佛教的语言学方法 A Philological Aprroach to Buddhism K. R. Norman

SCHOOL OF ORIENTAL AND AFRICAN STUDIES
A PHILOLOGICAL APPROACH TO
BUDDHISM
THE BUDDHIST FORUM, VOLUME V
A PHILOLOGICAL APPROACH
TO BUDDHISM
The Bukkyō Dendō Kyōkai Lectures 1994
K.R.Norman
SCHOOL OF ORIENTAL AND AFRICAN STUDIES

UNIVERSITY OF LONDON
1997
Published by the School of Oriental and African Studies (University of London), Thornhaugh
Street, Russell Square, London WC1H OXG
This edition published in the Taylor & Francis e-Library, 2005.
“ To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of
thousands of eBooks please go to http://www.ebookstore.tandf.co.uk/.”
© K.R.Norman 1997
British Library Cataloguing in Publication Data A catalogue record for this book is available
from the British Library
ISBN 0-203-98571-0 Master e-book ISBN
ISBN - (Adobe e-reader Format)

ISBN 0 7286 0276 8 (Print Edition) ISSN 0959-0596
Contents
Foreword vii
Abbreviations ix
Bibliography xi
I Buddhism and Philology 1

II Buddhism and its Origins 18
III Buddhism and Oral Tradition 35
IV Buddhism and Regional Dialects 50
V Buddhism and Writing 65
VI Buddhism and Sanskritisation 80
VII Buddhism and Aśoka 96
VIII Buddhism and Canonicity 111
IX Buddhism and the Commentarial Tradition 127
X Philology and Buddhism 143
Word Index 158

Foreword
The lectures contained in this book are slightly revised versions of those I gave as
Bukkyō Dendō Kyōkai Visiting Professor at the School of Oriental and African Studies
from January to March 1994. My thanks are due to the Bukkyō Dendō Kyōkai for
instituting the Professorship and for the authorities of the School of Oriental and African
Studies for appointing me to it.
The lectures were intended for an audience which knew something about Buddhism
but little or nothing about philology. The notes which I have added are intended for the
general reader, rather than the specialist, to supply the sources for the information I give.
Abbreviations
AMg Ardha-Māgadhī
AO Acta Orientalia
B.C.E. Before Common Era
BHS Buddhist Hybrid Sanskrit
BHSD Buddhist Hybrid Sanskrit Dictionary
BHSG Buddhist Hybrid Sanskrit Grammar
BSO(A)S Bulletin of the School of Oriental (and African) Studies
BSR Buddhist Studies Review
BUp
C.E. Common Era
CP I–V K.R.Norman, Collected Papers, Volumes I–V (1990–94)
CPD A Critical Pāli Dictionary, Copenhagen
CUP Cambridge University Press
DEDR T.Burrow & M.B.Emeneau, Dravidian Etymological
Dictionary (revised edition, 1984)
DPPN G.P.Malalasekera, Dictionary of Pāli Proper Names
(1937–38)
EV I, II K.R.Norman, Elders’ Verses I (1969), II (1971)
GD I, II K.R.Norman, The Group of Discourses I (1984), II (1992)
GDhp Gāndhārī Dharmapada
IIJ Indo-Iranian Journal
IT Indologica Taurinensia
JA Journal Asiatique
JBuRS Journal of the Burma Research Society
JOI(B) Journal of the Oriental Institute (Baroda)
JPTS Journal of the Pali Text Society
JRAS Journal of the Royal Asiatic Society
m.c. metri causa
MIA Middle Indo-Aryan
MRI Minor Readings and Illustrator
MW M.Monier-Williams, Sanskrit-English Dictionary (1899)
OUP Oxford University Press
PDhp Patna Dharmapada
PE Pillar Edict
PED The PTS’s Pali-English Dictionary
Pkt Prakrit
PTS Pali Text Society
RE Rock Edict
SBE Sacred Books of the East
Skt Sanskrit
SOAS School of Oriental and African Studies (London)
SWTF Georg von Simson (ed.), Sanskrit-Wörterbuch der
buddhistischen Texte aus den Turfan-Funden (1973–)
Überbl O.von Hinüber, Das ältere Mittelindisch im Überblick
(1986)
WZKS Wiener Zeitschrift für die Kunde Südasiens
Abbreviations of the titles of Pāli texts are those adopted by the CPD. Editions are those
of the PTS, unless otherwise stated.
Bibliography
Adikaram, E.W., 1953, Early History of Buddhism in Ceylon, Colombo.

Allon, M., 1994, “Some oral features of the prose portions of Pali canonical sutta texts”,
unpublished conference paper.
Almond, P.C., 1988, The British discovery of Buddhism, Cambridge.
Alsdorf, L., 1962, “Uttarajjhāyā Studies”, IIJ VI, 2, pp. 110–36.
——, 1965, Les études jaina: état présent et tâches futures, Paris.
Balbir, N., 1993, “Titres et Canons dans l’Hétérodoxie Indienne”, in La formation des canons
scripturaires, Paris CERF, pp. 47–59.
Beal, S., 1884, Buddhist records of the western world, London.
Bechert, H., 1978, Buddhism in Ceylon and Studies on Religious Syncretism, Göttingen.
——, 1980, The language of the earliest Buddhist tradition, Göttingen.
——, 1985, Zur Schulzugehörigkeit von Werken der Hīnayāna-Literatur, Erster Teil, Göttingen.
——, 1987, Zur Schulzugehörigkeit von Werken der Hīnayāna-Literatur, Zweiter Teil, Göttingen.
——, 1991A, “Methodological considerations concerning the language of the earliest Buddhist
tradition”, BSR 8, 1–2, pp. 3–19.
——, 1991B, The dating of the historical Buddha, Gottmgen.
——, 1992, “The writing down of the in Pāli”, WZKS XXXVI, pp. 45–53.
Bendall, C., 1883, Catalogue of the Buddhist Sanskrit manuscripts in the University Library,
Cambridge, Cambridge (reprinted Stuttgart 1992).
Bernhard, F., 1965, Udānavarga, Vol. I, Göttingen.
Bhattacharya, K., 1989, “Brahman in the Pali Canon and in the Pali Commentaries”, in Amalā
Prajñā: Aspects of Buddhist Studies, Delhi, pp. 15–31.
Bloch, J., 1950, Les inscriptions d’Asoka, Paris.
Bloomfield, M., & Edgerton, F., 1932, Vedic Variants, Vol. II (Phonetics), Philadelphia.
Bode, M., 1909, The Pali Literature of Burma, London.
Bollée, W.B., 1974, “Buddhists and Buddhism in the earlier literature of the Śvetāmbara Jains”, in
Cousins et al., 1974, pp. 27–39.
Brough, J., 1953, The early brahmanical system of Gotra and Pravara, CUP.
——, 1954,“The language of the Buddhist Sanskrit texts” (Review of Edgerton, 1953A & 1953B),
BSOAS, 16, 2, p. 351–75.
——, 1962, The Gāndhārī Dharmapada, London.
——, 1964, “The Chinese pseudo-translation of Ārya-śūra’s Jātaka-mālā”, Asia Major NS 11.1, pp.
27–53.
——, 1980, “Sakāya Niruttiyā: Cauld kale het”, in Bechert, 1980, pp. 35–42.
Buddharakkhita, Acharya, 1985, The Dhammapada: the Buddha’s path of wisdom, Kandy.
Burford, Grace G., 1991, Desire, death, and goodness: the conflict of ultimate values in Theravāda
Buddhism, New York.
Burghart, R., & Cantlie, A., 1985, (eds.): Indian Religion, London.
Burrow, T., 1973, “The Proto-Indoaryans”, JRAS, pp. 123–40.
Burrow T., and Emeneau, M.B., 1984, (second edition of) A Dravidian Etymological Dictionary
(1961), Oxford.
Caillat, C., 1989, Dialectes dans les littératures indo-aryennes, Paris.
——, 1991, Middle Indo-Aryan and Jaina Studies, Leiden.
Carter, J.R., & Palihawadana, M., 1987, The Dhammapada, New York.
Chakravarti, N.P., 1930, L’Udānavarga sanskrit, Paris.
Childers, R.C., 1875, Dictionary of the Pāli language, London.
Clough, B., 1824, A Compendious Pāli Grammar with a Copious Vocabulary in the Same
Language, Colombo.
Collins, S., 1982, Selfless Persons, CUP.
——, 1990, “On the very idea of the Pali canon”, JPTS XV, pp. 89–126.
——, 1992, “Notes on some oral aspects of Pali literature”, IIJ 35, 2/3, pp. 121–35.
Cone, M., 1989, “Patna Dharmapada I”, JPTS XIII, pp. 101–217.
Cousins, L.S., 1983, “Pali oral literature”, in Denwood and Piatigorsky, 1983, pp. 1–11.
Cousins, L.S. et al. (eds), 1974, Buddhist Studies in Honour of I.B.Horner, Dordrecht.
Cribb, J., 1985, “Dating India’s earliest coins”, in South Asian Archaeology 1983, Naples, pp. 535–
54.
Cutler, S.Mellick, 1994, “The Pāli Apadāna Collection”, JPTS XX, pp. 1–42.
Denwood, P., & Piatigorsky, A. (eds), 1983, Buddhist Studies: ancient and modern, London.
Deraniyagala, S.U., 1990A, “Radiocarbon dating of early Brahmi script in Sri Lanka”, Ancient
Ceylon 11, Colombo, pp. 149–68.
——, 1990B, “The proto- and early historic radiocarbon chronology of Sri Lanka”, Ancient Ceylon
12, Colombo, pp. 251–92.
Dhadphale, M.G., 1980, Synonymic collocations in the : a study, Poona.
Duroiselle, C., 1911, Review of Bode 1909, in JBuRS I, 1, pp. 119–21.
Edgerton, F., 1953A, Buddhist Hybrid Sanskrit Grammar, New Haven.
——, 1953B, Buddhist Hybrid Sanskrit Dictionary, New Haven.
Ehelolf, H., 1916, Ein Wortfolgeprinzip im Assyrisch-Babylonischen, Leipzig.
Elizarenkova, T.Y., 1989, “About traces of a Prakrit dialectal basis in the language of the
”, in Caillat, 1989, pp. 1–17.
Erdosy, G., 1993, “The archaeology of early Buddhism”, in Studies on Buddhism in honour of
Professor A.K.Warder, Toronto, pp. 40–56.
Geiger, W., 1994, A Pāli Grammar, Oxford.
Gethin, R., 1992A, “The Mātikās: Memorization, Mindfulness, and the List”, in The Mirror of
Memory. Reflections on Mindfulness and Remembrance in Indian and Tibetan Buddhism,
Albany, pp. 149–72.
——, 1992B, The Buddhist Path to Awakening, Leiden.
Gombrich, R.F., 1971, Precept and Practice, OUP.
——, 1980, “The significance of former Buddhas in the Theravādin tradition”, in Somaratna
Balasooriya et al. (ed.), Buddhist Studies in honour of Walpola Rahula, London, pp. 62–72.
——, 1988, “How the Mahāyāna began”, Journal of Pali and Buddhist Studies 1, pp. 29–46.
——, 1990, “Recovering the Buddha’s message”, in Skorupski, 1990, London, pp. 5–20.
——, 1992, “The Buddha’s book of Genesis”, IIJ 35, pp. 159–78.
Goody, J. (ed.), 1968, Literacy in Traditional Societies, CUP.
Grierson, Sir George A., 1926, (second edition of) Bihar Peasant Life (1885), Patna.
Hallisey, C., 1990, “ ”, JPTS XV, pp. 155–95.
——, 1993, “Nibbānasutta: an allegedly non-canonical sutta on Nibbāna as a great city”, JPTS
XVIII, pp. 97–130.
Hara, M., 1969, “A note on the folk-etymology of rājan”, The Journal of the Ganganatha Jha
Research Institute, XXV, pp. 489–99.
Hardy, E., 1902, The Netti- , London.
Harvey, P., 1990, An introduction to Buddhism, CUP.
Heesterman, J.C., 1969, Pratidānam, den Haag.
Hinüber, O.von, 1971, “Die ‘dreifache’ Wirkung des Karma”, IIJ 13, 4, [1972], pp. 241–49.
——, 1977, “Zur Geschichte des Sprachnamens Pāli”, in Beiträge zur Indienforschung Ernst
Waldschmidt zum 80. Geburtstag gewidmet, Berlin, pp. 237–46.
——, 1981, “The ghost word dvīhitikā and the description of famines in early Buddhist literature”,
JPTS IX, pp. 74–86.
——, 1982, “Pāli as an artificial language”, IT 10, pp. 133–40.
——, 1986, Das ältere Mittelindisch im Überblick, Vienna.
——, 1989A, Der Beginn der Schrift und frühe Schriftlichkeit in Indien, Wiesbaden.
——, 1989B, “Origin and varieties of Buddhist Sanskrit”, in Caillat, 1989, pp. 341–67.
——, 1991A, “Linguistic considerations on the date of the Buddha”, in Bechert, 1991B, pp. 175–
81.
——, 1991B, The oldest Pāli manuscript. Four folios of the Vinaya from the National
Archives, Kathmandu, Mainz.
——, 1993, “From colloquial to standard language: the oral phase in the development of Pāli”, in
Premier Colloque Étienne Lamotte, Louvain, pp. 101–13.
——, 1994, “Die neun Angas: Ein früher Versuch zur Einleitung buddhistischer Texte”, WZKS
XXXVIII, pp. 121–35.
Horner, I.B., 1954, Middle Length Sayings, Vol. I, London.
——, 1957, Middle Length Sayings, Vol. II, London.
Horner, I.B., & Jaini, P.S., (tr.), 1985, Apocryphal Birth-Stories (Paññāsa-jātaka), Vol. I, London.
Hultzsch, E., 1925, Inscriptions of Asoka, Oxford.
Jacobi, H., 1884, Jaina Sūtras, Part I, SBE XXII, Oxford.
Jaini, P.S., 1986, Apocryphal Birth-Stories (Paññāsa-jātaka), Vol. II, London.
Jayasekera, U.D., 1992, The Dhammapada, Colombo.
Jayawickrama, N.A., 1962, The Inception of Discipline and the Vinaya Nidāna, London.
Jones, J.J., 1952, The Mahāvastu, Vol. II, London.
Jong, J.W.de, 1974, “A brief history of Buddhist Studies in Europe and America (Part II)”, The
Eastern Buddhist New Series VII, No. 2 (October 1974), pp. 49–82.
Kalupahana, D.J., 1986, A Path of Righteousness: Dhammapada, Lanham.
Karashima, S., 1991, in Papers in Honour of Prof. Dr. Ji Xianlin on the occasion of his 89th
Birthday, Beijing, pp. 607–43.
——, 1992, The textual study of the Chinese versions of the Saddharma- in the
light of the Sanskrit and Tibetan versions, Tokyo.
Kloppenborg, R., 1973, The Sūtra on the Foundation of the Buddhist Order, Leiden.
La Loubère, S.de, 1693, The kingdom of Siam, London (reprinted 1969).
Lamotte, É., 1958, Histoire du Bouddhisme Indien, Louvain.
——, 1970, “Le Buddha insulta-t-il Devadatta?”, BSOAS XXXIII, pp. 107–15.
——, 1988, History of Indian Buddhism (English translation of 1958), Louvain.
Law, B.C., 1957–58, “The ”, Ceylon Historical Journal, pp. 1–266.
Lévi, S., 1912, “Observations sur une language précanonique du bouddhisme”, JA,pp. 495–514.
Luce, G.H., 1974, “The advent of Buddhism to Burma”, in Cousins et al., 1974, pp. 119–38.
Lüders, H., 1940, “Sanskrit muktā, muktāphala, phala”, in Philologica Indica, Göttingen, pp. 179–
90.
——, 1954, Beobachtungen über die Sprache des buddhistischen Urkanons, Berlin.
MacQueen, G., 1984, “The doctrines of the six heretics according to the Sūtra”,
IIJ 27, pp. 291–307.
Malalasekera, G.P., 1935, , London.
——, 1937–38, Dictionary of Pāli Proper Names, 2 volumes, London.
Matsumura, H., 1989, “ ”, in Sanskrit-Texte aus dem buddhi-stischen
Kanon: Neuentdeckungen und Neueditionen, Göttingen, pp. 69–100.
Mayrhofer, M., 1976, Kurzgefasstes etymologisches Wörterbuch des Altindischen, Heidelberg, Vol.
III.
Monier-Williams, M., 1899, Sanskrit-English Dictionary, Oxford.
Mori, S., 1988, “Chronology of the ‘ sources’ for the Pāli commentaries (II)”, Buddhist
Studies (Bukkyō Kenkyū), Vol. XVII, pp. 119–67.
Morris, R., 1884, “Notes and queries”, JPTS, pp. 69–108.
Müller, F.Max, 1881, The Dhammapada, SBE X, Oxford.
Nakatani, H., 1987, Udānavarga de Subaši, Parts I & II, Paris.
Nārada, U, 1969, Conditional Relations, London.
Norman, K.R., 1967A, “Notes on the Aśokan Rock Edicts”, IIJ X, pp. 160–70 (=CP I, pp. 47–58
[50–51)].
——, 1967B, “Notes on Aśoka’s Fifth Pillar Edict”, JRAS, pp. 26–32 (=CP I, pp. 68–76).
——, 1970, “Some aspects of the phonology of the Prakrit underlying the Aśokan inscriptions”,
BSOAS 33, pp. 132–43 (=CP I, pp. 93–107).
——, 1973, “Middle Indo-Aryan Studies X”, JOI(B) XXIII, pp. 64–771 (=CP I, pp. 160–69).
——, 1976A, “Notes on the so-called ‘Queen’s Edict’ of Aśoka”, Studies in Indian Epigraphy III,
pp. 35–42 (=CP II, pp. 52–58).
——, 1976B, “Middle Indo-Aryan Studies XIII: the palatalisation of vowels in Middle Indo-
Aryan”, JOI (B), Vol. XXV, pp. 328–42 (=CP I, pp. 220–37).
——, 1976C, “Pali and the language of the heretics”, AO, 37, pp. 117–26 (=CP I, pp. 238–46).
——, 1977, “The Buddha’s view of devas”, in Beiträge zur Indienforschung Ernst Waldschmidt
zum 80. Geburtstag gewidmet, Berlin, pp. 329–36 (=CP II, pp. 1–8).
——, 1978, “The role of Pāli in early Sinhalese Buddhism”, in Bechert, 1978, pp. 28–47 (= CP II,
pp. 30–51).
——, 1979A, “Māgadhisms in the Kathāvatthu”, in Studies in Pali and Buddhism (in Honor of
Bhikkhu Jagdish Kashyap), Delhi, pp. 279–87 (=CP II, pp. 59–70).
——, 1979B,“Dhammapada 97: a misunderstood paradox”, IT 7, pp. 325–31 (= CP II, pp. 187–
93).
——, 1980A, “The dialects in which the Buddha preached”, in Bechert, 1980, pp. 61–77 (=CP II,
pp. 128–47).
——, 1980B, “Four etymologies from the Sabhiya-sutta”, in Buddhist Studies in honour of Walpola
Rahula, London, pp. 173–84 (=CP II, pp. 148–61).
, 1981, “A note on attā in the Alagaddūpama-sutta”, in Studies in Indian Philosophy, Ahmedabad,
pp. 19–29 (=CP II, pp. 200–209).
——, 1982, “The four noble truths”, in Indological and Buddhist Studies (Volume in honour of
Professor J.W.de Jong), Canberra, pp. 377–91 (=CP II, pp. 210–23).
——, 1983A, “The Pratyeka-buddha in Buddhism and Jainism”, in Denwood and Piatigorsky,
1983, pp. 92–106 (=CP II, pp. 233–49).
——, 1983B, Notes on the Ahraurā version of Aśoka’s First Minor Rock Edict, IIJ 26, pp. 277–92
(=CP II, pp. 250–68).
——, 1983C, Pāli Literature, Wiesbaden.
——, 1984–85, “A dialect variation in Aśoka’s Minor Rock Edict I”, Journal of Ancient Indian
History, XV 1–2, pp. 1–14 (=CP III, pp. 126–36).
, 1985, “Pāli Lexicographical Studies III”, JPTS X, pp. 23–36 (=CP III, pp. 83–94).
——, 1988, “The origin of Pāli and its position among the Indo-European languages”, Journal of
Pali and Buddhist Studies 1, 1988, pp. 1–27 (= CP III, pp. 225–43).
——, 1989A, “Common terminology in early Buddhist and Jain texts”, in Jainology: manifold
facets (Pt. Jaganmohanlal Shastri Sadhuvad Grantha), Rewa, pp. 393–97 (=CP IV, pp. 264–
70).
——, 1989B, “The Pāli language and scriptures”, in Skorupski, 1989, pp. 29–53 (=CP IV, pp. 92–
123).
——, 1989C, “Dialect forms in Pāli”, in Caillat, 1989, pp. 369–92 (=CP IV, pp. 46–71).
, 1990A, “Aspects of early Buddhism”, in Ruegg and Schmithausen (eds): Earliest Buddhism and
Madhyamaka, Leiden, pp. 24–35 (=CP IV, pp. 124–38).
——, 1990B, “Why are the four noble truths called ‘noble’?”, in Ānanda: essays in honour of
Ananda W.P.Guruge, Colombo, pp. 1–13 (=CP IV, pp. 170–74).
——, 1990C, “Pāli philology and the study of Buddhism”, in Skorupski, 1990, pp. 31–39.
——, 1991A, “Brahmanical terms in a Buddhist guise”, in Skorupski, 1991, pp. 193–200 (=CP IV,
pp. 271–80).
——, 1991B, “Observations on the dates of the Jina and the Buddha”, in Bechert, 1991B, pp. 300–
12 (=CP IV, pp. 185–201).
——, 1991C, “Syntactical compounds in Middle Indo-Aryan”, in Caillat, 1991, pp. 3–9 (=CP IV,
pp. 218–25).
——, 1991D, “Death and the Tathāgata”, in Studies in Buddhism and Culture (In honour of
Professor Dr. Egaku Mayeda), pp. 1–11 (=CP IV, pp. 251–63).
——, 1993, “The metre of the Buddhist Hybrid Sanskrit Upāli-sūtra”, in Studies on Buddhism in
honour of Professor A.K.Warder, Toronto, pp. 113–23.
——, 1994, “Mistaken ideas about nibbāna”, in Skorupski, 1994, pp. 209–24.
Oberlies, Thomas, 1993, Āvaśyaka-Studien 2, Glossar ausgewählter Wörter zu E.Leumanns “Die
Avaśyaka-Erzählungen”, Stuttgart.
O’Flaherty, Wendy Doniger, 1975, Review of B.L. Ogibenin: Structure d’un mythe védique; le
mythe cosmogonique dans le , in BSOAS 38, pp. 173–75.
Oldenberg, H., 1879, The , London.
Oldenberg, H., & Pischel, R., 1883, The Thera- and Therī-gāthā, London.
Paranavitana, S., 1970, Inscriptions of Ceylon, Vol. I, Ceylon.
Pinault, Georges-Jean, 1989, “Reflets dialectaux en védique ancien”, in Caillat, 1989, pp. 35–96.
Potter, Karl H., 1981, Encyclopaedia of Indian Philosophies, Vol. III (Advaita Vedānta up to
and his pupils), Princeton.
Pruitt, W., 1987, “References to Pāli in 17th-century French books”, JPTS XI, pp. 121–31.
Rādhakrishnan, S., 1950, The Dhammapada, London.
Rahula, Walpola, 1956, What the Buddha taught, London.
Rhys Davids, C.A.F., 1974, (third edition of) Buddhist Psychological Ethics (1901).
——, 1909, Psalms of the Early Buddhists. I—Psalms of the Sisters, London.
Rhys Davids, T.W., 1881, Buddhist Suttas from the Pāli, London.
Rhys Davids, T.W., & C.A.F., 1921, Dialogues of the Buddha, Part III, OUP.
Roth, G., 1980, “Particular features of the language of the
Lokottaravādins and their importance for early Buddhist tradition”, in Bechert, 1980, pp. 78–
135.
Saksena, B., 1936, “Pāli bhūnahu”, BSOS 8, pp. 713–14.
Sander, Lore, “ und in Vinaya- und Hīnayāna-Sūtra-Texten”, in Bechert, 1985,
pp. 144–60.
Sasaki, Genjun H., 1986, Linguistic approach to Buddhist thought, Delhi.
Schmithausen, L., 1981, “‘Liberating Insight’ and ‘Enlightenment’ in Early Buddhism”, in Studien
zur Jainismus und Buddhismus (Gedenkschrift für Ludwig Alsdorf), Wiesbaden, pp. 199–250.
——, 1987, “Beiträge zur Schulgehörigheit und Textgeschichte kanonischer und postkanonischer
buddhistischer Materialien”, in Bechert, 1987, pp. 304–81.
Schubring, W., 1962, The doctrine of the Jainas, Delhi.
Schwarzschild, L.A., 1961, “The indeclinable je in MIA”, Bhāratīya Vidyā, XX– XXII, pp. 211–
17.
——, 1991, Collected Articles of L.A. Schwarzschild on Indo-Aryan 1953–1979, Australian
National University, Canberra.
Shukla, N.S., 1979, The Buddhist Hybrid Sanskrit Dharmapada, Patna.
Simson, G.von, 1973–, Sanskrit-Wörterbuch der buddhistischen Texte aus den Turfan-Funden,
Göttingen.
Skilling, P., 1993, “A citation from the of the Abhayagiri school”, JPTS XVIII,
pp. 165–75.
Skorupski, T., 1989, The Buddhist Heritage, Tring.
——, 1990, The Buddhist Forum, Volume I, London.
——, 1991, The Buddhist Forum, Volume II, London.
——, 1994, The Buddhist Forum, Volume III, London.
Southwold, M., 1985, “Nirvana in village Buddhism”, in Burghart & Cantlie, 1985, pp. 15–50.
Stargardt, J., 1990, The Ancient Pyu of Burma. Volume I, Early Pyu cities in a man-made
landscape, Cambridge.
Takakusu, J. (tr.), 1896, A record of the Buddhist religion, London.
Tambiah, S.J., 1968, “Literacy in a Buddhist village in North-East Thailand”, in Goody, 1968, pp.
85–131.
Thomas, E.J., 1949, (third edition of) The life of the Buddha (1927), London.
——, 1951, (second edition of) History of Buddhist Thought (1933), London.
Trautmann, Thomas R., 1981, Dravidian Kinship, Cambridge.
Tsuchida Ryūtarō, 1991, “Two categories of Brahmins in the early Buddhist period”, The Memoirs
of the Toyo Bunko, 49, pp. 51–95.
Turner, R.L., 1973, “Pāli phāsu and dātta”, BSOAS 36, pp. 424–28 (=Collected Papers, 1975, pp.
430–33).
Vetter, T., 1985, “The most ancient form of Buddhism”, in Buddhism and its relation to other
religions (Festschrift for Dr. Shozen Kumoi), Kyoto.
Vogel, C., 1970, The teachings of the six heretics, Abhandlungen für die Kunde des Morgenlandes
XXXIX, 4, Wiesbaden.
Walleser, M., 1914, Prajñāpāramitā. Die Vollkommenheit der Erkenntnis nach den indischen,
tibetischen und chinesischen Quellen, Göttingen.
Walshe, M., 1987, Thus have I heard, London.
Warder, A.K., 1970, Indian Buddhism, Delhi.
Wiltshire, M.G., 1990, Ascetic figures before and in early Buddhism. the emergence of Gautama as
the Buddha, Mouton de Gruyter, Berlin and New York.
Winternitz, M., 1933, A history of Indian literature, Vol. II, Calcutta.
Young, F., 1992, (third impression of) Can these dry bones live? (1982), London.
Yuyama, A., 1969, “The bodhi tree in the Mahāvastu-avadāna”, in Heesterman, 1969, pp. 488–92.
I
Buddhism and Philology
In the autumn of 1993 I attended a conference in America on the State of the Art in
Buddhist Studies.1 At that conference speaker after speaker said that they felt
marginalised, in effect left out on the edges of their subject. In some cases this meant that
their interest in Buddhism was what might be called “on the fringe”, often combined with
another discipline, e.g. social anthropology; in other cases it meant that in the
departments of humanities in the various universities to which they belonged they were
regarded with some sort of suspicion as being teachers of subjects which were not as
important as other subjects, so that when money was allocated for teaching or library
purposes they got the smaller share, or when posts were sacrificed in the interests of
economy it was posts in their subjects which went.
Even those who might have been thought to be in the mainstream of Buddhist Studies
said that they felt marginalised when they went as part of their field work to the country
of their own specific interest only to find that they were regarded in some way as
outsiders by the Buddhists in those countries.
My contribution to the conference was a paper on “Pāli studies in the West: present
state and future tasks”,2 delivered in a session on Textual and Philological Studies. It was
greeted with no great enthusiasm. People I met at various social functions thereafter
during the course of the conference said “Ah, you are Mr Norman, aren’t you? Your
paper was about texts, wasn’t it?” and hurriedly changed the subject, although one or two
people did express interest, in a way which showed that they had never thought seriously
about textual and philological studies before. Despite this reception, I felt in no way
marginalised, as did the many participants who spoke about what I might describe as the
“trendier aspects of Buddhism”. This was strange because there were very few people at
that conference who would claim to be philologists first and foremost, and to the best of
my knowledge I was the only philologist there who disclaimed all knowledge of
Buddhism. Nevertheless, I felt at the very core and heart of the subject—although this
may be the equivalent of saying that I was the only one in step—because the branch of
learning which I represented, i.e. philology, was to my mind the
1 “Buddhist Studies: The State of the Art”, University of California at Berkeley, October 28–31,
1993.
2 Subsequently published in a slightly revised form in Religion (1994) 24, 165–72.
A philological approach to Buddhism 2
very basis of all the work which my fellow participants in the conference were doing,
although my specific area of philology, i.e. early Indo-Aryan dialects, bore no
relationship to the geographical areas of most of the participants’ interests.
It is, in fact, not surprising if some scholars in the field of Buddhist Studies feel
marginalised, because to some extent they have marginalised themselves. There is a
tendency in modern scholarship to look always for the new—scholars entering the field
are not content to tread the paths well-worn by their predecessors, even when it is clear
that the work of their predecessors needs reworking. The cry is to find something new,
something which has not been done before. I am confronted with this tendency all the
time. Prospective research students visit me or write to me and ask what they can do for
their doctoral thesis in the field of Pāli studies. I say: “What has not been done needs to
be done, and what has been done needs to be done again”. Of these the second is the
more important, because, by and large, the most important Pāli texts were published first,
when little was known about the Pāli language—there were only inadequate dictionaries
and grammars, and only a few manuscripts had come to Europe. Now more and better
manuscripts are available, and we have superior grammatical and lexicographical aids,
and so vast improvements can be made in editions and in translations based on those
editions, and in books on Buddhism based on those translations. Unfortunately, the
enquirers all want to do something new, so the study of the fundamentals is abandoned
while they go after trendy trivia which they hope will have an earthshaking effect upon
the world of Pāli and Buddhist studies when the result of their research appears.
Why did I feel the way I did at that conference? I will answer that question with
another question.
If you hear about a religion for the first time, and want to know about it, how do you
set about gaining the knowledge you desire? You can of course read a book about it, but
if you are not entirely satisfied with the answers you get, and read another book on the
subject, you may well find that although the authors seem to be talking about the same
thing they are not necessarily saying identical things about it.
An alternative method is to question a follower of the religion, but you will almost
certainly find that precisely the same inconsistency in the views expressed appears, if you
go on to question another follower. Sooner or later any serious investigator into a religion
must examine the texts which the followers of the religion say they are taking as the basis
for their religion. He must see how far what the texts say and what the followers do
coincide, i.e. he must measure the difference between precept and practice.
If, by chance, the religion is one which still maintains an oral tradition, that is to say
its “sacred” texts have not yet been written down, then problems may arise from the fact
that the modern adherents of the religion are precisely those who are the authorities for
their religious beliefs, and what they say and do is their religion, i.e. the practice is the
precept, or the precept is the practice.
Such a situation presents difficulties of its own, but such difficulties are mercifully
absent from the study of Buddhism. Here we are faced with a situation where the founder
of the religion lived approximately 2½ millennia ago. There is a large body of material in
an assortment of languages which is ascribed to the founder and much of it has been
edited and published in a form which makes it easily accessible and available to
investigators. Those wishing to know about Buddhism can consult these ancient sources,
Buddhism and philology 3
and there is a battery of ancillary material—dictionaries, grammars, translations, etc.—

which enables them to do so.
There is, of course, the problem that if we set out to understand what the earliest texts
say, i.e. those ascribed to the Buddha himself, or his followers during his lifetime, we
have to consider the fact that the language which we find in such texts is not necessarily,
and almost certainly is not, the language of the Buddha himself, i.e. the language has
been changed both synchronically—it has been translated or transformed into other
languages as the need arose, perhaps as Buddhism spread into neighbouring areas—and
also diachronically, i.e. as the language of the readers or recensionists developed in the
course of time, this had an effect upon the language of the texts. It is also possible and
indeed probable that changes took place in what the Buddha is reported to have said and
done, i.e. the tradition changed, unconsciously, the Buddha’s views because, as certain
words fell out of use and were no longer understood, they were “brought up to date” and
made more intelligible by having an interpretation inserted into the texts in their place.
The account of what the Buddha said or did might also be changed consciously by having
interpolations inserted, for various reasons. Sometimes it is because a passage seemed
appropriate to the context. For example, when in the Mahāparinibbānasutta the Buddha
has given eight reasons for an earthquake occurring, a number of other sets of eight
phenomena are added. Sometimes an interpolation occurs because a person or a city or a
sect wished to have some dogma or action authenticated, and a reference to the Buddha
doing something or saying something was inserted into the text to give the authentication
they desired.
Much of our knowledge about Buddhism, at least in the early days, came in a rather
haphazard way, as the result of reports by travellers, envoys, etc. One such envoy was
Simon de La Loubère, who in 1687–88 went to Siam as an envoy of the king of France
(King Louis), and on his return wrote a book about the Kingdom of Siam, which was
published in France in 1691. Two years later an English translation was published in
England.3 La Loubère gave a fascinating account of many aspects of Siam and Siamese
culture, and he also included an account of the Siamese religion—Buddhism—and the
sacred language of the Siamese. He noted the differences between that language, which
he called Balie, and the Siamese language, and noted, correctly, the relationship between
the former and Sanskrit. He also gave French translations of a few Buddhist texts.
It was until quite recently believed that he probably made these translations from
Siamese translations from Pāli, but it has now been shown that, before La Loubère went
to Siam, French Christian missionaries were already active in that country, and had
compiled dictionaries of Siamese and Pāli, now unfortunately lost, and even a Pāli
preface and postface to a translation of St Luke’s Gospel into Thai, which still survives in
France in the archives of the Missions étrangères.4 It is therefore possible or even
probable that La Loubère gained his knowledge from those missionaries, rather than from
the man in the street. We may assume that those missionaries learned Pāli and studied
texts in Pāli in order that they might better confute their Buddhist opponents and convert
them to Christianity.
3 La Loubère, 1693.
4 See Pruitt, 1987, 121–31.
Such was, indeed, the purpose of British missionaries in Sri Lanka a little more than a
century later. Men like Benjamin Clough studied Pāli and wrote grammars, together with
lengthy word lists based upon indigenous dictionaries.5 If the original intention was to
learn the language so that the religion it supported might be demolished, we may suspect
that familiarity begat not contempt but affection, and translations of the Pāli texts, e.g. the
Dhammapada, which those missionaries produced did more to make English readers want
to learn more about Buddhism than to convert the followers of Buddhism to Christianity.
In fact the impact of Buddhism upon nineteenth century Britain reveals a very varied
response, well depicted recently by Philip Almond.6
It is possible, however, that such study of Pāli would have been as ineffectual outside
missionary circles as the French endeavours—although they sent manuscripts back to
France there is little evidence of their being studied—had it not been for the sudden rise
of a new field of study: comparative philology. In the few years since Sir William Jones’
well-known statement about the relationship of Latin, Greek, Sanskrit, etc., the subject
had made rapid progress.
Although Jones was not the first to notice such relationships, his pronouncement
seems to have attracted more attention than the suggestions of others. The need to study
Sanskrit and to obtain manuscripts of Sanskrit works led to the Danish philologist
Rasmus Rask going to India and Ceylon (Sri Lanka) c. 1820 and bringing back a fine
collection of manuscripts, which is now housed in the Royal Library in Copenhagen.
Other manuscripts of Buddhist texts came to the West in different ways. Brian
Hodgson was the British resident in Kathmandu in the 1830’s and began to send Buddhist
Sanskrit manuscripts to Europe. These formed the basis of Burnouf’s studies of
Buddhism. After the conquest of Upper Burma in 1885 a large part of the royal library at
Mandalay came to England, It is unfortunate that the complete holdings of the library
were not sent, for those manuscripts which did not reach England at that time have been
lost. Daniel Wright was the brother of William Wright, the Professor of Arabic at
Cambridge. When he became surgeon to the British Residency in Kathmandu in the
1870’s, he was asked to obtain copies of Sanskrit manuscripts for Cowell, the Professor
of Sanskrit at Cambridge at the time. Wright found it cheaper and easier to buy originals,
which are now in the University Library at Cambridge.7 Once professors began to be
appointed in European universities to teach Sanskrit and Pāli, the academic study of
Buddhist Sanskrit and Pāli manuscripts made great progress, and Spiegel and Fausbøll
had published Pāli texts by 1855.
These editions and studies were, inevitably, made from the very limited number of
manuscripts which were available to the editors, and were restricted to the tradition of the
country in which they were procured, in the case of Rask and Clough to Ceylon (Sri
Lanka). Only gradually did it become possible to make use of materials from other
traditions, and this produced surprises. Although Fausbøll had been able to make use of
Burmese manuscripts for his edition of the , for the most part he
took the readings of his Sinhalese manuscripts for the text, and gave the readings of the
5 Clough, 1824.
6 Almond, 1988.
7 Bendall, 1883.
Burmese manuscripts in the footnotes. When, however, he came to Volume VI, the
Mahānipāta, he reported8 that he had not taken full notice of his Burmese manuscript Bd,
as the text for the Mahānipāta in that manuscript had been very much enlarged
throughout, so as to make it, in many places, very different from the Sinhalese tradition.
He reported that the aim of the Burmese redactor seemed to have been to make the tale
more lucid and intelligible, but the consequential difference was in many particulars so
great that he expressed the hope that some scholar would give a separate edition of the
Mahānipāta according to the Burmese redaction. Fausbøll had been forced to realise the
need for comparative studies, on a philological basis, of parallel versions of Buddhist
texts.
I have entitled this lecture “Buddhism and philology”. I might just have well have
entitled it “Buddhism and the philologist”, since I want to set out the pattern which I, as a
philologist, shall follow in these lectures. The course is in effect an apologia pro vita
mea, explaining my encounter with Buddhism through the medium of philology, and at
the same time setting out some of the things which philology has told us about the
Buddha and various aspects of Buddhism: its origin, the way in which the texts were
transmitted, first orally and then in a written form, the way in which a selection of texts
was given canonical status, and how there was need for commentaries upon them. In my
last lecture I shall suggest what further contributions philology can be expected to make
to the promotion of Buddhism by giving a better understanding of the texts which lie at
the heart of Buddhist Studies.
I personally came to this body of material, or at least to a portion of it, not from my
desire to investigate Buddhism but from purely philological considerations. I was trained
as a classicist and studied classical philology, in the form which was current in my
student days, i.e. the investigation of the relationship between Latin, Greek and Sanskrit
in particular, and between other Indo-European languages in general. I went on to study
Sanskrit and the dialects associated with Sanskrit—the Prakrits—and was appointed to
teach the Prakrits, or Middle Indo-Aryan, as they are sometimes called, lying as they do
between Old Indo-Aryan, i.e. Sanskrit, and New Indo-Aryan, i.e. the modern Indo-Aryan
languages spoken mainly in North India.
I was interested in the relationship between the various Middle Indo-Aryan dialects,
and began to study them comparatively, trying to use forms in one dialect to shed light on
forms in other dialects. Because Middle Indo-Aryan includes Pāli, the language of the
Theravādin canon and the commentaries upon it, as a matter of course I included the
study of Pāli in my investigations. My interest in Pāli led to my being invited to join the
Council of the Pali Text Society and I was persuaded to help with the continuation of the
Concordance, which had come to a standstill after the death of its first
editor, E.M.Hare. This was essentially a philological undertaking, in that it consisted of
presenting for publication the contents of the Pāli canon, word by word, with the material
analysed and set out by case forms, tenses, etc., as was appropriate. The words were
given a meaning merely for referenee purposes, but no interpretation of meaning, or
discussion from a doctrinal point of view was involved.
That work in turn led to an invitation to become involved with A Critical Pāli
Dictionary, based in Copenhagen. That involvement led eventually to becoming Editor-
8 Preliminary remarks 7.
in-Chief of that project, until the portion for which I had assumed responsibility (to the
end of Volume II, i.e. to the end of the vowels) was completed at the end of 1990, some
66 years after the first fascicle of the dictionary was published by Helmer Smith and
Dines Andersen in 1924. That work for A Critical Pāli Dictionary was also primarily a
philological undertaking, since it involved an analysis of the usage of each word listed
and the assignment of meanings for each usage, but not, for the most part, a discussion of
the doctrinal importance of each word and the part it played in Buddhism.
What do I mean by saying that in A Critical Pāli Dictionary there was no discussion
of the doctrinal importance of each word and the part it played in Buddhism? I mean that
it is sufficient in a dictionary to state, for example, that jhāna means “meditation”,
without any discussion of how one meditates, or the importance of meditation to a
Buddhist or to Buddhism. As a result of my activities in this field, extending over nearly
twenty years, I am in the habit of saying that I know nothing about Buddhism, but I do
know a little about some of the words used in Buddhism and some of the languages of
Buddhism. I regard one of my purposes in life as being an adviser to those who know a
lot about Buddhism, but not a great deal about the languages of Buddhism. From time to
time I receive enquiries from those wishing to write about aspects of Buddhism. They tell
me what they think a particular passage means, and how they propose to incorporate their
interpretation into their discussion or description of Buddhism, and they ask me whether I
think that their proposal is possible, philologically speaking.
The difference between a student of Buddhism and a student of Buddhist philology is
very considerable. It may be possible for someone to be an expert in all types of
Buddhism. Since the age of the polymaths is not quite dead, it may be possible for
someone to be an expert in all types and aspects of Buddhist philology. I do not know
such a person, although I have been fortunate enough in my life to meet one or two
scholars who come close to it.
But I am not one of them. Such expertise as I have is confined to the dialects of
Middle Indo-Aryan used primarily in North India between about 500 BC and 1000 AD,
but taken from there south to Ceylon (Sri Lanka) and north to Chinese Turkestan. My
knowledge of Buddhist philology is therefore restricted to Indian Buddhism, and
especially to those schools of Buddhism which made use of a dialect of Middle Indo-
Aryan for their texts, or used a variety of Sanskrit which was greatly influenced by
Middle Indo-Aryan. Nevertheless, much of what I shall say applies to the way in which
we should approach the texts which form the basis for the study of any aspect of
Buddhism,
It is time to say what I mean by a philological approach to Buddhism. By that I mean
approaching Buddhism not merely by language—we all do that in one way or another—
but by means of what we can learn from language. Not “what does it mean?” but “why or
how does it mean it?”. Let me give an illustration. We might, in a philosophical passage,
have a sentence which says “X arises because of Y”. We can give tentative translations of
both X and Y, and the task of interpreters is to refine these translations until they have
arrived at what they think the author of the statement meant when the statement was
made. There is no need for philology in my sense of the word there. Philology in my
sense is required at an earlier stage. What is the structure of the sentence? Are we correct
in thinking that it actually says “X arises because of Y”? In many cases, perhaps most
cases, it will be clear that the structure of the sentence is sound as it stands, and there is
no need for the philologist to consider the matter further.
Let me give you another illustration from my own experience. When I was first
appointed as a Lecturer in Middle Indo-Aryan Studies, I found in my very first term that
there was an option available called “Sanskrit and Indo-Aryan philology” and I was
responsible for teaching the Middle Indo-Aryan part of the latter, centring the course
around the Aśokan inscriptions. More than that, I discovered that there were, in my very
first term, candidates taking this option. And so at high speed I set about producing a
course of lectures on Middle Indo-Aryan philology beginning with the Aśokan
inscriptions. I consulted all the editions of the inscriptions available to me, read all the
secondary literature on which I could lay my hands, and I produced a course of lectures
during which I dealt, in my opinion, in a satisfactory way with all the many problems in
those inscriptions. I showed those attending my lectures how the inscriptions should be
translated and interpreted. I was so pleased with my achievement that I inserted a course
of lectures on the Aśokan inscriptions into my own standard teaching. And so, year after
year, I lectured on the Aśokan inscriptions, using those same notes I wrote so long ago.
Well, not quite the same notes, because as my understanding of Middle Indo-Aryan
philology increased, and as my appreciation of the philological approach to these matters
grew, I realised that I had tried to say what the words meant, but not how or why they
meant it. As I came to grapple with the question of how they meant it, I discovered that,
in many cases, I did not know how the inscriptions could possibly mean what I had said
they meant, and as a result of not knowing how they could mean what I had said, I had
great doubts about what they did actually mean. And so my study of the Aśokan
inscriptions led to a situation where every year I understood less and less. To paraphrase
the words of another honest seeker after truth: the only thing I know about the Aśokan
inscriptions is that I know nothing about the Aśokan inscriptions.
I can almost hear some of you thinking, “If that is what philology does for you then
thank goodness I am not a philologist”. And of course there is a great deal to be said for
that view. If you are setting out to explain and translate a text then it is very frustrating,
even humiliating, to have to confess that you do not understand how the words can
possibly mean what you would like them to mean. And consequently many translators do
not confess this, possibly because they do not wish to show their ignorance, or possibly
because they do not realise their ignorance. Instead they translate by intuition. If you
examine many translations of Buddhist texts the impression you will gain is that the
translator has looked at the words, has perhaps looked them up in the dictionary and has
ascertained meanings for them, and then has used his intuition to put those words
together to reveal the meaning of the Buddhavacana, the word of the Buddha. Sometimes
intuition leads a translator to the right result, and sometimes not. Many translators have
worked on the basis of intuition, and for the most part it has got them along very well.
Often self-taught, or taught by native scholars whose standard of proficiency in teaching
was not necessarily high, they have been forced, especially in the days before such aids
as good dictionaries and grammars in a European language existed, to translate by the
“intuitive” method, whereby they examined the context and deduced from that what the
meaning must be. As has been said by Professor O’Flaherty in the context of
studies, where the situation is very similar: “Many of our most valuable insights into
otherwise obscure terms have come from scholars who have seen what the meaning must
be from [the] context, from an understanding of Vedic thought processes”.9 The wonder
is not that these intuitive translators were sometimes incorrect, but that they were correct
so often.
In what I propose to say in these lectures I will be more specific and restrict myself to
a field where I think, I hope correctly, that I know a little about the subject I am
discussing, i.e. the fields of Sanskrit and Pāli. In those fields, the habit of translating by
intuition—I could be unkind and say “by guesswork”—has been in the past, and to some
extent still is, the standard way of proceeding when faced with a difficult passage. When
in doubt, make a guess. This is all very well, if the guesswork leads to the right answer,
but guessing does not seem to me to be the right basis for the study of anything, whether
it is religion or social conditions, political affairs or history.
The more we study the work done by our predecessors the more, I am sorry to have to
say, we find that our idols have feet of clay. Thomas William Rhys Davids founded the
Pali Text Society in 1881, and by his editions and translations of Pāli texts and by his
contribution to Pāli lexicography he probably did more than anyone else to promote Pāli
studies in Europe. My attention was drawn recently to a passage in the Dīgha-nikāya, or
more precisely, to Rhys Davids’ translation of the passage. It comes in the -
suttanta, and outlines the sort of treatment which might be meted out to any non-human
creature who approached someone who knows the spell. One sentence in that
passage10 consists of four words or phrases; a phrase meaning “empty bowl”, a word
meaning “head”, a particle which has several meanings of which the concessive
“although” and the emphatic “indeed” are perhaps the most common, and a verb meaning
“to bend, to turn, to turn upside down”. Rhys Davids took these words and translated
them as “They would bend down his head like an empty bowl”,11 without perhaps
reflecting on whether there was any significant difference between a head being turned
down like an empty bowl and being turned down like a full one. In fact, the translation is
wrong. Rhys Davids was translating by intuition, or as we might say, he was guessing the
meaning. The words for “empty bowl” and “head” are not in the same case, which is
what his translation would require, and the particle does not mean “like”. The sentence
means “they would turn an empty bowl upside down on his head”—the word for head is
in the locative case, and the particle is used here to give emphasis to the word which
precedes it. The emphasis is upon the fact that the bowl was empty. The bowl had to be
empty, so that it would slip down on to the recipient’s shoulders. The tormentors then hit
the bowl with an iron bar, as the commentary explains,12 and the resultant reverberations
would presumably do the victim’s eardrums and brain no good at all.
The method of translation which leads to this type of misunderstanding of the text is
very similar to the way in which some young children read. They learn the value of some
of the letters, and the context in which they find the letters on the page—they are perhaps
9 O’Flaherty, 1975, 173–75 (173).
10 api ssu mārisa amanussā rittam pi sīse , D III 203,21–22.
11 T.W. & C.A.F.Rhys Davids, 1921, 195.
12 rittam pi pattan ti, patta-sadisam eva loha- hoti. sīse
yāva gala- bhassati. atha majjhe ayo-khīleana , Sv 968, 37–
39.
accompanied by a picture illustrating the story—enables them to work out what is going
on, and so to guess appropriate words which include the letters they can recognise and
will also fit the story. I am reminded of a child, who at the age of nine was diagnosed as
being dyslexic, with a specific reading problem. He had been “reading” in this way, and
his problem had only come to light because he had reached the stage in arithmetic where
he no longer had to add 193 to 476, but was beginning to get questions which had to be
read, e.g. of the “if two men can dig a hole five feet deep in two days how long would it
take ten men to dig to Australia”-variety—the sort of question which you cannot guess
from the context. That is exactly parallel to the intuitive translator. He knows the
meaning of some, or even all, of the words, and he deduces from the context the way in
which the words should be construed. Once the context gives no help for his intuition, he
is reduced to blind guesswork, and he may not always guess correctly.
You may say that the example I have quoted from Rhys Davids is a trivial matter, and
think that a trifling mistranslation of this nature is of no consequence. I cannot agree with
that view. The important point about this matter is that the Pāli words are incapable of
bearing the translation which Rhys Davids assigned to them—a translation which, I may
add, was copied, perhaps unthinkingly, by a more recent translator.13 The translator gives
the impression of having made no attempt to see how the words could have the meaning
he gave them, or to put it more bluntly, he showed himself unable to construe the words
in a simple Pāli sentence. If he adopted such a method of translating in a simple sentence
where we can see his error, how can we trust him in a more important context, where the
language may be more complicated, and every atom of philological expertise is needed to
unravel the meaning intended by the author of the text?
The situation can be seen very clearly from the way in which some people make
translations. Hardly a year goes by without a new translation of the Dhammapada
appearing. There are now probably forty or more available in English alone. And yet, if
one compares a new translation with its predecessors, one notes that for the most part the
new translation differs only in very minor details: the order of words, perhaps, or the
choice of translations adopted to serve for specific technical or semi-technical terms.
Anyone reading a translation of a Pāli text is dissatisfied with the translations given for
words like dhamma, kamma, nibbāna, āsava, etc., and, sooner or later, is fired with the
desire to make a new translation giving what they think is the “proper” or “correct”
translation of these terms. Having done this, translators believe that they have made a
better translation, without giving any thought to the question of whether they have gained
a better understanding of the meaning of the phrase or the sentence as a whole. It has
been pointed out14 that quite reputable translators have been known to take over
translations from their predecessors, forgetful or perhaps ignorant of the fact that the text
they print with their translation follows a different reading, and cannot possibly mean
what they say it does.
At this point, I should make it clear that the fact that we do not see a philological
problem in a text does not mean that there is no such problem, because it sometimes
happens that those unversed in philology do not understand that the language they are
13 Walshe, 1987, 477.

