is only known in part. As for the eastern portion of that zone, and it is with it alone that we are now concerned, the very fact of this irregularity of altitude along routes which run so close together proves that the surface features are very uneven. In some places the country is very flat, in other places it rises into particularly lofty swellings, and in yet others it is strangely broken by irregular stretches of hills. The latitudinal valley which occurs in the southern portion of the area expands into one of the largest internal-drainage basins in Tibet, forming a homologue to the Kum-köl basin in the north. In the same zone we find also the biggest river of the internal-drainage region of Tibet, namely the Satschu-tsangpo. Were the surface conformation here the same that it is in the north, this river would be broken up into several smaller streams, belonging to various latitudinal valleys, each possessing its own terminal lake. But the great flatness of the country and the absence of water-dividing ranges are circumstances which directly facilitate the gathering of the waters into a big river. And it is owing to precisely the same causes that such a large and extensive lake as the Selling-tso has been able to come into existence. The western continuation of this latitudinal valley narrows considerably, and in consequence the lakes which lie in that direction, namely the Tschargut-tso, Addan-tso, and Dagtse-tso, are of far smaller area, the first-named being also very narrow. Nevertheless it is quite evident to the eye that in this particular latitudinal valley the ranges do run very consistently east and west, e. g-. on the shores of the Tschargut-tso and on the large island of the Naktsong-tso. But opportunities for guesses and speculations are exceptionally plentiful in this region, and it is very difficult to determine in which direction and in what manner these ranges merge into the world of big mountain-ranges amongst which the head-feeders of the Indo-Chinese rivers are situated. Bower's, Bonvalot's, Rockhill's, and my own routes, which are here interlaced together, are not sufficient to justify us in drawing safe conclusions. I believe however, that the ranges which our routes cross over, after having described a curve towards the north, incline towards the east-south-east and south-east. If that is so, the Satschu-tsangpo would break through several of them in its upper course; but for the greater part of its lower course it will flow south of the most southerly of them, and parallel to it.
On the provisional general map which I have drawn I have, with the help of itineraries, attempted to enter all the mountain-ranges that exist in this region, although in so doing I was haunted by a feeling of great uncertainty. Possibly after the more deliberate map on the scale of i : 1,000,000, which is to be included in my atlas, shall have been completed, it may then be easier to draw sound conclusions. My present attempt is, as I have already said, entirely tentative.
For practical reasons I call this region, which extends from the Tang-la ranges southwards to the latitudinal valley with the lakes, Tschang, the name which the Tibetans themselves give to the high, uninhabited plateau; and here I will endeavour to distinguish seven ranges. The one farthest north, Tschang I, seems to be the easiest to follow, that is assuming that the following altitudes belong to this range — 4713 (De Rhins), 5473 (Littledale), 5236 and 52 I I (my two routes), and 5350 (Bonvalot). Thus the mean pass-altitude for Tschang I is 5197 m. The mean altitude of the latitudinal valley south of this range is 4821 m. In Tschang II we have
Hedin, Tourney in Central Asia. IV. 73
IIIF Color HighRes
