________________
198
THE INDIAN ANTIQUARY
[SEITEMBER, 1917
But a systematic search of this and other ruins soon showed that the archæological riches of the site were by no means exhausted. By a careful clearing of the debris which covered the bases of Stupas and the interior of temple cellas we brought to light abundant remains of Buddhist manuscripts and block prints, both in Tibetan and the as yet very imperfectly known old Tangut language, as well as plenty of interesting relievos in stucco or terra-cotta and frescoes. The very extensive refuse heaps of the town yielded up a large number of miscellaneous records on paper in the Chinese, Tangut, and Uigur scripts, together with many remains of fine glazed pottery, and of household utensils.' Finds of Hsi-hsia coins, ornaments in stone and metal, etc., were also abundant, particularly on wind-eroded ground.
There was much to support the belief that the final abandonment of the settlement was brought about by difficulties of irrigation. The dry river-bed which passes Kharakhoto lies some 7 miles to the east of the nearest branch still reached by the summer floods. The old canals we traced, leading to the abandoned farms eastwards, are removed considerably further. It was not possible to determine by conclusive evidence whether this failure of irrigation had been the result of desiccation in the Etsin-gol delta or been caused by some change in the river-course at canal-head, with which the settlement was for some reason unable to cope. But there seemed to me good reason to believe that the watersupply now reaching the delta during a few summer months would no longer suffice to assure adequate irrigation for the once cultivated area. Even at the Mao-mei oasis, over 150 miles higher up the river, and with conditions of ground far more favourable for the maintenance of a system of canals, serious trouble had been experienced for years past in securing a sufficient discharge early enough in the season, and much of the once cultivated area seemed to have been recently abandoned.
With the rapidly increasing heat, work at the desert sites' had become very trying both for the men and our camels, upon which we depended for the transport of water. With the completion of our task at Khara-khoto, and of the surveys which had meanwhile taken Lal Singh to the terminal lake-basins of the Etsin-gol, I was glad to let the hard-worked camels depart for their much-needed summer holiday in the Kongurche hills north-eastward and to start myself with Lal Singh south to the foot of the Nan-shan. The new route, which we were able to follow for part of the journey, took us through hitherto unexplored portions of the desert hills to the east and north of the river of Kan-chou. But owing to the heat and the scarcity of spring it implied serious fatigues, and it was a relief when Kan-chou was safely reached before the close of June.
A short but refreshing halt in that large and pleasant oasis was devoted to the arrangements needed for the new surveys I had planned in the Central Nan-shan. Their object was to extend the mapping, which in 1907 we had effected in the high mountains near the sources of the Su-lo Ho and Su-chou River, hy accurate surveys of the high ranges further east. containing the headwaters of the river of Kan chou. In conjunction with our labours in the Etsin-gol region, they were intended to complete the mapping of that large north-western portion of Kansu which, inasmuch as it sends all its waters into drainageless basins, may well be claimed in respect of its hydrography and general physical conditions as belonging to Central Asia rather than to China. Knowing the reluctance of the local Chinese to venture far into those mountains, I was prepared for the difficulties experienced at the outset in securing transport. But a fortunate chance brought just then an old Chinese friend to the military command of Kan-chou in the person of worthy General Tsai, whose kindness I remembered so well from my visits to Su-chou in 1907, and his opportune help enabled us to set out for the mountains by the first week of July.