14 Brough, 1962, 195 (ad GDhp 75).

dealing with, which perhaps they think they understand very well, does not and perhaps
cannot have the meaning which they ascribe to it. It is, for example, not always
understood by non-specialists that an early Pāli canonical sutta is itself a translation, and
forms which were left untranslated when the Pāli recension was made from some earlier
version can sometimes be identified.15 Although it may be possible to translate such Pāli
texts into English, we must, if our aim is to find out how the words can have the meaning
we assign to them, first try to find out what the author actually said, i.e. we must “back-
translate” the text into a form of language as close as possible to that which we believe
was spoken at the time of the Buddha.16
This involves making use of all the resources of philological and literary criticism to
establish the original form of the text which we wish to translate, which requires a
knowledge of the languages of North India and Ceylon at the time of the Buddha and the
centuries immediately after his death. This in turn necessitates expertise not only in the
Middle Indo-Aryan languages, of which Pāli is one, but also in Classical Sanskrit,
Buddhist Sanskrit and Buddhist Hybrid Sanskrit, since much of the language of early
Buddhist texts is related to or taken over from Sanskrit, while parallel versions of many
Pāli canonical texts exist in Buddhist Sanskrit or Buddhist Hybrid Sanskrit
A very good example of a portion of text which seems to have no philological
problems is something which is fundamental to Buddhist belief—the so-called four noble
truths. Everyone who knows anything about Buddhism knows these four truths well. The
shortest form of them that I know is that which says simply: “Four noble truths: pain,
arising, cessation, path”.17 This is, of course, abbreviation to the point of error. The first
noble truth is not pain, but the realisation that “this (i.e. the whole of existence) is pain”. I
have seen “this suffering is a noble truth”, “this origin of suffering is a noble truth”, and
“this cessation of suffering is a noble truth” as translations of the first three noble truths,18
in complete disregard of the grammar and syntax. It is only when the philologist
examines the problem, analyses the relationship of the words, compares other versions of
the noble truths found in Sanskrit, and establishes the syntax of each phrase that the
correct translations “The noble truth that ‘this is suffering’”, “The noble truth that ‘this is
the cause of suffering’”, etc., become possible.19
Once the structure of the syntax and grammar of the four noble truths has been
understood, then it becomes possible to give consideration to a number of other sentences
and phrases found in Buddhist texts (and other Indian texts as well), which present
comparable grammatical and syntactical problems for the philologist, but not always for
other people, and to suggest explanations for them which enable satisfactory translations
to be given.20
What else can philology tell us about concepts which are not entirely unimportant in
15 e.g. in Th 1279 (sacce atthe ca dhamme ca āhu santo ), sacce is probably the locative
case, while atthe and dhamme are in the nominative case. See Norman, EV I, 292.
16 The latest views on this subject can be seen in Bechert, 1980.
17 Th 492.
18 Kloppenborg, 1973, 4.
19 See Norman, 1982, 377–91 (=CP II, 210–23).
20 See Norman, 1991C, 3–9 (=CP IV, 218–25).

the field of Buddhist Studies? The word nibbāna (or in its Sanskrit form) is so
common in discussions of Buddhism that it is very often not translated, but left in its Pāli
or Sanskrit form. It is, however, possible for the philologist to take the word and suggest
an explanation for the fact that while it appears to come from a root meaning “to blow”,
the past participle which is used in conjunction with it comes from a root of entirely
different meaning although of rather similar appearance.21 The connection between them,
when explained in this way, enables us to understand the word play which is found with
both the noun and the past participle.
We can also explain the structure of the adjectives which are commonly used to
describe and thus enable better (and more accurate) translations to be given, so
that we can better understand the reluctance of the Buddha to attempt to describe the
nature of , or of someone who has gained .
For example, one of the epithets used of is (in Sanskrit), or amata (in
Pāli) “not dead”, or “deathless”, or “undying”, or “immortal”, and we therefore
sometimes find the compound amatapada “the amata place” or “the place of amata”
being translated as “the immortal place” or “the place of immortality”. The former
translation raises the difficulty of conceiving of a place that does not die, as opposed to a
place which is mortal and does die. The latter translation implies that those who have
gained nibbāna live for ever, which, as has been pointed out,22 seems to be incompatible
with the rest of the teachings of the Buddha, and must therefore be an untenable view in
Buddhism. It is, however, possible to gain a better understanding of the meaning of the
word, if we try to find out how it gets its various meanings. We can, in the light of the
distinctions we can draw between the meanings of , better understand the situation
with regard to : it is ajāta “without birth” and consequently it is “without
death”.23 That is to say: no one is born there, and therefore no one dies there.
To give another example. The concept of the pratyeka-buddha is well known as the
middle element in the triad: buddhas, pratyeka-buddhas and śrāvakas, and the usual
translation “a buddha (awakened) for himself” satisfies most people. This usage of the
word pratyeka (pacceka in its Pāli form) is, however, unusual, and it is not at all clear
how it could have the meaning given to it. When consideration is given to the fact that the
concept of a similar type of Buddha is also found in Jainism, but there, in Jain Prakrit, the
name is patteya-buddha, then it becomes possible for the philologist to suggest an
etymology for the word which can explain both the form in Prakrit and the form in Pāli—
explaining the form in Sanskrit presents no difficulty since it is merely a Sanskritisation
of the Pāli form or something similar to it—and on the basis of that explanation to
provide a more satisfactory translation.24
The basic work of philology is, of course, going on all the time when we consider
Buddhist texts. We have to ask ourselves “do we know what each word is” before we can
ask “what does each word mean”, and start to think about translating the words or
interpreting them. In many cases we can see that the understanding of the structure of a
21 See Norman, 1994, 209–24.

22 Kalupahana, 1986, 161 (ad Dhp 21).
23 See Norman, 1991D, 1–11 (7) (=CP IV, 251–63 [261]).
24 See Norman, 1983A, 92–106 (=CP II, 233–49).

particular word had already been lost before the commentaries were written, because it is
clear that the commentators do not understand its structure. In this connection, the
discovery of Sanskrit and Prakrit Buddhist texts in Chinese Turkestan, Afghanistan and
India in the course of the last hundred years has provided an invaluable tool for the
philologist. Quite often these new discoveries date back to a time before the confusion or
corruption or whatever it was came into being; sometimes they present a corruption of a
different nature which enables us to suggest a solution to the problem. A close
comparative study of (say) the Pāli Dhammapada, the Gāndhārī Dharmapada, the so-
called Patna Dharmapada and the Sanskrit Udānavarga is very rewarding, since it enables
us to use each and every version as a control upon the others, and we can quite often
identify the corruption—even when, in the absence of parallel passages, no corruption
was hitherto suspected—and, more important, suggest how it came into being.
Such philological studies have enabled us to identify nominal and verbal forms which
were no longer understood, to identify verbal roots which are not otherwise attested in the
language,25 and with our newly-found knowledge to correct corrupt readings and restore
something which we hope might approximate to the original reading, and make better and
more accurate translations as a result. It is very regrettable that many of these philological
discoveries are still unknown to, or are perhaps ignored by, some translators, who
consequently may well make mistakes in their translations, and in their understanding of
Buddhism based upon those translations.
For example, philological investigation has shown that in Pāli the ending which is
identical in form with the accusative singular can sometimes, although very rarely, stand
also for the accusative plural or the ablative singular.26 Nevertheless, some recent
translations reveal no knowledge of an accusative plural with this ending (in - ) in
Pāli. There is in the section about the wise man ( -vagga) in the Dhammapada a
verse27 about abandoning the , literally “the black dhamma”, and
cultivating the [ ], “the white (or bright) dhamma”. Some translators
take these forms as singular, and translate “Having forsaken a shadowy dhamma, the
wise one would cultivate the bright”,28 or “A wise man should develop a bright character
abandoning the shady one”,29 or “Let the wise man leave the way of darkness and follow
the way of light”,30 whereas the commentary on the verse where it occurs elsewhere in
Pāli,31 makes it clear that it is referring to the very common categorisation of dhammas,
in the plural, as akusala and kusala. I mentioned just now parallel versions of the Pāli
Dhp, and if we examine the forms of the verse which occur in them,32 we find that they
have plural forms.
25 See Norman, 1989B, 29–53 (52, note 139) (=CP IV, 92–123 [122, note 1]).
26 See Lüders, 1954, §§ 188–219.
27 vippahāya, bhavetha , Dhp 87ab.
28 Carter & Palihawadana, 1987, 170.
29 Kalupahana, 1986, 121.
30 Rādhakrishnan, 1950, 87.
31 ti, akusala- . sukkan ti, kusala- , Spk III 132, 20–21 (ad S V 24, 21).
32 kinhe dhamme viprahāya śukre bhāvetha , Patna Dharmapada 263ba;
viprahāya bhāvayata , Udāna-v 16.14ab.
Such translations seem even more remarkable when translators reveal knowledge of
parallel versions, but do not take advantage of their knowledge. There is in the Flower
Section of the Dhammapada33 a verse which describes how a bee takes nectar from a
flower and then flies away without harming it. The word for “flower” ( )
appears to be in the accusative case, and this has caused problems for some translators,
who find it difficult to fit an accusative form into the sentence. Recent translators34 state
that “from a flower” would be a better translation for , and they point to the
existence of the ablative forms and in the parallel texts, but do not follow
their own suggestion in their translation. They do not say that might actually
be an ablative, and they show no hint of any knowledge of the existence in Pāli of an
ablative singular in - .
Another way in which philology, the study of why words mean what they do, can be
helpful is that we sometimes find that those who made the first English translations of
Buddhist texts gave a particular meaning to a word which we have for the most part
followed without change ever since. When we come to look at the words themselves we
find that the meanings which we have accepted for so long are very often not the only
possible meanings but in some cases not even the most likely meanings. Take for
example, the phrase “noble truth”, which I mentioned a few minutes ago. It has become a
commonplace to talk about the four noble truths, and this is a perfectly acceptable
translation of the compound ariya-sacca: ariya means noble and sacca means truth, so
ariya-sacca means noble truth. This translation is so common and so fixed in our minds,
that it seems almost like blasphemy to have to point out that not only is this not the only
possible translation, but it is in fact the least likely of all the possibilities. If we look at the
commentators we find that they knew this very well. They point out that the compound
can have a number of meanings. It can mean “truth of the noble one”, “truth of the noble
ones”, “truth for a noble one”, i.e. truth that will make one noble, as well as the
translation “noble truth” so familiar to us.35 This last possibility, however, they put at the
bottom of the list of possibilities, if they mention it at all. My own feeling is that it is very
likely that “the truth of the noble one (the Buddha)” is the correct translation, although
we must never lose sight of the fact that in Indian literature multiple meanings are very
often intended, so that it is not always possible to say that there is a single correct
meaning.
To give another example: we all know at least the title of the text called
, and following Burnouf we have always translated this as “the
lotus of the good law”. Once again this is not the only possibility. It could, for example,
be “the good lotus of the law”, or “the lotus of the law of the good one”, or “…of the
good ones”. Since in fact we find that there is in the Dhammapada a verse talking about
the good teaching the dhamma to the good,36 we could very reasonably translate the title
33 yathāpi bhamaro paleti rasam ādāya, Dhp 49 abc.

34 Carter & Palihawadana, 1987, 443, note 14.
35 See Norman, 1990B, 1–13 (=CP IV, 170–74).
36 satañ ca dhammo na upeti, santo have sabbhi pavedayanti, Dhp 151 cd.
as “The lotus of the doctrine of the good one or of the good ones”, i.e. the Buddha or the
Buddhas.
Another place where the early translators were probably wrong is in the translation
which is given of the title of the text called the Mahāparinibbānasutta. Rhys Davids
translated this as “The book of the great decease”. This is possible, although we might
wonder what would constitute a small decease. We can, in fact, be fairly certain that the
translation is not correct. We find quite commonly in the Pāli canon pairs of suttas, one
called mahā “great” and the other “small”, so that we might have a large sutta on a
particular subject and a small sutta on the same subject. And we may assume that
originally the small version was in fact shorter than the large one, and represented a
contraction of the large, or alternatively the large represented an expansion of the small
version. It does not, however, in the forms in which we have them, always work like that.
Sometimes the small sutta is larger than the large sutta, and we must assume that further
contraction or expansion has taken place since the suttas received their names. It is
probable, therefore, that a more correct translation of Mahāparinibbānasutta is “the large
text about the decease”, and it is just chance that we do not find a
in the Pāli canon.37 I may note in passing that Rhys Davids’
translation of the title is all the more strange because the preceding sutta in the Dīgha-
nikāya is the Mahānidānasuttanta, which he correctly translates as “The great discourse
on causation”.
Those who mistranslate in such a way are in good company, because there is another
blatant, if I may be permitted to use the word, mistranslation of a text title of exactly the
same type, namely that adopted for the Mahāprajñāpāramitāsūtra, which that great
Belgian scholar Étienne Lamotte translated as “Le traité de la grande vertu de sagesse”
instead of “Le grand traité de la vertu de sagesse”. To do Lamotte justice we must
remember that he himself realised his mistake some time before his death, but at a time
when it was not deemed feasible to change the title.
Another important point which philology has been able to clarify is the translation of
the noun parinibbāna which Rhys Davids translated as “decease” in the title of the sutta I
have just mentioned. Because that sutta deals with the nibbāna which the Buddha
obtained at death, parinibbāna is often translated as “final nibbāna”, reserving the simple
“nibbāna” for the experience which the Buddha had at the time of his awakening.
Because of the use of the word parinibbāna in connection with the Buddha’s death, it is
assumed by some that it can only be used of nibbāna at death. This interpretation is,
however, based upon a misunderstanding of the significance of the prefix pari-. It can be
shown very easily that it does not imply “final”, and parinibbāna, at least in its original
usage, cannot mean “final nibbāna” because there are many references in the texts to
living beings who are described as parinibbuta “having gained parinibbāna”.38
Another example of an on-going mistranslation is the word kusala, which I mentioned
a few minutes ago. This is usually translated “skilled” or “skilful”. The word certainly
37 See Norman, 1983C, 36.

38 In the Theragāthā we find that Dabba refers to himself as parinibbuto (Th 5) as does Ānanda
(Th 1047). In the Therīgāthā we find the same of Ubbirī (Thī 53), and the Pañcasatā (Thī
132).
has this meaning, but in Sanskrit it is the secondary meaning, the primary meaning being
“good”. In the Pali Text Society’s Pali-English Dictionary, however, “clever”, “skilful”,
and “expert” are given as the first meanings, although its opposite akusala is given only
the sense of “bad, evil”. The two terms kusala and akusala are frequently combined as
epithets of the dhammas “the (mental) phenomena”, sometimes translated as “states”, and
we often read in translations of the abandoning of “unskilled states” and the arising of
“skilled states”, although translators rarely try to explain what a skilled or unskilled state
is. The commentaries make it clear that we are, in fact, dealing with good and bad
(mental) phenomena.
Publication of the Pali-English Dictionary was completed in 1925, so the last fascicles
are nearly 70 years old, while the earlier fascicles are more than 70 years old. At the time
of its publication it represented the summit of philological achievement in the field of
Pāli studies, but even so it contained numerous errors, as well as misleading statements of
the sort I have just mentioned. Since 1925 editions of many hitherto unpublished Pāli
texts have appeared, so that it is woefully incomplete, while better editions of many of
the texts which are referred to in it have been published, so that many corrections of
forms and meanings are needed. The Pali Text Society is fully aware of this, and a
revised version of the Pali-English Dictionary, to be called the New Pāli-English
Dictionary, is in preparation.39
Until that is published, the Pali-English Dictionary is the best complete dictionary of
the Pāli language we have, and despite all its failings it is still used by most Pāli scholars.
I have already mentioned the way in which would-be translators have perhaps looked
words up in the dictionary and have ascertained meanings for them, and then have used
these meanings as the basis for their intuition. It is, however, difficult to persuade many
scholars that they must be very wary of taking everything they find in the dictionary as
infallible. I remember asking one scholar how he proposed to solve some of the problems
which I knew existed in a text he was proposing to translate. He told me that he would
give every word in his text the meanings given in the PED, and he asked me very
indignantly how I expected him to translate the text if the PED meanings were not
correct.
These problems arise because there is a strange sanctity attached to the printed word.
In the same way that people used to say, “It must be true, because I read it in The Times”,
so when I offer critical comments on suggestions which people send me, and hint that
their translation of some Pāli word or phrase should be changed, they reply that I must be
wrong, because they took whatever it is they have written from the PED. When I come
back with the retort, “Yes, I know it is in the PED, but it is wrong, and we are going to
change it in the revised edition”, they are still very reluctant to believe that anything
which has appeared in print can be incorrect. I do not know what sort of persons they
think the editors of the first edition were, but they have certainly imparted some sort of
infallibility to them. It is quite clear that they have never read the comments about Rhys
Davids in the Afterword to the Dictionary, which William Stede wrote when the final
39 It is being prepared by Dr Margaret Cone, Assistant Director of Research in Pāli Lexicography,

in the Faculty of Oriental Studies, University of Cambridge.
fascicle appeared, some three years after Rhys Davids’ death: “His mind was bent on
other aims than dictionary work, which was not his strongest point”, or Stede’s statement
about the joint editors’ motto: “Better now and imperfect, than perfect and perhaps
never!”40
There is also, I believe, a reluctance to believe that present-day scholars, mere mortals,
can improve on anything in the Dictionary. I can only hope that when the revised edition
appears it too will gain the veneer of infallibility which the printed word imparts. I am
encouraged to think that it will do so, because I find that the same sanctity is becoming
attached to A Critical Pāli Dictionary, not simply the early fascicles produced by Helmer
Smith and Dines Andersen, who were perhaps the greatest Pāli philologists that Europe
has produced, but also more recent fascicles. In something that I was writing recently I
had occasion to quote from an article I had earlier written for A Critical Pāli Dictionary,
as part of my involvement in the dictionary. However, in the time since writing the article
I had thought of a way of expressing my solution to a particular problem in a slightly
different way. I showed what I had written to a colleague, and he returned my material to
me with the comment that he thought I should quote what A Critical Pāli Dictionary said
about this problem.
I could go on at some length about the light which a philological approach has shed
upon many of the key words in Indian Buddhism, with a resultant improvement in our
understanding of such terms, and a greater accuracy in the translations we can give for
them, but I have probably said enough for one lecture.
40 PED, 738.
II
Buddhism and Its Origins
In the first lecture I gave a general account of the way in which I personally became
involved in the philological study of Buddhism, and I gave some illustrations of the way
in which philologists have been able to shed light upon a number of important concepts
in Buddhism, and as a result of the shedding of that light to understand the concepts
better, to give better translations for them, and to show more clearly how they fit into the
general pattern of Buddhism.
I want in this and in the remaining lectures to deal with more specific aspects of
Buddhism, and to continue to show how, if we approach them through the medium of
philology, always asking “why or how do these words get their meaning”, we may be
able to add something to the picture which we get through other means of approach.
Since I shall in these lectures be giving translations of many of the terms I shall
discuss, it would perhaps be sensible to start now by giving some idea of the approach to
translation which I have evolved over the years.
It is very difficult to give a one for one translation of Sanskrit and Pāli words into
English. It is very rare that one Sanskrit or Pāli word has exactly the same connotations,
no less and no more, as one English word. This means that if I wish to give a more
adequate translation I am forced to give a phrase in English, or perhaps even a whole
sentence, or in the case of a very difficult word with a wide range of connotations, even a
whole paragraph. Consequently, if I am translating a Buddhist text into English, it is very
difficult to produce something which approximates closely to the meaning of the original,
and yet appears in good, clear, concise and readable English.
It is for this reason that many translators do not translate the difficult words, but leave
them in their original Sanskrit or Pāli form. And this is fine, for them. As they read
through their translations, every time they come across a Sanskrit or a Pāli word, they
know, within limits, what it means and they can mentally substitute that meaning. It is
not, however, so good for the rest of us, who have little idea of what these scholars have
in mind, since their translations may consist of little more than strings of Sanskrit, Pāli or
Tibetan words linked together with “ands” and “buts”. Reviewers sometimes complain of
translations which are so literal and so full of foreign words that they hardly read as
English.1 Such translations are of little value, and one might just as well leave the whole
thing in the original. Only an expert can understand all the words left in the original
language, and the expert needs no English words at all.
1 See Gombrich’s review of Cousins (et al.), 1974, in JRAS 1977, 132–33.
Buddhism and its origins 19
And so what I have favoured over the years, when translating, is to leave a minimum
of these difficult terms in their original form, but the first time they occur, to include a
note as lengthy and as detailed as I think is required, giving some idea of what I think the
word means, and why I think it means it. In effect I am saying that, every time the reader
comes across this word thereafter, he must remember to consult the relevant note to find
out what it means in the context. This system does have defects. As anyone who has
looked at any of my translations knows, the actual translation is only a small proportion
of the book. The notes are far longer, and then there are all the indexes which will enable
the reader to find where I dealt first with the problematic word.
However, in the context of lectures such as these, that system is clearly not possible,
and so I shall have to give a single word translation, for the technical terms, etc., that I am
talking about, but this will be simply for the purpose of identification. In no way do I
mean to imply that the one word translation I give is an adequate translation of the
technical term we are discussing. And, with that proviso, I will go on now to talk about
the information that philology can give us about Buddhism and its origins.
We are all familiar with the account of the origin of Buddhism which we find in the
Indian tradition. The Buddha-to-be was the son of an Indian king. Despite his father’s
attempts to ensure that his son should see no sign of old age, sickness or death, he
became acquainted with the suffering existing in the world and the advantages of the
ascetic life by seeing four signs at the age of twenty-nine. He left his wife and new-born
son and became a wanderer. He tried severe ascetic practices, and followed various
teachers, but found that he could not obtain the goal he was seeking. By meditation he
obtained , and then began to teach to others the way which he had found to be
successful, beginning with those with whom he had earlier practised asceticism.
What light can a philological approach throw on this narrative ?
It goes without saying that the origins of Buddhism lie in the political, economic,
social and religious environment of the time.
The political and economic picture which we gain from early Pāli texts is one where
the urbanisation of the Gangetic plain was well under way. There were large, well
fortified cities, with powerful rulers. Movement between those cities was easy, and trade
between them was flourishing. We read of merchants setting out with large caravans, and
there are frequent references to the coinage which must have facilitated the growth of
trade.2 The , the ruling class, had gained a more imposing identity—they were
no longer minor chieftains—and the vaiśyas, the mercantile class, had begun to gain
wealth, and power arising from that wealth, as opposed to an earlier situation where they
were merely itinerant traders. This must have lead to a situation where the and
the vaiśyas would be very open to a religion which gave them a social position equal to,
or even superior to, that of members of the brahmanical caste. It is very clear that the
brahmanical caste was regarded as superior, at least by the members of that caste. As a
the Buddha might be expected to oppose the brahmanical caste and much of his
teaching was devoted to defining the word as a moral term, and denying that
one became a brahman simply by birth. He insisted that it was actions which made a
2 Note the story of and the purchase of the Jetavana (Vin II 158 foll.).
brahman. A consequence of the Buddha’s teaching about this was that (although there are
many references to becoming his followers3) the main support for his religion
came from and vaiśyas, particularly from the latter. This was presumably
because they were wealthy and were well placed to gain merit by dāna “giving,
generosity”. They also travelled widely, and were able to act as missionaries, taking the
message to other vaiśya communities. It is a striking fact that, as Buddhism spread, it
followed the trade routes, being propagated either by “missionary” traders or by bhikkhus
who travelled with the caravans under the protection of the traders.
The Buddha was born in Nepal, and his name was Siddhattha. The traditional story
states that his father Suddhodana was a king, that is to say a rājā. This word, however,
may mean nothing more than a man of the royal tribe or the military caste, i.e. a
,4 and in this context, in a place some way away from the Gangetic plain, it is
probable that it still meant a minor tribal chieftain, at the head of the Śākya clan.
Siddhattha’s gotra name was Gotama, but Gotama is not a name, so it probably
represents a borrowing of the family purohita‘s gotra name5. This suggests that the
Śākyas were a fairly recent entrant into the caste system, which in turn suggests that
perhaps the Buddha’s family was not in origin Indo-Aryan. There are other examples of
clans or tribes being assimilated into the caste system in a comparable way.6 It has been
pointed out that important parts of the commentarial tradition concerning the Buddha’s
family relations followed Dravidian marriage patterns, which is taken, by some, as
proving that the commentarial tradition must have been composed in areas where
Dravidian marriage patterns prevailed, i.e. in the southern half of India or in Sri Lanka.7
The tradition, it is suggested, must therefore be a late story from South India. It is,
however, quite arguable and, I think, more likely that it represents the actual clan
relationships at that time among non-Indo-Aryan tribes in the North, or among tribes
which had until recently been non-Indo-Aryan.
What do we know about the religious beliefs of the Buddha’s contemporaries, those
beliefs which he found around him? When we come to examine the teaching of the
Buddha we find that we can make certain deductions about those beliefs from his
reactions to them. Some beliefs he accepted as they were, others he accepted nominally,
but gave a changed meaning to them, others he opposed outright.
There are, for example, traces in the Pāli texts of a belief which we may assume goes
back to an early date. We find references to what one does in this world, and what one
consequently experiences as a result of that action, when one has “passed away” (pretya).
For example, when one has passed away one who has done good rejoices, and one who
has done evil laments.8 This seems to go back to an earlier reward-and-punishment idea
of the after-life. There is no implication that this rejoicing or lamentation will be endured
again and again for the whole of eternity. A simple reference in a Pāli sutta9 to going
“beyond this shore and the far shore”, caused great problems for the commentators. My
3 For an assessment of this, see Tsuchida Ryūtarō, 1991, 51–95.
4 See MW, s.v.rājan.
5 See Brough, 1953, 5, note 3.
6 e.g. the Rajputs into the caste.
7 See Trautman, 1981, 316–30, referred to by Gombrich, 1992, 161.
8 See Dhp 17–18.
9 Sn 117.
personal belief is that this statement was first formulated in a situation where the author
was considering two stages only, i.e. this world and the afterlife, rather than the endless
stream of . The commentators,10 however, found the statement difficult to
explain, because when they wrote many centuries later, this shore and the far shore meant
and , and to pass beyond was a Mahāyāna idea which had no
place in a Theravādin text.
At the same time we find abundant evidence that this earlier stage of religion had been
replaced, or at least complemented, by a belief in , the journeying on in a series
of existences with no beginning and no end, with the precise nature of each existence
dependent upon the actions which one had done in a previous existence, i.e. one’s karma.
The origin of this doctrine of reincarnation is not clear. There are those who think that the
Indus Valley civilisation was the source. It can nevertheless perhaps be explained as a
development of the old idea that there was reward or punishment at the end of life. It is
possible that the idea grew that the reward or punishment could be another (good or bad)
existence, and if, at the end of that existence, there was an imbalance in the reward or
punishment, yet another existence, and so on and so on.11 Others think that the idea was
imported into India from some other culture, perhaps from the Greek cities of Asia
Minor, where we know that the idea of a cycle of births was current. Whether there was
any connection between these two areas and, if so, which way the borrowing went, is
dependent to some extent upon the dates which we assign to the first appearance of the
idea in both areas.
The Buddha accepted the Vedic gods, but although he accepted their existence, he
denied them any causal role in the universe. When they had been deprived of their causal
role, there was no longer any point in sacrificing to them, and they became simply super-
men, enjoying the benefits of the good karma they had amassed in a previous existence.12
The Buddha made it clear that the devas, like all others in , were subject to
death and rebirth. As super-men, they had a longer life-span than men, and Sakka, the
king of the devas, had a longer life-span13 than the other devas. Nevertheless, there was
an important difference between devas and men. When a man died, that was the end of
him in that existence; at the death of a king, his successor, a different person, took his
place. Not so with the devas. When Sakka dies, as he must do,14 being finite, his place is
immediately taken by another Sakka.15
There are references to an individual being reborn as Sakka a number of times. The
Buddha, for example, stated that he had been reborn 36 times as Sakka.16 The
10 For the cty see Brough, 1962, 202 and Pj II 13, 1 foll.
11 See Collins, 1982, 46–47.
12 Norman, 1977, 329.
13 Sakka’s lifespan is said to be ca vassasatasahassāni tisso ca (Ja II 312,
19–20).
14 Sakko hi indo aparimutto jātiyā jarāya sokehi…aparimutto dukkhasmā ti
vadāmi (A I 144, 24–26).
15 Sakko…cavi, añño Sakko nibbatti, so pi kāretvā āyukkhayena cavi. eten’ upāyena
Sakkā (Ja II 312, 19–22).
commentator Buddhaghosa tells a story that Sakka once died while listening to a
discourse from the Buddha, and was immediately reborn again as Sakka,17 so that he
could continue to hear the sermon. It is clear that the singularity of this occurrence was
not lost upon Buddhaghosa, for he proceeds to explain how it was that this was not
noticed by the other devas.18
We find echoes of statements in the Buddha’s sermons, and it would
therefore seem likely that any technical terminology he employed which has parallels in
the would be heard by those who were already conversant, if only to a
limited extent, with the usage. It is, for example, clear from the way in
which the Buddha was able to assume that his hearers understood such concepts as nicca
“permanent”, anicca “impermanent”, sukha “happiness”, and dukkha “misery”,19 that
they had already heard teachers speaking about such things. In the absence of any
evidence to the contrary, it would therefore seem likely that any mention the Buddha
made of attā “self” and anattā “not self” would be interpreted, rightly or wrongly, by his
hearers in the light of contemporary usage, and that, as far as we can tell, is
usage.
This being so, it is hard to see why almost all writers about Buddhism accept the
statement often made that the Buddha makes no mention of the concept of a
Universal Self, an ātman or brahman.20 When the Buddha stated that everything was
anattā “not self”, we should expect that the view of attā “self” which he was denying was
that held by other teachers at that time. We can, in fact, deduce, from what the Buddha
rejected, the doctrine which the other teachers upheld. Once we know that the Buddha
was using words in this way, then we are aided in our attempt to understand them and
translate them.
The Buddha vigorously denied the brahmanical idea of the existence of the ātman, the
idea that there is no difference between us and a world spirit, the standard advaitavāda
“non-dual” doctrine,21 which is expressed in the with the words tat tvam asi
“You are that”. You are that. You are identical with that world spirit—the view that we
all have a portion of that spirit in us and when everything which hides that identity is
removed then we can be absorbed into ātman/brahman.
The Buddha, on the other hand, specifically condemned the view that the world and
the self were the same thing, and that after death one might become permanent, lasting,
eternal and not liable to change, and he rejected the idea that one could look at the
various aspects of the world and say “that is mine, I am that, that is my self”, which is a
clear echo of tat tvam asi, expressed from a different point of view. When the Buddha
said: “rūpa ‘form’, etc., are not mine”,22 he was denying the view that there is no
16 devindo (A IV 90, 5).
17 Sakko pana sotāpanno jāto sotāpanno va hutvā, Bhagavato purato yeva cavitvā -Sakko
hutvā nibbatti (Sv 732, 29–31). The same episode is referred to at Dhpa III 270, 15–16.
18 hi attabhāvassa nāma na paññāyati, dīpasikha-
viya hoti. tasmā sesa-devatā na (Sv 732, 33–34).
19 See Norman, 1981, 19–29 (22) (=CP II, 200–209).
20 See Thomas, 1949, 35.
21 See Potter, 1981, 6.
22 M I 136, 6.
distinction between knower and known.
The Buddha’s rejection of the existence of the attā, i.e. his view that everything was
“not self” (anattā), was based upon the brahmanical belief that the ātman was nitya
“permanent” and sukha “happiness”. Hence the Buddha could refute this by pointing out
that the world, which was supposed to be part of ātman, was in fact anicca
“impermanent” and dukkha “misery”23—his belief that the world was dukkha was, of
course, the first noble truth.
The Buddha’s teaching about this is, however, not always understood. The word
anattā is sometimes translated “having no soul”, and various things which are specified
to be anattā are thought of as having no soul. We can see the need for better
understanding of this vital concept of Buddhism when we survey the range of translations
given for the phrase sabbe dhammā anattā which occurs in the Dhammapada24 and
elsewhere: “All forms are unreal” (Max Müller);25 “All the elements of being are non-
self” (Radhakrishnan);26 “All things are not self” (Acharya Buddharakkhita);27 “All
phenomena are non-substantial” (Kalupahana);28 “All things are ego-less” (Jayasekera);29
“All dhammas are without self”, says a very recent translation (Carter and
Palihawadana).30
Translations such as “without self” and “having no soul” cannot be correct, because
the grammar and syntax show that anattā is not a possessive adjective, which it would
need to be to have such a meaning. It is a descriptive compound, and if the correct
translation for attā is “soul”, then the word would mean that these various things are “not
soul”. This, however, cannot be correct, because the Buddha sometimes exhorted his
followers to regard these things as parato, i.e. “as other”. We cannot, however, co-relate
“as other” and “not soul”. It is clear that the only translation which it is possible to co-
relate to “as other” is “not self”. We are not to regard these things as part of the self, and
to clarify the point the Buddha asked his followers whether, when they saw wood being
burned, they felt any pain. The answer was “No”, and the explanation given was that they
did not feel any pain because the wood was not part of the self.31 It was other than the
self. We might regard the Buddha’s refutation of the ātman/brahman idea as being
somewhat empirical—as was Dr Johnson’s refutation of Bishop Berkeley’s theory of the
non-existence of matter—but it was no less effective for all that.
The Buddha also rejected the alternative brahmanical view of ātman as brahman.
There seems to be no occurrence in Pāli of the uncompounded neuter word brahma32 in
the sense of the brahman, but brahma is used in compounds, apparently in
23 See Norman, 1981, 22, and Gombrich, 1990, 14.

24 Dhp 279.
25 Müller, 1881, 69.
26 Rādhakrishnan, 1950, 147.
27 Buddharakkhita, 1985, 52.
28 Kalupahana, 1986, 139.
29 Jayasekera, 1992, 92.
30 Carter & Palihawadana, 1987, 312. Strangely enough they translate anattā in the gloss as “not
self”.
31 M I 141, 11.
32 For the occurrences of brahman in the Pāli canon, see Bhattacharya, 1989.
the sense of “excellent, perfect”. In its basic brahmanical sense brahma-carya means “the
practice of a ”, i.e. the living of a celibate life, learning the Vedas. The
Buddha used the phrase in the more general sense of “to live the best life, i.e. a holy,
celibate (or in the case of married couples, a chaste and moral) life”. In the
brahma-patha means “the way to brahman or Brahmā”. The Buddha used it in the sense
of the way to the best, i.e. nibbāna, and it is explained as being the same as
brahmavihāra.33
It is possible that brahma-vihāra was in origin a brahmanical term.34 It would literally
mean “dwelling in brahman or with Brahmā”, although it is not attested in that usage in
Sanskrit. It perhaps shows a trace of its original meaning in a sutta35 in which the Buddha
speaks to young brahmans who were disputing the correct way to obtain brahma-
sahavyatā. In the context this would seem to mean “union with brahman”, but the
Buddha, perhaps jokingly, interprets it as meaning a state of union with the god Brahmā.
He explains that someone who practises the four types of concentration36 called brahma-
vihāra is reborn as a Brahmā in the Brahma-world.37 It is to be noted that this means only
being born in the same heaven as Mahā Brahmā, not union with the
brahman.
Contemporary with the Buddha there was a growth of non-brahmanical

“ascetic” movements, and we find in Buddhist texts a list of names of six teachers and
something about their teachings.38 This seems to be a very old list because the texts are
not consistent about which beliefs they ascribe to which teacher,39 and as we have them
they are an unreliable guide to what was really going on at the time of the Buddha. We
can, however, see that some of the teachings were a reaction to the idea of , the
endless series of rebirths. We can then see a pattern of development of thought in Indian
religion: first, the view that there is a single existence at the end of which one was judged
and punished or rewarded; then a view that there is an endless series of punishments or
rewards in an endless series of existences; this led on to attempts to gain release from this
endless series. We can see that some teachers taught a way out of it, by specifying that
is finite, so that, when we have finished a certain number of rebirths, that will
be the end. Others taught various methods of gaining “release” from the endless
series of rebirths.
The Buddha’s way to release, as we shall see, was by means of meditative practices,
and this method followed closely and was developed from, it seems, the teachings of
33 Brahma-pathe ti catubbidhe pi brahma-vihāra-pathe, brahme va phala-
samāpatti-pathe samāpajjana-vasena . Th-a III 9, 9–11 (ad Th 689).
34 Thomas, 1949, 126.
35 The Tevijja-sutta (D I 235–53).
36 mettā, , muditā, and upekkhā.
37 so cattāro brahma-vihāre bhāvetvā kāyassa bheda Brahmalokūpago ahosi, D
II 196, 7–8.
38 See D I 52–59.
39 See MacQueen, 1984, 291–307.
other . The tradition tells us that he went to two teachers, Kālāma and
Uddaka Rāmaputta, but he left them both after a short while because their teachings did
not lead to the goal he desired, and they had only reached certain stages in their
meditative practices. He went away, and one version of the tradition40 tells us that he
remembered that as a boy he had entered the first jhāna, the first state of meditation. This
would seem to presuppose that there was a series of meditative practices which had
already been categorised and numbered or—perhaps more likely—that the Buddha was
employing a later categorisation and imposing it on an earlier occurrence, i.e. he had had
some sort of meditative experience, as a boy, which he equated with the first stage of the
code of meditative practices we read about later on. Repeating his boyhood experience,
the Buddha then went on to a second and a third and a fourth jhāna. I would personally
doubt that at this early time the four jhānas were so rigidly delineated. I would assume
that his meditative experience simply flowed on, and it was only later, when the Buddha
came to teach these meditations to his followers, that they were codified and categorised
as the four rūpa-jhānas “the meditations about form”.
From the fourth jhāna he gained bodhi. It is not at all clear what gaining bodhi means.
We are accustomed to the translation “enlightenment” for bodhi, but this is misleading
for two reasons. First, it can be confused with the use of the word to describe the
development in European thought and culture in the eighteenth century, and second, it
suggests that light is being shed on something, whereas there is no hint of the meaning
“light” in the root budh- which underlies the word bodhi. The root means “to wake up, to
be awake, to be awakened”, and a buddha is someone who has been awakened. Besides
the ordinary sense of being awakened by something, e.g. a noise, it can also mean
“awakened to something”. The desire to get the idea of “awakened” in English
translations of buddha explains the rather peculiar Victorian quasi-poetical translation
“the wake” which we sometimes find.41
It is not clear what the Buddha was awakened to, or at what particular point the
awakening came. In some texts he stated that he was awakened to the destruction of the
āsavas “the influxes”. He was therefore “one who has destroyed his āsavas”,
an epithet of an arahat. Elsewhere the Buddha said that he was awakened to the
knowledge and insight that this was his last existence.
The shortest account of the Buddha’s bodhi in Pāli is that found in the
Ariyapariyesanasutta,42 and for that reason some scholars believe that this is the earliest
account available to us.43 We may assume that in the shortest account of his bodhi the
Buddha would deal with the most important part of the experience, and it appears fom
this version that this was the gaining of nibbāna. This view is supported by the fact that
he left the teachers he had before his bodhi, because their teachings were inadequate, as
they did not lead to nibbāna. Of each of them he said, “This doctrine is not conducive to
disgust (with the world), nor dispassion, nor cessation, nor quiescence, nor super-
knowledge, nor awakening, nor nibbāna”.44
40 In the Mahāsaccakasutta (M I 237–51).
41 See C.A.F.Rhys Davids, 1909, v.
42 M I 160–75.
43 See Norman, 1990A, 24–35 (33, note 17) (=CP IV, 124–38 [127, note 1]).
44 dhammo nibbidāya na virāgāya na nirodhāya na upasamāya na abhiññāya na
sambodhāya na nibbanaya , M I 165, 10–12=166, 29–31.
We may deduce from this that the concept of the attainment of nibbāna existed, even
though the Buddha-to-be and his teachers were unable to achieve it, that is to say that
people knew that there was a state to which they gave the name nibbāna, even though
they could not attain it. Whether this meant that some had indeed already attained it, but
had not passed on their method to others, or whether it was a concept of some sort of
utopia which had been proposed and named, and towards which men were struggling, is
not clear.
We may also deduce that the words in the Buddha’s statement are in the order in
which the various states mentioned in it are to be realised, starting with disgust with the
world, and going on to awakening and nibbāna. This would support the belief that the
Buddha’s aim was to free himself from , and all aspects of his teaching were
concerned with the acquisition of means to do this, either in this life or a later one, and
with finding out how best to dwell in until release was obtained.45
There is an interesting point which arises in connection with the four jhānas which the
Buddha practised at the time of his bodhi. I have already mentioned the account of the
Buddha’s pre-bodhi visits to Kālāma and Uddaka Rāmaputta. With them he
practised meditation and reached “the state of nothingness” (ākiñcaññāyatana)46 and “the
state of neither perception nor non-perception” (nevasaññā-nāsaññāyatana),47
respectively. It is strange and noteworthy that, although he rejected both of these as not
leading to nibbāna, nevertheless in his own teaching after his bodhi he included them as
stages on the way to nibbāna. As taught by the Buddha, they are the third and fourth of
the arūpa-jhānas “the meditations about non-form”, i.e. they are devoted to the
contemplation of things outside the physical world of form in which we live, and they are
therefore the seventh and eighth of the samāpattis “attainments”, since they come after
the four rūpa-jhānas “the meditations about form”. It would appear, then, that the
Buddha had already attained the first four attainments with those teachers before he
gained the seventh and eighth attainment, and we have the statement of the commentator
Buddhaghosa to this effect.48 This would make the story of his boyhood memory seem
very strange, and we should perhaps follow the view that the four rūpa-jhānas and the
four arūpa-jhānas were originally two quite separate sets of states of meditation.
It would seem very probable that the four arūpa-jhānas were not discovered by the
Buddha, and were not in origin Buddhist, and that is why they were included in the
accounts of the non-Buddhist teachers’ views. If the suggestion of some scholars that the
story of the Buddha being taught by these teachers has no historical basis, we must
conclude that the inclusion of a mention of the arūpa-jhānas in the Buddha’s life history
was intended to show that they were inadequate, when compared with the Buddha’s
method. They did, however, lead to a state which seems to be equal to nibbāna, which
presumably means that some, at least, of these non-Buddhist teachers had also succeeded
in finding a way out of . It was possibly because the arūpa-jhānas were
successful in gaining the desired end that they were incorporated into the Buddhist
45 Norman, 1990A, 24–35 (26) (=CP IV, 124–38 [128]).

46 M I 164, 15.
47 M I 165, 35.
48 ākiñcaññāyatanapariyosānā satta samāpattiyo jānāpesi, Ps III 171, 22–23.
scheme of jhānas, not as simultaneous means (which would have been better, because
they are really an alternative) but as consecutive.
In the Buddha’s accounts of the eight attainments, however, we read of a ninth state, that
of the “cessation of feelings and perceptions” or “cessation of the feeling49 of
perceptions” (saññāvedayitanirodha).50 In this state, for one seeing with perceptive
knowledge (paññā), the āsavas are destroyed (paññāya c’ assa disvā āsavā
honti).51 This would seem to imply that, if we equate the destruction of the āsavas
(āsavakkhaya) with nibbāna, this was another way of attaining nibbāna, and some
scholars believe52 that this state of the “cessation of feelings and perceptions”
(saññāvedayitanirodha) and nibbāna were originally identical.
We must, however, note that there is no reference to the four arūpa-jhānas in the
accounts of the Buddha’s own attainment of nibbāna at the time of his bodhi. In the story
of his death, in the Mahāparinibbānasuttanta of the Dīgha-nikāya,53 we read that the
Buddha went through all the stages of the rūpa-jhānas and the arūpa-jhānas, and then
entered the “cessation of feelings and perceptions” (saññāvedayitanirodha). He was then
thought by Ānanda to have attained nibbāna.54 Anuruddha, however, pointed out that he
had only attained the “cessation of feelings and perceptions”,55 which clearly, as far as
Anuruddha was concerned, was not identical with nibbāna, but was probably some sort
of death-like trance. From there the Buddha went back, in due order, to the first jhāna,
and then up to the fourth jhāna, from which he died, and presumably attained nibbāna.
The root budh- in Sanskrit means not only “to be awake, to be awakened”, but also
from the earliest texts onwards “to perceive, to notice, to learn, to understand,” and
buddha in non-Buddhist texts means “intelligent, clever, wise”. Bodha as an adjective
means “knowing, understanding”, and bodhi probably has the idea of “knowledge,
understanding”, possibly the knowledge that release from is possible. It is
also very likely that the knowledge itself is efficacious, i.e. “I know that it is possible to
be released and merely by knowing I am released”. This theory fits in very well with the
fact that immediately after his awakening the Buddha rehearsed the 12-fold pratītya-
samutpāda, the chain of interdependent elements, the so-called chain of dependent
causation—the arising of things dependent upon other things.
The 12-fold chain of causation is probably not the earliest form of the chain, and
certainly portions of it with less links are found elsewhere in the canon. In discussing the
49 Although saññāvedayita is usually translated as a dvandva compound, this is not necessarily

correct. Grammatically, it could as well be taken as a tatpurusa compound, with the past participle
vedayita being used as an action noun. This interpretation would depend upon the occurrence of
saññā with the verb vedayati. This combination seems not to occur in the Pāli canon as we have it
now, but it is possible that it existed at an earlier date, when the precise signification of technical
terms had not yet been fixed.
50 M I 165, 35.
51 M I 175, 3–4.
52 See Schmithausen, 1981, 249 (addendum to ch. H), quoting Nagasaki.
53 D II 72–168.
54 D II 156, 17.
55 D II 156, 18–19.
12-fold chain I am not implying that that was necessarily the form in which the Buddha
rehearsed it immediately after the awakening, but it is clear that that was the standard
form at some time or other. It is interesting because it seems to be a calculated attempt by
the Buddha to work out what he had just achieved. It goes, as is well known: avijjā
“ignorance” produces “compounded formations”, these produce
“consciousness”, this produces nāmarūpa “name and form”, this produces
“the six senses”, they produce phassa “contact”, this produces vedanā “feeling”, this
produces “craving”, this produces upādāna “clinging”, this produces bhava
“existence”, this produces jāti “birth”, and this produces “old age and
death”.
In other places the chain is set out in reverse order, and I think that that is how the
Buddha must have enunciated it when he was trying to explain what had happened to
him. It is a statement of where the Buddha was, and how he had got into that situation. It
ends, in the traditional form, with old age and death. That I think is where the Buddha
started. He started from the position in which he found himself: he knew that he would
grow old and die. Why was he going to grow old and die? Because he had been born.
Why had he been born. Because of existence. Why was there existence? Because of
clinging. Why was there clinging? Because of craving. Why was there craving? Because
of feeling. Why was there feeling? Because of contact. Why was there contact? Because
of the six senses. Why were there six senses? Because of name and form. Why was there
name and form? Because of consciousness. Why was there consciousness? Because of
the compounded formations. Why were there compounded formations. Because of
ignorance. That is to say: the beginning of all this existence, which we know is suffering,
is ignorance. We can now see what has happened to the Buddha. If any link in the chain
is removed, then whatever depends upon it cannot arise. It is very clear that if I am not
born, then I cannot die. If we look at the pratītya-samutpāda as a whole we can see that if
there is no ignorance, then there are no compounded formations, and therefore everything
else which depends upon the compounded formations will not arise. If instead of
“ignorance” we translate avijjā as “lack of knowledge”, then we can see that avijjā is
destroyed by vijjā “knowledge”, and knowledge is what the Buddha had just acquired.
His knowledge, his bodhi, has therefore destroyed all the subsequent links of the chain.
He has therefore destroyed future birth, and can exclaim triumphantly: “This is my last
birth. I shall not be born again”.
One of the most interesting aspects of Buddhism and one of the strong points of
Buddhism—one which undoubtedly appealed to converts who were and
vaiśyas—was that it presented two ways to salvation. This perhaps reflects the fact,
which I have already noted, that the accounts are not consistent in saying what the
Buddha was awakened to. One way, the immediate way, was the one which the Buddha
himself had employed, the so-called jhānic way, by meditation. This was in effect,
although not in theory, restricted to those who had abandoned the world to become
wanderers, and who had the time, and the inclination, to meditate. When the Buddha
began to preach, however, he preached about the four noble truths, not about the
destruction of the āsavas. The fourth noble truth is about the path which leads to the
destruction of suffering, and this was a more gradual way to release, making use of the
precepts of the eight-fold path to gain a better rebirth. This is the so-called kammic way.
One might hope in time, after entering the stream, by amassing good kamma, to get to the
point where the number of future birth in this world would be limited. Rebirths after that
would be heavenly rebirths, leading at last to release from .
The whole point about Buddhism is that by its very nature it requires these two
methods of attaining release. The system whereby some abandon the world and become
wanderers in the hope of gaining nibbāna in that same life can only work as long as there
are those who have decided not to abandon the world, not to become wanderers, but to
continue as householders and make donations to the community of begging wanderers,
the or bhikkhus. A system which depends upon dāna, “giving, generosity”,
depends upon there being those who are able to make dāna, to give generously, to be
donors.
Another of the other non-brahmanical movements was Jainism, or to be
more precise the teaching of Vardhamāna, known as Mahāvīra, or the Jina “the
conqueror”. Although we get a picture of Jainism from the Pāli canonical texts which
indicates that Jainism and Buddhism were strongly opposed to each other, of all the
śramana religions of which we have knowledge, Jainism comes closest to Buddhism, in a
number of ways. When Western scholars began to investigate Buddhism and Jainism in
the nineteenth century they found that the two religions had so much technical and other
terminology in common, that Jainism was in fact thought to be an offshoot of Buddhism,
although the precise meanings of such terms did not always coincide.56
One such word is āsava,57 for the Buddhist usage does not fit the etymology of the word,
while the Jain usage does. The etymology of this word (the preposition ā “towards” + the
root sru- “to flow”) implies something flowing in, and this suits the Jain usage well, since
there the āsavas are influences which flow into a person, and discolour his soul.58 We
find illustrations of this in Jain manuscripts, with people ranging from white, through
yellow, red, blue and green to black, depending on the amount of āsavas which has
flowed into them. This does not suit the Buddhist idea, where the āsavas are not
attributes which are capable of flowing into a person. They are, in fact, identical with the
four oghas “floods”,59 and it seems clear that in Buddhism the word has lost its original
meaning. So, although the translation “influence” or “influx” suits the Jain usage well, on
etymological and exegetical grounds, it is not entirely satisfactory for Buddhism. This
accounts for the number of translations which have been suggested for the word,
including “passions”, “intoxicants”, “cravings” and “cankers”.60 The latter to me is a
56 Alsdorf, 1965, 3–6, and Norman, 1989A, 393–97 (=CP IV, 264–70).
57 Alsdorf, 1965, 4–5.
58 Schubring, 1962, §§ 84, 97.
59 The āsavas and the oghas are kāma “sensual pleasure”, bhava “existence”, “speculative
view”, and avijjā “ignorance”.
60 See Norman, EV I, 133–34.
disease of dogs’ ears and of roses, and I am always surprised when I find the arahat, the
one whose āsavas have been destroyed ( ), being described as “canker-
waned”.
A comparative study of the terminology of the two religions gives some idea of the
religious and cultural background in which Buddhism and Jainism came into being. The
explanation for such parallels in terminology as āsava can sometimes be seen as a
borrowing from one religion to the other or, perhaps more often, a common borrowing by
both from a third religion or from the general mass of religious beliefs which we may
assume were current at the time the two religious leaders lived, i.e. the beliefs of the
.
It is to this general background of religious thought that we can probably assign most
of the vocabulary of the ascetic type of religion, e.g. such words as “ascetic”,
pravrajyā “going forth”, pravrajita “one who has gone forth”, tapas “mortification”, and
“sage”, found in both Buddhism and Jainism, and we may assume that these were
terms which were common to many of the religious movements, in which the adherent
went forth from the life of a householder and became a wanderer. Much of the
terminology used for their religious experiences was also common to the two religions.
They had terms such as nibbāna ( ) in common and there is the strange fact that
they both use with it the past participle of another root (nibbuta, nivvua), with a different
meaning, which suggests that, just as I have suggested that nibbāna as a concept was pre-
Buddhist, the word-play on the two words was also earlier than both religions.
It was long ago noted61 that the Buddhists and Jains “give the same titles or epithets to
their prophets”, e.g. (in their Sanskrit forms) arhat, Sugata, Tathāgata, Jina “conqueror”,
Mahāvīra “great hero”, sarvajña “omniscient”, Siddha “perfected”, Buddha “awakened”,
Sambuddha “id.”, “gained ”, mukta “released”.
Two words in this list in particular merit close attention, namely Buddha and Jina. We
are accustomed to think of these words as distinctive of their religions. Buddhism is
called after the Buddha, and Jainism after the Jina. If this specific distinction was
attached to the words at the time of the founding of the two religions, one might have
expected that the other religion, in each case, would have avoided the words, or at least
have said “our buddha is better than your buddha” or “our jina is better than your jina”.
From the fact that they did not do so but continued to use the terms Buddha and Jina in
both Jainism and Buddhism, we may deduce that these words were in common use prior
to the origin of both religions and were taken into both of them with a non-distinctive
sense. That is to say that there were those who were spoken of as buddha “awakened” to
some sort of truth—doubtless about the possibility of release from , and those
who were called jina “conqueror”, doubtless conquerors of , before the words
were taken over into both religions, and it was then a matter of the historical development
of the terminology of both religions that the specific distinction which those words
denote now in those two religions arose. That is to say, there were buddhas and there
were jinas before the beginning of both Buddhism and Jainism. The fact that Gotama was
not the first buddha, and Mahāvīra was not the first jina helps us to
61 See Jacobi, 1884, xix.

understand how both religions evolved a theory of previous Buddhas and Jinas.
It is not known where the idea of a specific number of previous prophets came from,
but it may be no coincidence that the Jains have 24 Jinas, while the Buddhists have 24
previous Buddhas,62 plus Gotama Buddha. The addition of three extra Buddhas, which
we find in the , is clearly a late extension of the general idea in
Buddhism.
I have already mentioned in my first lecture the fact that the philological explanation
for the difference between Pāli pacceka-buddha and Prakrit patteya-buddha lends support
to the idea that the concept of this particular class of buddhas was also earlier than
Buddhism and Jainism. The information which the two religions give about these
buddhas,63 including the details down to the names and the causes of their awakening,
suggests that the concept was also something which was part and parcel of the
movement and not exclusive to any one religious group, and was taken into
those two religions from a third, earlier, source.
From the Pāli texts then we see that there was simultaneously a brahmanical orthodoxy
and a non-brahmanical movement. At a slightly later time Aśoka could refer to
all religious persons as being either or , as the compound
- which he uses in his inscriptions shows. That movement
produced not only religious ideas which went against the brahmanical way of thinking,
but also literature making the same point. We find that both the Buddhist Jātakas and the
Jain -sutta contain stories referring to the way in which
mis-treated who came to their sacrificial enclosure to beg for food. The stories
clearly had no specific class of in mind, which made it easy for both religions
to take over such stories and incorporate them into their collections of texts. Both
religions shared texts (once again, probably of a common origin) defining a
by his conduct, not by his birth. Such literature emphasised the fact that it was support of
the movements which would bring merit, not support of the brahmans.
The Jains too objected to the brahmanical idea of the ātman, and refuted it in a way as
empirical as the Buddhists. There is a Jain verse which says: “As a whole mass of earth,
with all its manifold nature, is seen as one, so the whole world, with all its manifold
nature is seen as the intelligent principle. Some fools, intent on their (bad) activities, say
that it is so with the individual. (But) the individual who does an evil deed goes by
himself to a harsh misery”, i.e. if we were all part of the same intelligent principle,
namely the ātman, we as a whole would be responsible for evil deeds as a whole, and
would suffer the punishment for them as a whole. It does not happen like that. Only the
evil-doer suffers the punishment.
Nevertheless, the Jains differed from the Buddhists in believing that, although there
was no world ātman, there was nevertheless a permanent, everlasting, individual ātman
62 For the number of Buddhas see Gombrich, 1980, 62–72 (68), where he suggests that the number
24 was taken over by the Buddhists from the Jains.
63 See Norman, 1983A, 92–106.
, which was the element which transmigrated. The Jains were able, therefore, to say that
when the personal ātman gained release it was called siddhattha “that which has gained
the goal”, and it went to the place of siddhi “perfection” at the top of the world, where all
the siddhas “perfected ones” went. In the diagrams of the world which we find
sometimes in Jain manuscripts, the universe is very often depicted in the shape of a man,
with the earth at his waist and the hells below the waist and the heavens above the waist,
the place of siddhi is right at the top, on his forehead.
The Buddha, on the other hand, had problems. Not only did he deny the existence of a
world attā, but he also rejected the idea of a permanent individual attā. It was therefore
very unclear what it was that transmigrated, and he found it very difficult to describe the
state of one who had gained nibbāna. It would probably be more accurate to talk of the
non-state of one who had gained nibbāna, because it is not possible to say what nibbāna
is like, or what someone who has gained nibbāna is like. How can you describe the
condition of a non-being in a non-state? It is that conundrum which led to the Buddha
refusing to discuss the existence or non-existence of the Tathāgata after death. All we can
do is say what nibbāna is not like. It is not like . Consequently it is often
defined in terms of negatives or opposites. It is “blissful” (siva) or “happy” (sukha) as
opposed to the dukkha of existence. It is “unmoving” (acala) as opposed to the endless
movement of . It is “without death”64 (amata) as opposed to the repeated deaths
of . It is “without birth” (ajāta), “without beings” (abhūta), “without made
things” (akata), and “without formed things” ( ) as opposed to the world,
which has birth, beings, made things and formed things.
I said at the beginning of this lecture that the origins of Buddhism lie in the political,
economic, social and religious environment of the time.
What time? I do not wish to say much about the date of the Buddha, but I must say
something about it, because it has some bearing upon what I have been discussing. There
are various ways of calculating the date of the Buddha’s death, and the one which is
perhaps most commonly accepted in the West is c. 486 B.C.E. This depends upon a
statement found in the Pāli chronicles that the Emperor Aśoka was consecrated 218 years
after the death of the Buddha. Aśoka in his inscriptions mentions the names of a number
of contemporary Greek kings, which enables us to fix Aśoka’s dates within fairly narrow
limits, and by adding 218 to the date we can calculate for his consecration (c. 268 B.C.E.)
we get a date c. 486 B.C.E. The Pāli chronicles also give the regnal years of the kings of
Magadha between the death of the Buddha and Aśoka, and these dates too support the
theory of an interval of 218 years, although they allow only 22 years for the ten sons of
Kālāsoka and a similar period of 22 years for the nine Nanda kings who followed them,
which seems to represent a manipulation of the chronology to make things fit.
64 Vetter (1985, 74) may not be correct in translating amata as “immortality”. This translation
perhaps gives the wrong impression, since the Buddha was presumably trying to gain release from
, i.e. he was trying to find a state where there was no rebirth, and therefore no dying
leading to rebirth. For this reason nibbāna is described as being without birth, without death,
without gati, etc.
This date of 486 B.C.E. causes difficulties, because the archaeological evidence, such
as it is, suggests that, at that time, some of the places which the Buddha is said in the
tradition to have visited had barely, if at all, been founded, while the evidence for a
monetary economy65 suggests that it is to be dated somewhere around 400 B.C.E. If this
archaeological dating is to be believed, then the picture of social and political life which
we get from the early Pāli texts is misleading, since it reflects the conditions of a later
time, not those contemporary with the Buddha.
Fortunately, however, philology has come to our aid. Recently attention has been drawn
to the fact that although the Pāli word sata and the Sanskrit word śata do undeniably
mean “one hundred”, they are also used to mean a large number, any large number. It has
also been pointed out that the number “eighteen” has some sort of auspicious
significance. Kings are very often said to reign for eighteen years, or important events to
happen in the eighteenth year of their reign. To say, then, that something happened 218
years after the death of the Buddha, probably means no more than saying that it happened
after a large number plus a large number plus an auspicious number of years, and this can
in no way be taken as firm evidence that the whole period of time came to 218 years.
Other ways of calculating the date of the Buddha’s death have therefore had to be
employed, and recent attempts making use of the information we find in both Buddhist
and Jain texts about the life spans of theras have to some extent settled on about 400
B.C.E.66 If we accept the tradition that the Buddha was 80 years old when he died, then
we are talking about a life-span covering the period 480–400 B.C.E. and a teaching
period of c. 445–400 B.C.E., which fits the archaeological evidence better.
For the forty-five years prior to 400 B.C.E., then, the Buddha moved around Magadha
and the surrounding areas, preaching his doctrine which grew out of the general
tradition of opposition to the brahmanical view of the social superiority of the
brahmanical caste, and to the brahmanical belief in the ātman/brahman world spirit. The
Buddha’s meditative techniques and terminology owed much to other
movements, and his establishment of a mendicant community followed the general
pattern of abandoning the world, and relying for food and other necessities of
life on the generosity of householders who were lay-followers. With those other
movements he shared technical terms and epithets to describe those who had
escaped from , as well as anti-brahmanical literature.
Looked at in this way we can see that, although the Buddha’s experience and his way
of making known his experience to others were unique, there is far more that Buddhism
held in common with the movements which were contemporary with it than we would
normally assume, and we can see that there is an element of truth in the well known
statement that the only original feature of the Buddha’s teaching was his combination of a
belief in the transmigration of the self with the denial that there was a self to
transmigrate.
The Buddha’s message about release from and the attainment of nibbāna,
as explained in his sermons, was remembered by his hearers, and repeated by them, as
65 See Erdosy, 1993, 40–56 (51). See also Cribb, 1985, quoted by von Hinüber (1993,
111).
66 See Norman, 1991B, 300–12 (=CP IV, 185–201).
they had heard them. After his death, collections were made by the expedient of asking
his leading followers to recite what they remembered the Buddha saying. What they said
was in turn recited by the rest of the monks, and arrangements were made for the safe
keeping of these texts.
III
Buddhism and Oral Tradition
In the second lecture I spoke about the way in which the Buddha’s teaching evolved from
the general religious movement, which had grown up in opposition to the
caste, as both a social and a religious entity. After seeking—unsuccessfully—
release from by means of severe ascetic practices and the meditation techniques
taught by contemporary teachers, he succeeded by means of his own meditative
practices—unique as far as we can tell, since no one else is known to have gained release
in precisely that way before him. His message that it was possible to gain release (
) from , and to attain , in this unique way, was taught, by word of
mouth, first to his former associates in ascetic practices, and thereafter it was spread by
the Buddha and his followers throughout Magadha and the surrounding areas.
There is no agreement among scholars about the date when writing first came into use
in India but everyone, I think, agrees1 that during the early period of Buddhism, even if
writing was available, all teaching was by oral methods, and the Buddhist scriptures were
transmitted orally, as was also the case with the brahmanical texts.
If writing was in use during the early period of Buddhism, we should have expected to
find rules laid down in the Vinaya governing the proper use and storage of writing
implements and materials, in the way in which we find instructions about everything else
which concerns a monk’s daily life. There are no such instructions in the Vinaya, and
there are in fact only two mentions of writing in the whole text, both of them in the
Parivāra, the last section of the Vinaya.
The Theravādin tradition confirms the absence of writing by stating that the canonical
texts were first written down during the reign of Abhaya, in the first
century B.C.E.2 in Sri Lanka, implying that before that date they had not been written
down—at least in their entirety. As we shall see in the fifth lecture, there is no good
reason for doubting that the two statements about writing in the Vinaya, which I have just
mentioned, are interpolations, added when the Parivāra, together with the rest of the
canon, was written. The absence of writing is also confirmed by the terminology
1 See Bechert, 1991A, 3–19 (10).

