284 ANCIENT SITES OF ENDERE [Chap. VII
preservation of the walls could not be taken as an indication of late date, was made clear on further reflection by the fact that a big sand-cone covered with living tamarisk on the top (Fig.82) adjoined the west wall, and rose fully twenty feet above it. Judging from what experience has taught me at other old sites by the southern edge of the Taklamakân, it would appear very difficult to account for the imposing height of thirty-eight feet which this tamarisk-cone has attained above the original ground-level, unless the ruined fort dates back to a period considerably earlier than Tang times.8 It must be borne in mind that this accumulation of drift sand could have started only since the surrounding ground had been abandoned to the desert. It is further obvious that the defensive purpose of the structure completely excludes the idea of its having been originally built in the immediate proximity of an already existing sand-cone ; for this was bound to impair its safety, as is forcibly demonstrated by the present condition when the summit of the cone has overtopped the fort wall by some twenty feet and, of course, completely commands the interior.
By the combined force of these indications I am led to conclude that this small fort belongs to the earliest remains of the whole site, and that its relatively good preservation must be ascribed to the special protection against wind erosion afforded by the closely set tamarisk-cones which grew up around it after this part of the site had ceased to be occupied. It is probable enough that physical conditions representing initial stages of those now observed, prevailed here already in' Tang times. That the post was erected to afford temporary protection against attacks rather than permanent shelter for a small garrison is suggested by its structural character. But whether it marks the point where the ancient high road of Han times passed the Endere settlement, as I felt tempted to surmise, must remain a mere matter of conjecture. When returning from my first visit to this ruin on November 9 to my camp at the Tang fort I followed throughout a dry depression which had all the characteristics of an ancient river-bed. As sheet No. 4o of the map shows, this approaches within a mile of the ancient post. It looks like a continuation of the wide bed which branches off from the present Endere river-course near Kök jigda-öghil, and though now dry supports a luxuriant vegetation of Toghraks, scrub, and reed-beds. "I'o judge from the remains of dead orchards I traced along the left bank of this old river-bed for about a mile south of the Tang fort, it is likely that it received water down to the time when the fort itself was occupied. But no structural remains could be traced on this ground.
When the excavation of the last dwelling-places within the Tang fort was completed on November i 2, I moved my camp back to Korgach, higher up the river. I utilized the march there for another visit to the fortified post in the south already described, and for the examination of the scanty remains which Mihman, the guide from the Endere Tarim, could show me in its vicinity. A ruin, reached after going about half a mile to the south-south-east, was found to be that of a dwelling hidden away between closely packed tamarisk-cones ten to twenty feet high. Portions of two walls, built of sun-dried bricks and clay, still survived to a height of about seven feet. One running from east to west had a length of about eight feet ; the other, standing at a distance of twenty-six feet and at right angles to the former, was a little longer. Between them the mud floors of rooms could be traced below the light cover of sand. The extant walls showed bricks of the same size as found in the ruins of the earlier settlement south of the Tang fort, viz. 19 by 13 inches, with a thickness of 32. They were laid, exactly as in the ancient ` Sipil ' near the great Stûpa and in the
g It has been shown above, p. 199, that in the case of two datable ruined shrines at Khâdalik and Farhâd-Begyailaki, the growth attained by the tamarisk-cones which now partially cover them has been sixteen and eighteen feet respectively since the end of the eighth century A. D.
At the Niya Site the maximum height observed in any sand-cone adjoining remains of dwellings was about twenty-five feet, as reckoned from the original ground-level ; cf. p. 238, above.