of the rocky »gateways» that are so common in that region, that is to say a breach in the range which separates the two lakes. Yet, strange to say, the culminàtingpoint of the isthmus lies at least a hundred meters to the south of the gateway, and we doubled the crags at its western shoulder. These mountains constitute therefore a detached group on the neck of land between the lakes. Towards the east they plunge down perpendicularly in places on the north; and in the face of these cliffs and at their foot we discovered concave excavations or shallow grottoes, to to 12 m. high and a couple of meters deep. One of them appeared to be sometimes used as a dwelling-place or temporary shelter, probably by shepherds, for just outside of it somebody had built a low stone wall. The descent from the top of the divide down to this craggy projection was relatively steep, but the rest of the way right down to the shore had a flat and gentle slope, and here I counted six extraordinarily distinct beach-lines in the shape of terraced steps. The highest of these, which runs just under the precipice containing the grottoes, is the most developed, and appears to extend farthest in both directions, following faithfully the northern foot of the range. With regard to the grottoes, there cannot exist a doubt but that they also mark a higher level of the Selling-tso, a level at which the lake remained for a relatively long period, whereas the terraced steps mark successive and relatively rapid subsidences. Nor need it occasion any surprise, that the 1 ke should scoop out such grottoes as these; their existence is easily accounted for by the single fact of the cliff being precipitous, and consequently directly exposed to the wave-beat and ice action of the lake. The rock itself is brittle and weathered, and yields easily to attack. Moreover at the time when the Selling-tsa reached up to the foot of the range, the lake spread out to an incomparably greater distance in every direction than it does now, as is pretty evident from the flatness of its shores. When the wind blew from the north the waves would wash a respectable height up the cliff and would have a powerful auxiliary agent in the ice; for it may well be assumed that this lake was formerly even less salt than it is now; the water along its southern shore had a sp. gr. of only 1.0245. When the Selling-tso reached up to the foot of the range, or at any rate to the highest of the six terraces, it would then be at the same level as the existing Naktsong-tso; and, being at that level, it would require but an insignificant rise to give origin to a sound of approximately the same appearance as that the shallow threshold of which we paddled across in the Naktsong-tso. This circumstance contradicts, it will be observed, the supposition which I made above, that the threshold in question is a moraine-ridge. However that may be, there exists nothing which goes to prove that any such connection did exist formerly between the two lakes. The Naktsong-tso has for so long retained its position, that at any rate no older mark remains on its shores. Meanwhile the Selling-tso has dropped step by step, and still continues to fall, so that the difference of level between the two lakes goes on constantly increasing. In view of the imperfect knowledge which we possess with regard to both lakes, I venture to think that the following view is perhaps the most reasonable. The Naktsong-tso is fresh because it has somewhere an outlet; the Selling-tso is salt, because it is the terminal lake, the last recipient in a self-contained drainage-basin. Along the north-west, north, and north-east shores of the Naktsong-tso we found no emissary issuing from
IIIF Color HighRes