2 89–77 B.C.E. according to Bechert, 1991A, 3–19 [9].
employed in the texts themselves, which enables us to deduce a great deal about the
manner of teaching.
The vocabulary of the early texts is centred around the words for “hearing” from the
root śru “to hear”, and for “speaking” from the root vac “to speak”. So a learned man is
spoken of as bahuśruta (Pāli bahussuta—“having much hearing [śruti]”), and the word
for “to teach” is vāceti “to make someone say something, to recite something (after his
teacher)”. Such examples can be multiplied endlessly, and collections have been made of
the words and phrases which imply reciting (an orally transmitted) text, rather than
reading (a written) one.3 With the aid of such collections we can make certain deductions
about the transmission of texts before the writing down of the canon. It must, however, be
noted that this terminology does not, in itself, prove that that the texts were not in written
form, because we know that the Aśokan inscriptions, which by definition are inscribed,
i.e. written, on rocks and pillars, employ the same type of terminology. Aśoka says at the
end of some of his inscriptions, “This edict is to be listened to” ( lipī sotaviyā),
suggesting that perhaps only the administrators were able to read, and their duty was to
recite the contents of the edict to an audience, on the dates and in the manner specified by
Aśoka.
We find, indeed, that the early terminology—the use of the verbs śru and vac, for
example—was widely used at a much later time when writing was well known, because it
was the standard terminology. That is to say: the technology changed, but the
terminology did not, in the same way as the French for “pen” is “plume”, or I still find at
the bottom of copies of letters which are sent to me “carbon copy to Norman”, when what
I receive is not an almost illegible letter written on some sort of flimsy tissue paper, but
an impressive document printed on a LaserWriter, and indistinguishable in every respect
from the top copy sent to the original addressee.
We must assume that in the early days the Buddha’s followers spread the message as they
had heard it from the Buddha himself, and from his chief disciples, and as they had
remembered it. We have no idea of the amount of divergence which began to creep into
the message as a result of poor memory, or other defects in such a method of
transmission, but it would be surprising if transmission in such a haphazard way had no
effect upon the words and form of the message, if not its basic content. It is possible that
in the very earliest stages of the oral tradition the situation was much the same as we are
told exists in the recitation of modern oral epics, where no two performances are exactly
the same, and the order of episodes may be changed from recitation to recitation, and
episodes may even be omitted. It does not, however, matter very much if the reciter
leaves out a portion, or adds something, as long as the main narrative is maintained and
the desired end is achieved. The same may well have been true of early recitations of
Buddhist suttas. The name of the place where the sermon was preached, and the identity
of the audience, and other such relatively unimportant details, may well have varied from
recitation to recitation.
The way in which groups of synonyms were used to explain or elaborate concepts,
which we find in particular in the commentarial portion of the Vinaya, in the Niddesas
and in some of the Abhidhamma texts, also suggests that texts of this type were
3 See Collins, 1992, 121–35.

Buddhism and oral tradition 37
composed and then transmitted orally. In their earliest form, they must have represented
the attempts of individual bhikkhus to explain something, with a resultant variation in the
details of the sermon, although the main theme would doubtless have been retained
unchanged. We can imagine that if someone was reciting, for example, the
, he might insert synonyms and quotations from other texts, as they
occurred to him, and on some occasions he might remember more or less of them, but it
made little difference if synonyms were omitted or their order was changed. It is possible
that, as the result of such inconsistencies, the need for some sort of codification was
realised, and it might even have begun to take place during the lifetime of the Buddha.
We know that the Pātimokkha, the body of rules which governed the monks’
behaviour, already existed as a collected corpus of material because the monks came
together twice a month to recite it, and we may assume that each recitation was identical,
unless new rules had been promulgated, or old ones modified. The fact that the
Pātimokkha was structured so early doubtless accounts for the fact that the formulation of
the monastic rules, although not their number, is very similar in the Pātimokkhas of the
various schools. We also know that four times a month the monks preached sermons to
the lay-followers, so we could speculate that such public recitations, presumably in the
presence of other monks, might have led to some inconsistencies in recitation being noted
and as far as possible eliminated. This would have been the beginning of codification. We
can only guess at the amount which had taken place during the Buddha’s lifetime.
We do, however, know that the tradition records that after the Buddha’s death a
meeting was held at which the Buddha’s teachings were recited at a joint recitation
( ). In the form in which the tradition tells the story, one thera recited
the Vinaya rules, and another the Suttas, and they were accepted by the rest. This was the
beginning of canonicity, about which I shall speak in the eighth lecture, but I must
anticipate my future remarks by saying that it seems to me to be most unlikely that all
these sermons were already in the shape, and in the order, in which they are said to have
been recited on that occasion, and it is probable that even the use of the names of the
nikāyas at the first joint recitation is an anachronism.
The tradition tells us that, after the first recitation, measures were taken to ensure that
this body of material was preserved from then on. The Buddhavacana was divided up
into parts and given to groups for safe keeping. Buddhaghosa tells us that the Vinaya was
entrusted to Upāli and his pupils. Upāli was the expert who had replied to Mahākassapa’s
questioning about the rules of the Pātimokkha, giving information about where the
offence was laid down, with respect to whom, on what subject, etc. In the same way the
Dīgha was entrusted to Ānanda (who had recited the Sutta- ), the Majjhima to
Sāriputta, the to Mahākassapa, and the to Anuruddha, and their
respective pupils.4 The way in which the texts were said to have been shared out to these
various groups implies that they were already organised into the nikāyas as we know
them: long (Dīgha), middle length (Majjhima), those linked by associated subject matter
( ) and those arranged in numerical order, smallest first, with each section
increasing by one ( ).
It is generally accepted that this distribution was probably the beginning of the
4 Sv 13, 23–24; 15, 2–13.

system. The word means “speaker”, from the root “to speak”,
and is another of the items of vocabulary which suggest that the early Buddhists used an
oral tradition. There are references in the Pāli commentaries to of the first four
nikāyas, and also of the Jātaka and the Dhammapada, but it is probable that there were
also of other individual Khuddaka-nikāya texts.5 There seems to be only one
reference in the early literature to the Khuddaka- . Since the Jātakas are part of
the Khuddaka-nikāya, the relationship between the Khuddaka- and the Jataka-
, who are mentioned in the same sentence by the author of the Milinda-
pañha,6 is not clear.
If the material entrusted to the groups of was at first of a somewhat haphazard

nature, as I have suggested, then the first task would have been to start some sort of
editorial process, to make the material more consistent and to devise ways which would
ensure that the consistent whole which the produced could easily be handed
on to their successors. We can surmise that the language was homogenised to a large
extent. Once we have determined the nature of the language which we call Pāli we can
see that in the canon as a whole there are very few non-Pali characteristics and most of
those are due to a consistent introduction, at a later date, of Sanskritisms, which are
restricted in number, the most obvious being the absolutive in -tvā. By eliminating
anomalous dialect forms, such an editorial process may be presumed to have made
learning the texts easier, by simplifying forms which at the beginning were probably
quite divergent.
Once the language was homogenised, then the lists of synonyms, etc., which I have
just mentioned, would probably have been fixed both in number and in order, and we can
see signs of this in some of the categories of the , where we
commonly find a trio of words together, in a list of synonyms—a short a- or long ā-stem
noun, then an action noun made from the same root with the -ana suffix, and then an
abstract noun made by adding -tta or -tā to the past participle of the verb—always in the
same order.
It is clear from a comparison of the way in which the nikāyas are formulated that there
are certain differences between them, presumably arising from the fact that they were
remembered in a slightly different way by the responsible. We would expect
the editorial process to vary from group to group, so that it is not
surprising to find that formulation also varied from one group to another.
Nevertheless, it is interesting to note that, by and large, the differences in the language
and other features of the various Pāli nikāyas, to which I have just alluded, are not great.
Considering the disparate nature of the material with which the must have
started, this is rather surprising, because, although we know little about the relationship
between the various groups of in the Pāli tradition, we can deduce that they
did not consult with each other to the extent of making their recitations of individual
suttas or groups of verses identical, as we can see, for example, in the case of the thera
5 See Norman, 1989B, 29–53 (33) (=CP IV, 92–123 [98]).
6 Miln 342, 1.
verses, for a comparison of the versions of the verses ascribed to him in

the Theragāthā, the -nikāya and the Suttanipāta, shows that the versions
preserved by the of the - and Khuddaka-nikāyas do not agree in
every way, although it is possible that repeated his verses in different ways on
different occasions.
The also had different ideas about matters of Buddhist history, e.g. whether
the four nimittas were seen on the same day or not, and why Ānanda arrived late at the
recitation,7 and also about the distribution of texts between the various .
Nevertheless, despite these differences, the fact that in general duplicated texts do not
differ so very much, when they occur in different nikāyas, would indicate that there was,
or had been, some sort of co-operation between the or their predecessors. In
the case of the belonging to different traditions, or their equivalents in other
system of transmissions, we can assume that the co-operation between them was less,
although recent investigation has shown that communication between some schools, at
least in later times, was closer than has sometimes been thought.8
If, however, the had started their conservation work immediately after the
first recitation, without consulting one another, it is not easy to see how the
homogenisation of the language of the texts, and in particular the translation from one
dialect to another, which must have happened on more than one occasion in the history of
the Theravādin canon, as the Buddha’s message spread—as we shall see in the fourth
lecture—could have happened in such a consistent way between the various nikāyas, with
no one nikāya showing a greater proportion of anomalous sound changes than another.
This would suggest that a certain amount of homogenisation had already taken place
before the took over, and the delay entailed in this standardisation would also,
of course, have allowed time for the texts to be collected, classified and codified into
nikāyas.
The fact that differences of dialect are detectable in the Pāli canon shows that the form
of the texts was certainly not fixed unalterably immediately after the first recitation. If
changes could be made, this too would suggest that the system was not yet in
operation, or at least not in the form of caretakers of an immutable body of material. The
possibility that change could still take place would suggest that similar changes could
occur when comparable texts were being remembered by the monks belonging to other
traditions. This would perhaps account for the differences which we find in related suttas
belonging to other schools.
We must accept, then, that the story that the system was instituted at the
first joint recitation creates great difficulties. Although it seems clear that the origin of the
system must have been on the lines that have been suggested, it clearly cannot
have happened in exactly that way. It is most unlikely, to say the least, that within a very
8 In particular the Sarvāstivādins and the Mūla-Sarvāstivādins; see Schmithausen, 1987.

short time of the Buddha’s death suttas had already been collected and categorised by
length and subject matter into the form in which we have them in the Theravādin canon,
and it is most unlikely that the Sutta- was in its present form at that time. It is
obvious that if the Dīgha-nikāya, Majjhima-nikāya, etc., had not yet been formulated and
named, there could hardly have been Digha- and Majjhima- , etc.
The Khuddaka-nikāya presents even greater problems, e.g. there are references in the
Apadāna to the Kathāvatthu. The Kathāvatthu is, however, acknowledged by the
Theravādin tradition to be a very late text, composed on the occasion of the third
recitation in Aśoka’s time. It is obvious, then, that portions, at least, of the Apadāna must
be very late additions to the Khuddaka-nikāya, and yet the Apadāna is included in the
lists of texts recited at the first recitation. As I have suggested, it is probable that even the
use of the names of the nikāyas at the first recitation is an anachronism.
It is therefore possible that the information about the texts which were recited at the
first recitation was given with the benefit of hindsight, by those who knew what form the
canon had in their day, and who thought or at least maintained, perhaps for the purpose of
authenticating their own school, that that form was precisely what had been recited at the
first . Similarly, the tradition was perhaps a later invention in the form
in which we hear about it, being restricted to the Theravādin tradition, and the story of its
early foundation was invented to give authenticity to the Theravādin canon. As to when
this was done, we cannot say much, except to draw attention to the occurrence of the
names of the nikāyas and the word (of the majhima, eka-uttiraka and śayutaka
nikāyas9) in early Sinhalese inscriptions probably datable to the second century B.C.E.
The institution of the system might then be the result of a decision taken after
the second recitation, or even as late as the third recitation, i.e. the system arose
before the formation of the Abhidhamma- .
On the other hand, we should perhaps note that there is no reference to the existence of
Abhidhamma- in the sentence in the Milinda-pañha which I mentioned earlier.
We find ābhidhammikas included there, presumably experts in the Abhidhamma, but
there is no mention of . This may mean that the system was closed,
with no new groups being set up by the time the Abhidhamma was formulated, so that the
Abhidhamma as a whole was composed too late to be incorporated in the
system. The Milinda-pañha sentence might indicate that there was a difference between
an ābhidhammika and an Abhidharma- . It is probable that the were
something more than mere caretakers of the texts entrusted to them. It has been suggested
that they were also professional reciters, and this seems to be the sense of the word in
Buddhist Sanskrit, where it occurs most commonly in the compound dharma-
“a preacher of the dharma”.10 They could perhaps be called upon to deliver a sermon
when required, and someone asking for a sermon to be recited could specify the type of
sermon by length. As I mentioned in the first lecture, we sometimes find two version of a
sutta, one long and one short, and so it would be possible for someone to say “I would
rather like a middle length sermon, about something or other, and by the way I would
9 See Norman, 1989B, 33, note 26 (=CP IV, 97, note 3).
10 See BHSD, s.vv. and dharma- .
prefer the shorter rather than the longer version of that sermon”.
The differences of view between the about historical matters, which I have
mentioned, perhaps indicates that a did not merely recite, but could also put his
recitation into some sort of context. The same distinction between a caretaker-cum-
historian and a reciter is perhaps to be made about the word “knowing the three
”, which is used to describe, for example, the monks at the third . There
was presumably a distinction between knowing the and being a member of a
group whose purpose in life was to teach a text to others and to be able to recite it to
order. If the purpose of the was to recite sermons to lay-followers at their
request, then we may assume that few lay-followers wanted a recitation of a portion of an
abhidhamma text. The teaching of abhidhamma was perhaps still possible by means of
the elaboration of the mātikās—which I shall mention in a moment—on an ad hoc basis.
We can only speculate about the way in which the system operated. We
might assume that the junior sat around their seniors and learned the texts
from them. Since the oral tradition is still strong in the Buddhist countries of South and
South-east Asia it might be thought to be a simple matter to visit a monastery, and see
just how the oral tradition is preserved, and we might have hoped that descriptions of the
way in which texts are remembered and recited in modern times would have thrown light
upon the situation at earlier times. This unfortunately does not appear to be the case.
There is, for example, an account, by Tambiah, of the way in which texts were learned by
junior monks when they were first admitted to the circle of reciters. He reported11 that, in
the monastery which he visited, texts were chanted by monks morning and evening.
Newcomers repeated what they heard and memorised the chants fairly quickly, e.g.
chants used in the daily worship of the Buddha—parittas such as , and other
texts such as the Pātimokkha (which as we have seen was one of the first Buddhist texts
to be recited).12 This would seem to be a good guide as to the way in which memorisation
was done in early times. Unfortunately, the picture is somewhat ruined by Tambiah’s
discovery13 that when the monks practised chants individually, they did so with the aid of
printed texts, to ensure that they got them right.
We do not know how long the system remained in being. Buddhaghosa refers
to it as though it was still in operation in his time, although by then the canon had been
written down for some hundreds of years. This is, however, not conclusive, because when
Buddhaghosa says that the Dīgha- and the Majjhima- do such and
such he may simply be repeating what he found in the commentarial literature available
to him in the Mahāvihāra in Sri Lanka, and it seems very clear that this commentarial
literature was composed some centuries before the time when Buddhaghosa wrote and
presumably referred to conditions at that time, which perhaps no longer pertained by
Buddhaghosa’s time.
11 Tambiah, 1968, 100.

12 Tambiah, 1968, 99.
13 Tambiah, 1968, 100.
I said at the beginning of this lecture that one of the first acts of the , or
their predecessors if they were a late institution, must have been to attempt to
homogenise the language. Side by side with that imposed consistency of language we can
see that there is a certain consistency of structure.
There has been in the past a certain reluctance to believe that all early Buddhist
teaching could be based entirely upon an oral/aural tradition, because of the sheer
quantity of the material which would have to be remembered. In this connection we must
have regard for the tradition, which I have mentioned, that the texts were divided up into
nikāyas and shared out to different theras and their pupils for safe keeping and onward
transmission, so that no thera was responsible, as a , for more than one nikāya.
When oral literature became the object of academic study, and research was carried
out into the recitation of oral epics in various parts of the world, it was discovered that so-
called “primitive” peoples were able to remember and recite very long texts. Doubts
about the ability of the early Buddhists to memorise their material were thought, by some,
to have been dispelled by reference to such feats of memory, since it was thought that
similar recitation might explain how the oral tradition of Buddhism and other religions
could be maintained.
It became clear after a while that the situation was not as simple as might appear. The
oral literature which was being studied was essentially of a verse nature and therefore the
comparison would seem to apply only to Buddhist verse texts, but even there the situation
was not entirely comparable, because the very nature of Buddhist verse texts, and the
metres in which they were written, demanded complete accuracy of memorising, whereas
the oral literature which has been studied is essentially of an epic nature where, as I said
earlier, it is alleged that no two performances are ever identical because the reciter is free
to insert, at any point, material of a formal nature, the so-called formulae which can be
used to keep the recitation going while he remembers what happened next in the story.
The great majority of Pāli canonical texts, however, are in prose, and complete accuracy
of reproduction is required at each recitation. In these circumstances the findings of
modern investigators of oral epic literature seem to have little relevance.
It seems that we must seek elsewhere for the explanation of the way in which the Pāli
material could be remembered and handed on. We know that elaborate systems of
recitation were and to a limited extent still are employed in Vedic recitation, but there is
no evidence known to me for the employment of such methods by the Buddhists. Nor
would they be entirely appropriate for prose texts.
I have already mentioned a consistency of structure as being one of the features which
we can detect in the Pāli canon, and if we examine the canonical texts we can, in fact,
find many features which we may assume were employed for mnemonic purposes to
make the memorisation, and therefore the recitation, of these texts easier.
Rhys Davids a century ago14 drew attention to some of the features which, he
suggested, aided the power of memory for Buddhist Sutta and Vinaya texts. He pointed
out “firstly, the use of stock phrases, of which the commencement once given, the
remainder followed as a matter of course and secondly, the habit of repeating whole
sentences or even paragraphs, which in our modern books would be understood or
inferred, instead of being expressed”.
14 T.W.Rhys Davids, 1881, xxiii (quoted by Tambiah, 1968, 100–101).

It is not precisely clear what he meant by stock phrases, and I suspect that he was
referring to the way in which suttas tend to start in the same way—“Thus have I heard”—
this is said by the cties to be a reference to the way in which Ānanda at the first recitation
repeated what he could remember of the Buddha’s sermons15—“At that time the Buddha
was staying at such and such a place with a group of bhikkhus, and one day the bhikkhus
decided to do something, or ask a question, etc.” Many of the introductory paragraphs to
these sermons carry on with stereotyped phrases—someone approached the Buddha, and
having approached him sat down at one side; to him seated at one side the Buddha said
something or other.
In a paper read at the conference of the International Association of Sanskritic Studies
in Australia in 1994,16 it was shown that the consistency in the way in which these
introductory paragraphs are structured is, in fact, more meticulous than might at first
appear. It can be seen that the wording changes subtly in conformity with a fixed pattern
to specify who is approaching whom. In each case the wording is slightly different and a
further result of this is that, once a story teller has remembered that the particular sermon
he is about to recite is, say, about one or more bhikkhus approaching the Buddha, to ask a
particular type of question, the form of wording to be used is prescribed. And therefore
he does not, so to speak, have to remember the form of words to use because the
circumstances fix the form for him.
Interestingly enough, it has been pointed out that if we examine such stereotyped
phrases in one nikāya and compare them with the phrases in another, we find that the
forms which are employed do not necessarily agree—something which leads us to the
conclusion, I think, that, as we would expect, once the texts had been distributed among
groups to preserve and hand them on to their successors, the precise methods of
stereotyping which were employed, in an attempt to make remembering easier, were not
necessarily the same for each set of .
Another way in which we can see that these these texts were recited rather than read is
the way in which they include in them lists of contents or indexes—the so-called
mātikās,17 sometimes translated as matrixes. It has been pointed out that these lists play a
mnemonic role,18 which, as we have seen, is important in a tradition which is founded
essentially upon oral transmission. The word mātikā is sometimes used to designate the
precepts of the Vinaya,19 and mātikās form, as it were, compendia of the doctrine. They
are, however, also lists, especially in the Abhidhamma, which are presented in a
numerically progressive form and they have a creative role, in as much as they allow the
Abhidhamma to be composed and thus, it has been suggested, the idea of “mother” is
contained in both mātikā and matrix. It has been said with regard to the , for
example, that if one knows the Mātikā to that text with its 24 conditional relations, and
relates them to the 22 triplets and 100 couplets of the mātikā to the ,
in all the permutations and combinations in which these can be taken, one can reconstruct
the whole of the .20
16 Allon, 1994.
17 Cousins, 1983, 1–11.
18 See Gethin, 1992A, 149–72.
19 See Norman, 1983C, 126.
20 See I.B.Horner, in the Foreword to Nārada, 1969, viii.
Something similar can be seen in the uddānas (the lists of contents) which are found
in, or at the end of, a number of texts, e.g. in the Dhammapada where the uddāna
givesthe names of the vaggas, and in the Thera- and Therī-gāthās, i.e. the verses ascribed
to male and female elders, where we are told how many elders there are and how many
verses as a whole they have recited. Despite their position in our texts, these uddānas
were probably intended originally for use at the beginning of the recitation, and even as
the reciter progressed. Someone reciting the Dhammapada, for example, had a guide to
tell him which vagga came after which, so that he could keep them in order. It must be
pointed out that the system is not foolproof, because the numbers given in the uddānas to
the Thera- and Therī-gāthās do not agree entirely with the numbers as we have them
now,21 and they presumably refer to an earlier recension of the text, where such numbers
were relevant and of value to the reciters. They have been retained in a written recension,
even though they are no longer of any value.
Listing things by numbers is a very common mnemonic device in India, and both the
Buddhists and Jains make use of this idea. The Jains have the and the
Samavāya, the contents of which are listed numerically, and the Buddhists have
something very similar in the -suttanta of the Dīgha-nikāya. The name is
reminiscent of the at which the texts were recited in a joint recitation, and the
name of the sutta suggests that it represents a recitation of doctrinal matters, arranged in a
numerical way, and intended for chanting together, perhaps in an attempt to provide a
summary of the doctrine as a precaution against the confusion among the Jains after the
death of their leader Nāthaputta, which was the occasion for the preaching of the
. Whether the recitation was at one of the great or at some other
chanting is a matter for investigation.
We also find this “more by one” principle used in the -nikaya, which is
listed numerically for the same reason. It is quite clear that this method of enumeration is
purely for artificial purposes, because we find that although the lists of ones and twos and
threes and fours, etc., are genuine groups, when we get to the higher numbers the authors
of the text were obliged to make combinations, e.g. one group of ten is made of the five
fears (bhaya), the four elements of stream-entry ( ), and the ariyan
method (ariya ñāya).22
Another guide for oral recitation which we can find in the Pali texts is the principle
which appears in the list of abbreviations in A Critical Pāli Dictionary as wax. comp., i.e.
the rule of waxing components. It is a translation of a German term, which was first used,
as far as I know, in the field of Assyrio-Babylonian studies.23 It perhaps sounds odd in
contemporary English, where the word “waxing” in the sense of “growing larger” is
almost entirely restricted to the moon growing larger, as opposed to its waning phase. If
we are to retain the word “waxing” then it is perhaps clearer if we call it the Waxing
Syllables Principle, because the phrase refers to the fact that strings of words which make
up a group, e.g. a number of Pāli epithets describing a city as “large, rich, prosperous,
21 See Oldenberg & Pischel, 1883, xiv.
22 A V 182.
23 Ehelolf, 1916.
flourishing, crowded”, are very often put into order depending upon the number of
syllables in each word, with the words with the fewest syllables coming first, and then the
word with the next fewest syllables next, and so on until the word with the most syllables
comes last.
Such a principle clearly guards against the way in which the order of such a string of
adjectives might be shuffled around. They were put into a fixed order depending on this
Waxing Syllables Principle and therefore they were always remembered in that fixed
order. If anyone recited the string of words and put one word in the wrong place, then the
change in the number of syllables would immediately reveal it. Once this principle
became second nature, so to speak, then a reciter could not go wrong, because he would
automatically recite words in the right order.
Something similar in English, but based on an entirely different principle, is the way
in which we might talk about, say, a big red armchair. For some reason, of great interest
to linguists, we never say a red big armchair, and so if we are telling a story where there
is frequent reference to a big red armchair, to some extent our task of remembering that
part of the story is eased, because we do not, each time, have to remember the order in
which to say the words. The order is fixed for us in a pattern by something we do not
know about and do not have to worry about. It is, so to speak, automatic.
The interesting thing about this is the way in which we occasionally find that words
are not in the Waxing Syllables Principle order. If we find a set of words which we might
suspect was at one time arranged on that principle—and if the suspicion is supported by
the fact that in another tradition or in a comparable phrase in a Jain text it does occur in
that order—but it no longer is, we may well be able to find a reason for the change.
Perhaps one term has been replaced by a synonym, with a different syllable length, or a
word in its Pāli form with a svarabhakti vowel is out of place, but if we calculate what the
Sanskrit form or the form in another dialect was, we can see that the principle is indeed
retained intact. This gives us information about the dialect in which the phrase was first
composed. Sometimes we may suspect that the replacement took place at a time when the
Waxing Syllables Principle was no longer operative, i.e. when the oral tradition which
required memorisation had given way to a written tradition, which did not require it, and
the Waxing Syllables Principle was of less importance.
Another principle to which Rhys Davids drew attention a hundred years ago is the way
in which we find in Buddhist texts frequent examples of repetition.
The way in which repetition operates varies considerably. Sometimes it is exact
repetition. If something happens twice or more, or something is said twice or more, then
the exact passage is repeated verbatim on each occasion, e.g. stock sets of words, such as
the description of a city, which I have just mentioned. Sometimes the repetition is partial,
for example, we may find a statement saying that the good man does a series of actions,
followed by a statement that the bad man does the opposite of these, each word being the
same words used for the good man, with a negative prefix a- added to each. The reciter
had therefore in effect only to remember one set of adjectives, and to put the negative in
front of them, e.g. the good man does kusala deeds, while the bad man does akusala
deeds. Once again the effort involved in memorising the sermon which contains these
phrases is reduced. Similarly, there might be a passage followed by its opposite with the
negative particle na in front of each verb. The good man does this, does that, does
something else. The bad man does not do this, does not do that, does not do something
else. Or there might be a set of phrases in each of which one word is changed: “We go to
the Buddha for refuge, we go to the Dhamma for refuge, we go to the for
refuge”.
Although it is unlikely that the circumstances were completely identical, the stories in
the Majjhima-nikāya about the Buddha-to-be visiting the two teachers, which I
mentioned last week, starts off with identical phrases, with changes made only to cover
the different level of meditation reached, and the different response of the teacher when
his pupil has successfully imbibed the teaching. Sometimes translators get carried away
when they find such repetition. It is not, in this particular example, as exact as one would
believe from reading Miss Horner’s translation, in which she makes the stories even more
parallel, even more repetitive, than the Pāli justifies.24
Repetition reaches its highest (or lowest, depending on how you look at it) level in
Pāli, in my experience, in the Alagaddūpamasutta of the Majjhima-nikāya.25 In that sutta
a bhikkhu called developed the erroneous view: “In so far as I understand the
dhamma taught by the Bhagavat, it is that, in following those things called stumbling
blocks by the Bhagavat, there is no stumbling block at all”.26 The text tells us that other
monks heard that had developed the erroneous view that “In so far as I
understand the dhamma taught by the Bhagavat, it is that, in following those things called
stumbling blocks by the Bhagavat, there is no stumbling block at all”, and so they went to
him and said; Is it true that you have developed the erroneous view that “In so
far as I understand the dhamma taught by the Bhagavat, it is that, in following those
things called stumbling blocks by the Bhagavat, there is no stumbling block at all”, and
replied; Yes, I have developed the view that “In so far as I understand the
blocks by the Bhagavat, there is no stumbling block at all”. The text goes on to say that
despite all their efforts, maintained his view that “In so far as I understand the
blocks by the Bhagavat, there is no stumbling block at all”. So the bhikkhus went to the
Buddha, and told him that had developed the view that “In so far as I understand
the dhamma taught by the Bhagavat, it is that, in following those things called stumbling
blocks by the Bhagavat, there is no stumbling block at all”, and that they had heard that
he had developed the view that “In so far as I understand the dhamma taught by the
Bhagavat, it is that, in following those things called stumbling blocks by the Bhagavat,
there is no stumbling block at all”, and so they had gone to him and asked him if it was
true that he had developed the view that “In so far as I understand the dhamma taught by
the Bhagavat, it is that, in following those things called stumbling blocks by the
Bhagavat, there is no stumbling block at all”; and he had replied that he had developed
the view that “In so far as I understand the dhamma taught
24 Horner, 1954, 207–210.

25 M I 130–42.
26 Bhagavatā ājānāmi yathā ye ’me antarāyikā dhammā vuttā
Bhagavatā te antarāya, M I 130, 6–8.
by the Bhagavat, it is that, in following those things called stumbling blocks by the
Bhagavat, there is no stumbling block at all”, and yet despite their efforts, he had
persisted in his view that “In so far as I understand the dhamma taught by the Bhagavat, it
is that, in following those things called stumbling blocks by the Bhagavat, there is no
stumbling block at all”. And so the Buddha sent for , and asked him if he had
developed the view that “In so far as I understand the dhamma taught by the Bhagavat, it
is that, in following those things called stumbling blocks by the Bhagavat, there is no
stumbling block at all”, and replied that he did have the view that “In so far as I
understand the dhamma taught by the Bhagavat, it is that, in following those things called
stumbling blocks by the Bhagavat, there is no stumbling block at all”. Whereupon the
Buddha condemned it as a wrong view.
That is to say that, if my arithmetic is correct, the identical passage occurs 12 times,
and it was partly for that reason that I used to include that sutta very early in my Pāli
course—the doctrinal importance of its contents was another reason. For beginners in
Pāli the sheer size of the vocabulary to be mastered is a deterrent, since every line of
every sentence presents new words which have to be understood and committed to
memory. The amount of repetition in the Alagaddūpamasutta means that students
suddenly find that they already know the words, and they begin to think that they are
making progress as the same sentence describing the erroneous view occurs again and
again, until they are able to chant out the heretical statement as well as any ,
proving the value of repetition when memorising.
To us, repetition carried to such an extreme length is ludicrous, and we should
certainly try to avoid it, introducing such phrases as “This view”, “such a view”, “the
view which you mention”, etc., yet quite clearly such repetition was not regarded as
otiose at the time of the oral tradition, although as we shall see in the fifth lecture when
we consider Buddhism and writing, the situation changed somewhat at a later time.
Stock phrases, lists, the waxing syllable principle, repetitions, are all things which
make the memorising and recitation of prose texts easier. In addition to these there are the
usual literary features of alliteration, etc., which help to determine the choice of words to
use, and to assist in remembering them.
But despite the help which such aids gave, and despite the care which the
may be assumed to have taken in reciting the texts allotted to them, it is inevitable that
mistakes would creep into the tradition by reason of the type of error which is inherent in
oral recitation, particularly when monks from different parts of India, with different
pronunciations of Pāli, were assembled together. Buddhaghosa gives a list of errors
which would invalidate a kammavācā—an official action of the : uttering an
aspirated consonant when it should be unaspirated, and vice versa, de-voicing a voiced
consonant and vice versa, etc., all of which would be of general application in oral
recitation.
In addition to such errors, the dangers inherent in an oral tradition are obvious.
Handing texts on orally depends on two essentials: a donor to hand them on, and a
recipient to accept them. If interest in a religion wanes, and no one wants to hear a text, it
dies out, or if the number of donors of a text are reduced in number to a few old men,
with failing memories, then the text is likely to be handed on partially or incorrectly, or
not at all.
At some time after the introduction of Theravādin Buddhism into Ceylon, war and
famine and the destruction of vihāras led to a breakdown in the system, and to
a situation where some texts were known to a very few bhikkhus, Buddhaghosa records
the fact27 that there came a time when only one bhikkhu knew the Niddesa, and urgent
measures had to be taken to have him repeat his text to receivers before he died. From
fear of the Niddesa disappearing completely, the thera Mahārakkhita was persuaded to
learn it from this one bhikkhu, and other theras learnt it from Mahārakkhita, so that the
future transmission of the text was assured.28
Such incidents no doubt had an effect and gave a warning to the , and made
the theras in Ceylon realise that the whole canon could disappear if the oral tradition died
out. This was probably one of the factors which persuaded the that it was time
they made use of the new-fangled writing. The interaction between the oral tradition and
the written tradition, and the effect which that had upon Buddhism are the subjects of my
fifth lecture.
Before then I want to talk about the way in which the Buddha’s message spread from
Magadha, where he had first delivered it, throughout the North of India, down south to
Sri Lanka, and North through Chinese Turkestan to China and elsewhere. And it is the
first part of that movement, away from Magadha, which will be the subject of the fourth
lecture. In that lecture I will deal with the philological information which we have about
the form of the language which was used in the very early stages of Buddhism, and the
way in which we can interpret philological material to give us some idea of the way in
which Buddhism began to spread from the boundaries of its origin.
27 Sp 695–96.
28 See Norman, 1983C, 87; and 1988, 1–27 (15) (=CP III, 225–43 [236]).
IV
Buddhism and Regional Dialects
In the third lecture I dealt with the way in which the early Buddhist texts were transmitted
by an oral tradition, and I mentioned the difficulty I had in believing that the
tradition could have come into effect within a very short time of the Buddha’s death,
because this did not seem to leave time for the homogenisation of language, which would
seem to have been necessary if the Buddha preached, as is generally accepted, in a
number of dialects. In this lecture I want to talk about the philological information which
we have about the form of the language which was used in the very early stages of
Buddhism, and the way in which we can interpret philological material to give us some
idea of the way in which Buddhism began to spread from the boundaries of its origin.
We very commonly find in books and articles about early Buddhism such statements
as: “The Buddha preached in the Prakrits, the language of the common people, and
resisted the suggestions of some of his ex-brahman followers to translate his sermons into
Sanskrit”. There is frequently no hint that these statements are anything other than
accepted fact, but readers need to be very wary, because such statements are frequently
not fact, and are anything but accepted by all scholars working in the field.
To tell the truth, there is a great deal of the Bellman principle in the academic world.
You all know about the Bellman in Lewis Carroll’s The Hunting of the Snark, who
maintained that “what I tell you three times is true”. I am as guilty in this respect as
anyone else, I fear. I may have an idea about something, and so I incorporate my idea, as
a suggestion, in an article I am writing, and wait for someone to reject or disprove it. No
one does, and I repeat the idea, still as a suggestion, in another article. Again, no one
rejects it. I do this a third time, and if there is still no reaction, it becomes fact in my
mind—I have said it three times, so it must be true, and I consequently refer to it in future
publications as an established fact. The thought that no one ever reads my articles and so
no one has ever seen my suggestion, and so no one has had any desire to reject it, or the
alternative explanation, that those who read my first article thought that the idea was so
preposterous that it was not worth wasting paper and ink refuting it, so that the second
and third repetitions were dismissed as “I see Norman is still pushing that stupid idea of
his”, does not enter my head.
This reference to the Buddha refusing to allow his sermons to be translated is based
upon a passage in the Vinaya,1 but to translate the passage in that way is to ignore the
1 na bhikkhave chandaso , Vin II 139, 13–14

Buddhism and regional dialects 51
very real difficulties which are presented by two words in the original Pāli, chandaso
and nirutti. The first has been translated “into Vedic”, or “into Sanskrit”, or “metrically”,
or “as desired”. I cannot believe that “into Vedic” is a possible translation, but if “into
Sanskrit” is a possibility, then this means, of course, that the Buddha’s words were not
already in that language. In support of this translation, it has been pointed out that the
bhikkhus who made the suggestion were brahmans by birth, who might be thought to
favour Sanskrit in preference to the local vernaculars.
On the other hand, we might have thought that such learned people would have the
sense to realise that the Buddha preached in Prakrit because that was what the local
people spoke and understood, and it would be an act of folly to turn his sermons into
something which would be unintelligible to them. There is the additional point that, if
chandaso does mean “into Sanskrit” this would appear to have been unknown to later
translators who did turn the Buddhavacana into that language. Buddhaghosa seems to
have understood the bhikkhus’ suggestions as meaning: “Let us translate, as we translate
the Veda into Sanskrit”, presumably referring to a situation where brahmans would be
able to explain difficult words in the Vedas, when they recited them, in the same way as
we know that Sanskrit commentaries were written upon the Vedas. This, then, might
perhaps mean that the ex-brahmans were asking permission to gloss, i.e. explain, the
Buddha’s words as they recited them.
If, however, we accept either of the translations “metrically”, or “as desired” for
chandaso, then this tells us nothing about the language which the Buddha used. In view
of this uncertainty, we should perhaps not put too much weight upon this problematic
passage, and we should instead try to work the matter out for ourselves in another way.
Is there any evidence that there were, in fact, dialects in use at the time of the Buddha?
We have Buddhist texts in both Sanskrit and dialects of MIA. How do we know that the
Buddha did not preach in Sanskrit, and the translations from Sanskrit into MIA were not
made later?
We must assume that the language(s) of early Buddhism reflected the linguistic
background of the times. The beginnings of Prakritic tendencies in Indo-Aryan can be
placed very early. In his examination of the history of the Proto-Indoaryans, Burrow
pointed out2 that in the traces of their language found in the Hittite archives, the word for
“seven” is šatta (cf. Skt sapta), and he referred to this “as having evolved beyond the
Proto-Indoaryan stage”.3 We find in the a number of characteristics which we
would regard as Prakritic, i.e. they do not present the form which our knowledge of Indo-
European philology tells us we should expect in early Sanskrit. Many lists of such forms
have been published.4 They include the semi-vowel appearing as u in muhur and in
pitus; disappearing and making a following dental -t- retroflex in , beside the
expected form . We can see that in some places the metre demands the insertion of
a svarabhakti vowel, e.g. . Even though we must make it clear that we are talking
about the recension of the which we possess now, which is perhaps no earlier
2 Burrow, 1973, 123–40.

3 Burrow, 1973, 125.
4 See Elizarenkova, 1989, 1–17; and Pinault, 1989, 35–96.
than the seventh century B.C.E.,5 nevertheless we can see that such features were already
in evidence well before the time of the Buddha. If there were Prakritic features in the
literary language of the , the assumption is that the spoken languages of the
ordinary people had many more of these features.
Our view about the prevalence of MIA features at this time is reinforced by the fact
that about 400 B.C.E., i.e. about the time of the Buddha’s death, the Sanskrit grammarian
produced his grammar, the prime aim of which was, we may assume, to establish
the form of Sanskrit inflexibly, so that MIA could not affect it—although, as will be clear
from what I have just said, to some extent he was closing the stable door a little late.
Patañjali’s , dating from the second century B.C.E., actually quotes some
of the forms which must be avoided, and they include forms which are well attested in
MIA.6
We need then have no doubt that the MIA dialects had been formed, and were in use
during the time when the Buddha was teaching, but we have little knowledge about the
dialect geography of the fifth century B.C.E. If we ask: “What was the language of the
Buddha?”, then we have to admit that we do not know for certain what language or
languages the Buddha spoke.
Any movement, whether religious or social, which opposed the power and status of
the brahmanical caste might be expected to make use of a non-brahmanical language. Not
only would this be an anti-brahmanical gesture, but it would also aid the non-Sankrit-
speaking element of the population, i.e. those who spoke a vernacular language. It
therefore seems very likely that the Buddha’s sermons were preached in a non-Sanskritic
language, i.e. a Prakrit, and from the fact that he moved about preaching in various places
we can assume that he preached in a number of dialects, varying his language to suit his
audience. Much of his teaching life was spent in Magadha (although none of the great
events of his life occurred in that area). We can therefore assume that on some occasions,
at least, he used the Magadhan dialect of the time, which we may call Old Māgadhī.
This belief might appear to be supported by the fact that the Pāli commentarial
tradition tells us that the Buddha’s language was Māgadhī. Unfortunately the
commentators use this name about the form of the canon which they had before them, i.e.
the language which we call Pāli, and as we can be quite certain that Pāli is not Māgadhī,
we have to examine carefully just what they meant when they said this. It is probable that
they believed that the Buddha spoke the words as they were in the canon, i.e. in Pāli, and
as they knew he lived and uttered the words in Magadha, they believed that Pāli was
Māgadhī.
We do not know precisely the form which Māgadhī had at the time of the Buddha. We
can deduce to some extent what form that dialect had at the time of Aśoka, and we know
what the grammarians writing some centuries later said were the characteristics of the
dialect. Extrapolating from this, we can gain what we may regard as a fairly accurate idea
of the main features of that dialect at the time when the Buddha was teaching. Its main
5 For other evidence see Bloomfield & Edgerton, 1932, who give evidence for voicing and
unvoicing (§§ 44–79), aspiration and de-aspiration (§§ 80–124), /ts/ps >cch (§§ 183–86), y>j,
and j>y (§§ 192–93), the development of - ->-a-/-i-/-u-(§§ 631–44).
6 e.g. , , ( [ed. F.Kielhorn], I, 259).
characteristics would have been: l for r, palatal ś for all sibilants, -e as the nominative
singular ending.
If this is correct, then, as I have said, the language of the Theravādin canon which we
have is certainly not Māgadhī. That would indicate that the language has been translated
on at least one occasion, presumably to meet the needs of the situation. As Buddhism
moved from the land of its origin into areas where different dialects or languages were
spoken, and as those dialects or languages developed and changed over the course of
time, it would seem to be inevitable that some sort of translation process was needed if
the Buddha’s teaching was not to become unintelligible to those to whom it was
preached. Are there any traces of this? Yes, there are. If we analyse the language of the
Theravādin canon, i.e. the language which we call Pāli, we can see that for the most part
it has features which we would class as western, using this in a linguistic rather than a
geographical sense, since we find eastern forms in the version of the Aśokan inscriptions
in the West at Sopārā. Nevertheless, we can detect anomalous forms in it, i.e. forms
which do not seem to follow the western patterns of phonology and morphology of the
language. Some of these have features which are more appropriate to Sanskrit, e.g.
consonant groups containing -r-, and the absolutives in -tvā. Disregarding these for a
moment, we can see that there still remain a number of other anomalous forms. There are
forms with eastern kkh where we should expect western cch, e.g. bhikkhu and bhikkhunī;
forms with l where we should expect r, e.g. verbs with the prefix pali- instead of pari-;
forms with -e where we should expect -o, e.g. bhikkhave instead of bhikkhavo, and also
some nominative singular endings in -e, instead of -o; examples of the voicing of
consonants, e.g. yādeti; forms with intervocalic y instead of k or t; forms with j where we
should expect y, e.g. jantāghara; dental n instead of retroflex , e.g. in nibbāna; v where
we should expect y, e.g. āvuso from the noun āyu(s); and a small group of words where a
consonant group including a nasal has developed in an unexpected way, e.g. nt>nd,
nd>nn and mb>mm.7
It is generally agreed that these are dialect forms, from one or more dialects, and they
may be regarded as being remnants of dialects through which the Buddhavacana was
transmitted before it was translated into the language which we find in the Theravādin
canon, i.e. Pāli.
In what I say in this lecture I am, to some extent, a prisoner of my own nomenclature.
I shall be using “Pāli” as the name of a language, and it has become conventional to do
so. But Pāli is an abbreviation of pāli-bhāsā, which means the language of the pāli, i.e.
the texts, or the canon. It follows, then, that every feature of the language of those texts,
however strange, and however inconsistent with the rest of the language of the texts, is
Pāli. We can define it, if we wish, as a western dialect with some eastern features,
although this in itself is a matter for debate, as we shall see, but it would, technically, be
incorrect to talk about these possible eastern features as anomalous forms. It is even more
incorrect to call them non-Pāli forms. Nevertheless, I must call them something in the
course of my discussion about what to call them, so I shall in general call them
“anomalous” or sometimes “eastern”, and when I want to stress that the Pāli being used is
free from such eastern forms, I shall call it western. This may seem confusing, but I hope
that all will become clear as I go along.
7 See Norman, 1989C, 369–92 (=CP IV, 46–71).

The presence of these anomalous forms in the Theravādin canon was, of course,
noticed a long time ago, and their value as indicators of an earlier form of the
Buddhavacana has been much studied and debated. We have to ask ourselves what their
significance is. Why they were retained from an earlier version, and what conclusions can
we draw from their retention? Some scholars have regarded them as evidence of an Ur-
kanon, (“a primitive canon”),8 while others have seen them as pré-canonique (“pre-
canonical”).9 The first of these views pre-supposes that there was a canon in existence at
some pre-Pāli time, and while this is possible and indeed quite likely—although this
depends to some extent upon the definition which is given to the word canon (something
to which I shall return in the eighth lecture)—these forms cannot be taken as proving this.
All they prove is that some texts, i.e. the ones in which we find the anomalous forms,
existed at an earlier date in a dialect or dialects other than Pāli. Even this is overstating
the case, because it is quite clear that the use of some of these forms was extended by
analogy. I shall return to this in a moment. At this point I will only say that there is
probably no one reason for their retention.
If these anomalous forms are remnants of dialects through which the Buddhavacana
was transmitted, then our task is to identify the dialects and the areas in which they were
used, in the belief that this will give information about the regions of North India through
which Buddhism spread. The problem is to see whether our anomalous forms exhibit any
of the characteristic features of any of the dialects about which we have knowledge, in
the hope that this will tell us about the regions through which the texts, or parts of them,
were transmitted. To guide us in our search we have two main aids: the inscriptions found
in India, from the time of Aśoka onwards, and the statements of the grammarians. Both
sources are to some extent unreliable, and in any case refer to a time later than the
Buddha. I mentioned in the second lecture the belief of some that the Buddha’s death is
to be dated c. 400 B.C.E. This means that the Aśokan inscriptions are about 150 years
after that date, and so we have to allow that much time for linguistic change, and we have
to accept that the Aśokan inscriptions cannot be entirely satisfactory as a guide to earlier
dialects and dialect forms. Furthermore, although we can surmise that Aśoka intended his
inscriptions to be carved in the dialects appropriate to each site, there is clear evidence
that this was not always done.
The grammarians were writing many centuries later, from the fifth century C.E.
onwards, when the dialects had, in any case, been standardised by being used for literary
purposes, especially in the dramas, although a literary dialect or language very often
represents a fossilised version of a language used as a vernacular long before, and so may
retain very old features. As a third aid, there is also the information which can be gained
from Jain texts, which underwent the same type of language changes as the Buddhist
texts, i.e. different dialects were used as Jainism spread. As in the Buddhist texts, there is
the same difficulty of identifying the languages and the areas where they were used.
Nevertheless, we can, to some extent, use Jain texts to help with the identification of the
dialects of Buddhist texts, and vice versa.
We know nothing about the translation techniques which were employed by those
taking the Buddhavacana to an area where a different dialect was used, and who were
8 e.g. Lüders, 1954.
9 e.g. Lévi, 1912, 495–514.
therefore faced with the task of translating into a new dialect If the donor dialect (i.e. the
dialect from which the translation was being made) and the receiving dialect (i.e. the
dialect into which the translation was being made) had differences of phonology and
morphology, then much of the translation process could have been done almost
mechanically. For example, if the translation was being made from a dialect which did
not voice intervocalic consonants into one which did voice them, than the translator
simply had to voice all unvoiced intervocalic consonants. If the translation was being
made from a dialect where the nominative singular of short a-stem nouns was in -e into a
dialect where the nominative singular was in -o, then the translator had simply to change
all nominative -e endings into -o.
There would, however, have been certain items which remained unchanged: 1) words
which had a specific sanctity attached to them because they were regarded as technical or
semi-technical terms, and were therefore, despite the air of strangeness which they must
have presented, too important to change; 2) words which had no equivalent in the
receiving dialect, and which therefore had to be retained; 3) words retained by an
oversight.
The accuracy of such a translation process depended on the knowledge and ability of
the translator. Clearly, the translator had to know something about the characteristics of
the donor and receiving dialects. If, by any chance, the donor dialect was one which
voiced intervocalic consonants, and the receiving dialect did not voice them, then an
mechanical translation technique was likely to produce errors, because it would lead to a
situation where consonants which should be voiced in the receiving dialect might become
unvoiced. It was all very well changing all past participles in -ida in the donor dialect into
-ita in the receiving dialect, but what about a form like uppāda? Since there is a root pad-
as well as a root pat-, should this be translated into uppāta, or left as uppāda? Only the
sense of the passage could determine this, and if the passage was ambiguous, or for some
other reason it was difficult to understand the meaning of the passage, then a translator
was left to guess, and his guesses might not always be correct. If the receiving dialect had
the nominative singular of short a-stem nouns in -o, then to turn all -e endings into -o
may well have meant that forms which should have had the -e ending in that dialect, e.g.
locative forms, were also changed, incorrectly, into -o.
Let us look at some of these anomalous forms and see what we can deduce from them.
It has been pointed out that the Buddha’s mother Māyā was probably not called
“Delusion”, but “Mother” (Mātā).10 This development of intervocalic -t->-y- is a
characteristic of , but it is scarcely conceivable that this detail of the
Buddha’s life story was transmitted from Magadha to (where our evidence
is in any case from a later time—probably after the writing down of the Pāli canon)
before it entered the Theravādin tradition.
The change of intervocalic consonants to -y- is not a feature of the eastern dialect of
the Aśokan inscriptions—there are only one or two examples, which can probably be
explained otherwise11—but the -ikya forms found at Kālsī, where the other versions have
-ika, may reflect some such change. We do not know precisely what the writing of the ky
10 von Hinüber, 1991A, 183–93 (187).

