the master mind, even measuring its mastery on the Buddhist standard of relevancy. The texts of the Khuddaka Nikāja still less admit of reference to any one period; they contain old matter and new, even within the limits of one and the same text. That we have in the Udāna or Itivuttaka the actual words of the Buddha is wholly implausible, and old as the Suttanipāta is reckoned, we find already the Buddha surrounded with an elaborate mythology. The Jātaka book is a strange conglomerate of old and new verses with new prose; some of its tales, as we know from Buddhist sculpture and a stray citation or two, go back to the Asokan epoch or shortly after; as folk-lore its contents are often of undeniable age, but as Buddhist fables their antiquity is uncertain.

The case of the Abhidhamma Pitaka is far worse than that of the Sutta and Vinaya Pitakas. These two texts know nothing of such a Pitaka; the division recognized is that of Dhamma, i. e. Sutta, and Vinaya with Mātikā, 'lists' as the third, the germ, doubtless, of the actual Abhidhamma.' None the less the most grandiose claims have been advanced for the Abhidhamma books. Mrs. Rhys Davids' claims that the Dhammasangani is to be dated about 885 B. C., basing the assertion on the internal evidence of form and content as compared with the Kathāvatthu (247 B. c.). Professor Walleser3 denies the validity of the argument, but resorts to the view that the Dhammasangani is the Dharmasaṁgraha of Çariputra among the northern Abhidhamma texts, and is referred to by Asoka as the Upatisa-pasine, Upatissa being the other name of Sariputta; the suggestion is impossible of credit, and it is certain that the Dharmasamgraha is wholly different from the Dhammasangani. A useful antidote to these extravagant estimates is afforded by Professor Rhys Davids who places the four Nikayas at the period assigned to the Dhammasangani and brings that text down later than the late texts of the Khuddaka Nikaya. He himself, however, asserts that in the subjects of which it treats and its style the Kathavatthu, the latest text of the Pitaka, accords perfectly with all we know or expect of Asokan India, but the assertion is idle and is unsupported by any evidence. On the 2 Psych. Ethics, pp. xviii. f. SBB. II. xi.; CHI. i. 194.

Geiger, PD., pp. 118 f.
PGAB., pp. 20 ff.

CHI. i. 197.

contrary the work shows the clearest signs, as Mrs. Rhys Davids' candidly allows, of much addition, which has deprived it of coherence or order. The admission is necessary in view of the obvious confusion of the work, and it deprives of all point the only serious attack made on Minayeff's criticism of the traditional origin of the Kathāvatthu; the inclusion in the text of a heresy known to have been held by the Vetulyakas in the second century A. D. in Ceylon and later does not prove that the work was only begun then, but it indicates that in all probability it was still open to additions in the second century A. D. and later. Moreover the theory that it is the latest of the Abhidhamma books is without foundation; it ignores three of them, Dhatukathā, Puggalapaññatti, and Yamaka. The scholastic character of these works suggests that they are divided by no small space from the other Pitakas, and are very possibly younger than the older portion of the Milindapanha, which refers to the Abhidhamma only in passages certainly late. The lateness of the Abhidhamma is confirmed also by the Ceylonese tradition itself. When it tells of the Great Council held by the heretical Vajjiputtakas, it says that they rejected the Abhidhamma books, along with the Patisambhidā and the Niddesa and portions of the Jātaka from the fifth Nikaya, and the Parivara appendix of the Vinaya. Now it is not disputed that the Patisambhida and Niddesa, which are really commentaries or Abhidhamma, are late, that the Jātaka is full of late matter, and that the Parivara is not original; we are confirmed, therefore, in the view that the Abhidhamma has no claim to the antiquity asserted for it. This is supported by the undeniable fact that, while the Sutta and Vinaya Piṭakas have parallels in other schools, based on a common tradition, the Abhidharma of the Sarvāstivādins, of which we now have information, utterly disagrees with the Pali Abhidhamma. As this Abhidhamma existed at the time. of the Council of Kaniska, it is doubtless to it and not to the Pali Abhidhamma that we have to refer the term Trepitaka, which Points of Controversy, p. xxxi; cf. Poussin, ERE. iv. 184.

2 Barth, RHR. xlii. 73.

4

Takakusu, JPTS. 1904-5, pp. 67 ff. A Dharmagupta Abhidharma also existed. JA. 1916, ii. 20, 38.

◄ EI. viii. 176. It is important to note that DN. no. 83, Sañgīti Suttanta, is

appears in an inscription of this time, and we are left to conjecture a date for the Pali Abhidhamma. All that can be said is that in the third century A. D. the Abhidhamma Pitaka seems to have been studied in Ceylon,' and that in the commentators of the fifth century we find the Abhidhamma Piṭaka treated as authoritative, as also in the late additions to the Milindapañha. But we are without means of judging precisely at what date the old Matikās were formed into our present texts, possibly after the Christian era.

The place of the production of the Pali Canon is uncertain; it comes to us as that of the Vibhajjavädin school of the Mahāvihara of Ceylon; and its connexion with Ceylon. is recognized in Sanskrit texts. But that is not to say that the Pali Canon was redacted in Ceylon; we need credit the Ceylonese tradition of the early conversion of the people to Buddhism as little as we do any other part of its legends. We may, therefore, treat the Canon as the work of an Indian school, and note the Ceylonese tradition recording its reduction to writing' under Abhaya Vaṭṭagāmani, perhaps at the close of the first century B. C., as perhaps applicable to the two older Pitakas, though not necessarily in their present form in detail; certainly not in the case of the Jataka book. It is a different question whether the Abhidhamma Pitaka was a product of Ceylon; there is nothing to prove its northern origin, and accordingly the matter must remain open, in view of our almost total ignorance of the facts on which a conclusion of value could be framed.

