PHYSIOGRAPHY OF NORTHWESTERN CHINA. 221 
The evidence in support of the assumption that the original course 
of the streams was by the valley abd, and not by the valley abc, has been 
given in describing the features of the region. It is found in the fact 
that the section bd is an old aggraded valley, in which there is a divide 
established on the Huang-t’u formation, and the channel bc is a character- 
istically youthful canyon. 
There is nothing in the conditions of rock structure to help us toward 
a solution, but if we draw a line MN, from southwest to northeast, between 
6b and d, we find that the canyon of the main river and a rather sharply 
cut gorge in the lower abandoned valley lie to the southeast of it, whereas 
a more open valley on the upper northwest fork and the aggraded valley 
of Li-yiian-p’u lie northwest of it. These relations of narrow, incised 
channels on the one side and widened, aggraded valleys on the other are 
intelligible if we assume that there was a local upwarp, of which the line 
MN was the axis. In consequence of such an upwarp, both of the streams 
abd and cd would be retarded on its northwestern side and accelerated 
on the southeastern slope. If they were strong enough to maintain their 
courses as antecedent streams, both would be characterized by canyons 
in the respective stretches bd and cd. If, however, either of them was 
unable to maintain its course while the other succeeded in doing so, the 
weaker would be reached by a tributary of the stronger and diverted to 
a course along the northwestern side of the upwarp. This is what appears 
to have taken place. In consequence of an upwarp, giving rise to the 
notable heights which form conspicuous outliers of the main range north- 
west of Féu-p’ing-hién, the main Sha-ho, cd, though retarded in its upper 
reaches, cut the canyon which characterizes it, and maintained its course. 
The southwestern fork, abd, on the other hand, failed to accomplish the 
task of canyon cutting with sufficient rapidity, and was so checked that 
it deposited not only gravel but also loess in its upper valley. While 
in this condition, whether absolutely ponded or merely rendered sluggish, 
it was reached by the southwestern branch which joined the main river 
at c, and the diverted waters developed the latest of the several channels, 
the canyon of the section bc. 
This explanation involves the assumption of the development of a 
relatively narrow upwarp having a northeast-southwest trend, and if the 
assumption applied only to this individual case it might be regarded as 
having slight foundation, but we shall see in the further discussion of 
the features of the mountains that there has been pronounced warping 
along axial lines having the same general trend, and the upwarp, as a 
feature common to the region, may be regarded as a reasonably probable 
local detail. The test of the hypothesis would lie in a closer study of 
