off in an almost southerly direction. This ruin, too, has suffered much ; but the outer wall to the south still stands to about io feet in height.
The ruined station R.R. xiv, half a mile to the SW., showed a ground plan (P1. 59) somewhat larger than the rest and differing in the internal arrangement. The interior has suffered through a drainage channel which has formed in the centre. The outside dimensions of the enclosing walls are about 77 feet by 54. They are strengthened by small towers at the four corners, all badly decayed, and a fifth projecting on the eastern face. There seems in addition to have been some kind of projection on the south, perhaps intended to protect the entrance on that side. A second entrance with a pointed arch is recognizable near the SE. corner and probably gave access to a staircase. Similar vaulting survives over part of the central room on the north, having a span of close on 15 feet. The masonry exposed on the east and south faces shows bricks about 24"X 13"X4", set on edge with the longer side upwards. They contain much straw. The eastern face still rises to a height of over 15 feet. Shepherds visiting the neighbourhood during the few weeks of spring vegetation have used the shelter of the ruin, as proved by the droppings of their flock, and this accounts for the glazed potsherds of recent appearance that I noted there. A small structure, about - mile to the ESE., built of stamped clay and comprising three rooms, appeared also to be of late origin ; it may date from the last period when land near the debouchure of the Rûd-i-biyabân is known to have been cultivated.
The ruin R.R. xx, less than i miles due south, is that of a post of the smaller size, measuring 45 feet by 42 outside and showing the typical arrangement of vaulted chambers inside, as seen in the sketch-plan, Pl. 59. It is situated above the right bank of a wide scrub-filled depression. This comes from the debouchure of the Rûd-i-biyaban and contains the canal which until a century ago carried water to Machi and Hauzdar, as already noticed at R.R. iv, v. The room, i, to the left of the entrance was less heavily filled with debris from fallen walls than the rest, and this I was able to get cleared during the two days' halt that we made by the,side of another old canal-bed, not far from the Muhammadan sepulchral domes which the map marks as Yak-gumbaz. The clearing of room i brought to light a sitting platform running round three sides (see Fig. 493 and detailed plan, Pl. 59), and underneath it a hypocaustic passage obviously intended for the warming of this room. At the openings, A, B, of this towards the entrance the plaster was found calcined. The whole strikingly recalled the k`ang arrangement in Chinese houses. The floor of the room lined with bricks, i6"x 15", was raised 2 feet above that of the entrance passage, and the sitting platform another I' io". The little recess in the raised masonry bench above the opening A may have been intended for the heating of vessels containing water or some similar purpose. Some lumps of clay, each about 22" high and of pyramidal shape, were found inside this hypocaust opening and were, perhaps, meant to be fired and then to serve as loom-weights.
On the other side of the wide bed and about 400 yards to the WSW. of R.R. xx we discovered the badly decayed remains of another post of the same type. The southern face, which was the easiest to trace, measured about 4o feet and showed bricks of the regular size, 24" x z 2"x 4". The structure had obviously suffered through moisture. The line of posts that we had succeeded in following so far led from R.R. xi' onwards more or less due south, and naturally it was in that direction that I searched for its continuation. But here at R.R. xxi, where we had struck the northern limit of the still recognizable delta of the Rûd-i-biyaban, it seemed to come to an end. So far one ruin sighted from the other had afforded easy guidance. The careful reconnaissance I made from this last point to the south took me in succession across four old river-beds, all branching from near the point where our camp stood, close to the pillar-marked Perso-Afghan boundary. But though the low gravel-covered plateaus separating these dry river branches were perfectly