them.

Additions to the Text

The text of Mahāparinibbānasuttaṁ is divided into six sections for recitation (bhāṇavāraṁ). A bhāṇavāra is calculated as being the same as 250 Siloka verses in length. As a Siloka is normally considered to have 4 lines of 8 syllables each, that would give 32 syllables to the verse. In fact Silokas are no so rigid, lines are often 9 syllables long, rarely 7, and there are sometimes seen 5, 6 or more lines in a verse. But the ideal is 4 x 8, and that is what is used for the purposes of the calculation. A bhāṇavāra therefore is 250 x 32 syllables long, which is equal to 8,000 syllables in length. In fact, of course, as the prose is not so definite in length as a verse, it follows that many times this is only an approximate length. Generally in prose we find that a bhāṇavāra is in fact somewhat shorter than the ideal.

To anyone reading the text as it stands today it is clear at once, that certain sections seem to be much longer than others. It therefore seemed worthwhile to count the syllables and compare their length against the ideal in the hope that this would throw light on the text as it is currently being passed down. Here is a table showing the length of the 6 sections:

We can immediately see from this that the 4th and 6th Chapters approximate very closely to the ideal. The 2nd Chapter is somewhat shorter than expected, but perhaps not unusually so. The remaining Chapters (1st, 3rd & 5th) are longer than we would expect them to be, especially the 3rd Chapter.

If we examine these more closely now we find that certain Teachings seem to interrupt the flow of the narrative. In the first section the Buddha talks about the seven conditions which will prevent decline in the Vajjīs. This is followed by applying the same teaching to the Saṅgha. That the Buddha would apply the teaching in this way seems quite natural, as throughout the discourse he shows concern that the Sāsana should be stable now that he is approaching his Final Emancipation.

It is followed however, by another seven conditions, and then another seven, and then another seven, and then another seven, and then another six. The interesting thing here is when we remove these 5 Teachings from the syllable count we find the Chapter is then 8,115 syllables in length (101%), which is very close to the expected syllabic count.

In the Aṅguttaranikāya (7.24-27) we see that the first four of these Teachings occur there also in exactly the same order. The last section also occurs in Aṅguttaranikāya at 6.11 and 12. It seems to me that there are two possibilities here. Either these Teachings have been brought in from the Aṅguttara text, because they were found to be further conditions that the Buddha (at one time or another) stated were true and proper conditions for the Saṅgha to develop to prevent decline. Or they were linked in the oral tradition by number and theme.

The 3rd Chapter is the most extraordinary, as at present it contains no less than 12,869 syllables, which is more than 50% longer than we would expect for a bhāṇavāra. In this Chapter the Buddha gives up the will to live on, at which point there is an earthquake, which is followed by the Buddha explaining to Ānanda the eight conditions which cause an earthquake. Rhys-Davids questioned the genuiness of this Teaching, P. 113, n. 3. but it seems to me to follow on quite naturally from what is happening in the narrative, and also serves to prepare the ground for the rest of the story.

It is then followed, however, by three further sets of eights, the eight assemblies, the eight means of mastery and the eight liberations, which appear to be completely unrelated to what is happening and arbitrarily inserted at this point. It is not quite arbitrary though, because as we saw above, material may come into the text through numerical association, and that seems to be the case again here. It may come of no surprise then that we find in the Aṅguttaranikāya that these Teachings occur in the same vagga. AN Bhūmicālavaggo, 8.70, 69, 65, 66. When we remove them from the syllabic count though we have 9,730 syllables, which although much closer to what is expected is still long.

In section 22 the Buddha tells Ānanda how he has now given up the will to live, and will attain Final Emancipation in 3 months time. Ānanda then begs the Buddha three times not to pass away so soon, but the Buddha asks why Ānanda is insisting on this? Ānanda explains that he had heard the Buddha say that whoever had mastered the Four Paths to Power would be able to live on for the completion of his lifespan.

The Buddha blames him for not asking sooner, and says that if he had done so he may have rejected it the first couple of times, but not the 3rd time, but now the Buddha has made up his mind and he is unable to go back on his word. So far the story, though drawn out, seems genuine enough. But then there are 2 sections where exactly the same explanation, questioning and recrimations are repeated in regard to other times and places where Ānanda had a similar opportunity and failed to make the request. In my edition I have filled in the ellipsis passages, believing that the recitor (bhāṇaka) was only using the written text as an aide-de-memoir, and would have filled them out himself when reciting. If we leave them out, though, the same problem would still arise even if to a lesser degree. These sections simply repeat what has gone before, and do not seem to add anything to the narrative. When we remove those also from the syllable count, we find the Chapter now has 8,374 syllables which would be about the right length (104%).

The 5th Chapter is also much longer than we would expect, but the arguments here are by no means so clear-cut. It does seem to me that there is evidence of additions to the text, though exactly where they begin and end is not so straightforward. Section 37 begins: “After this was said, venerable Ānanda addressed the Fortunate One, (saying): ‘Reverend Sir, may the Fortunate One not attain Final Emancipation in this small town (of Kusinārā).’ ” This seems to follow on from the Buddha’s announcement that he would attain Final Emancipation in Kusinārā, which occurred towards the end of section 34.

The intervening sections 35 and 36 on the four places for devotional worship, Ānanda’s marvellous qualities, the Universal Monarch’s marvellous qualities, and those considered worthy of a Shrine, interrupt the flow of the narrative, and introduce much matter that may be late in origin. It may come as no surprise then that most of the material that occurs here also occur as discourses in Aṅguttaranikāya. AN 4.118: Saṁvejanīyasuttaṁ, 4.129: Ānanda-acchariyasuttaṁ, 4.130: Cakkavatti-acchariyasuttaṁ, 4.247 Thūpārahasuttaṁ. Again when we remove the two sections and check the syllabic count we find we have a more reasonable 7,196 syllables (90% of the ideal).

I believe that the quantative analysis made above establishes that there have been additions to the text. Just what material is additional also seems clear in Chapters 1 and 3, but it is less so in Chapter 5. In any case by suggesting that these Teachings are additional to the text I am not suggesting that they are in any way alien to the teaching of the Buddha, indeed I do not think this is so, and most of the Teachings also appear elsewhere. I only wish to point out that the text we now receive very much looks like it has been supplemented in various ways as it was passed down in the oral tradition.

Some of the additions must have been made very early indeed, as they occur in the Sanskritised Mahāparinirvāṇasūtra also, a comparative edition of which I will prepare soon and which will hopefully throw more light on the textual tradition by which this material has come down to us.

Ānandajoti Bhikkhu
May, 2008