³ L. de Vallée, LaPoussin, Dynasties et histoire de l'lnde depuis Kanishlca jusqu'aux invasions musvlmanes, 1935, p. 1.Google Scholar

⁴ I do not mean to imply that in the book in question de La Vallée Poussin was himself misled in this way; and obviously he was thoroughly well acquainted with the generalities stated in the present paragraph. Their restatement may nevertheless be worth while, partly because his emphatic words, ‘ne vaut pas’, could be a distorting influence, and partly because later research has enabled us to understand more clearly the importance of the North-West. On the language, for example, see in particular Bailey, H. W., ‘Gāndhārī’, BSOAS, XI, 4, 1946, 765: ‘The preliminary studies of P. Pelliot [and others]… have hardly realized the importance of this North-Western Prakrit’.Google Scholar

⁵ Bisutun rock inscr., col. I, 6; Persepolis e; etc.

⁶ The few cases where later invaders penetrated much further into the subcontinent, and established their rule, are not directly relevant to the present discussion.

⁷ HHS, 118, Chav. 1907, 194, ff. The interpretation of the name Yüeh-chih is a vexed question (if the reader will forgive the meiosis) which, mercifully, lies outside the scope of the present article.

⁸ The precise locality varies in different sources.

⁹ HHS, 118, Chav. 1907, 194, 219. The latter passage also exemplifies the tendency of educated Chinese at this period to think of Buddhism as a variety of Taoism. Because of this accidental circumstance, we can hardly hope that Chinese sources might give reliable information on the date when Buddhism first arrived in China.

¹⁰ We leave aside as unproved the excessively early date (A.D. 67 or 75) traditionally ascribed to T, XVII, no. 784. There is still no evidence to contradict Demiéville's statement, ‘Les premières traductions de textes datées de manière relativement sụre remontent è la deuxième moitié du IIe siècle ap. J.-C’ (L'Inde classique, II, p. 398).

¹¹ Maspero, H., ‘Le songe et 1'ambassade de l'empereur Ming, étude critique des sources’, BEFEO, X, 1910, 95–130CrossRefGoogle Scholar, where the relevant texts are translated and discussed. Zürcher, E., The Buddhist conquest of China, 1959, 22, in citing this, mentions only the country of the Yüehchih (perhaps implying that ‘India’ in the other versions is a later ‘correction’, whether by authors or copyists ?). Another passage, from the Later Han Annals, has been held to prove that ‘dès l'année 65 de notre ère, le Bouddhisme était déjè en Chine une religion ayant ses moines et ses dèvots’ (Chav. 1905, p. 550, n. 1); this may be over-optimistic. See also Zürcher, op. cit., 19.Google Scholar

¹² Chav. 1905, 546; text in commentary to San-kuo chik, 30 (Peking, Chung-sua shu-chü, 1962, III, 859). I am admittedly hardly qualified to criticize the arguments which led Chavannes to take seriously the date in question. But, on general grounds, I cannot escape the feeling that such a date may legitimately be doubted. The text of the Wei-lüeh seems at first sight to say that it was the Chinese who taught the Yüeh-chih, which would be unbelievable. For a detailed discussion of this problem in relation to the versions in other texts, see Chavannes's note (ibid., 547–8). Whether the story refers to an embassy to China, or, as has sometimes been held, from China to the Great Yüeh-chih, is a question not relevant here. See also Zürcher, op. cit., 24–5.

¹³ And in addition, some six or seven Chinese. The figures given represent only the traditional ascriptions of translations which have survived, and are based on the information given in the ‘Table des auteurs et traducteurs’, Hôbôgirin, fasc. annexe (Tables du Taishd Issaihyô), pp. 127–52. Note also—to take a single example—that Chih-ch'ien eq4 (ibid., p. 148, s.v. Shiken), to whom some 50 translations are attributed, is said to have been born in a Scythian (Yüeh-chih) family which had already been domiciled in China for at least two generations. Even without regard to points like this, it would be naive to assume that all the ascriptions are equally trustworthy, and it must be made clear that no precision is claimed in this matter. Nevertheless, the tendency indicated seems beyond cavil. A farther point which may seem at first sight to be relevant, namely, that apparent ethnic designations could in some instances denote nothing more than the trivial relationship of spiritual or doctrinal ancestry (see Demiéville, in L'Inde classique, II, p. 400), does not materially affect the argument.

