A first inspection sufficed to show me that this terrace, rising about 35 feet above the surrounding, but little eroded, ground, was a Mesa differing distinctly in size, character, and direction from
the familiar Yârdangs. As the plan, P1. 12, shows, it measured about 56 yards in length from north-east to south-west, with a maximum width at its foot of about 32 yards. Just as in the case of Yârdangs the end facing the prevailing wind-direction rose very steeply, while the opposite end tailed away with a gentle slope. But, quite apart from the height of the terrace and its isolation on comparatively flat ground, the difference of its longitudinal bearing from the regular ENE.—
WSW. direction of the Yârdangs was sufficient to show that its origin must have been geologically different from that of the latter. It was in fact the westernmost of those Mesas of which we were
to meet thereafter so many around the northern and eastern shores of the ancient Lop Sea bed. From the first my attention was attracted by the fact that the terrace was composed of salt-impregnated clay of greater consistency than the ground around it, and further that wind-erosion, instead of undercutting at its foot, as regularly happens with Yârdangs, was here mainly at work along the top edges of the Mesa.2 The total absence of any structural remains on the top clearly showed that the terrace could not owe its height to the protection which some massive building might have afforded, but must already in ancient times have risen high above the level of the neighbouring ground.
I could not doubt from the first that it was the elevation of this Mesa terrace, a marked feature even at an early historical period when moisture and vegetation still protected the adjoining soil, that had caused it to be selected as a burial-ground. Some grave-pits, half exposed on the top edges of the mound, could at once be made out.3 But only few sepulchral objects were found on the slopes below, and most of these immediately beneath the still traceable graves. Had the top of the terrace represented only the surviving portion of what was once a cemetery situated on the general level of the area, we might reasonably have expected to find the slopes of the terrace as well as the eroded ground near it strewn with hard debris, such as metal objects, human bones, planks of coffins, &c., just as similar debris is always found round the remnants of wind-eroded dwellings. Yet the immediate vicinity of the terrace showed none of this ` Tati ' appearance.
The reconnaissance reports had led me to expect here merely such relics as wind-erosion might have spared. I was therefore all the more delighted to find on the first rapid inspection that the summit of the Mesa retained a number of graves, apart from those on its edges, quite untouched by that destructive agent. They were all marked by rough tamarisk posts fixed, as subsequent examination showed, around the edges of the graves or pits, while the latter themselves were covered with layers of reeds almost entirely exposed. Until the heavily laden men arrived I had to be content with examining such relics as the partially eroded graves on the edge of the terrace disclosed. Here my eye was caught at once, amidst human bones and broken boards from decayed coffins, by some rags of beautifully woven silk fabrics. Their brilliant colours were excellently preserved, even where the crumbling away of the steep slope of clay had left them lying on the surface, exposed to sun and wind. Their survival under such conditions seemed a particularly encouraging augury.
My hope of finding other fabrics in graves that had escaped erosion was naturally still further stimulated when in one of these figured silk fragments (L.C. 031. b, Pl. XXXV) I recognized a
Serindia, i., and Figs. 146, 149, 153, 156, &c., here, show wind-erosion at work on terraces that have been formed as a result of protection by ruined structures, will help to illustrate the difference above indicated.
3 See i, ii, vi, vii, in the plan, Pl. 12.