(cf. Fig. 68). But owing to erosion the top of the cone now stood fully twenty-one feet above the immediately adjoining ground.
I must note as a specially curious feature the frequent appearance of thick layers of sheep-dung at several of these ruins within what must have been living quarters. The fact seemed puzzling at first until, in the light of my subsequent observations at Miran, the ` Lou-lan ' Site, and elsewhere, it became evident that these ruins, at some period subsequent to their abandonment, had been used to shelter flocks grazing in the jungle, which had grown up around them at a time when the summer flocks still reached this nearest portion of the old site. There were, however, no archaeological indications to aid us in determining that period, if only approximately, and calculations based on the conjectured rate of growth of the sand-cones, the age of the neighbouring wild poplars and the like, might be very misleading, seeing how different the determining factors may be, even in areas closely adjoining. But so much is certain, that the abandoned dwellings could have invited use by shepherds only as long as their ruins were in some state of preservation. It is also noteworthy that I found traces of such use only in the southernmost portion of the site. The reason for this distinction is made clear by what I shall have presently to record about the ancient terminal course of the Niya River.
On October 3o there remained only the clearing of the modest dwelling to the south, N. xLI (for plan, see Plate 18), from which, on the first day of my return to the site, I had extracted three Kharosthi tablets.1 Its scanty remains, badly eroded and showing also traces of recent burrowing, refused to yield more than another tablet, much decayed, and some miscellaneous small implements. Among these it may suffice to mention a small stick, N. XLI. oo8, likely to have been used as a pen, and a longer one, N. xLI.005, which, judging from its rounded and charred ends, had probably served as a fire-drill. But the surroundings of this ruin, as a careful inspection soon showed, were to reveal features of far greater interest. The panoramic view, taken from a sandy ridge to the east and reproduced in Fig. 75, will help to explain the details as well as to convey the general impression of the ground. All round the ruined dwelling there could still be traced lines of fences bordered by rows of poplars and enclosing small arbours and orchards. Only some seventy yards to the west there still stood a rectangle of dead mulberry-trees, raising their trunks to a height of ten feet or more, which had once cast their shade over a tank still clearly marked by a depression.
The stream from which the canal once feeding this tank must have taken off was not far to seek ; for behind the nearest ridge of sand westwards, the very one from which the photographic panorama was taken, there still lay the remains of a foot-bridge, traceable for about ninety feet and stretching
across an unmistakable ancient river-bed, as clearly seen in Fig. 75. Of the trestles which
had carried the bridge two still stood upright, near what must have been the eastern head of the bridge, one of the posts rising to a height of over ten feet. The extant portion of the bridge was formed near its eastern end by one large flattened trunk of poplar (Poj5ulus alba), one and a half feet broad and thirty-nine feet long, as far as it could be cleared of the sand which covered what was once the right river bank. This part of the footway was continued by three narrow trunks, about forty feet long, lying side by side and once, no doubt, joined together. Beyond there could be traced detached fragments of trunks for another ten feet. They lay on the eroded slope which leads down from the west to the depression marking the lowest part of the ancient river-bed. The bottom of the bed lay fifteen feet below the level indicated by the extant eastern end of the bridge.
At a distance of sixty-eight feet from the westernmost portion of the footway now extant, the slope terminated in a steeply eroded bank of hard loess. This, clearly shown by the photograph along with several similar and parallel banks above it, has undoubtedly been ` washed out ' and
1 Cf. above, p. 2 x3.