Buddhism’s Oldest History Revisited: A New Text of the Dīpavaṃsa

Abstract

The Dīpavaṃsa (Dīp), the first historical account of the Buddhist religion that has survived in Pali, is widely known through Oldenberg’s late-19th century edition (designated hereafter O). The editor himself admitted it was faulty due to the quality of his Sri Lankan manuscript sources, all of which he thought were derived from a faulty Burmese exemplar. This problematic edition prompted new printed editions of Dīp in Sri Lanka and Myanmar in the 1920s, but Western scholarship established it as a ‘problem’ text, and it was thus generally neglected in favour of the later Mahāvaṃsa. A new edition of Dīp has long been a desideratum, and in 2004 Frasch pointed out the existence of a Burmese manuscript of a different text of the work, which, for the purposes of the present discussion, we designate B1. The present authors identified two further mss. of this version and have begun editing a new edition based on this in comparison to Oldenberg and other Burmese mss. The Burmese sources reveal an occasionally faulty but widely disseminated text, designated B2, that is not dissimilar to O, plus the rather ‘better’ text of B1. In addition, we have also identified the so-called ‘Dīpavaṃsa-ṭīkā’, properly named the Sāsanajotikā, as a commentary on B1 by the major 19th century Burmese scholar Jāgara. The present article will give details of this analysis.

1. Introduction to Dīpavaṃsa

The Dīpavaṃsa (Dīp) is widely known as the first historical account of the Buddhist religion to have survived into the modern era. It relates in verse the Buddha’s awakening followed by an account of him taking his teaching to the island of Sri Lanka. The text consists of 22 chapters, covering the transfer of relics to the island, early Buddhist sectarianism and also what might be described as political information, for example, royal lineages, and mythology. Some topics are covered more than once, and this has given rise to a perception shared by a number of scholars that its composition is “haphazard”.1 In considerable part this feature of the text is put down to its pioneering status in historiography: “It is the first Pali text known to have been composed in Ceylon…” (Hinüber 1996, p. 89). As such, it is a remarkable source of data to be used in our understanding of the transmission and growth of Buddhism in Sri Lanka.

Establishing its date has been, to some extent, relatively simple for such an old text. Oldenberg points the following out in his introduction: “The Dīpavaṃsa cannot have been written before A. D. 302, because its narrative extends till that year [and] … Buddhaghosa was acquainted with a version of the Dīpavaṃsa …” (Oldenberg 1879, pp. 8–9).

This gives us a range from 302CE up to whenever Buddhaghosa is thought to have worked, which, nowadays, tends to be pushed back to the 5th century CE (Hinüber 1996, p. 102). Thus, Oldenberg and, following him, Law both cite a range between the beginning of the fourth century CE and the ‘first third’ of the fifth century. Hinüber places it not long after 350CE, which seems to simply split the difference in the range offered by earlier scholars (Law 1947, p. 2; Hinüber 1996, p. 89).

The early generation of Western scholars judged that it was itself a composition that amalgamated material from an ‘ancient’ porāṇa commentary composed at the Mahāvihāra in Sri Lanka in Sinhala. Frauwallner went on to determine that it was in fact drawn from two separate sources: on the one hand, a history of the Saṅgha and, on the other, a secular history. Both sources, he judged, were subdivided between material relating to the sub-continent and earlier material relating to historical events within Sri Lanka (see Frauwallner 1994).

Hinüber summarises the content of Dīp as covering four different but overlapping areas: the Buddha’s visits to the island; a history of the saṅgha from the First Council; a royal lineage; and a history of events within Sri Lanka (Hinüber 1996, p. 90). To this we can add the opening account of the Buddha’s awakening. Hinüber comments that “The different sources are only loosely knit together, and therefore easily recognised and separated” (Hinüber 1996, p. 90).

It is perhaps because of these features, not to say problems, that Dīp came to be replaced, within 100 years of its own composition (Hinüber 1996, p. 91), by a much more elegant historical record written by Mahānāma, who wanted to deliver a more comprehensible text, free from the faults of his predecessor:

mahāvaṃsaṃ pavakkhāmi nānānūnādhikārikaṃ

porāṇehi kato p’eso ativitthārito kvaci

atīva kvaci saṃkhitto anekapunaruttako

vajjitaṃ tehi dosehi sukhaggahaṇadhāraṇaṃ

pasādasaṃvegakaraṃ sutito ca upāgataṃ

“I shall relate a Great Chronicle which will be authoritative in variety and completeness. The one composed by old authors is sometimes too detailed and sometimes too succinct, and sometimes repetitive. [My text] is free from those faults, easy to comprehend and recall, inspires faith and urgency, and has descended from tradition” (Mv vv. 1–3).

Taken together, these difficulties with Dīp also came to be reflected in modern assessments of its text. There are, however, other difficulties, to which we must now turn.

2. Oldenberg’s Edition and Method—‘the Problem Text’

Oldenberg was working on his edition of Dīp through the 1870s and had at his disposal for this purpose 11 manuscripts. Of these, two came from Burma. This small number probably reflected the only recent annexation of Lower Burma by the British in 1853 (Upper Burma was not taken until 1885). The first of these manuscripts was collected by a Major Fryer, a possible spoil of war, and although it was copied in 1826, one third of it was missing. The second Burmese source to which he had access was a record made for T.W. Rhys Davids by a Sri Lankan, Dewa Aranolis, who did not copy the manuscript in full but simply supplied variant readings from it as marginal annotations (in Roman and Sinhala scripts) to a new full copy he had made of a Sri Lankan ms. The source text for this had been a gift from the King of Burma to the Colonial Library in Colombo. Aranolis’ double copy is now in Cambridge. Oldenberg himself expressed concern at the reliability and fullness of the second source. Where Aranolis indicated no variant, Oldenberg judged that the assumption that the Burmese manuscript therefore aligned with the Sri Lankan ms was “by no means always … safe” (Oldenberg 1879, p. 10).

