710 The Chinese Loess Puzzle. [ December, 
formed concretionary layers exercise an important influence in 
determining a terrace-like form of the sides of the gorges worn 
down in the loess, which often rise in a succession of steps, hav- 
ing this peculiarity : that there is little or no talus or sloping pile 
of débris along the line where the vertical and horizontal surfaces 
meet. The natural tendency to this condition of things is as- 
sisted in its development by the labors of those who cultivate the 
soil, and for whose advantage it is to gain as much flat surface as 
possible. 
The first person to notice and describe these remarkable de- 
posits was Professor Pumpelly, who, in 1864, journeyed over a 
portion of the loess-covered districts lying along the southern 
border of Mongolia. His observation, however, did not, by any 
means, extend over so wide an area as those of Richthofen, and 
his theory of the origin of the loess appears to be insufficiently 
supported by the facts which a much larger and more exhaustive 
investigation of the country brought to light; still, it must be ad- 
mitted that there are difficulties which no theory seems able fully 
to overcome. Professor Pumpelly considered the loess, which he 
describes under the name of “ terrace deposit,” as a lacustrine 
formation, each of the basins in which this material occurs hav- 
ing been once the bed of a lake, a series of large bodies of fresh 
water being assumed as formerly extending along the course of 
the Hwang-Ho, which did not then occupy its present position, 
but ran in a pretty direct line, connecting the different basins, 
from Ning-hia-fu to Peking. This theory, therefore, demands. 
that there should have been a considerable diminution in the 
quantity of water formerly covering the region in question. It 
seems the most plausible one at first sight, however, and other 
observers —as for instance Kingsmill —have not hesitated to 
adopt and defend it. 
The difficulties which this theory presents seem, as developed 
by Richthofen, almost, if not quite, insurmountable. The mam 
arguments urged against Pumpelly’s theory will be readily m- — 
ferred from what has been stated in the preceding paragraphs. 
The absence of indications of stratification, and the constant pres- 
ence in the material of land instead of fresh-water shells, and 
of bones of land animals, are facts which it seems impossible to 
set aside, since they result from prolonged observations made by 
a most skillful geologist. Besides, the loess indicates by its 
structure the growth on its surface during deposition of an abun- 
- dant vegetation. The plants themselves — grasses, chiefly — are 
