Sec. i] RUIN N. i AND FIRST FINDS OF INSCRIBED TABLETS 327
inquiry such as the wedge-shaped tablets above specified, belong to this category must be clear to any one familiar with the routine of an Eastern ` Daftar'.
The fact that, among all this wealth of records on wood, not a single scrap of paper had been found seemed from the first a strong confirmation of the early date which palaeographic
considerations indicated for my discoveries. It was evident that the use of paper, so much
more convenient than wood, however old it may have been in China, had not yet spread to Eastern Turkestan at the period when this ancient settlement was abandoned. The possibility,
however, of its having known another writing material besides wood suggested itself when, among scanty débris of old pottery, ropes of twined rushes, and timber found under the sand in room N. ii. and outside it, I had come upon small cut pieces of thin sheepskin which looked as if prepared for writing. Yet it was left for another rich mine of ancient records (N. xv.) to verify the surmise which I noted down at the time.
The undisturbed condition in which I had found the contents of N. iv., though lying so close to the surface, was reassuring proof that the ruins of this site could not have suffered
so much as those of Dandan-Uiliq from burrowings of ` treasure-seekers '. But, on the other
hand, with the view of the ruined building before me as seen in Fig. 37, it was impossible to ignore the fact that the havoc wrought here by wind-erosion had been distinctly greater. The
extent to which the destructive power of the desert winds had asserted itself in the course of
long centuries could be measured by the 12-15 feet difference of level separating the small plateau occupied by the ruin from the immediately surrounding ground. There could be no
doubt that the former, protected by the walls and débris of the structure, had retained the
original level, while the open surface near by had been steadily lowered by erosion. As the drift-sand carried at present over this portion of the ancient site is insufficient to fill the
depressions scooped out or to cover the ruins, the raised ground bearing the latter is being
steadily cut into and undermined, just as if it were exposed to the action of running water. The result finally produced by this slow but steady process of destruction is illustrated by the
photograph just referred to ; for the heavy timber débris which is seen there, strewing the slope of the foreground or in places overhanging it, represents the last remains of the foundations and superstructure of a part of the original building, which has completely fallen owing to the soil beneath having been eroded.
Exactly similar conditions were observed by me around all other extant ruins of this site, the strips of ground occupied by them rising island-like above the level of the adjoining area,
which in some places was found to have been eroded to a depth of 25 and even 3o feet".
A reference to the plans reproduced in Plates X XI X-X X X I I I, X X X V, as well as to the photographs (Figs. 39, 4o, 41, 44), will illustrate this. The only apparent exception was that of
the northernmost ruin (N. v1I1) and some structures closely adjoining it, which were found deeply embedded in dunes, and where consequently the effects of former erosion were hidden from view. But even in this vicinity the loess soil, wherever bare, showed proofs of having been greatly lowered. The photograph, Fig. 46, taken of a completely-eroded ruin near N. xi will help to illustrate the final stage of destruction to which all ancient buildings are doomed in the zone lying further south and less protected by drift-sand. With such striking archaeological evidence before my eyes, I could not for a moment remain in doubt as to the true origin and significance of the broad ravines, about 15 to 3o feet deep, stretching across many places where the excavating force of the winds could freely assert itself in the bare loess.