The nine remaining mss, held in Paris, Cambridge and London, are all Sri Lankan. Of these, he judged at least six to be of the lowest value, and of the whole of the Sri Lankan sources, he commented the following: “All our MSS. are derived from the same original source which was very incorrectly written in Burmese characters, as we may infer from some of the blunders common to all of our MSS. Perhaps this was the MS. brought in 1812 from Siam to Ceylon by the Modliar George Nadoris” (Oldenberg 1879, p. 11).

The combination of these problems with Oldenberg’s strict methodological approach resulted in the following rather dismal reflection on his own edition: “It appeared desirable to print not the text corrected as far as possible, but the text of the codex archetypus, and to give in the footnotes my own emendations as well as those tried already in the MSS. In many passages I have refrained from correcting manifest grammatical blunders, errors in numbers of years etc., because I was afraid of correcting not the copyist but the author himself. Many passages also appeared to me too hopelessly corrupt for me to try to correct them. Of the various readings I could give, of course, only a selection, or the work would have increased to its threefold extent” (Oldenberg 1879, p. 12).

Clearly, Oldenberg was so uncertain of his text that, perhaps excessively adhering to purist method, he did not care to correct or emend the text because he could not differentiate possible transmissional errors from compositional awkwardness. All he tried to do was establish the bad text from which all his manuscripts were copied—his codex archetypus. It seems fair then to write that even the editor himself had no great faith either in his edition’s accuracy in representing the original text or in its credibility as a grammatical or factual text. He was clearly of the opinion that the Sri Lankan text as he knew it was in toto derived from a low-quality Burmese source, this conclusion driven by copyist errors that make sense in the script of that language and not in that of Sinhala. He even has in a mind a candidate for that source—a manuscript brought to Sri Lanka in 1812. It seems then that, in a historical sense, he saw the text that he worked on as a Burmese text, not by way of composition but by intermediate transmission. It is therefore remarkable, and presumably a result of the political situation of the period perhaps combined with a prejudice that only Sri Lanka could be the source of the most reliable information about a chronicle of its own history, that he could not have acted upon his insights and acquired further Burmese manuscripts to consult for his edition. This was clearly not an option for him.2

Geiger worked with Oldenberg’s edition for his study of Dīp and Mahāvaṃsa. The first section of his work is a fascinating study of the compositional character of the text focussing on its repetitions, its gaps, its fragmented character and on what could be described as its mnemonic verses—a result of its recitational dynamics as a form of ākhyāna.

“The Dīpawaṃsa, the discussion of whose form I now enter upon, can hardly be called a production of artistic merit, in spite of its rather bombastic proem. It gives the impression, not of an evenly worked out whole, but rather of a stringing together of fragments, which are arranged in the manner I have just described. One finds therefore a clumsiness and an incorrectness of metre, and a number of other peculiarities which require to be specially referred to”.3

Thus, he seems to move from the broader fragmented textual character of the text as a composition into referencing its detailed textual and linguistic features: its ‘clumsiness’, bad metre and ‘other peculiarities’. We take it that here he assimilates the kinds of granular text editorial problems that Oldenberg is very concerned with in his Introduction to perceived infelicities in the composition in general, as if they were a single problem.

Writing in 1958, Malalasekera links the contents and structure of the text to its ‘problems’: “The Dīpavaṃsa represents the earliest attempt, so far as we know, to treat of these subjects in a compact, concise manner, forming one continuous story. It is a conglomeration of myths, legends, tales, and history, and the further we go back in time the more mythical it becomes, put together from various traditional sources, in an unaided struggle to create a composite whole from materials existing scattered in various places. This accounts for the outward form of the Dīpavaṃsa, its clumsiness and incorrectness of language and metre, its repetitions, omissions and general fragmentary character. And this very incompleteness of its composition, and its want of style help us in fixing the date of its compilation” (Malalasekera 1958, pp. 134–35).

Lamotte, who not surprisingly uses Dīp widely in his seminal History of Indian Buddhism, while commenting on Dīp’s dating of the Second Council, laconically remarks that this is “one passage which we would like to think of as corrupt” (Lamotte 1988, p. 272) Typically, he does not comment on the reliability of his emic historical sources, and we assume that this remark is his acknowledgement of the lack of reliability he felt regarding Dīp’s edited text.

Commenting on his own observation that Dīp is the earliest Pali text to be composed in Sri Lanka (above), Hinüber goes on, rather graciously given Oldenberg’s own comments on the text, to say “this is probably the reason for a certain awkwardness of formulation and style. Even the grammatical construction of quite a few verses is difficult, often due to the fact that parts of verses or formulas have been put together rather mechanically. … Any literary pretentions are missing” (Hinüber 1996, pp. 89–90).

We can thus see a trend in scholarly assessments, which brings concerns with the composition of Dīp that begin with Mahānāma, together with problems in the modern edited text—problems that arise from the poor quality of sources available to its editor. The politics of the day did not allow for Oldenberg to acquire more reliable Burmese sources, and since 1879, Western scholarship has taken Oldenberg’s somewhat faulty edited text as the source for access to what is also, in terms of broader compositional features, already a mildly problematic text. Mahānāma’s revised composition, Mahāvaṃsa, has therefore drawn far greater attention than Dīp in subsequent discussions of the historiography of Buddhism in Sri Lanka.

