436 THE ENDERE RUINS [Chap. XII
establishment in it of a fortified post, for the accommodation of a small Chinese garrison, and perhaps of some local authority which the former was to support and control, would in that case have probably followed as a matter of course. Or if the tract remained deserted, as Hsüan-tsang saw it in 645 A.D., we should have to recognize in the Endere fort a small stronghold established on the great route from China to Khotan for the express purpose of assuring its safety.
It seems difficult to decide between the two alternatives thus presented. The Tang Annals furnish us, indeed, with an itinerary for the old route leading from Sha-chou to Khotan as it existed during this period of Chinese supremacy13. But though M. Chavannes' notes on it clearly show that the route passed through Chü-mo (Tsiu-mo) or Charchan, where a Chinese garrison is specially mentioned, it is impossible to identify with certainty any of the stages recorded on it further west until we get relatively close to Khotan, as the distances between them are indicated only in one instance 14. On the whole, I am inclined to favour the first supposition, in view of the analogy offered for it by the case of Chü-mo (Tsiu-mo) or Charchan, which, though wholly deserted in Hsüan-tsang's day, was again inhabited and the seat of a Chinese garrison in Tang times15.
Marco Polo's description, too, ` of the Province of Charchan ' would agree with the assumption that the route west of Charchan was not altogether devoid of settlements even as late as the thirteenth century. After referring to the numerous towns and villages of Charchan', he tells us : ' The whole of the Province is sandy, and so is the road all the way from Pein, and much of the water that you find is bitter and bad. However, at some places you do find fresh and sweet water 1ß.' This account of the route agrees accurately with the conditions now met with between Niya and Charchan. Yet in the passage immediately following the Venetian tells us how ` when an army passes through the land, the people escape with their wives, children, and cattle a distance of two or three days' journey into the sandy waste ; and, knowing the spots where water is to be had, they are able to live there, and to keep their cattle alive, while it is impossible to discover them '. It seems to me clear that Marco Polo alludes here to the several river courses which, after flowing north of the Niya-Charchan route, lose themselves in the desert. The jungle belt of their terminal areas, no doubt, offered then, as it would offer now, safe places of refuge to any small settlements established along the route southwards.
The presumption that the Endere fort had by its side a cultivated area with a settled population would, of course, be definitely established if it were proved that the débris-strewn area to the north-west of it, together with the Stûpa, dated from the same epoch. U nfortunately, conclusive evidence is not at present available on this point. No coins were found
13 Comp. Chavannes, Voyage de Song Yun, p. 1z, note 9.
" We are told that the route westwards of Tsiu-mo 'passed the Hsi-li-chih (Si-li-Iche) wells, the Yao wells, the river Hu-chi' (Hou-tcho), and, after Soo li, arrived at the military post of the town of Lan, which is east of Yü-t`ien'. If by the latter expression the eastern border of Khotan territory is meant, i. e. Niya, we might be tempted to suggest the location of
the military post of the town of Lan ' at the Endere site. But in the absence of exact evidence there is little profit in such identifications. The mention of wells as stages west of Tsiu-mo suggests that the country traversed was without permanent habitations, yet just as at the present day offered water in wells dug at regular halting stages.
15 It is noteworthy that Sung Yün, travelling in 519 A. D., describes Tso-mo, which M. Chavannes has shown ( Voyage de Song Yun, p. 13) to be identical with Hsüan-tsang's and
the Tang Annals' Chü-mo (Tsiu-mo), as ' a town ' inhabited by a hundred families only. Charchan, in fact, seems to have passed repeatedly through such successive periods of contraction and expansion, having developed during the last fifty years from a small penal station in the time before Yaqûb Beg's rebellion into a flourishing settlement of steadily increasing prosperity ; comp. Yarkand Mission Report, p. 34 ; Hedin, Reisen in Z.-A., p. 178. Examples of such fluctuations in importance and size of detached oases in this region could probably be greatly multiplied if our records were ampler. They may serve as a warning against any a priori assumption that steady decrease in the size of these oases must always have been the rule during historical times.
1fi See Yule, Marco Polo, i. p. 194. For the location of Pein in the close vicinity of the present oases of Gulakhma and Domoko west of Keriya, see below, chap. xIII.