EXPLORATION IN CHINESE TURKESTAN. 763 



Dandan-Uilik. Turdi, an old and, as experience showed, reliable 

 member of that fraternit}', had brought me from there some interest- 

 ing- relics, including fragments of Buddhist sculptures, an inscribed 

 piece of fresco, and a small but undoubtedly genuine scrap of paper 

 with ancient Indian Brahmi characters. Further inquiries made it 

 certain that Dandan-Uilik was identical with the ruined site which 

 Dr. Sven Hedin had seen on his march to the Keriya Darya, and 

 which in the narrative of his travels is spoken of as "the ancient city 

 Taklamakan." 



After hurriedly completing in Khotan the preparations for our 

 winter campaign, 1 started on December 7 for Tawakkel, a small oasis 

 on the outskirts of the forest belt which accompanies the Yurung-kash 

 on its course through the desert. Thanks to the stringent orders 

 issued by Pan-Darin, the kindly amban of Khotan, I speedily secured 

 there the 30 laborers 1 wished to take with me for purposes of exca- 

 vation, as well as a four- weeks' food supply. Owing to the reluc- 

 tance of the village cultivatoi's to venture far into the desert, it would 

 otherwise have been difficult to obtain sufficient labor, especially in 

 view of the rigors of the winter. The ponies, for which the desert 

 offered neither sufficient water nor food, were sent back to Khotan, 

 while we set out on foot, the heavily-laden camels carrying the food 

 supplies, together with the indispensable baggage. Marching in the 

 drift sand was slow work, though the dunes amidst which we passed 

 as soon as we had left the east bank of the river nowhere rose above 

 15 feet. Within live days Turdi had safely guided us through the 

 sandy waste to the area where the trunks of dead poplars, rising 

 shriveled and gaunt from between low dunes, indicated the vicinity 

 of ancient cultivation. On the following day (December 18) I had 

 my camp pitched in the middle of the ruins I was in search of. 



1 soon found that the structural remains of the site consisted of 

 isolated groups of small houses scattered over an area about 1^ miles 

 from north to south and three-quarters of a mile broad. The walls, 

 constructed throughout of a wooden framework covered with plaster, 

 were either broken down within a few feet from the ground, if exposed, 

 or, where covered b}^ low dunes, could be made out by the wooden 

 posts of the fj'amework sticking out from the sand. The structures 

 left more or less exposed had already been searched by native "treas- 

 ure seekers." Their operations repeated in successive seasons had, 

 together with the erosive action of the wind, caused great destruction 

 among these ruins. But the scanty remains left on some walls of 

 frescoes representing Buddhas, or Bodhisattvas, showed at once plainh^ 

 that the ruins belonged to the Buddhist period, and that some of them 

 must have served as Buddhist places of worship. 



Luckil}^ the native "treasure seekers" are prevented by the diffi- 

 cult^' of carrying sufficient supplies from stopping longer than a few 



