Sec. i] THE DESERT MARCH TO THE SITE 241
a timber framework with thick layers of plaster over a kind of wattle, were exposed to view, broken down to within a few feet from the ground. Elsewhere in places, covered by low dunes not exceeding 6—zo feet in height, the walls could still be made out by the rows of wooden posts rising in splintered stumps above the sand. Pottery débris was seen strewing many small patches of ground where the dunes had left the loess soil bare in the vicinity of the ruins. All structural remains left exposed showed signs of having been explored by `treasure-seekers', and the marks of the damage caused by their operations was only too evident. Yet even thus the ruins, on a first hurried inspection, furnished unmistakable evidence of their character and approximate date. The remains of frescoes visible on the much-injured walls of one of the exposed structures indicated that Buddhist places of worship were amidst the ruins. The style of these frescoes, clearly derived from Indian models, seemed to mark the last centuries preceding the introduction of Islam as the probable date when the settlement had been deserted. And already on the first day this conclusion received support from two Chinese copper coins of the K`ai-yüan period (713-741 A. D.), which were picked up under my eyes from the débris-strewn ground near the ruined buildings.
Old Turdi felt quite at home among these desolate surroundings, which he had visited so often since his boyhood. It was the fascinating vision of hidden treasure which had drawn him and his kinsfolk there again and again, however scanty the tangible reward had been of their trying wanderings. Yet the familiarity thus acquired with the silent relics of the past had developed in him an instinctive interest in all traces of the life that once reigned here. His shrewd observations on some of its more practical aspects, and his excellent memory for matter-of-fact details noticed in the course of his diggings made him distinctly useful at the site. It was evident from his communications that the conditions of the dunes were changing very slowly at this site. Consequently Turdi had no difficulty in recognizing the places where he and his companions had been at work during previous visits. Luckily their scanty resources had never allowed them to overcome the difficulty presented by the transport to this distant site of supplies sufficient for a prolonged stay, or to bring there working parties of more than a few men.
Hence the structures more deeply buried in the sand had a good chance of escaping unopened. It was important to select these in the first place for my excavations, and I felt grateful for Turdi's good memory and topographical instinct which enabled him readily to indicate their positions. Guided by this first rapid survey I chose for my camp a place from which the main ruins to be explored were all within easy reach. Practical considerations compelled me to make my choice carefully. For in order to keep the men at work as long as possible every day they had to be spared tiring tramps through the drift-sand. Still more important was it to keep to a spot where fuel could be readily obtained in quantities sufficient for our prospective long stay. Tamarisk-covered hillocks were rare, and subsequent observation showed that we had almost reached the northernmost limit of their occurrence in this region. I-Ience the dead trees of ancient orchards could alone be expected to furnish an adequate supply of fuel, and they were found to be very unequally distributed over the ruined area. As soon as the spot which seemed to answer these conditions best had been decided upon, and the baggage unloaded, I hastened to dispatch the camels on their three days' journey to the Keriya Darya, there to find the fodder they so badly needed. The donkeys, too, which had carried the men's food supplies, with the meagrest rations for themselves, were now sent back to Tawakka