The Origin of Drumlins. — Terr. \VM 
one has discovered what they were, nor attempted to explain 
them. 
The more important objections to the hypothesis that drum- 
lins are forms produced by the partial destruction of some 
mass of moraine as a result of a re-advance of the ice, are 
found in the following- sentences from Upham's recent paper.* 
He says.— 
"If this view were true, the till of the drumlins could not 
have its nearly uniform character, but would contain here and 
there remarkable aggregations of boulders; and frequent ir- 
regular enclosures of sand and gravel would be found, repre- 
senting the kame deposits and lenticular beds of modified 
drift which so commonly make up considerable parts of the 
terminal moraines. Salisbury remarks that neither the dis- 
tribution nor the composition of the drumlins seems to favor 
this hypothesis, and he therefore believes that they were built 
up beneath the ice, not being fashioned from hills overridden 
by it."' These objections may be left unanswered for the 
present, since they are considered below. 
3. — Indications of Destructional Origin. 
(a) Distribution. — There is every gradation between sheets 
of till and typical drumlins. Chamberlinf divides these un- 
stratined drift deposits into (a) till tumuli, (b) mammillary 
and lenticular hills, (c) elongated parallel ridges trending with 
the ice movement, (d) drift billows, (e) crag and tail, (f) 
precrag and combings, (g) veneered hills, and unclassifiable 
till hills. Pointing out that they are somewhat allied to mo- 
raines, he says that there is "a richness of variety and inter- 
gradation that almost defy classification. " 
This gradation, while in places very Striking, is not always 
present and is not of itself proof that the drumlins are merely 
a stage in the accumulation of drift. It would be equally well 
explained by the opposite theory, that they are hills of de- 
structional origin derived from a preexisting moraine or from a 
drift sheet of -irregular thickness. Indeed, the peculiarity of 
distribution of drumlins suggests to my mind the latter rather 
than the former origin. While there is every gradation be- 
tween the most perfect lenticular hill and the undifferentiated 
*Proc. Boston Soc. Nat. Hist.. 1892, vol. xxvi, |>. 7. 
IIYoc. Am. Assoc. Adv. Sci.. 1886, vol. \\w. \>. 204. 
