Texts of the early Buddhist tradition are preserved in Pāḷi, Gāndhārī, and various other forms of Sanskritised Prakrit; This term seems to me to be preferable to the usual Buddhist Hybrid Sanskrit or Mixed Sanskrit, which makes it sound as if we are dealing with forms of Sanskrit, whereas the languages underlying all the early texts, including Pāḷi, are forms of Prakrit that have been Sanskritised to a greater or lesser degree.01 besides being found in Chinese and Tibetan translation. Of the MIA versions the only one that has come down to us in anything like a complete recension is preserved in the Pāḷi Tipiṭaka, The Tipiṭaka, of course, contains texts of various ages, but undoubtedly in the four main nikāyas it preserves a fairly reliable recension of the original teaching.02 and it is certainly the Pāḷi texts, mainly in modern translations, that constitute the best known version of the early teachings, which is not surprising as they belong to a living Buddhist tradition.

Of the Pāḷi texts the Dhammapada is perhaps the best-loved collection of the Buddha's teachings. There have been many editions, The main editions in Roman script were made by V. Fausböll (1855, 2nd ed. London, 1900); S. Sumangala Thera (London, 1914); D.J. Kalupahana (Lanham, 1986); J.R. Carter & M. Palihawadana (Oxford, 1987); O. von Hinüber & K.R. Norman (Oxford, 1994). I also made a New Edition myself in 2002, which is the text used here.03 and almost innumerable translations of this ever-popular text in nearly all modern languages. For a survey of the Dhammapada translations up to 1989, see Russel Webb: The Dhammapada - East and West (Buddhist Studies Review 6.2 1989: 166-175).04 And in countries that have a Theravāda tradition there is a copy of the book in most Buddhist homes, and many people know at least some of the verses by heart.

Apart from the Pāḷi Dhammapada, however, there are comparable collections of the Dharmapadāni verses available in complete, or very nearly complete, editions in three other MIA recensions, and these parallels can often throw light on the early teachings, and act as a complement, and sometimes also as a corrective, to the Pāḷi verses. There are also parallels to individual verses found in other texts belonging to the early Buddhist traditions of other schools. According to Bhikkhu Kuala Lumpur Dhammajoti, in The Chinese Version of Dharmapada (Colombo, 1995, p. 26), there are "2 versions of the Dharmapada and 2 versions of the Udānavarga in Chinese ... and two or three versions of the Udānavarga in Tibetan".

Moreover, there are parallels in the Jaina and Brahmanical traditions. Jaina Parallels have been enumerated by W.B. Bollée, in his Reverse Index of the Dhammapada, Suttanipāta, Thera- and Therīgāthā Pādas with Parallels from the Āyāraṅga, Sūyagaḍa, Uttarajjhāyā, Dasaveyāliya and Isibhāsiyāiṁ (Reinbek, 1983).

The Brahmanical parallels have been collected by W. Rau in his essay: "Bermerkungen und nicht-buddhistische Sanskrit-Parallelen zum Pāli-Dhammapada", which was published in Jñānamuktāvalī. Commemoration Volume in Honour of Johannes Nobel... edited by Claus Vogel (New Delhi, 1959).

None of these, however, have been dealt with here, as we are concerned in this work with the relationship of the recensions of the verses in the various collections made in MIA.

Of the collections, the closest to the Pāḷi is what is now known as the Patna Dharmapada. There are 4 editions of this text at present. The first, made by N. S. Shukla under the name The Buddhist Hybrid Sanskrit Dharmapada, was published in Patna itself in 1979; a much more reliable version of the text was made by G. Roth, and published as a part of The Language of the Earliest Buddhist Tradition, ed. by Prof. Heinz Bechert (Göttingen, 1980); the third was made by Margaret Cone as part of her doctrinal thesis, and published in the Journal of the Pali Text Society, Volume XIII (Oxford, 1989); the fourth was made by K. Mizuno in A Study of the Buddhist Sanskrit Dharmapada in Buddhist Studies Vol. 11 & 19 (Hamamatsu, 1982, 1990).06 All the editions are based on a manuscript found in a Tibetan Monastery by Rāhula Saṁktyāyana some time in the 1930s. Exactly when is not clear, but it was probably during the trip to Tibet in 1934 or 1936.07 The photographs of this manuscript are now held in the K.P. Jayaswal Research Institute in Patna, which is how the conventional name for the text has arisen.

The language of the verses that have been collected in the Patna Dharmapada is very close to the Pāḷi version of the text. The morphology is slightly more Sanskritised, but anyone who can read Pāḷi, and has even a passing acquaintance with Sanskrit, should be able to read the text.