Most surprisingly, the appearance of a ‘Western’ edition in the Roman alphabet prompted printed editions in Theravāda countries, albeit only 50 years later. Ñāṇananda Thera’s Pali text and its paraphrase in Sinhala in 1927 and Thein and Hlaing’s Burmese edition of 1929 were both possible reflexes to Oldenberg’s edition—Ñāṇananda remarks on the errors in Oldenberg’s edition and provides its variants in the footnotes—but being printed in Sinhala and Burmese characters, respectively, neither have been used in Western scholarship, as far as we can tell (Ñāṇananda and Sīlananda 1927; Thein and Hlaing 1929).

3. The Present Project: A ‘New’ Recension, B1

The present authors began their engagement with the textual issues around this text in 2021 in the context of discussing Dīp’s problematic dating of the Second Council in Chapter 1. We have already mentioned Lamotte’s suggestion that the text is corrupted where it mentions this point. He does not enter into a discussion of the possible reasons for this, but from the review we have already given, it is clear that scholars have been inclined to think of Dīp as a problematic text in terms of its original composition. If we also take into account Oldenberg’s despairing account of the editorial challenges he faced and the likelihood that all his manuscripts were corrupted, we are presented with the second possibility, that Dīp is ‘corrupt’ for transmissional rather than compositional reasons. In other words, whatever the original text of Dīp may have said, the vicissitudes of multiple generations of copying have effectively damaged the text, and either some scribal error in the past or maybe insect or some other mechanical damage rendered the text incorrect or lacking where it mentions this topic.

It is worthwhile to review in more detail the problem touched upon here as it also acts as what we have been calling one of several ‘diagnostic criteria’ for differentiating recensions of the Dīp text—a topic to which we shall return later. The passage concerned occurs in the first chapter where the Buddha predicts the following:

anupādāparinibbāyi suriyo atthaṃgato yathā.

parinibbute catumāse hessati paṭhamasaṅgaho,

tato paraṃ vassasate vassān’ aṭṭhārasāni ca

tatiyo saṃgaho hoti pavattatthāya sāsanaṃ.

imasmiṃJambudīpamhi bhavissati mahīpati

mahāpuñño tejavanto Asokadhammo ’ti vissuto. 1.24–6

(Oldenberg 1879, p. 15).

… I shall reach complete Parinibbāna like the setting sun. Four months after my Parinibbāna the first convocation will be held …; a hundred and eighteen years later the third convocation will take place, for the sake of the propagation of the Faith. Then there will be a ruler over this Jambudīpa, a highly virtuous, glorious monarch known as Dhammāsoka (Oldenberg 1879, p. 119).

We give here Oldenberg’s edition and his translation. Note that ‘convocation’ is his term for what is nowadays widely known as a ‘Council’, and he refers to the convocations of the Saṅgha after the death of the Buddha. These events are widely known in Theravāda Buddhist historiography, and all other sources appear to agree that there were three such events within the first few centuries post parinibbāna. The first took place immediately after the Buddha’s death and cremation. There, his teaching was rehearsed and agreed upon by the monks attending, this being taken by tradition as the point of creation of the Tipiṭaka as a formal record of that teaching. By general agreement, this was followed one hundred years later by the Second Council, at which a major schism within the monastic community occurred. Finally, 118 years after the Second, the Third Council took place during the reign of Asoka, at which the Saṅgha was purified. Chronology is the issue here because the length of time between the Buddha’s parinibbāna and the Third Council is agreed to have been 218 years, i.e., the sum of the one hundred years from the First to the Second Council plus, from there, the 118 to the Third Council. Oldenberg uses the ellipsis in his translation to mark where he thinks reference to the Second Council is missing.

The Dīp text, as edited by Oldenberg, omits any reference to the Second Council at this point in the text, and it seems to locate the Third only 118 years post parinibbāna.4 This is the problem that Lamotte liked to think of, and therefore dismisses, as a corruption in the text—and there is nothing wrong with his judgement here. We can surmise the accidental omission of at least two half-lines, the equivalent of two pāda in terms of Pali prosody in the positions indicated below. This would be sufficient space for reference to the Second Council, and omissions of this type are very familiar to anyone who has worked with manuscripts—words, verses, lines, and even whole pages go missing all the time.

parinibbute catumāse hessati paṭhamasaṅgaho,

tato paraṃ vassasate ………………………………

…………………………………vassān’ aṭṭhārasāni ca

tatiyo saṅgaho hoti pavattatthāya sāsanaṃ.

It is entirely feasible, as Oldenberg suggests, that all his Sri Lankan manuscripts could ultimately have been copied from a single Burmese source in which the two pāda were missing due to some kind of accident of copying.

Kim’s interest in the dating problem led to a shared discussion between the present authors about the issues around it, and it was therefore of great interest to discover Tilman Frasch’s important article in which he reports his discovery of a manuscript of Dīp in his university’s library, the John Rylands Library, Manchester, in which this very passage seems ‘complete’.5 We eventually acquired high-resolution images6 of this manuscript and were able to see for ourselves what Frasch had mentioned in his article. The reading in this new manuscript for the same verses just quoted runs as follows (= diagnostic criterion 1, hereafter DC1):

parinibbute cātumāse paṭhamo hessati saṅgaho

tato paraṃ vassasate dutiyo hessati saṅgaho

tato dvisatavassāni vassānaṭṭhārasāni ca

tatiyo saṅgaho hoti pavattatthāya sāsanaṃ

It is immediately apparent that where one can reasonably deduce a gap in the printed text—the dotted lines inserted into Oldenberg’s edition above—here, we have a text that neatly both references the Second Council and corrects the faulty arithmetic of the text in the printed edition. The result can be translated as follows: “Four months after my final nibbāna there will be a First Council. One hundred years after there will be a Second Council, and two hundred and eighteen years after there will be a Third Council [held] for the purpose of [securing] the continuance of the religion”.

