QUATERNARY. 189 
slopes, whence every rivulet is now washing the loess, the formation could 
not now be laid down by streams. Neither does it exhibit in such situa- 
tions stratification such as we might expect of water-laid material, and 
this fact weighs against the possibility that these particular deposits repre- 
sent a former lake filling. They occur where wind drifts dust, as snow is 
drifted. On the other hand, the Huang-t’u formation, which constitutes 
the valley floor, is interbedded with frequent and irregular layers of 
gravel, with which it occurs in such intimate association that the entire 
deposit thus interbedded can be ascribed to one agent only, namely, 
torrential waters. [hus we are at once brought to recognize that portions 
of the Huang-t’u formation are laid down as alluvium, and that other 
portions, which lack the characteristics of alluvial deposits and occur in 
places where wind alone could drift them, are probably eolian. 
Immediately east of Liu-yiian and 1,300 feet, 400 meters, above the 
valley in which the village is situated, is a parallel valley filled with 
Huang-t’u. The occurrence of the formation at this high level is peculiar, 
and unfortunately our observation of its characteristics was not so detailed 
that we can now discriminate between the hypotheses of its origin. It may 
be a remnant of a once more wide-spread formation at this level; or it 
may be a wind-blown drift caught in this high valley in consequence of 
the funnel-shaped hollow formed by the hills. 
Following the highway southward from Téu-ts’un, we cross a rock 
divide into the narrow valley of a small stream, which flows through a 
sharp defile into an adjacent basin. In this valley a bed of stratified 
clay and marl, with a few inches of peaty material overlying Huang-t’u, 
occurs under 10 feet, 3 meters, of the same formation. The sequence of 
strata implies a temporary water body intervening between two episodes 
in the history of the valley, during which the Huang-t’u accumulated. 
The condition probably was local and may have resembled that which now 
exists in the larger basin adjoining on the southwest. 
By examining the topographic map C I, it will be seen that 4 miles, 
6.5 kilometers, east by north of Wu-t’ai-hién is a pond to which three 
streams are tributary and which has no outlet. It lies in a depression 
which isa little more than 100 feet, 30 meters, deep, 2 miles, 3.5 kilometers, 
wide, and 5 miles, 8 kilometers, long. The depression is part of the valley of 
a stream which appears to have normal development, and one is surprised 
to note that the brook which runs by T’ién-hua does not drain the basin. 
It will be seen that the outlet of the valley toward T’ién-hua is narrow, 
a fact which may be attributed, in part at least, to the limestones which 
there form the hills, the wider valley being excavated in slates. It is also 
true that this outlet is in line with points at which the parallel, adjacent 
streams enter canyons, and along which an upwarp is recognized in the 
general features of the topography. The upwarp is not, however, the 