11 See Norman, 1970, 132–43 (136–37) (=CP I, 93–107 [98–100]).
ligature means, but it is possible that the scribe was writing it because he received k in his
exemplar, but wanted to show that his own pronunciation or the pronunciation in his area
was nearer y. Lüders has given some examples of -ika/-iya alternations,12 and the fact that
the change of -k->-y- occurs after -i- seems to support the view that ikya at Kālsī does
indicate that the change >iya had taken place or was beginning to take place.13 It is,
therefore, possible that the sound change found in Pāli Māyā reflects the fact that in some
part of Magadha, perhaps in the West, towards the Kālsī area, this change was operative.
Moreover, it has been shown that some of the etymologies in the Sabhiya-sutta of the
Pāli Sutta-nipāta depend upon their being first pronounced, i.e. composed in, not just
transmitted through, a dialect where some, at least, of the intervocalic consonants had
developed into y.14 The age of the sutta is shown by the fact that a Buddhist Hybrid
Sanskrit version of it exists in the Mahāvastu, where the bhikkhu is called Sabhika—the -
iya/-ika variation of the name in the two versions being in itself an example of the very
change I am talking about. If the age of the text supports the view that the version
underlying the Pāli and Buddhist Hybrid Sanskrit versions was composed in Magadha,
then this is additional evidence for the existence of this sound change in Magadha at an
early date.
Similar considerations would apply to the examples of voicing found in Pāli. It seems
clear that they are evidence for transmission through a dialect where voicing was usual,
and such hyper-forms as uppāta<Sanskrit utpāda show that the translator had knowledge
of a dialect where voicing occurred. Voicing is a typical feature of Śaurasenī, but there
would be considerable problems if we had to assume transmission through that dialect.
Although it is not a consistent characteristic of any of the Aśokan dialects, nevertheless
there is some evidence for voicing in the Aśokan inscriptions, e.g. libi for lipi, thuba for
thūpa, and the hyper-forms which occur, e.g. the root pat- written for pad-, show that the
scribes had knowledge of a dialect where voicing occurred.
We find a very small number of forms which show an anomalous development of a
consonant group containing a nasal: hanta>handa; the Buddha’s charioteer name
Chanda>Channa; ālambana> . These changes are typical of the Gāndhārī
dialect as seen in the Gāndhārī Dharmapada, but although this text is said to show the
characteristics of the Gāndhārī dialect several centuries earlier than the manuscript itself
(which dates perhaps from the third century C.E.), these features are not found in the
version of the Gāndhārī dialect which we find in the Aśokan inscriptions in the North-
west They therefore seem to be a later development in that dialect, and it is most unlikely
that these forms could be borrowings into Pāli from the Gāndhārī dialect. I should rather
favour the view that since our knowledge of early dialect geography in India is so
unreliable, there may well have been unattested dialects which left some mark upon early
Buddhist texts.
The ending -e in place of -o is perhaps the most widely attested of the anomalous
forms. It is a standard feature of the Māgadhī dialect and the eastern versions of the
12 See Lüders, 1954, §§ 89–90.

13 See Lüders, 1954, §§ 88.
14 See Norman, 1980B, 173–84 (177) (=CP II, 148–61 [154]).
Aśokan inscriptions, and it also occurs in the Gāndhārī dialect and the Sinhalese
Prakrit. We find in the Dīgha-nikāya descriptions of the teachings of the six teachers who
were contemporary with the Buddha, and some of these descriptions include nominative
singular forms in -e,15 as well as a number of other anomalous forms.
Another place where such -e forms occur is in the framework of the Kathāvatthu, a
work which is acknowledged by the Theravādin tradition to be one of the last additions to
the canon, since it is said to have been recited at the time of the third which was
held at during the reign of Aśoka. We know, then, both where and when it
was composed, and it is therefore not surprising that we find in it a number of features of
both phonology and morphology which coincide with the eastern versions of the Aśokan
inscriptions, e.g, the way in which occurs with an emphatic h-, i.e. .16
One of the interesting features of this text is the way in which the translators seem to
have understood very well the difference between the predominantly western Pāli forms
and the eastern forms which are found on almost every one of its 600+ pages. The text
consists of a discussion of certain statements, of which 500 were orthodox and 500
unorthodox according to the commentary, although these numbers must have been used
in a typical “round number” way, since, in its present form the text, in fact, contains less
than 300 unothodox statements. They are set in a framework of question and answer. All
the statements, heretical or otherwise, are free from anomalous eastern forms in -e, except
for one stock phrase which is repeated a number of times. The framework, however,
shows two speakers using -e and -o dialects which were probably completely
differentiated at one time, but which, by the time the text had become fixed in its present
form, had become mixed in a consistent way. We cannot say whether from the time of the
text’s composition one of these dialects used exclusively -o forms, as it does in the
present version, If it did, then this would be evidence for the existence of a dialect with -o
forms in Magadha c. 250 B.C.E. The original differentiation could, however, have been
between two sub-dialects of Māgadhī, which later translators thought were too similar to
be easily distinguishable by non-Māgadhī speakers.
Among the anomalous forms in Pāli, attention has been drawn recently17 to the word
je, which is a particle used when addressing women of a lower class. It is explained as
being a shortened form of the word ajje, which would be an eastern form of
ayya<Sanskrit ārya “noble”. I have great doubts about this etymology. It may be possible
to find evidence for a word which originally meant “noble” being used in a pejorative
sense, but I would find it difficult to accept this derivation when the word ajje from
which it is said to have been derived was still in full use as an honorific form of address
in the eastern dialect in which je must have developed its pejorative sense. I personally
believe that je is the emphatic particle, which appears as ye in Pāli and the Aśokan
inscriptions,18 and I believe that it is an example of the very rare change of initial and
intervocalic y>j in Pāli, which is seen also in the word jantāghara. My doubt about the
15 See Norman, 1976C, 119–21 (=CP I, 240–42).

16 See Norman, 1979A, 279–87 (282) (=CP II, 59–70 [64]).
17 By von Hinüber, 1993, 101–13.

18 See Norman, 1967A, 160–70 (=CP I, 47–58 [50–51)].
derivation <ajje is supported by the occurrence in Jain Prakrit of , which is used

in a very similar sense to je.19
There are also hyperforms based upon the development of y>j. John Brough20 remarked:
“They were aware of the Prakritic tendency to voice intervocalic stops of the literary
language, and in attempting to combat this tendency, they occasionally overreached, and
produced monstrosities such as Yamataggi for Jamadagni”. From the existence of the
name Yamataggi in the Theravādin canon we can deduce that the person responsible for
the production of this form was aware of the fact that the dialect from which he was
translating sometimes showed initial j- where his own dialect showed y-, and intervocalic
-d- where his own dialect showed -t-. When he came across the form Jamadagni (or more
likely Jamadaggi), he did not know its correct form in his own dialect, doubtless because
it was a name new to him, whose meaning and etymology were unknown. In the absence
of any knowledge of the correct form of the name in his own dialect, he was obliged to
back-form by rule, which led to the hyper-form. This shows that there was a pre-Pāli
dialect where the changes y->j-21 and -t->-d- occurred. This conclusion supports the
possibility that the words showing the change of initial y->j- which I have already
mentioned are dialect words in the Theravādin canon.
There is another word in Pāli which seems to be an example of this anomalous sound
change, i.e. niya<Sanskrit nija “own, belonging to oneself”. This may simply be an
example of an intervocalic consonant being elided and replaced by a glide -y-, i.e. it may
be a genuine -type form <nija, but it is also possible to explain it as a hyper-
, i.e. a translator who knew that intervocalic -y- became -j- in the donor dialect,
wrongly back-formed the -j- which he found in his exemplar into -y-, presumably not
recognising the fact that in this particular case nija was correct in the receiving dialect
also.
Brough was doubtful about the date when this sound change was operative. He said:
“The mere existence of the [Pāli] form Yamataggi then forces upon us the conclusion that
parts at least of the Pāli canon were translated from a MIA dialect in which initial y- had
already become j-. This seems to demand a seriously late date. But I can only pose the
question…”.22
Brough’s problem was that he knew of no evidence for the existence of a dialect
which turned y into j, either initially or intervocalically, at the time of the Buddha. We
know that some at least of the Jain canon was transmitted in such a dialect, where the
stem of the relative pronoun is, for example, ja, where Pāli has ya, and the optative
ending is -ejja, where Pāli has -eyya, but the Jain tradition tells us that the canon was not
19 See Schwarzschild, 1961, 211–17 (=Collected Articles, 104–16).

20 Brough, 1962, 249.
21 The antiquity of the change of y->j- is shown not only by the examples given in Vedic Variants
(see Bloomfield & Edgerton, 1932), but also by the fact that at 7.4.12 the etymology
of implies a form with j- (cf. Pkt jakkha). See T.Burrow, review of U.P.Shah: The Valmiki-
, Vol. I, Baroda 1972, in JRAS 1974, 74.
22 See Brough, 1980, 42.
written down until some nine centuries after Mahāvīra’s death, in the fifth century C.E.,
and although parts, at least, of the Jain canon certainly existed before that date, there is no
reliable way of dating any of the phonological features of the languages of the Jain canon.
The answer to Brough’s question is that we must surmise that there was an earlier dialect
where this change took place, and we have, by chance, a minute portion of evidence to
support this view. It has been noted that there is one occurrence of an intervocalic y
becoming j in the Aśokan inscriptions,23 and that is in the word for “peacock” in RE I(G).
It occurs as majūla at Kālsī and , with the eastern l replacing the r of Sanskrit
mayūra, and as majura in Sh and M, with the expected western r. The appearance of j at
four sites (the fifth, Girnār, has mora, just like Pāli) suggests very strongly that j was
taken over without correction by the scribes at all those sites from the exemplars which
they received. The dialect at Sh has a tendency to turn -j- into -y-, and would therefore be
unlikely to do the opposite, unless there was a good reason for doing so.
The change of -y->-j- is, however, anomalous in the Aśokan inscriptions. It seems not
to be a feature of the dialect which Aśoka’s secretariat at used, and it
presumably represents a remnant of Aśoka’s own dialect, which was probably a sub-
dialect of Māgadhī, i.e. a dialect spoken in just the area where we would expect the early
language of Buddhism to be in use. It seems very probable, then, that these words with j
in place of y came into Pāli from an eastern dialect. This might have been the actual
dialect used by the Buddha, but it was at any rate one of the dialects into which his
teachings were translated at an early stage in the history of Buddhism.
It is to be noted that this deduction, made on the basis of one single word, implies that
Aśoka’s own dialect differed from the eastern dialect attested in his inscriptions, because
if we assume that his dialect changed y>j, then the nominative singular of his form of the
relative pronoun would be je, whereas we know that the eastern versions of his
inscriptions have a form without initial consonant, i.e. e. This implication need present no
difficulties, because there is also evidence that Aśoka’s own dialect had palatal ś for the
sibilant,24 whereas no version of the Aśokan inscriptions, other than those from the
North-west, has this sound except by scribal idiosyncracy.
From this, then, we can deduce that at the time of Aśoka, and earlier in the case of
some Pāli texts which we can, with great probability, date to a pre-Aśokan time, almost
all the sound changes which we have marked out as anomalous in Pāli can be shown to
occur somewhere in the East, although sometimes only as a single example in the Aśokan
inscriptions.
We must therefore conclude that the information we have hitherto had about dialects
in Aśoka’s time, let alone at an earlier time, is deficient, since it by no means tells us
about all the dialects which were in use. The fact that, if we look carefully, we can
augment this information by detecting hints of other dialects is, in itself, of great
importance, because it means that we can use the same technique of marking out
anomalous forms as we have employed in the Pāli canon, to mark out the anomalous
forms in the eastern versions of the Aśokan
23 See Norman, 1980A, 61–77 (74, note 43) (=CP II, 128–47 [144, note 1]).
24 See Norman, 1980A, 61–77 (65) (=CP II, 128–47 [133]).
inscriptions, in an attempt to find the dialect which Aśoka himself used—a dialect which
is not represented, in its entirety, in any version now extant.
If our conclusion is correct, then, we can assume that the dialects in use, even in a
limited area such as Magadha, showed variations. This should not surprise us, because we
should expect neighbouring areas, even neighbouring villages, to have very slightly
divergent dialects. We can also assume that traces of this linguistic diversity were
retained when the sermons, which had been preached in different areas, were first
collected together and their language was homogenised. We can guess—but it is nothing
more than a guess—that when the first collection was made and homogenisation began to
take place, examples of divergence were even more numerous, since, as sub-dialects of
the Magadha area, these variations were probably not sufficiently great to cause
difficulties for speakers of other sub-dialects, and there was therefore no need to remove
them.
Our question must be: on the assumption that the process of homogenisation was
continued, while Buddhism was still, for the most part, confined to the Magadha area, so
that fewer and fewer of the anomalous forms remained, why do we find that some of
those forms survived the major translation into a western dialect which must have taken
place at a later date?
Some of these anomalous forms that I have been discussing are called, by some,
“Māgadhisms”, because they conform to the pattern of Māgadhī, as described by the
grammarians. Others, however, say that the term Māgadhism is misleading, because it
takes for granted that these forms are taken over from Māgadhī.25 It is true that it is
probably not entirely accurate to regard all such anomalous forms as Māgadhisms, in the
sense of their being words which were originally in a Māgadhī version of the
Buddhavacana, but were retained deliberately or accidentally at the time when translation
took place into a non-Māgadhī dialect.
Some of these forms are certainly found in other dialects, e.g. nominative singular
forms in -e occur in the Gāndhārī Prakrit, probably as the remnant of a linguistic area
which was split into two by a later tribal movement, after which the two halves migrated
to very different areas, one to Gandhāra and the other to Magadha. Nevertheless, to
suggest that the North-west was the source of such forms would demand a complete re-
appraisal of our ideas about the way in which the Buddha’s message was spread.
It has also been pointed out that other explanations can be given for some of these
forms, but some of these alternative explanations seem unnecessarily complicated. To say
that pure<Sanskrit was taken over into Pāli as puro, but then became pure by
analogy with agge, etc., or by the dissimilation of the vowels u and o, leads to a situation
where we have to assume that a Māgadhī form was converted into the correct western
form, but then underwent a change which converted it back to the same form which it had
in Māgadhī, but we are not, nevertheless, to regard it as a Māgadhism.26 There is unlikely
to be agreement about such views, since there is no way of proving or disproving either
hypothesis. In such circumstances, I should like to propose a theory of “the economy of
development”, which means that an explanation of
25 See Bechert, 1991A, 3–19.
26 See Bechert, 1991A, 13.
any sound change by a single stage of development is more likely to be correct than an
explanation by a multiple development. If we adopt this theory, then to take it as a
Māgadhism provides a simpler explanation.
What is, however, clear is that every Māgadhism does not prove that the passage in
which the Māgadhī form occurs is old and dates from the Māgadhī period of Buddhism,
and certainly those who say27 that we can rely too much on such details are correct. Once
a standard procedure had been adopted, e.g. of writing bhikkhave in a specific context,
and it was applied to newly created texts, the occurrence of such Māgadhisms tells us
nothing about the original language of the text in question, i.e. the occurrence of a
Māgadhism in a text does not prove that the text was originally in the Māgadhī dialect.
There was a great deal of extention of use by analogy. In just the same way, as we shall
see in the eighth lecture, phrases in common use in canonical texts do not prove that a
text is canonical, because such phrases were used by medieval writers to give their texts a
veneer of canonicity.
A recent discussion of some of the features I have just been talking about was entitled
“From colloquial to standard language”.28 I find the use of the word “colloquial” slightly
strange, because to my ear it has something of a pejorative sense. If the Buddha preached
in the dialect of the local people wherever he went, then we would say, I think, that he
made use of the local vernaculars. Many of these features also occur, as we have seen, in
the language of the eastern versions of the Aśokan inscriptions, which is normally
identified as the administrative language of the secretariat, and which can scarcely be
described as colloquial. The same discussion describes the difference between the particle
je and the feminine vocative ajje as representing a distribution of colloquial and standard
language.29 Again, this usage of “standard” is strange, in the first place because it is not
justified, but it is simply taken as fact that there was a standard language in Buddhism,
and in the second place because the term “standard” is not defined. We are left to deduce
that it means a language without “colloquialisms”, on which, perhaps, all existing
versions of Buddhist Indo-Aryan dialects and languages are based. It is hard to see how
Pāli could fit into such a pattern, because, as we have seen, Pāli does include such forms.
Perhaps “standard” means a literary type of language, but no evidence is given that such a
dialect ever existed. I would doubt that changes were made by translators simply because
some words were thought to be “colloquialisms” and therefore “non-literary”. It seems
much more likely that eastern forms were unacceptable, for phonological and
morphological reasons, to the western Prakrits into which the early Buddhist texts were
translated, and they therefore had of necessity to be changed, except when there were
reasons for their retention.
It is clear that many of these anomalous forms were retained because they were
technical or semi-technical terms, e.g. bhikkhu, bhikkhunī, nibbāna, bhūnahū,30 with the
last three showing a replacement of - - by -n- which in the Aśokan inscriptions is typical

28 von Hinüber, 1993, 101–13.
29 von Hinüber, 1993, 102.
30 See Saksena, 1936, 713–14.
of the eastern dialect(s). These words were, in fact, part of the basic vocabulary of early
Buddhism. We can surmise that the spelling of <Sanskrit “opportunity”
was retained because of its repeated occurrence in the phrase “do not let the opportunity
pass you by”. The expected western form exists, but in the sense of a particular
sort of opportunity, namely “a festival”. The particle je was doubtless retained in its
eastern form because the genuine western form ye was not used in this particular sense in
the receiving dialect into which the sermons were translated, although ye is used as an
emphatic particle after infinitives in Pāli in exactly the same way that je is used in Jain
texts.31 A word like āvuso also had a semi-technical sense, in as much as the Buddha had
laid down the specific circumstances in which it was to be used.32
It seems to me that the words attributed to the six teachers probably reflect (despite the
view of some to the contrary33) the actual dialects of their teachings, at least as they were
remembered at the time of the composition of the texts. In support of this suggestion is
the fact that comparable views ascribed to heretical teachers in Jain texts show close
verbal similarities. Other forms were probably kept by mistake, or because the translators
did not recognise the verbal root, e.g. āvudha in place of āyudha “weapon”. In some
cases it seems clear that the sense was ambiguous, and the translator could not determine
the correct form, e.g. there is a verse in the Theragātha,34 where we have three forms
ending in -e, sacce, atthe and dhamme, which can be eastern nominative or western
locative case forms, and we cannot be certain how we should translate them. The Pāli
commentary takes all three as locatives, but this seems rather forced, and if we look
elsewhere for an interpretation we have a choice of translating “In truth the meaning and
the doctrine are grounded” or “Truth is grounded in the meaning and the doctrine”.35 The
BHS translator took it in the second way, and changed one -e form to -am, making a
nominative satyam.
After examining the anomalous forms in Pāli, we can therefore say that we have
evidence that the texts of the Theravādin canon was transmitted through a mixture of
dialects or sub-dialects, almost all of which can be shown, or can be surmised, to have
been employed in the East at the time of Aśoka, and probably earlier.
I said earlier that the great majority of forms in Pāli are what we would call western,
using this in a linguistic rather than a geographical sense. This would indicate that the last
recension before the writing down of the canon was in an area where a western style
Prakrit was in use, at least for literary purposes. This was likely to be in the West of
India, but not necessarily so, since, if an eastern dialect could be used in the West, as we
have seen at Sopārā, there is no reason why a western dialect should not be used in the
East of India. Theoretically, the recension could have been made at the time of writing
the canon down in Sri Lanka, but this suggestion causes problems, because the Sinhalese
Prakrit which we find in the inscriptions of the first century B.C.E. does not resemble Pāli
very closely—it has, for example, eastern nominative forms in -e—and if we are looking
31 von Hinüber is misled by the Āgamaśabdakośa (see Überbl § 49) into believing that this usage
is only attested once. As Schwarzschild (1961) makes clear, it is quite common. For additional
examples see Oberlies, 1993, 78, s.v. je. For Pāli ye after infinitives see Norman, EV II 418.
32 D II 154, 9–15.
33 Bechert, 1991A, 13.
34 sacce atthe ca dhamme ca āhu santo , Th 1229cd.
35 In EV I 1229, I took the former interpretation; Udāna-v 8.14 takes the latter, writing satyam.
for an area where a western dialect was used, then this does not seem to be a likely
candidate.
The presence of Sanskrit forms in Pāli used to be taken as evidence for the belief that
Pāli was one of the oldest of the MIA dialects. This idea was based upon a view that MIA
showed a steady progress from Sanskrit to the last form of MIA, i.e. —the
very latest stage before the emergence of the New Indo-Aryan languages. A dialect which
retained a proportion of Sanskrit forms was therefore thought to be closer to Sanskrit, not
only in form but also in time. As more was learned about MIA, however, and as more and
more texts were found in Sanskrit or various forms of Sanskrit, owing a smaller or larger
debt to underlying MIA dialects, it became clear that this view was wrong. As some of
these Sanskrit forms in Pāli were recognised to be incorrect, and not genuine Sanskrit
forms at all, e.g. attaja is Sanskritised as atraja, instead of ātmaja, it became clear that
the Sanskrit forms in Pāli were late additions to the texts, i.e. they represented the result
of a limited re-introduction of Sanskrit or quasi-Sanskrit forms into the MIA original, not
the preservation of old forms.
We have no direct evidence as to where and when the Sanskritisms were introduced
into the Theravādin canon. Although a start had probably been made before Theravādin
texts were taken to Sri Lanka, nevertheless, we believe that the greater part of the
Sanskritisms were introduced in Sri Lanka, if only because we would start to date the
start of Sanskritisation rather late, probably not before the second century B.C.E.
Whether other changes, beside Sanskritisation, were being made to the language of the
canon at that time we do not know. Since it is clear from the commentators’ explanations
that updating of the language did occur, e.g. present participles with the old historical
nominative singular ending in the canonical texts are explained by younger forms
with the analogous ending -anto in the commentaries, we may well be correct in
believing that such things could also happen in the canon, although not in verse texts,
where the metre would act as a constraint upon any changes which would alter the
metrical length of a word.
We have no evidence that Pāli, either with or without Sanskritisms, coincided with any
historical language or dialect. It is not clear what we should call such a language. Some36
call it “artificial”, and certainly there are artificial features, such as the incorrect back-
formations just mentioned. The English language, however, has similar artificial features,
e.g. the g in sovereign, by analogy with reign, the s in island by analogy with isle, or the p
in in receipt because it is derived from Latin receptum, but no one regards English as an
artificial language, reserving the term for such invented languages as Esperanto. Others
call the Pāli language “literary”, and certainly it is the language of a literature, although it
is not literary in the sense that it represents the refined form of a popular dialect.
However we choose to describe Pāli, the chronicles tell us that in the first century
B.C.E. the Pāli canon was written down in Sri Lanka, and it is with the use of writing and
the effect which it had upon Buddhist texts and Buddhism itself that the fifth lecture is
concerned.
36 See von Hinüber, 1982, 133–40 (140).

V
Buddhism and Writing
I want in this lecture to say something about the impact which writing had upon
Buddhism, and about the deductions which we can make about the changes which took
place when Buddhist texts began to be written down—first, the changes in those texts and
second, the changes in Buddhism itself.
The , our earliest authority for the statement that the canon was written
down, states: “Before this time (i.e. the reign of King Abhaya, in the first
century B.C.E.1), the wise bhikkhus had handed down orally the text of the three
and also the . At this time, the bhikkhus, perceiving the loss of
living beings, assembled, and in order that the religion might endure for a long time, they
recorded (the above-mentioned texts) in written books”.2 It is not clear whether
“perceiving the loss of living beings” is to be taken generally—“everyone is going to
die”—or whether it refers to the specific events of the time. If the latter, then it could well
be a reference to the partial breakdown of the system, I mentioned in the third
lecture that Buddhaghosa records the fact3 that, at some time after the introduction of
Theravādin Buddhism into Sri Lanka, there came a time when only one bhikkhu knew
the Niddesa, and from fear of its disappearing completely a thera was persuaded to learn
it from this one bhikkhu, and other theras learnt it from him. This probably made the
in Sri Lanka realise that the whole canon could disappear if the oral tradition
died out.
Nevertheless, it is probable that there were other pressures too. It is very likely that the
so-called famine, foreign invasion from South India and also the
political and economic circumstances of the times played a part in persuading the
bhikkhus to make this decision. It has also been suggested that the growing power of the
newly founded Abhayagiri vihāra, and the threat which this offered to the Mahāvihāra,
could not be ignored.
1 89–77 B.C.E., according to Bechert, 199A1, 9.

2 ca tassa pi ca\ pubbe bhikkhu
mahāmati\\ disvāna tadā bhikkhu samāgatā\ dhammassa
potthakesu \\ Dīp 20.20–21.
3 Sp 695–96.
Other, later, sources give more information about the writing down of the canon,
presumably making use of information either unknown to, or at least unused by, the
author of the . Some authorities talk of a joint recitation ( ) being
held in the Āloka-vihāra before the writing down, and this is referred to as being “like a
fourth ” by some sources.4 This may, of course, be an invention dating from a
later time when there was some doubt about the authority of the written canon, because it
appeared not to have been authenticated by a . In this connection we should
remember the pattern at the fifth in the nineteenth century and the sixth
in the 50-s of this century: The canon was recited and then what had been
agreed was carved and printed, respectively. This clearly was the pattern which was
expected if an authentic version of the canon was to be produced. It is, of course, possible
that the committing of the canon to writing was a rather drawn-out affair, and was not the
result of a single , which is why there is no mention of one in the earliest texts. If
this was so, then it is certain that the story of the was a later invention. There is,
however, little doubt that we can accept that the writing down of the during the
5
reign of Abhaya was an historic fact.
The gives no information about the way in which the was
written down. We must assume that what was currently being remembered and recited
was repeated in the presence of scribes, who wrote it down from dictation. Unless the
various sections of the tradition had been co-operating and had been making
simultaneous changes, e.g. in the Sanskritisations which had been introduced, we must
assume that immediately prior to writing down the canon there were divergent features in
the nikāyas. Writing down would have been an excellent opportunity for the
homogenisation of forms—all absolutives in -ttā being changed to -tvā, and the forms
containing -r- being standardised, etc. The references to the in the
commentaries suggest that the oral tradition continued to function alongside the written
, and if the continued to act in the same way as before, they had
perhaps to relearn or, at least, revise their nikāyas, with these homogenised readings in
them.
In the case of the Pāli tradition, then, we may assume that the writing down of the canon
may have had some effect on the inter-relationship of the various traditions, if
they were compelled to co-operate in this way. It is noteworthy that, as I said in the third
lecture, there are differences of readings in parallel passages in different nikāyas, as we
have them now, and we find that the use of stock phrases, e.g. the formulae used to
describe bhikkhus approaching the Buddha, etc., are not identical. We may conclude that
any editorial process which occurred was not complete, but was restricted to certain
particulars.
The date when writing might have first been used by the Buddhists depends, of course,
on the date of the introduction of writing into India, and there is no agreement about this

5 Bechert, 1992, 45–53 (52).
Buddhism and writing 67
among scholars. As is well known, writing was widely used by the emperor Aśoka c. 260
B.C.E. in two scripts: and Brāhmī. The former, it is agreed by all scholars, I
think, was derived from the Aramaic script which was used by the administrators of the
Persian Empire, and its use was confined to the areas of North-west India which were at
one time part of the Persian Empire, i.e. the region around Gandhāra, and also Chinese
Turkestan into which the script was taken from Gandhāra. We have no inscriptions in
earlier than Aśoka, but the references made by the Greek historians to the use
of writing in the North-west of India at the time of Alexander presumably refer to it, and
if refers to writing, then since he came from the North-west, it was probably
that he meant.
There are, to simplify the discussion, three views about the origin of the Brāhmī
script.6 It is undoubtedly the case that the script used by the Indus Valley civilisation c.
2000 B.C.E. contained a small number of characters which were identical with, or very
similar in appearance to, characters in the Aśokan form of Brāhmī (although, of course,
until the Indus script is deciphered we do not know if the phonetic values were the same).
Some scholars, therefore, think that writing continued from the Indus Valley civilisation
right through to the third century B.C.E. Others, basing their theory on a small number of
characters which seem to have parallels, of both form and phonetic value, in an early
Semitic script, believe that writing came to India from further west. The third view is that
the script was invented by Aśoka’s scribes, or those of his immediate predecessors,
specifically for the purpose of inscribing imperial edicts. There are objections to this
view, in as much as the variety of forms which we find in the Aśokan version of Brāhmī
suggest that there was already a history of development of the script before Asoka’s
scribes used it. Support for this idea of pre-Aśokan development has been given very
recently by the discovery of sherds at Anuradhapura in Sri Lanka, inscribed with small
numbers of characters which seem to be Brāhmī. These sherds have been dated, by both
Carbon 14 and Thermo-luminescence dating, to pre-Aśokan times, perhaps as much as
two centuries before Aśoka.7
It is hard to believe that traders, operating throughout the Near and Middle East and
meeting fellow-traders from countries where writing had been used for centuries, would
not see the advantages of adopting a writing system. The same applies to administrators.
Once the North-west of India had become part of the Persian empire, if not before,
writing would have been employed by the Aramaic scribes in that area, and it is likely
that neighbouring rulers would realise the advantages of keeping records of regnal years
and royal accounts, and treasury and armoury details, in some tangible form. Aśoka’s so-
called “Queen’s edict” (if my interpretation of it is correct)8 specifically states that certain
charges are to be set against the queen’s name, in a way which implies that this system of
accounting was well known.
My own belief is that although one correspondence in both shape and sound between
Brāhmī and an early Semitic script might be simply chance, it is beyond the bounds of
coincidence for what I regard as at least three certain parallelisms of both form and
6 von Hinüber, 1989A.

7 Deraniyagala, 1990A, 149–68; and 1990B, 251–92.
8 See Norman, 1976A, 35–42 (=CP II, 52–58).

phonetic value to have occurred.9 I therefore take it as certain that at least those three
characters of Brāhmī are connected with the Semitic script.
There are other similarities of form between Brāhmī and early Semitic, but the
phonetic values are not the same. This partial correspondence does not, however, present
an insurmountable difficulty. There is precisely the same problem with the Carian
alphabet which was based upon the Greek alphabet, but did not always have the same
phonetic value as the Greek. What was borrowed was the idea of writing, with a small
number of direct parallels. This is not to say that these Brāhmī characters were
necessarily borrowed from the Semitic script at the time for which we have attestations of
its use. There is, in principle, no objection to believing that they were borrowed from a
later, at present unattested, version of that script.
I would therefore support the view that the Brāhmī script was in use in India well
before the time of Aśoka, and was probably brought from the West by traders, who had
borrowed the idea of writing, and some of the characters, from Semitic traders.
There is, however, no certain evidence for the use of the Brāhmī script in India before
the time of Aśoka—there is no agreement among scholars about the handful of
inscriptions found in India which have been claimed to be pre-Aśokan. It is, therefore,
appropriate to ask why, if writing existed in India before the time of Aśoka, there are no
records of it. The probable explanation is that writing was not at first used for religious or
literary purposes, but exclusively for administration and trading. Since there was no need
for the documents of administrators and merchants to last for ever, they were written on
ephemeral materials, and therefore perished.
Similarly, although the Gandhāra area had been a part of the Persian empire since the
time of Darius, there are no remains of pre-Aśokan Aramaic inscriptions from that area,
presumably because such records were similarly written on ephemeral material. Only
kings had inscriptions carved on rock—“that they might long endure” “as long as the sun
and the moon”, as Aśoka says—and there appears to be no reason to reject the much-
repeate suggestion that the wording of some of his edicts seems to contain echoes of the
words of Darius, which suggests that Aśoka’s inscriptions were a direct result of his
having heard about the inscriptions of Iranian kings at Behistun and elsewhere.
Even if writing was not used until the time of Aśoka, it is surprising that the
Theravādin texts were not written down until 200 years after Aśoka’s time. Since many
converts to Buddhism were and vaiśyas, who would have been acquainted with
the use of writing for accounts and lists of merchandise or military equipment, etc., it is
inconceivable that not one of them made use of his earlier knowledge after he had
become a bhikkhu. The opposition of the would probably have prevented any
attempt to transmit entire texts by writing, but it would be odd if sermon notes or aids to
recitation10 were not written down. There could be no religious reason for not using
writing. As I said in the third lecture, there is no mention of writing in the Vinaya rules,
either to regulate or forbid its use.
Moreover, by the first century B.C.E., writing in a semi-religious context had been
9 Initial a- and aleph, ga and gimel, and tha and theth (with the cross in the middle of theth reduced
to a dot in tha). See MW, xxvii.
10 As Tambiah (1968, 85–131) reported of a modern Thai monastery he visited.
widely used for a century or more in Sri Lanka, for donative inscriptions, etc. It is hard to
imagine that monks sitting in their caves with an inscription on the drip ledge did not
think of the value which writing might have for the perpetuation of the Buddhavacana.
We must then ask if it is an incontrovertible fact that Buddhists did not make use of
writing until the whole canon was written down in the first century B.C.E., or whether
there is any evidence at all for the use of writing before the time of the so-called fourth
. There are, in fact, hints which have led some scholars to suggest that writing
was indeed used, if only to a limited extent. Following Lüders,11 John Brough12 compared
a pāda in the Gāndhārī Dharmapada13 which ends in so vayadi “he goes” with the
corresponding pāda in the Udāna-varga,14 which ends in sa vrajati, with the same
meaning, and contrasted these forms with va sayati “he lies down” in the Pāli
equivalent,15 and he pointed out that in the context the idea of moving is more likely to be
correct than lying down. He explained the Pāli form as showing the metathesis of the two
va and sa, and noted that although such a metathesis might occur in a purely oral
tradition, it would imply “an unbelievably slipshod paramparā”. In manuscript copying,
however, this is a common and readily understandable error. The conclusion he arrived at
was that the text was already being transmitted by manuscript copying, and not
exclusively by oral tradition, at a date earlier than the redaction of the Pāli version.
It is not always possible to be certain whether such an error is the result of an oral or a
written tradition. Even the belief that an error is due to writing does not, in itself, prove
that the error was earlier than the writing down of the canon, since it could well have
occurred at some later date when the manuscript was being copied. We should, however,
note, that if this is the explanation, we have to assume that it happened at a time when
manuscripts of any individual text were very few in number, and probably unique. It can
be shown that this might well have been the case, for Alsdorf has pointed out that in one
Jain text16 the whole tradition available to us, not only all extant manuscripts but even the
oldest commentaries, go back to one single individual manuscript in which, through an
ordinary clerical mistake, one had been omitted.17
Similarly, although Brough wrote of a slipshod paramparā, we cannot rule out
completely the suggestion that there might have been such an imperfect method of
transmission. If all manuscripts of a particular Jain text and the commentaries upon that
text go back to a single, unique manuscript, then, in theory, at least, all manuscripts of a
particular Pāli text and the commentary upon it, may go back to a single, unique
, and, as I have said, we know that that was the situation with regard to at least
one Pāli canonical text. In short, I agree with Brough that writing was already being used
before the whole canon was written down, but I am not certain that the evidence he gave
proves it.
11 Lüders, 1954, § 105.
12 See Brough, 1962, 218.
13 so vayadi, GDhp 144.
14 sa vrajati, Udāna-v 1.6
15 va sayati, Ja IV 494, 2*.
16 Utt 25.7.
17 L.Alsdorf, 1962, 110–36 (134).
Others too18 believe that writing had already begun to be used on a small scale, and we
may assume that its use increased, as it became more and more common in the secular
world. In the absence of firm evidence, however, it is always possible for alternative
suggestions to be made about the cause of errors in the transmission, and scholars rarely
agree either about the probability of such errors being due to transmission by means of
writing, or about the date when they might have occurred.
I mentioned in the third lecture some of the errors which, despite all the care which the
took in their reciting, might creep into the texts because of mispronunciations.
Writing, however, brought with it a whole new range of possible errors, of the sort
with which we are familiar from the manuscript tradition of classical studies: writing the
same syllable or group of syllables twice (dittography); omitting one of a pair of
consecutive identical syllables or groups of syllables (haplography); allowing the eye to
jump from words at the end of one line or verse to the identical words at the end of
another line or verse, with the resultant loss of the words in between (homoioteleuton),
etc., and a glance at the critical apparatus of any Pāli text will provide abundant examples
of all these errors occurring in the manuscripts. Such errors are found from the very
beginning of writing in India, and examples of all of them can be seen in the Aśokan
inscriptions.
There was another type of error, however, which was due to deficiencies in the early
Indian writing system. In the earliest form of the Brāhmī script, double consonants were
not written, and the marks for long vowels were frequently omitted.
There are, in the Pāli canon, a number of textual variations showing alternative forms
with single or double consonants, which are most easily explained as having been written
at some time in such a script. When, in a later form of the Brāhmī script, the facility of
writing double consonants was adopted, in places where the metre was of no help in
determining the length of a vowel, e.g. in prose or in metrically doubtful positions, a
copyist had to decide whether the single consonant in the exemplar, which he was
copying, stood for a single or a double consonant. In most cases the context would make
this clear, but in passages where both forms could be considered to yield some sort of
sense, it was a matter of personal preference which form he chose to write or how he
chose to interpret it. The tradition sometimes remained ambivalent, with both possibilities
being handed down to us.
The writing of single consonants after vowels in metrically doubtful positions can
explain how doublets of this kind could arise.19 In the Therīgāthā we find a verse which
states, “There is no release from dukkha for you, even if you approach and run away”.
This seems slightly odd. The word for “approach” is the absolutive upecca20 (<Sanskrit
upetya), and the commentary sees the difficulty and explains that here it means sañcicca
“having considered”, i.e. “after consideration” or, perhaps, “deliberately”. The
commentary, however, also states that there is another reading, namely uppacca, which it
explains as uppatitvā “having flown up”. This seems to make better sense in the context
“there is no release from dukkha for
19 See Norman, EV II, 109 (on Thī 248).
20 Showing the palatalisation of -a->-e- before -c-. See Norman, 1976B, 328–42 (=CP I, 220–37).
you, even if you leap up and run away”, and it is very likely that this is the reading we
should adopt, especially as it is the only reading found for the verse in some of the other
places where it occurs. Here we can postulate the existence of an earlier form with single
consonants, i.e. *upaca. Since the first syllable is in a metrically doubtful position at the
beginning of a pāda, the scribe had no way of telling whether the word began with up- or
upp-. In the absence of such knowledge, upaca developed to *upeca when -a- was
palatalised to -e- before -c-. When the possibility of writing double consonants arose, the
tradition interpreted *upaca as uppacca (<Sanskrit utpatya) and *upeca as upecca
(<Sanskrit upetya).
In a prose passage in the Niddesa21 we find a long list of adjectives describing ascetic
practices, e.g. hatthi-vattika “following the practice of elephants”. There are variant
readings for each epithet, with a single -t-, e.g. hatthi-vatika “taking the elephant vow”.
In Sanskrit we find -vratika in similar contexts,22 but since the difference in meaning
between vratika and is very small, a scribe who was accustomed to a particular
writing system, where single -t- was written for both vatika and vattika, might have
difficulty in deciding which was the correct spelling, when the possibility of writing
double consonants arose.
Sometimes we find that a variation for a word appears, not in the commentary, but in the
scribal tradition, e.g. in a verse in the Theragāthā,23 where the Pali Text Society edition
reads , with double -mm-, there is a variant reading , with single -
m-. The first syllable, being the first syllable of the pāda, is in a metrically doubtful
position, so the metre cannot help us to decide between single and double -m-. In Pāli,
nimināti “he exchanges” and nimmināti “he constructs” are sometimes confused,24
because in certain contexts their meanings converge. In this verse, however, it seems
certain that we should adopt the variant reading with single -m-, since
“exchange” makes better sense in the context, which talks of “exchanging the ageing for
agelessness”. The word was presumably written as * in an earlier version, and
because of the doubt about the metrical length of the first syllable, either nim- or nimm-
was possible as an interpretation, and different branches of the scribal tradition
transmitted different readings.
Not only did the early Brāhmī script write double as single, but it also
frequently omitted anusvāra, e.g. the active “he smears” and the passive lippati
(<lipyate) “he is smeared”, could both be written as lipati, since the passive ending -ate
was often replaced by the active ending -ati in Middle Indo-Aryan. Exactly the same
could happen with the active muñcati and the passive muccati<mucyate, both of which
might be written as mucati. This would have caused no confusion as long as the oral and
written traditions continued side by side, as they must have done for some time. But
21 Nidd I 89, 17–29.

22 cf. Skt go-vratika, -vratika, cāndra-vratika.
23 Th 32.
24 The same confusion is found in BHS (see BHSD, s.vv. nimināti and nirminoti), e.g. at Mvu II
176,12 where Jones emends (1952, 170, note 3) nimmin- to nimin-, because Ja III 63, 9, 10 has
nimini and nimineyya.
when the oral tradition ceased, scribes were then entirely dependent on what was written,
which might be ambiguous, unless the context gave help. It is certain from the nature of
some alternations that scribes were entirely dependent on written forms, with no help
from any oral tradition.
Faced with a form mucati, a scribe who did write doubled consonants and anusvāra
had to choose between writing muccati and . If the context did not make the
choice obvious, then mistakes might occur. In a verse in the Patnā Dharmapada,25 for
example, we find the passive mucceya, where other versions of the pāda have the active
stem muñc-. There is considerable doubt about the meaning of this particular verse
among modern scholars, and it is clear that some ancient copyists were equally
perplexed, and were uncertain about the way in which to interpret the reading in their
exemplars. It is noteworthy that this possible confusion of active and passive continued,
to some extent, even after the use of double consonants and anusvāra became standard,
and we find in Pāli that muñcati is sometimes used for muccati,26 and vice versa.
There is, similarly, doubt about the alternation between the active and the
passive lippati, to such an extent that seems almost to acquire a passive sense.
The root lip means “to smear”, and in the passive “to be smeared, to be defiled, to cling
to”. Although the pattern is not entirely consistent in Pāli, it would appear that the case
usage with this verb polarised, with the instrumental being used with the sense of “defiled
by (something)”, and the locative with the sense of “clinging to (something)”, e.g. water
clings to, or does not cling to, a leaf. In the second usage, where there was an absence of
an obvious sense of passivity, a doubt possibly arose in the copyists’ minds as to whether
the word lipati which they found in their exemplar was not perhaps an active form, and
they consequently wrote the active . We consequently find that the Pāli
manuscript tradition is quite uncertain about this verb, and limpati and lippati are
sometimes found as alternative readings.
We can get some idea of the difficulties which such a writing system caused if we
examine a verse in the Gāndhārī Dharmapada, which is written in the script,
which has similar deficiencies, in that it does not write double consonants or long vowels.
When we try to interpret the words kamaradu and kamaramu in the Gāndhārī
Dharmapada,27 we are unable, because of the nature of the script, to decide
whether the first part of these compounds stands for kāma- or kamma-. If we look for
parallel or near-parallel passages in Pāli, to help us make a decision, we find that both
kāmarata and kammarata occur there.28 The occurrence of the two forms would seem to
be a clear indication that at one time that portion, at least, of the Theravādin tradition was
transmitted through a script which did not write either double consonants or long vowels.
Consequently the first part of the compound was written as *kama-, which could be
variously interpreted.
It has been pointed out29 that the so-called ablatives in - in Middle Indo-Aryan
25 PDhp 46.
26 See PED, s.v. muñcati.
27 GDhp 63.
28 kāmarata at A IV 438, 19 and kammarata at It 71, 16.
29 Brough, 1962, 79.
may be due to a transmission through manuscripts which, like the script of the Gāndhārī
Dharmapada, did not mark vowel length or nasalisation or, like the Brāhmī script of the
Aśokan inscriptions, did not do so consistently, so that from the point of view of the
written form an accusative of a short -a stem noun, which should be - , and an
ablative, which should be -a, would be identical, i.e. -a. We should expect that, when a
text written in such a manner was rendered into a full orthography, whether at the time of
translation into Pāli or subsequently, this would for the most part be done correctly, since
the scribe’s knowledge of the language would enable him to interpret from the context.
Equally, however, it would not be surprising if, from time to time, either because his
attention was wandering, or because the passage was genuinely difficult or ambiguous,
the scribe misinterpreted his exemplar, and wrote - where he should have written -ā.30
We find similar problems arising from a type of orthographic ambiguity which occurs in
some later manuscripts, where the copyists used dots to denote anusvāra and also to
indicate doubled consonants.31 This is not, strictly speaking, the use of an anusvāra to
indicate doubling, but the use of a dot which could be confused with the dot used for
anusvāra. The end result of this scribal habit was similar to that which I have already
mentioned. Without help from the context, it is difficult in a Middle Indo-Aryan text to
tell whether, say, lipati, written with a dot above the -p- in such a manuscript, stands for
lippati or .
There is a similar potential ambiguity found in manuscripts written in Burmese script,
where the i-mātrā is a circle above the consonant, and the anusvāra is a (smaller) circle
above the consonant.32 This therefore sometimes leads to a confusion between -i and -
. This phenomenon sometimes gives us a very useful criterion when investigating the
early history of a particular manuscript tradition. In Sinhalese manuscripts we sometimes
find - where we should expect -i, and vice versa. Such an error should not occur in
manuscripts in the Sinhalese tradition, because in the Sinhalese script -i and - do not
resemble each other at all. If, therefore, we find in a Sinhalese manuscript an error which
we can explain as being due to the confusion of the two , we can be certain that
at some stage in its transmission the text was copied from a Burmese manuscript,33 and
the copyist was mistaken in his identification of what he was copying.
I spoke in the third lecture of the value which repetition had in an oral tradition. It is
likely that another result of writing the canon down was that scribes began to realise that
there was no need to write out all the repetitions in full, since it was quite easy for a
reader to look back to the place in the manuscript where the passages had occurred
before, whereas it was not so good to say, when reciting, “I am going to leave out the
next few paragraphs because I said them ten minutes ago”. There arose, then, the practice
which we find in the manuscripts of writing the syllables pe or la which are abbreviations
30 Brough, 1962, 266–67 (ad GDhp 292).

31 See Brough, 1954, 361.
32 See Norman, 1989B, 29–53 (48, note 120) (=CP IV, 92–123 [117, note 2]).
33 Sadd 374, 29 quotes “ datvā ti pariyodāpetvā” from Tikap-a 269, 20,
where, however, is missing in Ee. We can postulate that it had been written/read as ti because
of this Burmese phenomenon, and the second ti fell out by haplology.
for peyyāla (<Skt paryāya), which means something like “formula”. In a text then, it
means “(here comes a) formula, i.e. a repeated passage”, which is the equivalent of “and
so on down to…”. Relying on such precedents, early European Pāli scholars also made
abbreviations in their editions of Pāli texts, writing for pariyādiyati, etc. This was
condemned by a number of Sinhalese scholars34 who pointed out that “to interfere, either
with words or letters, otherwise than is done by the peyyālams made use of by the
Arahats (who recited the texts at the various recitations), has frequently been declared to
be not good”. Consequently, volume I of the Pali Text Society’s edition of the
-nikāya, which abounded in such abbreviations, had to be withdrawn and a
new edition made.
Such indignation was perhaps an over-reaction to “western tampering with sacred
texts”, because these pe and la are not always in the same place in different
traditions, e.g. the oldest Pāli manuscript we know of—four folios of the Vinaya from
Nepal, which are about 1,000 years old—sometimes agrees with the Pali Text Society
edition in writing pe or la in certain places,35 but in other places gives no hint that there is
an omission, even though the Pali Text Society edition does write la.36
It is sometimes not made clear in western editions whether the omissions which
editors have made are those authorised by this ancient abbreviation system, or whether
they are the editors’ own innovations. Because the passages which have been abbreviated
seem to vary, in this way, in the different manuscript traditions, someone who recently
wished to make some calculations about the amount of repetition there was in a single
sutta had to go to a great deal of trouble to compare different traditions of that sutta, to
try to establish something approximating closely to the original unabbreviated version,
with all its repetitions kept intact.
The etymology of the word peyyāla, with its eastern l for r, perhaps indicates that it
came into use in Buddhism while the texts were still in an eastern dialect. That may mean
that the texts had already been written down before the translation into a western dialect
took place, or alternatively that this was a technical word, used in this sense in non-
religious writing in both eastern and western dialects (as a borrowing from the East in the
latter), and taken over into religion from that secular source. In just the same way we find
āva (<Sanskrit yāvat) “(and so on) down to” used in the Aśokan inscriptions, and also the
Jain Prakrit form jāva used in the same sense in Jain texts.
The fact that some characters were very similar to others in the Aśokan Brāhmī script
led to errors being made by the scribes who perhaps misread a badly written in
their exemplar. So, for example, because of the close similarity of ca and va, we
sometimes find that one version of an Aśokan Edict has va where the others have ca, and
vice versa. The same is true of pa and sa, and ta and na, etc. Comparable errors are found
in texts written in the various scripts of South and South-east Asia, which are derived
from the Aśokan Brāhmī, and those working with manuscripts written in such scripts
quickly learn the characters which, because of their similarity, may be confused by the
copyists.
34 See Report for 1883, JPTS 1883, xii.

35 e.g. Vin II 103, 21.
36 Vin II 106, 4. See von Hinüber, 1991B.
As the shapes of the developed, it sometimes happened that pairs of

, which at an earlier date were capable of confusion because of their similarity, became
dissimilar in shape, while pairs of which had originally been quite different,
became more similar. Despite their common origin, the Sinhalese, Burmese, Thai and
Cambodian scripts sometimes show widely divergent patterns of development in the
shape of , so that two characters which are very similar in one script are perhaps
quite different in another. This enables us to gain certain information from orthographical
errors. If, for example, we find an error in script A which seems to have arisen from the
confusion of two which only resemble each other in script B, then this strongly
suggests that the tradition was dependent, at some stage of its transmission, upon a
manuscript written in script B.
The change from oral to written transmission led to certain changes being made to
features which were oral in origin, e.g. those which are dependent upon a variation, or
ambiguity, in pronunciation. The introduction of writing meant that it was no longer
possible to be ambiguous. In the Suttanipāta we find, in two successive verses, puns
dependent upon an non-aspirated and an aspirated form: pakkodano “I have boiled my
rice” says Dhaniya the herdsman, and akkodhano “I am free from anger”, says the
Buddha punningly. Writing this pun causes no difficulty, but if the pun depended upon
two pronunciations of the same word, then there are problems, because writing forced
copyists to choose one spelling or the other, in the same way that Sanskritising copyists,
as we shall see in the sixth lecture, had to choose between two Sanskrit forms when
translating puns which depended upon Middle Indo-Aryan homonyms. So in a verse in
the Dhammapada37 we find “For a bad deed done is not released immediately, like freshly
extracted milk”, where “is released” is muccati. Although we might think that the verse
has something to do with release from bad kamma, and milk being released from the cow,
this really makes little sense.
The commentary, however, explains the verb as meaning “develops” (parinamati),
and we can see that it must originally have been mucchati38 “it coagulates”, i.e. a bad
deed does not develop (its result) immediately, just as new milk does not coagulate. In
the parallel versions, the Patna Dharmapada39 reads mucchati and the Udāna-varga40
reads mūrchati, so it is clear that the idea was understood in those traditions. It would
appear that the Pāli tradition thought that the context of the first part of the line, with the
mention of kamma, needed the verb muccati, while the second part needed mucchati, i.e.
there was a pun dependent upon a non-aspirated and an aspirated pronunciation. The
copyist could, however, only write one of the two and he chose, wrongly, to adopt
muccati, and this incorrect reading is found, as far as I know, in all manuscripts of both
the Dhammapada and its commentary.
As is well known, the oral transmission of the Vedas continued for many centuries
after the time of Aśoka. As far as we know, although writing was available it was not
adopted for Vedic purposes. Similarly, although the Theravādin tradition states that the
37 Dhp 71.
38 See Morris, 1884, 92.
39 PDhp 107.
40 Udāna-v 9.17.
Pāli canon was written down in the first century B.C.E., oral recitation of Pāli texts
continued and still continues to this day. We cannot tell how long the official
sort of oral tradition and the written tradition continued side by side, but it was perhaps
not very long. The writing errors, which I mentioned earlier, and which seem to indicate
that the oral tradition had been lost, must have occurred at an early date, because they
could only happen as long as double consonants (and long vowels) were not written and
we know that the facility for doing this was soon developed.
Writing down also had an effect upon the contents of the Theravādin canon. There is
some doubt about the state of the canon when it was written down. We do not, for
example, know whether it was complete, i.e. whether it was in the form in which we find
it today, or not, and we may well wonder whether any texts contain anything which
would enable us to identify it as material added after the canon had been committed to
writing.
As far as I can judge, once the Theravādin canon had been written down, very little
further change was made to it. From the point of view of its language, we should have
expected anything added in Sri Lanka to show traces of the local Prakrit, but there are
few signs of borrowings from Sinhalese Prakrit being inserted into it, and most of the
borrowings which have been suspected can be explained otherwise.41 The process of
Sanskritisation remained incomplete, which suggests that nothing further was done after
the canon had been written down. Nevertheless, the explanations given by the
commentaries in later centuries show that the canon was not fixed absolutely by the
process of writing it down. The system of reciting and approving the form of suttas at
should have resulted in the elimination of all variations, but it is clear that the
writing down of the canon did not, in itself, lead to the disappearance of readings which
had not been preferred at the various , and this would support the view that the
system of continued, in some form or other, for some time, with some reciters
still keeping alive readings not found in the written texts. It seems that such
“unauthorised” readings, if still recited, could easily creep back into a text, if a scribe42
heard them and reintroduced them while he was copying the text. This would account for
the number of variant readings which we find mentioned in the commentaries.
Despite the separation of the Sinhalese Buddhists from North India,43 it seems that
literary material continued to reach Sri Lanka, but there is no evidence for the addition of
any complete text to the Theravādin canon after the Ālokavihāra . An origin in
North India is postulated for the Milindapañha,44 the and the
. The fact that these texts are highly regarded by the commentators, but
are not given canonical status,45 suggests that they arrived in Sri Lanka after the closure
41 See Norman, 1978, 28–47 (32) (=CP II, 30–51 [34]).