This type of loss to the text is commonly known in text editorial circles as a saut du même au même, meaning that the copyist’s eye jumps from the source manuscript down to his copy and then returns to the source, his eye erroneously catching a phrase or word there that is similar to, but not in the same place as, the one in the passage just copied—here, the phrases containing the word vassa.

The existence of this manuscript begged several questions which we set about trying to answer. Of course, such a discovery speaks to every textual scholar who yearns to find better manuscript sources, and this one seems to be offering a ‘correct’ text for this key, not to say infamous, problem passage. Could it be that the Manchester manuscript (ManDī, as we dubbed it) actually preserves a text of Dīp that is closer to the original composition and maybe resolves many other of Dīp’s ‘problems’? Does it offer a shortcut to bypass Oldenberg’s irresolvable editorial dilemma and take us to a ‘safer’ text, closer to the original redactor’s intentions than any manuscript from Sri Lanka and thus potentially rehabilitating Dīp from its errors and obscurities that had led to its relative neglect by modern scholarship in favour of Mahāvaṃsa?

In practice, these tempting fantasies are resolved into pragmatic questions concerning objective, measurable data. Is it an isolated unique manuscript? Is it simply the same old text with just this single passage ‘corrected’? Or is this a distinct recension of Dīp? When was it copied, by whom and where, etc.? The search for answers to these questions has resulted in an ongoing editorial project in which the present authors are editing this new recension and re-translating Dīp in light of this evidence.

The first of the answers to be determined concerned the date of the manuscript. It carries a colophon:

sakkarāj 1259 khu||tapui.tve la chan: 13 rak mū mra re: kū: rve||1898 khu||phe phau vā rī la 4 rak

“Copied from the source text on 13th day of the waxing fortnight of Tabodwe [the 11th month] in Sakkarāj era 1259, and on the 4th of the month of February, 1898”.

We thus know that this copy of the text of Dīp is from the late-19th century, but no information is given relating to its place of copying or who carried it out. The late date, postdating Oldenberg’s edition by 19 years, raised the rather deflating thought that this text could be a manuscript made in response to the Western printed edition, perhaps made by a scholar monk as a riposte to the work of the German editor. Finding a way to assess this possibility was therefore added to the tasks at hand.

Whether or not this was a unique manuscript was answered by the eventual identification, after extensive searching, of two further manuscripts of this same version.7 Their identity was determined by diagnostic criteria that we developed as we read its text in relation to Oldenberg’s edition—more on this in the next section. These confirmed that this was not a single text—perhaps an autograph manuscript of a monk who wanted to correct Oldenberg’s errors—but instead a version that circulated in multiple copies. We concluded from this, as much as from the fact of its divergence from Oldenberg’s text, that we were now dealing with a distinct recension of Dīp that deserved to be edited and translated in its own right. The third factor that confirmed its status as a distinct recension is the discovery of the commentary, the Sāsanajotikā “Illuminating the Religion”, which is a commentary on this ‘new’ recension and not that of Oldenberg.8 We should add at this point that we have also compared the text of B1 with the recension of Dīp quoted at some length by Buddhaghosa and have concluded that there is no evidence that it is related to the version he knew.9

These new manuscripts also helped establish the relationship of this text to Oldenberg’s edition. The second manuscript of this recension that we acquired came from the Myanmar National Library, but it had no colophon.10 The third we discovered, collected by the Fragile Palm Leaves Project, has a colophon with a date of 1877, i.e., two years prior to the publication of Oldenberg’s edition. This new recension could therefore not be a response to his work. It predates the Western edition. That all its manuscripts are Burmese indicates that it circulated there, and at the current time, we know of no manuscript of it from outside Myanmar. This raises another question—is this recension in fact a Burmese artefact? In terms of logical possibility, it could be an extraordinary survival of the ‘original’ Dīp that happens to have been transmitted in Myanmar, or it could be a Burmese text created or ‘revised’ there from faulty sources not dissimilar to those used by Oldenberg but prior to his work. These are questions to which we hope to provide the answers as our project proceeds.

4. The New Recension

We have already mentioned the diagnostic criteria used by us to identify manuscripts of this new recension. Our discussion above concerning the Second Council demonstrates the first of three major recensional divergences that characterise the text of B1, the newly discovered recension, in Chapter 1 of Dīp. Below we give two other major passages of divergence, but the reader should understand that, in addition to these macro-structural differences, there are numerous micro-structural divergences—variations in words or word order, etc.—that separate B1 from Oldenberg’s text. These minor differences are too numerous to record here and belong in the critical edition of the text under preparation. Some of this minor material is shared between B1 and B2, the broader Burmese manuscript record, and separating B1’s text from B2 and from Oldenberg’s is a major task of the project in which we are engaged.

The two passages that follow come from the middle of the chapter and from its end. We provide the text in parallel with Oldenberg’s edited text plus his translation. The translation of B1 is only provisional, not least because the Pāli text of B1 is also only provisionally edited here.

From the middle of Chapter 1, is found diagnostic criterion 2 (DC2; See

Table 1. DC 2.

