Sec. i] BY THE DESERT EDGE TO KHOTAN 99
The description which the Hsi yü-chi gives of the character of the route is easily accounted for if the wording is capable of the translation given by Beal, ` skirting along the high mountain passes and traversing valleys.' The mountains south of Kilian, Sanju, and Duwa, together with parts of the snowy main range on the Upper Kara-kâsh river, are fully visible from the route whenever the atmosphere is sufficiently clear. The streams from the mountains where they are crossed by the route show broad and well-marked flood-beds, in parts deeply cut into the loess terraces, as near Gitma, Chudda, Pidlma. Further towards the foot of the hills, the valleys from which these streams debouch and the ravines they have cut through the conglomerate strata are also in view 16. Thus Hsüan-tsang's reference to valleys' is quite justified. It is true that Julien renders the first words of the passage by ` it franchit de hauts passages de montagne '. But it must be observed that this description, if taken literally, would be equally inapplicable to any route by which Hsüan-tsang could possibly have travelled within eight days from the central part of the Karghalik District to Khotan Is.
SECTION II.-THE OASIS OF GUMA
The first trace of remains of archaeological interest was met with at the end of my first march from Karghalik. When nearing the massively built rest-house of Kosh-Langar, erected during Yâqûb Bég's reign in the midst of a completely sterile steppe, my attention was attracted by a mound visible from afar over the level horizon, which was known to my Karghalik guide by the general designation of Tim'. It was reached after a ride of about I â miles to the north-north-east of Kosh-Langar, and proved to consist of a solid mass of sun-dried bricks measuring about 75 feet in circumference at the base and rising to a height of about 3o feet. The appearance of the ruin suggests an originally conical shape for the upper portion, which would agree with the assumption that the mound marks the remains of a Stûpa. But its state of decay is too far advanced to permit of any certain conclusion. The bricks seem irregular in size, but generally smaller than those in the ruins of Kurghan-Tim and Mauri-Tim. The people frequenting the lonely station are inclined to recognize in the ruin a watch-tower of great antiquity, corresponding to the ` Potais ' built by the Chinese at intervals of ten li, approximately equivalent to two miles, along the greater portion of the modern high road between Kâshgar and Khotan 2. The fact that I could not trace any pottery débris or other remains in the vicinity of the ruin speaks against the site having once been permanently inhabited ; yet the very nearness of this ruin to the present route and halting-place is an indication how little the line of the former is likely to have changed for centuries.
Gûma, which I reached on the 4th of October, after two more marches over dreary barren Dasht, is the largest of the oases on the route from Karghalik to Khotan. It occupies an extensive loess terrace, which here overlies the sterile glacis of gravel and detritus washed down from the mountains. The greatest breadth of the terrace seemed about four miles, while its length in the direction from south to north, as marked by unbroken cultivation, is at least eight miles. Only a narrow strip of gravelly ` Sai ', some two miles broad, separates this terrace
16 Compare Hedin, Reisen in Z.-A., pp. 10-17, passim.
16 It might be thought that Hsüan-tsang first travelled nearer to the hills by the caravan road leading to Sanju, and thence joined the present main road via Zanguya. But this route would have been some twenty-four miles longer than
the direct one and also could not appropriately be described as anywhere surmounting hauts passages de montagne'.
1 See for this term, above, p. 74.
2 See Ruins of Khotan, p. 159.