382 THE ANCIENT SITE BEYOND THE NIYA RIVER [Chap. XI
caution ; for it is evident that the time when it was made and used must lie many centuries further back than the period to which the structural remains of the site belong.
It is thus obvious that the peculiar conditions created by erosion at this and also at other ancient sites of this region must impose limitations upon the conclusions of the archaeologist. Only where structural remains have survived can we hope to determine the period of simultaneous occupation and the approximate date when this occupation ceased. Where the ancient remains are of the ` Tati ' type, i.e. consist only of pottery fragments and other small débris, finds of coins, seals, and other approximately datable objects may help to show the latest period down to which the ground was in all probability occupied or cultivated. But not until we have learned to distinguish with certainty between the potsherds of different periods, will it be possible to form any definite conclusions as to how much of the débris really dates from that latest epoch, and how much represents remains of earlier habitations. Seeing that the latter need not have been built always in the same positions which were occupied by the dwellings of the period immediately preceding the abandonment, it is evident that the mere extent of débris-strewn patches of ground by itself furnishes no safe argument regarding the character and size of the settlement which had last occupied the site. It will, therefore, be well to restrict ourselves to the positive evidence which the examination of the structural remains has supplied.
There can be no doubt that all the ruins with which the excavations described in the preceding sections have made us acquainted were inhabited within the same period and also abandoned about the same time. The area over which they are scattered covers fully seven miles, from N. viii in the north to where the first dead fruit-trees were met with at the approach from the south, while its greatest width, as marked by N. i on the east and N. xII on the west side measures over two miles. It is possible that in those portions of the area where the dunes are higher, ruins of substantial buildings may lie completely hidden from view ; but even allowing for this, and for the complete disappearance of others through erosion, we must bé struck by the relatively small number of extant ruins of any size and by their wide dispersion. We have already had occasion to note the same fact at Dandan-Uiliq, and must hence attach to it the , more significance.
Two observations appear to me most likely to explain it. In the first place, it is evident that relatively substantial houses would necessarily offer longer resistance to the destructive forces of erosion than the mud-built dwellings of small cultivators. Hence we may recognize that the extant ruins represent only the few well-constructed residences of the ancient settlement. In the second place, the examination of any modern oasis too small to possess a town centre, such as Guma, Moji, Chira near Khotan, or else Tashmalik, Khan-arik, &c., near Kashgar10, would show us exactly the same scattering of the population in numerous small hamlets widely separated from each other by stretches of cultivation and gardens or even by strips of desert. In none of those oases is there any agglomeration of houses that with any approach to accuracy could be called a town. The ` Bazar ' which forms their commercial centre is nothing but a street of booths practically without inhabitants on any but market days, while the well-to-do ` Begs ' who, as Mingbashis or in other official posts, carry on the whole of the local administration, generally reside far away in the hamlets or in isolated farms nearest to their landed property 11.
1° Comp. Ruins of Khotan, p. i16, regarding Tashmalik, where I first had occasion to learn practically how widely separated the habitations even in a populous oasis may be, and how difficult in consequence it is to find their administrative centres. See also Yarkand Mission Report, p. 32, for
some judicious observations.
11 Compare for descriptions of such residences belonging to Bégs either in office or in spe, where I had occasion to take up temporary quarters, Ruins of Khotan, pp. 152, 153, 199, 257, 2ITr.
Extent of ancient site marked by structural remains..
Small number and scattered position of substantial dwellings.
Scattered hamlets and country residences in modern oases.