The same cannot be said of the Gāndhārī Dharmapada, the scribe of which did not regularly distinguish between short and long vowels; for the most part he didn't mark the difference between assimilated conjuncts and the simplex, or preserve niggahīta (anusvara) either; also the phonetic values of the text are considerably different to the other recensions. It is a desideratum that some attempt be made to restore the phonetic values of the text, so as to clarify the true nature of the language underlying the written remains.

The basis for this text is a set of manuscript remains that were found in the Gandhāra region in what is now Xinjiang Uyghur Autonomous Region of China in the late 19th century. There were initially a number of partial publications of this text which it appears had been broken into three pieces and sold off to various European explorers. One part found its way to France, another to Russia, and a third part, it appears, has been irretrievably lost. A transcription of the material in France was published by E. Senart under the title Le manscrit kharoṣṭhī du Dhammapada: les fragments Dutreuil de Rhins (Journal Asiatique, 1898); a 2nd edition of this material was made by B. Barua and S. Mitra in 1921, under the title Prakrit Dhammapada (University of Calcutta Press); a 3rd edition of the same material was made in 1945 by Prof H.W. Bailey under the title The Khotan Dharmapada (Bulletin of the School of Oriental and African Studies, volume xi, London). But it was not until Prof. John Brough's edition, entitled The Gāndhārī Dharmapada that all the remaining parts of the text were collated and published in London in 1962 (School of Oriental and African Studies, London; reprinted by Motilal Banarsidass in Delhi, 2001).08 According to Prof. Brough's estimate there must have been approximately 540 verses in the text in total, but the verses in Brough's edition only amount to 342, and many of those are very fragmentary.

The third major parallel to the text is not called a Dhammapada, but is known as the Udānavarga. However, it is clearly a recension of the same Dharmapadāni material, even if the collection has been greatly expanded. The Sanskrit remains of the text are fragmentary, but owing to the fact that there are many fragments to compare, it has been possible to restore the verses to a much greater degree than with the Gāndhārī. Chapter xxix was published by R. Pischel in 1908; part of chapter xxii, and chapters xxix - xxxii were published by L. de la Vallée Poussin in 1912; N.P. Chakravarti in Paris in 1930 published chapters i - iii; v - xxi; and B. Pauly published chapter xxxiii in 1961; but the first complete edition of the remains of the Sanskrit text were published by Dr. Franz Bernhard in 1965 in Gottingen.09

The text is evidently much more Sanskritised than the other versions we have discussed so far, but the degree of Sanskritisation is not standardised throughout and the text contains something of a mixture in terms of its language. The metre of the text is also a mixture of early and late forms, as I have shown in a separate study. See Udanavarga for two studies of the Udānavarga relating to its prosody and to the Sanskritisation of the text.10

Of the incomplete parallels, two chapters from yet another Dharmapada have been preserved in the Mahāvastu, one of the earliest of the Sanskritised Prakrit texts; one of the chapters is named as the Sahasravarga, and appears to be the whole of the chapter; the other is a selection that comes from an unnamed Bhikṣuvarga. Parallels also exist in the Divyāvadāna edited by E.B. Cowell & R.A. Neil (Cambridge 1886); the Gilgit Manuscripts edited by Prof. Nalinaksha Dutt (Calcutta, 1950); the Avadāna-śataka, edited by J.S. Speyer (St. Petersburg, 1902, 1906); and the Mahā-karmavibhaṅga edited by Sylvain Lévi (Paris, 1932).

In preparing this Comparative Edition of the Dhammapada I have used the Pāḷi Dhammapada as the basis, and collected the rest of the material around it. This should not, however, be taken as implying that the Pāḷi is the standard from which the other versions have more or less departed; nor should it be thought to imply that the Pāḷi edition is the original, or even the earliest version. Even if we could determine with certainty the relative ages of the various recensions of the Dharmapadāni material, that would tell us almost nothing about the genuineness of the verses contained in them. It may be that a verse that has been heavily Sanskritised reflects an authentic saying of the Buddha; and another verse in an early form of Prākrit is intrusive. As all the recensions are in fact collections of already extant verses we can be quite sure they contain material of various ages.11

* * *

In re-presenting this material my primary aim has been to present matter that actually throws substantial light on the verses in question. Because of this I have not presented every fragment that could possibly have been parallel to each of the verses, considering that it actually adds little to our understanding of the verses, but have normally only admitted material when it is at least parallel to a quarter verse. What counts as a parallel is often a purely subjective judgement, and no doubt others may disagree occasionally with what has been included and what excluded; however, there seems to be no way to avoid this. Similarly, given the great mass of material at hand, there may be some inconsistencies in the presentation. For these failings I can only beg the reader's indulgence.12

Similarly, I have not reproduced every parallel to every verse, which would entail an enormous amount of redundancy, but have reproduced the parallels at the place where they are closest to the Pāḷi verse, so that if there is a parallel that is equal to all four lines of one of the Pāḷi verses, and later is found equal to three lines of another verse, it is only reproduced on the first occasion. This was an essential policy to adopt, as there are many series of verses in all the editions that simply extend the number of verses by substituting keywords. If every parallel line had been printed everywhere it occurs, the situation would have been confused, I feel, rather than clarified.13

Presenting the material in this way should throw much light on the Pāḷi verses, and the relationship they have to the other versions. However, it does not reveal the whole picture, as the other texts sometimes have many verses that could be regarded as parallel to one of the Pāḷi verses, owing to repetition of a verse with the replacement of keywords. This applies especially to the Udānavarga.

