298 The American Geologist. November, i89t< 
rapid motion than during the cuhnination of the Glacial peri- 
od. Moreover, the streams formed on the surface of the ice- 
sheet by the summer melting before it was so thick as to have 
motion would be free from drift, so that they could readily 
find their way through crevasses, wearing pot-holes in the 
rock beneath, and thence flowing in subglacial courses; but, 
on the contrary, the superglacial streams during the departure 
of the ice, which then became more or less covered with the 
previously englacial drift, laid bare by ablation, were heavily 
freighted with the gravel, sand, and clay of the modified drift, 
which must have soon choked up the passages wherever these 
drift-laden streams found crevasses, causing them to flow in 
superficial channels walled and underlain by ice, until, near 
their mouths, the ice was melted through to the ground and 
kames and eskers there received the coarser parts of the riv- 
ers' burden. 
The rate of erosion of the giants' kettles, referred to a 
stage of incipient glaciation, and the rate of formation of 
kame knolls and hills and esker ridges during the wane of 
an ice-sheet, were surprisingly rapid, in comparison with the 
generally very slow rates of geologic action. Watch the 
artificial processes of granite and marble abrasion and pol- 
ishing, and there will be no need to doubt that the largest 
rock kettle of Christiania, Lucerne, or Archbald, could be 
hollowed out during the warm months of even a single year 
by a stream 20 or 50 feet wide and 2, 3, or 5 feet deep, fall- 
ing down a moulin 200 or 500 feet deep, and well supplied 
at the bottom with grinding boulders of granite and other 
very hard rocks. Crevasses and moulins would be formed 
in successive years at nearly the same situation, thus pro- 
ducing such a profvisely kettled surface as in the Glacier 
Garden. 
Much later, if the retreat of the icefields under ablation 
was so rapid as a tenth of a mile yearly, apparently its rate 
near Stockholm according to observations noted in the pre- 
ceding paper of this series, or about half a mile each year 
during centuries, as was probably true of the area of the 
glacial lake Agassiz and the vast plains of the Saskatchewan 
and Winnipeg country, we cannot doubt that the most mas- 
sive eskers, as in Sweden, and the highest kame hills, as the 
