remnants of monsoon moisture passing here across the high mountains.19 In view of the desiccation so plainly proved elsewhere by archaeological evidence in the Tarim Basin, it seems but reasonable to suppose that in earlier times forest was not altogether so exceptional in the mountains south of
Karghalik as it now is.
There is another local fact which deserves passing notice with regard to the historical topography of this region. My Pakhpu informants were well aware of a place near the junction of the Yulung and Chukshü Valleys where jade used to be regularly quarried from the hill-side ` in the old Khitai times', i. e. before Yaqûb Beg's rebellion. Now the record of the Former Han Annals speaks distinctly of Tzû-ho as a territory producing jade.20 We have seen above that the name Tzti-ho is likely to have originally applied to the submontane group of oases comprising Kök-yar and its neighbours. The Pakhpu tract has its easiest approach from the side of Kök-yar, and must always have been closely linked with it. Thus the mention of its jade in connexion with Tzû-ho becomes quite intelligible and helps to confirm still further the location proposed for this territory.
From Kök-yar I made my way to Khotan between July 25 and August 5 by the little-known route which passes along the foot of and through the barren outer hills of the Kun-lun. I had chosen this track, instead of the high road followed on my previous journey which leads by the edge of the desert,21 mainly for the chance of fresh surveys it gave me. It allowed me to visit in succession the oases of Kilian, Sanju, Duwa, all resting at the debouchure of snow-fed streams destined to lose themselves further down in the desert, and proved, as Chapter XIII of my personal narrative shows,22 in various ways geographically interesting. Incidentally the journey gave me complete proof that this submontane route, owing to its length and the nature of the ground it traverses, could not possibly be the one which Hsüan-tsang followed to Khotan.23 But otherwise there was little opportunity here for archaeological observations.
From the large and flourishing oasis of Sanju, which forms an important adjunct of Gama, the ancient Pi-shan, I reached on July 31 the debouchure of the Puski Valley (Map No. 16, C. 4). At the long-stretched settlement of about forty households which extends along the scanty stream of Puski, I first heard of a ` Tim ', or ancient mound, situated to the north on the route to Zanguya. According to the report of the local greybeards it had been repeatedly dug into, on the last occasion by three men of Puski, who all died shortly thereafter ! So it was easy to guess that it was the ruin of a Stûpa. As the distance proved too great, I had to leave my visit for the next day.
Riding along the left bank of the stream, which, characteristically enough, carried even at that season only ` Kara-su ', or spring water, there being no permanent ice or snow at its head, I reached, after about two miles, a colony of five or six households, founded about twelve years before and known as Jangal-bagh. It was interesting to find here, just below the area of new cultivation and partly within it, a small ` Tati ', one of those wind-eroded old village sites which are such typical marks for the extent of ancient oases along the southern edge of the Tarim Basin. Their characteristic features I have fully discussed in Ancient Khotan.24 The Jangal-bagh Tati' proved about three-quarters of a mile long and a quarter across. The ground, a soft loess, was thickly covered with fragments of ancient pottery of bright red colour and mostly of fine texture and remarkable hardness. Representative specimens brought away, some with lightly incised wave patterns, are described
" Cf. Desert Cathay, i. p. 140.
R0 Cf. Wylie, J. Anthrop. Inst., x. p. 32.
21 See Ancient Kholan, i. pp. 95 sqq., for a full account of this route.
22 See Desert Cathay, i. pp. 152 sqq.
23 I have fully discussed Hsüan-tsang's route in Ancient Khotan, i. pp. 98 sq. The distance from Kök-yar to Khotan
town proved by actual road measurement about 183 miles and would be slightly greater from Karghalik by the same route. This distance is greatly in excess of the 800 li reckoned by Hsüan-tsang from Chê-chü-chia to Yötkan, the ancient Khotan capital ; cf. also loc. cit., note x 6.
24 Cf. Ancient Khotan, i. pp. 107 sqq.