nun 
MERWIN: SHORE-LINES. 321 
stagnant ice. Terraces due to three successive periods of grading 
of the present stream are found below the gravel terraces. The 
upper one, at an elevation of 775 feet, is much broader than the rest, 
and dies out further down the valley. Inasmuch as I found no 
barriers down the valley to account for these periods of grading, 1 
have supposed that they were caused by successive lowerings of a 
lake which formed the local base-levels. Evidence of a lake at this 
775-foot level is found northward, 3 miles further down the valley, 
where a delta-like deposit occurs at the mouth of Jones Brook. ‘The 
upper terrace shown in Plate 3, fig. 1, is also at this elevation. It 
is across the river from, and about half a mile north of Jones Brook 
and the village of Northfield. The lower terraces shown in the view 
are of the same origin as those described at the beginning of this 
paragraph. Similar terraces occur still further north. 
In the north-south valley next east of the Dog River valley a lake 
formed north of the 890-foot divide near Williamstown. This divide 
is in a swamp which drains both northward into the Winooski River, 
and southward through the Second Branch of the White River into the 
Connecticut River. Immediately north of the divide the valley floor 
is occupied by a lakelet about one half a mile long, probably an ice- 
block basin, and by a small esker with a bordering belt of kames. A 
little further north the stream draining the pond falls over a 20-foot 
ledge into a swamp. 
South of the divide there are two other small lakelets in rock basins. 
Beyond them, southward, the valley floor has been swept nearly clear 
of glacial materials by the temporary discharge of the glacial lake 
north of the divide. Still further south the flat valley floor gives 
place to a narrow postglacial gorge with cascades. Below the cas- 
cades the gorge widens considerably and is floored with alluvial 
deposits over which the small stream meanders. The excavation of 
this part of the gorge was evidently done by a larger stream than the 
present one, for the present stream is aggrading here. Part of the 
excavation is probably preglacial, most of the remainder was done 
by a glacial stream while an ice-tongue still projected south of the 
divide. Only a small part seems to have been done by the water 
flowing from the glacial lake behind the divide, for erosion has not 
been great near the divide and the small lakes in the path of such 
drainage have not been filled. 
The ice-front stood for some time about 3 miles north of the divide 
, 
discharging debris into a lake which occupied the site of Williamstown. 
A section about 20 feet in depth near the railroad station shows by 
