442 183. COTTON
reduced to -pat should have retained the consonantic group -rp- of Skr. kärpâsi, instead of assimilating it in pp- as in Pali kappâsi. On the other hand, kupâi and *kappâi can easily be reconciled, since we have only to suppose a retrogressive action of the first -p-, finally melting into the new labial vowel. This is possibly what occurred in quite a different part of Asia with the modern form gupas of Burusaski. My conclusion is that the Indian name of the cotton which the Chinese received in the 5th cent. from the southern seas reached them neither through the intermediary of Jay. and Malay kapas, nor of an Indo-Chinese form similar to the modern Bahnar köpaih (where -h < -s, as in Cham), but as *kupâi *kappci , itself < kappâsi. It may be that *kupâi is to be traced to the Chams. In Cham, many words with a first -a- vowel have a doublet in -u : cf., for instance, kabav and kubav, kubau, « buffalo » (Jay. kébo, Mal. kérbav, Khmer krebéi, etc.), and, in the case of the word for « cotton », it is kapalt among the Chams of southern Annam, but kupah among those of Cambodia (CABATON, lot. tit. 57, 74). But I am far from being certain that kupah, the final -lt of which, like that of kapali, goes back to an earlier -s, can be ancient enough to provide the prototype of *kupâi (which moreover does not really represent kupah [--::*kapas], but is t` *kappâi ; kappâsi). As to the chi pei of the Liang shu, and any other which might be traced, they must be the result of scholarly corrections, after a graphic error had, in early Sung times, substituted in current use chi pei for ku-pei. If Jucen *kubu, Manchu kubun, Mongol kübüng and Goldi kufa are borrowed from ku-pei, we must not be surprised not to find kübüng in Mongolian during the Mongol dynasty, since the Mongols were then under the influence of Uighur culture, and adopted, as we have seen, the Uighur word käbäz. But ku-pei must have been borrowed from the Chinese at an early date, before chi pei became the only form in use in southern China. So it may be surmised that the *kubu which so far we know only in late Jucen was already in use when the Nü-chên ruled over northern China, even if it does not go back to the time of the Ch'i-tan.
PO-TIEH. — We have seen that the Liang shu mentions cotton as existing in the Turfan region in the first part of the 6th cent., and gives it the name po-tieh-tzü, tzic being a substantival affix of the spoken language; on the other hand, po-tieh can theoretically be conceived either as being a transcription (*b'vk-d'iep), or as simply meaning « white tieh (*d'iep) ». That the term was used at least by the numerous Chinese inhabitants of the country is expressly stated in the middle of the 7th cent. by Hsüan-ying, in a gloss repeated almost ad nauseam by himself and by Hui-lin; according to that gloss, which has a bearing on the various transcriptions of Prâkrit forms of karpâsa, « the people of Kao-ch'ang (> = the Turfan region) call [cotton] fit tieh » (Tokyo Tripit. of Meiji,
, VI, 2 b, 33 a, 59 a; Ix, 147 a); ftt tieh (*d'iep) is merely another graphic form of ' tieh. Of course, the use of tieh instead of po-tieh by Hsüan-ying is not in favour of the view that po-tieh is a transcription; it is, however, not conclusive, since it may be a scholarly abbreviation. In the same way, the author of the Hsin T'ang shu (222 c, I b), copying a passage in which the less literary author of the Chiu T'ang shu (197, I b) explained the difference between ku-pei and po-tieh, merely speaks of pei and tieh, although ku-pei is admittedly a transcription.
The main Western exponent of the theory according to which po-tieh would be a transcription was HIRTH (HR, 218), who proposed to see in this term, «certainly borrowed from one of the Turki languages » (this has passed into COULING, Encyclopaedia Sinica, I, 134), an old form similar to the
IIIF Color HighRes