B1—The New Recension	Oldenberg’s Text and Translation
54 gaṅgāya nadīyātīre yo padeso patiṭhito mahithaṅgaṇa thūpassa patiṭhāne bhūmaṅgaṇe 55 tasmiṃ padesasmiṃ ṭhito naruttamo samappito jhānasamādhim uttamaṃ jhānaṃ lahuṃ khippanisantikāro muni samāpajjati, cittakkhaṇe On a bank of the river, in the place where the lord stood, in the open area in which was established the Mahithangana Thupa, in that place, the best of men remained. He established the best Concentration (Samadhi) in Absorption (Jhana). The sage, who applies his attention quickly, speedily attained concentration in one mind-moment.	52-ab gaṅgātīre Mahiyāsu pokkhalesu patiṭṭhite thūpaṭṭhāne Subhaṅgaṇe 52-cd; 53ab tasmiṃ padesasmi ṭhito naruttamo samappito jhānasamādhim uttamaṃ. jhānaṃ lahuṃ khippanisantikāro muni samāpajjati cittakkhaṇe, 52. On the bank of the river, near Mahiya Pokkhala, on the site of the Subhaṅgana Thūpa, there the highest of men stood, and entered upon the highest ecstatic meditation. 53. The Sage, the awakener of quick attention, speedily entered upon that meditation (by revolving) in a moment by one thought (the whole system of qualities).
56 sahasā vata tuṭṭhāya samāpajjitajhāniko cittesu issaro buddho iddhihi pāramīgato For sure, he had achieved absorption as quickly as he pleased. In his stream of consciousness, the lord Buddha had reached perfection through supernatural powers.	53-cd; 54ab sahasā tam uṭṭhāti jhānakhaṇiyā samāpayi sucittehi pāramīgato. ṭhito naro iddhi vikubbamāno yakkho va mahiddhi mahānubhāvo, Suddenly he thence rose; he who had reached (all) perfections by his virtuous resolutions, … finished his meditation.
57 ghanaṃ meghaṃ pavassento sītalavātaduddinaṃ vuṭṭhivātandhakārehi tesaṃ saṃvejanaṃ akā Making a great cloud to rain with cold wind and thunder, he made them anxious with the rain and wind and darkness.	54cd khaṇiyaṃ ghanā meghasahassadhārā pavassati sītalavātaduddini. 54. There the hero stood, performing miracles by his (magical) power, like a Yakkha of high (magical) power and great (supernatural) faculties; gathering (?) thick clouds, containing thousands of rain drops, he sent rain, cold winds, and darkness.
58 sitaṭṭhitā vate yakkhā ayāciṃsu tapaṃ jinaṃ uṇhakaro jino āha yakkhe te tisitaṭṭhite The goblins were stuck standing in the wind. They begged the conqueror for heat. The conqueror who can make heat said to those goblins who were stuck standing in the three [rain, wind and darkness.]…	[no counterpart]

): a sequence of five verses, constituting vv. 54–58 in B1, partially parallels three verses in Oldenberg’s text, vv. 52–54 (Oldenberg 1879, p. 17).

And from the end of the chapter, diagnostic criterion 3 (DC3; See

Table 2. DC 3.

