been formed since the Wall was built. The city of San-yiian Hsien, situated 
twenty miles north of Hsi-an Fu, has been cut in two, since it was built, by a 
ravine 200 feet deep and over 200 feet wide. Again, in the Loess country of 
North Shansi one may often notice the remains of a succession of old roads, 
one below the other, along the side of a ravine. The ravine in these instances 
marks the original trace of the road. This was washed away, so that it 
became necessary to form a new road along the margin of the first. This in 
turn became dangerous or unfit for use, owing to the continual falling-away of 
great masses of loess, and again a new road had to be made. To-daya 
yawning chasm remains, which has swallowed up each successive road, leaving 
perhaps remnants of the last two or three. The whole process has taken 
place within the memory of the older iuhabitants of the district, who can 
testify to the facts. Places like this occur along the roads from T’ai-yiian Fu 
to Shou-yang Hsien, and Hsin Chou. Ofcourse the practice of the inhabitants, 
who each winter root up or cut down every trace of vegetation, greatly 
accelerates the denudation of the loess deposits; but even taking this into 
consideration, we must conclude that the rainfall, during the time when the 
loess plain was forming, was far less than it is now. 
It may be suggested that these loess deposits were laid down at the bottom 
of a lake, but there is no proof of this; nor is there any reason to suppose 
that it is a fluvial deposit such as the Chihli plains. The north Shensi basin 
is open to the north, and there can be no doubt that the loess was originally 
brought down from the Gobi Desert, of which the Ordos is but an arm. 
During our stay in Yii-lin Fu we had excellent opportunities of noting 
how the wind, which we found to prevail from the north, carries southward 
the material of which the Loess is composed. The inhabitants of North 
China are familiar with terrible dust-storms, which sweep down from the Gobi 
Desert at all times of the year. 
The Loess sometimes shows stratification. A good example of this 
appears in the accompanying photograph (Plate 56)*, which was taken a 
few miles west of the Chiao-ch’éng Shan in Shansi. The strata occur in 
alternate layers of brick-red, and light-coloured loess; the former having 
rather the constitution of Shao-t’w. 
In many places we noticed, embedded in the loess, shells of a little snail 
inhabiting the country at the present time; whilst bones of the Mole-rat— 
a rodent with an underground mode of life like the mole’s—were also found. 
No fossils of any kind were discovered in the Loess. 
* cf. Plate ss. 
I 129 
