the real figure, and the final result is, that Wellby's latitudinal valley lies higher than mine, though by a few meters only.
The mountain-system which separates these two valleys is the westward continuation of the Mongolian Koko-schili; and here again we are on somewhat uncertain ground. It is true there are six routes, two of which coincide in their passage across these mountains, though it is difficult or impossible to weld together into continuous chains the crests crossed over by the different travellers. When journeying south in 1901, I crossed over three distinct passes in the Koko-schili, and in the course of my 1900 journey, which lay very close to the former on the east, I crossed over five, and in my most easterly route I crossed over two, the ranges they belong to being identical with Bonvalot's immediately west of my third route just mentioned. Bonvalot's > Chaîn de Crevaux» is plainly the northernmost of the Koko-schili ranges, namely the one that bounds my latitudinal valley on the south. This range is the one that is easiest to trace, for there is reason to believe that each of the routes which cross my big latitudinal valley and which proceed from its lowest part towards the south, have to converge upon the pass in this northernmost range of the Koko-schili. I will however leave out of account the first of the passes that I crossed over in the middle one of my three routes, because there is no pass to correspond to it in the westernmost of those routes. On my general map I have endeavoured to trace what I take to be two parallel Koko-schili ranges, and for the northern one we have, proceeding from west to east, the altitudes 5020, 5337, 5151, and 5042 m., giving a mean of 5138 m. For the southern range we have only three altitudes, namely 5242, 5095, and 497o m., or a mean of 5102 m. In the case of both of these, we may say, that as a rule they grow lower towards the east. The numerous passes which exist in certain parts of the system point however to the presence of minor parallel ranges intercalated between the larger ones, unless indeed they are simply spurs of secondary importance. For this reason it is almost impossible to speak of the existence of anything at all approaching to a definite latitudinal valley between the two ranges that I fancy we may trace; and the figures 4884, 4904, 5172, 5074, and 5000 m. (this last an interpolated estimate) are no doubt the altitudes of depressions between the two ranges, but they need not on that account belong of necessity to identically the same valley. The datum 5172 m. may however quite well belong to a swelling, corresponding to a cross-threshold, from which the latitudinal valley slopes away towards both east and west, although in point of elevation it surpasses the mean pass-altitude in both the northern and the southern Koko-schili range. The mean altitude for the depressions mentioned is 5007 m.
However uncertain these mean values are, owing to the fewness of the data, they nevertheless serve to suggest one or two important reflections. In the first place, we see that the latitudinal valley between the two Koko-schili ranges is of quite a different rank from the latitudinal valleys which Wellby and I explored; for these are especially broad, sharply defined, and made up of a series of self-contained basins, each with a salt lake at the bottom of it. Thus they are real, independent mountain-systems which these great valleys separate one from the other. The valley however which lies midway between them, 1. e. that which lies between the Koko-
Hedin, Tourney in Central Asia. 1V. 72
IIIF Color HighRes