“Mandi”	Oldenberg
70 tesaṃ nivāsanaṭṭānaṃ vicintento ca addasa pubbadakkhiṇadisāya laṅkā nāmaka dīpato And thinking of where they might dwell, he saw to the southeast from the isle called Laṅkā, 71 giriṃ nāma ṭhitaṃ dipaṃ nadīta lāka pabbataṃ sita-salila-sampannaṃ vitapaṅka-jalāsayaṃ the isle called Giri, which has rivers, lakes and mountains, which has cool [weather] and water, and clean lakes.	67 ath’ aññadīpaṃ paṭirūpakaṃ imaṃ ninnaṃ thalaṃ sabbathānekasādisaṃ nadīpabbatatalākasunimmalaṃ dīpaṃ Giriṃ Laṅkātalasamūpamaṃ (He) then (thought of) another island, similar to this, with low ground and high ground, with many various aspects, beautifully adorned by rivers, mountains, and lakes, the island of Giri, most similar to the country of Lanka.
72 vanaggahanasambuṇṇaṃ kusumehi suphullitaṃ ucchu-kadali-sampannaṃ panasa-mātuluṅgikaṃ Which has groves and wildernesses, is full of flower blossoms, which has sugar-cane and bananas, jack fruit and citron. 73 situṇautusamattaṃ sabba-bhaya-sunibbhayaṃ pugatampulasampuṇṇaṃ gopitasāgarantakaṃ Where the cool and warm seasons are similar, where all the fears completely disappeared, which has many cows and wild animals, which is protected and surrounded by the ocean. 74 tālamba-nāḷikerehi phalarukkhehi chāditaṃ haritasaddalamahiṃ satātirekayojanaṃ Which is covered by palm trees, mango trees, coconut trees and fruit trees, where the earth is covered with green grass, and its size is more than 100 yojanas. 75 subhikkhaṃ bahutabbhakkhaṃ pubbaṇṇehi sayaṃjāti tatheva aparaṇṇehi laṅkādīpassa vitthāraṃ Where there is a good harvest, where there is plenty of food, where grain crops grow on their own, similarly, the vegetable crops [grow on their own,], a place about the size of Isle of Lanka. 76 yojanānaṃ sahassamhi matte ito patiṭṭhitaṃ yakkhānam atisappāyaṃ ārāma-rāmaṇeyyakaṃ [The place] is located about 1000 yojana from here. It is delightful like a park and exceedingly suitable for Yakkhas.	68 sunibbhayaṃ gopitasāgarantakaṃ pahūtabhakkhaṃ bahudhaññamākulaṃ utusamatthaṃ harisaddalaṃ mahiṃ varaṃ Giridīpam imassa uttariṃ 68. (It was) free from danger, well protected, surrounded by the ocean, full of excellent food and rich grain, with a well tempered climate, a green, grassy land, the beautiful island of Giri, superior to this (island). 69 rammaṃ manuññaṃ haritaṃ susītalaṃ ārāmavanarāmaṇeyyakaṃ varaṃ, santīdha phullaphaladhārino dumā, suññaṃ vivittaṃ, na ca koci issaro, 69. It was charming and delightful, green and cool, adorned by gardens and forests, exquisite; there were trees, full of blossoms and fruits; it was empty and solitary, subject to no master. 70 mahaṇṇave sāgaravārimajjhe sugambhīre ūmi sadā pabhijjare, suduggame pabbatajālamussite sudukkaraṃ attha aniṭṭhamantaraṃ. 70. (It was situated) in the great sea, in the midst of the ocean and of the deep waters, where the waves incessant! y break; around it there was a chain of mountains, towering, difficult to pass; to enter it against the wish (of the inhabitants) was difficult.
77 disvā taṃ amanūssānaṃ suphāsukaṃ yato tato dadāmahaṃ imaṃ tesaṃ dīpaṃ gīri ti nāmakaṃ Because after seeing it that it was very pleasant for the non-humans, therefore [the Buddha said] “I will give this island called Giri to them”.	71 paravānarosā parapiṭṭhimaṃsikā akāruṇikā paraheṭhane ratā caṇḍā ca ruddhā rabhasā ca niddayā vidappanikā sapathe idha ime. 71. Full of desire and anger towards other beings, backbiting, pitiless, given to injuring other beings, cruel and furious, violent, merciless,… 72 atha rakkhasā yakkhagaṇā ca duṭṭhā dīpaṃ imaṃ Laṅkāciranivāsitaṃ dadāmi sabbaṃ Giridīpaporāṇaṃ, vasantu sabbe supajā anīghā. 72. (Buddha thus spoke:) Ye Rakkhasas and ye wicked hosts of Yakkhas, I give unto you this island which is not far from Lanka, the whole old island of Giri; may they all inhabit it and multiply undisturbed.
80 giridīpo va yakkhānaṃ phāsuko pitivaḍḍhano laṅkādīpo manussānaṃ sappāyo sukhavaḍḍhano Indeed, the isle of Giri is comfortable for the Yakkhas and increases their delight. The isle of Lanka is suitable for human beings and increases their happiness. 81 giridīpo va yakkhānaṃ atthahito bhavissati laṅkādīpo manussānaṃ taṃ dīpaṃ parivattayi Indeed, the isle of Giri for the Yakkhas and the isle of Lanka for the human beings will be useful and beneficial. [He] exchanged that island for the human beings.	75 dīpaṃ ubho mānusā rakkhasā ca ubho ubhinnaṃ tulayaṃ sukhaṃ muni bhiyyo sukhaṃ lokavidū ubhinnaṃ parivattayi goṇayugaṃ va phāsukaṃ. 75. Weighing the prosperity and the high happiness of the two, the Sage who knew all worlds, interchanged the two islands and the two (kinds of beings), men and Rakkhasas, (as a peasant) easily (interchanges) his pairs of bullocks.

) can be found: the following passage of ten verses, 70–81 in B1, parallels a sequence of seven verses in Oldenberg, vv. 67–72 plus 75 (Oldenberg 1879, pp. 19–20).

As we can see from these data, the new recension B1 not only transmits variant readings not recorded here but, as seen here, adds new verses—its first chapter is six verses longer than Oldenberg’s text—and in addition does not record verses that are in Oldenberg’s text. While the verses from B1 not extracted above can be taken to more or less conform to the text we know from Oldenberg, the extent of clearly intentional variation unquestionably qualifies B1 as a distinct recension.

At the same time, reading other Burmese manuscripts shows that there are a substantial number of those that are closer to Oldenberg’s text than our B1 in the sense that they do not share these diagnostic criteria with B1. Provisional work on Burmese manuscripts of Dīp from the British Library and the Fragile Palm Leaves collection shows that these other Burmese manuscripts can themselves be subdivided according to their distinctive readings and errors that differ both from Oldenberg’s text and from each other. The Burmese transmission of this text is therefore complex and potentially contaminated in terms of possible borrowing between transmissional lineages. B1 is something of a numerical outlier even within the Burmese tradition—its manuscripts are fewer in number, and we can see an overall agreement, for example, with respect to DC1 (above) between the majority of Burmese manuscripts and Oldenberg’s text. Careful transmissional analysis is therefore needed to unpack this complex story, both to establish how B1 stands in comparison to Oldenberg and how it shares material with B2 as part of the broader Burmese tradition, separately from its identity as a distinct recension.

The second aspect of Dīp that we can use to differentiate the recension B1 from O is macro-structural—by this, we refer to the chapter divisions of the text as a whole. Oldenberg’s edition is set out in 22 divisions, numbered as chapters, although most are only marked with bhāṇavāra divisions rather than chapter titles. We give here the titles in Pali along with translations of them and, in square brackets, the titles in Ānandajoti’s version of Oldenberg’s text where they differ. The list is as follows:

• bhāṇavāraṃ paṭhamaṃ & Yakkhadamana

  -

  The Taming of the Yakkhas [Buddha’s subjection of the Yakkhas].

• Nāgadamanaṃ & bhāṇavāraṃ dutiyaṃ

  -

  The Taming of the Nāgas & second recitation section [The Conquering of the Nāgas].

• Mahārājavaṃso & bhāṇavāraṃ tatiyaṃ

  -

  The Great Lineage of Kings & third recitation section.