42 See Norman, 1984–85, 1–14 (13) (=CP III, 126–36 [135]).
43 The break between North India and Ceylon was clearly not an abrupt one, or even a complete
one, for Theravādins continued to reside near Bodhgayā for some centuries. The fame of the
Sinhalese commentaries was sufficiently widespread in North India to attract Buddhaghosa to
Ceylon.
44 See Norman, 1983C, 108, 110, 111.
45 For the question of the canonicity of these texts in Burma, see the discussion referred to in the
eighth lecture.
of the canon, which presumably occurred at the time when it was committed to writing.
This view is supported by the fact that these post-canonical works contain a number of
verses and other utterances ascribed to the Buddha and various eminent theras, which are
not found in the canon. Nevertheless, it seems that there was no attempt made to add such
verses to the canon, even though it would have been a simple matter to insert them into
the Dhammapada or the Theragāthā.
Although there are references to writing in the canon, there are only two which refer to
religious texts being written. They are in the Vinaya, and both of them are in the Parivāra,
the appendix to the Vinaya. At the end of the first eight (of 16) sections of the first
chapter, there is a statement that “these eight sections are written in a manner for
recitation”,46 while at the end of the book, after the words “the Parivāra is finished”,47
there is a statement that Dīpanāma, having asked various questions about the ways of
former teachers, thinking out this epitome of the details of the middle way of study, had it
written down for the bringing of happiness to disciples. It seems, however, unlikely that
if the whole of the Parivāra was composed after the time the canon was written down, i.e.
if it was a late addition to the canon, it would have had those two references to writing
inserted in the way they are. I can see no good reason for doubting that the two
statements are interpolations, added when the Parivāra, together with the rest of the
canon, was written down.
One result of the writing down of the canon was that the stranglehold of the
was broken, and the control which they had had over the contents of the
nikāyas was relaxed. At the beginning of the oral tradition they must have observed the
Buddha’s instruction about the four mahāpadesas,48 the four references to authority: the
Buddha, a community with elders, a group of elders, and a single elder, and checked
whether any teaching said to have been obtained from these sources was consistent with
the sutta and vinaya which the already had. As time passed, however, and less
and less hitherto unknown teaching was brought forward for acceptance, we can assume
that the canon, or at least the various nikāyas which were in the hands of the ,
was closed and no further additions could be made, except, perhaps, in such small ways
as adding extra apadānas to the Apadāna or perhaps extra verses to the Thera- and Therī-
gāthā. There was no way of inserting an entirely new sutta into a nikāya unless the
of that nikāya could be persuaded to accept it.
In the third lecture I spoke of the bhikkhu , who, you will remember, totally
misunderstood the purport of the Buddha’s teachings about stumbling blocks.49 As long
as the recited the sutta which made the position clear, then any attempt which
(or others who thought like him) might have made to pretend that his view was
authorised by the Buddha, by inventing a sutta which authenticated it, would be
unsuccessful, because the would not have admitted a new sutta which
included a view condemned by the Buddha and which was, therefore, not consistent with
the rest of the canon.
46 Vin V 48, 29.
47 Parivāro ,Vin V 226, 6–7.
48 D II 123–26.
49 Alagaddūpamasutta, M I 130–42.
We might also think of the bhikkhu Sāti, who, as we read in the Majjhima-nikāya,50 so
misunderstood the Buddha’s teaching that he thought it was “consciousness” ( )
which continued in .51 This would appear to be a recollection by Sāti of a
teaching similar to that found in the - that vijñāna
continues:52 “This great being, endless, unlimited, consisting of nothing but intelligence”.
This view was refuted by the Buddha, who pointed out that he had frequently taught that
“Without a cause there is no origination of consciousness”.53 Once again, Sāti had no
chance of inserting his view into a nikāya.
Nor would it be possible for such views to be inserted into the nikāyas, even after the
introduction of the use of writing, as long as the tradition continued alongside
writing, as I have suggested it did, for some time, in the Theravādin tradition.
On the other hand, once the various schools of Buddhism had started to make use of
writing, it was not difficult to produce a text and say that it was Buddhavacana, as long
as there was no tradition to refute the claim. You will know that the Buddha
stated, as reported in Theravādin texts, that there was no “teacher’s fist”,54 as far as he
was concerned, i.e. he was not keeping anything back for an élite, but was making his
teaching known to all of his followers who wished to listen. There developed, however, a
view in some Buddhist schools that what the Buddha had said openly was intended only
for the masses. There was really another, hidden, meaning which the Buddha imparted
only to a chosen few. For such schools, the adoption of writing enabled them to claim
that their views were indeed Buddhavacana.
Richard Gombrich has dealt with this in the context of the rise of the Mahāyāna,55 but
it is possible that not just Mahāyāna, but also dissident Hīnayāna sects benefitted from
the use of writing. Since no one ever accuses the Abhayagirivāsins, the opponents of the
Mahāvihāravāsins, of transmitting their scriptures in some language other than Pāli, it is
probable that their canon was in Pāli. Nevertheless, their version of the ,
as far as can be judged from the small portion which is preserved in Tibetan, does not
agree with the Theravādin , and must therefore have been added to their
body of scriptures at a time when the system had been by-passed. It is quite
possible that they were able to add to their scriptures in this way, by making use of
writing. The way in which the so-called quasi-canonical texts came into existence in the
Middle Ages in Sri Lanka and South-east Asia is another indication of the way in which
the fact that texts were written down enabled them to be accepted by some Buddhists, at
least, in a way which would have been impossible if the were still transmitting
all texts orally.
Paradoxically, then, if the writing down of the Theravādin canon may be presumed to
50 M I 256–71.
51 tad ev’ sandhāvati , , M I 256, 19–20.
52 idam mahad bhūtam anantam vijñāna-ghana eva, BUp 11.4.12.
53 aññatra paccayā n’ atthi sambhavo, M I 258, 20. 54 na tatth’ Ānanda
Tathāgatassa dhammesu acariya- , D II 100, 4=S V 153, 19.
55 Gombrich, 1988, 29–46.
have stopped the further Sanskritisation of Pāli and prevented the insertion of new suttas
into the nikāyas, it was writing which made possible the production and acceptance of
Mahāyāna and other texts. We may suppose that, to a very large extent, the advent of
writing meant that an already existent canon was fixed when it was written down, but
writing allowed new canons to come into effect because the authors did not have to point
to a long tradition of the texts, which alone, before the use of writing, could
prove that they were Buddhavacana.
In view of this connection which has been seen between writing and the Mahāyāna, it
is not unreasonable to believe that the writing down of the Theravādin canon was not due
simply to a threatened breakdown in the system of transmission in Sri Lanka,
and the social, political and economic conditions of the time, as the Pāli commentarial
tradition suggests, and as I proposed earlier, but also to a need to give an authenticity and
prestige to the Theravādin canon vis-à-vis the written texts of other schools. If the
beginnings of the Mahāyāna, and therefore religious writing, can be dated to the second
century B.C.E., then it is likely that Hīnayāna texts were also being committed to writing,
in North India if not in Sri Lanka, at that time.
Since the social and religious conditions which had existed at the time of the Buddha
had by then changed greatly, and since Pāli and the literary forms of other Middle Indo-
Aryan dialects were now almost as much out of touch with the languages of the common
people as Sanskrit had seemed 400 years before, it is probable that many such texts were
being written in Sanskrit—the language of culture—as opposed to being translated into it.
VI
Buddhism and Sanskritisation
In the fifth lecture I referred to the fact that the writing down of the Theravādin canon in
the first century B.C.E. seems to have put an end to the Sanskritisation of that canon
which had begun in the decades before the reign of Abhaya, although,
paradoxically, writing seems to have led to the appearance of other texts, which, as far as
we can tell now, were actually composed in Sanskrit.
In this lecture I want to consider the reasons for the process of Sanskritisation, the way
in which it was effected and the result which it had upon Buddhist texts and Buddhism
itself.
I must start with a definition. What exactly do I mean by Sanskritisation? I use the
word in two senses. In the first place, I talk, in a broad sense, about the Sanskritisation of
Buddhism, when I am discussing a particular phenomenon, namely the way in which
Buddhism, which had started as a revolt against the social and religious system which
was exemplified by the use of Sanskrit for literary and religious purposes, now began
itself to embrace Sanskrit as a medium for the propagation of the Buddhavacana.
In the second place, Sanskritisation means the use of Sanskrit in Buddhist texts as a
replacement for the dialects of Middle Indo-Aryan in which the Buddha’s teachings had
previously been transmitted for some hundreds of years. In this sense, the term is
applicable to the whole range of Buddhist texts starting from those in a Prakrit which
contains a very small amount of Sanskrit, or Sanskrit-like, forms in it, through a range of
texts which are in a variety of languages which might be regarded as Sanskritised Prakrit
or Prakritised Sanskrit, sometimes called Buddhist Hybrid Sanskrit, to texts which are in
pure classical Sanskrit, in accordance with grammar.1 The language of the last
group is classified by some as Buddhist Sanskrit, because the texts are written by
Buddhists about some aspect of Buddhism or Buddhist history, and perhaps contain items
of vocabulary which are specifically Buddhist. We can therefore classify Sanskritised
texts under three headings: (1) texts written in a Middle Indo-Aryan dialect into which
some Sanskritisms have been inserted; (2) texts originally written in a Middle Indo-Aryan
dialect which have been translated into Buddhist Hybrid Sanskrit or Buddhist Sanskrit;
and finally (3) texts composed in Buddhist Hybrid Sanskrit or Buddhist Sanskrit
1 For a survey see von Hinüber, 1989B, 341–67.
.
Buddhism and Sanskritisation 81
In this lecture I want to concentrate upon those aspects of Sanskritisation which are
exemplified in the first two of these classes. I shall be considering for the most part the
problems which arose when texts in Middle Indo-Aryan dialects were converted into
other dialects with a Sanskritic content, small or large, or into Sanskrit. I shall not be
referring to those texts which were composed as original works in Buddhist Hybrid
Sanskrit or Buddhist Sanskrit, except in so far as some of the vocabulary they employ
perhaps owes something to Middle Indo-Aryan.
This pattern of Sanskritisation was not restricted to Buddhist texts. We find something
very comparable in the language of the inscriptions found in North India. The earliest
inscriptions, those of Aśoka, are written in a variety of Prakrit dialects. Those of the early
centuries of the Common Era are in a mixture of Prakrit and Sanskrit. From the fifth
century C.E. onwards inscriptions are written in classical Sanskrit. If we examine the
inscriptions in the middle phase at Mathurā, we find that they do not show a steady
progression from Prakrit to Sanskrit. What we know of the languages of North India,
during the period covered by the inscriptions, indicates that the local population spoke
some dialect of Middle Indo-Aryan. Donors would presumably dictate in Prakrit what
they wanted to have carved on their donations, and as it became fashionable to write
inscriptions in Sanskrit the scribes would “translate”, to the best of their ability, into that
language. An inscription in bad Sanskrit is, therefore, not necessarily older than one in
good Sanskrit. The difference in quality may simply represent the ability of one donor to
employ a better educated scribe than the other. Similarly, the fact that one Buddhist text
is in better Sanskrit than another does not necessarily mean that it was Sanskritised at a
later date.
Why did Sanskritisation begin? If we consider the partial Sanskritisation of the
language of the Theravādin canon, we must, I think, agree that there is no obvious reason
why Sanskritisation should have started in Sri Lanka in the first century B.C.E. We have
no information to make us believe that there was a strong pro-Sanskrit movement in the
island then. There is, for example, no hint that the Abhayagirivihāra, the rival of the
Mahāvihāra, was making use of Sanskrit at this time. The Sanskritisation of Pāli can
hardly have started spontaneously, in the absence of any reason, and we must assume that
it started under the stimulus of Sanskrit elsewhere, presumably in North India. There was
no obvious reason why a language used in Sri Lanka should have been influenced by
anything happening in North India, and so we must assume either that the process of
Sanskritisation started in the Theravādin canon before it was transported to Sri Lanka, or,
more probably, that there was still a strong connection between North India and Sri
Lanka after the introduction of Buddhism to the island.
The Pāli chronicles report that Buddhism was taken to Sri Lanka during the reign of
Aśoka. We know that, for secular purposes at least, Prakrit was preferred to Sanskrit at
the time of Aśoka, and it was only after his death that Sanskrit began to regain its position
of predominance, although we should perhaps note the Sanskritising tendency of the
Aśokan scribe at Girnār, even if we are unable to decide whether the Sanskritisms there
are features of the local language (which might have been rather archaic) or insertions by
a Sanskritising scribe. In favour of the latter, we should note that there is a strong case for
believing that the -r- forms at Girnār owe more to the scribe’s views of what was
appropriate than to the actual vernacular spoken in the area.2 If we believe that the second
century B.C.E. saw the appearance of new Buddhist texts, setting out new teachings,
composed by new sects, in Sanskrit, the language of culture and literature, then we might
well believe that that was the time when, in competition with these new Sanskrit works,
those schools of Buddhism which had hitherto used a Middle Indo-Aryan dialect, began
to make changes in the languages of their teachings in order to rival the growing use of
Sanskrit by other sects, and to make their teachings available to the same classes of
readers.
There were various degrees of Sanskritisation: (1) in its simplest form it involved the
partial restoration of Sanskrit phonology, perhaps the restoration of a number of Sanskrit
forms to consonant groups, and possibly in addition the restoration of long vowels which
had been shortened in Middle Indo-Aryan, e.g. before consonant groups; together with
(2) the restoration of Sanskrit morphology, e.g. “correct” verbal and nominal endings;
and perhaps (3) the substitution of Sanskrit vocabulary in place of Middle Indo-Aryan
vocabulary. This might include the removal of Eastern forms which, as I pointed out in
the fifth lecture, represented some of the oldest elements of Buddhist vocabulary, e.g.
avuso.
Further levels of Sanskritisation would include: (4) the insertion of the correct Sanskrit
sandhi forms, i.e. those demanded by the rules governing word juncture. We find that
hiatus is avoided by the insertion of particles, or by the rearrangement of the order of
words, or by a change of vocabulary, (5) and, in metrical texts, there is the avoidance of
irregular metre by changes of word order, or of vocabulary.
When we come to examine the Sanskritisations in Pāli, we have first of all to decide
whether they are in fact Sanskritic features which were inserted by the redactors, or
whether they might not be original features of the language, i.e. are the Sanskritisms in
Pāli retentions or restorations of Sanskritic features ? If we thought that such
Sanskritisms were remnants of an earlier form of the language of the canon, then we
could regard them as genuine archaic features in the language, and we might then define
Pāli as one of the oldest (linguistically speaking) of the Middle Indo-Aryan dialects, in as
much as it was a form of Sanskrit with some Middle Indo-Aryan developments in it.
An investigation, however, shows that some of these Sanskritisms in the Theravādin
canon are incorrect back-formations, e.g. atraja, which is probably a misinterpretation of
attaja (<ātmaja).3 Although some of the forms with the consonant group br- are correct,
2 von Hinüber (Überbl § 15) writes of an archaising dialect I would rather think of an archaising
scribe. The apparent resemblances to the Girnār (=G) version of the Aśokan inscriptions are not
conclusive, since we face there the same problems as in Pāli. The G version represents a
“translation” of an Eastern version which we may assume was sent from Aśoka’s capital
. We have no certain way of telling whether the G version accurately represents the
vernacular spoke in Western India at that time, or whether it merely represents the scribe’s attempt
to produce what he thought was appropriate in the circumstances. It is arguable that the scribe at G
was trying to Sanskritise, to the best of his ability, what he had received.
3 See PED, s.v. attaja.

e.g. , others of them are non-historic, e.g. brūheti.4 Even some of the forms
which are correct, e.g. the absolutive ending -tvā and br- in , can be shown to
be unoriginal in certain contexts. There is evidence for the absolutive ending -ttā,5
which we can assume was the regular absolutive ending in the dialect before the
restoration of the -tvā ending, and both the metre and the etymologies which are given for
in the canonical texts6 show that at an earlier time the word occurred with
initial b-, not br-. We can therefore conclude that these forms, and probably all other
Sanskritic features, are deliberate attempts at Sanskritisation, made at some time during
the course of the transmission of the canon. It is therefore clear that it is not correct to
speak of them as retentions. They are features which have been restored to the texts by
scribes or reciters who were trying to change into Sanskrit the language which they had
received in their exemplars.
A close examination of the Sanskritic features in the Theravādin canon suggests that
the process took place in two phases, separated by some centuries.
To the early phase we can allot the Sanskritisation of some consonant groups, e.g. tv in
the absolutive ending -tvā, which I have just mentioned, and tr in the suffix -tra of the
locative pronouns, and the st in utrasta and bhasta, etc.7 Some of these changes were
probably made as early as the writing down of the canon, and they are certainly earlier
than the commentaries, because the commentators refer to the variation between the
suffixes -ttā and -tvā. It is noteworthy that r is not restored in groups with p. The fact that
there are no forms with pr in Pāli, in contrast to the Girnār version of the Aśokan
inscriptions, where pr- is the most common group containing -r-, suggests that, despite all
claims to the contrary, there is probably no connection between Pāli and the Girnār
dialect of the Aśokan inscriptions.
Other Sanskritic features include the retention or restoration of long -ā-, e.g. in certain
compounds and derivatives of vāk “speech”, even when the resultant form of the word
goes against the pattern of the dialect by producing a long vowel before a consonant
group, e.g. vākya, which by the rules might have been expected to develop into *vakka,
with a short -a-.8 The fact that vākya and comparable forms are back-formations is
4 Skt should have developed >* >*būhayati/būheti (with the - -/-u-

alternation), as in AMg.
5 See von Hinüber, 1982, 133–40 for absolutives in -ttā, and Norman, 1980B, 183, note 21, and
1985, 32–35, for absolutives in -tā. I assume that the latter are m.c., but von Hinüber (Überbl §
498) lists an absolutive in -tā.
6 e.g. bāhita-papo ti , Dhp 388.
7 See Norman, 1989C, 369–92 (377–79) (=CP IV, 46–71 [56–58]).
8 There are also examples of the retention or restoration of a long vowel before a doubled
consonant, which goes against the general rule of two morae. Probably a word such as dātta is a
borrowing from a North-Western dialect (see Turner, 1973, 424–28 [= Collected Papers, 430–35]);
Geiger, 1994, § 7 (despite Geiger, dātta is quotable from Mil 33, 3 foll.). We also find a long -ā-
occurring before a doubled consonant as a result of the crasis of two vowels in a compound, and
also in such words as yvāssa (=yo assa, M I 137, 11) and tyāssa (=te assa, Dhp-a I 116, 20) as
products of a non-historical sandhi process, as well as gavāssa ca (=gavā assā ca, Ja III 408, 21).
All these examples seem to be based upon a knowledge of Sanskrit, since they go against the
general rules of MIA dialects.
supported by the existence of other derivatives of vāk where the expected shortening of
long -ā- does occur before a double consonant. We can deduce that the redactor did not
recognise the existence of the word vāk in such compounds, and consequently did not
restore the long vowel.
A later phase of Sanskritisation took place under the influence of the grammarians,
centuries after the first phase, and probably for the most part after the appearance of the
Saddanīti, ’s grammar of Pāli which was written in the twelfth century.
Some of these Sanskritisations are in the field of vocabulary or morphology, and many
are concerned with sandhi. Sandhi in Pāli is quite different from Sanskrit sandhi, in that
itis much more flexible. Since final consonants have disappeared, sandhi in Pāli consists
of the contraction of the final vowel of a word, including a nasalised vowel, with the
initial vowel of the following word. Such contractions produce a wide range of crasis
vowels, which are not always predictable.
The grammarians, with their extensive knowledge of Sanskrit, tried to bring Pāli more
in line with Sanskrit grammar, and copyists inserted into the manuscripts the forms which
the grammarians prescribed. Final -i and -u which had remained in hiatus were
(sometimes) changed to -y and -v. If the final vowel had been -e or -o, then an indication
of this was sometimes given by prefixing y or v to the vowel which had evolved by
contraction. The fact that these changes are later additions to the texts, and are not the
result of the earlier phase of Sanskritisation, is shown by the fact that the manuscripts are
rarely consistent about such insertions. If the contraction of final and initial vowels led to
a long vowel occurring before a consonant group, this was sometimes kept, even though
it offended against the law in Middle Indo-Aryan which demanded a short vowel in this
position.
It is clear that the reasons for these two phases of Sanskritisation of Pāli were quite
different. It is probable that the reason for the first was the appearance of other schools
using Sanskrit, as I have suggested, but the reason for the second is that grammarians
who knew enough about Sanskrit to be able to model their grammars upon Sanskrit, and
in fact to quote extensively from the Sanskrit grammarians, thought that they should try
to make Pāli more like Sanskrit.
Sanskritisation caused problems of a different nature from those which arose in the
period of oral tradition. The nature of the development of Middle Indo-Aryan meant that
there was a increase in the number of homonyms. A glance at Sheth’s Prakrit-Hindi
Dictionary ( ) shows, for example, that there are at least ten
Sanskrit words which can develop into Prakrit saya-, and, if we look at compounds, there
are no less than 48 Sanskrit equivalents for the Prakrit compound para-(v)vāya. Although
a specific context would immediately rule out many of these, it is obvious that a redactor
translating into Sanskrit might well have difficulties when trying to decide between
homonyms.
Problems arose even in the limited amount of Sanskritisation which we find in Pāli.
The problem for the redactor was to decide which of two or more possible forms to
choose when Sanskritising. There is sufficient evidence for us to deduce that in the
language of the Theravādin canon before it was Sanskritised the absolutive ending was -
ttā, as I have mentioned. This would be identical with the form in Ardha-Māgadhī, the
language of the Jain canon. A form such as kattā<Sanskrit “having done” was
therefore identical in form with, i.e, was a homonym of, kattā, the agent noun “a doer”
from the same root -. There is a pāda which occurs twice in Pāli which in its earlier
form contained, we can deduce, the word kattā. In one context kattā was taken as a verb
and Sanskritised as katvā. In the other the tradition interpreted it as the noun kattā, and so
retained it in that form and explained it as a noun.9
Exactly the same happened with the word chettā, from the root chid- “to cut”. Here, in
one and the same context, the tradition is ambivalent. In the edition of the canonical text
we find the absolutive chetvā “having cut”, but the commentary reads the noun chettā “a
cutter”, and explains it accordingly.10
When a Sanskritising recensionist was faced with the word santa, he had the option of
assuming that it was the present participle of the verb “to be”, which gets the meaning of
“good”—you will know that the feminine of this (satī) is used of the good woman who is
so devoted to her husband that she immolates herself upon his funeral pyre—or the past
participle of the verb śam “to rest”, i.e. the equivalent of Sanskrit śānta “at rest,
peaceful”. In some cases it is possible, even probable, that in a Middle Indo-Aryan dialect
santa was intended to be ambiguous, but in Sanskrit one or other of the two meanings has
to be preferred.
The same problem can arise with the Middle Indo-Aryan word dīpa, which may be the
equivalent of Sanskrit dīpa “lamp” or dvīpa “island”. As a parallel to the verse in the
Dhammapada11 “a wise man should make a dīpa which the flood will not overwhelm”,
the Sanskrit Udānavarga12 has dvīpa “island, but the Chinese version13 has “lamp”,
showing that it is based upon a Sanskrit version which had dīpa “lamp”.
A verse in the Dhammapada14 tells us that someone who has pīti in the dhamma sleeps
happily. The word pīti is ambiguous since there are two words pīti in Pāli, one from
Sanskrit pīti “drinking” from the root pā “to drink”, and the other from prīti “joy” from
the root prī “to please”. It seems very likely that both meanings are intended in this verse.
The second pāda tells us that he sleeps with a clear mind, perhaps unfuddled by drinking
the dhamma as opposed to the intoxication he would have experienced if he had drunk
strong drink. In the last pāda of the verse, however, we read that the wise man delights
(ramati) in the dhamma,15 which suggests that pīti is also to be taken as “joy”. The Pāli
commentator presumably did not see the possibility of the word play, and explains it only
as drinking the dhamma.16 The redactor of the Sanskrit Udānavarga had to choose
between pīti and prīti, and perhaps because of the idea of “delight” in the last pāda, or
perhaps because he was following a different commentarial tradition, he decided to read
prīti.17
9 katvā ti…karitvā, Ja II 317, 21′ (ad katvā, 317, 14*); kattā kārako, Ja IV 274, 9′ (ad
kattā, 274, 2*).
10 chetvā, Th 1263; chettā chedako, Th-a III 199, 11.
11 kayirātha meddhāvī| yam ogho nābhikīrati, Dhp 25.
12 karoti medhāvī, Udāna-v 4.5.
13 Quoted by Brough, 1962, 209 (ad GDhp 111).
14 Dhp 79.
15 dhamme sadā ramati , Dhp 205.
16 dhammapītī ti dhammapāyako pivanto ti attho, Dhp-a II 126, 15.
17 The Gāndhārī Dharmapada is written in a dialect which retains many of the consonant groups of
Skt, and there we find dhama-pridi, GDhp 224. See Brough, 1962, 244.
There was some doubt about the Sanskritisation of the Pāli word nekkhamma. As the
Pali-English Dictionary points out, although the word is derivable from a hypothetical
Sanskrit form “going out”, from the root kram- “to go”, i.e. departing
from the householder’s state to the houseless way of life, there are also word plays which
seem to be based upon the meaning “the state of being without desire” ( )
from the root kam- “to desire”. I see no reason to doubt that we are in fact dealing with
two homonyms here. In Buddhist Sanskrit, however, we find only the form
,18 even in conjunction with the word for desire ( ).19
We sometimes find that Pāli and Sanskrit technical terms do not correspond, and in
such cases we may surmise that the lack of correspondence may be due to Sanskritisation
from a homonym. The Buddhist Hybrid Sanskrit equivalent of Pāli sammappadhāna
“right effort” is “right abandoning”. It has been suggested20 that the
Sanskrit version is an incorrect backformation from an Middle Indo-Aryan form
*samma-ppahāna which might stand for either -pradhāna or -prahāna. In view of the
fact that such word-plays are very common in Indian literature, we cannot exclude the
possibility that the authors of the passages which contain such words were making using
of the ambiguity, and, if this was so in this particular case, we can see that neither the Pāli
nor the Sanskrit form is capable of giving the double meaning which the author intended.
Sometimes commentators are aware of the fact that a word can have two meanings, and
they incorporate both senses in their explanations. It is suggested by some21 that the
Buddhist tradition deliberately capitalised on, or even exploited, any ambiguity which a
Middle Indo-Aryan form might have, but it seems to me that using such phrases as
“deliberately capitalised” is perhaps the wrong way of looking at the matter. In my view
it is rather taking advantage of the situation. The speakers of Middle Indo-Aryan dialects
were not likely to know that a given item in their vocabulary had developed from two or
more Sanskrit words. They merely knew that it had two (often quite different) meanings,
and the commentaries tried to make use of this fact when giving their explanations. It was
only when someone proficient in Sanskrit tried to Sanskritise that the need to choose
between two possible Sanskrit antecedents arose. We should not be too surprised if a
recensionist made the wrong decision, especially since,22 in the example I have just
mentioned, Sanskrit pradhāna does not have the meaning “endeavour”.
There are many other similar examples of wrong backformation in partially
Sanskritised dialects, such as the language of the Theravādin canon and the Patna
Dharmapada, and also in more highly Sanskritised languages. It should be noted that it is
not always the Sanskrit version which is incorrect, and we can surmise that there was
sometimes a commentarial tradition available to the translators, which gave information
about the meaning and interpretation of the words they were concerned with.
18 See Sasaki, 1986, 3.

19 Mvy 6444 “renunciation as regards desires”, quoted in BHSD.
20 See Gethin, 1992B, 70.
21 Gethin, 1992B, 72.
22 As Gethin notes (1992B, 71).
The need to choose between homonyms quite often leads to a situation where puns
and explanations which worked in Middle Indo-Aryan no longer worked in Sanskrit, or
even in a Middle Indo-Aryan dialect with partial Sanskritisation. In the Dhammapada the
word is explained by means of forms from the root bāh- “to remove”. A
brahman is one “who has removed, got rid of, his sins”.23 The connection between brāh-
and bāh- is not as close as one would like for an etymology, but the answer to the
problem is to deduce that this explanation was originally formulated in a Middle Indo-
Aryan dialect where the development from was , so that the pun on
the meanings of the two verbs, which in their Sanskrit forms are - “to be strong” and
- “to remove”, worked perfectly. The Sanskritisation to which occurred
in Pāli, as I have already noted, had already spoilt the pun. Another pun found in the
Dhammapada which does still work in Pāli, is that found in the etymology which is given
for the word “ascetic”. One is called a “because one’s sins have been
put to rest”,24 where the sam- of is punningly linked with the root sam- “to be at
rest”. In a Sanskritised form, however, becomes , which is from the
root śram- “to make an effort”, while the Sanskrit form of the root “to rest” is śam-. The
pun which worked well in Pāli is lost when we have explained by śam-.25
The concept of the pratyeka-buddha is well known as the middle element in the triad:
buddhas, pratyeka-buddhas and śrāvakas. The word is usually translated as “a buddha
(awakened) for himself’, but this usage of pratyeka (pacceka in its Pāli form) is unusual,
and it is not at all clear how it could acquire the meaning which is given for it. The
concept of a similar type of Buddha is also found in Jainism, but there, in Jain Prakrit, the
name is patteya-buddha. The variation in the name suggests that the concept of pratyeka-
buddhas was borrowed into both Buddhism and Jainism, and we can therefore look for an
etymology for the word outside Pāli. It seems most likely that the earlier form of the
word was pacceya-buddha, where pacceya is to be derived from Sanskrit pratyaya
“cause”. A pacceya-buddha was “one who is awakened by a specific cause, a specific
occurrence (not by a Buddha’s teaching)”, and both the Buddhist and the Jain tradition
give lists of the occurrences which caused the awakening of the four most famous
persons in this class. The Pāli form pacceka is an incorrect translation, probably a hyper-
form, of pacceya, and pratyeka is a Sanskritisation of pacceka.
Sanskritisation led to hyper-forms, as a result of misinterpretations. Many Buddhist
Sanskrit texts are entitled sūtra. To anyone who comes to Buddhist studies from classical
Sanskrit studies, this name comes as a surprise, because, in Sanskrit, sūtra literature is a
specific genre of literature, composed in prose, usually of a very abbreviated and concise
nature, while Buddhist sūtras have an entirely different character. This difference is due
to the fact that the word sūtra in Buddhist Sanskrit is a Sanskritisation of the Middle
23 bāhitapāpo tii , Dhp 388.

24 samitattā hi tti pavuccati, Dhp 265.
25 śamitatvāt tiu hi nirucyate, Udāna-v 11.14.
Indo-Aryan word sutta, which is probably to be derived from Sanskrit sūkta, a compound
of su and ukta, literally “well-spoken”.26 It would be a synonym for , which is
the word used of the Buddhavacana by the emperor Aśoka, as we shall see in the seventh
lecture, when he said: “All that was spoken by the Lord Buddha was well-spoken”.
Another wrong back-formation centres around the same word sutta. There is a verse in
the Pāli Dhammapada where we are exhorted to chinda “cut off the stream” (of
), and the Sanskritisation in the Udāna-varga gives the same sense: chindhi
. The Patna Dharmapada, however, was doubtless dependent upon a Middle Indo-
Aryan version which, through an orthographic variation, must have had the reading sutta,
via sŏtta. This was then Sanskritised as sūtra, so the pāda in that version tells us to “cut
off the thread”, presumably interpreted as the thread of rebirth, since we find in Pāli that
craving ( ) is described as “the seamstress” (sibbanī), which joins us to
by means of death and rebirth.27
One of the best-known examples of Sanskritisation is the word bodhisattva, which is a
back-formation from Middle Indo-Aryan bodhisatta. Anyone who knows anything about
Sanskrit will realise that the translation which is commonly given for this word “one
destined to be a Buddha”, or “one destined for awakening”, is, to say the least, unlikely,
and we should note that Monier-Williams gives the basic meaning as “one whose essence
is perfect knowledge”,28 which would be more appropriate as an epithet for a Buddha
than for a bodhisatta. It is noteworthy that the Pāli commentaries did not assume that the
second element of the compound was the equivalent of Sanskrit sattva. They give
derivations either from the root sañj- “to be attached to”, which should have given a
Sanskrit form “one attached to bodhi”, or from the root śak “to be able”
which should have given a Sanskrit form bodhiśakta “one capable of bodhi”.29
We should note, incidentally, that wrong Sanskritisations are not restricted to Buddhist
texts. One of the Sanskrit words for “pearl” is muktā, which literally means “released”. It
is, however, more than likely that the Sanskrit form is a wrong Sanskritisation from a
Middle Indo-Aryan form muttā, which occurs, for example, in Pāli, so that its
Sanskritisation is an example of a folk etymology: a pearl is called “released” because it
is released from the oyster. It has been suggested that Middle Indo-Aryan muttā is to be
derived from Sanskrit mūrta30 “coagulated, shaped, formed”, but in view of the wide-
spread existence of a word muttu in Dravidian,31 with the meaning “pearl”, it is perhaps
more likely that Middle Indo-Aryan muttā is a loan-word from Dravidian.
Resolved consonant groups, those consonant groups which have been separated into
their constituent members by inserting an epenthetic (svarabhakti) vowel between them,
were Sanskritised by the removal of the svarabhakti vowel, e.g. Sanskrit kriyā “action”
26 See Walleser, 1914, 4, note 1; von Hinüber (1994, 132, note 28) follows Mayrhofer (1976, 492,
s.v. ) in thinking this proposed etymology is unnecessary.
27 See EV I 663.
28 MW, s.v. bodhisattva.
29 See Bollée, 1974, 27–39 (36, note 27).
30 See von Hinüber, 1993, 113, referring to Lüders, 1940, 179–90.
31 See DEDR 4959.
became Middle Indo-Aryan kiriyā, and was then Sanskritised back to kriyā again.
Sometimes misunderstanding on the part of the translators led to forms such as
being Sanskritised32 from Middle Indo-Aryan parisā, which was derived from Sanskrit
, on the assumption that -i- was an epenthetic (svarabhakti) vowel, and similarly
nyāma “rule”, or “regulation” was, in a similar way, wrongly back-formed from Middle
Indo-Aryan niyāma, which was identical with Sanskrit niyāma. This type of error can be
seen in the oldest translations which we have in India, i.e. the Aśokan inscriptions, where
one of the scribes wrote hveyu in place of the form huveyu, from the root bhū-, which he
had received in his exemplar, presumably in the belief that -u- was a svarabhakti vowel.
Although I said earlier that the fact that one Buddhist text is in better Sanskrit than
another does not necessarily mean that it was Sanskritised at a later date, nevertheless, in
some cases we can see how Sanskritisation developed. We have, for example, an edition
of an earlier version of the Udānavarga from Chinese Turkestan,33 which is closer in its
form, especially with regard to word order, to its Middle Indo-Aryan original than the
later version from Chinese Turkestan,34 which, in its attention to metre, its observance of
sandhi rules, etc., represents a more correctly Sanskritised version of the text.35
It is sometimes thought that because a more Sanskritised version of a text is in better,
i.e. more correct, Sanskrit, it must also give a better, i.e. more correct, interpretation of
the underlying Middle Indo-Aryan text, and we should therefore interpret a Middle Indo-
Aryan text in the light of a later Sanskritised version. This, however, is not necessarily a
good principle to follow. Since all the versions we have of Hīnayāna texts are translations
of some earlier version, the correctness or otherwise of a Sanskrit version depends upon
the aids which the Sanskrit translator had available to him when he was making his
translation. Where there was doubt about the way to interpret, and hence to translate, a
word or passage, a Sanskrit translator who had better guides to interpretation than
someone translating into a Middle Indo-Aryan dialect, was likely to produce a Sanskrit
version which was more correct than the Middle Indo-Aryan version. If the Middle Indo-
Aryan redactor had better guides, then it is likely that the Middle Indo-Aryan version
would be more correct. If neither of them had any help, then their translations would be
based upon their ability to guess the meaning, and neither translator was likely to be
consistently better than the other at guessing.
Reviewers and others sometimes point out that in my translations or studies of Middle
Indo-Aryan words and works I have not taken account of the Chinese or Tibetan or
Khotanese, or whatever other, version they think is important. This is perfectly true, for a
number of reasons: (1) I am deliberately confining myself to translating or dealing with
the Middle Indo-Aryan material; (2) I was probably ignorant of the Chinese, Tibetan or
Khotanese version’s existence; (3) I could not handle it even if I knew about it; (4) such
versions are translations, usually from Sanskrit, which is itself a translation, from some
Middle Indo-Aryan version. It may therefore be interesting, as an interpretation which
one tradition has put upon the material it has inherited, but it is in no way an authority, to
be followed slavishly. It very often happens that such versions are incorrect, because of
32 See Sander, 1985, 144–60.

33 Nakatani, 1987.
34 Bernhard, 1965.
35 See Bechert, 1991A and de Jong, 1974, 49–82 (53).
errors or misunderstandings in the tradition.

A study of the metre of the Buddhist Hybrid Sanskrit version of Upāli’s verses has
recently been made.36 A Pāli version of these verses occurs in the Majjhimanikāya, and
portions of a Buddhist Hybrid Sanskrit version of the verses were discovered in Chinese
Turkestan and published in 1916. Since then other fragments have been identified and
published. The verses are of great interest, for a number of reasons. They give a list of
100 epithets of the Buddha in a metre which occurs only very rarely in Pāli—the old āryā
or old gīti metre. An examination of the structure of the verses of the Buddhist Hybrid
Sanskrit Upālisūtra shows clearly that the transforming into Sanskrit was made from a
Middle Indo-Aryan exemplar in a way which was almost entirely mechanical. We can see
that of the features of Sanskritisation which I mentioned earlier, the redactor restored
consonant groups, replaced non-historic case endings with more correct endings, and
made some changes of vocabulary. He did all this without any regard whatsoever for the
metre, and although he made some changes to the sandhi, the frequent examples of hiatus
have been reproduced almost without exception. It is obvious that the verses of the
Buddhist Hybrid Sanskrit version were originally composed in a Middle Indo-Aryan
dialect, but the differences in word order, and indeed in the order of the verses, show that
it was a tradition which differed somewhat from the Pāli.
There is also a Chinese version of these verses, which seems not to be based on the
Pāli version or any of the Sanskrit versions which we possess. Some of the epithets in the
list differ from those we find in the other versions, although it is not clear whether the
differences are due to variations in the tradition or to the Chinese redactor’s inability to
understand his exemplar. It is very probable that the Chinese translation was made from a
Buddhist Hybrid Sanskrit version, which in turn was made from a Middle Indo-Aryan
version, probably in a dialect of the North-western Gāndhārī variety, which must in turn
have come from another Middle Indo-Aryan dialect, which may have been the dialect
used by the Buddha, but was probably not. And yet, despite the long history of
translations in the tradition, each of which would have given opportunity for errors to
arise, people are very surprised when I say that the Chinese, Tibetan or Khotanese, or
whatever version it was, might not have been all that useful to me, if I had been able to
read it. I must make it clear that I am talking about Hīnayāna texts whose origin may be
presumed to have been in India.
Sanskritisations from the Gāndhārī dialect are always likely to be of doubtful value,
because of the wide discrepancy between orthography and pronunciation in that dialect. It
is clear that sometimes, at least, the orthography was carried over into translations,
irrespective of whether it made sense or nonsense. There is a story found in the Chinese
version of the Vinaya of the Mūlasarvāstivādins, where Ānanda heard a monk reciting a
Dharmapada-verse which ended with the words, “It were better that a man live only for
one day, and see a water-heron”. Ānanda’s efforts to persuade the monk that the verse
should have ended with the words “and see the principle of coming into existence and
passing away”, were unsuccessful. This Chinese version was obviously following a
tradition based upon a Sanskrit form *udaka-baka, which could only come from a
Gāndhārī-type dialect37 which inserted a non-historic -k- in place of a glide -y-, in the
36 See Norman, 1993, 113–23, where the relevant information is given.
37 cf. udaka-vaya, GDhp 317.

compound udaya-vyaya38 “arising and passing away”.

A recent study of the language of the earliest Chinese version of the
39
suggests that a number of the idiosyncrasies of the language of
that text owe something to the fact that the text seems to have been based not upon the
Sanskrit versions which we know now, but upon some other version in a dialect which
was very similar to the Gāndhārī dialect. The frequent confusion between jñāna
“knowledge” and yāna “vehicle” found in the text seems to indicate that there must have
been a variant jāna in place of jñāna in the donor dialect, and a development of this with
-y- in intervocalic positions in compounds would have produced forms such as Buddha-
yāna in place of Buddha-jñāna.40 Similarly, the fact that bhoti, the contracted form of
bhavati “he becomes”, appears to have been Sanskritised as bodhi “awakening” implies a
translation from a dialect which, like Gāndhārī, articulated the aspirate so weakly that it
could be written or omitted at will.
Among the features which are very common in Middle Indo-Aryan metrical texts is
the way in which verses, although hypermetric, can be made to scan by assuming the
very common phenomenon of the resolution of syllables, under the musical influence
which allows, in certain positions in many metres, the replacement of a long syllable by
two short syllables. When pādas showing resolution are Sanskritised without any further
change, and Sanskrit forms, etc., are restored, the redactor cannot compensate for such
resolution, and a verse results which has to be categorised by a modern editor as hyper-
metric,41 or irregular.42 We can sometimes see that the first stage of Sanskritisation
produced verses which barely approximated to Sanskrit in form and metre, and a second
stage of Sanskritisation was required to produce something which was more correct, by
changes of vocabulary or word order.
Another common feature in Middle Indo-Aryan metrical texts is the way in which forms
are evolved metri causa, by the lengthening or, less commonly, shortening of syllables,
e.g. we find nirūpadhi instead of nirupadhi “without sub-strate”, or anūdaka as a
replacement for anudaka “without water” in the cadence of a śloka verse, where the
metre requires the short-long-short-long pattern of syllables. When such metrical
adjustments are Sanskritised, the earlier Sanskritised Prakrit version writes forms which
are still not correct Sanskrit, e.g. niropadhi and anodaka. In a more correct Sanskritised
version we find that a different way of dealing with the problem is evolved, often
involving the use of vowels, e.g. niraupadhi.43
What effect did all these various aspects of Sanskritisation have on Buddhist texts? It
is obvious that, by definition, the Sanskritisation of Buddhist texts made them more like
Sanskrit, which meant that ambiguous Middle Indo-Aryan phonology and morphology
was replaced by precise Sanskrit forms, ambiguities of sandhi and grammar were
removed, and the metre was repaired, where it was recognised. This also had a negative
38 Dhp 113.
39 Karashima, 1992.
40 Karashima, 1992, 266 and 1991, 607–43.
41 See Matsumura, 1989, 80 (ad [26]).
42 See Matsumura, 1989, 81 (ad [32]); 82 (ad [35]).
43 Udāna-v 6.10.
side. Where a Middle Indo-Aryan form was ambiguous, the Sanskrit redactor may have
chosen the wrong alternative. Where the ambiguity was intentional, perhaps to make a
deliberate play upon words, e.g. when giving an etymology based upon a homonym, the
Sanskrit redactor could only translate the homonyms into Sanskrit, where they were not
homonyms, thus losing the point of the pun. If the redactor did not understand his
exemplar, and was forced to translate mechanically or by guesswork, then the results
were unpredictable, but were almost certainly wrong.
In the third lecture I spoke of the Waxing Syllable Principle, the way in which words
in stock phrases are often arranged in order according to the number of syllables in each
word, and I considered the help which this principle might be considered to have given in
the memorisation and recitation of texts. The Waxing Syllable Principle was sometimes
violated in Sanskritisation, because Sanskritisation changed the syllable count by
removing svarabhakti vowels, or by changing the vocabulary, i.e. by replacing Middle
Indo-Aryan words by Sanskrit synonyms.44 The Waxing Syllable Principle was a feature
of oral literature, intended as an aid to memorisation and recitation. Since Sanskrit and
writing went hand in hand, the fact that Sanskritisation made memorisation more difficult
was perhaps of less importance than it might have been at an earlier time.
What effect did Sanskritisation have upon Buddhism? I mentioned in the fourth
lecture the suggestion made by some scholars that in the often-quoted passage from the
Vinaya, where bhikkhus ask for permission to translate the Buddha’s teachings, the word
chandaso means “into Sanskrit”. If that is correct, we must note that the Buddha forbade
translation in this way. Centuries later, however, that instruction (if it was indeed his
instruction) was ignored, and Buddhists works were indeed translated into Sanskrit.
Whether those who made the translation were aware that they were ignoring the
Buddha’s command, is an interesting question.
The Sanskritisation of Buddhist texts meant that their readership or intelligibility was
not limited to monks who were acquainted with the dialect in which the texts had hitherto
been transmitted, and it must, therefore, have made them more accessible to a wider
range of readers. If a monk of any individual school could read and understand the (more
or less) Sanskritised version of the teachings of his own school, he would be able to read
the teachings of other texts too, if in his travels he came to a vihāra belonging to another
school, or met a monk of another persuasion. The effect which such increased
communication between sects and schools had is a matter for speculation. It has been
shown, for example, that the Sarvāstivādins and Mūlasarvāstivādins possessed a number
of different versions of the same text, e.g. the Udānavarga, and also remained in close
contact, so that a Sarvāstivādin version of a text might be influenced by a
Mūlasarvāstivādin version of the same text.45 Such influence could more easily be
effected if both versions were in Sanskrit, than if they were in two different dialects of
Middle Indo-Aryan.
This takes us back to the reason for Buddhism making use of the vernacular languages
in the first place. This must have been two-fold: as a rebellion against the brahmanical
caste and their language, and to make the Buddha’s teaching available to the common
people. The brahmans were the communicating channel between men and gods. They
44 See von Hinüber, 1993, 101–13 (109).

45 See Schmithausen, 1987, 304–81.
alone knew the rituals and the sacrificial methods to intercede with the gods on behalf of
their patrons. The Buddha, by denying the power of the gods, removed the need for a
priestly caste to intercede, and taught a system whereby every man could gain his own
salvation, either immediately or in a future existence. His teaching was available in the
language of the people. As the teaching became codified and its language became
ossified, and as the vernacular languages continued to develop, a gap arose between the
language of the teachings and the language of the people. Already by the second century
B.C.E. we can see that there was a considerable difference between Pāli and Sinhalese
Prakrit, as it appears in inscriptions, and a century or two later the difference was so great
that, apart from Pāli loan words in Sinhalese Prakrit, one would be hard pressed to see
that the two Middle Indo-Aryan dialects were related. If, therefore, the Buddhavacana
was to be propagated in a form which the common people could understand, then
bhikkhus would have to translate the texts which they had memorised and recited in Pāli
into the vernacular languages of the people. Ultimately, as we know, the Pāli canon had
to be translated into Sinhalese (Prakrit). Something similar must have been the
experience of all sects and schools of Buddhism.
Although the language of the Gāndhārī Dharmapada was archaic, and approximated to
the spoken language of two centuries before, nevertheless we may assume that it was still
intelligible, for the most part, to the people living in the Gandhāra area and along the
southern Silk Road in Chinese Turkestan where it was found. Texts written in this
language, or something very similar, were, however, then translated into a Prakritised
Sanskrit and then into Buddhist Sanskrit by the monks living in the Turfan area on the
northern Silk Road. We may wonder whether anyone, other than those monks, could read
the scriptures in that form.
If the teachings were no longer intelligible to the masses, then it did not matter much
which language they were in, and it is clear that Sanskrit, the language of culture and
literature, was an obvious choice. It was the literary language of North India, at least, and
translation into that language meant that educated people—mainly monks, but probably
some laymen as well—could read their own scriptures and also those of any other school,
if they were so minded. The result was, however, not hard to predict. Buddhism started to
become an academic study, where only the educated, who had learned Sanskrit, had
access to the teachings in their written form. Bhikkhus, studying in their vihāras, became
more and more remote from the people. We know of only one Buddhist work devoted to
the life and duties of a layman, the . When Buddhism came under
attack from the iconoclastic Muslim invaders, and the vihāras and their libraries were
destroyed, Buddhism in the land of its founder virtually disappeared.
There is a very obvious contrast with Jainism. There the monks and the laymen were
more closely integrated. Texts dealing with the duties of laymen are numerous. Many
translations of canonical texts were made, but into vernacular languages, not Sanskrit,
and there were also commentaries written in vernacular languages. Although the Jains did
use a form of Sanskrit, which resembled the Sanskrit used by Buddhists in its dependence
upon Middle Indo-Aryan forms, Sanskrit never gained the importance for Jains which it
had for Buddhists. Although both Buddhism and Jainism had royal patrons, Jainism also
had close links with, and was less remote from, the people, and we find, for example, that
the Jains made great use of popular literature, with Jain versions of the Mahābhārata and
the . It has been argued that it was as a consequence of its support
among the common people that Jainism, although hit no less than Buddhism by the
invaders, survived in North India. Buddhism did not.
VII
Buddhism and Aśoka
While re-reading recently a book on Buddhism by an eminent Buddhist scholar, I noticed

the following statement: “This notion of establishing the sāsana or Buddhism in a
particular country or a place was perhaps first conceived by Aśoka himself. He was the
first king to adopt Buddhism as a state religion, and to start a great spiritual conquest
which was called dharma-vijaya…. Like a conqueror and ruler who would establish
governments in countries politically conquered by him, so Aśoka probably thought of
establishing the sāsana in countries spiritually conquered by him”.1
In other publications I have seen such claims made as: “Aśoka was the first Buddhist
Emperor”, “Aśoka was connected with the popularisation of Buddhism, and with the
enthusiastic promotion of religious activities such as pilgrimage and the veneration of
relics through his involvement in the construction of stūpas and shrines”,2 and “Aśoka
was the greatest political and spiritual figure of ancient India”.3
Such comments are typical of the way in which Aśoka is described in books about
early Buddhism. As I stated in the first lecture, I have spent a large portion of my
academic life studying Aśoka’s inscriptions, and I do not find that the picture of the man
which emerges from his edicts coincides entirely with what we find written about him. So
in this lecture I want to consider the part which Aśoka played in the history of Buddhism,
and I shall compare what we learn about him from Buddhist texts with the information
which we can get from his own inscriptions.
It is probable that most people know about Aśoka from the information given about
him in the Pāli chronicles, and in particular the , although many of the
same stories are told in greater detail in Sanskrit and Chinese sources.
In the we read how, after the death of his father Bindusāra, Aśoka killed
99 of his 100 brothers, sparing only Tissa, and became the sole ruler of Jambudvīpa.4
After hearing the Buddhist novice Nigrodha preach, he was established in the three
refuges and the five precepts of duty, i.e. he became a Buddhist layman (upāsaka).5 We
are told of Aśoka being a firm supporter of Buddhism, to the exclusion of
1 Rahula, 1956, 54–55.

