Sec. vii] GENERAL OBSERVATIONS ON DANDAN-UILIQ SITE 287
which appears to exclude the possibility of the Keriya river having ever in historical times flowed in the immediate vicinity of the site : I mean the relatively high level of the latter. The hypsometrical heights ascertained by me show that the ancient ground-level of the Dandan-Uiliq site in the immediate vicinity of the depression referred to lies 200 feet higher than the bank of the Keriya Darya at Kochkar-Öghil, where the river is nearest. The approximate accuracy of this estimated difference of level may the more be relied upon since it agrees exactly with the difference shown in Dr. Hedin's map 12. Now it seems to me exceedingly difficult to believe that the Keriya river should until eleven hundred years ago—geologically speaking a mere yesterday—have at this point followed a course fully 200 feet higher than the one which it follows now, and which lies in the direct line of its drainage. With nothing but soft loess or drift-sand to resist its course after the debouchure from the mountains the river must have found its bed within the depressed area where it now flows long before any historical period. Within this depression it may shift its bed to the west or east as long as its waters keep to the lowest level, but only engineering works on a large scale could ever force its whole volume to the high ground of Dandan-Uiliq 13.
From the levels recorded by Dr. Hedin and myself, it is clear that the ruins of DandanUiliq with the lands once surrounding them are situated on a band of high ground which forms the watershed between the Yurung-kash and Keriya rivers, and this fact has undoubtedly its antiquarian interest. Most unlikely as the line of a river-course, this is just the ground which the makers of an irrigation canal would for preference follow, as any properly contoured canal map of the Punjab or similarly irrigated area would show at a glance. By keeping to the line of highest level, a canal not only commands the maximum area of land capable of irrigation, but is also more easily protected against breaches by natural drainage channels and similar disturbing influences.
In view of the facts just set forth, it is impossible to accept the theory of a great shift of the Keriya Darya as an explanation why the site of Dandan-Uiliq was deserted and has remained unoccupied ever since. The immediate cause of the abandonment at the close of the eighth century is indicated with sufficient clearness by the antiquarian evidence already specified ; but there remains still the question as to the physical causes which have helped to render that abandonment permanent. The question seems to me closely connected with a far larger one, that of the desiccation which is assumed to have gradually been proceeding over large portions of Central Asia within historical periods. The consideration of this problem with special regard to the Khotan region is a task for which properly established archaeological data may be most helpful, but which itself can be dealt with only on a geographical basis, and it must therefore remain outside the scope of this work. The case of Dandan-Uiliq would prove of particular interest in connexion with this problem if conclusive topographical evidence were to confirm the view which I have set forth above, and which appears to me by far the most probable, viz. that a canal fed by one or several of the streams now irrigating the oases of Chira, Gulakhma,
12 While my hypsometrical heights for Dandân-Uiliq and Kochkar-Öghil are 4,290 and 4,090 feet, respectively, Dr. Hedin's map shows the elevations of 1,3x9 and 1,258 metres for his camps in practically the same positions. The corresponding figures in feet are (approx.) 4,327 and 4,127, respectively, giving the identical difference of zoo feet.
13 We shall see below, in discussing the site of Kara-dong, that so far as antiquarian evidence goes, these shifts of the Keriya river bed seem within historical times to have been
relatively very moderate. The ruins of Kara-dong, though probably considerably older than Dandan-Uiliq, are yet only a little over five miles removed from a bed which still receives water at flood times. The area between the ruins and this bed shows plentiful remains of dead forest, while these strips of ketek (dead forest), always marking old river-beds, are completely absent between Dandan-Uiliq and the Keriya Darya, a distance of about twenty-eight miles.
Watershed near Dandan-Uiliq.
Causes of permanent abandonment.