Sec. in TUN-HUANG AND THE `CAVES OF THE THOUSAND BUDDHAS' 357
their position and the direction of the winds that prevail in the lower Su-lo-ho valley. These winds, as repeatedly stated elsewhere, blow mainly from the east and north-east.12 They are in all probability caused by the ` aspiration ' which draws the colder air of the Pei-shan plateau and of the high ground joining it to the Nan-shan down into the lowest portion of the Tarim basin, where the atmosphere is quickly heated in spring and summer.
The action of these winds upon the heavy masses of sediment which the Tang-ho or Tun-huang river and the numerous flood-beds east of it carry down from the barren northern slopes of the Nan-shan suffices to explain the heaping up of the high sands on the broad foot-hills south of the Tun-huang oasis. I t is probable, however, that wind-erosion, of which my observations have so abundantly proved the existence all along the lower Su-lo-ho valley, especially below An-hsi,13 is also constantly adding its quota of fine dust to these sand-hills. As Map No. 39. B. i shows, the deep-cut river of Tun-huang with its considerable and constant flow of snow and ice-fed water stops the westward progress of these accumulations of sand. But to the west of the Nan-hu oasis we again meet with a big belt of dunes swept up against the foot-hills by the same winds, and this belt extends along them right up to the point where they overlook the terminal basin of the Sulo-ho.14 Still farther to the west we may recognize a continuation of the same system of dune-covered foot-hills in the high sand ridges which flank the Besh-toghrak valley on the south and join up beyond it with the big sands of the Kum-tâgh.l5 We find an exact parallel in the Turfân depression to this accumulation of high sands over the foot-spurs of the westernmost outliers of the Nan-shan.16 But that is on a much smaller scale and must be left for discussion elsewhere.
On April and I was able to leave Tun-huang town for the ` Caves of the Thousand Buddhas ', after having secured the guides and the additional camels required to enable R. B. Lai Singh and Muhammad Yâqûb to carry out surveys by separate routes in the mountains and along the Su-lo-ho before rejoining me at An-hsi. An icy gale from the north-east accompanied us on the march and prepared me for the desolate wintry aspect that the sacred site still presented. Ice covered the shallow channels in which the small stream loses itself over the wide gravel beds at the mouth of the desert valley, and the murky dust-laden atmosphere helped to emphasize the utter barrenness of the conglomerate cliffs and sand slopes on either side.
Wang Tao-shih welcomed me cheerfully and showed with genuine pride the various new structures which his pious activity had created since I had last seen the sacred spot seven years before. Opposite to the cave-temple in which the great hoard of manuscripts and paintings had come to light, there rose now a spacious guest-house and a series of shrines filled with big gaudily painted stucco images. Near by, a garden well laid out with young fruit trees, rows of stables, brick-kilns, &c., attested the little priest's single-minded ambition to restore, according to his lights, the glory and popular attractions of the ancient sacred site. He told me that the new hospice had been built mainly with the gifts of silver made by me in 1907 in return for the ` selections ' I had then been able to carry away. The statement seemed to credit me with more ` merit ' than was warranted by the number of silver ` horseshoes ' which had then actually passed between us.'? But it was very welcome as an indication that fresh favours were expected of me on the same business-like basis. Anyhow there could be no doubt that in the fine red book of donations, kept by Wang-Tao-shih and now eagerly produced for my inspection, the sums I had successively disbursed were all duly entered.
12 Cf. Serindia, ii. p. 643 ; iii. pp. 1095 sqq., 1102 ; 14 See Maps Nos. 39. A. I ; 36. B—D. I.
Desert Cathay, ii. pp. 140 sq. 15 See Maps No. 35. A, B. 4 ; 32. C, D. 4 ; 33. C, D. I.
13 Cf. Serindia, iii. pp. 1095 sq., i loo sq. ; below, pp. 365, 16 See Maps No. 28. D. 3 ; 31. A. 3.