Sec. vii] THE ANCIENT CASTRUM L.E. AND REMAINS ON MESA L.F. 263
The Mesa L.F. on close approach proved an imposing ridge of reddish clay, rising, as Fig. 164 shows, very steeply to a height of over a hundred feet. Its bearing, as in the case of all Mesas in this area, was from north-east to south-west, thus differing slightly from that of the Yardangs. On all sides, except to the south-west, the several well-marked strata from which the ridge had been carved rose in almost vertical steps, making approach to the top very difficult, and near the northeastern end almost impossible. I soon realized that it was this easily defensible character of the ridge which accounted for its ancient occupation. To the south-west the Mesa tailed away, as usual, with a much easier slope, as seen in the plan, Pl. 12. Ascending this, I first reached a lower terrace and on this a grave, marked by remains of enclosing planks, which was subsequently cleared and found empty. On a level about fifteen feet higher we found the central portion of the ridge (Fig. 170) occupied by a small number of graves, of which some appeared to have been partially bared by wind-erosion, while others were evidently still intact.
This little cemetery was separated from the north-eastern extremity of the Mesa by a trench, about six feet wide, cut across where the flat top is flanked on either side by small gullies and has been reduced by erosion to a width of about fifty feet. Owing to the debris accumulated in it the original depth of this cutting remained uncertain. Beyond it, the rest of the Mesa top was crowned by a wall, five or six feet thick, roughly built, as Fig. 166 shows, of slabs of hard clay, a material corresponding to the kisek in use at the present day. These slabs had obviously been brought from the dried-up edge of a marsh or lagoon or else possibly quarried from the clay strata of the Mesa itself. The small stronghold thus formed had the shape of an irregular oblong and measured about 200 feet in length and about 8o across its greatest width. A little knoll in the centre, rising about fifteen feet above the foundation level of the wall, had suggested to Afraz-gul a Stûpa, but was found to consist of natural clay. It may have been purposely spared when the ground enclosed was levelled, in order to serve as a look-out.
The wall protecting the little stronghold on the side of the gap through which the trench passed contained the gateway, about five feet wide. Its rough timber frame of Toghrak wood still stood upright, as seen in Figs. 166, 170. To the right of the entrance passage two rooms were built against the inside of the wall in the same rough fashion as the latter (see Pl. 12). Both were found full of refuse, mainly reed-straw mixed with the droppings of horses and cattle. By clearing this we recovered from room i a small wedge-shaped Kharosthi record on wood, L.F. i. 05 ; the Chinese document, L.F. i. o6, written on a wooden slip,sa as well as the fragment of a Chinese paper record, L.F. ii. 07. Among other small finds described in the list below may be mentioned the gold finger-ring, L.F. i. 02 (Pl. XXIV), with a small ruby or carnelian held in a circular bezel ; three wooden writing-sticks, L.F. ii. 02. a—c (Pl. XXIX) ; the wooden fire-stick (female), L.F. ii. o6 (Pl. XXIX), resembling those found before at the Niya Site, L.A. station, and elsewhere ; a decorated wooden pin, L.F. ii. 04 (Pl. XXIV), of the type subsequently recovered from the graves ; and the leg of a kid which my men took for that of a deer.
The room iii to the left of the entrance passage still retained some of its massive roof beams in position, as seen in Figs. 165, 170. It contained nothing but an uninscribed Chinese coin of the Wu-chu type and a small heap of oats and oat-straw. This is a find of some antiquarian interest as it shows that ground with some kind of cultivation, perhaps as casual and intermittent as that still practised in the belts of riverine vegetation along the lower Tarim, was not very distant when this outlying post of Lou-lan territory was still occupied. The post lay moreover in the same north-easterly direction which, as the position of L.C. and L.E. showed, the ancient Chinese route from the Lou-lan station obviously followed. This seemed a clear indication that the Mesa top,
sa [M. Maspero notes in it the mention of a signal station called Ching-hu.]