372 IN SEARCH OF THE LIMES TO SU-CHOU [Chap. XI
southern confines of Mongolia. But our efforts to engage one from among the Mongols grazing in the valleys above Kara-shahr had been frustrated by the official obstruction directed from
Urumchi. ` Malûm ', the stop-gap now produced, proved a hardy and fairly intelligent fellow,
dressed in a monk's red garb and possessed of some education. In spite of a somewhat irascible temper which involved him more than once in quarrels with the Torgut herdsmen, &c., encountered
on the Etsin-gol, he served us well, both on the march and during the excavations at Khara-khoto. Carried, with all that he needed, by his sturdy Mongolian mount, he caused the minimum of encumbrance. At the same time, in order still further to lighten the loads of our camels, I dispatched Surveyor Muhammad Ya.qûb, whom there was no opportunity of employing independently on the way to Su-chou, by cart to that place in charge of all equipment not immediately needed.
The rest of us started from An-hsi on April i8th and at a point north-east of the town crossed to the right bank of the Su-to-ho. The river in its wide deep-cut bed held nowhere more than two
feet of water at the time. This was clear proof that the first spring flood from the snows between
the outer ranges of the Nan-shan had passed, while the melting of the winter snow on the high plateau-like valleys drained by the Su-lo-ho head-waters had not yet commenced. We marched
along the narrow fringe of scrub-covered ground that separates the river-bed from the bare glacis of piedmont gravel sloping down towards it from the southernmost Pei-shan range, and thus approached, on the evening of the second day, the well-marked defile of the Su-lo-ho ; the river passes through this about eight miles above the village of Hsiao-wan, which is situated on its left bank.
I have already explained in Serindia the reasons which led me, when passing here in September, 1907, by the high road to An-hsi, to the conclusion that the Limes line traceable from the west
to the vicinity of Hsiao-wan crossed to the right bank near this defile.3 But on that journey I had
not myself been able to visit the ground on the right bank. I had indeed noticed towers, which looked as if they might have belonged to the Limes, near the lower end of the defile. Yet the above
conclusion was based mainly on the obvious strategic advantages that the configuration of the ground would have offered for taking the border line across the river just at that point. That farther east the Limes lay actually to the north of the Su-lo-ho had been established by the remains of its wall that I had traced near Shih-êrh-tun, north of Yü-mên-hsien (Map No. 40. c. 5).4
The defile, as already described in Serindia, is formed by a low gravel-covered offshoot from
the southernmost Pei-shan range, which juts out here with its end close to the river. On the opposite side it is faced by a rugged and somewhat higher spur, known as Wan-shan-tz&, the north-eastern
continuation of that outermost hill range of the Nan-shan which separates the valley of T`a-shih
and Ch`iao-tzû from the trough of the Su-to-ho.5 This spur falls off precipitously towards the river and is crossed at an elevation of about 200 feet above the latter by the high road leading from
Yü-mên-hsien and Bulungir to An-hsi. Where this road passes over a western outlier of the spur
before descending to the flat ground towards Hsiao-wan village, I had found it guarded by two large towers. But these in their present condition showed no signs of antiquity. Nor had I been
able to trace any remains of the Limes wall on the scrub-covered ground, once probably cultivated, intervening between the western foot of the Wan-shan-tzû spur and Hsiao-wan. Thus definite archaeological evidence of the Limes having here crossed the river was still lacking.
When approaching this ground on the present occasion by the right bank, we had passed at a distance of about thirteen miles from Camp 121 some deserted shepherd huts, marked strangely