• Mahākassapasaṅgaha & Dutiyasaṅgaha & bhāṇavāraṃ catutthaṃ

  -

  Kassapa the Great’s Council, the Second Council & fourth recitation section [The First Two Councils].

• ācariyavāda & bhāṇavāraṃ pañcamaṃ

  -

  The Schools of the Teachers & fifth recitation section.

• bhāṇavāraṃ chaṭṭhaṃ

  -

  sixth recitation section [Asoka’s Conversion].

• saddhammasaṅgahaṃ navamāsaṃ & bhāṇavāraṃ sattamaṃ

  -

  The nine-month Council of the True Faith & seventh recitation section.

• bhāṇavāraṃ aṭṭhamaṃ

  -

  eighth recitation section [The Missions].

• bhāṇavāraṃ navamaṃ

  -

  ninth recitation section [Vijaya’s Story].

• bhāṇavāraṃ dasamaṃ

  -

  tenth recitation section [Paṇḍuvāsa].

• rājābhisekabhaṇḍaṃ & bhāṇavāraṃ ekādasamaṃ

  -

  Accoutrements to the Coronation & eleventh recitation section [Devānam piyatissa].

• bhāṇavāraṃ dvādasamaṃ

  -

  twelfth recitation section [The Coming of Mahinda].

• bhāṇavāraṃ terasamaṃ

  -

  thirteenth recitation section [The Earthquakes].

• Mahāvihārapaṭiggahaṇa & Cetiyapabbatapaṭiggahaṇa & bhāṇavāraṃ cuddasamaṃ

  -

  Acceptance of the Mahāvihāra and the Cetiyapabbata & fourteenth recitation section [Mahāvihāra and the Cetiyapabbata].

• bhāṇavāraṃ paṇṇarasamaṃ

  -

  fifteenth recitation section [The Relics, the Buddhas, and Queen Anulā].

• bhāṇavāraṃ soḷasamaṃ

  -

  sixteenth recitation section [The Bodhi Tree].

• bhāṇavāraṃ sattarasamaṃ

  -

  seventeenth recitation section [The Passing of a Generation].

• bhāṇavāraṃ aṭṭārasamaṃ & mahāvāraṃ

  -

  eighteenth recitation section & The Great Protection [The Bhikkhuni Lineage].

• bhāṇavāraṃ ekūnavīsatimaṃ

  -

  nineteenth recitation section [Duṭṭhagāmani].

• bhāṇavāraṃ vīsatimaṃ

  -

  twentieth recitation section [Tissa to Kuṭikaṇṇatissa].

• bhāṇavāraṃ ekavīsatimaṃ

  -

  twenty-first recitation section [Abhaya to Subha].

• dīpavaṃsaṃ niṭṭhitaṃ

  -

  twenty-second recitation section [Vasabha to Mahāsena].

Here, the ‘chapter’ colophons are given, followed by an English translation. Note that the only titles transmitted in the text itself are for Chapters 1, 2, 3, 4, 5, 7, 11 and 14—i.e., 8 out of 22. The remaining English titles, in square brackets, are inserted by Ānandajoti in his digital transcription of Oldenberg’s edition and are therefore not historically authentic to the text, except where the title is present within the text of the chapter.11 The use of the division bhāṇavāra is odd since the term originates as an indicator for a long passage of text—“a section of the Scriptures, divided into such for purposes of recitation, “a recital””. (PED sv. bhāṇa). However, some of these sections are extremely short—e.g., O’s Chapter 8 is only 13 verses in length but is marked as a bhāṇavāra—and so we must assume the term is being used in O’s manuscript sources in a different sense, although what that might be is not clear at the moment. Also, 5 of these chapters or sections—2, 3, 4, 5 and 11—have one or another of these colophons as internal subdivisions, but somehow or another, these do not qualify as ‘chapter’ endings. Why Oldenberg subdivided the text in the way that he did—treating bhāṇavāra as chapter divisions but ignoring occasional title colophons within those sections—is unknown to the present authors. All this said, Dīp has been seen as a 22-chapter text since 1879.

If we look at the B1 recension, the situation is rather different. An initial analysis suggests that this recension has 14 or 15 chapters. In the following table (See

Table 3. Chapter headings in manuscripts of the newly discovered recension, B1.

	Manchester	Myanmar National Library
1	yakkhaniddhamaṃ	yakkhaniddhamaṃ
2	nāgamadanaṃ	nāgamadanaṃ
3	maṇiakkhīkanimantanāgamanaṃ	maṇiakkhīkanimantanāgamanaṃ
4	mahākassapasaṅgahaṃ	mahārājavaṃso
5	dutiyasaṅgaha	bhāṇavāraṃ paṭhamaṃ
6	ācariyavādaṃ	dutiyasaṅgahaṃ
7	tatiyasaṅgahaṃ	ācariyavādaṃ
8	rājābhisekakhaṇḍaṃ	tatiyasaṅgahaṃ
9	bhāṇavāraṃ tatiyaṃ	rājabhisekakhaṇḍhaṃ
10	mahāvihārapaṭiggahaṇaṃ	bhāṇavāraṃ tatiyaṃ
11	cetiyapabbatapaṭigahaṇaṃ	mahāvihārapaṭiggahaṇaṃ
12	bhāṇavāraṃ catutthaṃ	cetiyapabbatapaṭiggahaṇaṃ
13	bhāṇavāraṃ pañcamaṃ	bhāṇavāraṃ catutthaṃ
14	niṭṭhito dīpavaṃso	bhāṇavāraṃ pañcamaṃ
15		niṭṭhito dīpavaṃso

), we give the chapter headings from the two complete manuscripts of this recension:

We should note that the chapters with true titles mainly coincide with the titles in O, but the proliferation of ‘false’ bhāṇavāra does not happen. Instead, the whole text is distributed across five ‘recitals’, which seems consistent in terms of volume per unit with the usage of this term elsewhere in the canon. Similarly, the chapter divisions employed in the Sāsanajotikā map well onto these titles, providing further evidence that it is a commentary on this recension (See

Table 4. Chapter divisions of the commentary on B1.

