THE OLD COUKSE OF FALL KIVEE. 621 



this section: Coarse sand and gravel, 2 feet; medium buflF sand, 4 feet; fine, 

 even-bedded sand, 5 feet. The ridge was 20 feet wide and the hi}'ers 

 crossed it horizontally, as if they had been eroded on either side. A little 

 farther west, down the hill toward the bridge, the gravels were fomid to be 

 coarse and scarcely bedded at all. 



Everything shows that the floods swept west through the liernardston 

 Pass and, joined by the waters coming down the extensive upper valley of 

 Fall River at Bernardston village, passed into the Greenfield basin. 



THE OLD COURSE OF FALL RIVER. 



Commencing high ui) in the valley north of Bernardston, Fall River 

 is bordered by a broad, flat plane (1 P) that has been cut in a heavy 

 sand deposit whicli once filled the bottom of the valley, forming the 

 lake bench, and which in part still remains intact on either side of the 

 alluvial liuttom of the river. At the bridge in the village of Bernardston 

 this plain leaves the river and skirts the west edge of the kame area, being 

 bounded on the west by West Mountain, and extends southwesterly into 

 Greenfield, where it merges with the lake bottom of the Greenfield basin, 

 above whicli the esker ridges project for a distance and then are finally 

 submerged. The river, on the other hand, does not follow this lower plain, 

 as would seem natural, but nins from the bridge due south, in a deep, 

 narrow channel, cut in the much higher kame area, and then among the 

 drumlins and the sandstone ridges to the Connecticut. 



It seems to me certain that when the waters of the Connecticut became 

 confined to the main valley to the east, the stream coming down the Ber- 

 nardston Valley continued to run southwestward by the now abandoned 

 channel, and cut down and flattened the kame material into the broad, flat, 

 lakelike watercourse which now remains, and which forms now, near the 

 town line, a low di\ade from which a small brook runs back into the main 

 stream and another on to join the Ley den Brook, in the west of Greenfield. 

 The continued melting of the ice beneath the kame area at last di-opped 

 the sands so low that the stream suddenly found a new course opened to it 

 directly south into the Connecticut. 



It can be clearly proved that the flow of the glacial waters in the 

 Bernardston Valley commenced long before the ice had lowered so far 

 that any connection with tlie main valley can have existed, for the esker (k) 



