292 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1926 
and opening Pliocene times. Dall’s determination of the San-Men 
fossils would narrow the lower limit to later than early Pleistocene. 
It is possible that the upper series of Sang-kan Ho beds will still 
further reduce the bracket," especially if they can be correlated with 
the beds observed at the base of the loess by Fathers Licent and 
Teilhard de Chardin in the Ordos region, who discovered at this 
horizon paleolithic stone implements strongly recalling the Moustier- 
ian stage of human culture.1*? The upper limit is set by the fossils 
found in the gravel of the redeposited loess. Of these it can only be 
said that they are comparatively recent, but are in some cases distinct 
types from the species living to-day. 
Judging from the data offered by the loess itself, it must be ad- 
mitted that no more exact date can be given. Andersson concludes 
that the fauna is “decidedly Pleistocene in type, and assuming that 
Dall is right in dating the mussels of the subloess San-Men beds as 
early Pleistocene, it will follow that the loess is of Middle Pleistocene 
age. It would then be the arid equivalent of the Pleistocene ice age ” 
(loc. cit.). 
From what has been said it will be clear that the loess is certainly 
as young as the date given by Andersson, and that it probably corre- 
sponds in time to the later stages of the Pleistocene ice age, but 
formed under more arid conditions. 
ORIGIN AND MODE OF ACCUMULATION OF LOESS 
The present position of many loess deposits, perched high up on 
the slopes of mountains, the uniformly fine grain, the absence of 
evidence of water action and the known semiarid conditions of the 
time, all point strongly to wind as the great agent responsible for the 
accumulation of the Asiatic loess. This was recognized by v. Richt- 
hofen in 1877. The fact that the geological study of erosion has for 
the most part been carried out in lands of moist climate has tended to 
give too little weight to the work of the wind in drier regions, 
whether cold or hot. The very fact that wide areas of such rock 
deserts as the Gobi have to-day no covering of sand implies that the 
eroded material of past ages has already been carried off as dust 
clear out beyond the margins of the desert, to sink and collect wher- 
ever moister conditions, the shelter of mountain ridges, or the pro- 
1°The studies in the Huai-lai Basin already mentioned together with later investiga- 
tions in the Sang-kan Ho area confirm this hypothesis. Moreover as an epoch of erosion 
intervened before the beginning of the loess epoch, the date of the latter was probably 
Middle Pleistocene. (Footnote added October, 1925.) 
2° Tf found in Europe, such implements would be taken to indicate that the overlying 
deposits were as young or younger than the interglacial period between the third and last 
(Wurmian) great advance of the ice sheet; following Osborn’s estimate that would fall 
well within the last 75,000 years. At present, however, no such exact correlation with 
China is possible.” 
