248 RESEARCH IN CHINA. 
Von Richthofen’s theory of the loess does not admit the interaction 
of winds and streams. He held that during the loess epoch dessication 
was so complete that the rivers ceased to flow, and wind alone was the 
sorting and transporting agency. From this extreme view we differ in 
assigning to rivers a practically continuous, though variable, activity as 
transporting agents. Had there been times of desert dryness, we should 
have deposits of desert sands and residual sheets of stony shingle, but 
these have not been observed in Chi-li or Shan-si. 
We recognize, however, certain evidence of climatic changes. The 
eolian drifts, which occur in the canyon of the Sing-ho and adjacent valleys, 
are deeply cut by growing canyons. The conditions of removal which 
now exist can not have been those of accumulation, and the one factor 
which may have changed is rainfall. With somewhat less precipitation 
than now occurs, the rivulets which are dissecting the drifts should have 
been ineffective, and the winds might pile up the continuous mass of 
loess which has accumulated in the hollow of the hills. Hence we infer 
a fluctuation of climate; and, since the drifts are younger than the moun- 
tains, the fluctuation is one which falls in the latest epoch, the Fén-ho. 
That other climatic variations occurred earlier in the Fén-ho epoch and 
perhaps still earlier in the Hin-chdéu is highly probable. 
This evidence indicates climatic change of a kind and order closely 
comparable with that inferred by Huntington for the Basin of Sistan, on 
the basis of river terraces.* 
Transportation of the Huang-t’u.—Whenever the climate was sufficiently 
moist, extensive watersheds must have yielded rivers, perhaps resembling 
the Huang-ho in size, which flowed from the continental interior to the 
sea. Among such streams rising in the plateaus of Mongolia, we would 
name the Hu-t’o-ho in its course as far down as Hin-chéu, and further 
through the Shi-ling to the Fén-ho valley. With our present meager 
knowledge of details of geography in northern Shan-si, we are unable to 
point to any other similar streams, but it is quite possible that stretches 
of the upper Huang-ho date from an early time, and that there were other 
streams flowing in a general southwest course, such as we have traced 
in our descriptions of the T’ang-hién epoch. 
Flowing from an area which was in process of elevation, and which 
was subject to climatic conditions extremely favorable to a rapid denuda- 
tion, such rivers must have been heavily loaded with silt, a large part of 
which differed from that commonly received by streams in having been 
thoroughly sorted by the winds and in being, therefore, the characteristic 

* Explorations in Turkestan, Expedition of 1903, under the direction of Raphael Pumpelly. Car- 
negie Institution of Washington, Publication No, 26. The Basin of Eastern Persia and Sistan, by Ellsworth 
Huntington, pp. 273 et seq. 
