QUATERNARY. I9I 
the canyon was cut out, and could not have been laid down by water so high 
on the mountain slopes. There seems good reason, therefore, to attribute 
this deposit on the left bank of the Sing-ho to the activity of the northwest 
wind, which sweeps the upper loess-filled valley and here rises over the con- 
verging heights, leaving its burden of dust in the concavity of their slope 
(see Fig. A, Plate XXVI). 
Southwest of the Sing-ho there is a parallel valley in which an inverted 
stream flows northwest. The brook cuts a canyon 150 feet, 45 meters, 
deep in a deposit of Huang-t’u, which rises on the slope of the mountains 
to an altitude of 4,500 feet, 1,375 meters. The inversion of the stream is 
attributed to warping, but the accumulation of the loess, which here almost 
exclusively forms the Huang-t’u, is an effect of the northwest wind, which 
has brought it from the upper valley and deposited it in a situation pre- 
cisely parallel with that on the left bank of the Sing-ho (see view on atlas 
sheet CII). A glance at the map southwest of Wu-t’ai-hién shows that the 
two wind drifts present similar features of slope and canyon, and that the 
plain south of Wu-t’ai-hién exhibits the form of a smooth hollow, which 
might be a plain of accumulation or a surface scoured by wind. In its 
detailed features it exhibits both aspects, for at times the rivulets carry 
back to it the dust which at other seasons the winds again sweep out. 
Among the limestone ranges southwest of Wu-t’ai-hién is a wide 
plateau, the Yau-t’6u district, which corresponds with a syncline in the 
Ki-chéu (Cambro-Ordovician) limestones. An ancient valley from which 
the former river has been entirely diverted is represented at an altitude 
of 4,500 feet, 1,375 meters, by a surface eroded across the limestone and 
the sandy strata of the Shan-si coal-measures. Upon the old valley floor 
occurs a deposit of Huang-t’u, 20 to 30 feet, 6 to 9 meters, thick and of 
peculiar character. It is redder than that formation usually is, is more 
like residual clay than loess, and exhibits an imperfect vertical cleavage 
only. Regarding it as the alluvium of a valley dissected during the Fén-ho 
epoch, we consider this deposit to be of early date. Its present elevation is 
an effect of warping. A similar deposit of like age occurs in the T’ién-hua 
basin and is shown in the view, Fig. B, Plate XXIV. 
In the valley of the Hu-t’o-ho near Tung-yii occur deposits of Huang- 
t’u closely resembling those already described, and we there observed the 
movement of the dust in a strong wind. It rose in a dun-colored cloud, 
which hid the hills at a distance beyond 2 miles, 3.5 kilometers. But 
the great body of material drifted near the ground, as snow drifts, and fell 
in sheltered spots where it rapidly gathered to a depth of many inches in a 
single storm. Had there been any grass or other vegetation, the surface 
thus covered would have been protected from the wind and would, to a 