¹⁴ The key passage is the remark in the Later Han Annals (trans). Chav. 1907, 167): ‘A partir de la p´riode yang-kia (132–134 p.C), le prestige impérial tom ba graduellement; les divers royaumes (d'Occident) devinrent arrogants et negligents; ils s'opprimèrent et s'attaquèrent tour a tour les uns les autres’. And a few sentences earlier (A.D. 127), ‘ …les pays occupant les Ts'ong-ling [Pamir] et les régions plus è l'Ouest, rompirent (toutes relations avee la Chine)’.

¹⁵ The use of Brāhmī in the KuṣāḄa. inscriptions in districts such as Mathurā was of course inevitable; but it does not contradict the above assumption. There is no evidence that Brāhmī was ever used at this period in Central Asia as an established means of writing, and apart from a few isolated scraps, almost all the Central Asian Brāhmī manuscripts known are of the Gupta period and later. The palm-leaf fragments in Kuṣāṇa-type Brāhmī discovered near Kucha are only an apparent exception. The use of palm-leaves shows that the manuscript was written in India, and may have been transported to the region of Kucha at a much later date. (See H. Lüders, Bruchstüche buddhistischer Dramen (Königlich Preussische Turfan-Expeditionen: Kleinere Sanskrit-Texte, Ht. I), Berlin, 1911, p. 3 ff.).

¹⁶ So translated by Beal, S., Buddhist records of the western world, I, 56. The original (T, Li, no. 2087, 873c) has eq6. We should doubtless be slightly less sceptical if it were possible to maintain that chih eq7 indicated written records; but, in relation to India, Beal was probably right in his interpretation of the phrase as referring only to oral tradition.Google Scholar

¹⁷ T, IV, no. 203, 484b; transl. Chavannes, Cinq cents contes et apologues, III, 85–6.

¹⁸ Chav. 1907, 205. [Note that on this page, line 6, the date 116, which unfortunately has been repeated in print, is a misprint for 119.'ao, transl. Chavannes, T'oung Poo, Sér. II, Vol. VII, 1906, 232–3.

²⁰ Many of the points mentioned in the preceding part of this article have been noted and discussed (though often from a different point of view) by earlier writers. Special mention should be made of Sten Konow's ‘Historical introduction’ to Corpus Inscriptionum Indicarum, II, 1, Kharoshthī inscriptions, 1929; and L. de Vallée, La Poussin, L'Inde aux temps des Mauryas, 1930, 310 ff. These provide a useful conspectus of the bibliography of earlier stages in many controversies of this period of history, and much that must still be given serious consideration, even although not a few interpretations of detail must now be adjudged to be mistaken.Google Scholar

²¹ Apart from its use as such in the Kharoṣṭhī documents, the very considerable influence of this Prakrit is attested by the numerous loan-words from it which survived in other Central Asian languages.

²² See below, § 20.

²³ For the former, see t he introduction to my edition, The Gāndhārī Dharmapada; and for the latter, ‘A Kharoṣṭhī inscription from China’, BSOAS, XXIV, 3, 1961, 517–30.Google Scholar

²⁴ Serindia, I, 243.

²⁵ Chavannes, in Ancient Khotan, I, 538, where the fourth character is noted as conjectural. We can presumably take the reading of the name on trust; but unfortunately, in the published photographs of the piece of wood in question (ibid., II, plates CV, CXIV), no writing can be distinguished except by the application of more imagination than is desirable in such matters.

²⁶ Kh. 571, 590, 640; Serindia, I, 230, 260, 262, 266; photograph of the seal on the first of these (printed upside down), Serindia, IV, plate xx.

²⁷ This appears also to contradict the assertion in modern dictionaries (Tz'ŭ-hai, p. 1347; Tz'üan, p. 1493)—which do not give the source—that Shan-shan became a chün under the Sui dynasty. Presumably the evidence from the ground is to be taken as more trustworthy. But this problem I must leave to specialists in Chinese history.

²⁸ Chav. 1905, 537. For a summary of the Chinese information about Shan-shan/Lou-lan from the Han period until the T'ang Annals, see Stein, Serindia, I, 318–45, and 416–27.

