440 183. COTTON
« pairs » (cf. PRZYLUSKI, in JA, 1919, I, 406). Then comes the Liang shu (ch. 54), which always speaks of chi-pei, but the Nan ship, in copying it, always gives ku-pei (except once, 79, 7 b, where it gives chi-pei in the notice of K'o-p'an-t'o west of Khotan; but it may be a misprint), and so does the quotation from the Liang shu in the T'ai-p'ing yü-lan (788, 15 b). It is also ku-pei, and ku-pei alone, which we find in the Sui shu (82, 3 a) and in both T'ang shu (Chiu T'ang shu, 197, 1 a and b; Hsin T'ang shu, 222 c, 1 a and b), as well as in the great compilations of T'ang and early Sung times, such as the T'ung tien (188, 4 b), the T'ai-p'ing huan-yu chi (176, 3 a, 3 b-4 a, 12 a and b; 179, 19 a, in the section of P'o-ni [Borneo], while the later Sung shih, in an otherwise identical sentence, gives chi-pei), the T'ai-p'ing yü-lan (ch. 788,
sometimes with a corrupt form a ku-chit, which is certainly a misprint) and the Ts'ê fu
yüan-kuei (959, 21 a). The statement in the P'ei-wên yün fu (s. v. «chi pei »), blindly followed by the Tz'û yüan, and also in the T'u-shu chi-ch'êng, that the Nan shih and the Hsin T'ang shu give chi-pei is nothing but an error. In a note in the China Review, xix, 191, PARKER
cites the A, t Chiu yü chih as speaking of ku-pei in a notice on an Indonesian state which
he supposed to date from the Sung dynasty (cf. also LAUFER, Sino-Iranica, 471; FERRAND, in JA, 1921, II, 287). This would be the latest independent use of the form ku-pei found hitherto. Unfortunately, PARKER'S information is certainly second hand, taken, it seems, from some dictionary or encyclopaedia which he did not trouble to mention. The only Chiu yü chih I know of is the Dian -fag chiu-yii chih, dated 1080 (cf. Ssu-k'u ... , 68, 7-9), but I can find in it no passage recalling the one mentioned by PARKER. In any case, the passage is not original, but, with a mistake in the name of the state 4 .tK P'o-ssû instead of 4 *1J P'o-li), is merely copied from the notice of P'o-li in Nan shih, 78, 6 b; the true title may have been Chiu-chou chi, a work written prior to 527, on which cf. infra, p. 444. In the Buddhist scriptures too, I have found only ku-pei, occurring in Paramârtha's translations of the Abhidharmakosa (NANJIO, No. 1269; ch. 9; Tokyo Tripit. of Meiji, ice, i, 54 b) and of the Fo a p'i-t'an ching (NANJIÔ, No. 1108; ch. 2; g, x, 16 b) which are of the end of the 6th cent.; on that account, ku-pei is commented upon by Hsüan-ying in his I-ch'ieh ching yin-i (ibid. n, vi, 71 a; vII, 72 a), and Hsüan-ying's text has been taken over by Hui-lin in his ch. 70 (ibid. ix, 189 b). Of course, chi and ku are graphically so much alike that one could easily have taken the place of the other; moreover Chinese texts have often been revised by scholars, and we know cases of wrong forms which have been substituted wholesale for the correct ones, such as t'ieh-lo which always occurs in ancient Chinese texts instead of the correct t'ieh-ch'in, Avar and Turk. tegin. But, in the present case, the agreement of so many sources is the more impressive because it extends to Buddhist texts. Nevertheless, on account of the chi-pei of the Liang shu and of the later use of chi-pei under the Sung, I should have hesitated to express any definite opinion, if it were not that the correctness of ku-pei can be established, I believe, by comparison with another transcription.
The word karpasa is of frequent use in Buddhist texts, so that it often occurs in their Chinese
translations, though in different guise, once asf j ft chieh-pei-so (*kivp puâi-sti; mentioned in
connection with the flower of the plant; cf. Tokyo Tripit. of Meiji, ,t , VIII, 164 a), more usually as
# chiehpo-yü (•kkvp-pud-iuk; cf. ibid., vt, 2 b, 33 a; vrl, 2 b; VIII, 117 b) and -ft chieh-
IIIF Color HighRes