Part of this ignorance arises from the uncertainty of the origin of Pali, the language of the Canon. One fatal objection to the orthodox theory, and to the attempts of recent authorities to defend it, is the fact that the language of the Canon is plainly and undeniably a post Asokan literary dialect, assuming much communication between the learned monks of different parts of India. Even if we

parallel with the Samgītiparyāya of the Sarvāstivādins; this suggests that it was the Sarvāstivādin example which induced the Vibhajjavādins to develop a separate Abhidharma. For other Abhidhamma matter in the Sutta Pitaka see MN. i. 299 ff.

Hardy, Eastern Monachism, p. 156.

This record is not favourable to the theory that Pitaka means box for MSS. (Bühler, Ind. Alph., pp. 86 ff.) Cf. SBE. xxxv. 28, n. 1.

* See Lévi, JA. 1912, ii, 495 ff. (Geiger's reply (Pāli, pp. 1 ff.) is ineffective.)

2

4

assumed that the Canon was established by Asoka, we would have to admit that it has been radically altered in form, and it would be absurd to claim that the alteration in form had taken place without change of substance. The complexity and artificiality of the language as it stands are shown by the extraordinary diversity of the suggestions as to the vernacular which must lie at the basis of Pali. Oldenberg1 found it in the region south of the Deccan; Rhys Davids, on historical grounds, in Avanti; Franke, on linguistic considerations, in the tract whose centre was Ujjayinī; while Grierson, accepting Windisch's contention in favour of the Indian tradition which makes Magadhi the basis of Pali, finds that Pali is the literary form of the Magadhr language, the then Koine of India, as it was spoken and as it was used as a medium of literary instruction in the Takṣaçila University, the vernacular of Takṣaçila being Paiçãcr Prakrit, whose home he places in the north-west though Konow locates it about the Vindhya region; or again it is held that Ardha-Magadhi underlies our Pali texts. The obvious deduction is that Pali came into being, such as we have it, by a slow and complex. process occupying some centuries, and variations of place.

2. The Conclusions attainable

If we adopt, as we must if not precluded by obligations of faith, the conclusion that the Pali Canon came into its present shape long after the death of the Buddha, the question presents itself whether any effective result can be achieved in selecting parts of the Canon as earlier and more authentic than others. The results attainable" in this regard are from our point of view of negligible importance; the fifth Nikaya contains miscellaneous

Cf. Grammont, Mél. Lévi, pp. 65 ff. Rhys Davids (SBE. xxxvi. 269) suggests a Ceylonese origin of the Parivāra.

1 VP., p. liv.

2 CHI. i. 187; cf. Buddhist India, pp. 140 ff.

3 Pali und Sanskrit, p. 188.

♦ Bhandarkar Commem. Essays, pp. 117 ff.; Windisch, Actes du XIV Congrès International des Orientalistes, i. 277. But see SBA. 1913, pp. 1003 ff.

' ZDMG. lxiv. 114 ff. Paiçācī is attributed to the Sthaviras by Wassilieff, Bouddhisme, p. 268; cf. Lacôte, Bṛhatkathā p. 44.

See references in Winternitz, Ind. Litt. ii. 51, 53, 365; Rhys Davids, Buddh. Ind., pp. 165 if.

matter, some old in part at least, like the Suttanipata, Udana, Thera- and Theri-gāthās, Dhammapadu, and Itivuttaka, some new like the Peta and Vimana-vatthu, the Buddhavansa, Apadāna, and Cariyapiṭaka; the Abhidhamma is clearly a late and deliberate working over of the Dhamma in its technical aspect. But the bulk of the Sutta Pitaka and the Vinaya Pitaka represents the same stage of ideas; they may have been redacted contemporaneously, and any attempt to trace strata of diverse ages involves a determination on other grounds of what elements should be early and what should be late. On the theory of Buddhist rationalism, we can decide that all the supernatural element is secondary, but we are faced with the insistent question whether we have any right thus to proceed. Granted that a preacher of a gospel of pure reason may be deified by some accident, it is at least certain that it is much more easy to deify an inspired seer who deems himself to be an embodiment of the divine, and that India in particular has been specially prone to accept as real such embodiments. If we reason a priori, and lay aside our natural desire to modernize and to find reason prevailing in a barbarous age, we should rather expect to find that the Buddha was one who was indeed human, but who at the same time felt himself to be, and was regarded by his followers as, something far superior to humanity, a great divinity in the eyes of his followers, a deity even to those who were not of the chosen circle. The conclusion is doubtless embarrassing to rationalism, but, if we are content to seek historical truth, we must be prepared to shed our personal predilections and to accept the conclusion which evidence indicates. The case for this view is greatly strengthened by the nature of the texts of the Pali Canon. The Vibhajyavādins were plainly prepared to rationalize as far as practicable;1 as opposed to other schools they minimize the supernatural element in Buddhism, and the salient fact is that even in the records of these would-be rationalists we find abundant proof that the orthodox and prevalent view of the Buddha made him far removed from ordinary humanity.

1 Cf. Barth, RHR. xlii. 57; Poussin, Bouddhisme, Études et Matériaux (1898), p. 42; Bouddhisme, pp. 216 ff; cf. ERE. i. 95.