²⁹ See Rapson, , Khar, inscr., III, 324–5.Google Scholar

³⁰ cf. EGP, 122. The alternative transcription in which the first syllable is written K.1056 Iâu is apparently attested only from later sources, and is not relevant for the original form of the name

³¹ See Henning, W. B., ‘The date of the Sogdian Ancient Letters’, BSOAS, XII, 3–4, 1948Google Scholar, 611. In view of this agreement, we need not attempt to decide here whether the Kharosthī spelling reflects a phonetic development or a derivative form with the suffix -ina. (For other examples of the latter, see Burrow, T., The language of the Kharosthī documents from Chinese Turkestan, § 77.)Google Scholar

³² Chav. 1905, p. 537, n.; Stein, Serindia, I, 325.

³³ Khar, inscr., III, 325.

³⁴ Burrow, op. cit., § 18.

³⁵ Quoted EGP, 231; T, I, no. 1, 47a. [Note that beside the development -mb- > -mm- Gāndhārī also has examples where the voiced stop is preserved, e.g., sabudha (Sk. sambuddha): see GDhp., p. 100.]

³⁸ Attested for the second century A.D., T, III, no. 167, 409c (An Shih-kao).

³⁷ This is doubtless to be preferred to the alternative reading Chu-mo, P. tsjo-mât; and if so, the rising tone here may in some manner reflect the -I- in the Prakrit spelling.

³⁸ Rapson, loc. cit.; EGP, 109.

³⁹ Serindia, I, 219. This seems reasonable on the map, and is not necessarily a t variance with the proposal to derive the name Charchan from the ancient form of Shan-shan: Hamilton, J., in T'oung Pao XLVI, 1–2, 1958, 121, cited in EGP, 109. I am in no position to judge the linguistic acceptability of this etymology; but this is not relevant here.Google Scholar

⁴⁰ Han shu, 96 (followed by Wei-lüeh), cited by Chav. 1905, 536.

⁴¹ Ancient Khotan, I, 435.

⁴² T, LI, no. 2087, 945c, where the footnote shows that the reduction to two syllables is a definite editorial choice.

⁴³ EGP, 90.

⁴⁴ There are admittedly discrepancies between the sources in the distances given; but no argument worthy of consideration can be based on figures which were, atthe best, very rough estimates in the first instance, and which have subsequently been subject to the risks of accidental corruption and the tampering of interpolators.

⁴⁵ Recognized as a place-name by Burrow, op. cit., p. 86. On the problems raised by the early Chinese transcriptions of the name of the Keriya region, see EGP, 88–9. The Hou Han shu, which calls the territory P. kiou-mye, gives the name of the capital as . As a possibility for consideration, it might be suggested that the second character here (being a common one in transcriptions) is a corruption, encouraged further by the name of the country, and that the name of the town might be emended to ney-ne/ (or *ney-nei/). This could represent the name of Niya, a town whose strategic importance must have been obvious. That it should sometimes have been held by the kingdom to the east, sometimes by that to the west of the river, would scarcely be a cause for surprise.

⁴⁶ EGP, 88, and Bailey, H. W., Khotanese texts, IV, pp. 37, 135 ff.Google Scholar

⁴⁷ T, LI, no. 2087, 945b.

⁴⁸ One of the king's titles, hinaza ‘general’ [in this context, ‘commander-in-chief’ would be more appropriate] is certain: Burrow, T., BSOS, VII, 3, 1935, 514Google Scholar; Khot. hīnāysä, Bailey, H. W., in BSOS, VIII, 2–3, 1936, 790–1.Google Scholar The idea that this title ‘would most probably have been conferred by the Chinese court’ is a guess without foundation, and should not have been allowed to appear in print (F. W. Thomas, BSOS, VIII, 2-3, 1936, 789). The vendor's name, Khvarnarse, has a decidedly Iranian appearance (cf. Pharnerseh, F. Justi, Iranisches Namenbuch, 93b), and Burrow, BSOS, VII, 3, 1935, 515, compared the first part of the name with Av. fCannah-, although he later withdrew this in favour of x"ar-, BSOS, VII, 4, 1935, 789.Google Scholar

⁴⁹ Henning, W. B., BSOAS, XII, 3–4, 1948, 615.Google Scholar

⁵⁰ Khar, inscr., III, 323 ff.

