ORIGINGOR THE LOESS OF THE MISSISSIPPI VALLEY °797 
Salisbury that the loess particles are composed in part of feld- 
spars, amphiboles, pyroxenes and other common constituents of 
the glacial clays. These silicates are decomposable under pro- 
longed weathering, and hence cannot well be supposed to come 
from residuary clays under the ordinary conditions of the Mis- 
sissippi valley. The presence of the calcium and magnesium 
carbonates, independent of the presence of shells, points in the 
same direction. This inference is strengthened ina peculiar way 
by observations in the lower Mississippi valley. Above the 
Lafayette gravels and below the loess there is a stratum of silt 
which does not habitually contain the characteristic silicate 
particles of the loess. This stratum has been by most observers 
associated with the loess, but it is separated from it by a soil 
horizon as abundantly affirmed by the observations of Salisbury . 
and the writer. On the other hand it graduates more or less 
freely into the Lafayette sands and gravels. The stratum is, 
as we interpret it, the last deposit of the Lafayette stage. It is 
a typical finishing deposit succeeding a fluvial sand and gravel. 
Now this has special significance in this relationship, in that it 
shows that in the stage closely preceding the loess deposit, the 
Mississippi did not lay down silts of the same constitution as 
the loess. The inference therefore is that the loess is not simply 
a fluvial silt brought down from the surface of the river basin, 
nor common wind drift borne into it, but that it had a special 
origin connected with glacial action which was competent to 
supply precisely the kind of silt of which the loess is made. 
It is hard to resist the force of this argument from the con- 
stitution of the loess taken in connection with the two distribu- 
tive relationships. Jointly they seem to force the conviction 
that the loess had its origin in some relationship to the ice of 
the Iowan stage and to the rivers that led away from the ice 
edge at that time. 
But the hypothesis that the loess is simply an outwash of 
glacial grindings distributed along the river valleys by the 
glacio-fluvial waters is attended by grave difficulties. This 
remains true whether the deposition be supposed to have taken 