	Sāsanajotikā
1	paṭhamagāthāvaṇṇanā
2	dutiyagāthāvaṇṇanā
3	tatiyagāthāvaṇṇanā
4	yakkhaniddhamanakhaṇḍavaṇṇanā
5	nāgamadanavaṇṇanā
6	nāgarājānimantanāgamanavaṇṇanā
7	paṭhamabhāṇavāravaṇṇanā
8	mahākassapasaṅgahavaṇṇanā
9	ācariyavādavaṇṇanā
10	dutiyabhāṇavāravaṇṇanā
11	moggaliputtatissasaṅgahavaṇṇanā
12	tatiyabhāṇavāravaṇṇanā
13	mahāvihārapaṭiggahaṇavaṇṇanā
14	cetiyapabbatapaṭiggahaṇavaṇṇanā
15	catutthabhāṇavāravaṇṇanā
16	dhātu-āgamanavaṇṇanā
17	bodhī-āgamanavaṇṇanā
18	pañcamabhāṇavāravaṇṇanā
19	vaṃsapakāsakavaṇṇanā

).

That these internal structural divisions are distinctive for B1 has been provisionally confirmed by comparison with a Burmese representative of B2, which appears to follow the divisions of O. In other words, B1 differs not only from O but also from other recensions of the Burmese tradition. A full study of these chapter variations must wait for another occasion.

5. Conclusions: The Burmese Tradition

The first step in examining this new recension began with Frasch’s article. In it, he discusses a minor publication of notes on Dīp, written by Pe Maung Tin, which Frasch had picked up from a second-hand bookstall and in which he realised Maung Tin was himself commenting on a recension different from the published text of Dīp (Frasch 2004, pp. 70–73). We now know that there are indeed different ‘versions’ of Dīp transmitted in Myanmar, of which our new recension is one but of which there may be a further two. To date, we have been able to locate 18 Burmese manuscripts of Dīp not used by Oldenberg for his edition. Of these, there are two that we have as yet been unable to consult (University Colleges Library, Yangon). We have accessed a single manuscript of the Sāsanajotikā, a commentary on this recension from the U Pho Thi Library, and we know of one further copy of it, again not yet accessible to us. From searches across all catalogues of Pali manuscripts that we could find, it is abundantly clear that the text of Dīp is mainly preserved in Burmese manuscripts. There are Thai manuscripts of Dīp in the Thai National Library collection (not yet accessed), but otherwise, the text seems not to have been popular there, plus we have a single Tham Lao manuscript.

Leaving aside the dozen or so manuscripts from Sri Lanka that Oldenberg consulted, which he judged to be all derived from a poor Burmese exemplar, we might not be far from the truth in suggesting that Dīp was, in general, insufficiently popular for it to have been copied very much at all throughout the Theravāda world in the early modern period, except in Myanmar where it obviously inspired a modest degree of copyist activity. It also inspired a commentary, on the B1 recension, made by the Burmese author Jāgara in the 19th century.12 Jāgara was one of the learned scholars tasked with the organisation of the Burmese Fifth Council, although we cannot at present determine whether his interest in Dīp was connected to his work on the Council. The existence of just three manuscripts of B1, suggesting its limited diffusion, one predating Oldenberg but postdating (we cannot tell for the undated copy) the Fifth Council, gives rise to the speculative thought that the B1 recension could be a revision of the text made for that council, possibly even by Jāgara himself, in the course of which he also compiled his commentary. But this is just speculation, and we mention it here for the sake of covering the logical possibilities—there is no evidence, other than the simple existence of Sāsanajotikā, for such a connection. It is possible that, of the manuscripts not yet consulted, there may be evidence that B1 predated that council.

A detailed discussion of the Sāsanajotikā must be left for another occasion, as must a more thorough analysis of the differences between B1, our so-called new recension, and the others transmitted within Myanmar. The result of such a more thorough analysis should be a ‘better’ text of Dīp. But in performing such a task, we must also be methodologically cautious, as was Oldenberg. This means, in our case, to establish clearly the outlines of all distinct recensions of Dīp, however many there may be—and we should note the evidence from Buddhaghosa that he too had a distinctive recension—and to keep these distinct recensions apart, rather than conflate them. This will then allow us to read real, historical texts that were known to living Buddhist communities. That these differed is just a fact of history, and to appreciate this truly, we must relinquish our desire for there to be one single authoritative text of Dīp. The somewhat problematic text of Dīp that has circulated in Sri Lanka during the early modern period has been preserved and disseminated by Oldenberg’s printed text for 146 years and, despite being judged inadequate by its own progenitor, has lived a strange life as ‘the’ Dīpavaṃsa for many generations of scholars. We are now in a position to evaluate the full transmissional state(s) of Dīp, as represented by as many manuscript sources as possible—certainly more than were available to Oldenberg for his pioneering edition of 1879. That the most important voices telling us of other incarnations of Dīp come from Myanmar rather than Sri Lanka is an irony that doubtlessly can be explained by the comparative histories of the two nations and their roles in the preservation of Theravāda itself.