Sec. in THE REMAINS OF AK-SIPIL AND KIGHILLIK 475
reproduced in the plan of Plate XXXIX, shows that the extant ruins form the northernmost segment, measuring about 36o ft. at its arc, of a circular rampart which must have originally enclosed an area a little over Boo ft. in diameter 3. Here, as at Endere, the lower portion of the circumvallation consisted of a rampart of hard stamped loess, such as is still used in Chinese fortifications throughout Eastern Turkestan, and popularly known as khilailik sighiz, ' Chinese earth-bank '. This rampart measured approximately 5o ft. at its base, and rose about II ft. above the original surface of the ground outside, which was still clearly distinguishable in places free of sand and uneroded.
The rampart was surmounted by a parapet built of sun-dried bricks, and resting on a supporting platform of the same construction about 2 ft. high. The thickness of the parapet was 8 ft. at its foot, while the platform below projected beyond it 3 ft. on the outside and 2 ft. on the inside. The parapet, more or less broken on the top, but otherwise in a fair state of preservation, nowhere rose higher than 7 ft. The large size of its sun-dried but relatively hard bricks, about 20 by 15 in., with a thickness of 4 in. on the average, as well as its solid construction, suggest considerable antiquity. The parapet, where intact to .a corresponding height, showed loopholes arranged in two uniform levels, one 16 in., the other 5 ft. above the base, but at irregular intervals varying from 5 to 10 ft. The loopholes were about 3 in. square in the lower row, and about 6 in. wide, with a height of 8 in. in the upper one. Only the upper row is likely to have been used for shooting arrows. The top of the rampart on the inside formed a banquette about 4 ft. broad, running along the base of the parapet, and evidently serving as a means of communication for the defenders. At two points of the extant wall-segment the parapet was strengthened by solid brick platforms, projecting about 3 ft. on either side of the base. These probably once bore constructions corresponding to watch-towers. Near the western one (6 in plan) the remains of a flight of stairs leading to the top from the inside were visible. In the angle formed by this with the adjoining parapet eastwards the red colour of the bricks indicated a sheltered spot where a fire must once have been lit, perhaps by men on guard.
The break visible in the parapet in Fig. 57 does not mark the position of a gate, as Mr. Högberg supposed ; for the rampart shows no sign of a cutting. It was easy to see that the parapet had fallen here from some cause. Inside the wall the ground showed considerable erosion down to a depth of 8 ft., as seen in the plan and section. It was this eroded hollow which the same visitor took for the remains of a ditch, with the result that in his description the inside area of the fort figures as the outside 4. To the north of the extant wall-segment the original surface of the ground was remarkably well-preserved, probably owing to the protection afforded by drift-sand, which the prevailing northerly winds heaped up here against the
s Owing to the ruined condition of the extant segment and its relatively small length, the position of the centre, as shown in the plan, could not be determined with absolute exactness. But the position of the isolated wall portion d in the reconstructed line of the circumvallation shows that the ascertained radius is sufficiently accurate for practical purposes.
' It is this confusion between inside and outside which explains, e. g., the curious observation : ' Near the gate and on the further side of the ditch, the remains of two towers (Stûpas?) are visible, &c.'; see Hoernle, Report on C.-A. ant., i. p. xv. In reality these ` towers' were two of the several
little isolated loess banks or ' witnesses ' which rise above the eroded ground within the wall. One of them, marked o in the plan and visible on the extreme right of the photograph in Fig. 58, was pointed out to me by Turdi (who had accompanied Mr. Högberg and Mr. Bäcklund on their visit), with a twinkle in his eyes, as a spot where the Sahibs' tried some digging for manuscripts, since Islam Akhün had indicated Ak-sipil as one of the find-places of his [forged] old books.' For a diagram of two similar natural loess banks which had been conjectured to have yielded some of Islam Akhûn's finds, see ibid.