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

both in 1907 and 1914 after fresh snow-fall
in October.

The historical topography of the Chinese
'highroad' from An-hsi to Hāmi has been
discussed by me in Serindia, iii. pp. 1141
sqq., and also in connection with the ad-
venturous desert crossing of Hsüan-tsang
(A. D. 631). The record of this in the
pilgrim's Life throws interesting light on
the physical features of the forbidding 'Gobi'
to the S.E. of Yen-tun; cf. Geograph. Journal,
November 1919, pp. 273 sqq. For a brief
account of the route followed in 1914 from
Mao-mei to the northern slopes of the
Karlik-tāgh, cf. Geograph. Journal, xlviii.
pp. 200 sq.

Here in its easternmost portion the
T'ien-shan, as elsewhere along its far-
flung line, forms a great geographical divide.
To the north of that part of the range which
raises its crest above the line of perpetual
snow and which is appropriately known as
Karlik-tāgh, 'the Snowy Mountains', we
find plateau-like ground (A. 2), typical of
Dzungaria, with abundant grazing and water
more than sufficient for the limited area
actually under cultivation. But eastwards
where the range steadily falls in height,
ultimately to merge in the plains of Mon-
golia, the character of its northern slope
rapidly changes and the wide amphitheatre
of piedmont gravel surrounding the small
village of Bai (C. 2) is as arid as any 'Sai'
of the K'un-lun.

Astronomically observed latitudes.
1906-08. K'u-shui, Camp 248 (south of station; B. 4) ... ... 42° 2′ 50″
1913-15. Chin-êrh-ch'üan, Camp 215 (to W. of springs; D. 4) ... 42° 28′ 58″
Bai, Camp 220 (near Bēg's house, by stream; C. 2) ... 43° 13′ 6″

NOTES ON SHEET No. 38 (TUN-HUANG, AN-HSI)

The greater portion of this sheet, in the
north, comprises the desert ranges and pla-
teaus of the Pei-shan. Here the survey was
confined to the ground along the Chinese
cart road leading from An-shi to Hāmi and
followed in 1907; only in the north-eastern
corner and in the south-west were hill
features observed from routes of the third
journey. The southern part of the sheet
shows the wide trough-like valley of the
lower Su-lo-ho course, together with the
foot-hills of the westernmost Nan-shan range.
This ground, owing to the archaeological and
historical interest of the remains I was able

The southern slope of the Karlik-tāgh
is extremely barren throughout, and only
where subsoil drainage from the snows is
available for the irrigation of naturally
fertile loess beds at the top of the gravel
glacis, is permanent occupation possible, as
marked by the string of small oases from
Khotun-tam to Tāsh-bulak.

The southern glacis of the T'ien-shan
sinks down to a trough-like depression which
is clearly marked from Chin-êrh-ch'üan
(D. 4) down to Yen-tun on the An-hsi–Hāmi
road (A. 4) and probably extends for some
distance beyond to the south-west. This
depression, not unlike that of the Su-lo-ho
but without running water throughout, di-
vides the T'ien-shan glacis from the desert
uplands of the Pei-shan. The northern-
most range of the latter projects into the
south-eastern corner of this sheet and is
crossed by the Shuang-ch'üan-tzu pass (D. 4).
The absolutely bare glacis of the same desert
range is crossed further west by the very
exposed and much dreaded portion of the
caravan road between K'u-shui and Yen-tun
(A, B. 4).

Corrections. B. 3. The road-line west of
C. CLXIV should be marked as leading to
Hāmi.

D. 4. A vegetation area with spring
should be shown 8 miles S. E. of Pan-tzu-
ch'üan, with a route diverging to the south.

A pass symbol to be added against
Shuang-ch'üan-tzu pass.

to trace of the ancient Chinese border-line
along the Su-lo-ho course, was surveyed on
numerous routes both in 1907 and 1914.

The various traverses were adjusted on
the position adopted for An-hsi (D.3). The
observed latitude is 40° 31′ 38″. The longi-
tude 95° 57′ was derived from the closely
concordant results of the three traverses
brought in 1914 from the side of Tun-huang
(or Sha-chou; B. 4). This longitude showed
a difference of only one minute from the
value assigned to An-hsi in Sheet No. 81 of
the 1906-08 Map and was hence finally
accepted. But inasmuch as the longitude