2 Warder, 1970, chapter 8.
3 Lamotte, 1988, 223.
4 Mhv 5.20.
5 cf. Dip 6.55; Mhv 5.72.
Buddhism and Asoka 97
other religions. He stopped giving food every day to the 60,000 brahmans whom his
father had fed, and in their place gave food to 60,000 bhikkhus. He gave orders for
84,000 vihāras to be built in 84,000 towns,6 of which the most famous was the one which
he himself founded in —the Asokārāma—and he built stūpas in the places
which the Buddha had visited.7 He was persuaded, by the thought of becoming an heir of
the doctrine,8 to allow his son Mahinda and his daughter to join the order in
the sixth year of his reign,9 and Mahinda was subsequently sent as a missionary to Sri
Lanka. At the time of the schism in the order he personally listened to the bhikkhus
expounding their views and was able to decide who were orthodox and who were
heretics. After the schism had been settled, the third was held under his
patronage.10
11
The says that because of his wicked deeds, he was known as
“fierce, or violent Aśoka” in his early days, but later because of his pious
deeds he was called Dhammāsoka. The change of name is, of course, intended to
emphasise the difference between Aśoka as a non-Buddhist and as a Buddhist.
We get rather different information about Aśoka from reading his inscriptions.
For example, the story of his conversion in the chronicles is somewhat at variance
with his own statements. The early history of Aśoka’s involvement with Buddhism is told
in the first Minor Rock Edict. Whereas, according to the Pāli sources, as I have just
stated, he had already been converted to Buddhism, had 84,000 vihāras built, and had
given permission for his son and daughter to join the order within six years of his
consecration as king, nevertheless, we can calculate from Aśoka’s own words that his
conversion to Buddhism occurred fairly soon after the war in , which he states
in the thirteenth Rock Edict [RE XIII(A)] took place when he had been consecrated eight
years. His conversion was presumably because of his remorse, not for his fratricide which
seems to be disproved by the references which he makes to his brothers and sisters in the
fifth Rock Edict [RE V(M)], but for the transportation of 150,000 persons, the killing of
100,000, and the death of almost that many, in . The chronicles show no
knowledge whatsoever of this carnage. At the time at which he promulgated the edict, he
had been a layman for more than two and a half years, including a year when he was not
very zealous—I assume that this means that after he had been converted to Buddhism, he
had not been a very energetic Buddhist for a while—and then more than one year when
he was zealous, after he had “approached the ” (which, perhaps, means that he
went on a refresher course) with good results—“I have made good progress”, he says.
When he issued the first Minor Rock Edict he was, therefore, at a point just short of the
eleventh year of his reign. He issued the third Rock Edict in the twelfth year, so the first
and second Rock Edicts, which are not dated, were issued either in the same year, or in
his eleventh year.
6 Mhv 5.79–80.
7 Mhv 5.175.
8 sāsanassa dāyādo, Mhv 5.197.
9 Mhv 5.209.
10 Mhv 5.280.
11 Mhv 5.189.
As the story is related in the first Minor Rock Edict, however, there is no direct
evidence that it was the Buddhist he went to. Although in one version of the
edict (the one at Maski) he is described as a Budhaśake, which Hultzsch translates as “(I
am) a Buddha-Śākya”, it seems fairly certain that the insertion of the word Budh(a) was
done by a single local scribe to “correct” the word upāśake which he had already written.
It is noteworthy that Budhaśake or its equivalent does not occur in any other version of
the first Minor Rock Edict. All the other versions have the word upāsaka.12
It is probable that the scribe, realising that, as I have just said, there is no indication of
the sect in which Aśoka was an upāsaka, and wishing to make the situation clear to all
readers of the edict, tried to insert the word “Buddha” before the word upāśaka, which
he had just carved on the rock, with only partial success.
Nevertheless, confirmation that Aśoka had become a Buddhist is provided by the
reference to his visit to the bodhi tree, which is described in the eighth Rock Edict [RE
VIII(C)] as happening when Aśoka had been consecrated ten years. This must have been
one of the first consequences of his conversion to Buddhism, and it perhaps coincided
with his visit to the which improved the quality of his religious life.
We can, in fact, be certain that, for Aśoka, means the Buddhist ,
because in the seventh Pillar Edict [PE 7(Z)], when he summarises all his achievements,
he states that he has set up mahāmātras “ministers” of morality to look after the affairs of
the , the , the Ājīvikas, the Jains, and various other religious sects.
In the context, with the other sects specified by name, the , by a process of
elimination, must be the Buddhist .
There is, then, no doubt that Aśoka was a Buddhist layman, but there is no reason to
believe that his mind was closed to other religions, and we read in the sixth Pillar Edict
[PE 6(E–F)] that he had honoured all sects with various forms of honour, and the best of
these, in his opinion, was a personal visit to them. As we shall see, his sort of dhamma
was moral and ethical, as opposed to spiritual, so that he could equally well have been a
Jain layman.
The edicts give a great deal of information about the way in which Aśoka propagated
his own dhamma. He writes, among other things, of dhamma-thambhas “dhamma
pillars”, dhammalipi “dhamma writings”, dhamma- “dhamma ceremonies”,
dhammadāna “dhamma giving”, dhammānuggaha “dhamma benefit”, dhammayātrās
“dhamma journeys”, dhammasavana “hearing the dhamma”, dhammamahāmātras
“dhamma ministers”, dhammavijaya “dhamma victory”.
The problem is to know if Aśoka’s dhamma was the same as the Buddha-dhamma. He
makes it clear that there is a difference between ordinary practices and institutions, and
the dhamma version of them. A pillar is a thambha. It becomes a dhamma-thambha if
Aśoka’s dhamma is carved on it. There were mahāmātras before Aśoka’s time. He was
the first to institute dhamma-mahāmātras to propagate his dhamma. Before his time
kings went on yātrās. He instituted dhamma-yātrās, so that he could practise his dhamma
while on journeys. People performed all sorts of ceremonies ( )—in case of
12 See Norman, 1973, 68–69 (=CP I, 166–67).

illness, and at weddings, to get children, before going on journeys, etc. The
, however, is the proper treatment of slaves and servants, honouring
teachers, self-restraint with regard to living creatures, generosity to and
, etc. [RE IX(G); cf. RE XI(C)].
Aśoka’s dhamma is set out clearly in several inscriptions, e.g. in a concise form in the
second Minor Rock Edict: “Obey one’s parents; obey one’s elders; be kind to living
creatures; tell the truth”. All this is said to be in accordance with ancient usage (porānā
pakati)—a third century B.C.E. version of “back to basics”. Elsewhere, in the third Rock
Edict, a slightly expanded version of this is given: “Obedience to mother and father is
good; liberality to friends, acquaintances and relatives, to and is
good; abstention from killing animals is good; moderation in expenditure and moderation
in possessions are good” [RE III(D)].
The series of seven edicts on pillars, which we call the Pillar Edicts, is devoted to an
explanation of Aśoka’s dhamma, with an account of how he himself has complied with it,
by planting trees for shade by the road-side and digging wells and building watering-
places for men and animals. Pillar Edict 1 tells of government by dhamma. Pillar Edict 2
states that dhamma consists of doing little sin, doing much good, showing compassion,
making donations, telling the truth, and purity. Aśoka has done much good by not killing.
Pillar Edict 3 tells of good and evil, and identified the latter as fierceness, cruelty, anger,
pride, and envy. Pillar Edict 4 emphasises the need for equality of justice and the
rehabilitation of prisoners. Pillar Edict 5 prohibits the killing of a number of animals
which are specified by name. Pillar Edict 6 states that the aim is to bring happiness to all.
All sects are to be honoured, especially by personal visits. Pillar Edict 7 seems to be a
summary of all that Aśoka has done. He explains how kings in the past had sought to
increase dhamma. Aśoka had decided to do it by preaching and instruction, and had
instituted dhamma-pillars (dhamma-thambhas) and dhamma-ministers
(dhammamahāmātras) to put this decision into effect. The dhammamahāmātras were
concerned with all sects. Dhamma is defined again as: obedience to parents, obedience to
teachers, respect to the old, and proper behaviour towards and , to
the poor and to slaves and servants. There had been an increase of dhamma as a result of
Aśoka’s legislation, e.g. about killing animals, but also because of an attitude of mind,
i.e. personal conscience (nijhati). In this way the next world is gained.
Elsewhere, in the series of major Rock Edicts, we read that one must obey the
dhamma and conform to it [RE X(A)]. The gift of the dhamma is defined as the proper
treatment of slaves, obedience to parents, etc., generosity to and ,
and non-killing. The dhamma gives endless merit [RE XI(E)].
Aśoka calls his edicts dhamma-writings (dhammalipis), and we can read them and see
exactly what dhamma each lipi contains. We can see that his dhamma is exclusively a
moral one, which is why we often translate dhammalipi as “rescript on morality”. Aśoka
promoted his dhamma widely, and instituted dhammamahāmātras to supervise it,
dhammathambhas to carve it on, and had messengers (dūtas) to carry it all over India and
even to the Greek kings to the West. Except in so far as the moral ideas are quite in
conformity with Buddhist moral teachings, there is no hint of anything exclusively
Buddhist in them, and in the insistence on non-killing ( ) his thought closely
resembles the Jain emphasis on this, and in fact parallels have been noted between the
lists of animals declared inviolable in the fifth Pillar Edict and lists of animals in Jain
texts.13
In the edict we find that, under his personal name of Priyadassi, Aśoka greets
the , and wishes it well. He goes on to say that it is known how great is his faith
in the Buddha, the Dhamma, and the . It is obvious that in this context Dhamma
has its usual Buddhist meaning as one element of the Triratna, but other than this one
reference, it is, in fact, very clear that Aśoka’s references to dhamma do not refer to the
Buddha’s dhamma, and Aśoka’s dhamma was not the same as the Buddha’s dhamma.14
The quotation I gave at the beginning of this lecture about establishing the sāsana in
countries which he had spiritually conquered seems, therefore, to be based upon a
misunderstanding of the nature of Aśoka’s dhamma.
That same quotation also seems to be wrong when it talks about countries being
spiritually conquered by Aśoka, at least if, by spiritual conquest, it is referring to Aśoka’s
dhamma-vijaya. As I understand the situation, Aśoka expanded his empire by force, but
thereafter devised the principle of victory by morality, and commended it to his
successors.
It seems to me that the guiding principle of Aśoka’s teaching was non-killing
( ), presumably as a result of his remorse over the killing in . His very
first Rock Edict is concerned almost entirely with the prohibition of killing in daily life,
including the killing of animals for sacrifice and for food.
In the thirteenth Rock Edict he tells the story of how, after the victory in ,
with all the death and suffering this involved, he had a great desire for morality
(dhamma) [RE XIII(C)]. It pained Aśoka that those who did obey his dhamma, including
and (J), suffered nevertheless (G). His hope for the forest dwellers is
that they too (like him) may repent (of past killing?) and not kill (in the future) (M-N).
What he now wished for was a dhamma-victory, and this consisted of security for all
creatures, self-restraint, equanimity and gentleness (O). Messengers were sent to preach
this everywhere, including the Greek kingdoms to the West (Q). His conquest by
morality (dhamma) is promulgated in the hope that his successors will not think of
another (military) victory, by force of arms which would entail slaughter, similar to that
in , and that in their own victory there will be mercy (khanti) and light
punishment ( ) (X).
And so it seems to me quite certain that the messengers (dūtas) who were sent to the
Greek kings were not charged with the propagation of Buddhism. It would seem clear
that they were sent in an attempt to persuade the rulers, probably despotic rulers, of the
neighbouring states that they too should give up their desire for conquest by war, and
should try to institute the reign of peace and tranquillity, based upon the principles of
Aśoka’s dhamma. In these circumstances, to talk, as some do, about the Aśokan
missionary expansion of Buddhism among the Greeks, seems to me to be a mistake.
Certainly we have no evidence from the Greek side which indicates that any Buddhist
missionaries had arrived among them c. 250 B.C.E.
13 See Norman, 1967B, 26 (=CP I, 68–69).
14 See Lamotte, 1988, 228.
It will be clear from what I have said so far about Aśoka’s dhamma that those who talk
of him making Buddhism the state religion are very wide of the mark.
In his edicts, Aśoka says little or nothing about Buddhism. There is no reference to any of
the basic tenets of Buddhism, e.g. , mokkha, nibbāna, anattā, the eight-fold
path or the four Noble Truths. In the Separate Edicts he stated that his aim was the
happiness of all (SepE I), and a number of inscriptions include the statement that his aim
was that his people may attain happiness in this world, and heaven in the other world. His
boast was that he had mixed men with gods, a statement which has been variously
interpreted. I take it to mean that he had succeeded in bringing men to heaven, where of
course they will be reborn as gods, i.e. mixed with other gods. This seems to me to be far
from the idea of an endless series of rebirths which we normally associate with
Buddhism.
His general failure to mention Buddhism has been variously explained. It was perhaps
due to ignorance, i.e. although he was nominally an upāsaka, he had very little
knowledge about Buddhist doctrines. Alternatively, perhaps he knew about Buddhism,
but he thought it would be favouring one sect unduly, and thus destroying the impartiality
which he aimed to show elsewhere, if he referred to it in detail. It is also possible that he
thought that it was irrelevant to his purpose in publishing his edicts, namely to spread
knowledge of his own personal dhamma, which was intended to bring peace to his
empire and enable all his subjects to live in harmony with each other.
One of the more bizarre explanations I have come across is the view that what Aśoka
conveys in his edicts is the state in which Buddhism was in his time. Hultzsch says:
“Aśoka’s dharma is in thorough agreement with the picture of Buddhist morality which is
preserved in the…Dhammapada. Here we find Buddhism in statu nascendi”—perhaps,
‘in its infancy’.15 He goes on to say, “In one important point Aśoka’s inscriptions differ
from, and reflect an earlier stage in the development of Buddhist theology or metaphysics
than, the Dhammapada: they do not yet know anything of the doctrine of , but
presuppose the general Hindu belief that the rewards of the practice of dhamma are
happiness in this world and merit in the next world”.16 Hultzsch’s statement raises an
interesting question about the nature of Buddhist theology. Is it possible that the doctrine
of is a later stage in its development? I cannot believe that that is so, and I
therefore think that Hultzsch’s reason for Aśoka’s failure to mention cannot be
correct.
On the face of it, the situation appears to be very similar to that found in Sri Lanka and
described by Richard Gombrich.17 He reports, “But most Sinhalese villagers do not want
…. They say that they want to be born in heaven.” They do, at least, know about
. Aśoka gives no hint of ever having heard
15 Hultzsch, 1925, xlix.

16 Hultzsch, 1925, liii.
17 Gombrich, 1971, 16–17 (quoted by Southwold, 1985, 30).
about it. Although the villagers’ statement may seem strange, it is in fact quite in
conformity with the statement which we find at the end of the Alagaddūpama-sutta of the
Majjhima-nikāya,18 where the Buddha states that those bhikkhus who act in accordance
with the dhamma and with faith, will gain awakening (sambodhi), while those who
merely have faith in, and love for, the Buddha will attain heaven. The villagers know
about nibbāna, but prefer the heaven which they will gain because of their love of the
Buddha. Aśoka makes no mention of loving the Buddha. For him, heaven is gained by
doing the things which he specifies in his dhamma. The statement is not even put in the
context of acquiring good karma. He makes it clear that this is, in itself, the summum
bonum: “What is more important than gaining heaven?” he asks in the ninth Rock Edict
[RE IX(L)].
The strongest indication of his connection with Buddhism is the edict at ,
which I have already mentioned, and we should note that the one place where he actually
refers to the Buddha’s teaching is in this edict addressed to the . In it he says that
everything said by the Buddha was well said, and he commends seven texts by name to
the . He refers to the Buddha’s teaching as the saddhamma, which perhaps was
an intentional action to distinguish the Buddha’s dhamma from his own. We must hope
that Aśoka was preaching to the converted. If his exhortation had been intended for the
common people, it would presumably have been in a Rock Edict. There is the problem
that we cannot be certain of the identity of some of the texts, but since we sometimes find
that the commentaries, e.g. that on the Sutta-nipāta, know of some texts under other
names,19 it is perhaps not altogether surprising that we cannot recognise all of Aśoka’s
choices.
In his statement in the eighth Rock Edict [RE VIII(C)] that dhamma journeys
(dhammayātrās) have replaced the pleasure trips that kings used to take, Aśoka says that
he went to the bodhi tree, but he says nothing about the need for others to go on
pilgrimage to sacred places, or about pilgrimage as a religious activity. As a result of his
visit, he defines a dhammayātrā as an opportunity to travel around in order to put his
dhamma into effect. He says precisely what it entails: giving audiences and making
distributions to and , giving audiences and distributing money to
old people, giving audiences to the country people and preaching to them and answering
questions about his dhamma.
What we read of Aśoka’s own visits to sacred places does not encourage us to think
that he thought that they were very important. It is interesting to note that of the four
places which are sacred to Buddhists—where the Buddha was born, achieved sambodhi,
gave his first sermon and died—we have only first-hand evidence of Aśoka’s visit to two
of them. Although the Aśokāvadāna tells of Aśoka being taken on a conducted tour of
these four places and others, Aśoka himself states that he went to the bodhi-tree in his
10th year, and to the Buddha’s birth-place at Lumbinī, where he had a pillar erected, ten
years later, in his 20th year. To some extent, of course, this is an argument from silence.
Perhaps Aśoka did visit the other places, as the texts say, and
18 M I 142.
19 See Norman, GD II, xxvii.
did raise pillars there. If so, they are now lost The discrepancy in the dates, however, is
less easily explained.
We should note that the visit to sambodhi has been interpreted by some as meaning
that Aśoka was so proficient a practitioner of Buddhism that he actually gained bodhi
himself. Such an interpretation would imply that meditation, bodhi and were, in
fact, known to Aśoka, even though there is no trace of that knowledge in the edicts. Even
as recent a writer as Jules Bloch toyed with this idea, but then decided that “go to
sambodhi” is philologically unlikely as a way of saying “gained awakening, became a
Buddha”, and he concluded that it means “visited the site of the bodhi tree”.20
We have an inscription which tells us that in his 14th year Aśoka enlarged the stūpa of
the previous Buddha Konākamana. It is not clear who built it. The inscription has been
interpreted as meaning that Aśoka enlarged it for the second time, but it is perhaps more
likely that it means “enlarged it to twice its former size”. We have no more information
from the edicts about Aśoka erecting or enlarging stūpas, although the Chinese pilgrims
record the existence of a number of stūpas which were attributed, in their day, to Aśoka,
including one built at the place where the held their assembly.21
Both the Northern and the later Southern Buddhist sources give the information that
Aśoka broke into caityas which he thought might hold relics of the Buddha, and when he
eventually found one with relics in it he re-interred them in 84,000 caityas in the 84,000
vihāras which he had had built. The words budhasa salīle “relics of the Buddha” which
occur at the end of the version of the first Minor Rock Edict found at Ahraurā in 1961,
might be thought to refer to this redistribution of the relics,22 but since the words are
found in only one of the seventeen versions of that edict so far discovered, it seems more
likely that the words are an invention of the scribe at Ahraurā, based upon a
misunderstanding of something which he found in the version of the edict sent to him.23
Aśoka devotes the whole of the twelfth Rock Edict to making it clear that he is equally
concerned with adherents of all religions, and he honours them all with gifts and other
sorts of honours. All sects must listen to each others’ dhamma, so that there may be an
increase of sālā (which I take to mean “communication”) between them [RE XII(I)].
Then there will be an increase in each individual sect and an illumination of dhamma [RE
XII(N)]. Aśoka wishes them all to live in harmony together, without self-aggrandizement
or disparagement of other sects.
Aśoka seems to use the compounds - and -
(in their various Prakrit forms) to mean all members of religious orders,
orthodox and heterodox. In most cases he puts the word - first, but in two
places he reverses the order of the words, and in the fourth Rock Edict he has both forms
of the compound. The scribes at some sites change - to -
, perhaps thinking that to have first was wrong, or perhaps
20 See Bloch, 1950, 112, note 6 and (for the use of the word in the Mahāvastu-avadāna) Yuyama,
1969, 488–92.
21 Beal, 1884, Part II, 164.
22 See Norman, 1983B, 277–92 (291, note 82) (=CP II, 250–68 [267, note 1]).
23 See Norman, 1983B, 288 (=CP II, 268).
believing that they ought to correct what appeared to be Aśoka’s inconsistency. In the
thirteenth Rock Edict he writes of and and members of other
sects ( ). He would surely not have referred to the sects in this way if he had in
fact rejected the completely, as the story suggests. The
variations of this word order are probably due to the regional scribes, who (depending on
their personal feelings) put one or the other first This applies especially to the scribe at
Girnār, who always prefers to have - first This may be connected with the
Sanskritisations we find at Girnār, and suggests that the scribe was perhaps a brahman. In
Pāli texts, it seems to be conventional to have the words in the order -
, which is not surprising. In the seventh Pillar Edict [PE 7(HH)], where we
have only one version of the edict, we find the compound in the order bābhana-samana.
His encouragement of all sects must mean that he did not stop feeding ,
and, as I have said, his dhamma in fact specifically includes giving to and
. His donation of caves to the Ājīvikas in his 12th year24 is additional
evidence that he was not devoted exclusively to Buddhism.
The item which most merits our attention, however, is the set of three versions of the
so-called Schism Edict, because these have been interpreted not only as showing that
Aśoka was sufficiently involved in the affairs of the to be able to settle the
schism in the order ( -bheda), as the says, but also to intervene to
the extent of removing dissenting heretics by force. The question is how far the Schism
Edict reflects the actual events of the schism
They then held the third at which Moggaliputta Tissa recited the Kathāvatthu.
The end of the is dated to the 17th year after Aśoka’s consecration.
The two versions by Buddhaghosa are earlier than the . They are very
similar to each other, and also to the version, with the detail, omitted in the
, that Aśoka gave the heretics white robes when he expelled them from the
Order. The heretics’ practices are said to include the tending of the sacrificial fire, from
which we can deduce that some of them were .
The versions in the are earlier than Buddhaghosa. We should note that
the is a strangely undisciplined text. It obviously represents a
conglomeration of source material bundled together uncritically, so that there are often
two versions of the same event, and sometimes three. There are, in fact, two versions of
the schism story:
(1) The first version26 states that the schismatics and heretics, among whom the
Jains and Ājīvikas are specifically mentioned, had lost gain and honour, and
consequently infiltrated the Order. For seven years the uposatha ceremony was
24 Bloch, 1950, 156.
25 Mhv 5.274.
26 Dīp 7.35–41.
carried out by incomplete groups (vagguposatha),27 since the noble ones did not attend
the ceremonies. By the time 236 years had passed since the death of the Buddha, 60,000
bhikkhus lived in the Asokārāma. The various sectarians ruined the doctrine, wearing
yellow robes. Moggaliputta convened a recitation, and having destroyed the different
doctrines and expelled the shameless intruders, he recited the text known as the
Kathāvatthu.
(2) The second version28 says there was a dreadful schism (bheda) among the
Theravādins 236 years after the death of the Buddha. The heretics (numbering 60,000),
seeing the honour being given to the , furtively attached themselves to it. The
Pātimokkha ceremonies in the Asokārāmavihāra were interrupted. A minister, who
ordered the Pātimokkha ceremony to be performed, killed some of the bhikkhus, which
led to the king consulting the elders about the killings. Moggaliputta presided over a
gathering of 60,000 Buddhists, which had assembled to destroy the sectarians. Aśoka
learned the doctrine from the thera, and is said29 to have destroyed the (bhikkhu)-
30
emblems of the intruders (rājā… …nāseti ). The heretics,
performing the pabbajjā rite according to their own doctrine, damaged the Buddha’s
utterances. To annihilate them Moggaliputta recited the Kathāvatthu. After that recitation
he held the Third Recitation.
If we examine all these versions, we can probably trace the way in which additions were
made to the basic version of the story. It is likely that the first account in the
is the earliest version, It dates the occurrenee, and states that sectarians
whose honour and gain had been reduced because of the growing prestige of the Buddhist
Order infiltrated the order and wore the yellow robe. For seven years the true Buddhists
would not perform the uposatha in their presence. Moggaliputta destroyed the various
doctrines and removed the shameless ones. There is no mention of Aśoka, nor of the
giving of white robes. The second version in adds the statement that there
was a schism (bheda) in the Theravāda. It does not specifically mention the uposatha, but
states that the Pātimokkha ceremony in the Asokārāmavihāra was interrupted, although it
does not say for how long. A minister tried to settle the matter, but his intervention
caused bloodshed. The king asked about the bloodshed, received religious instruction,
and destroyed the sectarians’ (bhikkhu)-emblems.
Buddhaghosa introduces the story of Aśoka becoming so involved in the matter that he
sends a minister, rather than the minister acting on his own responsibility. That minister
tries to settle the matter by forcing the bhikkhus to perform the uposatha, and killed a
number of them in the process. After a week’s training in the doctrine, Aśoka was able to
27 At Dīp 7.36 vagga (<Skt vyagra) is opposed to samagga, according to PED (s.v. vagga2).
Oldenberg (1879, 157) translated vagguposatha correctly, and it is not clear why Law (1957–58,
183) differed from him and, by dividing the compound vaggu (< Skt valgu)+posatha (instead of
vagga+uposatha), translated “pleasant uposatha”, although this is highly inappropriate in the
context.
28 Dīp 7.44–54.
29 Dīp 7.53.
30 Dīp 7.53. It would appear that bhikkhuno is a genitive plural form (= ). For
genitive plural forms in -o, see Norman, 1976C, 124 (=CP I, 244).
discern that the intruders had heretical views, and he consequently made them wear white
robes and expelled them from the Order. The Order is then said to be in harmony
(samagga). The version adds the detail that no uposatha ceremony was
held in Jambudīpa for seven years, nor the ceremony in all the ārāmas.
We can probably reconstruct the account of the matter in the following way.
Sectarians (probably those who had fallen out of favour when Aśoka began to show a
preference for Buddhism) infiltrated the Asokārāma, and the true bhikkhus refused to
celebrate the uposatha ceremony while they were there. There was therefore bheda in the
Asokārāma . It has been suggested that this must have been a
very serious event, which carried a heavier penalty than that laid down for
in the Vinaya- ,31 namely, expulsion from the order, which is what
wearing the householder’s white robes implies. I would suggest, however, that it was not
a question of bhikkhus being forced to wear the householder’s white robes, but of
infiltrators being forced to give up the emblems to which they were not entitled, and
being made to depart from the vihāra, where they had no right to be. The Vinaya
penalties would not be appropriate for those who were not genuine bhikkhus.
I see no reason to believe that Aśoka himself carried out the expulsion. The earlier
version in the states that Moggaliputta removed the heretics, and makes no
mention of Aśoka. It is, however, not unlikely that, as the chronicles say that the
bhikkhus were unable to restrain the sectarians by the rules of discipline, Moggaliputta
was unable to enforce the order of expulsion from the vihāra. In this case, recourse to the
civil power was perhaps inevitable, and a minister had to deal with the matter. This action
would not be a case of one of the king’s ministers intruding into a religious matter, since
those to be evicted were not true bhikkhus.
There seems to be no reason to doubt that this part of the story is historically true. The
version, however, has Aśoka himself becoming involved, doubtless because it
was “his” ārāma. According to this version, he personally sent his minister, and became
further involved after the bloodshed which was caused. Aśoka’s commitment to the
Theravādin cause is emphasised by the story that he personally decided who held the
heretical views, and expelled them from the Order. The story is, however, given a slightly
unreal element by the insertion of a detail whereby Aśoka, after recognising the heresies
of the dissidents, and the correct views of the orthodox bhikkhus, then asked the thera
Moggaliputta Tissa what the Buddha actually taught (Mhv 5.271). When the sectarians
had been removed, the in the Asokārāmavihāra became samagga “united, in
harmony”. The final expansion of the story adds the detail that no uposatha ceremony
31 Causing schism is dealt with in the tenth rule. Anyone attempting to cause
schism should be told to desist. If after three admonitions he still persists, then it is a breach of the
rule. The penalty for this is laid down at Vin III 185, 37–38: va tassā āpattiyā
deti mūlāya deti abbheti; “placing on probation, sending back
to the beginning, inflicting the mānatta discipline, rehabilitation”. It appears that the schismatic
bhikkhus at Kosambī needed to be re-ordained (bhedānuvattakā bhikkhū puna
, Vin II 201, 1–2).
was held in Jambudīpa for seven years, nor any ceremony in all the ārāmas.
These additional details presumably represent an attempt to make the matter appear far
more widespread than it really was.
The shortest version of the three versions of Aśoka’s Schism Edict, which we may
assume gives the gist of the edict, states that the had been made united
(samagga), and that any monks and nuns, who caused schism in the future, should be
made to live outside the dwelling (āvāsa), i.e. the vihāra, and to wear white robes. There
is no information about where the had been made samagga or by whom, and the
order to remove schismatics refers to the future and does not say that any had already
been removed. Consequently, it is by no means obvious that Aśoka’s edict and the story
in the chronicles refer to the same event. Nevertheless, as we have seen, the removal of
the (bhikkhu)-emblems or the wearing of white robes, the expulsion and the
being made united (samagga) are mentioned in some of the Pāli accounts. I believe that it
is too much of a coincidence for there to be no connection whatsoever between the edict
and the Pāli accounts. I conclude, then, that the references in the Pāli texts must go back
to a very early tradition, brought from India and preserved in the Mahāvihāra, that Aśoka
did, or at least wrote of doing, these things. It is interesting to note that the references to
white robes and the being samagga do not occur before Buddhaghosa’s account
of the matter, which implies either that these details were not available to the author of
the , perhaps because they did not yet exist, or else that he chose to omit
them for some reason.
It is obvious that some of the statements made about Aśoka by modern writers can be
verified by reference to his inscriptions.32 Aśoka was a Buddhist and he was emperor of
India, or at least the sole ruler of a large proportion of it. We know of no Buddhist ruler
of this, or any other comparable, territory before Aśoka, so it is not incorrect to call him
the first Buddhist Emperor. On the other hand, some of the statements made about him
seem to be rather extravagant, and not capable of being verified. A case can perhaps be
made for saying that “Aśoka was the greatest political figure of ancient India”, but, in the
absence of any first-hand information about his spirituality, it seems unjustified to say
that he was also the greatest spiritual figure. Aśoka’s own words about the function of
dhamma journeys (dhammayātrās) seem to make his alleged enthusiastic promotion of
religious activities such as pilgrimage and the veneration of relics less than certain.
There are three questions to answer. The first question is: why did the Buddhists claim
that Aśoka was exclusively pro-Buddhist? The second question is: why did the
Theravādins claim that he favoured them at the time of the schism? The third question is:
why do modern writers make claims about him which are not supported by his own
words? The last is easily answered. Modern writers say what they do because they have
either read only the Buddhist sources or been misled by other modern writers. In either
case they have not actually read the Aśokan inscriptions themselves. Those who have
heard that Aśoka recommended certain suttas, i.e. portions of the Buddhist dhamma, and
know that Aśoka set up dhamma-writings and sent out messengers, have put the two
32 See Lamotte, 1988, 253.

pieces of information together, and have assumed that Aśoka set up, i.e. popularised,
Buddhist teachings and sent out Buddhist missionaries. This incorrect view may owe
something to the editor of the Dictionary of Pāli Proper Names,33 who had perhaps
confused Aśoka’s messengers with Moggaliputta’s missionaries, and actually states that
Aśoka sent out the missionaries after the third , although the texts clearly state
that they were sent out by Moggaliputta.
It is clear that the Buddhists appropriated Aśoka for their own use. It is possible that the
earlier version of the story in the goes back to a very early form of the
tradition, when there was no need to “invent” Aśoka’s involvement, whereas the second
story was formulated later, when the political situation had changed. There
was undoubtedly rivalry with the brahmanical caste, when the power of the
grew again after Aśoka’s death, as supported a brahmanical reaction.
Samprati, Aśoka’s grandson, is said in Jain sources to have been a great supporter of
Jainism and, in face of the royal support for both those religions, it was perhaps
inevitable that the Buddhists, building upon the undoubtedly correct fact that he was a
Buddhist layman, would maintain that the great king Aśoka favoured only the Buddhists,
to the exclusion of other sects, and in fact had 18,000 Jains put to death in a single day,
according to the Aśokāvadāna.34 In later times the story was embroidered even more, and
I-tsing reports that an image of the Buddha was dressed in a monk’s robes of a particular
pattern,35 implying that Aśoka was more than just a layman.
It is probable that it was only after the schism in the Buddhist , and indeed
because of the schism, when the inclusion of the word mahā- in the name
seemed to imply that the Theravādins were only a minority sect of Buddhism, that it
became necessary to prove that the Theravādin view of the Buddha’s teaching was the
correct one. The Theravādin school consequently began to make statements about Aśoka
himself favouring their view of the Buddha’s teachings, in order to legitimise their claims
to be the true exponents of the Buddhist tradition.
It has been said that Aśoka’s patronage was responsible for establishing Buddhism
over a far wider area than could have been imagined before the founding of the Mauryan
empire.36 We must note that we can find no evidence in the edicts that there was any
greater patronage of Buddhism than of any other sect. It is probably pure chance that we
have little or no information about the patronage which Aśoka bestowed upon other sects,
although I have mentioned the caves which he gave to the Ājīvikas. It is also probable
that some of the things which the Buddhist texts claim Aśoka did were not done by him
personally, but by the mahāmātras appointed to look after the affairs of the .
There is no reference to Asoka quelling schism in the order ( ) in the
seventh Pillar Edict, which is dated to Aśoka’s 27th year, and seems to be a summary of
Aśoka’s activities as a ruler, nor is there any mention there of the third . The fact
that Aśoka says nothing about the being held under his patronage suggests that
33 See DPPN, s.v.Asoka.
34 Divy 427.
35 See Takakusu, 1896, 73.
36 Cutler, 1994, 33.
he did not include among his achievements any detailed information about the
mahāmātras’ activities. It would have been the mahāmātras who made
whateverarrangements were needed to ensure that the could be held. Since the
Buddhists recognised that they did so on behalf of the king, they were able to claim that
he was their patron.
Similarly, the instructions in the covering letter, which is attached to the version of the
so-called Schism Edict at Sarnath, about mahāmātras coming on every uposatha day to
read and understand the edict, probably refer to the mahāmātras whose duty it was to
look after the . We may not be too wide of the mark if we also assume that it was
not Aśoka himself who officiated at the voting, recorded in Chinese sources,37 after the
dispute which resulted in the arising of the , but the mahāmātras, acting
on Aśoka’s behalf.
If it is true that Buddhism expanded during the reign of Aśoka, then it seems to me
that, rather than this being the result of his patronage, or to his deliberate attempts to
propagate it (for which, as we have seen, there is little or no direct evidence), it is more
likely that it was a result of the peace which he established, leading to greater prosperity
and the expansion of trade. As I said in the second lecture, Buddhism followed the trade
routes. It was undoubtedly those same trade routes which were followed by Aśoka’s
emissary dhamma-mahāmātras, who, as we read in the fifth Rock Edict, were instituted
in Aśoka’s 13th year to look after members of all sects, and to spread Aśoka’s dhamma
through all parts of his territories, and we may be sure that the religious missionaries sent
out by Moggaliputta followed in the footsteps of the mahāmātras.38
I mentioned the doubt about the identity of some of the seven texts which Aśoka
recommended to the . Beside that doubt, there is also a dispute about the
significance of the list, some believing that Aśoka’s ability to mention texts by name
implies that a canon was already in existence in his day. I will deal with this problem in
the eighth lecture.
37 Lamotte, 1988, 172–73.

38 Lamotte, 1988, 301.
VIII
Buddhism and Canonicity
We are accustomed to talking loosely of Buddhist texts being canonical or non-canonical.

Perhaps we ought to try to define what precisely we mean by “canonical”, because when
discussing the period of early Buddhism to which these lectures have been mainly
devoted, scholars tend to use the terms in different, and therefore confusing, ways. For
example, Étienne Lamotte asserted that there was no question of there being either a
canon or a before the end of the Mauryan period, and he stated categorically
that there was no canon, Magadhan or otherwise, before the period of Aśoka.1 Since it
has been pointed out that a western term such as “canonical”, although convenient, must
nevertheless be used with circumspection,2 we must recognise, when considering
Lamotte’s statement, that much hinges upon the definition of “canon”, because there is a
danger that we may be perpetrating an anachronism by trying to read the wrong concept
of the word “canonical” into the period immediately preceding and following the time of
the Buddha’s death.
I want in this lecture to restrict myself to the canons of the Indian schools of
Buddhism, and specifically to the Theravādin canon, although there is no reason to think
that, in principle, the canons of other Hīnayāna schools were formulated in a different
way. We simply do not have enough information to be certain about it.
To define the word “canon” briefly, we may say that it is a collection of scriptures
(oral or written), which gives a certain authority to those texts included in it. The
collection may be open, giving the possibility of other texts being added to it, or closed,
which implies that the texts listed in it, and no others, are documents fundamental to the
religion concerned.
For the most part, except where the founder of a religion produces a closed body of
inspired scripture, many religions start with a body of texts which have been collected
together over a period of time during the life and after the death of the founder, and it is
only later, if at all, that the list is closed, sometimes after the weeding out of texts which
were at one time thought of as authoritative, but about whose nature doubts have arisen.
So in the case of the Christian scriptures, it was only some centuries after the foundation
of the church that the present contents of the
1 Lamotte, 1988, 562.

2 Brough, 1962, 33.
New Testament were given canonical status, in the second, closed, sense. The situation in
Jainism is confusing. One section, the Digambaras, believe that all the early teachings of
the founder Mahāvīra have been lost, and what they regard as authoritive is a number of
texts written by eminent Jain authors from the second century C.E. onwards. The
Śvetāmbara section of the Jain community, however, believe that not all the early
scriptures were lost, but even they are not in total agreement about those which remain.
One sect rejects a number of texts which are accepted by another, and even the latter
cannot agree about the texts which are to be included in the “minor” class,
with the manuscript tradition rarely agreeing about the texts in this category.
If we speak of the Theravādin canon, in the form in which we have it now, it clearly
comes into the second category. It is closed. The concept of canon for Mahāyānists and in
Tibet and China differs. Some traditions include texts which are canonical and non-
canonical by Theravādin standards, and there was the possibility of making additions at a
late date, so that it was, for example, possible for a translation, and a bad translation at
that, of the Jātakamālā to be added to the Chinese canon at the end of the eleventh
century C.E.
Perhaps we should begin by considering the categories which do not necessarily
constitute a canon. There are in the Theravādin canonical texts traces of a different
division of the dhamma from the one we have now—one into nine . The number
is increased in some traditions, with a list of twelve being found in the
Mahāvyutpatti.3 Not all the divisions in the list (
)4 are easily
5
explicable. According to the , the change from the old ninefold
classification into the later one took place at the first joint recitation ( ). This is
therefore taken by some as implying that there was a nine-fold canon already in existence
during the Buddha’s lifetime. The seem, however, not to refer to precise portions
of the canon, but are rather a classification of types of texts, at least if we follow
Buddhaghosa’s interpretation of their meanings,6 and it does not seem possible to take
them as evidence for the existence of a canon in the second sense of the word.
In the quotation which I gave at the beginning of this lecture, Lamotte made a distinction
between canon and . He was right to do so, since it is clear that the word
“ ” is not synonymous with “canon”. is a division of texts rather than
an assessment of their authority, and this and other terms, such as nikāya, only came into
use at some time after the Buddha’s death.
The word “having three ” occurs in the canon in the Parivāra to the
7
Vinaya, and and occur in the Milindapañha.8 The Vinaya reference
occurs in the list of theras prefixed to the Parivāra, which is probably an addition made in
Ceylon as late as the 1st century C.E.9 It is difficult to date the Milindapañha, but it must
3 See Thomas, 1951, 276 foll.

4 e.g. M I 133, 24–25.
5 Dīp 4.18–20.
6 Ps II 106, 8 foll. See von Hinüber, 1994.
Buddhism and Canonicity 113
have been composed long enough before the time of Buddhaghosa for it to be regarded as
authoritative, for he seems to quote it on the question of tradition, authority, etc.10
Nevertheless, these texts, however they are dated, must be some centuries before
Buddhaghosa, and the word , and the idea of , must therefore be old.
We are accustomed to refer to the Buddha’s teaching as the “the three
baskets”, but this is not the only way in which the Buddhavacana was divided up. Hsüan
tsang records the fact11 that the collected the Sutra- , the Vinaya-
, the Abhidharma- , the miscellaneous (Khuddaka-nikāya), and the
- , and thus they distinguished five . It is noteworthy that they
call the Khuddaka-nikāya a , but this arises from the fact that the word is
not only used of the baskets of the Buddha’s teachings. There is a Theravādin canonical
text called the Cariyā- , and is used in the Theravādin canon itself to refer
to the collections of other teachers, including brahmans.12
It is clear that the words and simply denote a type of text or an
arrangement of texts, and do not, in themselves, imply any sort of canon, open or closed,
although of course the words can be, and are, applied to a body of scriptures which is
regarded as canonical.
The use of the word “pāli” in this connection is both confused and confusing.
Although it is often translated as “canon”, it is more correctly translated as “text”, since it
is the complement to (“commentary”), being the text on which the
commentary is written.
For the most part, the we have are on canonical texts, in the sense of the
closed canon which we know of from Buddhaghosa’s list of texts in the three .
There is, however, an upon the Netti- , and in it the author
Dhammapāla refers to the Netti as a , and the also refers to Netti- .13 The
bibliographical text also calls the commentary upon the Netti an
.14
This same criterion applies to certain portions of the Jātakas “the birth stories of the
Buddha”, which are for the most part in verse, the prose being later and commentarial.
7 Vin V 3, 14 (Khemanāmo [m.c.]).

8 Mil 18, 15 ( ); 19, 14 (thero ); 90, 4 ([Nāgaseno]…so pi
[m.c.]).
10 Sp 230, 28 quoting Mil 148, 6 foll. (but in a slightly different form from the text as we have it).
11 Beal, 1884, Part II, 164–65.
12 M I 10 and II 169, 13. See Collins, 1990, 89–126 (108, note 15).
13 Quoted by Hardy, 1902, x, note 6 (on xi) and xi, note 1.

14 Collins, 1990, 108, note 11.
There are, however, certain portions in prose upon which there is a commentary,
implying that these prose passages are pāli. We in fact can accept that this is so, because
almost without exception15 such prose passages occur in the -jataka, and they
are, for the most part, composed in a very archaic form of rhythmical prose known as
, quite different in style from the ordinary commentarial narrative type prose.
Nevertheless, the whole discussion about pāli and is made uncertain by
the fact that there are certain passages where pāli means nothing more than text, with no
implication whatsoever about canonical or non-canonical status.16
The fact of the matter is that there is no single Pāli word which exactly coincides with
our word “canon”. It is probable that what we call a canon in this context is what early
Buddhists would describe as Buddhavacana “the words of the Buddha”, since they were
probably concerned only with what the Buddha taught.
I mentioned in the fifth lecture the four mahāpadesas,17 the four references to
authority: the Buddha, a community with elders, a group of elders, and a single elder,
mentioned in the Dīgha-nikāya and elsewhere in the Pāli canon,18 which were concerned
with finding out whether a particular teaching was Bhagavato . The Buddha’s
criteria were clearly intended to be applied after his death, when the Buddha himself was
no longer available as an authority. In effect, anything that was said to be Buddhavacana
and was consistent with the sutta and the vinaya was acceptable. The importance of
obtaining such a sutta is shown by the existence of a rule in the Vinaya stating that a
monk may interrupt his rains retreat for up to seven days to get a text which may
otherwise be lost.19
If, then, by canon we mean a body of texts regarded as having a specific kind of
authority,20 there seems to be no reason to doubt that such a body of texts began to come
into existence very soon after the Buddha’s death, and indeed might have already existed
in embryonic form during his lifetime.
Buddhaghosa talks of bringing together in the Vinaya- what had been both
recited and not recited at the first , recognising that many sayings of the Buddha
had escaped the attention of the , and also that many additions to the
collections were made at a date subsequent to that of the first .21
If, however, the canon is what is Buddhavacana, then we have problems with the
various texts which are ascribed to various persons other than the Buddha. We find that
notice is taken of this in some of the early canonical texts, where an enquirer received his
15 The cty is also found at Ja I 7.

16 von Hinüber (1977, 237–46 [244]), draws attention to: catunnam pi catasso gāthā bandhitvā
pālim eva uyyoesi, Ja VI 353, 11–12: “Composing four verses, and making the
four learn them, he sent them off”.
17 D II 123–26.
18 D II 124, 10 foll.=A II 168, 11 foll. See Collins, 1990, 109, note 18.
19 āgacchantu bhaddantā pur’ suttanto palujjati, Vin
I 140, 37 foll. See Norman, 1988, 1–27 (14) (= CP III, 225–43 [236]).
20 See Collins’ Preface to Horner and Jaini, 1985, ix.
21 Jayawickrama, 1962, 100 (§ 20, note 1). ca ca (Sp
18, 3–4).
reply, not from the Buddha, but from one of the Buddha’s close followers, and we are
told that this is in effect the word of the Buddha because the follower has answered in
exactly the same way as the Buddha would have answered had he been present, and it
was therefore approved of by him.22
The summaries of the story which we find in some texts, e.g. the
Mahāparinibbānasutta, including the information about the distribution of relics, the
portions of verses which we find in, say, the Suttanipāta, which are ascribed to the
,23 all go to show that the concept of a closed canon or a canon consisting
entirely of Buddhavacana cannot be entirely correct.
The very fact that the Kathāvatthu was recited at the third , in the reign of
Aśoka, somewhere around 250 B.C.E., i.e. a century and a half or so after the death of the
Buddha, however, produced a controversy about the propriety of regarding it as Buddha-
bhāsita,24 and the Mahāvihāra tradition, reported by Buddhaghosa, had to explain that it
was Buddha-bhāsita because the Buddha had drawn up the table of contents, foreseeing
that it would be elaborated by Moggaliputta Tissa at a future date.25 The reciter was
therefore was merely giving words which the Buddha had already put into his mind, so to
speak. The Sarvāstivādins faced a similar problem by saying that the Buddha composed
the Jñānaprasthāna, but Kātyāyanīputra edited it after the Buddha’s death. Buddhaghosa,
in his account of the first , states that the fifth nikāya, the Khuddaka-nikāya,
included whatever sayings of the Buddha were not included in the first four nikāyas.26
Since he lists the texts included in that nikāya, he must have been aware of the fact that
many of them, e.g. the Niddesa, Theragāthā, Therīgāthā, Vimānavatthu, Petavatthu, and
portions of the Apadāna, are attributed to others. It is, however, noteworthy, that, in the
Theragāthā and Therīgāthā, many of the verses are said to have been uttered by the
Buddha in the first place. They only became the verses of the thera or therī after they had
repeated them.
According to the Saddhammapajotikā,27 the commentary upon the Niddesa, the
Niddesa was composed by the Buddha’s disciple Sāriputta, but if this was so, then it is
strange that he gives three different explanations of his own words as recorded in the
Suttanipāta. This would seem to indicate that although some of the Niddesa material
perhaps goes back to the time of the Buddha, the work as a whole was later.28
What, however, are we to make of a text like the Apadāna, where we find that some
individuals are actually acquainted with the Kathāvatthu,29 therefore making (those
22 The justification for the of the Majjhima-nikāya (18) and other suttas is
that (in that particular case) Mahākaccāna had answered in exactly the same way as the Buddha
would have done, had he been asked.
23 See Norman, GD II, xxxv.
24 pan’ aha: kasmā ? As 3, 25–26.
25 As 5, 32.
26 Sp 16, 14–15.
27 Nidd-a I 1, 14–15.
28 See Norman, 1983C, 84–85.
29 Kathāvatthuvisuddhiyā, Ap 37, 1; Kathāvatthuvisārada 550, 21, both times in conjunction with
Abhidhammanayañña.
sportions at least of) the text even later than Aśoka?

It becomes clear that Buddhavacana was interpreted in a broader sense.
Buddhavacana, in this broad sense, means the sayings of the Buddha, but sometimes it
means what the Buddha would have said, had he been there, or sayings about the Buddha,
or sayings in accordance with the Buddha’s teaching.
How did the Theravādin canon develop from being an open collection of
Buddhavacana into a closed canon?
As far as the contents of the Theravādin canon are concerned, we have the information
which Buddhaghosa gives us. He lists three : Vinaya-, Sutta-and Abhidhamma-;
the Sutta- has five nikāyas, the fifth of which, the Khuddaka-, contains fifteen
texts, although he also mentions a division into fourteen parts,30 without specifying which
text is omitted under this scheme; the Abhidhamma- contains seven texts. He tells
us, however, that some had different ideas, and also put texts into different
categories.31 He was giving information about the situation in his own time, but the
situation in earlier times must have been rather different.
In the third lecture I spoke about the tradition that a collection of teachings of some
sort existed and was recited immediately after the Buddha’s death, and I dealt with the
institution of the , the reciters of various parts of the Theravādin canon. I
suggested that this event was not as early as the tradition suggests. Although we can point
to the references to the foundation of the system as support for the existence of
the nikāyas—the constituent part of the Sutta- —at an early date, it is not in the
canon itself but only in the commentaries that we find mention of these reciters,32
although there are references to them in early inscriptions.33 There are also inscriptional
references to nikāyas, although we have no idea of the contents of those early nikāyas.
We must accept, then, that the suggestion that the were instituted at the first
recitation creates difficulties because the Sutta- was probably not in its present
form at that time. It is obvious that if the Dīgha-nikāya, Majjhima-nikāya, etc., had not
yet been formulated and named, there could hardly have been Digha- and
Majjhima- , etc.
There is also some doubt in the tradition as to when the Abhidhamma came
into existence. In place of the expected abhidhamma-dhara “remembering, i.e. an expert
in, the abhidhamma” to go with vinaya- and sutta-dhara “expert in the Vinaya and the
30 As 26, 3.
31 See Norman, 1983C, 9. The Digha- put the Khuddaka-gantha into the
, classifying the texts in the following way: Jātaka Mahāniddesa
Cūlaniddesa Suttanipāta Dhammapada Udāna Itivuttaka Vimānavatthu
Petavatthu Theragāthā Therīgāthā. The Majjhima- , however,
32 See Norman, 1983C, 8–9.
33 See Paranavitana, 1970, civ, cvi.
Sutta- ”, in a number of places in the canon we find mātikā-dhara, implying

that only the mātikās “the indexes” to the Abhidhamma texts existed at an earlier time.
It is quite clear that, as long as the canon was open for any text to be added to those
texts, by following the principles enunciated in association with the four mahāpadesas,
then the canon in so far as it consisted of the word of the Buddha was not closed. At some
stage, however, each of the must have become closed and the question is to
decide whether they all became closed at the same time, or whether at different times,
and if at the same time, what that time was.
Ever since the time when the discovery of different sets of canonical texts belonging
to different schools of Buddhism undermined the claim of any one version to be the
authentic word of the Buddha, some scholars have nurtured a hope that, by delving
behind the texts which we have available to us, we can find put these texts in the Sutta-
, together with Apadāna and . Both groups of
thus omitted the from the canon. We must assume that the
Digha- closed their list of the Khuddaka-nikāya before Apadāna
and were added to it, while the Majjhim- closed their list before
the was reckoned as being canonical. the texts upon which they are all
based, perhaps the very words of the Buddha himself. As part of the task of discovering
that underlying canon, scholars have subjected the various versions of the Buddhist canon
we possess to careful scrutiny, and by seizing upon the various anomalous factors they
have discovered therein they have been able to find traces of earlier languages.
I mentioned, for example, in the fourth lecture that there were different views taken of
the non-Pāli features of the Theravādin canon, which were obviously remnants from
earlier dialects through which the texts had passed. S.Lévi,34 when he discussed them,
spoke of them as “pré-canonique”. Lüders,35 however, decided that they were remnants of
an Ur-kanon “a primitive or original canon”, the language of which he called “Old
Buddhist Ardhamāgadhī”. He regarded this as the language of the original canon of
writings. Both scholars had the idea that there was something earlier than the Pāli canon,
but as can be seen from the way in which they described the material, it is uncertain
whether the earlier form should be described as a canon, or whether it should be
described as pre-canonical. Most scholars now, I think, would be very wary about this
whole matter. These forms, in themselves, show nothing more than that the material had
existed in another dialect before it was taken over into the Theravādin canon. It is, of
course, possible for both Lévi and Lüders to be correet, if they were following different
definitions of “canon”.
The problem is to decide when the open canon became a closed canon. When did the
body of the Buddha’s teaching gain the status of canonicity? In general, scholars are far
from unanimous about the time when the Theravādin canon became closed, and we are
not likely to be able to come to any definite answer about this in this lecture. I would
hope, however, that we will be able to narrow things down a little.
There are those those who say that the Pāli canon, as we know it, was not fixed until
34 Lévi, 1912, 495–514.

35 Lüders, 1954.
the time of Buddhaghosa who, by listing the texts which he regarded as forming the
various constituent parts of the , in effect defined and limited the scope of the
. There are, on the other hand, others who believe that the writing down of the
canon in the first century B.C.E. had effectively done that already.
Lamotte seems to have held both views. He stated that the Pāli canon was open until
the fifth century C.E., but this is at variance with another statement which he made with
reference to the writing down of the Theravādin canon in the first century B.C.E.: “From
that moment”, he said, “the text of the in Māgadhabhāsā was drawn up in its
final form”.36 Although it is possible that he made this statement as a summary of the
tradition in the Pāli chronicles, and not as his own point of view, this precise form of
words does not occur in any text known to me. I mentioned in the third lecture the fact
that Buddhaghosa used commentarial material which was perhaps centuries old, and it is
possible that the number and names of texts which make up the canon were indeed fixed
at the time of writing the canon down.
Nevertheless, when Lamotte said that, with regard to the Pāli canon, the collections
were not closed before the time of Buddhaghosa in the fifth century of the Christian era,37
we might think that, while this might be true of the collections as a whole, it was
probably not true of the greater part of them. It is hard, for example, to think of the
Vinaya- being open until so late a date. On the other hand, it could well have been
true of the Abhidhamma- , since, as we have seen, the Kathāvatthu was not
composed until the time of Aśoka.
We can give tentative dates to the closure of some parts of the Theravādin canon. The
fact that the Theravādin version of the Vinaya and the Vinayas of other schools of
Buddhism include accounts of the events which led up to the second , and an
account of the second itself, shows that those accounts were composed after that
event. Traditionally it took place 100 years after the death of the Buddha, but in view of
the fact that 100 is a round figure, not intended to be accurate, we can probably assume
that the second was somewhere around 60 years after the death of the Buddha.
There is, however, no reference in the Vinaya to the third , held during the reign
of Aśoka, i.e. about 150 years after the Buddha’s death. This would suggest that the
Vinaya had not been closed by the time of the second , but was closed before the
third .
The fact that the Theravādin canon includes the Kathāvatthu shows that the
Abhidhamma- was not closed before the third . Similarly, since the
Niddesa is a commentary upon a canonical text, it must be later than that text. If the
Kathāvatthu was the last text added to the Theravādin canon, then the Niddesa is earlier
than this. The states that the Niddesa was one of the texts rejected by the
after the second .38 This perhaps is not to be taken literally, but
36 Lamotte, 1988, 558.
37 Lamotte, 1988, 562.

