260 ACROSS THE DESERT 



the yamen. It wjr, very late, however, so we de- 

 clined. They had had no rain here for two years, 

 and the road, as may be imagined, was exceedingly 

 dusty as we set out the next morning. 



At 'J'urfan, the most important place on the 

 Nan-lu, we only remained one day. My chief 

 recollection is of a \^ery good new inn, and a large 

 number of children flying kites amid prodigious 

 quantities of dust. The next stage was smooth 

 easy-going for 70 //. The stages were usually 

 either 70 or 1)0 //, but at times stretched to 110, 

 120, 180, and once 180 /?. We must have averaged 

 25 to 30 miles a day. Once, on asking the length 

 of the next stage, we were told it was 70 //' there, 

 but 90 li back ! The return journey — of course, 

 on the same road — was uphill ; hence the dis- 

 crepancy ! As we only occupied thirteen hours on 

 the 180-//' stage, I fancy some such error must 

 have accounted for the estimated distance, though 

 the road ran on a bare plain on which no snow 

 lay. The relative positions of the 5-li posts, tliough 

 they had dwindled away almost to nothing after 

 leaving Kansu, were as uncertain as the scansion 

 of my Latin verses, which is saying a good deal. 



A few days after leaving Turfan I had a long 

 walk after a single gazelle, which led me a tiring 

 chase and eventually defeated me altogether. I 

 reached our destination, San-kech-wan by name, 

 to find the carts had already arrived. It is a 

 desolate spot, noted for the terrible wind, which 

 makes it one of the most dreaded places on the 

 road. Recently fourteen camels had been frozen 

 to death, and for men to perish by the same fate 

 is by no means unconnnon. Our inn we shared 



