116 SURFACE GEOLOGY. 
The principal outlet from the part of this basin in Hillsborough county 
appears to have been through Greenfield south-easterly to Souhegan 
river. South from this pass the east border of the Contoocook valley 
is formed by Pack Monadnock, Temple, Kidder, and Barrett mountains, 
which extend in a continuous range through the west portions of Temple 
and New Ipswich. Northward this valley has a high eastern water-shed 
two to four miles from the river, with no deep depression till we reach 
the pass through which we have supposed a former outflow towards Pis- 
cataquog river. The culminating points of this water-shed are at its 
south and north ends, in Crotched mountain and Craney hill. 
When the melting of the ice-sheet had advanced so far as to open an 
avenue from this valley through Greenfield, we may suppose that large 
streams descended from the glacier to this point, by which the kames on 
the east side of the railroad south of the village, those between Hogback 
and Pollard ponds and along the road northward between Greenfield and 
Bennington, and those at Bennington station and for a mile north-west on 
both sides of Contoocook river, were in succession deposited. The fine 
alluvium of these streams was at first spread out in the level plain east of 
Cragin pond, while ice still remained over the area now occupied by this 
pond. A small lake was afterward formed by the melting of the ice on 
the north-west side of the pass. This lake received the finer drift brought 
down by the glacial rivers, producing the alluvial plain west and north- 
west from Greenfield. 
A channel appears next to have been formed farther to the north-west, 
skirting the hills upon the east side of the valley and walled on the west 
by ice. This became filled by the nearly level-topped and terrace-like 
gravel and sand seen on the east side of the Manchester & Keene Rail- 
road south from Bennington station, which seem to belong to the same 
date with the kames at this station and about Whittemore pond. The 
kames were probably formed in ice-channels which were narrow and 
somewhat higher than the former, with so rapid a descent that only 
coarse gravel was deposited in them by the summer floods, the sand be- 
ing carried onward to the quiet waters of the channel below, which was 
an arm of the lake. With the full melting of the ice, however, such of the 
kames as had been formed over the middle of the valley sank to its bot- 
tom, and are found at a lower level than the principal deposits of fine 
