FROM THE PLATEAU TO THE BASIN FLOOR 135
rain and hail began to fall. When the hailstones melted on a piece of dark cloth, each one left a spot of dust which it had gathered in the air, typical yellowish loess, of precisely the same nature as that which covered the moraines and mountain slopes. The dust could not have been of local origin, for the region nearby was well covered with grass, and was receiving accessions of loess rather than giving them up. Apparently, the dust came from the desert far to the north.
During the next few days, we descended twenty or thirty miles into the narrow Sanju valley. Even at so short a distance from the main mountain range there was a distinct decrease in the amount of rain, as we could see from the vegetation, and also from the clouds which gathered thickly over the main range, but only thinly over us. The loess, too, began to grow thinner, and in many places was being dissected by wind and rain. It soon became evident that though dust from the desert was deposited everywhere, it accumulated into thick deposits of loess only in places where there was rain enough to support an abundant growth of vegetation, able to hold all the dust that fell.
On June 21, 1905, I left the main caravan and went down the Sanju valley. Thereafter, in order to cover more ground, Mr. Barrett and I worked independently most of the time, until our final separation in September. Taking a single servant, Ibrahim, and a Khirghiz guide, I rode down 'the narrow gorge where, as already related, we raced so successfully with the daily flood due to the snow melted by the hot sun on the mountains. We forded the river twenty-three times that day among huge boulders, where the horses con-
IIIFカラー高解像度
