198 ANCIENT SITES OF THE KHOTAN OASIS [Chap. VIII
aware that a widely accepted theory, first propounded by Richthofen, ascribed the formation of loess (or adobe') to the long-continued drifting and deposit of fine dust by wind, i.e. to purely ` aeolian ' action 16, I was anxious that the view I had been led to form of the character and origin of the covering-layer at Yôtkan should be tested by geological evidence. I accordingly secured a considerable number of specimens of soil taken from various portions of this layer as well as from the culture-strata' below. In 1903 I had the satisfaction of submitting these specimens, together with a number of others brought away from ancient sites in the desert, to Professor L. L6czy, of the University of Budapest, who, owing to his extensive travels and researches in the loess regions of Western China, and his subsequent investigations into the loess formations of the Danube Basin, was specially competent to deal with them. The result at which the distinguished geologist and geographer has arrived, by the careful microscopical examination of these specimens and by their comparison with other similar materials, available to him, will, I hope, be soon made accessible [see now Appendix G]. In the meantime I am enabled by Professor L6czy's kindness to indicate here the main facts as summed up in his letter dated December 4, 1903.
The specimens submitted from Yôtkan show all the characteristics of riverine loess, i.e.
a deposit formed of river-carried fine silt. It consists chiefly of distinctly angular grains of
quartz, showing scarcely any trace of that attrition which is undergone by particles long subject to movement by wind, such as those contained in true drift-sand. Mica flakes, evidently originating from the granite and gneiss of the mountains drained by the Khotan rivers, are found plentifully in it. On the other hand, the deposit contains very little fine dust and practically no clay ; it is but slightly calcareous, and lime concretions are rare. The constituents revealed by microscopical analysis are thus all such as might be expected in the sediment of the Khotan rivers ; but there is reason to believe that the deposition of this sediment in loess layers such as seen at Yotkan has not throughout taken place direct from the water, but to a very considerable extent also by subaerial (` aeolian ') transport from the river-bed. The fine silt carried up from the latter by the wind must necessarily reach first the tracts immediately adjoining the river course, and its retention there must be all the more certain and constant the moister the soil is kept by irrigation. Thus cultivated ground in the oasis is assured steady accretion in level by a simultaneous deposit both from the water brought to it and from the atmosphere, the material accumulating by either process being always the same silt of the rivers.
The explanation thus furnished makes it easy to understand why Professor L6czy's examination has revealed no difference whatsoever in mineral constituents between the layer of pure riverine loess and the culture-strata of Yotkan. Whatever of the latter does not consist of still recognizable hard débris or decomposed vegetable and animal matter, can only be loess soil once used for sun-dried bricks and mud walls or else river-silt in the form of dust brought to the site while it was still inhabited. In either we must expect the identical mineral composition which• the more recently formed loess of the oasis presents. The difference in colour is caused solely by the abundance of decomposed organic materials in the ancient culture-strata. It is evident that the definite explanation thus obtained of the deep layer of earth that covers the remains of the old Khotan capital applies equally to other ancient sites within the oasis which have left no trace above the present ground-level and which thus still await discovery and excavation 16.
'b Comp. Geikie, Text-book of Geology, i. pp. 439 sq.
16 We shall see hereafter that there is at least one other site of Khotan—I refer to Tam-öghil—which has actually