date corresponding to 94 B.C. It mentions the local name, Ta-chien-tu, by which this westernmost wall section is designated also in documents found elsewhere. As among these one is dated in 96 B.C., we have here definite evidence that the extension of the Limes must by that time have been carried right through to the extreme end of the wall.
This conclusion was fully confirmed when we came to explore those watch-stations which, from the terminal point of the wall, stretched away to the south-west along the edge of the great marshy basin (Fig. 72) . The distances at which they were placed from each other clearly showed that they were meant to serve mainly as signalling posts along a portion of the line for which a natural flanking defence was provided by impassable marshes. High detached ridges of clay stretched out here like fingers from the gravel plateau into the wide marsh-filled basin. These offered ideal positions for conspicuous signalling stations, and the Chinese engineers did not fail to make the most of them. They had placed their towers here for more than twenty-four miles in an almost straight line as if they had fixed their position by sighting with a diopter.
There were interesting relics to recover at almost all these watch-towers. But at none were they so abundant as at the ruin of a small post which was placed about two miles behind the line and evidently served for a kind of sectional headquarters. The arrangement of the modest quarters could be made out very clearly, as the plan shows. The wooden door-posts at the entrance were still in position; the fire-place or oven, enclosed by a thin clay wall burnt red, still retained its ashes. Yet among the wooden documents recovered in the room, which was probably meant to
IIIF Color HighRes
