National Institute of Informatics - Digital Silk Road Project
Digital Archive of Toyo Bunko Rare Books

by the erosive force of wind and driven 'sand', i. e. disintegrated loess. At all such points small
terraces of fairly hard loess soil, rising to heights of six to thirteen feet above the eroded depressions
close by, served as witnesses approximately indicating the ancient level of the ground. Thin layers
of pottery fragments on their surface helped to explain why these terraces had withstood the
unceasing attack of wind-erosion. The progress of the erosion could be observed on the sides of
the terraces, which everywhere showed the effects of under-cutting as plainly as a river-bank which
is being washed out by the setting current. What remains of modest dwellings I could still trace
on such ground, as at Kushuk-aste about a mile to the west of Khādalik, or at Kök-jigda, another
small 'Tati' about two miles south, consisted of low rush walls or fences which by their very weakness
offered less scope to the grinding force of driven sand, and on the contrary were apt to catch
and retain it as a cover.

Formation
of tamarisk-
cones. The same phenomenon was illustrated by the structure of the closely packed sand-cones which
covered most of the ground and encircled all open areas. Their height usually varied from about
twelve to fifteen feet above the original ground level ; but in places I observed cones rising twenty-
five feet and more. Tangled masses of tamarisk scrub, usually dead at the foot but still flourishing
on the top, invariably covered these hillocks. There could be no doubt that the latter owed their
origin to the tamarisk bushes, which had first overrun the fields when cultivation slackened and
ceased, and had then served to catch and collect the drift sand passing over the ground with the winds
of the spring and summer. The low tamarisk thickets I saw spreading over the fields of 'Old
Domoko' and Ponak, abandoned during the nineteenth century,⁴ fully illustrate this initial stage of
the process. The struggle for light and air, which the tamarisk bushes, once rooted on level ground,
have to carry on against the sand steadily accumulating around them, forces their head branches to
rise ever higher and higher. The sand, the smothering embrace of which they try to escape,
naturally follows this rise, and the cones formed by it thus grow correspondingly in size and height.

Rate of
growth of
tamarisk-
cones. The structural process here briefly indicated must have been essentially the same wherever the
great desert of the Tārim Basin is edged by this characteristic zone of tamarisk-cones. But the rate
of growth may have varied considerably according to local conditions connected with climate, sub-
soil water, prevailing winds, and the like. Hence it was of considerable interest that an archaeo-
logical observation enabled me to determine with quasi-chronological exactness the time which this
building-up process here took. At the edge of a small area of open eroded ground about half-way
between Khādalik and Balawaste my guides showed me a few fragments of small relievos in hard
white stucco emerging from the slope of eroded loess soil immediately at the foot of a big tamarisk-
covered sand-cone. Such slight digging as the masses of sliding sand permitted us to make into the
side of the hillock, brought to light more stucco fragments which once formed part of the relievo
decoration of some big halo in a Buddhist shrine. That this dated from the same period as the
temples of Khādalik was made certain by the style and technique of these fragments of which speci-
mens will be found described in the list below.

Relievos at
foot of
tamarisk-
cone. The abraded condition of the relievo fragments representing small standing Buddhas and floral
borders left no doubt that they had been exposed for a long time to corrosion by driven sand, until
the expansion of a neighbouring sand-cone came to provide protection for these humble remnants of
a shrine otherwise completely destroyed. The level on which they were found was about three feet
higher than the top of the nearest 'witnesses', while the witnesses themselves rose six to ten feet
above the bottom of the eroded depressions adjoining them. This difference of about three feet agrees
with that noted at Balawaste and clearly represents the progress made by erosion since the relievo