38 Dīp 5.37.
it means that the Niddesa was not included in the canon, either because
it did not exist at the time of the schism, or because it had not yet attained canonical
status at the time the split from the Theravādins, at some time before
the third .
With these exceptions, we probably have to agree in principle with those who date the
closure of the canon to the time of Buddhaghosa. There is, however, some evidence
which suggests that the canon had been closed at an earlier time. As I said in the fifth
lecture, despite the separation of the Sinhalese Buddhists from North India,39 it seems
that literary material continued to reach Sri Lanka, but there is no evidence for the
addition of any complete text to the Theravādin canon after the at the
Ālokavihāra in the first century B.C.E. An origin in North India is postulated for the
Milindapañha,40 the and the . The fact that these texts are
highly regarded by the commentators, but are not given canonical status, suggests that
they arrived in Sri Lanka after the closure of the canon. This view is supported by the fact
that these post-canonical works contain a number of verses and other utterances ascribed
to the Buddha and various eminent theras, which are not found in the canon. It was
possibly such material that Buddhaghosa had in mind when he referred to some of the
Buddha’s utterances being pāli-muttaka “not included in the texts”.41 Nevertheless, there
was no attempt made to add such verses to the canon, even though it would have been a
simple matter to insert them into the Dhammapada or the Theragāthā. This perhaps
supports the suggestion I have made that Buddhaghosa was sometimes giving the views
of commentators who lived long before him, so that the canon had effectively been
closed before his time.
Another point to consider, and I will return to this in the ninth lecture, is the fact that
some of the best known stories in Buddhism, e.g. the story of the maiden
Kisāgotamī,42 are known in the Theravādin tradition only in the commentaries, although
they are found in texts which are regarded as canonical in other traditions. It would make
no sense to say that these stories are inventions of the Northern Buddhists which reached
Sri Lanka too late to be included in the canon, if it was really open until the time of
Buddhaghosa, who translated many of the Mahāvihāra commentaries into Pāli. We have,
therefore, to assume that at least the Vinaya- and Sutta- had been closed at an
earlier date.
39 The break between North India and Ceylon was clearly not an abrupt one, nor even a complete
one, for Theravādins continued to reside near Bodhgayā for some centuries. The fame of the
Sinhalese commentaries was sufficiently widespread in North India to attract Buddhaghosa to
Ceylon.
40 See Norman, 1983C, 108, 110, 111.
41 See Collins, 1990, 92.
42 As Collins points out (1990, 117, note 55); cf. Mvu II 157, where the lady is called .
There is some doubt about the status of four texts (the , Sutta-
, and the Milindapañha) in Burma. In a survey of Pāli
literature in Burma they are said to be recognised as canonical in that country,43 and
certainly the , and Milindapañha are added to the usual
fifteen texts of the Khuddaka-nikāya in the edition of the canon which was carved on
stone by King Mindon after the fifth in the nineteenth century, and also in the
printed sixth ( ) edition in this century. A reviewer of the
survey,44 however, said that this statement was mistaken, and commented: “No educated
Burman, lay or monk, ever included these four works among the books of the
Khuddakanikāya. They are placed after the books of the Khuddaka-nikāya only because
of their intrinsic value (this applies to the first and the last two, the second [the Sutta-
] being merely a collection of suttas) as a help to the study of the Scriptures”.
The itself includes the statement that it was recited at the first
, in which case we must ask why Buddhaghosa did not include it in his lists, and
it is perhaps the designation as , which I have mentioned, which has led to the Netti
being added at the end of the Khuddaka-nikāya in Burma. We can also see why the Sutta-
might be included. It is in fact a selection of the Buddha’s suttas and is,
therefore, Buddhavacana. We are back to our original question of what constitutes a
canon. If these texts are published with the canon, how are we to decide whether they are
regarded as canonical or not?
I have just said that the fact that the Kathāvatthu could be added to the
at the time of the third shows that that at least had
not been closed before the reign of Aśoka. Do we have any firm evidence that there was a
canon of any sort in existence at that time?
Aśoka’s inscription, in which he recommends seven texts to the ,45
has been variously interpreted. There is the problem that we cannot be certain of the
identity of some of the texts which Aśoka recommends, but since we sometimes find that
the commentaries, e.g. that on the Sutta-nipāta,46 know of some texts under other
names,47 it is perhaps not altogether surprising that we cannot recognise all of Aśoka’s
choices. It has been suggested48 that, as Aśoka specified the “Exhortation to Rāghula in
regard to lying”49 as a text to be studied, he must have known of the existence of the
43 Bode (1909, 4–5) states that the Burmese tradition adds to the fifteen ancient texts of the
Khuddaka-nikāya four other works—the Milindapañha, the , the
and the .
44 Duroiselle, 1911, 119–21.
45 imāni : vinayasamukase anāgatabhayāni
munigāthā moneyasute upatisapasine e ca lāghulovāde adhigicya bhagavatā
budhena bhāsite. There are many discussions of these titles. See Winternitz, 1933, 606 foll.
46 For such alternative titles, see Norman, GD II, xxvi–xxvii.
47 See also Balbir, 1993, 47–59 (53).
48 Winternitz, 1933, 607.
49 M 61.
other exhortation to Rāghula,50 but this does not follow—it may have been his teacher
who specified the sutta in this way. The question is whether this reference to texts
indicates that the Buddhist canon was already in existence in Aśoka’s time, as some have
suggested.
He clearly had knowledge of some individual texts. In what language were they
written? The titles are in an Eastern Prakrit, but that may simply be as a result of Aśoka
using such a Prakrit. It tells us nothing about the dialect of the texts—presumably an
Easterner could read a Western dialect—but on balance it seems more likely that he knew
the texts in an Eastern form. And it was therefore not the Theravādin canon as we know
it. It seems to me that all that Aśoka’s inscription proves is that texts were still thought of
as individual suttas, not as parts of nikāyas or other large collections, and that these seven
texts were already thought of by some, including Aśoka (and presumably his teacher who
recited them to him), as of particular value.
The fact that Aśoka specified suttas by name, rather than referring to nikāyas or
need cause no surprise. Although there are very few references in the
Theravādin canon to portions of the canon by name, what references there are, are to
individual suttas, and there is evidence that texts were often spoken of as individual
suttas, by name, rather than as parts of nikāyas.51 We should remember that the Niddesa
is a commentary on two vaggas and one sutta of the Sutta-nipāta, which suggests that
those vaggas were still thought of as individual vaggas, when the Niddesa was
composed. As can be seen from catalogues of manuscripts, what was transmitted in
manuscripts was frequently individual suttas rather than the whole sutta- , or even
a single nikāya.52 Similarly, it is interesting to note that when texts were exported to other
countries, e.g. Burma, it would seem that they were taken as individual texts or as groups
of texts, rather than as nikāyas or .53
When Lamotte wrote of the impossibility of there being “a canon or a ”
before the end of the Mauryan period, it is not clear what distinction he was making
between the two, but if by “canon” he meant a closed canon, then there can be no doubt
that he was correct. Nevertheless, the fact that Aśoka could quote titles of suttas means
that those texts were available for recitation to laymen, and in that sense we can say that
some sort of collection of the Buddhavacana was in existence in Aśoka’s time.
We know little about the closure of the canons of other schools of Hīnayāna Buddhism.
As I said earlier, there is no reason to think that, in principle, the canons of other
Hīnayāna schools were formulated in a different way. We can surmise that, like the
Theravādin canon, they were at first open, but were closed at some later date.
The tells us something about the contents of the canon:
“They rejected the Parivāra, Abhidhamma, , Niddesa, and a portion of
50 M 62.
51 Balbir, 1993, 50.
52 See Hallisey, 1990, 155–95 (162).
53 See Luce, 1974, 125–27.
Jātaka”.54 This probably means, not that they actually rejected these texts at the time of
the split from the Theravādins, but that they were not in the canon.
Since all these texts, with the exception of the Jātakas, are accepted as late compositions,
on stylistic and other grounds, we can surmise that they were not in the
canon because they did not exist when the split took place. We may therefore date their
composition some time after the second .55 In the case of the Jātakas, the
collection available to the would have represented the size of the
Jātaka collection at that time. Further additions must have been made by the Theravādins
before the canon was finally closed,56 just as the presence of Jātaka stories in the
Mahāvastu57 (which is described as being based on the redaction of the Vinaya-
made by the noble , the Lokottaravādins of the Middle Country)58
which have no counterpart in the Pāli collection shows that the also
continued to add Jātaka stories to their canon.59
The gives a description of the ’ canon: “Altering the
original redaction, they made another redaction.60 They transposed suttas which belonged
to one place (in the collection) to another place; they destroyed the meaning and the
doctrine which were in the five nikāyas. The monks, not knowing what was taught with
exposition or without exposition, neither the natural meaning nor the recondite meaning,
placed in one place the meaning which was spoken with reference to another. Those
monks destroyed much (true) meaning under the shadow of the words”.61 These words
might have been written with the Mahāvastu in mind, for the inclusion in it of suttas
which were in the Theravādin Sutta-nipāta, e.g. Sabhiya-sutta, Nālaka-sutta, and
-sutta, and Jātakas which were in the Theravādin Jātaka collection, e.g.
Kusa-jātaka and -jataka, would inevitably give the impression to any
54 Dīp 5.37.
55 “Council” is misleading as a translation of . It was a recitation carried out by the
Theravādins alone. The tells us that most of the Northern Buddhists sects separated
from the Theravādins in the second century after the Buddha’s death. Few of these sects are likely
to have known of the , much less have regarded it as authoritative.
56 “Closed” implies that no new texts were admitted to the canon after that time, not that the texts
themselves underwent no change. The Abhidhamma texts, excluding the and
Kathāvatthu, probably consisted of nothing more than the mātikā lists, with a few examples to
show how they should be elaborated. I assume that the , since it incorporates a
portion of ancient cty, is earlier than the Third Council in its present form. I accept the tradition that
the Kathāvatthu was compiled by Tissa and was recited at the Third Council.
57 e.g. -jataka (Mvu II 271–76), -jataka (III 33–41).
58 Mvu I 2.
60 bhikkhū \ bhinditva
, Dīp 5.32.
61 Dīp 5.32–35.
Therāvadin who examined the canon that the Theravādin collections

had been re-arranged.62 The phrase “placed in one place the meaning which was spoken
with reference to another” describes very accurately the displacement of some of the
pādas in the Sabhiya-sutta from their original position, so that they are no longer in close
contact with the words they are defining.
An example of the uncertainty about the meaning of the word “canon”, which I
mentioned earlier, can be seen in Lamotte’s use of the word when referring to the
Gāndhārī dialect. He stated63 that the mere existence of the Gāndhārī [Dutreuil de Rhins]
Dharmapada “does not allow us to infer the existence of a canon in North-Western
Prakrit”. The fact that one text does not prove the existence of a closed is so
self-evident that it seems unlikely that Lamotte could have been using the word in that
sense. If he was using the word in the other sense of “canon”, then we must remember
that he was writing before the publication of John Brough’s edition of the Gāndhārī
Dharmapada, in which the idea of such a canon was considered in detail. Brough
concluded that it was difficult to believe that a group of monks, whom he identified
tentatively as Dharmaguptakas, might have possessed a Dharmapada, without at the same
time possessing at least some stock of Sūtra and Vinaya works.64 It has also been shown
that some Chinese translations of canonical texts appear to have been made from the
Gāndhārī dialect, or something very similar to it. This does, however, raise the question
of whether such a collection merits the name of “canon”, and clearly this depends on
which definition of “canon” we adopt.
If we accept the latest date proposed for the closing of the canon, i.e. the time of
Buddhaghosa,65 then we should note that, in some cases, this may have been effective
only so far as the titles of texts were concerned, not necessarily the contents.
The form of some texts, including the Kathāvatthu, Dhammapada, Apadāna,
Theragāthā, and Therīgāthā, is such that even when the texts themselves had been
accepted as canonical and included in an individual nikāya, or of the
Buddhavacana, additions could still be made. As I mentioned in the third lecture, the
uddānas to the Theragāthā and Therīgāthā, the index verses found at the end of the text,
give a list of the number of theras and therīs and their verses which does not accord with
the texts as we have them now, showing that changes have been made since the uddānas
were formulated, doubtless at one of the , by those who were holding the
, the -karas.
It is probable that the Kathāvatthu which was composed at the time of the third
62 It seems likely that at the time of the Second Council individual suttas and Jātakas still had an
independent existence, and after the schism the sects were able to arrange them as they wished. It is
clear that at the time of the composition of the Niddesa the Sn collection was still regarded as
separate vaggas, not as a whole. This would explain why cties were made on two vaggas only. The
existence of the cty on the -sutta presumably shows that this was still regarded as
an individual sutta.
63 Lamotte, 1988, 573.
64 Brough, 1962, 43.
65 Lamotte, 1988, 562, 140 foll. (especially 163 foll.).
, was in fact only the core of the text as we have it, and its form was not finally
fixed until the early commentators, whose work Buddhaghosa made use of, wrote their
commentaries upon it. Even after that time we might believe that minor additions could
be made. The first part of the Kathāvatthu deals with items which we can identify with
the five points of Mahādeva, which the other Buddhist traditions state caused the schism
in the Buddhist which led up to the third , at which the Kathāvatthu was
recited. Because of the form of the text, there would have been no problem whatsoever in
adding further material dealing with other points of doctrinal interest, over the next few
centuries, until the text was finally fixed.
As a general principle, we may note that once a portion of a text had been commented
upon, then the presence of canonical words in the lemmata, and in the explanations of
those words, meant that changes were very unlikely to be made. We must, however, note
that not every word in every text is commented upon, and we must not assume that
because something is not commented upon it must be a later addition. All we can say is
that a word or phrase, which is commented upon, was in the text in front of the
commentator. There are, however, in the Suttanipāta verses which have no old
commentary upon them,66 and it might be argued that they were not commented upon
because they did not exist at the time when the commentaries were being compiled.
We can conclude that the form of the Theravādin canon, and the texts it comprises, are
fixed by the information Buddhaghosa gives. He records other views about this, but
clearly his version prevailed. The latest date for the closing of the canon, i.e. the date of
the establishing of its canonicity, is, then, the fifth century C.E. We can, however, to a
certain extent, narrow down the date of the closing of the canon. We have seen that the
Vinaya includes an account of the events which led up to the second , but not to
the third , and it must therefore have been closed at some time between the two.
The Abhidhamma- includes the Kathāvatthu, and cannot therefore have been
closed until the time of the third in the reign of Aśoka. The canon was in all
probability closed some time before the time of Buddhaghosa because stories and texts
and verses from North India did not find their way into the canon.
I suggested in the fifth lecture that the writing down of the Theravādin texts prevented
any further changes in the language of the texts, and it is highly likely that the same event
also prevented any changes in the contents of the canon, with the exception of additions
to those texts whose structure permitted them. Such additions were finally stopped by the
comments made upon them by the commentaries, mainly in the fifth and sixth centuries,
although we must note that these commentaries were mainly based upon earlier material,
and so in effect the prevention of additions dated from an earlier period. A text like the
Therīgāthā-Apadāna, having no commentary, was only partially limited by the quotations
from it in other texts, especially the Therigatha- .
If these guesses are correct, then the canonicity of the Theravādin texts was effected
some time after the writing down of the canon in the first century B.C.E.—perhaps in the
first century C.E.—and we should perhaps see writing and canonicity together as two
66 The author of Pj II (477, 13–14) notes that there was no old commentary upon Sn 677–78:
avasane eva pana Maha- n’ atthi. See Norman,
1978, 28–47 (34) (=CP II, 30–51 [36–37]).
parts of an attempt by the Theravādins of the Mahāvihāra to legitimise their claims to be

the true recorders of the Buddhist tradition. It seems clear that the Theravādin school
have used the canon for their own purposes. Their aim was doubtless to make it clear that
their texts, and their texts only, represented the true Buddhavacana.67
There is, however, an interesting post-script to this discussion of canonicity. As we
have seen, the Buddha allowed for texts being added to the collection of suttas,
presumably after his death, and Buddhaghosa accepted that not all texts had been recited
at the first . Although it is clear that the sources which Buddhaghosa was
following had defined exactly the texts which were, according to their belief, included in
the , it is not certain whether Buddhists at a later time thought that their canon
was closed, and that nothing beyond what had been listed could be regarded as canonical.
The fact that the Buddha had given criteria for determining whether a newly found sutta
was authentic or not, meant that it was always possible for another text to be brought
forward and added to the body of material. If something looks as though it is
Buddhavacana, i.e. it follows the pattern of earlier accepted texts, contains the typical
formulae, includes a reference to the Buddha preaching, and has in it nothing which is at
variance with the sutta and the vinaya, then why should this not be Buddha-vacana, in
the broadest of the senses I have mentioned?
It is probable that by medieval times, once the Mahāvihāra had triumphed over the
Abhayagirivihāra, the need to maintain the closed nature of the canon became less
important, and the distinction between canonical and non-canonical became blurred. This
had already received an impetus from the fact which I have already mentioned, that some
of the best known stories about the Buddha were in non-canonical texts. It is certainly the
case that canonical and non-canonical texts were often written in the same manuscripts,68
and preserved by the same tradition, in the same place, at the same time.69
Consequently we find individual Jātaka stories and collections of such stories in south-
east Asia which are not identical with the collection of stories in the Theravādin canon,
but are clearly modelled upon that collection in their formal structure.70 They are
sometimes called “Apocryphal”.71 There are also individual texts found in the same area.
By title they are suttas; they have the standard canonical opening me
…; the narrative attributes their contents to the Buddha; they meet the
requirements of the Buddha’s criteria, in as much as they contain nothing which is not in
conformity with the Theravādin version of the Buddha’s teaching, but they are not
included in any edition of the Pāli canon. These are sometimes called “quasi-canonical”
or “allegedly non-canonical”.72
It is probable that the purpose behind such works was merely the desire to produce
teaching material, or sermons. Having composed something which was suitable for these
67 See Collins, 1990, 89: “The Pāli canon…should be seen as a product of [the Theravāda] school,
as part of a strategy of legitimisation by the monks of the Mahāvihāra lineage in Sri Lanka in the
early centuries of the first millennium A.D.”68 As Collins (1990, 116, note 55) reminds us.
69 Collins, 1990, 111.
70 Collins, in the Preface to Horner and Jaini, 1985, x.
71 See Horner and Jaini, 1985, and Jaini, 1986.
72 See Hallisey, 1993, 97–130.
purposes, based upon scriptural material, there perhaps seemed to be no reason why the
usual scriptural trappings—the opening formula and standard phraseology, etc., should
not be added. There was no intention to deceive.
I know of no evidence that would make me think that those who composed such works
were deliberately forging material which they hoped to pass off as Buddhavacana. There
is nothing which would make such a thing desirable. The fact that the newly found sutta
had to be compatible with the sutta and the vinaya would mean that the possibility of
anyone forging a document to support a new heresy he had just thought up was ruled out.
It was not like the situation in Chinese Turkestan at the turn of the century when the
presence of European
explorers searching for ancient documents meant that there was a market for forged
documents. Nor is it like the Chinese translation of the Jātaka-mālā, which is not so much
forged (as its description as a “pseudo-translation”73 implies), as made by incompetent
scholars.
I have already mentioned the fact that one of the canonical texts, the Niddesa, is a
commentary upon parts of the Suttanipāta, made—it would seem—so early that it could
be accepted as part of the canon. I will deal with this at greater length in the ninth lecture.
73 Brough, 1964, 27–53.

IX
Buddhism and the Commentarial Tradition
Commentaries were probably needed from the very beginning of Buddhism. We can
assume that the earliest form of commentary was simply the explanation of one word by
another, and we need not doubt that the Buddha on occasion replaced words by
synonyms, to make his teachings more easily understood. We find some sort of evidence
of this in the words of the Buddha himself, where he gives, for example, seven different
words all meaning “bowl”,1 which he says are “country language” (janapadanirutti). This
is presumably something which he found from experience, i.e. as he moved from one
place to another, he found that the word used for a particular object in one area was quite
different from that used in another area for one and the same thing. Anyone who has ever
looked at Sir George Grierson’s Bihar Peasant Life,2 will know exactly the problems
which the Buddha faced. Grierson moved around the area for which he had responsibility
as a civil servant, recording the wide range of different names which the villagers of
Bihar used for all the items of everyday domestic and village life. The Buddha must have
found a situation which was very similar to this 21/2 millennia ago.
The earliest extensive portion of Theravādin commentary material we have is that
found in the Vinaya- . We know that, at the suggestion of King Seniya Bimbisāra
of Magadha,3 the Buddha instituted a fortnightly recitation of the Pātimokkha (following
the practice of other sects which recited texts). We are therefore certain that the
Pātimokkha existed from the very early days of Buddhism. The Pātimokkha does not
have a separate existence in the Theravādin tradition, but is embedded in the Vinaya-
. If it is the earliest text for which we have direct evidence of recitation, it is not
surprising if it was the first text that needed explanation. And so we find a simple type of
commentary upon it, whereby words are for the most part explained by synonyms or
clarification of terms, e.g. the first word of the first pārājika rule is yo “whoever”, and it
is explained: “‘Whoever’ means he who on account of his relations, his social standing,
his name, his clan, his morals, his dwelling, his field of activity, an elder, or a novice, or
1 M III 234, 34 foll.

2 Grierson, 1926.
3 See Vin I 101, 18 foll.
one of middle standing, this is ‘whoever’”.4 The nature of that commentary is made clear
by its name: pada-bhājaniya “analysis of words”. The commentary is so early that, like
the Pātimokkha itself, it is embedded in the Vinaya- after each of the Pātimokkha
rules, and therefore the text and the commentary have been handed down together.
The earliest Theravādin commentarial text is one which is actually given separate
canonical status. It is the Niddesa, a commentary upon two vaggas and one sutta of the
Sutta-nipāta. Here again we find a great deal of explanation by means of synonyms. We
also find another characteristic feature of this type of commentary. Whenever a particular
word occurs in the text, the same explanation is given verbatim, e.g. whenever there is a
reference to kappa “figment”, or a verb based upon the root kapp-, we are told that there
are two types of kappa, i.e. -kappa and -kappa, and the identical explanation
is given, even if the words recur in successive verses. In the third lecture, I pointed out
that this kind of fixed repetition is typical of the oral nature of an early text, giving
support, if it were needed, for the assumption that the Niddesa is very old and was
composed well before the writing down of the canon.
Even the early commentarial passages in the canon itself are not merely lists of
synonyms. Commentaries have two functions, although not all commentaries perform
both of them all the time: they explain the meaning of the words and they explain the
meaning of the phrase or sentence in which the words occur. In the
“The discourse on the undefiled” of the Majjhima-nikāya,5 the
“undefiled person” is defined in one paragraph by the Buddha. The remainder of
the sutta is devoted to explaining the meaning of each sentence of that paragraph.
Similarly, the Niddesa is not merely a list of synonyms. It also includes exegetical
passages, giving us some idea of what the early commentarial tradition thought was the
meaning of the relevant portions of the Sutta-nipāta. It is, however, not a organic
structure of exegesis, but a series of disconnected phrases, which serve as explanations of
the individual words, not in the particular context of the Sutta-nipāta, but in any setting.
The chronicles tell us that, at the time of Aśoka, Mahinda took Buddhism to Sri
Lanka.6 The story, as related in the , the Samantapāsādikā, and the
, tells how King Tissa met Mahinda, exchanged greetings with him and had
a discussion. All three sources agree that the first sermon which Mahinda preached to the
king was the “The short sermon on the elephant’s
footprint”.7 The 8
states that Mahinda preached the true doctrine in two
places “in the speech of the island” (dvīsu bhāsitvā dīpabhāsāya),
implying that in the other places he did not “translate”.9 The
4 Vin III 23, 37 foll.

5 M III 230–37.
6 Dīp 12.9; Mhv 13.21.
7—M I 175–84.
8 Mhv 14.65.
9 I use “translate” to mean the change from one dialect of Middle Indo-Aryan to another, however
closely related.
Buddhism and the commentarial tradition 129
and the Samantapāsādikā make no mention of this “translation” process, and

although there are instances of the including authentic historical material
which was either unknown to, or consciously omitted by, earlier chroniclers, it is possible
that the need for “translation” was deduced by the author of the from the
statement repeated at the beginning of Buddhaghosa’s commentaries, that they were
translations into Pāli of the commentaries which were brought to Sri Lanka by Mahinda
and put into Sinhalese for the benefit of the islanders.10 The author of the
may well have thought that if Mahinda needed to “translate” the commentaries then he
must have “translated” the canonical sermons too. The fact that the “translation” process
is said to have happened in only two places suggests, however, that the statement is based
upon tradition. If it is true, we may deduce that in some places Mahinda preached the
suttas in their canonical form, whereas in two places he added an exegesis, in Sinhalese
Prakrit.
There is no evidence about the language of the commentaries which Mahinda is said
to have brought with him to Sri Lanka. They could have been in the same language as the
canonical texts, whatever that was, or they might have been in the language of the
missionaries, in so far as they may have spoken a different Prakrit. We have no evidence
earlier than Buddhaghosa’s statement which I have just mentioned,11 that Mahinda did
bring commentaries with him, but we have no reason to doubt that commentaries were
brought from India either by Mahinda or other missionaries coming after him. The
commentaries seem to have contained much information about India, and particularly
North India, which could only have been brought from there. As we shall see, the Pāli
commentaries sometimes employ a form of words in their explanations which is closer to
the versions found in Sanskrit or the Gāndhārī Prakrit than the Pāli version which is being
commented upon,12 and this can only be explained by assuming that the Pāli commentary
and the Sanskrit and Gāndhārī versions go back to a common source, which must have
been in North India.
There is evidence for dialect differences in the early commentarial material. We know
from the Aśokan inscriptions that in the third century B.C.E. there was a dialect variation
in the treatment of consonant groups. In the East the dialects tended to resolve consonant
groups by inserting an epenthetic (svarabhakti) vowel. In the West such consonant
groups were simplified by assimilation. We occasionally find that a word is found in the
Pāli canon which shows one regional variation, while the explanation of the word which
is preserved in the commentaries shows a cognate form from another dialect, e.g. we find
in the Dhammapada13 the phrase samādāya, usually translated
“having adopted the whole law”, assuming that vissa is from Sanskrit viśva “all”.14 The
commentary, however, explains this as “vissa means taking up a dhamma that is uneven
10 pana ābhatā ’tha vasinā Mahā-Mahindena

dīpavāsīnam atthāya (Ps I 1, 23*–24* and Lamotte, 1988, 557).
11 Ps I 1, 21* foll.
12 See Brough, 1962, 226.
13 Dhp 266.
14 It is so taken by PED, s.v. vissa 1.
(visama) or a dhamma that is foul-smelling, concerned with physical activities, ete.”.15

The alternative explanation signified by “or” (vā) shows that the commentator was
uncertain about the meaning of the word, but nevertheless the tradition of the Sinhalese
he was following had preserved an alternative dialect form.
One Sanskrit version of the verse has veśma-dharma16 and it is probable that the
Gāndhārī17 version also has veśma-. The meaning therefore is “domestic manner (of
life)”.18 In the Pāli version vissa stands for vĕssa, where -ss- results from the assimilation
of -śm->-ss-. The first explanation in the commentary shows the alternative development
of an epenthetic vowel being inserted, i.e. veśma>*viśma>*visma>visama, which the
commentator took to be the equivalent of Sanskrit “uneven”. Another Sanskrit
19
version of the same verse actually has , which is clearly a backformation from
the same tradition which preserved the Pāli commentarial reading visama. We may
presume that the verse was uttered at least twice, once in an area where consonant groups
were assimilated, and once where they were resolved. These explanations in different
dialects must therefore antedate the introduction of the canon into Sri Lanka, and they
show that the dialect variation already existed at an early date, perhaps at the time of the
Buddha.
Other explanations which depend upon non-Pāli and non-Sinhalese Prakrit forms must
20
also antedate the introduction of the canon into Sri Lanka, e.g. the explanation of
“top-knot” by cora “thief” must depend upon a source which, like Māgadhī, confused l (
) and r. Similarly, abhibhāsana21 “illuminating” is explained as tosana “delighting”,22 but
this is really the explanation of abhihāsana,23 and shows that the explanation must have
been made in a dialect where both abhibhāsana and abhihāsana appeared as abhihāsana.
The Critical Pāli Dictionary24 refers to nominative singular forms in -e in two
commentarial passages25 and claims that these are from the < -> ,
because Proto-Sinhalese, like Māgadhī, was an -e dialect. These two passages are
virtually identical, and it seems unlikely that, of all the hundreds of Sinhalese Prakrit
forms in -e which must have been in the Sinhalese commentaries, Buddhaghosa not only
forgot to “translate” into Pāli the first time he gave the explanation, but also repeated his
error a second time. We may therefore deduce that Buddhaghosa deliberately quoted the
passage with the -e forms (attributing it to the the second time), perhaps
15 vissan ti, vissa- vā kaya- samādāya, Dhp-a III
393, 2–4.
16 Udāna-v 32.18,
17 GDhp 67.
18 Brough, 1962, 266.
19 Mvu III 422, 13*.
20 Th 170.
21 Th 613.
22 abhibhāsanan ti -hetutaya cittassādhippamodanato, Th-a II 260, 2–3.
23 abhihāsana<*abhihassana< .
24 CPD, s.v. avitakka.
25 ese eke same samabhāge tajjāte tañ ñevā ti, Mp I 71, 13; ese eke same sabhāge
tajjāte tañ ñevā ti, II 273, 16–17.
because it was invested with such authority in the Sinhalese commentary that he felt
obliged to quote it verbatim. We may note that something very similar to the passage
occurs in the Kathāvatthu,26 and there is evidence27 for believing that it was a common
phrase in the early Buddhist tradition in Magadha. I conclude, therefore, that these forms
are not remnants of the - .
Childers was of the opinion28 that the commentaries brought by Mahinda were in Pāli.
If Childers was right, we have to assume a progression from Pāli to Sinhalese Prakrit and
back to Pāli again. While this is not impossible, it does raise the question of why the Pāli
of the canon was sacrosanct, while the Pāli of the commentary could be translated. It also
raises the question of the nature of Pāli. If it means the language of the Theravādin canon,
then it is unlikely that at the time of Aśoka, when Mahinda came to Sri Lanka, the
language of that canon was already a western Prakrit with a certain amount of
Sanskritisation introduced into it. I think that the Sanskritisation came later, perhaps
because of the rise of the Mahāyāna. It would seem more likely that the commentaries
already represented a heterogeneous mass of material, in various dialects, and probably
including comments on readings which differed from those in the canon as established in
Sri Lanka.
In the fifth lecture, on Buddhism and writing, I quoted the passage in the
which stated that prior to the reign of Abhaya the
and the were transmitted orally, but in his reign they were written down. We
tend to concentrate on the fact that it was the which was written down, without
thinking much about the commentaries because, to us, the word “commentaries” implies
the works composed, in particular, by Buddhaghosa and Dhammapāla in the fifth and
sixth centuries C.E. respectively. The reference in the must, however, be to
whatever commentarial material was available in the first century B.C.E., and which had
hitherto been transmitted orally in conjunction with the canonical texts to which it
referred.
The fact that the commentarial material was already of a disparate nature would
probably have led to an attempt to impose homogeneity upon it, and also to make it more
intelligible to the Sinhalese bhikkhus by translating it into the vernacular language.
Because it was in the vernacular it would have been easy for additions to be made to it.
That there was a continuing commentarial tradition in Sri Lanka itself is shown by the
fact that Buddhaghosa quotes (from the Sinhalese commentaries he was using) the names
of individual Sinhalese theras whose views he is accepting or rejecting.
According to the , there were no commentaries available in India in the
fifth century C.E., and so the thera Revata suggested to Buddhaghosa that he went to Sri
Lanka and translated the commentaries into Māgadhī.29 At the beginning of the
26 ese se same samabhāge tajjāte ti, Kv 26, 20–21 et passim.

27 See Norman, 1979A, 279–87 (281) (=CP II 59–70 [61]).
28 Childers, 1875, x.
29 Mhv 37.230.
Samantapāsādikā, Buddhaghosa states that his work will be based on the Maha-
and the Mahāpaccariya, while also taking into account such commentaries as
the Kurundi. He also quotes from the Andhaka and the , and the Paccarī, but it
is not clear whether this last named is the same as the Mahāpaccariya.
There has been much speculation about the Andhaka- . It is said30 that it
31
was handed down at Kāñcipura in South India. It is further stated that it was very likely
written in the Andhaka language. Probability then becomes certainty, and we than find
statements such as “the references relate to…the Sinhalese and some of the Dravidian
commentaries”,32 and “Buddhaghosa drew his material not only from Sinhalese and
Dravidian but also from…Pāli”.33 Comparable statements are made about Dhammapāla:
“Dhammapāla…very probably was a Dravidian by birth. It is also likely that he made use
of Dravidian commentaries”.34
There was, in all probability, some connection between the Andhaka commentary and
either the Andhaka country or the Andhaka sect, and it is very likely that the sect was so
called because it came from the country. Probably the Andhaka commentary came from
the Andhaka country originally, for Buddhaghosa quotes it35 as referring to conditions in
that country. It must, however, like the other commentaries, have had a core of material
which came from North India, for Buddhaghosa quotes the Andhaka- when
36
he is talking about the Magadha- “a particular measure” but quotes the Mahā-
for the - in the same passage. The fact that the Andhaka
commentary is usually quoted only to be refuted37 tends to support the view that the
basis, at least, of the Andhaka commentary belonged to the Andhaka sect.
About the language of the Andhaka commentary nothing can be said definitely, but it
would seem clear that by Buddaghosa’s time it was no longer available in South India.
The statement that no commentary was available in India,38 which is given as the reason
for Buddhaghosa to go to Sri Lanka, was not likely to have been included in the
Sinhalese commentaries upon which the was based while there were South
Indians such as Dhammapāla available to refute it, if it was false. It is in any case clear,
from Buddhaghosa’s statement,39 that the view of the thera Mahāsumma was regarded as
authoritative in the Andhaka- , that the commentary must have been
introduced to Sri Lanka some time before Buddhaghosa, for Mahāsumma is a Sinhalese
thera who is datable to the first century C.E. Since, however, Buddhaghosa stated plainly
that he translated Sinhalese commentaries into Māgadhī, we can be fairly certain that if
the Andhaka commentary was originally composed in a Dravidian language, it had
already been translated into Sinhalese Prakrit by Buddhaghosa’s time.
30 C.A.F.Rhys Davids, 1974, xxviii.

31 Adikaram, 1953, 12.
32 Adikaram, 1953, 14.
33 Adikaram, 1953, 16.
34 Adikaram, 1953, 9.
35 Sp 747, 23.
36 Sp 702, 23–27.
37 e.g. Sp 697, 1.
38 Mhv 37.227.
39 Sp 646, 11.
It is usually stated40 that no material was added to the Sinhalese commentaries after
the first century C.E., but a very careful study41 of the individuals mentioned in the
commentaries has shown that King Mahāsena is mentioned by name in the
Samantapāsādikā.42 Since this king is datable to 276–303 C.E., this shows that additions
to the - continued to be made until the very end of the third century
or the beginning of the fourth,
The author of the commentary upon the Sutta-nipāta notes that the Sinhalese
commentary did not comment upon two verses in the Sutta-nipāta,43 and he therefore
excludes those verses from the number of verses in the original sutta.44 This statement
has been taken45 as pointing with more or less certainty to an addition made to the canon
in Sri Lanka. The verses could equally well, however, have belonged to a different
recension from that being used by the Maha- , and been either unknown to
the author of the commentary or ignored as being less authentic. It is clear that
different recensions of some texts did find their way to Sri Lanka. Buddhaghosa and the
other commentators frequently record the existence of variant readings, and it is
questionable whether these would all have been in one and the same version of the canon.
It is interesting that the commentator on the Sutta-nipāta did not invent a commentary
upon these verses. It would seem that the commentary was, as far as he was concerned,
closed.
If the Buddha himself explained X by Y in one place, but Y by X in another, then we
can see that X and Y were both Buddhavacana, and this situation might lead to a
different sort of commentarial tradition, whereby the commentarial explanation in one
tradition was the canonical reading in another tradition, and vice versa. I have just
mentioned the comparable situation with regard to vissa, veśma and visama.
It has been shown46 that in a number of cases other (non-Pāli) traditions preserve as
canonical forms which are akin to those in the Pāli commentary, showing that those
traditions thought that they were Buddhavacana. For example, where a verse in the
Dhammapada47 includes the words adhisessati, chuddho, and apeta- , the
Gāndhārī Dharmapada equivalent48 has three different words in their place (vari sa’
, tuchu, and avakada- ). The glosses in the Pāli commentary, however, are the
equivalents of the Gāndhārī forms (upari sayissati, tuccha, and apagata- , Dhp-a
I 320–21), which clearly shows acquaintance with the same tradition as underlies the
Gāndhārī version. A possible explanation of this is that the Buddha uttered the same
sermon on more than one occasion, sometimes making changes, as the occasion
40 e.g. by Adikaram, 1953.

41 Mori, 1988, 119–67 (143).
42 Sp 519, 26.
43 Sn 677–78.
44 avasāne eva pana Mahā- n’ atthi,
tenāvocumha “vīsatigāthāsū” ti, Pj II 477, 13–14. This seems to be the only statement of this kind
in the cty.
45 Adikaram, 1953, 12.
46 Brough, 1962, 192.
47 Dhp 41.
48 GDhp 153.
or locality demanded. When tradition preserved more than one version, it sometimes kept
one as an explanation of the other.
Another example of this parallelism between canonical and commentarial traditions,
which seems to cross the boundaries of sect, can be seen in Buddhaghosa’s commentary
on the incident where bhikkhus ask for permission to change the Buddha’s words
chandaso. He glosses: “Let us translate chandaso means, let us translate into Sanskrit (or
into the refined language) as (we translate) the Veda”.49 Here “into the refined language”
(sakkata-bhāsāya) seems to correspond to phrases in two of the Chinese versions of the
same incident. The Dharmaguptaka version includes the request in the form “in
accordance with the fine language (or perhaps texts) of the world”, while in the Vinaya-
version the Buddha states that his doctrine “is not concerned with beautiful
language”. It seems very likely that both these canonical phrases are based upon originals
which included the equivalent of or . 50
It is an interesting fact that in his commentary on the heretics’ views which are quoted
in the Sāmaññaphala-sutta,51 Buddhaghosa seems to include phrases which are akin to
those in the Tibetan version of the Pravrajyā-vastu.52 He glosses: “cooking means beating
others with a stick”,53 which seems to be a combination of two phrases in the Tibetan:
‘Who grills and lets grill, who beats and lets beat’.54 At the end of his explanation of
Ajita’s doctrines, he comments: “fools give, the wise take”.55 The Tibetan version states:
‘Thus (only) the fool accords instruction, the sage receives instruction’.56
These parallels suggest that the commentarial and canonical traditions which underly
them are equally old, and the material upon which they are based must pre-date the
separation of the sects. It seems probable that some of this material, since it is accepted as
canonical by some sects, must go back to the earliest days of Buddhism, perhaps to the
time of the Buddha himself. A close comparison of the Theravādin commentaries with
non-Theravādin canonical texts might well bring to light other parallels of this nature.
This might help us to date the various strata of the commentarial tradition, which would,
in turn, enable us to estimate its value more accurately.
Sometimes there is mention of an event in a commentary, with no reference at all to it
in the canon, although it appears in the canon of another school. As I mentioned in the
eighth lecture, a problem arises from the fact that some of the best known stories in
Buddhism are known in Pāli only in the commentaries, e.g. the story of the
maiden Kisāgotamī.57
In some cases, portions of Buddhist history seem not to have reached Sri Lanka at all,
e.g. the early history of Devadatta’s hostility to the Buddha, and the reason why the
49 chandaso āropemā ti, viya sakkata-bhāsāya āropema (Sp I

214, 16–17).
50 See Brough, 1980, 35–42 (39).
51 D I 52 foll.
52 Vogel, 1970.
53 pacato ti, pare (Sv 159, 16–17).
54 Vogel, 1970, 26.
55 bālā denti, , ti dasseti (Sv 166, 16).
56 Vogel, 1970, 22.
57 See Collins, 1990, 89–126 (117, note 55).
Buddha called him “lick-spittle”,58 .59 Buddhaghosa does not know the reason,
60
and the incorrect explanation he gives is followed by modern translators and
commentators, so that, for example, Miss Horner translates it “to be vomitted like
spittle”. Some of the Chinese versions of the Vinaya, however, do know the correct
meaning of , and tell a story, perhaps invented to explain the word, of
Devadatta eating Ajātaśatru’s spittle.
Sometimes the commentators give explanations based upon individual syllables of
words, e.g. bhikkhu: ikkhanatāya, “because of seeing danger in
” Vism 16, 21; bhagavat vanta-gamano bhavesu, “he has rejected going in
existences” Pj I 107, 26*; King Pasenadi is so called because:
jināti, “he conquers the hostile army of another” Ud-a 104,27. Since the
Sanskrit form of his name was Prasenajit [Pkt ], the commentarial explanation is
correct to include “he conquers”, and this makes it clear that when this explanation was
composed (which might have been long before Dhammapāla) the name still had the
spelling with -j-, or the tradition at least knew that the correct form was -j-.61
The fact that the commentary explains the syllable -di by jināti, shows that the text
and the cty were transmitted separately,62 with neither having an effect upon the other.
The reason for the -d- of the text form is not clear. It is claimed by some as a
Sinhalesism, because j>d in Sinhalese. If this were the explanation, I should have
expected the commentary to show the change, not the text. The disagreement between the
canonical text and the commentary is probably due to the fact that they were in the
keeping of different groups, i.e. the responsible for the canonical text were not
also responsible for the safe keeping of the commentary upon that text.
Other explanations are given in accordance with the type of etymologising in Sanskrit
literature known as nirukti or nirvacana, which was taken over into Buddhist literature. It
often resembles folk etymology in its mode of operation, but this is to misunderstand its
purpose. Quite frequently words are explained by means of others which are similar in
appearance but are, in fact, in no way related to them. Such etymologies are found in the
canonical texts from an early period, and are then taken over into the commentarial
tradition, e.g. , based upon a root brah- meaning “to be strong”, is explained
in a Dhammapada verse by means of a different root brah- “to remove”.63 A
was a remover—of his evil deeds, of course. Similarly a is so called because he
has put to rest his evil deeds, explaining sam- (Skt śram- “to strive”) by sam(Skt śam- “to
be quiet”).64 The word rājā “king” is explained in the Dīgha-nikāya by “he pleases others
58 At Vin II 188, 37 and Dhp-a I 140, 1. See Lamotte, 1970, 107–15.

59 See Norman, 1979B, 325–31 (329, note 23) (=CP II, 187–93 [191, note 2]).
60 Sp 1275, 17.
61 has a rule to deal with this: Sadd sūtra 88 (622, 24): do jassa: Pasenadi.
62 For similar separate transmission of text and commentary, see von Hinüber, 1981, 74–86 (78).
63 bāhita-pāpo ti , Dhp 388.
64 samitattā hi ti pavuccati, Dhp 265.
by righteousness”, explaining rāj- “to rule” by means of rañj- “to be delighted”.65 The
fact that a comparable etymology is given, or alluded to, several times in the
66
Mahābhārata and shows that this is not a Buddhist invention.
Some of these explanations seem ludicrous in western eyes, and this has led to some
modern commentators trying to see the joking nature of the Buddha at work in them. It
has been claimed that the Buddha is here parodying the etymologies (nirukti) found in
brahmanical texts.67 I cannot rule out the possibility of the Buddha joking, although it is
noteworthy that joking is not a well-known or highly approved Buddhist pastime, but I
can see no evidence that these etymologies were really intended as jokes. Certainly in a
series of etymologies of this type in the Sabhiya-sutta of the Sutta-nipāta, dealing with
the words kusala, , muni, etc., no joke seems to be intended, and I find it hard to
understand why we should think that some etymologies are meant to be humorous, and
others not.
I am reminded of talking to various people about Sanskrit, and being asked by one if
Sans-script was so called because it had no script and could not be written down, and by
another if Sand-script was so called because the early Indians had no writing materials
and had to scratch the characters in the sand. These folketymologies seemed comical to
me, but both questioners were very serious about them.
These nirvacana etymologies are sometimes of value (beside the etymological
theorising they contain) because they give information about the dialect(s) in which they
were first given, and such information about the earlier history of the text may not be
available in other ways.68 For example, to explain by using the form bāhita
shows that in the dialect in which the etymology was invented the word had
the form , while some of the etymologies in the Sabhiya-sutta make sense only
if we assume that they were first composed in a dialect in which -t- and -j- became -y-.
The fact that, despite the displacement of pādas, etc., which makes the etymologies in
the Sabhiya-sutta very hard to understand, the commentary still contains the correct
traditional explanation of some of the words included in the sutta, e.g. kusala and
ājāniya, shows that the sutta tradition, which is very corrupt, and the commentary
tradition, were quite separate, as in the case of the name Pasenadi which I have just
mentioned, and the commentary went on transmitting the correct explanations even
though the text which was being commented upon no longer had the correct readings.69
The growth of homonyms in MIA often meant that two or more explanations were
possible for a word, depending upon which etymology was being considered. This
sometimes enables a Pāli commentary to give a richer and fuller exegesis than would be
possible in Sanskrit, where a redactor had to decided upon one form or another. The BHS
equivalent of sammappadhāna “right effort” is “right abandoning”. I
mentioned in the sixth lecture the suggestion70 that the Sanskrit version is an incorrect
backformation from an MIA form sammappahāna which might have either meaning.
65 dhammena pare rañjeti, D III 93, 14. See Gombrich, 1992, 159–78 (174).
66 See Hara, 1969, 489–99.
67 See Gombrich, 1992, 174.
68 See Norman, 1980B, 173–84 (177–78) (=CP II, 148–64 [154–55]).
69 See Norman, 1980B, 173–84 (178)=(CP II, 148–64 [155–56]).
70 See Gethin, 1992, 70.
This view is supported by the fact that Buddhaghosa’s explanation of the term
actually includes the idea of abandoning,71 and this suggests that the commentarial
material was first composed in a dialect where the form of the word was sammappahāna.
Where the commentators no longer knew the meanings of words, they sometimes had
to deduce them by devising etymologies which we can see are incorrect. In Pāli the
consonant cluster in the word -jña “knowing” becomes -ññ-. The consonant cluster -ny-
also becomes -ññ-. When dealing with the word vadaññu, which is derived from Skt
vadānya “bountiful, liberal, a munificent giver” or “eloquent, speaking kindly or
agreeably, affable”,72 the commentarial tradition wrongly took the final element -ññu to
be from -jña, and the word is explained as vacana-vidu,73 i.e. understanding vadaññu as
the equivalent of vada-jña “knowing the utterance (of the Buddha)”.
Another development from Sanskrit -jña is Pāli -jina, with an epenthetic (svarabhakti)
vowel evolved between -j- and -ñ-. The resultant form is identical with the word jina
“conqueror”, and the commentaries consequently explain compounds ending in -jina as
“conqueror”, e.g. magga-jina74 “conqueror (of the defilements) by means of the road”,75
whereas we are probably to see its meaning as “knowing the road”. The same type of
explanation is given for khettajina “conquering the field”. Here we can see that -jina is
certainly from -jña, because there is a Sanskrit equivalent -jña “knowing the
field”.76
We are inclined, in western philology, to believe that there is only one correct answer
to a question of etymology. In India, however, there was a custom of seeing more than
one meaning in any word or phrase—the so-called . So, instead of saying the
meaning is either this or that, as we would do, commentators very often say that the
meaning is this and that. Sometimes the meanings they give include what we would
regard as the correet etymology, but sometimes they are all, from our point of view,
incorrect. These are not, however, intended as western-style etymologies, but they have
rather a religious, or even mystical, purpose.
So the word bhagavat is explained with the words: “The reverend one (guru) has
blessings (bhagī), is a frequenter (bhajī), a partaker (bhāgī), a possessor of what has been
analysed (vibhattavā), has caused abolishing (akasi ), is fortunate (bhāgyavā),
has well maintained himself in being (subhāvitattano) in many ways, has gone to the end
of being (bhav-anta-go), thus he is called bhagavat”.77 These multiple explanations
would not be regarded in the tradition as specifying the meaning of the word bhagavat,
but as attributes or epithets of the bhagavat himself, and would form the basis of a sort of
litany “the bhagavat has blessings, etc.”, i.e. they would be taken as facts, not as
explanations of the term bhagavat.
71 Gethin, 1992, 71.

72 See MW, s.v. vadānya.
73 P II 415 (ad Sn 487).
74 Sn 84.
75 Pj II 162: maggajino ti maggena sabbakilese vijitāvī.
76 See Dhadphale, 1980, 205, and Norman, 1980B, 173–84 (183, note 6) (=CP II, 148–61 [149,
note 3]).
77 Pj I 107, 19 foll; translated at MRI, 116.
Similarly with arahat, which is explained: “One is an arahat because one’s defilement
are far from one (āra-kilesa)”.78 “The bhagavat is arahanta because of remoteness
(āraka), because of the destruction of his enemies (ari) and of the spokes (ara=rāga,
etc.), because of his being worthy of requisites, and because of the absence of secret ill-
doing ( rahābhāvā).79 The inclusion of an explanation based on ari is
interesting because the Jain form is ari-hant, i.e. with a svarabhakti vowel -i- instead of -
a-.
A number of non-Buddhist terms were taken over and applied in a Buddhist sense, e.g.
the Buddha is called a and a nāga. Since it was thought inappropriate to call the
Buddha by the name of a minor supernormal being, i.e. a“snake-god”, other explanations
were invented. One is a nāga because one does not perform sin (na+āgu), or one will not
go to rebirth again (na+gam “to go”), or will not come back again (na+āgam “to come”).
Sometimes the explanations which the commentators give include forms which
suggest that they were first given in some other dialect. So, among the explanations of
tathāgata, we find some based upon forms with gada instead of gata, i.e. they are taken
from, or through, a dialect which voiced -t- to -d-. Tathāgata is explained as: “He is thus
gone (tathā gata), thus come (tathā āgata), he has thus known (tathā ājānana [verbs
meaning ‘go’ also mean ‘know’]), has uttered what is real (tathā gadanato)”.80 We are
also told that just as a physician overcomes snake-bite by an antidote (agada), so the
tathāgata overcomes the world with the antidote of truth.81
Commentators not always well informed, and we can see that they or the sources they
were following were sometimes mistaken. For example, in the commentary on the Sutta-
nipāta,82 Ādiccabandhu is said to be the name of a pratyeka-buddha, presumably because
it occurs in the verses ascribed to the pratyeka-buddhas. The word is, however, used so
commonly as an epithet of the Buddha that I find it hard to believe that it does not apply
to him in this verse.83
Ignorance about rare grammatical forms also led commentators astray. The word
phalesin is a future active participle,84 and means “about to fruit”. It occurs in a verse in
the Theragāthā85 where the trees are described as being in blossom and about to fruit. The
commentator, unacquainted with the form, analyses phalesin as coming from phala
“fruit” and esin “seeking”, and describes the trees as “seeking their fruit”. The word
occurs again in a verse86 where a foolish action is likened to a man who cuts a tree down
just when it is about to fruit. Here the error in the translation “a man seeking fruit cuts a
tree down” is masked by the fact that this too would be a foolish action.
78 Ps I 42, 19, quoting M I 280.

79 ārakattā arānañ ca hatattā arahattā pāpa- rahā-bhāvā ti
imehi tāva so bhagavam arahan ti veditabbo, Sv 146, 10–12; cf. Vism 198, 11 foll. cf.
ārakattā arahā hoti, A IV 145, 2; hatattā arahatā (instr. sg.), Mp V 84, 22.
80 MRI 217 foll.; cf. Ud-a 131, 15.
81 See also Ud-a 132, 9 for agada/āgada.
82 Pj II 105, 27.
83 Sn 54.
84 See Geiger, 1994, § 193A.
85 dāni dumā bhadante phalesino vippahāya, Th 527ab.
86 yathā phalesī mūle chettu tam eva icchasi, Th 1121ab.
Although some of the commentarial material must go back to very early times in
Buddhism, perhaps to the time of the Buddha himself, nevertheless, as we know well
from other religions, all texts are interpreted in the light of the age and culture of the
reader, which may differ considerably from the age and culture of the original composer
of the text. We have, therefore, to decide how we should interpret a canonical text. Do we
interpret it in the light of our knowledge of the culture of the time, the culture of the first
commentary upon it, the culture of the second commentary on it, the culture of present-
day Buddhists in South or South-east Asia, or our own western culture?
An examination of the Niddesa and the explanations it gives enables us to see that the
composer of that text was already interpreting the ideas of the Sutta-nipāta in a different
way from the original hearers of the text, as far as we can judge.87 We also have a
commentary upon the Sutta-nipāta attributed to Buddhaghosa, and he too interprets the
Sutta-nipāta in the light of his own culture, or at least the culture of those who wrote the
material in the Mahāvihāra which he used in his work. Again there are differences of
interpretation, as the commentaries tried to interpret things in their own way, although, of
course, this does not necessarily mean that their interpretation differed from that of the
Buddha’s contemporary audiences.
The way in which interpretations changed can be seen easily from an examination of the
meaning of the phrase tasā vā thāvarā vā.88 If we base our translation on the meaning of
the related Sanskrit words, then Sanskrit trasa means “moving”=“the collective body of
moving or living beings” (as opposed to sthāvara),89 and Sanskrit sthāvara means
“standing still, not moving, fixed, stable, immovable”.90 The phrase, therefore, means
“(all creatures) moving or unmoving”. By the time the Pāli commentary was written,
however, the interpretation of the words had been changed. The commentator states:91
“tasa is a synonym for those who have cravings or fears: thāvara means that they are
standing still. This is a synonym for those who are rid of their cravings, i.e. arahats”. To
the commentator, therefore, the phrase had taken on a Buddhist flavour: “Ordinary
individuals who still had cravings, and arahats who were freed from them”. This explains
why Miss Horner translated the compound tasa-thāvara as “those who have craving and
those who have none”,92 although it is very likely that to the original hearers the meaning
was far less restricted, and more on the lines of the meaning in Sanskrit.
There are many other comparable examples of development of meaning. In the Uraga-
sutta of the Sutta-nipāta we find the refrain: “That monk leaves behind this shore and the
far shore as a snake leaves behind its old skin”.93 What does the phrase “this shore and
the far shore” mean in this context? In some Pāli texts “this shore” means “this existence”
87 See Burford, 1991, 9.

88 Sn 146.
89 See MW, s.v. trasa.
90 See MW, s.v. sthāvara.
91 tasā, sa- sa- c’ , ti thāvarā,
, Pj I 245 (ad Khp IX.4= Sn
146).
92 Horner, 1957, 290 (=M II 105).
93 so bhikkhu jahāti urago iva , Sn 1–17.
and “the far shore” means “nibbāna”. By the time the commentator wrote in the fifth century C.E.,
the idea of leaving behind the far shore in the form of nibbāna was a Mahāyāna idea, which as a
Theravādin he was very reluctant to accept. To understand the phrase therefore entails the
discussion of the question of whether the Mahāyāna idea could have been in existence at the time
when the sutta was composed, and, if not, what “far shore” could mean in this context. My own
belief is that the references is not to and to the far shore of , i.e. nibbāna, but
to this world and the next, and I believe that the verse was first formulated in a situation where the
author was considering this world and the afterlife, rather than the endless stream of .
When the Buddhists took the verse over, however, they had to make it fit into the
system. In an attempt to avoid the Mahāyāna idea, the commentary on the Sutta-nipāta gives seven
explanations,94 including the lower fetters and the higher fetters, and the world of men and the
world of gods.95 It is interesting to note that, in the Sanskrit Udāna-varga, where the editions of
Chakravarti and Nakatani read “near and far shore”,96 Bernhard’s edition reads
“this shore”, which suggests that the redactor of that version also thought that the
abandoning of pāra was unacceptable.
The Buddhaghosuppatti97 states that the commentaries given by Mahinda, and the
additions thereto, were burned in a great bonfire when Buddhaghosa had finished writing
his commentaries. Although this has long been recognised as being merely an
exaggerated way of saying that the Sinhalese commentaries fell into disuse, there is
evidence to show that the earlier commentaries were not, in fact, completely superseded.
It is clear that the commentaries which Buddhaghosa used to write the historical
introduction to the Samantapāsādikā were not destroyed, for they were available to the
author of the and even to the writer of the - some
centuries later.98 The same applies to the commentaries on canonical texts. The -
refers to the - on the Majjhima-nikāya and includes
information not given in the Papañcasūdanī.99 The on the Sāratthappakāsinī includes
a variant reading for a passage in the Majjhima-nikāya,100 which is not only not given by
Buddhaghosa,101 but which shows by its form ahunā (<adhunā)= idān’ eva that it was
taken from a dialect other than Sinhalese Prakrit and Pāli, probably a North Indian
Prakrit, and perhaps even from a commentary brought over by Mahinda. Nevertheless, at
some unknown date the Sinhalese must have fallen completely into disuse,
and they seem now to be irretrievably lost. Only the Pāli commentaries which were based
upon them are now extant.
In time, of course, the commentaries themselves became less intelligible, and required
94 Pj II 12–14, discussed by Brough, 1962, 201 foll.

