advancing again into ground long abandoned to the desert jungle, I may refer to my Personal Narrative.3 Next morning after taking some twenty labourers from Malak-alagan we moved a little south of east, crossed the stream of Domoko-yar.near where the canal of Malak-alagan takes
off, and after marching for about three and a half miles through a maze of conical sand-hills overgrown by tamarisk scrub arrived at the Khadalik Site. The appearance of pottery débris forewarned me within the last half-mile or so.
The site at first glance seemed to hold out little archaeological promise. There was a little plain about 400 yards from east to west and less than that across, fringed all around with tamarisk-
covered sand-cones (see Fig. 39). The ground, in parts wind-eroded and elsewhere overrun by low dunes, showed no indications whatever of structural remains except one or two quite low mounds near the centre, with scattered fragments of stucco and timber on the surface. Considering how near the site was to an area still occupied and how exposed its remains must have been from early times to constant exploitation, I could not feel surprised by the absence of those gaunt remnants of timber-built houses and of ancient orchards which had at once struck the eye at sites like Dandanoilik and Niya, far out in the desert. But the appearance of the extensive mound which Mullah Khwaja pointed out as the provenance of the manuscripts suggested such multifarious burrowings that it seemed as if none of its layers could have escaped undisturbed.
After a rapid preliminary survey of the whole site, including a small detached débris-area about half a mile to the east, I set the men to work where a shallow eroded depression approached the
south face of the mound. Small broken pieces of stucco from a frescoed wall, evidently belonging to a Buddhist shrine, were discovered almost immediately on the sand-covered slope, and with them little fragments of paper manuscripts written in bold Brahmi script of the Central-Asian Upright Gupta type. Within half an hour the first important ' Khat ' was brought to light from a depth of
about two and a half feet, in the shape of three almost complete leaves of paper, fifteen inches long by four and a half in height, which I could recognize as belonging to a Pbthi of some Buddhist
Sanskrit text. More finds of the same kind, but in far greater number, followed in rapid succession.
These consisted of detached leaves, sometimes even of small packets from the same Pbthi, mostly broken, or of mere torn fragments. All the manuscript remains were in Brahmi script, but plainly
belonged to a number of different texts, either in Sanskrit or that ' unknown ' language of ancient
Khotan for which recent researches have established Iranian origin. Among the latter finds was a convolute, containing the major portion of ten leaves each made up of two sheets of thin yellowish
paper which were pasted back to back and bore writing on one side only, after the fashion of Chinese printed books. With them turned up, though far more rarely, oblong wooden tablets of small size, inscribed in the same non-Sanskritic language. The total number of individual ' finds' of these kinds exceeded a hundred by the evening.
Fragments of painted stucco, evidently from frescoed walls, of appliqué relievos and of painted panels, were also discovered in plenty, all closely recalling in style and technique the remains found
among the Buddhist shrines of Dandan-oilik in 1900. The assurance thus conveyed as to the
character and date of the ruin was doubly welcome at the start ; for vainly did I watch that first day for the appearance of any structural remains in situ. The excavation was, indeed, carried
down through the layers of sand and plaster débris to the original floor of the building ; but it still left me without guidance as to its shape or extent. One thing, however, was clear, that the temple had been a large one, and that the burrowings of Mullah Khwaja and his associates had by no means exhausted the débris heaps left behind by the destructive operations of a far earlier time. However
s Cf. Desert Cathay, i. p. 238.