⁵¹ In Kh. 418(see Rapson, loc. cit.) the editors were uncertain whether the figure written was 36 or 46. In his translation, Burrow read it as 36, and this is provisionally accepted in the present discussion, since it fits better into the chronological schema outlined in the following part of this article. If, on the other hand, the higher figure is correct, this would not disprove the suggested chronology, but would only entail slight adjustments in the later part of the period. In the absence of other information, probability would naturally favour the shorter period: reigns of 46 years or more are by no means unknown elsewhere, but they are comparatively rare. A fresh inspection of the original is desirable, but, with the likelihood of progressive fading of the ink since it was read by Rapson and his colleagues, it would be only good luck if the figure could now be read with complete certainty.

⁵² It would be mere irresponsibility to suggest that Tajaka is the equivalent of a Sanskrit *Tarjaka ‘Threatener’, or to suppose that the existence of varsman- would justify a derivation of Vaṣmana from Vārsmazṇia ‘Possessor of virility’: Indian kings did not use personal names such as these. A Sanskrit verse in an anthology (Subhāṣitaratiiakosa, ed. D. D. Kosambi and V. V. Gokhale (HOS, XLII), 1957, no. 1448) is ascribed to an otherwise unknown Arigoka—a curiosity, but scarcely relevant here.

⁵³ The spelling Tajaka implies either -jj- or -rnj-, in contrast to the use of in the same document for the earlier single intervocalic stop, as in raja.

⁵¹ Quoted in commentary to San-kuo chih, 30 (1962 edition, in, 837–8) from a Wei shu. This passage does not permit us to decide whether the third syllable is part of the personal name or a transcription of a title in the indigenous language: cf. also Huai-t'au 1fa , ibid., 838.

⁵⁵ Wei-lüeh (ibid. 859, with K.tśiua\ (✶ca\), but this may be a misprint); Chav. 1905, 526, where T'an-t'o is preferred, although the alternative reading of the second character, tche (chih), is mentioned in a footnote. Karlgren, GSR, no. 795, gives only t'âk, but the other reading is noted in the Kuang-yün, spelt , i.e. cak (in Karlgren's notation, tśiäk). T'an-shih-huai is said to have died between A.D. 178 and 184 (W. Samolin, East Turkistan to the twelfth century, 1964, 45), and is therefore not the same person as T'an-chih, if the Wei-lueh is correct in referring the latter to ‘the juncture of Han and Wei’: . This is an elastic expression, it would seem, since Chavannes explained it in a footnote as ‘entre 200 et 220 environ ap. J. - C.’. But these matters are outside our province. What is relevant is that in each case the reference is to a leader of the Hsien-pi tribe ; that the older forms of T'an-shih and T'an-chih could be transcriptions of the same foreign name; and that the Hsien-pi had at this period extended their territorial control to the south-west as far as Tun-huang and the Lop region—close enough in space and time to admit the suggested comparison of names as at least a possibility.

⁵⁶ In Kh. 709 the editors presented a further alternative, aṃgoṃka, but the published facsimile (Stein, Innermost Asia, plate XVIII) does not seem to justify the second anusvāra. The spelling aṃgoṅka, which appears twice in the edition, is not in any case to be taken as identical in intention, and was rejected by Burrow, Translation of the Kharoṣṭhī documents, no. 418. A form of the name with a nasal in the second syllable can thus be considered as unsupported up to the present.

⁵⁷ Thomas, F. W., in Ada Orientalia, XIII, 1935, 49, 50, quoted by Burrow, Lang, of the Khar. documents, p. 71. Because of the date, the reference could surely not have been to the An-chou ?ân-ciu mentioned in Accounts of Western nations in the History of the Northern Chou dynasty (transl. R. A. Miller, 1959, 8) who attacked Shan-shan in A.D. 442.Google Scholar

⁶⁸ Chav. 1907, 173 and 171.

⁵⁹ Or so it would appear: de L a Vallée Poussin, L'Inde aux temps des Mauryas, 371.

⁶⁰ CII, II, 1, p. lxxiv.

⁶¹ See § 5(e), with n. 18.

⁶² CII, II, 1, p. Ixxiv.

⁶³ San-kuo chih, 3 (1962 edition, I, 97).

