772 EXPLORATIOlsr TN CHINESE TURKESTAN. 



''Uzuii-Tati'' ("the distant Tati"), in the desert north of tlie oasis of 

 Gulakhnia. Far-advanced erosion and tlie operations of treasure 

 seekers from tlie neighboring- villages have left little of structural 

 remains, but tlie usual debris of broken pottcny, glass, china, etc., was 

 plentiful. 



A close inspection of the conditions under which cultivation is carried 

 on in this vicinit}', along the edge of the desert, was very instructive 

 from the point of view of historical topograph}". 1 found that, owing 

 to a difficulty of conducting the irrigation water sufficiently far, some 

 villages of this oasis had, within the memor}" of living men, l)een shifted 

 as much as 6 to 8 miles farther to the south. The crumbling ruins of 

 the old village homesteads, stripped of all that could be of use, are 

 still to be seen. Over miles of ground, which the desert sand is slowlj" 

 overrunning, the lines of empty canals, embanked lields, etc., can be 

 made out with ease. It was the best illustration 1 could have of the 

 process wdiich many centuries ago nmst have followed the abandonment 

 of ancient localities like the Niya River site and Dandan-Uilik. 



Increasing heat ])y day and recurring dust storms warned me that 

 the season was close at hand when work in the desert would become 

 impossil)le. So, as soon as I had ret'urned to the outskirts of Khotan 

 on April 5, I set out for the ancient sites whicli still renuiined to be 

 examined in the desert northeast of the oasis. There a discovery of 

 unexpected importance awaited me; for when, after examining Aksipil 

 and other del)ris areas, I arrived at llawak, of which Turdi, ni}' hon- 

 est old guide, had spoken merely as "an old house,''' 1 found l)efore 

 me a large Stujxi, forming, with its inclosing (piadrangle, l)y far the 

 most imposing of all extant ruins of this region. The excavations I 

 at once conmienced along the massive walls of the great stupa court 

 revi^ded a remarkable series of colossal statues in stucco, representing 

 liuddhas or Bodhisattvas, with many smaller relievos between them 

 (pi. vi). The walls were further decorated with e]al)orate plaques 

 forming halos, as well as with fresco paintings. The whole of the 

 relievo work had originally l)een painted. 



The careful excavation of this wealth of sculpture was a difficult 

 matter. The interior framework of wood, which once su})ported the 

 masses of stucco, had rotted away, and, deprived of this support, the 

 heavy images threatened to collapse wdien the protecting sand was being 

 removed (pi. vi). The risk was considerably increased l<ythe Burans, 

 which were t)lowing with more or less violence during the whole of 

 my stay. Extreme care was needed in clearing the statues, and their 

 lower portions had to be covered up again as soon as they had been 

 photographed. An attempt to remove the larger sculptures wascjuite 

 impractical)le owing to the extremely friable condition of the stucco 

 and the difficulties of transport. But of the smaller ones and of pieces 



