108 THE ROUTE FROM KARGHALIK TO KHOTAN [Chap. V
ruins of the timber and plaster buildings of the Niya River Site were almost always found occupying such terraces, while all round them depressions were formed through erosion. To whatever specific cause or combination of causes we may have to attribute the preservation of loess banks at Tatis, it is certain that we have in them evidence of the original ground-level. As such, they are equally interesting to the antiquarian and the geologist, and it seems appropriate to apply to them the designation of witnesses' (or Zeuge') familiar to geological terminology,
These witnesses' are, of course, of special value wherever we find on or near them archaeological remains which are approximately datable ; for such remains clearly mark the highest chronological limit for the commencement of that process of erosion of which the results are seen in the depressions formed around. Such datable remains must on ` Tatis ' proper be necessarily of rare occurrence. Owing to the fact that they are ordinarily within easy reach of localities which have remained inhabited, these sites at all times have been particularly exposed to destruction from human agency.
My observations at deserted villages of modern date near Domoko may serve as a typical illustration to show how soon after their abandonment such sites were bound to be cleared of everything of any value or that might serve as building material or fuel 8. Timber has always played an important part in the construction of Turkestan houses. It is easy to see how its early abstraction must have converted all remains of ordinary dwellings into crumbling mud heaps within a short period after their abandonment. The regular visits of ` treasure-seekers ', whose profession appears to have flourished at all times about Khotan, not only helped to complete the destruction of all perishable materials, but gradually led to the removal of any small objects of value, such as coins, metal ornaments, cut stones, &c., which might have been hidden in the ground by design or accident, and which by their harder substance were proof against decay or erosion.
Other ancient objects, such as written records, sculptures in plaster, wood-carvings, &c., which the deserted settlements probably once contained, and about which the ` treasure-seeking' visitors of ` Tatis' in former days certainly never troubled, would now, no doubt, be most valuable to us for the archaeological dating of these sites. But we have seen already that the physical conditions absolutely preclude the possibility of such relics surviving 9. Fragments of ancient pottery, indeed, strew the ground of all ` Tatis ' in abundance ; but in the present state of our knowledge it is impossible to attempt a proper classification of these potsherds with a view to ascertaining the chronological relation of particular types, materials, &c. Even hereafter, when some progress may be made by the detailed examination of ancient pottery at sites where it has been protected by drift-sand or silt deposit, the proper classification of potsherds found at Tatis will present particular difficulties, owing to the small size of the fragments and the effacement which the surface of the rare decorated pieces has undergone through the ` grounding ' already referred to.
Thus the only relics at Tatis likely to afford guidance as to the period from which they date, are coins, cut stones, and small objects of metal work. Unfortunately I was unable to
8 See below, chap. mt.
9 In the accounts given by Islam Akhün of his alleged discoveries, his written or block-printed ' old books ' in unknown characters ' often figured as having been found exposed on the bare surface at certain desert sites. As far as these sites had any existence at all, Tatis were meant suçh as the forger had occasion to see near Güma, Ak-sipil,