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

of surface water, has been scooped out and sculptured by the same powerful east wind which is
constantly at work in the Su-lo Ho basin into small ridges and trenches invariably running east to
west.⁴ These reproduce the Yārdang formations of the Lou-lan area in the Lop Desert with
striking fidelity, though on a smaller scale. The lesser depth of the scourings, usually between two
and five feet, is accounted for by the fact that, as we shall presently see, the time that has elapsed
since the disappearance of surface water and protective vegetation is here fully a thousand years
less. On the other hand, the thickness of the surface layer of loess is here limited, and it is obviously
due to this that I found it in places completely abraded and carried off right down to the underlying
fine gravel. Much of the sand brought down originally from the detritus-covered hill range is coarse,
and its corrasive action here must considerably aid deflation, i. e. wind-erosion pure and simple.
The broad belt of eroded loess stretched away eastwards as far as the eye could reach, bare of all
sand. But within the area of the old oasis surviving scrub has helped to retain small dunes in
places, as already stated, and within the ruined town the protection still afforded by the massive
walls has caused the drift-sand to accumulate in big dunes.

In connexion with the physical features just described it may be as well to state at once what Traces of
my survey of the ground, rapid as it had to be from want of time, allowed me to ascertain as to the ancient
water-supply once available for the ruined town and the cultivated area near it. That it could not irrigation.
have been derived from the spring-fed marshes, the drainage of which irrigates the present oasis of
Ch'iao-tzŭ, is certain. A look at the ground was enough to show me that all these springs lie con
seen below the level of the area of deserted occupation. No clear trace of any stream could be
seen from So-yang-ch'êng on the bare gravel glacis that slopes up for miles to the foot of the hill
range. But when I was riding, towards the end of my visits, to the site across the wind-scoured
ground to the east of the ruined town,⁵ my eye was caught by a low gravel-covered ridge coming
from the south-east. It proved, as expected, to carry the line of an old canal with its banks still
clearly marked on the top. The coarse sand and pebbles which the water used to bring down from
the foot of the hills had necessarily in the course of time helped to raise the canal bed, as noticed
in every oasis from Khotan to An-hsi. Subsequently, when the canal ceased to carry water, the
same heavy deposits protected it from the force of the winds which have, since irrigation ceased,
been continually cutting up and lowering the ground on either side. Thus the top of the canal
banks has come to lie 10–12 feet on the average above the level of the eroded ground on either
side.

The line of the canal was seen to throw out several smaller branches at about one mile's Line of
distance from the ruined temple α. Thence it could be traced quite clearly up to a point about ancient
4 miles to the south-east of the ruined town, where a solid mound of clay (marked ε in Map No. 83. B. 4) canal.
rose by its side. Beyond this the line disappeared on ground completely furrowed by Yārdangs.
On the south of this eroded belt, about half a mile wide, the bare gravel glacis was seen to rise
steadily, and on this no traces of the canal bed could be expected to survive. But far away in the
distance to the south-east I sighted a line of white clay cliffs which suggested a river-course now
dry and sunk into the gravel glacis. A broad gap seen south of it in the nearest hill range made
me think that this course had been cut by the drainage from the next higher range, which, as
subsequent surveys showed, rises to heights of 12,000 feet and more, and which, though carrying no
permanent snow, is yet likely to receive even now at times a fair amount of moisture.⁶ There could