1120 THE NORTH-WEST FRONTIER 6F KAN-SU [Chap. XXVII
spur which flanks the debouchure of the stream on the right bank. It ends where the cliffs on this side too become impracticable. As I followed the little valley, cultivable all along its bottom, upwards for about a mile I found it closed, where it narrows to about 200 yards across, by a battlemented wall which resembles in construction the main wall of the Chia-yü kuan barrier ',4 and is in any case of no great age ; for the many loopholes piercing it, a true ` chiusa ', were manifestly intended for fire-arms. This wall, which was in good preservation except for the middle portion exposed to damage from the stream, faced with its parapet to the south-west, i. e. up the valley. Its purpose could not be doubtful. The precipitous rock walls, rising on either side to heights of 300 to 400 feet with unscalable faces, and the rugged nature of the hill range in general completely protected this closing wall in the Hao-shan-k`ou gorge from being turned. Thus that portion, too, of Huang-ts`ao-ying cultivation which lies outside the main wall was made safe from raids, and the guarding of the latter itself greatly facilitated.
But a very puzzling fact faced me when, only some hundred yards higher up in the gorge, I found the remains of another and distinctly older wall on either side of the low ground watered by the stream. It, too, had undoubtedly been meant to close the passage through the valley ; but strangely enough its parapet, as seen in Fig. 253 cm the left, faced to the north-east, i. e. down the valley. This second wall, though massively constructed, had suffered far more decay than the first. It was built of stamped clay with a thickness of II feet at its base, where it rested on a solid stone foundation covered with a layer of rushes ; its height, where greatest, was still I I feet. The parapet, about 2 feet thick and 3 feet high, had no loopholes. Had the gorge been situated somewhere on the Indian North-West Frontier, or on ground similarly affected by internal feuds in the past, it might have been easy to account for two fortified positions facing each other at such close range. But here, on the North-West Frontier of China, such an explanation would find no support in anything that we know of its history. Hence the close proximity of such inverse lines of defence, and both by their very nature manifestly of Chinese origin, was bound to appear from the first in the light of a problem.
Before indicating what I now consider its likely solution, it will be advisable to complete the account of my Chia-yü kuan survey with reference to another puzzling feature. I mean the line of wall which on my first approach to Chia-yii kuan I had already noticed striking off at right angles from the modern barrier wall and stretching far away to the north-east. It was at the very first of the watch-towers guarding the wall north of the Gate, and only about a mile and a half from the latter,' that I found the junction of this distinct line. Its remains were far more decayed, as, in fact, I had noticed from a distance, and closer inspection at once produced evidence of a different, and probably earlier, date of construction. This wall, also built of stamped clay like the Chia-yü kuan barrier, shows at its base a thickness of 82-9 feet only as against the I I feet of the latter. Its height, exclusive of the parapet, is about 8 feet. Of the badly broken parapet only about a foot remained here. The examination of the clay layers also disclosed a striking difference. Whereas in the wall forming the Chia-yii kuan ` chiusa ' they are 4 to 5 inches in thickness, those in the older wall touching it here at right angles have a height of Io to II inches.
The line of this wall, running south-west to north-east, is, as subsequent inspection showed, broken by many great gaps. But though the aggregate length of these in the section here examined is probably quite as great as that of the parts of the wall still standing, the alignment of the wall