IS92.; 
21 [Davis. 
when looking at large drunilins in the field, is the one that coni- 
]>ares them to sand bars in rivers or to sand dunes on windy deserts. 
They are, according to this explanation, accumulations of sub- 
glacial till at places where more material was Iwought than could 
be carried on. The conditions by which such excess of accumula- 
tion are determined are so various that it would be as difficult to 
predict the location of adrumlin among our rugged New P^ngland 
hills as to predict the location of a sand dune on the uneven sur- 
face of the Sahara. The excess of subglacial drift may be due to 
the jn-esence of ]U'eglacial soils, to the weakness of the rock 
foundation, to the ])resence of earlier drift deposits. The de- 
ficiency of transporting power may be due to some inequality of 
weiglit or motion of the ice-sheet, not dependent on the immedi- 
ate locality of the drumlin, but determined by the integration 
of accelerations and resistances both northward and southward of 
the locus of accumulation. It seems to me that in our hilly coun- 
try, where the distribution of the rocks and the variation of their 
texture and structure are so irregular, and where the probability of 
several advances and retreats of the ice-sheet is so strong, it is 
entirely unwarrantable to rule out this hypothesis simply because 
we cannot give immediate reason for the location of individual 
drunilins. There is no such hesitation in ascribing sand dunes to 
the variation in the load and carrying power of the wind, even 
though the distribution of sand dunes on desert surfaces appears 
to be fortuitous. 
Mr. Upham finds reason to conclude that his explanation of the 
origin of drumlins confirms a growing belief in the essential 
unity of the glacial period, although the ice-sheet may have had 
many oscillations back and forth ; and that it therefore contra- 
dicts the various arguments that have been brought forward in 
favor of its division into strongly separated epochs. I fully agree 
with him that the occurrence of washed gravels and of forest beds 
between layers of till, by which the complexity and subdivision of 
the glacial period was first argued, does not inevitably lead to 
such a conclusion ; but on the other hand I believe that evidence 
of quite another character, based on the work done by rivers be- 
tween two ice advances, as well as on the greater or less weather- 
ing and general wasting of different drift deposits, gives good 
reason for separating successive advances of the ice-sheet by a 
measure of time fully as long and probably much longer than 
