that these petty chiefships must have been situated close together near the present Karghalik. The passage about Hsi-yeh mentions that its ruler was called king of Tzû-ho ; and in apparent agreement with this indication of a special connexion between the two places, modern Chinese geographers were prepared to identify Hsi-yeh and Tzü-ho with the closely adjacent village tracts of Yül-arik and Kök-yar.4
Hence it was of interest for me to be able to familiarize myself with the local conditions. At Kök-yâr the area capable of cultivation is restricted to a narrow strip of ground less than half a mile across and under five miles in length, enclosed between absolutely barren slopes at the bottom of a narrow valley. The people, reckoned at about two hundred households, depend largely for their sustenance on cattle and sheep kept far away in the mountains. Nor is the configuration of the valley such that much extended cultivation can be assumed even for an earlier period when a moister climate prevailed. Conditions are somewhat more favourable at the small oasis of Yül-arik, situated at the mouth of the Akchik-jilga some six miles in a direct line eastwards. Here the whole irrigated area—and there did not seem to be any water to spare—was said to support about 26o households, including the village of Rôwush. Ushak-bâshi, the adjacent oasis eastwards, which receives irrigation from the stream known as Ulagh-astang and is fed by permanent snow-beds, appeared to be slightly larger and was reckoned at over 300 households .6 Fertile as the thin cover of loess is, which overlies the Piedmont gravel beds at this little cluster of oases, it seems difficult to suppose that, situated so near to each other and so limited in resources, they could ever have figured as separate ` kingdoms'. A glance at the map (No. i 2) will help greatly to strengthen this doubt ; for it shows how relatively small is the cultivated area of these submontane villages when compared with the fertile expanse of the main Karghalik oasis.
In view of the topographical conditions it may be safely asserted that Karghalik, with its ample supply of water from the Tiznaf River and its thick and fertile loess terraces, must always have been the most populous and important of the oases south of the Yarkand river. It is impossible to assume that it can have remained without mention in the Chinese survey of which the Han Annals have preserved us a record. An explanation of the apparent omission is afforded by the more lucid notice which Chap. CXVIII of the Later Han Annals contains of these territories.6 There it is stated that travelling from Khotan by the route leading westwards ` one passes through Pi-shan and arrives at Hsi-yeh, Tzû-ho, and T6 jo'.7 That all these must be looked for in the vicinity of Karghalik is clear ; for the identity of P`i-shan with the present Gama is certain,8 and the route thence westwards necessarily leads to what is now the Karghalik district.
Of the kingdom of Hsi-yeh j -ge it is stated that it also bore the name Lu-sha O and
that it had then a ruler distinct from Tzû-ho, ` the [CHien] Han shu stating wrongly that Hsi-yeh and Tzû-ho are a single kingdom. The king of Tzû-ho- â has his residence in the gorge of Hu-chien itlf itt at a thousand ti's distance from Su-16 (Kashgar).' Both the distance indicated and the situation in a confined valley point to one or another of the submontane oases south of Karghalik as the Tzû-ho capital here referred to. Accepting this location we are necessarily led to identify Hsi yeh as Karghalik ; for only on the assumption that this great oasis is meant can we account for the striking difference in population which the notice of the Later Han Annals indicates by stating the number of households as 2,500 at Hsi-yeh and only 35o at Tzû-ho. This proportion is about the same as a modern census would be likely to reveal between the
1907, p. 174)•
7 Cf. Ancient Khotan, i. pp. 97, 103.
8 See Chavannes, Pays d'Occident, p. 28 (Toungpao, 1907, p. 174).