⁶⁴ The identification of the name has occasionally been doubted, but the general opinion seems to be that the Chinese form is at least a possible transcription of Vāsudeva, or rather, of a Prakritic version of the name, even if it may not be an entirely adequate attempt. (Note that the initial pâ has a parallel in the common transcription , pâ-lâ-nâi\, Sk. Vārāṇasī.) The real difficulty is not this. Accept that the name Vāsudeva was meant; but consider whether, from a comparable (hypothetical) reference in an Oriental text to a Western European king called Henry (who might be Heinrich), and with no more data than we have for Vāsudeva, we would dare to guess which person was meant.

⁶⁵ Texts in Konow, ‘The oldest dialect of Khotanese Saka’, Norsk Tidsslcrift for Sprogvidenskap, XIV, 1947, 156 ffGoogle Scholar. For the first of these documents, see also Bailey, H. W., ‘Languages of the Saka’, Handbuch der Orientalistik, IV, 1, 1958, 153. Konow proposed a date not earlier than the seventh century for these documents (loc. cit., 160): and it is of some interest to observe that, while the royal titles appear in Iranian, the first document includes among them jezdampura, a translation of the old Kuṣāṇa title devaptitra, and so rendered by Konow.Google Scholar

⁶⁶ See Rapson, , Khar, inscr., III, 325; Kh. 213–98; Stein, Ancient Khotan, I, 338 ff., 399 ff.; Chinese documents edited and translated by Chavannes in Ancient Khotan, I, 537 if. The original of the dated document is clear and unambiguous, ‘fifth year of the period t'ai-shih ’:Google Scholaribid., II, plate CXII.

⁶⁷ Kh. 327: the name of the king is lost, but, as Rapson showed (Khar, inscr., III, 326), the document must belong to the reign of Amgoka, because of the occurrence of the name of the kitsa'itsa Varpa (and, it may be added, other proper names).

⁶⁸ EGP, 67–8.

⁶⁹ In appendix to Stein, Ancient Khotan, I, 537.

⁷⁰ Cinq cents contes et apologues, II, p. 336, n.

⁷¹ Chin shu, 3 (Ssŭ-pu pei-yao ed., second ed., 1936–7), Shanghai, Chung-hua shu-chü, n.d., 326); mentioned by Chavannes, Ancient Khotan, I, 537.

⁷² In BSOAS, XXIV, 3, 1961, p. 530, n. 2, I remarked briefly that, on the basis of Rapson's arguments, the period covered by the ‘Niya documents’ could hardly be earlier than approximately A.D. 180–275 or later than approximately A.D. 230–325. It will be seen that the arguments now put forward are in favour of a dating very close to the later of these. The earlier, it would seem, can now be excluded, since anything at all close to this would imply that Amgoka's acknowledgement of the overlordship of China would have taken place during the last declining decade of the Han dynasty, which would be incredible. Some time ago, Professor D. C. Twitchett kindly informed me of an article by Nagasawa Kazutoshi, in Shigaku Zasshi LXXII, 12, 1963, 1–26, where a still earlier dating was proposed: between A.D. 112 and 264. Unfortunately, the volume in question is not accessible to me at the time of writing. The final date given seems to have been suggested by the fact that the Western Chin came to power in A.D. 265; but such an assumption, and indeed the whole theory of so early a date, would appear to be impossible to reconcile with numerous points of the evidence.Google Scholar

⁷³ The proportions of the two are reversed, Kharoṣṭhī documents from Lou-Ian being very few in comparison with the Chinese. The site L.A.vi.ii, which produced a small number in Kharoṣṭhī as well as many more in Chinese, was a large rubbish-heap measuring about 100 feet by 50 feet (Serindia, I, 381), and is not strictly comparable to the waste-room N.v.xv at Caḍota. One paper fragment has Chinese on one side and Kharosthi on the other (Kh. 699; Chav. Docs., 918), but from what little can be read, there seems to be no connexion between the contents of the two sides. The two Kharoṣṭhī fragments from Lou-Ian found by Hedin (Conr., p. 191) add no useful information.

⁷⁴ The statement that the earliest date in these documents is A.D. 252 (Conr., p. 5) is misleading. The document in question (Conr. pp. 93–4, no. 16, 1) does in fact contain this date, but in addition the date A.D. 264. The year of writing, therefore, cannot be before 264, and the earlier date is merely referred to as part of the contents of the letter.

