` treasure ', and in the course of five or six days' fruitless wanderings he had come upon ` ten or twenty old houses ' to the east or north-east of the Stûpa. Vague as the treasure-seeker's topography seemed to be, the Surveyor was to endeavour to locate the ruins and then to push due north into the desert as far as he could in a day without exhausting the camels. I myself with the rest of my column tramped on under Ibrahim's guidance over the high sands to the north-north-west, in quest of the ruined dwellings upon which my old guide had lighted in the course of a prolonged but avowedly profitless ` treasure-seeking ' expedition in the winter of 1904.
Soon after passing the ruin N. xi' we left behind the last few living Toghraks and had to make our way over steep dunes. These rose closely packed amidst a maze of sand-cones, crowned on their top with still living tamarisk scrub. Progress was necessarily slow here, and it took fully an hour to cover the two miles which brought us first to a small ` Tati ' with pottery débris and then to the nearest of Ibrahim's promised ruins. It proved that of a timber-and-plaster dwelling (N. xxii), half-covered by a big dune, just beyond the line to which living tamarisk growth extended. A tank
surrounded by fallen poplars lay close by. Marching on over absolutely bare dunes for another mile and a half, I passed one after another of the ancient houses reported. They lay in a line along what manifestly had been the extreme north-western extension of a canal once fed by the Niya River. In
full accord with the indications which Ibrahim had given when first talking to me about these ruins
at Niya, their line proved to be situated within two miles to the west and north-west of the northernmost group of ruins, N. viii, we had been able to trace in 1901 (cf. the site plan in Plate 7). But the
high swelling ridges of sand intervening had then kept them from view. For my camp I selected
a patch of open eroded ground conveniently near the northern end of the ruins I had come to explore, and lost no time in starting my day's work at the farthest ruined structure we could trace.
Patches of pottery-strewn ground cropped up for some distance further north. But though the view over the bare expanse of dunes in that direction was open and reached far, we could sight no more structural remains, such as posts rising above the sand would usually indicate from a distance.
The ruin first excavated, and numbered N. mil in continuation of the work done in 1901 (see the site plan, Plate 7), occupied the narrow northern end of what, owing to the depression produced
around by wind erosion, looked like high ground. The latter extended in continuation of the line of
a small irrigation canal still marked by fallen rows of dead poplars. On the north and east sides the ground had been scooped out to a depth of fully fifteen feet below the original level ; to the west
the depression produced was about ten feet. The ruin was that of a comparatively small dwelling, measuring about fifty feet square in its extant portions, as seen in the detailed plan (Plate Io). The sand which had accumulated in it rose nowhere more than four feet above the floor level. But this had sufficed to protect not merely a number of interesting small objects but also enough of the walls to show the essential features of construction.
In the main rooms the walls were built of timber and plaster exactly after the fashion which my excavations of 1901 had proved to prevail in all more substantial houses of the site.1 Hence
no detailed description is needed here. The matting which served as core or backing for the plaster of the walls, was fixed to an elaborate and carefully fitted framework of posts in ` Terek ' wood, and generally consisted of diagonally woven tamarisk strips. This is clearly seen in the photograph (Fig. 47) which shows the main rooms to the south after clearing. Only in the room iii which seems to have served as a kitchen, this diagonal matting was replaced by horizontally fixed layers of reeds, a system which I had met with also elsewhere at this site, though more rarely. The small outside apartments on the west side had walls of much rougher çonstruction, the core consisting here