国立情報学研究所 - ディジタル・シルクロード・プロジェクト
『東洋文庫所蔵』貴重書デジタルアーカイブ

left side of the valley, where 3 fluvial terraces, each about 3 m. high, are distinctly
visible. The rest of the way goes up and down over hills and more or less deep-
cut ravines. In the largest of these ravines Serolung-gompa is situated at a brook
which soon disappears underground. At Camp 212 the open level plain between
the water-line and the foot of the hills is only 84 m. broad. Just above the Camp,
where the hills are very steep, six terraces or old beach-lines are faintly visible, the
highest being 49.5 metres above the surface of the lake. It is difficult to account
for them, without a very detailed survey being undertaken all round the lake.
Only the highest of the 3 above-mentioned terraces in the valley of Samo-tsangpo
may be attributed to the lake and not to the river, as a certain divergence appeared
to exist between it and the two lower terraces. At Na-marding I have mentioned
some terrace formations with perfectly horizontal surfaces. The difficulty is to bring
these old remainders in accordance with the erosion in the valley of the effluent from
the Rakas-tal. The basin of both lakes may, however, have been filled with æolian
deposits during a dry period of steppe climate, simultaneous with the period when
the great loess deposits were formed in China and Mongolia, and the valley of Guge
was also filled. During the fluvial period which must have been contemporary with
the ice age in the north, the rivers broke their ways through these soft beds. As
soon as the valley of the effluent was established, the surface of the Manasarovar
cannot possibly have stood more than a few, perhaps 3 or 4 m., higher than now,
as the lake was constantly drained by the effluent. But as I said before this prob-
lem can only be solved after a very detailed examination of the whole region round.

Following the shore by boat south-westwards from Serolung-gompa one sees
how the last hills along the lake become lower, until they are hardly more than
10 m. high near the mouth of Tage-tsangpo. At certain intervals ravines open out
to the lake, most of them dry in 1907. The lake is shallow the whole way and
the isobathe of 1 m. is about 50 m. from the shore.

Tage-tsangpo enters the lake without forming a delta; but at the right side
of the mouth there is a long, narrow cape with a mud island in its prolongation.
At the left side there is another mud island in the form of a crescent. Even now,
in spite of the dry summer, the water of the river was rather muddy, which, how-
ever, does not depend upon the fact that it is chiefly fed by glaciers, for at Lang-
chen-kamba the water was very clear, pointing to the fact that it does not come
direct from the glaciers, but probably passes some moraine lakes or becomes filtered
in old moraines. The dimensions of the river were now: breadth = 17.30 m., average
depth = 0.727 m., average velocity 0.911 m., and volume 11.26 cub. m. a second, on
August 1st, 1907. Thus it had nearly 3 cub. m. more than a few days earlier
north of Tso-nyak. The difference did not depend on any changes of weather, but
obviously on tributaries joining the river from the Gurla below Tso-nyak.

The river makes a very curious bend before entering the lake. Instead of
reaching the shore at right angles, it turns off sharply to the north keeping parallel