⁷⁵ Dated in the last two-year periods of the Wei, e.g., Chav. Docs., 738, ching-yüan 4 (A.D. 263); no. 721, [chinq]-yüan 5 (A.D. 264); nos. 722, 730, hsien-hsi 2 (A.D. 265). The date in Conr., p. 127, no. 52, hsien-his 3 (misprinted in text as ) is illegible in the published photograph, but the same date occurs also in no. 64 (ibid., p. 129), where the photograph is not subject to doubt. This is an unsolved problem. In spite of Conrady's footnote (p. 127), the situation was not parallel with that responsible for the aberrant style of date in Chav. Docs., 886. (See below, p. 604, n. 77.)

⁷⁶ In addition to notes concerning grain-rations issued to soldiers and others, the Chinese documents contain frequent references to various officials concerned in the administration of the ‘Department of Corn Supplies’, ts'ang-ts'ao : Masp., 214; Conr., p. 126, and in many others. (One of the taxes mentioned in the Kharosthi documents is the tsamghina/tsamgina tax. This is one of the few words in the administrative vocabulary which may be of Chinese origin, though presumably borrowed at an earlier period. If so, we might translate the term as ‘granarytax’, assuming the addition of the adjectival suffix -ina to *tsaṃgha, as a loan-word from Chinese tshâ ‘granary’.)

⁷⁷ On the later dates in the series, see Chav. Docs., p. iv. In Conr., p. 98, in document no. 20, 1, of the year A.D. 310, the date 330 in the translation is a misprint: cf. text, p. 99 (where is an independent misprint for ). The date A.D. 330 is in Chav. Docs., 886, where it is written as ‘the eighteenth year of chien-hsing’, which period, however, was in fact A.D. 313–16. In his note on this use of a year-period long since past, Chavannes remarked that the Chin dynasty ‘a dû vraisemblablement renoncer à toute visée politique dans les pays d'Occident, [et] il est assez naturel que le petit poste chinois abandonné à ses propres ressources au Nord du Lop nor ait continué à se servir du nien-hao dont l'abrogation n'avait pu lui etre notificiée’. This seems to be the only possible explanation of the anomaly; and it follows as a corollary that these were lean years for the Shan-shan country also.

⁷⁸ San-kuo chih, 2 (1962 edition, I, 79), third year of huang-ch'u = A.D. 222. (In Conr., p. 4–5, the year 220 is doubtless a misprint.)

⁷⁹ ‘Anc. L', II, 37, ’ yntcwt, swySykt: Henning, W. B., BSOAS, XII, 3–4, 1948, p. 603, n. 3, where other connexions between the Sogdians and the Lou-Ian country are noted.Google Scholar

⁸⁰ On the interpretation of the name, see Henning, ibid., p. 606, n. 7, and 609.

⁸¹ Ancient Khotan, I, 540.

⁸² Serindia, I, 411.

⁸³ Chav. 1905, 527.

⁸⁴ Chav. Docs., 892, where the word is printed huân/, with a note to say that in the document it is written with radical 140, ‘comme cela a toujours lieu dans ces textes’.

⁸⁵ Chav. Docs., 846, ‘le tche hou Yin Tō’, the first character of the name having the alternative reading Cien (GSR, no. 450).

⁸⁶ Chav. Docs., 763.

⁸⁷ In Ancient Khotan, I, 542.

⁸⁸ Conr., p. 106, no. 28, 2.

⁸⁹ See Demiéville, , in L'lnde classique, II, p. 411; Hôbôgirin, Supplément au premier fascicule, p. xv; Mochizuki Shinkō, Bukkyō daijiten, I, 729c.Google Scholar

⁹⁰ Completed in A.D. 807: Mochizuki, op. cit., VI, 182.

