and that this country is not Yärkänd, but a place farther to the south, Qaryalïq according to STEIN, CHAVANNES and THOMAS, Kök-yar according to HERRMANN (STEIN, Ancient Khotan, 28-29, 88-89; CHAVANNES, in BEFEO, III, 398, and Doc. sur les Tou-kiue, 311, 342; THOMAS, Tibetan Texts and Documents, I, 25; HERRMANN, in Southern Tibet, VIII, 60, 450). That Cukupa, the ancient Tzü-ho, is to the south of Yârkänd is evident; it was on the direct road from Khotan to Tag-quryan. The case of *Cakuka, Hsüan-tsang's Chê-chü-chia, is not so plain. Hsüan-tsang passes it on his way from Ka yar to Khotan, which makes one think of Yärkänd quite naturally. Moreover there is a certain phonetic analogy between So-chü (*Sâku), the name of Yârkänd under the Han, and *Cakuka, which might seem to be the same name with a Sanskrit derivative -ka suffix. It goes without saying that, in such a case, *Cakuka < *Sliku could have nothing to do with *Cukupa < Cugapa < *Cigap. But the alternation of -a- and -u- occurs for both names, since we have met with a form Cukuka on the one hand and a form *Cakupa on the other. Both *Cakuka and *Cakupa occur in the lists of the Candragarbha, and what Jnânagupta connects whith *Cakuka is located at *Cakuban in a later compilation. It looks as if both *Cukupa and *Cakuka had been used almost indifferently, or rather as if the first represented the name with a suffix in one language (perhaps the non-Iranian and non-Indian language responsible for most of the non-Indian elements in the Kharosthi documents), while the second would be an indianized form in -ka.
Such reasoning would be, however, of little avail if there were not geographical difficulties in the way of the identification of Hsüan-tsang's Chê-chü-chia with Yârkänd. The pilgrim, on leaving Ch'ieh-sha (Khasa, Käsyar) travels to the south-east, crosses the Sitâ River, passes a great sand mountain ( 5- 7% a to sha-ling) and arrives at Chê-chü-chia. It seems difficult not to admit that the Sitâ River here is the Yârkänd River, but this precludes the identification of Chê-chüchia with Yärkänd, since a traveller coming from Kâgyar would reach the Yârkänd River only after having left Yârkänd. That is why I concur with the above-named Western scholars in locating Chê-chü-chia south of Yârkänd and the Yârkänd River.
But I find it hard to decide between Qaryalïq and Kök-yar. In my opinion, the general trend of the itineraries from Khotan to Tag-quryan is decidedly in favour of *Cukupa being Qaryalïq. On the other hand, Kök-yar would be quite out of the way for any one going from Kâsyar to Khotan; but, if *Cakuka (and *Cukupa) were Kök-yar, we might easily suppose that Hsüan-tsang followed a round-about route in order to visit the sites of *Cakuka which Jnânagupta had made famous in China. But there is nothing decisive here either way. In favour of Kök-yar, HERRMANN justly said that there was no room in the cultivated area between the Yârkänd River (south of Yârkänd) and Qaryalïq for the « great sand mountain » spoken of by Hsüan-tsang. But I am not certain that it is easier to find it between Qaryalïq and Kök-yar; Kök-yar is in a mountainous, not in a sandy district. Without lending too much weight to what may be a pure coincidence, I must remark that, among the few geographical names other than those of tribes and towns which are given by Käsyari, there is Bayram-qumi, a sand hill or down between Käsyar and Yârkänd (BROCKELMANN, Kâ§yari, 29, 241). It must have been a site of some renown to be thus mentioned. I can only suppose that it lay in the fairly high sandy region, which extends east of Yangï-li§àr and is marked on STEIN'S map as having at least two Mussulman shrines, perhaps continuing earlier Buddhist ones. But to connect Käsyari's Bayram-qumi with Hstian-tsang's « great sand mountain », one
IIIF Color HighRes
