EXPLORATION IN CHINESE TURKESTAN. 767 



After a march of about 30 miles through the desert northward, I 

 arrived on the evening of eTanuary 27 at the southern edge of the 

 wide area over which are scattered the ruins 1 was in search of. The 

 subsequent explorations showed that it extends for over 11 miles from 

 north to south, with a maximum breadth of about tti miles. 



Pitching- my camp near a small stupa half buried in the sand, I pro- 

 ceeded next morning to the ruined house where Ibrahim, the young 

 Niya villager alread}^ mentioned, had unearthed his inscribed tablets. 

 He declared he had left more in situ. It was a moment of cheerful 

 excitement when I approached the timber debris, rising like the 

 remains of a wreck from the eroded ground around it. On the sandy 

 slope I found at once some tablets actuall}^ exposed, and many more 

 scattered about under a slight layer of drift sand within the small room 

 where Ibrahim had originally unearthed them (pi. iv). The house 

 which contained it had, like the rest of the buildings at this site, been con- 

 structed of a wooden framework of massive beams and posts. Between 

 the latter rose the walls of hard plaster, strengthened internally by thick 

 mattings of rushes. These walls had completely decayed where not 

 actually covered by sand, but the posts, now l)lanched and splintered, 

 still rise high above the surface. In the building first explored, the 

 sand, which during former centuries must have protected it, had largely 

 drifted away. The remarkable state of preservation in which man}" of 

 the inscribed tablets were found was hence all the more surprising. 

 Over 100 were cleared from the little room already mentioned, and 

 the excavation of a large room of the same building, on the day fol- 

 lowing, more than doubled that number. Unfortunately the protect- 

 ing la3^er of sand was here only about 2 feet deep, and in consequence 

 all materials not lying quite flat on the floor had decayed completely. 



The present condition of this ruin, which originally appears to have 

 been used as a monastic building, illustrates strikingly the destructive 

 effect of erosion on this and other structures of the site. The actual 

 remains of the building occupy a small plateau raised now 12 to 15 

 feet above the immediatel}^ surrounding ground. The lower level of 

 the latter is the unmistakable result of erosion. While the strip of 

 ground actuallv protected by the debris of this and similar structures 

 retains the original level, the open surface near by, consisting of mere 

 loess, has been gradualh^ lowered by the action of the wind. The 

 drift sand carried along this portion of the desert is not sufficient at 

 present to All the depression thus created. From the geological point 

 of view, not less than from the archaeological, it would })e interesting 

 to study the exact conditions under which the power of the desert 

 winds asserts itself in its two main lines of action — -erosion and the 

 movement of drift sand. But I am convinced that it will take years 

 of minute and systematic observation before any safe conclusions can 



