Sec. vii] THE ANCIENT CASTRUM L.E. AND REMAINS ON MESA L.F. 261
work once revetting the south gate, while the other, only a kind of postern, masked by a structure in timber and wattle, showed traces of having been lined with masonry of sun-dried bricks.
The chief interest of the ruined fort proved to lie in the remarkably solid construction of its walls. They were built, as the photographs (Figs. 161, 163) show, of layers of closely tied fascines, made up of tamarisk twigs and about one foot thick, alternating with strata of stamped clay, five to six inches thick. This, owing to salt-impregnation, had consolidated into cement-like consistency. The several layers were provided with a revetment of longitudinally fixed fascines. These survived in places where the position of the wall faces or else accumulation of drift-sand had afforded them some protection, but elsewhere had been loosened and carried off by the erosive action of the wind. The structural method employed was exactly that observed by me along the different sections of the Tun-huang Limes and fully described in Serindia.5 The proportionate thickness of the alternating layers of tamarisk fascines and clay was the same as I had noted in the well-preserved stretch of the Limes wall near the watch-station T. xxxv, north-east of Tun-huang, where the fascines had similarly been made from the tamarisk growth of the adjoining ground.6 Both structures conveyed the same impression of solid regularity and neatness ; and this impression helped to convince me from the first that there can have been no great interval between the construction of L.E. and the time when the patient builders of Han Wu-ti's ` Great Wall ' were at work in the desert of Tun-huang.
The thickness of the walls appeared to have been about twelve feet where they rested on the ground. Their inside face still rose in places almost vertically, as seen in Fig. 163 ; but outside, the grinding action of erosion had caused the originally steep wall face to recede in what had come to look almost like a succession of steps. Thus the width of the wall at what is now the top was reduced to five or six feet. Towards the middle of the east face the wall still rose to a height of over ten feet, with seven double layers of fascines and clay intact. Elsewhere it had been worn down a good deal more. That its height had been originally considerably greater was evident from a portion of the west wall. Undercutting of the ground through erosion had caused it to subside bodily towards the interior, but in spite of the consequent distortion some ten successive double layers could still be counted here.' Near the south-eastern corner and again towards the middle of the west face the foundation layers of the wall showed a thickness of about eighteen feet, probably indicating the place where stairs had led to the top. That the top once carried a parapet is probable ; but it must have been particularly exposed to the corroding force of the sand driven across the walls, and no remains of it could be traced.
The walls of this castrum, built as carefully as those of the Tun-huang Limes and a good deal thicker, were strong enough to withstand any attack that a local rising or raiding bands of Huns could have directed against it. The castrum could thus well serve its obvious purpose as a safe resting-place for any missions, military detachments, convoys, or trade caravans reaching the eastern edge of the once habitable area of Lou-lan from the side of China or preparing for the trying desert journey across the waterless wastes of gravel, salt, and sand in the opposite direction. The materials and methods of construction used for its defences were those best adapted to local conditions. Even the relentless, if slow-grinding, force of wind-erosion, ever at work in this desolate region, had failed to overcome them completely. But if the winds and the driven sands could not efface this ancient circumvallation, they seemed to have done their work of destruction all the more effectively upon the interior and whatever once stood there. The whole of the enclosed area
s Cf. Serindia, ii. pp. 568 sq., 605 sq., 736. similar effect produced on a small section of the Limes wall
6 See ibid., ii. p. 605, Fig. 157. near T. xiv. a, as seen in Serindia, ii. Fig. 189.
7 Compare the contorted wall line in Fig. 162 with the