Sec. vii] CONCLUSION OF WORK AT TURFAN 715
occupation of Yâr-khoto had continued down to the Uigur period.9 That the rooms on either side of the court and cella had served as monastic quarters was shown by niches in the walls, which had evidently been used as cupboards, as well as by blackened passages cut into the clay of the outer walls, at corners suggesting the position of fireplaces. Holes for rafters in the massive clay walls, still rising to 17 feet and more, showed that there had been at least one upper floor. It was instructive to note that the enclosing main walls up to a height of 5 feet consisted of the natural clay left to form a base, while the ground on either side was dug away, and also that the rooms near the main gate had their floor considerably below the level of the court. This method of providing basement rooms not needing any masonry by excavation from the hard clay may be observed in most of the ruined dwellings of Yâr-khoto. These basement rooms were obviously resorted to during the fierce heat of the Turfân summer, just as the tai-khânas' are at Peshawar and elsewhere in the extreme north-west of India, on account of their comparative coolness.
On the flat ground clear of ruins towards the northern extremity of the plateau it was interesting to note how the bare clay surface had been furrowed into rudimentary Yardangs from one to two feet in depth. Their bearing from north-west t(E) south-east showed clearly that wind-erosion was here due to the violent winds which 'aspiration ' carries down into the Turfân depression during the spring and summer through the gap in the Tien-shan marked by the Ta-fan-ch`êng saddle on the road to Urumchi. That this erosion has not proceeded farther is to be attributed to the fact that the Yârs along the plateau stop the direct access of drift-sand, though, of course, they do not prevent particles being carried through the air and swept over the plateau surface 10 The tombs to be seen on this ground were arranged in small groups, as they are near Astâna, but the approach trenches to some of them were wider. In one place a small sunk court gave access to the passages leading to several tomb chambers. All tombs appeared to have been opened long ago and repeatedly searched. The extensive cemeteries scattered over the Sai to the west of Yâr-khoto appeared to have been similarly treated. A number of small tomb chambers seen exposed along the lower portion of the cliffs on either side of the western Yâr were quite empty.
That portion of the plateau which bears the ruins of the ancient town of Chiao-ho may be divided into two fairly well defined areas. The northern and larger one, of which Fig. 326 gives a comprehensive but rather distant view, is closely occupied by the ruins of what manifestly were for the most part dwellings, often considerable in size. Several clearly recognizable roads pass between them, the two widest extending along the longitudinal axis of the plateau and connecting with shorter transverse roads. Near the top end of the western main road rises a very conspicuous pile of structures marked II in P1. 35. It comprises a large temple in the centre, and grouped on its flanks and along the sides of a big rectangular enclosure, measuring about 6o by 8o yards, a series of minor shrines and halls, as seen in Fig. 331. Within the main temple stands a massive tower-like brick pile,1' of the type of the Tura of Sirkip and the Taizan of Astâna. Some of the niches arranged in rows on all four faces still retain the seated Buddha images in plaster with which they were adorned. There can be no doubt that this big Vihâra, which from a distance presents
9 Cf. Serindia, iii. p. 1168.
10 It may be conveniently noted here that I was able to observe the effects of wind-driven sand also on the ground to the east of Yâr-khoto. Riding across the village lands towards the most fertile portion of the Yâr-mahalla tract where my camp C. 243 stood (Map No. 28. c. 3), I noticed about half a mile from the eastern ghol' (or Yar) of Yâr-khoto a belt of uncultivated ground, in places overrun by small dunes. A clay wall had been built to stop their farther extension.
Farther on I came upon fields which, instead of being manured with loose earth brought from the ruins of Yâr-khoto, had big lumps of clay scattered over them. According to Zahid Beg, the local headman accompanying me, these clods are left after the autumn flooding of the fields for the purpose of protecting the surface from erosion and of catching the dust which the ` Burans ' drive along here. This dust, when thus caught, was said to suffice as a fertilizer,
11 See Serindia, iii. Fig. 278.