95 manussaloko devaloko, Pj II 13, 9.
96 Udāna-v 18.21, etc.
97 Bu-up 7.
98 Malalasekera (1935, cix) dates Mhv- to the eighth or ninth century.
99 Mhv- 193, 305.
100 M I 255, 16 (which is quoted at Spk I 13, 29).
101 Both PTS editions read ahu tañ ñeva aññatarassa. Spk- (Be 1961) 41, 25 reads ahunā idān’
eva. (Information received from L.S.Cousins in a letter of 25 December 1973).
explanation. This led to the appearance of , which were in effect commentaries

upon the . Sometimes, as I have just said, these commentaries, although
late, nevertheless preserved readings and explanations which are superior to anything
found in earlier commentaries. In time came a further expansion of commentarial
material, with , and other commentaries being written, and the production of
such works has continued right up until the present time, as Buddhist commentators have
used all the facilities of Indian philology to explain the meaning of the Buddhavacana, as
it has been re-interpreted again and again in the light of ever-changing cultural patterns.
How are we to evaluate the commentarial tradition? It is clear that sometimes it is very
useful:
(1) The commentaries sometimes explain something we could not otherwise
understand. Mrs Rhys Davids, for example, paid tribute to the indispensibility of the
commentary on the Therīgāthā,102 when she was translating that text. Despite all its
errors, it nevertheless gives correct explanations of many passages which would
otherwise be quite unintelligible. The same is probably true of almost all commentaries.
(2) Commentaries sometimes contain readings which are better than those in the
canonical texts we possess. So in the descriptions of famine which occur in the Vinaya
and the -nikaya, there is a word dvīhitikā, which is explained as being made
up from du “two” or “difficult”+īhita “effort” or “activity”, i.e. the word means either
“there is two-fold effort, i.e. begging will be either successful or unsuccessful” or “life
will be difficult”. The commentary upon the -nikaya, however, gives a variant
reading duhitikā, which it says has the same meaning as dvīhitikā. Nevertheless, if duhita
is taken as the opposite of suhita “satiated”, then we can see that *duhita would mean
“hungry”, and duhitika “stricken by hunger, or connected with hunger”.103
(3) As I have stated, commentaries show us how Buddhist thought has developed
since the time when the canonical text they are commenting on was composed.
They were not always quite so useful:
(1) Sometimes the tradition has lost the meaning, and the commentator resorts to
giving several explanations, all unlikely, and all very difficult to understand. This
produces the familiar cry: “I can understand the text, but not the commentary”.
(2) Sometimes the explanations are of a circular nature, e.g. “wise means possessed of
wisdom”,104 and are only intelligible if the meaning is already known.
(3) Sometimes the explanations are wrong, as I have already pointed out.
(4) Sometimes the commentary explanation has had an insidious effect upon the
canonical text, i.e. what was originally written in the commentary was sometimes
included in the text (as “glosses”), or had an effect upon the words in the text, in that the
text was changed to fit the meaning given by the commentary. It has been pointed out105
that the threefold categorisation of kamma (karma), which is found in some Theravādin
canonical texts, is due to the misunderstanding of the absolutive upapajja or
102 C.A.F.Rhys Davids, 1909, xvi–xvii.

103 von Hinüber, 1981, 74–86.
104 ti samannāgatā, Sp 552, 24.
105 von Hinüber, 1971, 241–49.
.106 This was thought to be incorrect, and it was consequentially “corrected”

to the “locative” upapajje. As a result of this, what had originally been a two-fold
classification, i.e. “one feels the result [of a bad deed] in the here and now or, having
been reborn, in some future period” became”…in the here and now, or in (a future)
rebirth, or in some future period”. This misinterpretation seems to have come into
existence in a 15th century on the , from which it was introduced
into manuscripts of the Netti itself, and then into manuscripts of the Majjhima-nikāya and
the -nikaya, on which the Netti passage was based.
106 The extension of an absolutive by a nasal can be found elsewhere in Pāli. See Geiger, 1994, §
214.
X
Philology and Buddhism
This is the last of the series of lectures on “A Philological Approach to Buddhism”. A

number of people have asked me what the difference is between “Buddhism and
Philology”, which was the title of my first lecture, and “Philology and Buddhism”, which
is the title of this lecture. My answer is “In principle, none”, but that does not mean that I
am going to repeat the first lecture.
If I had entitled the two lectures “Buddhism and the achievements of Philology” and
“Buddhism and the future aims of Philology”, respectively, it would perhaps would have
given the best indication of my intention, because that really is the point of the reversal of
the order of the two words. In the first lecture I looked back at some of the achievements
of philology in the field of Buddhist Studies, and looked forward to the subjects which I
was going to cover in this course of lectures. In this last lecture I want to consider the
subject from the opposite point of view. I want to look back over the series of lectures, to
summarise what I have said, and to look forward to see what contribution philology may
be expected to make in the future.
In these lectures, I have tried to show some of the results which a philological
approach to Buddhism has helped to produce. One such result is a better understanding of
texts, as, of course, is to be expected if there is a better understanding of the language in
which the texts are written, or if it is possible to make a comparison with other versions
of the texts in different languages or dialects. This, however, is not the only result A
philological approach can also lead to a better understanding of the way in which the
Buddhist tradition was transmitted, how the various versions of the Buddhist canon were
developed, how cultural developments had an influence on Buddhist texts, and
consequently how Buddhism itself developed.
In what I have said in these lectures I have been deliberately vague, avoiding as far as
possible the presentation of a plethora of Sanskrit, Pāli and Prakrit words, although, even
so, some of you may think that I have given far too many. I have very rarely quoted
authorities for the statements I have made, and I have never given specific references to
publications. I thought that such details were out of place in a course of lectures which
was intended to be of general interest, and which, I hoped, would be intelligible to those
who knew little or nothing about languages or philology. They will be added when the
lectures are published.
In this final lecture I want to say something about my hopes and fears for the future. I
want to suggest what further contributions I think philology can make to the study of
Buddhist texts and to the study of Buddhism itself, and I want to point to some of the
difficulties which lie in our path, and some of the problems which must be overcome if a
philological approach is to achieve the goals of which I think it is capable.
(1) To do this, I must summarise what I have said in previous lectures. I started these
lectures by explaining how I personally had become involved in philology and in the
study of Buddhist texts, and I said something about the light which a philological
approach had already thrown on some of the major problems of vocabulary and
interpretation which occur in Sanskrit, Pāli and Prakrit versions of Buddhist texts.
(2) In the next lecture I spoke about the political, economic, social and religious
background to the Buddha’s teaching. I commented briefly upon the date of the Buddha,
and said something about his background. I described the evolution of his
teaching from the religious movement which had arisen in opposition to the
caste, and I discussed the relationship between Buddhism, Jainism and
brahmanical Hinduism, and pointed to some of the vocabulary which they held in
common, although frequently with changed meanings.
The work of identifying traces of other brahmanical terminology in Buddhism
continues, and scholars are currently devoting time to considering the echoes of the
and the which can be heard in some of the Buddha’s teachings,
e.g. in the Aggaññasutta of the Dīgha-nikāya. There are certainly more features common
to Buddhism and Jainism which await detection. They may be small points, such as the
explanation I mentioned in the ninth lecture of the word arahant as “destroyer of
enemies”, which suits the Jain spelling of the word (i.e. ari-hant), but not the form which
we find in the Theravādin canon. Nevertheless, collected together, lots of small points of
this nature may help to establish a significant body of evidence.
(3) In my assessment of the oral aspects of the Buddha’s teaching I mentioned some of
the work which has been done to identify the similarities, and also the differences,
between the oral nature of the Buddhavacana in the fifth century B.C.E. and the
recitation of oral epics in modern times. Investigation is continuing into the features in
the Theravādin canon which seem to be mnemonic devices, designed to help in the task
of remembering canonical texts and reciting them accurately. If we can assess the
significance of such devices, and if we can understand the way in which modern oral
literature is recited, we shall be better able to understand the transmission system by
means of which the early Buddhavacana was handed down. Consequently we shall be
able to see what effect that system had on it, and it will, perhaps, be possible to deduce
something about the form of the Buddhavacana before the editorial process which we
associate with the system was instituted.
(4) As long as our knowledge of Middle Indo-Aryan dialects is limited to the
languages of the inscriptions, starting with those of Aśoka, and the literary languages,
there will be gaps in our knowledge of the dialects in which the Buddhavacana was first
promulgated and through which it was transmitted. Nevertheless, as I suggested in the
fourth lecture, on regional dialects, careful examination of what evidence there is in the
Theravādin canon, e.g. anomalous forms with unusual sound changes, will enable us to
make deductions about what lies behind our earliest tangible evidence, and to postulate
something about the characteristics of the dialects which were in use at a very early stage
in the development of Buddhism, and also about the development of translation
Philology and Buddhism 145
techniques which were used when the Buddha’s teaching was being taken to areas where
different dialects were spoken.
A linguistic problem which awaits solution is the nature of the early terminology of
Buddhism, and the way in which the various schools make use of terms which seem
similar enough to suggest that they are based upon common ancestors, and yet different
enough to defy an easy solution to the problem, e.g. apadāna/avadāna;
ekabījin/ekavīcika; anupadisesa/ ; -sesa/ ;
1
/ ; sammappadhāna/samyak- . Further
investigation may enable us to decide whether these variants are genuine dialect forms,
incorrect back-formations, or whether they have been modified to comply with the
interpretations which were handed down in the various traditions.
(5) In the lecture about Buddhism and writing, I said something about the theories
which are held about the introduction of writing to India, and the effect which early
Indian writing systems had on Buddhist oral literature when it was first written down. I
pointed out how the deficiencies of the writing system had led to ambiguous forms,
which were then interpreted in varying ways. Confirmation of the provisional dating of
the recently-found material from Anuradhapura in Sri Lanka is eagerly awaited, because,
if it is as old as it is claimed to be, then, providing there is enough of it, we should be able
to confirm whether our conjectures about the nature of early writing are correct. If they
are, then we should be better able to assess its effect upon texts.
(6) In the lecture on Sanskritisation, I gave some examples of the effect which the
introduction of Sanskrit forms into the Middle Indo-Aryan dialects had upon the structure
of the languages of Buddhism. I pointed out how redactors sometimes chose the wrong
one of a pair of homonyms when translating from Middle Indo-Aryan into Sanskrit, and
also how the necessity of making a choice deprived them of the word-plays which
homonyms had made possible. As more Sanskrit and Sanskritised texts from Turfan and
elsewhere are published, and as comparative studies of them are made, then we shall be
better able to assess the way in which Sanskritisation was carried out. This should enable
us to identify, more accurately, the precise form of the Middle Indo-Aryan versions
which underlie the Sanskrit texts.
(7) When considering Aśoka and the contribution which he made to Buddhism, I gave
information about what the Theravādin chronicles said about him, and also what he had
said about himself in his edicts, and I suggested that some of the claims made about him
were erroneous or at least overstated, in some cases because researchers had not made the
distinction between Aśoka’s dhamma and the Buddha’s dhamma. The time has perhaps
arrived when a new assessment of Aśoka should be made, in which philology and
archaeology are linked together to assess more carefully the area over which Aśokan
influence was felt, and the nature and effect of that influence.
(8) In the lecture on canonicity, I noted that this is essentially a western concept, and
that it is still not clear whether there was anything which could be called a canon in early
Buddhism or, if there was, when the various canons were closed. There is uncertainty
about the contents of the canons of some schools, and there is a continuing debate about
1 See Norman, 1989C, 375 (=CP IV, 53).
2 See, for example, Skilling, 1993.

the status of some Theravādin texts. There is need for more work of the sort which has
been done recently, i.e. the extraction from Sanskrit, Tibetan, and other sources,2 of
material which is alleged to belong to various schools of Buddhism, to see what we can
deduce about the texts which non-Theravādin traditions regarded as canonical.
(9) In the ninth lecture I spoke about the commentarial tradition, and showed that it
was often transmitted separately from the text being commented upon, so that the
commentary sometimes kept correct readings and explanations, even when the text was
corrupt. I pointed to the evidence that material, which occurred in commentaries in one
tradition, was regarded as canonical in another tradition, and I suggested that at an earlier
time the distinction between text and commentary was not always so clear-cut, and that at
one time text and commentary were possibly of equal importance. More comparative
work of this nature between commentarial and canonical texts of different schools may
be expected to unearth more examples of this phenomenon, and thus shed light on the
way in which the Buddha taught, and the means by which his teachings were transmitted
and explained.
In particular a thorough investigation of the commentaries should be made, because in
the past the need for such an investigation has tended to be overlooked. Since their
purpose is interpretive and exegetical, their language is often difficult to understand.
Consequently, the scribal tradition often found them problematic, with the result that the
text of commentaries is frequently corrupt. Some of the Pali Text Society’s editions are,
in fact, so bad that the Society is unwilling to reprint them. New editions are needed, and
also translations, and critical studies, in an attempt to establish what the relationship is to
the texts they comment on, and to versions in other traditions and languages. What little
study has been done has centered mainly upon the explanatory portions, the etymologies,
etc., but the commentaries also contain long exegetical passages, which sometimes seem
irrelevant to the subject matter being commented upon. We may presume that the reason
for the inclusion of these passages perhaps varied depending on the type of text, i.e.
material in a Vinaya commentary would have had a different purpose from that in a
commentary on a Sutta- text, and that would have differed from anything in the
commentary on an Abhidhamma- text. A careful study might help to reveal just
what purpose the commentators had in mind when introducing such material.
The individual subjects of these lectures were not, of course, clear-cut and separate.
They were designed to be inter-dependent to some extent, at least, and they were intended
to blend together to make a whole. My hope is that any future work done along the lines I
have suggested will lead to a situation where:
(a) A careful investigation of the Theravādin texts, and an assessment of the
philological and cultural information contained therein, will enable us to unravel the
problems of the development of the Middle Indo-Aryan dialects, and as a result of this, to
date, if only relatively, the composition of certain portions of early Buddhist texts.3 This
3 von Hinüber (1993, 113): “By using first of all the invaluable evidence of the Theravāda
and linking it to the events of general cultural history, it will not only be possible to trace
the history of MIA within this general development, but, at least to a certain degree, to find even
approximate dates for certain passages of early Buddhist texts”.
will also require us to take into consideration philological evidence available from other,
non-Theravādin and non-Buddhist, sources, and to co-ordinate it with archaeological and
any other evidence, which sheds light on the development and spread of Buddhism.
(b) And even more ambitiously, any conclusions to which we come as a result of this
philological work will not be restricted to Buddhism, but will enable us to define more
accurately the development of the religious milieu as a whole during the period between
the fifth century B.C.E., and the fifth century C.E., i.e. between the time of the Buddha
and the time of Buddhaghosa, and also the literature, in the very broadest sense, which
was composed in that religious environment, whether Buddhist, Jain, brahmanical or
secular.4
In my first lecture I repeated the statement which I usually make to those who are just
entering the field, and wonder what tasks remain to be done, and which task they
personally should select as their contribution to Buddhist Studies. In this last lecture I
will repeat my statement yet again: “Everything that has not been done needs to be done.
Everything that has been done needs to be done again”. So everything that I have said
about the achievements of philology, every suggestion I have made, and all the
suggestions of other scholars, all need to be kept permanently under review to see if new
evidence can help to prove or disprove what has been proposed.
So much for my hopes. What about my fears? You will have noticed that all the tasks I
have just mentioned demand considerable philological skills. They all require a wide
knowledge of Indian languages, and in some cases Tibetan and Chinese too. And this is
the problem. Where are the philologists who can do this now, and who will train the
philologists of the future?
What I should therefore like to do in this final lecture is to make a plea for a bigger and
better use of philology in the field of Buddhist studies. Unfortunately, I know that for the
most part my plea is destined to fall on stony ground. For a number of reasons the
number of philologists capable of working in the field of Indian languages and in the
field of Buddhist studies based on those languages—I will say nothing about the
comparable situation in Tibetan, Chinese, Khotanese, and a host of other languages in
which Buddhist texts are written—is decreasing and is likely to decrease even more in the
future. There are various reasons for this. One of them, perhaps the most important, is
that because the number of teachers of Indian philology is small, and the number of
students of philology is also small, compared with other subjects, in any set of
circumstances where the financing of teaching is in direct proportion to the number of
students taught, then the amount of money available for teaching philology will decrease
as the number of students decreases proportionally. Less money for teaching means less
teachers, which means fewer classes, and as the numbers of classes and students continue
to decrease, in comparison with other subjects, the amount of money available for
teaching philology will decrease even further. This is a circle of the most vicious nature.
And this is very unfortunate, because by and large the number of those interested in
Buddhism is growing and I think will grow even more now that Buddhism is one of the
4 Bechert (1991A, 19): “If we view our question in its broader ramifications, its answer will prove
to be an important element in the task of elaborating an accurate understanding of the entire
linguistic, literary and religious development in India during the 5th to the 1st century B.C.E.”
religions to be taught in British schools as part of Religious Education. And so I would

hope that as the numbers of those studying Buddhism grow, the amount of money
available for the teaching of Buddhism will increase and the number of teachers of
Buddhism will increase. But if the possibility of learning the languages of Buddhism has
diminished, and if the teachers of Buddhism have not been able to gain an adequate
philological training in Buddhism, then I suspect that the situation that pertains at the
moment, where much of the work in the field of Buddhist studies is done by those who
have not had an adequate philological training in the languages required, will continue
and will in fact get worse.
I suspect that what I have said in these lectures will have little effect upon some
scholars, who would probably say that I am overstating the need for philological training.
They would perhaps not describe themselves as philologists, but would nevertheless
regard themselves as sufficiently competent in philology to be able to handle all the
textual material they need, because in their books and articles about Buddhism they do
include Pāli and Sanskrit terms. This, however, is not what I regard as philology. Any
work on Buddhism, which makes a claim to be based on original language sources, will
be worthless unless it is based upon a full understanding of those sources. This may
appear to be stating the obvious, but it seems to me that much of what is alleged to be
based on such sources is, in fact, not so based. Some of it seems to be based on
translations, with Pāli and Sanskrit equivalents inserted in brackets, to give a veneer of
scholarship.
Anyone who writes about Buddhism can sprinkle his article with Pāli and Sanskrit
equivalents, but this is not the same as knowing what the words mean, or why they mean
it. I say “anyone can do it”, but this is not entirely true. I have read books, published
recently, by teachers in recognised academic institutions, where the Pāli and Sanskrit
words in brackets do not actually coincide with the English words which are supposed to
be their translation. Even if the authors have succeeded in tracking down the correct place
on the page in the text, the way in which they quote the reference, with compounds
wrongly divided, or case forms misunderstood, makes it clear that tsome of them do not
fully understand Pāli and Sanskrit.5
Diacritical marks are often a complete mystery to those who write about Buddhism. One
gets the impression that some authors assume that they are just dots and dashes which can
be added or removed at will. It sometimes appears that they are added merely to placate
critics like myself, who in reviews regularly complain about their omission.6 They are
inserted in such an inconsistent and incorrect way that it is obvious that the authors have
little idea of what they are doing, and do not understand that diacriticals are a matter of
spelling, not of aesthetics, or of whim, and result from an attempt to make the Roman
alphabet cope with a sound pattern that needs an alphabet with more than 26 letters.
Omitting diacritics is therefore a matter of misspelling and inserting them is a matter of
correct spelling. It is not just a way of placating some pedantic academic who is a stickler
for accuracy.
What is perhaps the most discouraging feature of all is that some of these publications
5 See, for example, the review of Wiltshire, 1990 (BSOAS 55, 1, 1992, 144–45).
6 See, for example, the review of Stargardt, 1990 (JRAS, 2, 1, 1992, 114–17).
are in fact based upon doctoral dissertations, which means either that the examiners
concerned did not detect such blatant errors, or that, hardly less culpable, they did not
inform the doctoral candidate of the errors, if they did spot them.
When I began to read Indian Studies more than 40 years ago, the situation in Buddhist
studies seemed very simple. If you look at some of the older books on Buddhism
available in English about that time you will find that everything seemed cut and dried.
We knew exactly when the Buddha lived and died; we knew what he had preached and
how he had preached, and we knew how his followers had recited his words and handed
them on to their followers; we knew how in three joint recitations held during the two
centuries or so after his death this tradition had been continued; and so on. I remember
that I was advised to read E.J.Thomas’s The Life of Buddha and The History of Buddhist
Thought, because those two books would tell me all I needed to know about Buddhism—
you will understand that I was training as a philologist, not as a Buddhologist.
In general, books about Buddhism written in English at that time tended to be based
on Pāli source materials, partly because of the long history of Britain’s involvement in
Burma and Sri Lanka, which had brought British missionaries and civil servants into
contact with Theravāda Buddhism. It was thought that with all the Pāli canonical texts
and commentaries published and most of them translated, and with the Pali-English
Dictionary and other ancillary works available, all problems in Buddhism had been
solved, or nearly so. There was a tendency to think that Pāli sources gave the most
reliable picture of early Buddhism, being the oldest and the most complete, while
Sanskrit sources, being mainly Mahāyāna, were late and suspect, with their grain of truth
overlaid with a thick cover of mythology.
This view had, of course, been quite untenable since the discovery of Sanskrit
Hīnayāna material in Chinese Turkestan and Gilgit, etc., and to do Thomas justice, he did
take account of Sanskrit materials, and he pointed out the danger of neglecting the works
of schools preserved in Sanskrit, and in Tibetan and Chinese translations from Sanskrit.
As he said, such Sanskrit works needed to be analysed, no less closely than the Pāli. But
Thomas wrote his books in 1927 and 1931 respectively, when comparatively little of the
Hīnayāna material from Chinese Turkestan had been published, and the Gilgit
manuscripts had not yet been discovered. It was in the third edition of The Life of
Buddha, published in 1949, that he drew attention to the fact that our knowledge of
Buddhism during the previous 20 years had consisted chiefly in establishing the fact of a
Sanskrit canon parallel to the Pāli, and he was able to include a reference to
Waldschmidt’s analysis of the in his Bibliography. In the
preface to the second edition of The History of Buddhist Thought (published in 1951), he
again referred to Waldschmidt’s analysis of the , and its value
for determining more of the actual relations between Pāli scriptures and the recensions of
the Indian schools, although he did not actually include any material from Waldschmidt
in that revised edition of his book. Thomas’s books have been reprinted and are still
available, and are still referred to, despite the absence in them of any detailed
consideration of all the texts published since they first appeared more than 60 years ago.
Outside Britain, the situation was rather different, and there was less emphasis on Pāli
sources. Étienne Lamotte commanded a wide range of languages, and his Histoire du
Bouddhisme Indien7 was based upon Sanskrit, Tibetan and Chinese as well as Pāli
materials. In Germany a strong school of Sanskrit Buddhist Studies was emerging, based
mainly upon the German Turfan material, which was providing material for doctorate
after doctorate. Although much more of the Turfan material has been published since
1951, it has not all been translated, or if translated, it is usually into German and only
rarely into English. The differences between this Sanskrit material and the Pāli tradition
have still not been adequately studied, and hence it has not been adequately taken into
account in any English books which deal with Buddhism in a general way. So a recent
publication, intended, according to the publishers, as a textbook specifically for students
of Religious or Asian Studies, deals with the four noble truths and other Buddhist
doctrinal matters in their Pāli versions, without making it clear that these are not
necessarily the original forms of these terms, and that in some cases the Sanskrit
equivalents are perhaps nearer in form and meaning to the Buddha’s own words.8
I had myself studied with Professor Sir Harold Bailey, who was well acquainted with
the material from Chinese Turkestan and referred to it constantly in his lectures, but since
his main orientation was towards Iranian, it was the Khotanese material which he tended
to refer to most often, and the impact of this new material upon Indo-Aryan studies, and
particularly the Middle Indo-Aryan studies on which I had decided I would concentrate
after graduation, was somewhat less emphasised.
The biggest influence upon Middle Indo-Aryan studies came from the posthumous
publication of Heinrich Lüders’ Beobachtungen über die Sprache des buddhistischen
Urkanons by Ernst Waldschmidt in 1954, with its collections of eastern dialect forms in
Pāli and Buddhist Sanskrit, including those which had been misunderstood by the later
tradition, its observations on the phonological and morphological features which Lüders
had identified as belonging to the Urkanon, and its comments upon such details of
nominal inflection as the ablative singular and accusative plural endings in - . As I
have sometimes noted in studies I have made of various phenomena in Middle Indo-
Aryan, a few examples of something can frequently be explained away piecemeal, but
when dozens of examples are put forward together, it is more difficult to do this. All the
constituent examples of a phenomenon have, together, a holistic effect, where the
evidence as a whole is greater than all the individual parts taken singly.
Sir Harold had already referred to many of these individual features in his lectures,
and I find that my student notes from his lectures on the Pāli Dhammapada are full of
references to the published portions of the Udānavarga and to the Gāndhārī Dharmapada
of which he had made an edition, at least of the portions which were available to him at
that time. It was not, however, the individual forms which were so impressive, but the
fact that all of them put together really made a coherent whole. Whether they proved that
there had been an Ur-kanon or not was beside the point. What was clear was that there
was evidence which could be extracted from the texts to show that there had been
dialects, not attested in any literary works now extant, with characteristics which could
help to explain some of the problematic forms which existed in Buddhist texts in
Sanskrit, Prakrit and Pāli.
At that time, there was about half the Udānavarga available in Chakravarti’s edition of
the French Pelliot material,9 and a few other vargas that had been published separately.
7 Lamotte, 1958.
8 See the review of Harvey, 1990 (BSOAS 55, 1, 1992, 142–43).
9 Chakravarti, 1930.
Some years later our knowledge of this text was immeasurably increased by the
publication of Bernhard’s edition of the German manuscripts of the Udānavarga,10 at
almost the same time as John Brough published all the Gāndhārī Dharmapada material
which was available,11 thus revising and augmenting the portions of this in the French
and Russian collections which had been made known more than 60 years before. Then
came the discovery that there was another version of the Dharmapada available in
Buddhist Sanskrit, the so-called Patna Dharmapada, photographed by Rāhula
in Tibet in the 1930’s. This disclosure was closely followed by no less
than three editions of this text,12 followed finally by an edition of all the French
Udānavarga material.13 This means that there is now enough material available to make a
full-scale comparative investigation of the Indian source material of the Dharmapada
genre, but despite the fact that new translations of the Pāli Dhammapada appear almost
every year, few translators seem to realise the help which such a comparative study could
give, or if they do realise it, they do little about it.
Other branches of Indian Buddhist textual studies are not so well provided for, but the
continuing publication of material from the German Turfan collection and the on-going
appearance of the Dictionary of the Turfan Sanskrit material14 means that more and more
comparative studies of suttas in the prose nikāyas can be made. Once again, few
translators of Pāli suttas make such studies before producing their translations.
As a result of the publications of all these texts in Sanskrit and Prakrit, much of what
has been written about the early days of Buddhism, the assembling of texts, the way in
which they were transmitted, and the languages in which they existed is now, to a greater
or lesser extent, incorrect or misleading. Consequently a meeting was held in Brussels in
1989 to see what should be done to bring Lamotte’s work up to date. Bizarrely, the
meeting was timed to coincide with the release of an English translation of his Histoire.15
This was a straight translation from the French, with almost no attempt, except for the
Bibliography, to bring it up to date, despite the protests of some of us that the book
needed revising, not translating. It seemed nonsense to me, when I was asked to act as a
consultant on the section which Lamotte had devoted to the formation of the Buddhist
languages,16 to be told that I must make no changes whatsoever to the portion sent to me,
even though I knew well, in the light of knowledge gained since 1954, that some of it was
misleading, or even incorrect.
Consequently, those who now rely on the English translation of Lamotte are making
use of something which is not quite as old as Thomas’s books, but still more than 35
years out of date. Reading the books listed in the Bibliographical Supplement will, of
course, help to overcome this defect, but a list of publications, most of them prefixed by
“see…”, “see now” or “new study” is not very helpful, since few of the entries give a
summary of the conclusions arrived at in the new publications. The very aim at
completeness in some of the entries is self-defeating. For example, there is a
10 Bernhard, 1965.
11 Brough, 1962.
12 (1) Shukla, 1979; (2) Roth, 1980; (3) Cone, 1989.
13 Nakatani, 1987.
14 SWTF, 1973–.
15 Lamotte, 1988.
16 Lamotte, 1988, 549–93.
reference to a dozen publications about Aśoka’s Greek-Aramaic bilingual inscription

from Kandahar,17 which have appeared since Lamotte’s Histoire was first published. This
leaves the reader uncertain about where to start in his search for extra information about
this inscription. In the fifteen lines taken up in listing these works, useful information
could have been given about the relative value of some of these publications.
I should strike a warning note. Although comparative studies, e.g. those on the
Dharmapada genre which I have just mentioned, can bring positive results, the field to be
covered is so large, especially if it is extended to include Chinese and Tibetan
translations, that one person can scarcely be expected to cover all of it. Such projects
must therefore be joint ones, and although in theory such a solution to the problem should
bring good results, this will only happen if there is complete and harmonious co-
operation between all participants, and if there is give and take between the proponents of
each tradition. If I choose to think that the Middle Indo-Aryan component is nearer the
Buddha’s own words, and I therefore disparage everything outside that field, as being
merely translations—and sometimes faulty translations—of Middle Indo-Aryan originals,
then I shall be the loser. If scholars working in the Buddhist Sanskrit field are content to
edit their texts and make statements such as “metre incorrect”,18 whenever the metre of
their verses does not fit the classical Sanskrit pattern, then they will be the losers. The
task of reading fragments written in a difficult script, and of finding parallels in Pāli,
Tibetan or Chinese texts which will help with the reading of illegible and the
understanding of problematic readings, is in itself very considerable. Nevertheless, a
study of related fields might well enable those scholars to understand the problems better,
to realise how the imposition of Sanskritisation upon a Middle Indo-Aryan original has
produced these incorrect forms, to deduce the forms which lie behind their texts, and thus
to interpret them better. If those dealing with Tibetan or Chinese believe that their
traditions are more faithful transmitters of the Buddha’s teaching than any of the Indian
language versions, which have in some way become corrupted, then they are the losers.
There is, however, no reason why any of us should be losers. This is a situation where we
can all be winners if we are willing to work together, and to stop disparaging those whose
interests do not coincide exactly with our own.
I am afraid that some of you may have found this course of lectures hard going. I am
sorry about that, but the fact remains that philology is hard going, which is why
philologists are rather rare beasts, perhaps even an endangered species. The purpose of
this course of lectures has been to show that the study of Buddhism based upon texts is
not a simple matter of taking an edition and a translation and assuming that all problems
have already been solved.
I will repeat myself, and say again that, in my opinion, no worthwhile original work
on early Buddhism, perhaps any sort of Buddhism, can be done by anyone who does not
have a good grasp of the relevant languages. And by a good grasp I mean the ability not
only to say what the words mean, but also why they mean it, so that a researcher is not
restricted to working with translations, but can actually refer to the original text and see if
the translator has gone astray. As part of his work he should be able not only to look
17 Lamotte, 1988, 743.

18 Matsumura, 1989, 69–100 (e.g. “Pāda b unregelmäßig”, 81, note [32].2).
words up in a dictionary, but also to judge whether the meanings given there are likely
to be correct Dictionary writers are only human, and they too, like other people, are liable
to make mistakes, including mistakes of judgement.
For example:
(1) There is in Pāli a word meaning “thorn”, and then, metaphorically,
“obstacle”. In several Theravādin texts there is a statement that “sound (sadda) is an
obstacle to the first meditation” ( jhānassa saddo ), where the
word for “obstacle” is . There is a compound “sound or noise is an obstacle” or
“having sound as an obstacle” (sadda- ), and we find statements such as
“meditations have sound as an obstacle” ( jhānā). A Critical Pāli
Dictionary, however, although listing the meanings “thorn” and “obstacle” for ,
has a separate meaning for the examples I have just given: “a sharp and annoying sound”
and it compares the Sanskrit usage.19 It does not specify what Sanskrit usage, nor is the
reader given any guidance about the way in which the sentences are to be translated if
this sense is adopted.
(2) Again, reference to A Critical Pāli Dictionary will show that it gives two meanings
for the compound kañcana-de-piccha. On the face of it, this word should mean “having
two golden tail feathers”, but the commentary explains it as “having two wings like
gold”. The Dictionary entry states that the literal meaning does not make sense, but no
reason is given for this statement. The editors have clearly been unable to agree about the
interpretation of this word, and have therefore agreed to differ and, more surprisingly,
they have agreed to let the world see their difference, even to the point of writing two
separate entries, each signed by its author.20 The two separate entries are mutually
incompatible, with different word divisions proposed for the verse in which the word
occurs. One editor says “pāda a must contain a finite verb”, and refers to a parallel
passage in the Jātakamālā as an authority, but gives no guidance as to how the
problematic compound is to be translated. The other editor gives no hint of believing that
there must be a finite verb in the first pāda, and says that the verse in the Jātaka-mālā has
been reformulated, and is therefore, we assume, no guide as to the original form of the
verse. What is a user of A Critical Pāli Dictionary to do? Few will have the
lexicographical expertise to solve the matter themselves, and, not unreasonably, will
expect the editors to have solved the problem for them, They will be disappointed,
because the editors would seem to have abandoned their editorial responsibility. It is
unique, in my experience, for two editors of a dictionary to be unable to agree upon a
meaning, or, if there is genuine doubt about the meaning, to be unable to agree upon a
form of words which would make the difficulty of finding a meaning clear.
So those working in the field of Buddhist Studies must have genuine expertise, the
ability to make real judgements. They must be able to tackle textual problems, and not
propose emendations to texts simply in order to produce an easier reading. Still less must
they give arbitrary meanings to words because the attested meanings do not fit in with
19 CPD III, 58.

20 CPD III, 29–30.
their theories of what the Buddha said. Theories must be based on facts and must fit the
facts. Facts must not be twisted to fit a theory.
My optimism about an increased interest in the study of Buddhism, because it is one
of the religions to be taught in British schools, is perhaps outweighed by my pessimism
about the way in which Buddhism will be taught. Despite all the efforts put into this
project by advisory committees and bodies whose aim is the promotion of Buddhism,
there will inevitably be shortcomings in the teaching. I suspect that there will be a market
for school-level text-books on various religions, where opinion will be represented as
fact, and everything will be set out clearly as unambiguous and undeniable. I am well
aware of the fact that at school level the teaching must of necessity be somewhat cursory,
but I hope that the teaching will not be so cursory as to obscure the fact that there are
many problems remaining in the interpretation of quite vital aspects of Buddhism. The
problematic nature of these concepts is rarely stressed in popular literature, and the views
of different schools of Buddhism are rarely given, even, as I said earlier, in a book
intended as a handbook for university students.21
If you investigate the translations and interpretations given of many of the basic terms
of Buddhism in such text-books, you find that what had been presented as a fact is in
many cases only the writer’s opinion, said three times, of course, so that it is true. Very
few things in life are clearly black or white. Most of them are some shade of grey, and
this is true for Buddhism also. Even for such fundamental concepts as Buddha,
, , āryasatya, etc., the translations which are usually given, and the interpretations
which are made on the basis of those translations, are frequently misleading or, if not
misleading, tell only half of the story.
It is alleged, and I have no reason to doubt the accuracy of my informants, that when
teaching posts in Indian Religion are being filled in Departments of Comparative
Religion in some American universities, preference is given to those who have no
language ability, so that applicants who do have some philological expertise are forced to
hide the fact, in order to have any hope of being appointed. What hope is there for the
subject in such circumstances? Even in British universities, if there is no language
component to be taught in courses on comparative religion, then the ability to teach
languages will certainly not be a qualification when applying for jobs, and might even be
taken to be a disqualification by an appointments committee which can see no relevance
in the ability to read source materials in the original language. Such an appointments
committee might perhaps fear that anyone with that ability would, if appointed, waste his
time teaching this irrelevant subject.
I can only view with horror the prospect of more and more courses on Buddhism being
taught to those who have no knowledge of language, by those who have no knowledge of
language, so that if those being taught want to learn the language they cannot do so,
because there is no one to teach them.
Some years ago I read a paper to the Buddhist Forum22 in the School of Oriental and
African Studies, in which I drew attention to the great difference to be seen between the
state of affairs in the editing of Pāli texts as opposed to editions of Christian gospels in
New Testament Greek. I pointed out that whereas we might have an edition of a Pāli text
21 Harvey, 1990.
22 Norman, 1990C.
based upon two or three manuscripts or printed editions, or even less, New Testament
scholars might be able to make use of 200 or even 2000 manuscripts and other sources. I
contrasted the number of New Testament scholars who could handle the Greek of the
New Testament, as well as the Greek of the Old Testament and the Hebrew of which that
is a translation, with the comparatively small number of scholars capable of handling
Pāli, Sanskrit and Prakrit. I was therefore surprised to find, in a recent publication by a
Professor of Theology at a British university, a reference to the state of affairs in one
branch of the Christian church—and by no means a moribund branch. It was stated that
one of the aims of the book was “to stimulate…renewed interest in serious theological
study, particularly among ordinands and clergy”,23 where it would appear that interest
was waning. There was a clear statement that academic theology at its best is concerned
with the central questions, not the proliferation of marginal theories,24 which reminded
me of the marginalisation of Buddhist scholars to which I referred in my first lecture. The
book also included the revelation that many ordinands were informing the author that
they were giving up Greek, i.e. New Testament Greek, because they could not see its
relevance to their future ministry.25 They apparently felt no need to read the primary
source for their religion in the original language. What was needed, the writer stated, was
a revival of serious interest in learning the original languages in which the great treasures
of the tradition had been written. It seems, then, that even in a subject so well established
in Great Britain as the study of Christianity, there are problems when it comes to training
people adequately in philological method. We are then not alone in having problems, but
if that is the situation in a field where the facilities are infinitely greater than in the field
of Indian and Buddhist Studies, then the outlook for us is bleak indeed.
The difference is that, in the field to which I have just referred, the study of the
tradition has always in the past been based upon philology, and the mechanism for
teaching philology still exists, and will continue to exist, for a time at least, despite the
flight from philology by those newcomers to the field who do not realise its relevance. In
the field of Buddhist Studies, however, in many universities the subject has never been
taught on a philological basis, even if the teachers were capable of doing so, because the
syllabuses and the examination system have not demanded it. The situation in the future
can only get worse, because those graduating from such universities—those who will be
the teachers of the future—will not have received the necessary training to remedy the
situation. I am glad to be able to say that my own university still thinks that competence
in language is essential for religious studies, and those reading theology and religious
studies have compulsory papers in Hebrew, Greek, or Sanskrit. I hope that this will
continue to be the situation, although I know that this deters some applicants for
admission, who consequently choose to go to other universities, where life looks a little
easier.
Nevertheless, even if knowledge of the relevant languages is not compulsory, I hope
23 Young, 1992, 3.
24 Young, 1992, 16.
25 Young, 1992, 2.
that there are those who realise that it is not satisfactory to study Buddhist texts
exclusively in English translation, without being able to read anything of the teachings of
the Buddha in an oriental language. If anything I have said in any of these lectures has
inspired anyone to learn a little more about philology, then my time has not been wasted.
Word Index
akata 38
akusala 15, 17, 53
akkodhano 89
agada 162
agge 71
132
44
acala 37
ajāta 13, 38
ajje 68, 72
77, 133, 134, 152, 153, 154, 165
attaja 74, 98
attā 26, 27, 37
atthe 74
atraja 74, 98
advaitavāda 26
adhisessati 156
adhunā 165
anattā 26, 27, 119
anicca 26, 27
165
anudaka 108
169
anupādisesa 169
anusvāra 85, 86, 87
anūdaka 108
anodaka 109
apagata- 156
apadāna 92, 169
apāra 164
apeta- 156
abbhutadhamma 132
abhidhamma 48, 137
abhidhamma-dhara 137
abhibhāsana 153
abhihāsana 153
abhūta 38
amata 13, 37
Word index 159
amatapada 13
13
ayya 68
ara 161
150
arahat 30, 35, 161, 163
arahant 168
ari 161, 162
ariya 16, 52
ariya-sacca 16
ari-hant 162, 168
arūpa-jhāna 31, 32
arhat 35
avakada- 156
avadāna 169
avijjā 32, 33
38
117, 118
ahunā 165
ākiñcaññāyatana 30
ājāniya 160
ātmaja 74, 98
ātman 26, 27, 28, 37, 39
Ādiccabandhu 162
ābhidhammika 47
āyudha 73
āraka 161
āra-kilesa 161
67
ārāma 123, 125, 126
ārya 68
āryasatya 181
āryā 107
ālambana 67
āva 88
āvāsa 126
āvudha 73
āvuso 63, 73, 97
āsava 11, 30, 31, 34, 35
āsavakkhaya 31
itivuttaka 132
Indara 61
īhita 165
utpatya 84
utpāda 66
utrasta 98
udaka-baka 108
udaya-vyaya 108
udāna 132
uddāna 51, 145
Word index 160
upaca 84
upapajja 166
166
upapajje 166
upari 156
upādāna 33
upāśake 115
upāsaka 114, 115, 119
upeca 84
upecca 83, 84
upetya 83, 84
uposatha 123, 124, 125, 126, 128
uppacca 84
uppatitvā 84
uppāta 65, 66
uppāda 65
35
eka-uttiraka 47
ekabījin 169
ekavīcika 169
67
esin 162
ogha 34
orapāra 164
kañcana-de-piccha 179
179
15
kattā 100
katvā 100
kappa 150
kamaradu 86
kamaramu 86
kamma 11, 34, 89, 166
kammarata 86
kammavācā 56
kammic 33
karma 24, 120, 166
kāma 102
kāmarata 86
kiriyā 105
kusala 15, 17, 53, 159, 160
100
kriyā 105
73
23, 33, 81, 140, 158, 168
-jña 161
73
khanti 118
Word index 161
30, 35
khettajina 161
158
gata 162
gada 162
gadana 162
gātha 132
gīti 107
guru 161
geyya 132
16
caitya 121
cora 153
153
73
Chanda 67
chandaso 60, 109, 157
Channa 67
chuddha 156
chettā 100, 101
chetvā 101
janapadanirutti 149
jantāghara 63, 68
Jamadaggi 69
Jamadagni 68, 69
33
jātaka 132
jāti 33
jāna 108
jāva 88
Jina 35, 36
-jina 161
jināti 158
je 68, 70, 72, 73
-jña 160, 161
jñāna 108
jhāna 7, 29, 31, 32, 179
jhānic 33
ñāya 52
134, 165, 166
33, 104
-kappa 150
tathāgata 35, 37, 162
tapas 35
tasa-thāvara 163
tasā 163
48, 78, 132, 133, 138, 146, 154
133
Word index 162
133
tuccha 156
tuchu 156
48, 133
tosana 153
trasa 163
thambha 116
thāvarā 163
thuba 67
thūpa 67
- 155
dāna 23, 34
-kappa 150
Dīgha 44
dīpa 101
dīpa-bhāsā 151
du 165
dukkha 26, 27, 37, 83, 84
duhita 166
duhitika 166
duhitikā 165
dūta 117, 118
deva 25
dvīpa 101
dvīhitikā 165, 166
dhamma 11, 16, 27, 54, 55, 101, 116, 117, 118, 119, 120, 122, 127, 129, 132, 152, 170
dhamma- 127
15
dhammathambha 116, 117
dhammadāna 116
116
dhammamahāmātra 116, 117, 129
dhammayātrā 116, 120, 127
dhammalipi 116, 117
dhammavijaya 116, 118
dhammas 15, 17
dhammasavana 116
dhammānuggaha 116
dhamme 74
dharma- 47
dharma-vijaya 113
nāga 162
nāmarūpa 32
nikāya 44, 45, 46, 47, 49, 50, 78, 79, 91, 92, 93, 133, 136, 142, 143, 145, 177
nicca 26
nija 69
nijhati 117
nitya 27
nibbāna 11, 13, 17, 28, 30, 31, 32, 34, 35, 37, 39, 63, 73, 119, 120, 164
Word index 163
nibbuta 35
nimitta 45
nimināti 84
84
84
nimmināti 84
84
niya 69
niyāma 105
nirukti 158, 159
nirutti 60
nirupadhi 108
nirūpadhi 108
niropadhi 109
niraupadhi 109
nirvacana 158, 159, 160
13, 22, 24, 41, 119, 181
35
nivvua 35
nekkhamma 102
nevasaññā-nāsaññāyatana 30
102
102
nyāma 105
pakkodano 89
pacceka 13, 103, 104
pacceka-buddha 36
pacceya 104
pacceya-buddha 104
paññā 31
169
159
patteya-buddha 14, 36, 103
pada-bhājaniya 150
pabbajjā 124
paramparā 82
para-(v)vāya 100
pari- 17, 63
paritta 48
parinamati 89
parinibbāna 17
parinibbuta 17
35
35
105
parisā 105
paryāya 87
Word index 164
105
pali- 63
123, 125, 126
158
Pasenadi 158, 160
pāda 160
pāra 164
pārājika 149
pāli 63, 133, 134
134, 141
pāli-bhāsā 63
pāli-muttaka 140
122
45, 48, 77, 133, 136, 137, 141, 142, 145
pitus 61
pīti 101
15
71
pure 71
puro 71
purohita 23
15
15
peyyāla 87, 88
peyyālams 87
132
169
pratītya-samutpāda 32, 33
pratyaya 104
pratyeka 13, 103, 104
pratyeka-buddha 13, 103, 162
pradhāna 102, 103
pravrajita 35
pravrajyā 35
Prasenajit 158
prahāna 102
prīti 101
pretya 24
phala 162
phalesin 162
phassa 32
47
bahuśruta 42
bahussuta 42
bābhana-samana 122
103, 160
bāhita 160
Word index 165
buddha 13, 29, 32, 35, 36, 103, 115, 181

Buddha-jñāna 108
Buddha-dhamma 116
Buddha-bhāsita 135
Buddha-yāna 108
Buddhavacana 44, 60, 63, 64, 71, 81, 92, 93, 95, 104, 110, 133, 134, 135, 136, 141, 142, 145, 146,
147, 148, 156, 165, 168, 169
Budhaśake 115
bodha 32
bodhi 29, 30, 31, 32, 33, 105, 108, 115, 120, 121
bodhiśakta 105
105
bodhisatta 104, 105
bodhisattva 104
brahma-carya 28
brahman 26, 27, 28, 39
brahma-patha 28
brahma-vihāra 28
brahma-sahavyatā 28
23, 28, 36, 41, 98, 103, 122, 159, 160, 168
- 122
- 36
brūheti 98
bhagavat 158, 161
bhagī 161
161
bhajī 161
bhaya 52
bhava 33
bhavati 108
bhav-anta-go 161
bhasta 98
bhāgī 161
bhāgyavā 161
44, 46, 47, 48, 49, 55, 56, 59, 77, 78, 81, 82, 83, 90, 91, 92, 93, 136, 137, 158, 169
bhikkhave 63, 72
bhikkhavo 63
bhikkhu 34, 62, 73, 124, 125, 126, 154, 157, 158
bhikkhunī 63, 73
34
bhūnahū 73
bheda 124, 125
bhoti 108
Magadha- 155
magga-jina 161
116
majura 70
majūla 70
Word index 166
Majjhima 44
majhima 47
mayūra 70
mahā 16
mahāpadesa 91, 134, 137
mahāmātra 115, 116, 128, 129
Mahāvīra 35
mātikā 48, 51, 137
mātikā-dhara 137
Mātā 66
Māyā 66
mukta 35
muktā 105
mucati 85
muccati 85, 89
mucceya 85
mucchati 89
mucyate 85
muñcati 85
85
muttā 105
muttu 105
muni 159
muhur 61
mūrchati 89
mūrta 105
mokkha 100, 119
29, 41, 181
mora 70
162
Yamataggi 68, 69
yātrā 116
yādeti 63
yāna 108
yāvat 88
ye 68, 73
yo 149
rañj- 159
ramati 101
rahābhāva 161
rāga 161
rāj- 159
rājā 23
rājā 159
rūpa 26
rūpa-jhāna 29, 31, 32
118
lipati 85, 86
lipi 67
lipī 42
Word index 167
lippati 85, 87
libi 67
85, 87
vakka 99
vagga 142
vagguposatha 124
vacana-vidu 160
vatika 84
vattika 84
vada-jña 160
vadaññu 160
vadānya 160
vayadi 81
vari 156
vā 152
vāk 99
vākya 99
vāceti 42
61
61
vijjā 33
vijñāna 92
32, 92
vinaya 92, 134, 147
vinaya-dhara 137
vibhattavā 161
viśva 152
152
visama 152, 156
vissa 152, 156
vihāra 56, 78, 110, 111, 114, 121, 125, 126
84
134
vedanā 32
vedalla 132
132
veśma 152, 156
veśma-dharma 152
vĕssa 152
vaiśya 23, 33, 81
vrajati 82
vratika 84
šatta 61
sa’ 156
śata 38
śayutaka 47
śānta 101
29, 34, 35, 36, 37, 39, 41, 103, 122, 168
Word index 168
122
śrāvaka 13, 103
śruti 42
161
44
24, 25, 28, 30, 31, 32, 34, 36, 37, 38, 39, 41, 92, 104, 119, 164
157
sakkata-bhāsāya 157
32
78
43, 47, 48, 52, 67, 78, 81, 90, 91, 114, 123, 127, 128, 132, 135, 136, 139, 140, 141, 143,
145, 146
135, 145
135
56, 77, 92, 115, 120, 122, 124, 125, 126, 128, 141, 145
122, 125, 128
169
169
sacca 16
sacce 74
sañcicca 84
sañj- 105
saññāvedayitanirodha 31, 32
sata 38
satī 101
157
sattva 105
satyam 74
sadda 179
sadda- 179
saddhamma 120
santa 101
sapta 61
samagga 123, 125, 126
103, 159
- 122
samāpatti 31
Sambuddha 35
sambodhi 120, 121
sammappadhāna 102, 160, 169
sammappahāna 102, 160
102, 160, 169
saya- 100
sayati 82
sayissati 156
Word index 169
sarvajña 35
32
salīle 121
sālā 122
sāsana 113
Siddha 35, 37
siddhattha 37
siddhi 37
sibbanī 104
siva 37
sukka 15
sukha 26, 27, 37
Sugata 35
sutta 16, 43, 46, 47, 50, 52, 54, 55, 88, 90, 92, 93, 104, 127, 132, 34, 141, 142, 143, 144, 146, 147,
150, 151, 156, 160, 164, 177
sutta-dhara 137
subhāvitattano 161
104
suhita 166
sūkta 104
sūtra 104
sota 104
sotaviyā 42
52
sŏtta 104
stūpa 114, 121
sthāvara 163
104
svarabhakti 61, 105, 109, 152, 161, 162
68
hatthi-vatika 84
hatthi-vattika 84
hanta 67
handa 67
huveyu 105
67
hveyu 105

佛教的语言学方法 A Philological Aprroach to Buddhism K. R. Norman

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

佛教的语言学方法 A Philological Aprroach to Buddhism K. R. Norman

Uploaded by

Copyright:

Available Formats

SCHOOL OF ORIENTAL AND AFRICAN STUDIES

SCHOOL OF ORIENTAL AND AFRICAN STUDIES

ISBN 0-203-98571-0 Master e-book ISBN

ISBN - (Adobe e-reader Format)

I Buddhism and Philology 1

Word Index 158

Adikaram, E.W., 1953, Early History of Buddhism in Ceylon, Colombo.

and there is a battery of ancillary material—dictionaries, grammars, translations, etc.—

13 Walshe, 1987, 477.

14 Brough, 1962, 195 (ad GDhp 75).

20 See Norman, 1991C, 3–9 (=CP IV, 218–25).

21 See Norman, 1994, 209–24.

24 See Norman, 1983A, 92–106 (=CP II, 233–49).

33 yathāpi bhamaro paleti rasam ādāya, Dhp 49 abc.

37 See Norman, 1983C, 36.

39 It is being prepared by Dr Margaret Cone, Assistant Director of Research in Pāli Lexicography,

distinction between knower and known.

23 See Norman, 1981, 22, and Gombrich, 1990, 14.

Contemporary with the Buddha there was a growth of non-brahmanical

45 Norman, 1990A, 24–35 (26) (=CP IV, 124–38 [128]).

49 Although saññāvedayita is usually translated as a dvandva compound, this is not necessarily

61 See Jacobi, 1884, xix.

1 See Bechert, 1991A, 3–19 (10).

3 See Collins, 1992, 121–35.

4 Sv 13, 23–24; 15, 2–13.

If the material entrusted to the groups of was at first of a somewhat haphazard

verses, for a comparison of the versions of the verses ascribed to him in

7 See Norman, 1983C, 9.

8 In particular the Sarvāstivādins and the Mūla-Sarvāstivādins; see Schmithausen, 1987.

11 Tambiah, 1968, 100.

14 T.W.Rhys Davids, 1881, xxiii (quoted by Tambiah, 1968, 100–101).

21 See Oldenberg & Pischel, 1883, xiv.

24 Horner, 1954, 207–210.

1 na bhikkhave chandaso , Vin II 139, 13–14

2 Burrow, 1973, 123–40.

7 See Norman, 1989C, 369–92 (=CP IV, 46–71).

10 von Hinüber, 1991A, 183–93 (187).

12 See Lüders, 1954, §§ 89–90.

15 See Norman, 1976C, 119–21 (=CP I, 240–42).

17 By von Hinüber, 1993, 101–13.

derivation <ajje is supported by the occurrence in Jain Prakrit of , which is used

19 See Schwarzschild, 1961, 211–17 (=Collected Articles, 104–16).

27 See Bechert, 1991A, 12.

36 See von Hinüber, 1982, 133–40 (140).

1 89–77 B.C.E., according to Bechert, 199A1, 9.

4 See Norman, 1983C, 11.

6 von Hinüber, 1989A.

8 See Norman, 1976A, 35–42 (=CP II, 52–58).

21 Nidd I 89, 17–29.

30 Brough, 1962, 266–67 (ad GDhp 292).

34 See Report for 1883, JPTS 1883, xii.

As the shapes of the developed, it sometimes happened that pairs of

41 See Norman, 1978, 28–47 (32) (=CP II, 30–51 [34]).

1 For a survey see von Hinüber, 1989B, 341–67.

3 See PED, s.v. attaja.

4 Skt should have developed >* >*būhayati/būheti (with the - -/-u-

18 See Sasaki, 1986, 3.

23 bāhitapāpo tii , Dhp 388.

32 See Sander, 1985, 144–60.

errors or misunderstandings in the tradition.

36 See Norman, 1993, 113–23, where the relevant information is given.

37 cf. udaka-vaya, GDhp 317.

compound udaya-vyaya38 “arising and passing away”.

44 See von Hinüber, 1993, 101–13 (109).