To help rectify this situation, in the Appendix some important verses that have found their way into the parallel versions, but are missing in the Pāḷi recension, have been gathered together, and are presented from four different angles. The first follows the parallels to the Pāḷi with their variants, then the Patna, Gāndhārī, and Udānavarga texts with their parallels in turn. This shows both the texts and parallels to the verses in all the editions, and also reveals how the verses have been organised in the various versions. Time permitting, I hope to produce a similar collection using the Udānavarga as the basis, which has the advantage of being the largest collection of verses, and therefore offering the largest amount of material to compare.14


Texts employed in this Edition

The Dhammapada, A New Edition edited by Ānandajoti Bhikkhu (Ancient Buddhist Texts website, 2002).

The text of the Dhammapada in this new edition has been established through a comparison of the Sinhalese, Burmese, Thai, and European editions.

Changes in presentation made in this edition:

Patna Dharmapada edited by Margaret Cone (JPTS Vol XIII, Oxford 1989) (reproduced by permission of PTS).

I am grateful to Prof. K.R. Norman, who sent a few small corrections to be made to the printed text. These include the accidental omission of a half-verse (145cd), and a repetition of a quarter-verse (149b).

Changes in presentation made in this edition:

The Gāndhārī Dharmapada edited by John Brough (Motilal Banarsidass, Delhi, 2001) (reproduced by permission of Motilal Banarsidass).

I am very grateful to the Early Buddhist Manuscript Project at the University of Washington, and particularly to Dr. Andrew Glass who prepared the database, for making the text available to me. This re-presentation of the text was made with the help of Dr. Mark Allon.

Changes in presentation made in this edition:

In the original text of the Gāndhārī Dharmapada as printed by Brough certain conventions were employed which have had to be dropped here. These include:

Also note that:

Udānavarga edited by Franz Bernhard (Vandenhoeck + Ruprecht, Göttingen, 1965) (reproduced by permission of the Akademie der Wissenschaften zu Göttingen, and Mrs. Sobotzik the editor's sister and literary heir).

Changes in presentation:

Divyāvadāna edited by E.B. Cowell & R.A. Neil (Cambridge 1886). The transliteration of this text has been modified to accord with the other texts presented here.

Excerpts from the following have been transliterated by the present writer: 

Mahāvastu, edited by E. Senart (Paris, 1882, 1890, 1897).

Gilgit Manuscripts, edited by Prof. Nalinaksha Dutt (Calcutta, 1950).

Avadāna-śataka, edited by J.S. Speyer (St. Petersburg, 1902, 1906).

Mahā-karmavibhaṅga, edited by Sylvain Lévi (Paris, 1932).

The Arthaviniścaya Sūtra & its Commentary (Nibandhana), edited by N. H. Samtani (Patna, 1971)

Jātakamālā, edited P.L. Vaidya (Darbhanga, 1959)

Uttarādhyayanasūtraṁ, edited Muni Virendar (Bikaner, 1973)

Isibhāsiyāiṁ, edited Mahopadhyay Vinayasagar (Jaipur, 1988)


Additional Texts (GRETIL):

When I started this work, around 20 years ago, there were very few transcriptions of texts available, nearly all included here were made by myself working from printed editions. Since that time there has been an enormous amount of work put into transcribing Buddhist texts, especially later Mahāyāna works.

The leading publisher has been the GRETIL (Göttingen Register of Electronic Texts in Indian Languages, project at the University of Göttingen in Germany; but also the Digital Sanskrit Buddhist Canon ( organised by the University of the West in California has contributed greatly to this work, and also made their work available to GRETIL.

The texts from GRETIL, which are what are used here and are for the most part, transliterations of printed editions. From my experience I would say they are not always 100% accurate, and sometimes they have changes introduced that are unrecorded. Also they are sometimes amalgams of more than one edition of a text.

Occasionally they are editions that were originally made available online by various editors. Because of the nature of the work then, I give here links to the texts as they existed in early 2020. The texts were found through advanced AI and neural network techniques on the Buddha Nexus website.

