strength and good proportions, serves to illustrate the architectural skill which has survived in Sistan till very recent times. Two or three other residences visited farther to the south show an interesting ground plan with a cruciform central hall, as illustrated by the sketch-plan in Pl. 58.
About 2 miles to the SSW. of the fortified mansion rises a mound, about 20 feet high and measuring about 8o yards by 30. This old Mesa is now covered with Muhammadan graves, but the abundance of potsherds found on it conclusively proves that it had previously served through successive ages as a site for habitations. The range covered by the pottery fragments from this mound extends from ` chalcolithic ' painted or plain pottery, as shown by the specimens, Machi. 0I—I2, Pl. CXIII, to well-made plain red ware of the type associated with Ghâgha-shahr and Shahristân. Pieces of blue-glazed pottery were rare, and none were found with painted design under glaze. The fragment of a stone pot, 016, such as found elsewhere with ` chalcolithic ' pottery in this area, also belongs to this earliest period of occupation. The remains of this occupation, so abundant at other points of the southern delta, will be discussed in the next section. The List in section iii also includes specimens of pottery and of glass fragments picked up at other points of the site. Of these many, if not most, probably belong to modern times. But it must be remembered that wind-erosion has probably, in places, brought to the surface also relics of earlier periods. This was obviously the case in the southern portion of the Machi area ; for in the vicinity of the mound just mentioned the ground showed clear marks of wind-scouring, and rudimentary tamarisk-cones could be seen in course of formation, just as at sites but recently abandoned to the desert along the southern edge of the Taklamakan.
It was while proceeding from Machi towards the well and post of Girdi-chah, on the trade route some i6 miles to the SW. from Hauzdar, that I first passed a belt of regular Yardangs, or kalward, as they are called in Sistân. They were only from 4 to 5 feet in height, but duly prepared me for the effects of wind-erosion as exhibited by the ruins of the southern group of Muhammadan sites. These are scattered over an area which extends for about 9 miles to the south-east of Girdi-chàh with a maximum width of about 3 miles. This area once received water from branching outlets of the Rûd-i-biyâbàn, still clearly recognizable, which trend to the west and south-west from where the Perso-Afghan boundary line crosses them between the pillars marked B.P. 17, i8 b on the map. The old canals fed by these outlets can still be followed in places.
In the vicinity of Ramriid, the nearest of these sites and the best known, cultivation had, as Site of already stated, been carried on as late as the beginning of the nineteenth century. But the examinaRâmrùd.
tion of the ruined fort village to which that name is principally applied (Fig. 494), as well as of certain other remains, soon showed me that this Iate occupation meant merely a partial reclamation
of land which had previously for a considerable period been abandoned to the desert. The circum-
vallation of the fort, as shown by the sketch-plan, Pl. 57, has been so badly breached—in places on the N. and NW. it is almost completely effaced—that exposure to the eroding force of wind-
driven sand must reasonably be supposed to have extended here over more than one century.
The same impression is conveyed by the appearance of the ground around the walls, which, as the photograph in Fig. 503 shows, has been cut up into regular Yardang trenches and ridges. I am
hence inclined to believe that this fort, and probably also a smaller one about half a mile to the SE.,
were already ruins when water was again brought to this neighbourhood in Malik Bahram Khan's time. On the other hand, a group of eight or nine domed tombs, some of good size, which stand on
ground but very slightly eroded, about i miles to the east of Râmrûd fort, may perhaps date back only to that latest occupation. A stretch of perfectly flat ground covered with a crust of hard silt, which we passed within a mile or so of Râmrûd fort to the SE., probably also marks land that was then under cultivation.