490 SlNO-lRANICA 



is not part of the transcription, any more than the word $$& kin, which 

 precedes it in the Sui Annals; but the combination of both po and kin 

 with tie indicates and confirms very well that the latter was a brocaded 

 silk. HiRTH 1 joins po with tie into a compound in order to save the 

 term for his pets the Turks. "The name po-tie is certainly borrowed 

 from one of the Turki languages. The nearest equivalent seems to be 

 the Jagatai Turki word for cotton, pakhta" There are two fundamental 

 errors involved here. First, the Cantonese dialect, on which Hirth 

 habitually falls back in attempting to restore the ancient phonetic 

 condition of Chinese, does not in fact represent the ancient Chinese 

 language, but is merely a modern dialect in a far-advanced stage of 

 phonetic decadence. The sounds of ancient Chinese can be restored 

 solely on the indications of the Chinese phonetic dictionarie^and on the 

 data of comparative Indo-Chinese philology. Even in Cantonese, 

 po-tie is pronounced pak-tip, and it is a prerequisite that the foreign 

 prototype of this word terminates in a final labial. The ancient pho- 

 netics of & H is not pak-ta, but *bak-dzip or *dip, and this bears no 

 relation to pakhta. Further, it is impossible to correlate a foreign 

 word that appears in China in the Han period with that of a com- 

 paratively recent Turkish dialect, especially as the Chinese data rela- 

 tive to the term do not lead anywhere to the Turks; and, for the rest, 

 the word pakhta is not Turkish, but Persian, in origin. 2 Whether the 

 term tie has anything to do with cotton, as already stated by CHA- 

 VANNES, 3 is uncertain; but, in view of the description of the plant as 

 given in the Nan Si* or Lian $u? it may be granted that the term po-tie 

 was subsequently transferred to cotton. 



The ancient pronunciation of po-tie being *bak-dib, it would not be 

 impossible that the element bak represents a reminiscence of Middle 

 Persian pambak ("cotton"), New Persian panpa (Ossetic bambag, 

 Armenian bambak). This assumption being granted, the Chinese term 

 po-tie ( = Middle Persian *bak-6ib = pambak dip) would mean "cotton 

 brocade" or "cotton stuff." Again, po-tie was a product of Iranian 

 regions: kin siu po tie 4k SI S & is named as a product of K'aii (Sog- 

 diana) in the Sasanian era; 5 and, as has been shown, po-tie from Parthia 



1 Chao Ju-kua, p. 218. 



2 STEINGASS, Persian-English Dictionary, p. 237. 

 8 Documents sur les Tou-kiue occidentaux, p. 352. 

 4 Ch. 79, p. 6 b. 



6 Ch. 54, p. 13 b. Cf. CHAVANNES, ibid., p. 102; see also F. W. K. MULLER, 

 Uigurica, II, pp. 70, 105. 



Sui $u, Ch. 83, p. 4. Hence *bak-dlb may also have been a Sogdian word. 



