curve without a trace of terracing or any indication of old lake shorelines. Here and there, however, the smooth stretches of the swamp are broken by low, rounded hills or by broad, flat tables, 20 to 4o feet high and several miles wide, which are distinguished by a sparse growth of knotted poplar trees. On the soft slopes of these elevations are found most of the springs that make it possible for the poor Kirghiz to inhabit the wretched swamp. Jai Tebeh (Devil Hill), at the western end of the lake, is one of the best examples. Here on the sides of a small rounded hill five small springs bubble gently up at heights of from io to 4o feet above the green swamp which stretches for 3 or 4 miles on every side. The material of the hill and of the swamp seems to be identical, although possibly that of the hill is a little more sandy. In neither is there the least sign of gravel. A few miles west of Jai Tebeh a formation of apparently the same sort as that of the hill assumes the shape which has above been called a table, although it might better be described as a low and very broad promontory. It rises from the swamp as a rude terrace with an irregularly dissected front, on which there is a suggestion of smaller terraces. Near the borders the top is somewhat rough, but it soon becomes very even and stretches back smoothly toward the mountains, near the base of which it merges into the slope that rises more steeply, although still very gently, toward the mountains from the edge of the adjacent swamp. On this low promontory are two or three springs like those of Jai Tebeh. A few miles farther to the west, at Dongjigdeh, another bill stands in the middle of the portion of the swampy plain lying west of the lake. The hill is about 4o feet high and has a spring near the top. bike Jai Tebeh, it appears to be mostly made of silt, but there are gravel and some sand scattered here and there. Six other streams were seen in different parts of the basin, welling up on small hills at a height of from 20 to 3o feet above the surrounding swamp or neighboring lake. In all the springs the water was sweet, and the material from which they flowed was the same silt as that of the swamp, so far as the eye could detect.
It seems hardly possible, however, that the underlying part of these hills can be composed of this sanie silt. The one hill where sand and gravel occur seems to furnish the key to their structure, and the suggestion afforded by this one is borne out by evidence which will be presented after the statement of the conclusion to which it leads. Shor Kul appears to have been twice expanded to a size much greater than that of to-day. At the first of these expansions it reached a height of about 35o feet above the present water level. It then retired, so that the lake deposits were first covered with sand and gravel and later were well dissected. It next rose again, although to a less extent, and covered some of the remnants of the old lake deposits with a new layer of silt (fig. 146). The remnants of the old deposits are the hills and tables which have already been described, and which are now left exposed by a second retreat. The springs occur on them because the layer of sand and gravel on top of the earlier lake beds is easily pervious to water, while the silts above and below are impervious. Consequently water from the base of the mountain beyond the end of the silts gathers in the sand or gravel and percolates gently downward toward the lake. In the little hills the upper layer