Sec. i] THROUGH THE DESERT RANGES OF THE PEI-SHAN 527
about 300 yards across (Fig. 288), at an elevation of a little over 7,000 feet.* None of the higher
points of the range to the west was hère visible. The ground continued almost level for about
a mile, and here we came upon a herd of mountain sheep peaceably grazing.
For about three miles the descent led down gently to the north-west in a slightly widening Approach
valley, brightened by the autumnal tints of the still comparatively plentiful scrub and grass. Beyond to Sim-
thatthe valley debouched on to a wide alluvial fan, cut up by very numerous shallow drainage well,
channels, all dry, and all trending towards NNE. The detritus slope was broken here and there
by low outcropping ridges of dark red or black rock, suggesting granite or some closely allied
rock. Isolated little hillocks also, like those met with south of the range, were to be seen in places
emerging from the detritus. On approaching a low terrace marked by some cairns, we came
upon a delightfully green patch, about a quarter of a mile across, with a large well 6 feet deep in
the middle, holding excellent water. We subsequently ascertained that this tiny oasis encircled
by detritus hillocks is known as Sha-thing-tzt1.
Our route on September r 5th led first over detritus slopes of the same type as those crossed Sgpulurin.gs of
on our descent from the Ma-tsun-shan. Remnants of completely decayed rocky ridges cropped out Tsagan-
here too at intervals and seemed of the same composition as that met with to the south of Sha-ching-
tzù. After about 10 miles' progress to the NW. we came upon another small patch of luxuriant
scrub with a shallow well, at an elevation, like the last, of about 6,roo feet. From here our soi-
disant guides would persist in going astray to the north. But fortunately after a long search we
managed to strike again the track still leading to the north-west. It took us first across a bare and
almost level plain of detritus and then over gently falling ground to a long depression with abundant
grazing and some springs. Near its western edge, below a well-marked rocky plateau, we found
six Mongol families encamped with their sheep and cattle, and from them learned that we had
reached Tsagan-gulu. These Mongols had come from the grazing ground to the south-east along
the Ma-tsun-shan and were to return there for the winter. They said they knew little or nothing
of the routes towards Barkul or Hâmi, and our endeavours to secure a guide failed. They, however,
told us the true names of our last two halting-places, and explained that the well of Lo-teo-ching,
for which our Chinese ` guides ' had mistaken Sha-ching-tzû, lay in reality to the south, on the
route usually taken by caravans from Chin-tea towards Ming-shui. This was said to skirt the
western end of the Ma-tsun-shan and probably passes through a depression in the line of Futterer's
I had found the name of Tsagan-gulu marked in the sketch-map of MM. Grum-Grizhmailo Caravan
as a point where, evidently according to information gathered from Mongols, a route leading L1 âssutai.
towards Uliassutai, the former Chinese administrative centre far away in outer Mongolia, was
supposed to cross a route that appeared to be the one we were following. It was therefore of
interest to learn, just as we were setting out next morning, that a large camel caravan had arrived
overnight carrying rice and flour from An-hsi to Uliassutai. If the information was correct it would
point to there being wells or springs to be found also on a line cutting transversely from An-hsi
across the Pei-shan routes of Futterer and Grum-Grizhmailo (Obrucheff).
Having secured welcome supplies in the form of milk and a couple of sheep, we made on March to
September 16th an easy march of about 16 miles to a well that a Chinese-knowing Mongol had Zell iu-kou
spoken of as Liu-kou. The track for the first five miles or so led through abundant scrub with
4 It is of interest to note that the pass by which Professor ' Grum-Grizhmailo's sketch-map seems to mark such
Futterer's route crosses this identical range, the third of a gap, but evidently only ` from native information ' and
his reckoning, shows an elevation practically the same, in a position too far east relative to Tsagan-gulu.
viz. 2,130 metres (6,983 feet) ; see his W'üsle Gobi, p. 17. .