⁹¹ T, LIV, no. 2128, 675a. On the initial i- (P. y-) see EGP, 115, where a transcription value of a voiced palatal fricative [A] is accepted for early Buddhist transcriptions. This is readily justified in many instances: e.g., a Kharoṣṭhī spelling *kośka doubtless implies a pronunciation such as [koźiya]. (It may be prudent, nevertheless, to leave open the possibility that in other instances i-, was used to represent the actual value of the Indian ‘semivowel’ y.) The same Chinese initial, however, also represents a voiceless Indian ś- in the transcription of Śuddhodana already quoted ( §9), where the expected voiceless fricative is confirmed by the Khotanese spelling Śśādūtana. Against this, a few transcriptions are cited (EGP, 68) where an unexplained initial Chinese ź- (K. dź'-) occurs instead of the regular and frequent Chinese ś- for Indian initial ś-. There is a problem here which awaits further investigation.

⁹² T, LIV, no. 2128, 420a.

⁹³ Naturally, this is not always so. One would expect considerable variation between individuals both in aptitude and in the degree of importance which they attached to accuracy in this matter. Some ‘transcriptions’ in the editions are so fantastic that they must have suffered serious corruption in the textual transmission. Other reasons for divergent transcriptions include such factors as regional differences in the pronunciation of Chinese at the time (see for example EGP, 214), and the not infrequent tendency of copyists and editors to make adjustments in order to give a closer approximation to the practice of later periods. Some transcriptions, imprecise if from Gāndhārī, would be regular if they represent some other Prakrit; but only sporadic examples of this have been observed so far, and would seem to depend on loan-words within Gāndhārī texts: cf. suyi (Sk. śuci), more frequent than suyi (see GDhp., p. 101). Further research will be needed before a more confident answer can be given to this question.

⁹⁴ It is to some extent a matter of accident that the Dirghagama has figured somewhat prominently in discussions of this question. Waldschmidt, E, Bruchstucke buddhistischer Sūtras, I, pp. 231 ff., compared some of the transcriptions with the language of the Dharmapada; but the position remained unclear, since, alongside the reconstructions based on Karlgren, he printed a conventional approximation to the underlying Indian forms, but with frequent use of t for d, for example, and other inconsistencies (op. cit., 166 ff.). See also GDhp., pp. 50 ff.Google Scholar

⁹⁵ The comparative studies of parallel versions of passages from these and other texts, in Przyluski, Le Concile de Rājagrha, and Hofinger, Étude sur le Concile de Vaiśāī, were of course written with other problems in view; but they contain a number of names where the earlier texts can be seen to transcribe Prakritic forms, while the later tend more towards the Sanskrit.

⁹⁶ For references to the texts, see Akanuma Chizen, Indo Buklcyō hoyū meishi jiten, s.v.

⁹⁷ T, XII, nos. 360, 361, 362. If the traditional ascription of T, 361, to the Yüeh-chih ‘Lokaksema’ is correct, this would be roughly a century earlier than the other two; but the attribution has been doubted.

⁹⁸ Professor É. Lamotte has suggested to me in a letter that the sense of ‘knowing, understanding’ may be due rather to the prefix Ājnāta-, and has kindly drawn my attention to later renderings of the full name (Akanuma, op. cit., 43) where a transliterated version of Kauṇḍinya is preceded by or , i.e. by a translation of Ājnāta-. It could then be suggested that f or are parallel to these. This is of course a possibility, but it would not affect the essence of the argument presented above. It would indeed imply that the element ‘knowing’ was doubly represented, for I can see no way in which a form corresponding to Kauṇḍinya as a whole could give rise to . Either, then, the Chinese translators in question merely ignored the final syllable, or were content to understand the whole of this part of the name as koṭi, or else they ignored Ajnāta-. I suspect, however, that the parallelism is only apparent. It may be noted that T, 362, which transliterates in full many longer names, gives here only , and this suggests that all three of the early translations had a recension of the text in which Ājnāta.- was missing.

⁹⁹ Suzuki, Index to the Lankāvatāra Sūtra.

¹⁰⁰ For references, see Critical Pali dictionary, s.v. anta.

¹⁰¹ For example, Fa-hsien (T, LI, 857a); Sung-yün (T, LI, 1018c). I am indebted to Mr. S. Weinstein for a reference to T, L, 331b, and Chin shu, 114 (K'ai-ming edition, II, 301c: Shanghai edition, 901b), which mentions the kings of Shan-shan and Chien-pu in the year A.D. 381. By this date, Shan-shan must have designated a country of very much smaller extent than that of the period of our documents—presumably only the eastern portion of the old kingdom.