Sec. ii] PAST THE MAO-MEI OASIS AND ITS OUTPOSTS 413
of the eastern wall on the outside. The walls of the fort had suffered but little except on the north
and south faces. There they showed breaches which did nbt appear to be due to wind-erosion.
A much larger outer enclosure adjoined this fort on the north and east ; its walls, also of stamped
clay, but only 5 or 6 feet thick, could be traced for about 700 and 500 feet, respectively. These
walls were badly decayed, but towers of varying sizes intended to strengthen them survived in
several places. The position of the end of the northern wall towards the river could not be deter-
mined, and the western wall had disappeared completely, having probably been carried away by
a flood of the river.
Within the inner fort, walls of two small structures survived to a height of a few feet, the masonry Uncertain
showing courses of vertically set bricks. Excavation brought to light only remains of the roofing ; date of fort.
nor did the refuse found in the south-
west and south-east angles of the en-
closure yield any datable objects. The
clearing of two shallow depressions, prob-
ably marking the position of wells, one
in the inner, the other in the outer fort, """''
could not be attempted. So for chrono-
logical indications I had to fall back on
what could be gathered from the remains
of pottery ; these were very plentiful, especially within the outer circumvallation. Among the potsherds fragments of hard grey ware, with or without
string or mat marks, prevailed, distinctly suggestive of antiquity. Pieces of plain whitish-grey glazed ware were few. Most significant, perhaps, was the
total absence, so far as I could observe, of porcelain and of such glazed ware as
I had found at the sites of the Sung R~
period that I had previously examined.
The irregularity in the plan of the " 't ° SO .00 "° '°°
do. do. broken +syg
inner fort speaks against its attribution
to Han times, at least in its present shape, while the absence of such pottery as is common at Khara- Origin of
khoto or Ch`iao-tzû makes it difficult to believe that the stronghold could have seen prolonged defences.
occupation either during the period of Sung or that of Hsi-hsia domination. I was therefore led to conjecture that the construction of the Taralingin-dürüljin defences might originally date, perhaps, from the troubled epoch which prevailed between the close of Han times and the advent of the Tang dynasty, and again after the Tibetan conquest of the Kan-su marches (c. A. D. 75o), when inroads of Turkish and other nomad tribes from the north must often have threatened the security of the Chinese settlements along the Kan-chou river and the foot of the Nan-shan. But some of the repairs may well be due to later temporary occupation.
A ride of two miles diagonally across the widening bed of the river brought us back again to First
the Etsin-gol route on the left bank.. There in a small Mongol encampment, the first met with, mongol
` Mâlum ', our itinerant Làma, discovered a relative in the owner of one of the few felt tents. This unexpected meeting was not without its interest. Mâlum had drifted years before to the Tun-
SKETCH ?IAN OF
T. XLVIII. d
ON ETSIN - GOL