134 REMAINS OF THE KHOTAN OASIS [Chap. IV
of light dunes, I arrived at a ruined mound in which I speedily recognized the Stûpa of Arka-kuduk Tim, already visited in 1901 from the side of Hanguya.' Though our route crossed plentiful patches of bare loess soil clear of sand, the only vestiges of ancient occupation I noticed consisted of a small and completely decayed brick mound, five feet high, passed at a distance of about six miles from our camp at Târ-kuduk, and of a fairly large stretch of ` Tati ' which we traversed before reaching Arkakuduk ` Tim '. It formed part of that extensive débris area, generally designated as Hanguya Tati. From the description of my faithful old guide Turdi, now no more, I knew this to reach northwestwards as far as Ak-sipil, and a subsequent visit in March, 1908, showed it to spread quite as far on the east.2
I was aware that the ruin forming my goal was situated at some point of the narrow strip of desert which projects from the Hanguya Tati' to the south, dividing the cultivation of the Hanguya and Yurung-kash tracts, and which in April, tgot, I heard designated as Arkalik, the back lands'. But I well realized, as I made my way to the south-west across five miles or so of sterile dunes, how difficult it would have been to search for it without expert guidance. Here and there small patches of bare ground strewn with potsherds cropped out among the sand ridges. But on our left the trees of the outlying hamlets of Hanguya were already well in view before we struck the edge of a larger débris area only about a quarter of a mile to the north-east of the promised site.
Of the temple ruin, which Raze's description of a ` Bat-khâna' and the decorative relievo pieces brought away had led me to expect, no structural remains whatever were visible above ground. But plentiful fragments of the same type could be picked up near a small dune which my guides had taken the precaution to mark by a rag-topped staff.3 There were small relievo Buddhas seated within vesicas of lotus leaves, portions of wreaths, cloud scrolls, flame-bundles, and other relievo fragments closely resembling in character and style the appliqué stucco decoration of the halos which in 1901 I had discovered at Rawak round the colossal images. To find the friable stucco of the Rawak relievo work replaced in these fragments by what seemed terra-cotta was in itself very curious. But still more puzzling at first sight was their presence among the pottery débris of a regular ` Tati ' ; for of structural remains, such as temple walls, the level surface of sand showed no trace whatever. Like the potsherds around, these relics of relievos, once decorating the walls of an ancient temple, now rested on nothing but soft eroded loess.
In the absence of any surface indications such as I had to guide me at Dandân-oilik and other sites previously explored, the search for more substantial remains seemed here to demand systematic trial trenches. For such the dozen ` treasure-seekers ' who had tramped after us in the broiling sun were an inadequate force. But in order not to lose time I set them to dig near the north foot of the dune where the terra-cotta fragments lay most numerous. Chance for once favoured us ; for after burrowing down in the loose sand for only two feet, the remains of a fairly thick wall in reddish clay was struck, and from the débris layer near it, covering a plastered floor some two feet lower down, more relievo fragments of the same description emerged in rapid succession. These finds continued as far as the wall could be followed before dusk overtook us, and thus furnished assurance that the wall indeed belonged to a temple.
From the lowness of the wall laid bare and the entire disappearance of its outer facing it was clear that the temple had undergone almost total destruction in its structural features. Yet if larger
' See Ancient Khotan, i. p. 471, with Fig. 54.
2 See Ancient Khotan, i. p. 470 ; also below.
3 All relievo fragments, potsherds, beads, and other small remains which were picked up at the site in my presence bear the marks A. T. oor—ool 2 z. Pieces brought to Khotan