RECONNAISSANCE IN CENTRAL TURKESTAN. I95 
kettles are drained. On the outer edges are valleys where streams probably flowed 
along the two sides of the glacier, as they often do. The inner slopes of these two 
portions of an old moraine are verj' steep, as they are now being undercut by the 
ice, C, on which lies part of the present moraine, D. The slopes show that the 
material of the old moraine is truly glacial in its angularity and irregularity of size, 
but utterl)- different from the present moraine in that it is well weathered and that 
the soil produced by weathering fills all the interstices. In fact, the slopes seem 
to contain decidedly more soil than rock. At present the modern moraine lies 50 
feet more or less below the top of the older ridge, but there are many places where 
it formerly rose to the top and o\-erflowed, as at D. 
These facts seem to lead to the conclusion that between the deposition of mo- 
raine No. 3, or 4, as the case may be — it is quite immaterial which — and the 
deposition of the present moraine, the ice retreated to a position farther up the 
valley than that which it now occupies, a.s the following considerations will show. 
It may, perhaps, be taken as beyond question that a moraine such as B can not 
ha\e become thoroughly weathered, graded, and covered with grass without long 
exposure to the air; nor does it require discussion to show that where graded slopes, 
such as those of B, are being undercut, they must once have extended farther in the 
direction of the agency which undercuts them. Therefore the ice must for a long 
time have occupied a smaller space than at present, and since that time it must 
have widened. But this could not have been possible with a continuously retreat- 
ing glacier, for it should have suffered a continuous narrowing. Moreover, on the 
supposition of continuous retreat, with or without pauses, but without readvances, 
each moraine ought to lie above the one that preceded it, and this seems to be the 
crux of the whole question. A portion of the third or fourth moraine — let us say 
the fourth for convenience — lies from 1,000 to 2,000 feet above the rest of that 
moraine, and from i "^ to 3 miles farther upstream. Between the two portions of 
the fourth moraine lies the whole of the fifth and most of the sixth moraine. It 
seems impossible to explain the facts on the theory of one retreat whether at a 
uniform rate or with pauses. 
If, on the other hand, each moraine represents an advance and retreat of the 
ice, the difficulty disappears. The old glaciers were probabh- co\ered with moraine 
stuff just as the present one is, and as each retreated it would leave a trail of 
moraine behind it. The glacial stream would carve a \alley in the abandoned 
moraine during the interglacial epoch. The next glacier would follow this valley 
at first, though it would widen it greatly, and in most cases utterly obliterate it 
But each succeeding ice sheet was smaller than its predecessor, and where the 
valle\- was wide it might happen that portions of the older moraine would be pre- 
served. This is what appears to have taken place at Kan Su. If tliis interpretation 
is correct it means that after the fonnation of the main portion of the fourth or 
possibly the third moraine, the ice retreated so far as to end at least 1,200 feet 
above the level of its moraine, and so reached a point 700 feet above the present level 
of the glacier front; that is, the fourth glacial epoch was followed by an inter- 
glacial epoch decidedly wanner than the present epoch. 
