Position of ruins N. III, N. iv.
Depth of sand in parts of N. III.
Protection from fallen garden trees.
Refuse layers over loess-banks.
830 THE ANCIENT SITE BEYOND THE NIYA RIVER [Chap. XI
promising ruins. The change already noticeable in the climatic conditions supplied a forcible reason to hurry on work. The minimum temperature of January 31 was still —3° Fahr., but at noon the thermometer rose to 45° Fahr. in the shade, and the warmth of the days was steadily increasing. It was clear that I could not depend for many weeks longer on the rigour of the desert winter to facilitate our water-supply in the form of ice, while even for other sites, where the difficulties raised by this question of water might be less serious, the available season was distinctly limited by the approach of spring with its sand-storms.
For my next excavations I selected the remains of the two large dwelling-houses which I had passed on the evening of my first arrival, about two miles due south of the Stûpa (see
plan in Plate XXVII), Both ruins were prominent from their position on isolated high banks of loess, due to the erosion of the surrounding ground, and from the rows of large fallen poplars marking the ancient gardens and avenues which once surrounded them (see Figs. 40, 41). The number and size of the apartments, together with the careful construction of the timber framework, showed that they were the remains of substantial residences. In the case of the ruin to the east (N. iii), the pieces of an elegantly carved wooden chair which I had previously noticed lying on the surface were a specially promising indication. As this building was also buried far deeper in sand than the other, and thus likely to prove better preserved, its excavation was commenced first. The work of clearing it occupied my band of labourers fully four days, though by this time they had been reinforced by the last able-bodied men from Itnâm Ja`far Sadiq, summoned in haste as soon as I realized the great extent and importance of the site.
The difficulty of the work was mainly due to the depth of the sand filling the apartments
that lay on the south side of the ruin and were thus, as the plan (Plate XXX) shows, well protected from the erosion attacking the edges. The large central room which, judging from its great size, 38 by 26 feet, probably served as a kind of reception hall, held no less than 9 feet of sand. The four massive beams of poplar wood, which stretched across the whole length of the hall and once supported the roof, were found in their original position resting on the sand that had accumulated beneath them, as seen in the centre of Fig. 43, which shows this part of the building before excavation. The smaller rooms adjoining the central hall westwards were also filled with deep sand, and owed to it the relatively good preservation of the timber framework of their walls, as seen in Fig. 42. The apartments lying to the north had lost most of their cover of sand by erosion, and their walls consequently were found standing only a few feet above the foundations.
That the building had once extended further in this direction was proved by the plentiful
débris of timber strewing the north slopes. That erosion had not succeeded in attacking the ruin from its southern side also was due mainly to the protective belt formed here by the large trunks of ancient poplars, which were found lying in groups and rows as they had fallen long centuries ago. A number of these trees, even in their shrivelled and splintered condition, still measured well over 5o ft. in length. The strength left in their wood was attested by the positions some of them were found occupying, raised high into the air where they lay heaped up over other trees, or else jutting out far beyond the eroded slope, as seen in Fig. 41. The weight. of these rows of fallen trees had effectively protected the ground beneath, which accounts for the fact that the plateau-like raised area occupied by N. Hi was found to comprise not only the actual site of the ruined building, but also a considerable portion of the orchard and fenced courtyard once surrounding it (see Plate XXX).
Where the edges of this plateau marking the ancient ground-level had been cut away sharply and steep banks formed in the loess as towards the north-west (seen to the right in