123. CASCAR 211
like the Käncäk» (BROCKELMANN, 104). Part of the Käncäk may have settled in the region of Talas, but Kàsyari's text leaves no doubt that Käncäk people also lived in the districts of Kâsyar, and this is confirmed from other sources. The name occurs after that of a man from Qaml = Qamul (see « Camul ») in the Manichaean fragment studied by HANEDA (Mem. of the. . . Toyo Bunko, No. 6, 3), although the editor does not seem to have seen in it a proper name. In the Tibetan chronicle of Khotan, The Annals of the Li Country (THOMAS, Tibetan Texts and Documents, i, 118), the « king of the Ga-lag» invades Khotan territory, but is defeated; the king of Khotan sends him back to Su-leg, i. e. Kâsyar. That Ga-'jag is meant to render a pronunciation *Ganjag, and that the king of the Ga-'jag is the king of Kâsyar do not seem open to doubt, and CLAUSON (JRAS, 1937, 178) has already connected Ga-'jag with the name given by Kâsyari to the language spoken in the Kàsyar area. CLAUSON adds that, as a consequence, the name in Kâsyari ought to be transliterated Ganjak, and no longer Käncäk. «The name », CLAUSON adds,
is clearly Iranian, and the same as Ganja, the name of another well-known town at the other end of the Iranian area in the Southern Caucasus ». I am afraid that such far-reaching conclusions are still premature. It is true that there are no -j- in Kâsyari's Turkish, or, to be more correct, that Kâsyari does not distinguish between _ c and : y, and always writes I, but with the value of e"; J was foreign to Middle Turkish. Consequently, we are right when we read Käncäk in Turkish that which is written Känjäk by Kâsyari, even though the original name, if it be non-Turkish, may have actually been Känjäk (moreover, a foreign -j- often gave in Middle Turkish -z- or -z-). Very much the same may he said of g and k; an initial g- is quite exceptional in Middle Turkish; but in such a case, and also sometimes to prevent a confusion between words written identically, Kâsyari indicates that the word must be read with g-, not with k- (cf. BROCKELMANN, V, 61) ; no such remark is made about Käncäk. On the other hand, the Tibetan spelling does not permit a choice between Käncäk and *Ganjag. There are many cases when a surd initial of a foreign name is rendered as a sonant in Tibetan; for instance the very name of the Turks becomes Drug, Dru-gu, Drug-gu in Tibetan. Tibetan writing does not admit of any final surd occlusives, so that a final -k is always transcribed -g in Tibetan. The vocalization Ga-'jag and not *Ge-leg may be more significant; still, Tibetans have not always been very strict in that respect; their e is not the Turkish ä, and they have borrowed Uiy. iimZi with the two spellings .am-ei and .em-ci (LAUFER, in TP, 1926, 489). In my opinion, Käncäk may represent an original Iranian name, which in its turn may be *Kanjak or *Kanjag, but this is only an hypothesis, and we have no reason to connect it with a name of the far-away Caucasus.
Of that ancient language of Kâsyar, Iranian or not, we do not possess any text, or none at
least has been recognized as such hitherto. As far as I can remember, only one word is anciently said to be a specifically Kâsyarian form, and even this case is open to some doubt. As is well known, the Skr. upâdhyâya > Prâkr. uvaj jhâa, « master », through an apheretical form of the
same type which has given Tamil v¢ddyar, is the remote original of fu ho-shang or fl1 L
ho-shang (*yuâ-ziang), the respectful title with which Buddhist monks are usually addressed in Chinese. From the point of view of Chinese phonetics, ho (*yud), which is pronounced wa in Japanese, always renders a va- of Central Asian languages (but with a prothetic laryngeal opening) ; the transcription was made on a form with a nasalized final. Although the derivation is not open
IIIF Color HighRes
