By the • Tibetan highlands with internal drainage, or the central plateau, I understand the whole of that part of Tibet which is bounded on the north by the water-divides of all the rivers that flow down to East Turkestan, on the east by the water-divides first of the rivers that flow into Tsajdam and secondly of the great Indo-Chinese rivers, on the south by the water-divides of the rivers which flow into the Tsangpo and upper Indus, and on the west by the water-divides of the rivers that empty themselves partly into the upper Indus, partly into the Jarkent-darja and Chotan-darja. The only district about which uncertainty can exist, as to whether it belongs to this highland region or not, is the basin of the Kum-köl. By reason of its peculiar situation and its relatively unimportant absolute altitude that basin may be considered to form a transitional zone between the central and the peripheral regions.
At all events this central internal-drainage country resembles in shape an isosceles triangle, having its base or greatest breadth in the east, from which it tapers away towards its apex in the west. We have found however that this region cannot be regarded — as hitherto has been the practice — as a plateau country in the ordinary acceptation of the term, but it is traversed by a great number of mountain-ranges, which in general diverge somewhat towards the east. When we come to consider the mean altitude of this region, our first procedure is to divide it, in accordance with its general surface configuration, into several different altitudinal zones. And first of all we have to deal with the true, dominating plateau, or fairly level base or socle upon which the mountain-ranges have been reared up; and its mean altitude is dictated by the mean altitude of the latitudinal valleys. Then we have to deal with the mean altitude of the depressions, and that is obtained from the mean of all the lakes; the absolute mean altitude for the depressions is obtained from the mean for the salt lakes, which occupy the very lowest absolute levels in all parts of the country. Finally, we have to consider the mean altitude of the cross-thresholds in the latitudinal valleys, as well as the mean altitude of the passes which cross the east-west mountain-ranges. For the next highest altitudinal level above that there exists no possibility of deducing a mean value, owing to
IIIF Color HighRes
