Sec. vii] GENERAL OBSERVATIONS ON SITE 383
It is to the residences of these Begs, often remarkably well-built, judged by Turkestan standards, and comfortable in spite of their wholly rural surroundings, that I think we must look for the modern counterparts of the more substantial ruins of the Niya Site. Thus guided by what observation at the present day teaches, we can readily understand the relatively high standard of comfort which these ruins indicate, even though the description of their surroundings as a ` town' or ` city ' must appear inappropriate.
As to the population which this ancient settlement once contained we could only form vague guesses, but a look at the map shows that in size the area for which simultaneous cultivation can be proved by definite archaeological evidence approximately corresponded to the present oasis of Guma, Chira, or Tawakkel. It is still more certain that the ancient settlement, like the first two of the modern localities just named, must have been a typical terminal oasis 12. In contrast to Dandan-Uiliq, no question whatever can arise here as to the source of its water-supply. It can only have come from the Niya river, the course of which still bears straight in the direction of the ruined site, but now finally ceases, even during the summer floods, some seven to eight miles from the southernmost point of the latter. Taking the direct distance between the northernmost structural remains explored by me and the last point down to which I could trace the flood-bed cut in recent times by an extreme off-shoot of the river, we find that the belt fertilized by the latter has receded some fifteen miles since the abandonment of the ancient site. This distance is reduced to about eleven miles if we take our measurement from the northernmost ruins to the limit of living trees as marked on the map.
That the abandonment of the site must be connected with this shrinkage of the river course, whatever its explanation, is an assumption which naturally suggests itself first. But the historical student, and I think also the geographer who is prepared to scrutinize his evidence critically, may well pause before committing himself to the conclusion that this undoubted recession of the river's terminal area must necessarily have been the immediate or sole cause of the abandonment. In the case of all rivers and streams that flow from the Kun-lun range into the Tarim Basin and lose themselves in the sands of the Taklamakan, we find the final portion of their course liable to great and frequent shiftings laterally, i. e. to the east or west. The phenomenon has long ago been noted by all competent observers and finds its natural explanation in the masses of silt, turning into drift-sand, which these rivers deposit in their terminal course, and which in turn force them to divide their waters delta-like or periodically to seek new channels.
For a typical illustration we need not look further than at what my map shows for the terminations of the Yartunguz and Endere rivers nearest to the Niya river on the east. As I shall have occasion to mention in the next chapter, the relatively recent colony of J igdalikBulung and Kala-sulaghi, forming the terminal oasis of the Yartunguz river, had for several years previous to my visit been in danger of losing its water-supply, not from any want of water in the river but owing to the head of the river having shifted considerably to the west and thus failing to fill the canal from which the fields of Yartunguz-Tarim are irrigated. The people of the small colony being far too few to cope with this danger by strengthening the head of their canal or otherwise confining the vagaries of the river, have resorted to the expedient of starting fresh cultivation in the new terminal area known as Yilba-Sarigh (see map). That the same river had at different periods followed courses also to the east of the present Yartunguz-Tarim was demonstrated when I passed the broad depressions shown by