| || || || || |
714 THE ANCIENT CEMETERIES OF ASTANA [Chap. XIX
120 and 200 yards wide on the average, adjoin the long sides of the plateau and unite where its south-eastern end tapers into a prow-like edge of cliffs. At the opposite end the plateau is cut off from its continuation to the north-west by a natural fosse, about 5o to 6o yards wide, formed by a small branch which the more easterly of the two ` Yârs ' has thrown out towards the other. Small streams flow in both Yârs. These are ordinarily fed by springs, in which subterranean drainage from the valleys of Yoghan-terek and Shaftalluk comes again to the surface ;6 but they occasionally also receive water from heavy floods descending from the mountains. The bottom of these deep Yârs lies, near Yâr-khoto, about a hundred feet, on the average, below the level of the plateau and of the cultivated area to the east ; they owe their formation, no doubt, to the greater volume and eroding force of such floods at a former period. The alluvial clay through which these streams have carved their way falls off along the edges of the plateau in vertical cliffs, unscaleable in most places and from 95 to II() feet high where I measured them. These cliffs provided the town with natural walls of almost impregnable strength, easily defended from above and practically proof against breaching or mining, though, of course, affording no protection against projectiles discharged from the ground on the other side of the 'Vars.
Approach to the plateau occupied by the town lies only at two points. Near its south-eastern end a winding track, evidently going back to ancient times and artificially improved, gives access to a spot where the plateau top slopes down to about 75 or 8o feet above the bottom of the adjoining Yâr. The only other track by which the plateau can be gained now lies in a steep little ravine ascending near the middle of the north-eastern edge. This ravine communicates above with a narrow and deeply cut gully forming a lane and lined by ruined houses, and is flanked lower down by massive remains which look like watch-towers cut from the clay. It hence appears probable that this approach likewise dates from ancient times.
The lower and broader half of the plateau lies towards the south-east and is closely occupied by the ruins of the town. The other half is a waste of bare clay but for scattered remains of ruined shrines, obviously Buddhist, and some small cemeteries of a type resembling those of Astâna. Among the former the most striking is a ruined structure, marked iv in the sketch-plan and seen in Fig. 279 of Serindia, which in its ground-plan and style must, when intact, have closely resembled the curious temple (P) at Idikut-shahri, described by Professor Grünwedel.7 Of the eighty small tower-like Stûpas ranged in four symmetrical groups round the central group of a large and high Stûpa with four others flanking it, all but two had been reduced to shapeless heaps of debris. These appeared to have been little dug into for manuring earth, and a systematic clearing of the debris might possibly still have revealed remains of votive deposits.
Two other conspicuous ruins found in this open area are evidently Vihâras, containing within a rectangular enclosure of high walls of stamped clay a cella once occupied by cult images as well as rooms grouped around a court. Both these two ruins had evidently been searched. But in the one marked I, of which a plan is shown in Pl. 26, and a photograph in Fig. 333, careful clearing of the passage i along the north-western side of the cella brought to light from among the debris thrown there several fragments of block-printed Chinese and Uigur texts, including two with glosses in Brâhmi script. Other manuscript and block-print remains in Chinese and Uigur were purchased from men engaged in digging for manuring earth at small ruins near the northern edge
of the town area.8 They all help to confirm the conclusion, to which other evidence points, that the
6 See Map No. 28. B. 2, 3. bead orn. ; then border of Byzantine acanthus-leaf scroll
7 Cf. Idikutschari, pp. 31 sq., Figs. 24-27. pattern in panels mitred at corners, and in centre narrow
8 From them was acquired also the following fragment : vertical panels containing similar acanthus scrolls running
Yâr. oz. Fr. of clay mould for rectang. tile. Grey, vertically, each panel framed by raised moulding (in cavetto
burnt hard. Within high outer edge comes, first, row of in mould). One corner of mould preserved. 3}" x 5".