I have made some changes in presentation for consistency here. GRETIL uses the IAST system of transliteration, whereas I use ISO/Unicode, the changes mean that ṃ there becomes ṁ here, and ṛ there is represented by  here. Normally the texts on GRETIL use forward slash as the divider in verses / and //, whereas I have preferred the bar | & ||. I have repunctuated and parsed the texts for consistency, and to make comparison easier for the student.

Many of the texts simply represent themselves as quoting the verses, sometimes with attribution of the original text and sometimes just as something the Buddha has said. Others are from chantings of traditional texts found at the end of other recitals, for instance, of the Prātimokṣa. While others quote for the authority the Buddha’s words lend to a particular teaching.

Abhidharmadīpa, based on Jaini, 1959 (

Abhidharmakośabhāṣyam, based on Pradhan, 1975 & Shastri, 1998 (

Abhidharmakośavyākhyā, based on Wogihara, 1932-36 (

Abhisamācārika-Dharma, ?? (

Bhaiṣajyavastu, based on Dutt, 1939-1959 (

Bhikṣuṇī Vinaya, based on Roth, 1970 (

Cittaviśuddhiprakaraṇa, based on Patel, 1949 (

Dharmasamuccaya, based on Caube, 1993 (

Dvāviṁśatyavadānakathā, based on Okada, 1993 (

Ekottarāgama-Fragmente, based on Tripathi, 1995 (

Garuḍapurāṇam, based on Venkatesvara, ?? (

Kalpanāmaṇḍitikā, based on Lüders, 1926 (

Kośāmbakavastu, based on Dutt, 1984 (

Mahābhārata 05, based on Bhandakar, 1999 (

Mahābhārata 10, based on Bhandakar, 1999 (

Mahābhārata 12, based on Bhandakar, 1999 (

Mahābhārata 13, based on Bhandakar, 1999 (

Mahāsubhāṣitasaṅgraha, various, ?? (

Manusmti, based on Yano & Ikari, 2020 (

Prajñākaramati, based on Vaidya, 1961 (

Pravrajyāvastu, based on Vogel & Wille, 1984-2002 (

Prātimokṣasūtram (Mā), based on Pachow & Mishra, 1956 (

Prātimokṣasūtram (Mā-L), based on Tatia, 1976 (

Prātimokṣasūtram (Mūl), based on Banerjee, 1954 (

Prātimokṣasūtram (Sū), based on von Simson, 1986 & 2000 (

Ratnāvalī, based on Vaidya, 1961 (

Satyasiddhiśāstram, based on Sastri, 1975 (

Saṅghabhedavastu 1, based on Gnoli, 1977-78 (

Saṅghabhedavastu 2, based on Gnoli, 1977-78 (

Suvarṇavarṇāvadāna, based on Roy, 1971 & Rajapatirana, 1974 (

Suvikrāntavikrāmiparipcchā, based on Vaidya, 1961 (

Yogalehrbuch, based on Schlingloff, 1964 (

Āyuḥparyantasūtra, based on Matsumura, 1989 (

Śaraṇagamanadeśanā, based on Negi, 1992 (

Śarīrārthagāthā, based on Enomoto, 1989 (

Śrīghanācārasaṁgrahaṭīkā, based on Singh, 1983 (

Method of Presentation

When lines in a verse of one of the parallels are not found in the Pāḷi Dhammapada, I have included those lines, so as to represent better the form of the original, they are written in grey coloured text.

It should also be pointed out here that a verse that is only partially paralleled in the Dhammapada may find a complete parallel elsewhere in the Pāḷi Tipiṭaka, though it appears that normally when a verse as a whole is absent from the parallels it is also absent from the rest of the Canon.

I have occasionally included more than one parallel where it seemed to me that the inclusion was justified by the light it throws on the Pāḷi verse. But normally I have only chosen the closest parallel to the Pāḷi verse that I could find - this applies particularly to the Udānavarga, where there are often a number of parallels to choose from.

I have given the reference numbers by chapter and verse number to the 3 main parallels, even though Brough’s edition of Gāndhārī and Cone’s edition of Patna give only sequential numbers, In these cases I give the sequential numbers also; Udānavarga has sequential numbers, but these have been omitted here, as any quotation can easily be found under the chapter and verse number. as this allows us to see at a glance whether verses in the parallels are, for instance, at the beginning of a chapter.

The layout adopted in regard to the parallels has been maintained throughout this presentation:





This layout has normally been preserved even when it leads to a lot of white space, as it makes referencing a lot easier if one is looking for a parallel from a particular text; also it seems useful to be able to see where no parallel exists in any of these texts, in which case the source text title will be greyed out accordingly, e.g.: Gāndhārī

There are now (2020) many parallels from later Sanskrit and Jaina works included here, and the layout has been expanded with the additions placed below the basic layout, to whatever extent has been necessary. Sometimes we now have up to 10 parallels to a verse.