separating the Lou-lan station and the fort L. K. one would expect some sort of signal-towers as the important road to Miran must have run here. The region is not untouched by man, because stray finds both of stones and metal do occur here, but the water supply was most likely insufficient to sustain any permanent settlements or allow any construction of watch-towers.
A little to the east of this line lies the route along which Dr. HEDIN carried out his levelling from the Lou-lan station to Qara-qoshun, and where he discovered a flat depression, which he identified with the bottom of the lake Lop-nor of Loulan's time. The extension of this basin is too limited to have been the only lake, and it was hardly inundated at the time of Lou-lan as HÖRNER'S finds K. 13356-8 are made in the deepest part of it. We cyan therefore hardly count with the presence of a lake here as an explanation of the absence of ruins.
As seen from the above the ancient settlements in the Lop desert are not limited to any special still traceable shore-line but have been grouped along the river branches of a delta. This kind of settlement has many parallels among the present-day oases of Eastern Turkistan. The same kind of settlement existed in the Han and the Sung-Yuan periods at lower Edsen-gol.
Taking the known ruins as a starting point one can thus, to a certain degree, reconstruct some features of the configuration of the Tarim delta at the time of Loulan. Thanks to the surveys carried out by STEIN and HÖRNER certain parts of ancient river beds have been put on record. In many cases it is very hard to determine the location of the old river courses as the ground is so strongly wind-eroded. Only some trifling spots of the actual land surface of the Lou-lan time now remain. All the rest has been ground *off by the violent force of the north-eastern winds. STEIN'S criteria of ancient river beds, i. e. long depressions lined by rows of trees, are not conclusive in all instances. Some of his river beds as shown on his map (Sheet B) can hardly be proved until more extensive surveys have been made.
When surveying the area immediately to the south of Qum-darya, between the height of Yardang-bulaq and The Small River, I encountered several features, which at first sight seemed to be old river beds. They consisted of river-like depressions, 8-10 m. deep and 50-100 m. wide, and the edges were bordered by groves of dead poplars and tamarisks. But they were anything but old river courses. Closer study proved them to be oblong, but closed, depressions, i. e. wind-hollows of very considerable dimensions.
In some instances the bottom contained moist, black clay and even water of the utmost salinity, Pl. V b.
If I had only crossed these depressions at right angles, without studying their extensions, the river-like appearance would certainly have led me to mark them on my map as old river beds.
The distribution of ruined dwelling sites does not necessarily mean that all of them were within reach of water at one and the same time. It is very probable that they