a straight line than my northernmost camp of 19ot. His testimony as to the complete absence of ancient structural remains in that direction was all the more important, because I could rely on his accuracy and keen eyesight. He had found all ground covered with dunes, generally ten to fifteen feet in height, and had noticed pottery débris only at one patch of eroded soil, about a mile or so beyond N. wit. Curiously enough he had found a small group of Toghraks still living beyond the high ridge of sand reached on his reconnaissance of 1901.
It was evident that either Islam Akhûn, like his namesake, the Khotan forger unmasked at the close of my first explorations, had indulged in romancing, or his topographical memory was defective. Yet his statement as to the ruins to the east of those previously explored by me could not be left untested merely because he had chosen a wrong bearing by mistake—or otherwise. So next morning Ibrahim ` the miller', with Saduk Äkhûn, the intelligent shepherd from the Mazar, and another enterprising companion, was sent out to reconnoitre independently eastwards. I had heard more than once of cases when men, shiftless and often enfeebled by opium, whom the hope of finding ` treasure ' tempted to old sites in the desert, had lost their lives by mistaking the direction which might have guided them back to safety after their scanty store of water was exhausted. Perhaps this had been the fate of the man whose skeleton I came upon, about halfway between N. xix and vut, stretched out on the sand. It was lying face downwards on a low dune, with no remains whatever near by. The preservation of most of the bones showed that it could not have been exposed to erosion for a protracted period. This discovery was not without its use to me ; for it first induced Ibrahim to mention to me the cemetery, or ` Mazar' he thought, which he had come upon in the course of his last wanderings at a point to the south-east of N. xtx. About this more further on.
October 22 was mainly occupied by the clearing of N. xx, a relatively large house, of which Plate 12 shows the ground-plan and Fig. 53 the appearance before excavation. The complex of its rooms extended over ninety feet from north to south, and sand filled them up to seven feet in height. Whereas the east side of the building had suffered from erosion the vicinity of a dune had protected the south-west portion and, in fact, prevented its complete excavation within the time I was able to afford. Room ii was filled completely with sand up to its roofing, which was six feet seven inches above the floor and made of short rafters with brushwood layers above, the whole resting on a rough longitudinal beam. The room, i, adjoining eastwards must have served for a kitchen. It is seen in the foreground of the photograph (Fig. 48). Here two big jars were found, one complete measuring three feet in diameter where widest, with a height of two feet eight inches and a mouth ten inches across. This jar had been cracked while in use and was found enclosed in a rope netting to secure it. Two branching posts in a recess of the same room served for trivets. The walls in this northern portion of the house were constructed either of timber and plaster with diagonal tamarisk matting or of vertical rush bundles covered with plaster.
The apartments in the southern part of the house (see photograph, Fig. 54) were more carefully built, and the central posts of the framework showed their roof to have had a height of about eight feet. An inset in the plan (Plate 12) shows the constructive details of the timber and wattle framework in the east wall of room iii. In the other walls of this room and of the small apartment, iv, the wattle consisted of horizontal reed bundles fixed to the posts of the framework. The wall plaster of these two rooms was of unusual hardness and retained a pink colour-wash. Near the north door of iii a graffito showed the outlines of a hand with the five fingers spread out. Two low doors in carefully fitted wooden frames, with jambs slightly slanting, opened from iii into rooms which could not be cleared owing to the height of the dune overlying them. One of these doors, with its lintel five feet above the floor, and another to the south were approached across the sitting platform of the room, an unusual arrangement. In the small apartment, iv, a small slip-shaped