534 A STRANGE OLD LAKE BED CH. XLVIII
well-sinking luck having unduly limited the supply. But at least there was plenty of fuel to give warm camp-fires to the men when the bitter wind rose in the evening—this time from the south-east—and there had been just enough water for tea to go round.
Notwithstanding a missed dinner I sat content in my little tent ; for the day's march, fatiguing as it was, had yielded interesting observations in abundance. It had served to illustrate strikingly Marco Polo's account of the difficulties and dangers of the route. The track, soon after leaving the Besh -toghrak grazing, had become practically effaced over long stretches of sand, and as the Kuruk-tagh range recedes here northward and flattens, it would have
been easy to lose the right route. By continuing too
long eastwards travellers coming without guides from Besh-toghrak might easily stray away into the hopelessly barren desert north of the depression to which we had to cling. This wilderness of fantastic clay terraces and sand dunes would necessarily impress those who tried to make their way across from the side of Tun-huang. In fact, it is like a curtain, and most effectively hides the well-marked valley which leads south - west from Besh - toghrak and affords the natural route to Lop-nor.
The most interesting feature, however, of the day's march was the discovery of the large lake bed along and through which we had been passing for fully ten miles. Its aspect and its position relative to the Khara-nor made me already feel convinced that it must have served once as a terminal basin for the Su-lo Ho and Tun-huang rivers. But when did the latter's waters cease to reach it ? Parts of the lake bed we had crossed showed a thin salt crust which looked quite recent. Yet the lake shown as Kharanor in the maps, where the Su-lo Ho and Tun-huang rivers were hitherto believed to end, and from which alone sufficient water could come, lay, according to Roborowsky and Kozloff's map, still more than a degree farther to the east.
What seemed a puzzle at the time luckily found its explanation two months later. Then, in the course of resumed surveys to be related hereafter and illustrated