In describing these defences ° it will be convenient first to follow those which form the outer enceinte. Starting from the height at the eastern end of the main ridge, i, we find the edges of its rocky plateau-like top bounded by a brick wall in close proximity to the cliffs with which the ridge falls off to the plains both on the east and north. Owing to the strength of the natural defence provided by precipitous rock slopes, over 200 feet in height at this end, the wall here was only 3 to 32 feet thick and has suffered much decay. It is built of sun-dried bricks of an average size, as in most portions of the outer enceinte, of 14 to 15 inches by io to i i, with a thickness of 3-1 to 4 inches. The wall and the towers both round and square, with which it was strengthened at irregular intervals, had their brickwork raised on low stone foundations. The loopholes, of which nowhere was more than one row here traceable in the surviving brick masonry, were usually placed close above the stone foundations. Their height varied, being up to 3' 3" inside and decreasing to about 2' 3" outside. Their width narrowed towards the outside to 7 or 8 inches. A peculiar feature traceable along the eastern and northern portions of this circumvallation was a second wall, running parallel to the outer one and at a distance of 6 feet from it. It had a thickness of only II-2 feet and had decayed badly. No loopholes could be found in it. An exactly corresponding arrangement had been observed at the main circumvallation of Zamr-i-atish-parast.5 Its exact purpose remains doubtful.
On the northern face the general height of the main ridge gradually decreases towards the west. But the cliffs towards the plateau outside are very steep and about the middle of this face still rise to a height of about 6o feet. Where the elevation of the ridge lessens the place of the comparatively weak wall with towers crowning the rock wall is taken by a remarkably solid rampart built of sun-dried bricks outside and apparently of stamped clay within. Erosion has scored it with numerous fissures descending towards the interior. But the outer face still presents a strikingly solid front except at the two breaks marked ii in the sketch-plan (Pl. 49). This rampart is about 16 feet across near ii, and gradually increases to a thickness of about 33 feet towards the NW. corner, guarded by a massive circular tower, iii (Fig. 41 i). Its height above the wall of live rock outside is about 25 feet on the average. The parapet which, no doubt, it carried is no longer traceable ; nor could the position of any towers be made out.
At the NW. corner, iii, the rampart turns to the SSW. and crosses the gap between the two ridges (Fig. 41 i). Its thickness here is about 22 feet at the top except between the two square bastions, iv, shown in the sketch-plan (Pl. 49), where it is reduced to about 12 feet. The bastions built on massive stone foundations project about 20 feet beyond the curtains on either side and may possibly have been intended to protect a gate or postern between them ; decay of the wall, however, no longer permits one to distinguish this. Close examination at this point showed that the bastions and the curtains both on the inside and on the outside were faced with solid masonry in bricks, I 6" x 9" x 4", while the interior space was filled with layers of stamped clay divided at irregular intervals of 6 to io inches by thin strata of brushwood. The same method of construction, re-
sembling that described in the forts of Kansir and Kiz-kurghan,e was noticed also elsewhere at Qala-i-Qa`ga. The use of this method, unknown locally at the present time but widely applied
in early structures of Chinese Turkestan, appears to me a clear indication that the fortress was of considerable antiquity. In a few places it was also possible to make out layers of rubble near the top of the rampart covered with bundles of brushwood 8 to io inches thick.
The curtain closing the gap ascends south to the foot of the precipitous rocks which form the
unsupported by critical evidence. 5 See above, ii. p. 867.
6 Cf. Serindia, i. pp. 69, 75.