284 THE RUINS OF DANDAN-UILIQ [Chap. IX
clearness. But here, too, it may be safely asserted that there is no object that would suggest for its origin a date later than the eighth century. Nor is there, with the possible exception of some fragmentary Pathis, any among them which is likely to prove of an appreciably earlier period than the last decades preceding the abandonment. In the case of the wall paintings, the stucco images, and other objects of worship found in the small shrines, the perishable nature of the materials precludes the idea that they were already of considerable age previous to their burial under the drift-sand, while for the documents in Brahmi: script everything, as we have seen, suggests an origin contemporary with the Chinese records.
The value of the latter, however, is not limited to supplying us with a firm chronological basis. They also establish the interesting fact that Chinese administrative control and a Chinese garrison within the Khotan region was a reality down to the end of the eighth century. M. Chavannes, in his concluding remarks on the Chinese records from Dandan-Uiliq, has called attention to the remarkable agreement which the dates of these documents present with what the historians of the Tang period tell us of the end of Chinese sovereignty over the ` Four Garrisons '. The historical synopsis given in a previous chapter has already acquainted us with the fact that it was the year 791 which saw the final abandonment of the whole of Eastern Turkestan to Tibetan invasion, after a long period of isolation and struggle B. Thus our Dandan-Uiliq records, closing with the year 790, represent, as M. Chavannes has justly pointed out, the last trace of Chinese political influence under the Tangs in that great region 7.
But may we not, apart from this general historical interest, attribute to these relics of Chinese occupation a further specific significance for the history of the site ? The letter which the Chinese commandant of Li-hsieh addressed to the king of Khotan in 768 A. D. (Document A) has shown us in most authentic form that already at that time the settlement to which DandanUiliq belonged had lost a part of its population, which had retired to the main oasis owing to depredations of bandits 8. Seeing how seriously local insecurity had affected the settlement in the immediately preceding period, it seems difficult not to connect its final abandonment after 790 A. D. with the great political upheaval of the years next following. The collapse of Chinese authority and the successful Tibetan invasion must have meant for Khotan a period of exceptional trouble ; for Tibetan power is, from all that we know, likely to have asserted itself mainly in unchecked raids of large marauding bands, such as had already devastated extensive parts of Western China. In a previous chapter I have endeavoured to show that the constant struggle with the drift-sand of the desert, which the outlying parts of every oasis on the edge of the Taklamakan have to carry on, cannot be maintained successfully except with an effectively supervised system of irrigation and an adequate population. Both conditions are likely to fail during prolonged periods of political trouble, and in no part of the cultivated area are the effects of such failure bound to appear so surely and rapidly as in isolated colonies like Dandan-Uiliq.
For those who are familiar with tracts like the greatest part of the Western Punjab, where cultivation is wholly dependent upon an elaborate system of canal irrigation, it needs no great effort of imagination to realize the effects of prolonged political troubles and insecurity on the cultivated area. As soon as the want of a firm central authority or a diminution in the available supply of labour stopped the effective working of the canals the reduced water-supply would force the cultivators to retire from all outlying lands however fertile by nature. Then