National Institute of Informatics - Digital Silk Road Project
Digital Archive of Toyo Bunko Rare Books

to fifteen feet. This comparatively plentiful vegetation is probably supported by subsoil moisture
brought down into this part of the depression by the flood-bed of the Po-yang-ho, which debouches
from the transverse hill range farther south. When this zone was passed, bare patches of clay
increased and gradually gave way to a surface of gravel or very coarse sand, and here the well
of Ko-ta-ch'ûan-tzŭ was reached after a total march of seventeen and a half miles (Map No. 42.
A. 4). Aneroid readings showed for our camp at this place an elevation of 3,370 feet, fully three
hundred feet lower than that of Ying-p'an-p'u-tzŭ (3,670 feet). This, together with the direction
of the dry shallow beds met with in the course of our reconnaissances to the north-east and north,
justifies the conclusion that the lowest portion of the Hua-hai-tzŭ basin lies in this direction.

Reconnais-
sances
towards
Limes line. These reconnaissances were carried out on April 29th by R. B. Lāl Singh and myself indepen-
dently under weather conditions unusual for this desert region. Clouds hung over the basin,
hiding from view both the gravel glacis of the Pei-shan northward and the steep and rugged hill
chain to the south, and we actually started in the morning under light drizzling rain. Both recon-
naissances were rewarded by interesting discoveries. The Limes wall was struck by Lāl Singh
at a distance of over eight miles to the north, while I myself, to the east and north-east, traced
a line of ancient watch-towers stretching towards the wall but quite distinct from it. In describing
these remains, the survey and exploration of which kept us busy for a second day, I may con-
veniently start from the northernmost of the watch-towers mentioned.

Limes line
struck
north of
T. XLIV, a. We reached this tower, T. XLIV. a, after covering a distance of close on five miles from Camp
129, across clayey ground dotted with sparse tamarisk-cones. Numerous shallow beds coming
from the north-west showed surfaces of cracked mud on their banks, and suggested that flooding
had occurred during recent years. To the north of T. XLIV. a we crossed for about four miles an
area covered by low dunes and holding in its centre a narrow belt of small Yārdangs. We then
entered a zone of tamarisk-cones gradually growing in height, and at a point about five miles
from the tower, where these rose to about 30 to 40 feet, we came upon the line of the wall skirting
their foot.

Agger
formed of
tamarisk
faggots. As seen in Fig. 219, it presented itself here as a low mound half-smothered by drift-sand, but
unmistakably disclosing its character by a thick layer of flat tamarisk branches covering its top.
Where we cleared the agger of sand, it showed a width of 9 feet and still reached a height of about
4 feet. It was found to consist entirely of closely packed faggots of tamarisk wood, with plenty of
sand between them but no layers of stamped earth. The thickness of the tamarisk branches
employed indicated that old tamarisk growth must already have been plentiful near by when
the wall was constructed ; while from the absence of stamped clay layers anywhere along this
section of the Limes it may safely be concluded that water for forming pisé layers was difficult
to obtain.

Line of
agger hidden
by dunes. The agger, at a distance of about a furlong, disappeared eastwards among the tamarisk-cones ;
but Lāl Singh, when returning here under my instructions, succeeded in tracing its line at intervals
for about two miles farther east to a point where it became completely hidden under ridges of high
dunes. Westwards, too, the agger was lost, about 300 yards away, among tamarisk-cones. Going
half a mile farther, we reached the north-eastern end of a regular 'Dawān' of drift-sand which with
its dunes, 20 to 30 feet high, completely smothered the Limes line. Continuing, however, for another
mile westwards we got clear of these high dunes and then without difficulty picked up the dark
line of the agger again on open bare ground overrun by light sand only.

Continua-
tion of line
traced
westwards. It could be seen from the last high sandy ridge running for over two miles westwards with one
or two slight bends. At the spot where the photograph reproduced in Fig. 218 was taken, about
half a mile on, the agger stood to a height of about 10 feet and was 6½ feet wide at the top. Here and