68 GLACIAL GRAVELS OP MAINE. 



margin of a river which then filled the whole valley to the height of the 

 terraces. The water rose by successive stages, and the central parts of the 

 valleys were never filled by a sheet of drift, as postulated by the erosion 

 theory. The erosion theory postulates a water channel along the valley, 

 and a pretty large one, but by no means so large as the whole space 

 included between the terraces. Professor Dana's theory requires several or 

 many times the amount of water required by the erosion theory, i. e., the 

 stream must have been swift enough to keep its supposed channel (the 

 space between the terraces) free of sediment. 



The presence in several of the river valleys of a central ridge, so 

 evidently an uneroded portion of a once continuous plain, strongly favors 

 the erosion theory as to the formation of the broader terraces of valley 

 drift up to the level of the central ridges. Tliis includes most of the ter- 

 races. Perhaps I have not seen the terraces at very high level, noted by 

 Jackson and Hitchcock in the Kennebec Valley, though I looked for them; 

 but I have notes of a few narrow terraces above the erosion terraces which 

 seemed to have been deposited in substantially their present shapes. Their 

 material resembles that of the glacial gravels, but is not much rounded. 

 These terraces were at first judged to be ordinary glacial gravels, but they 

 preserve so nearly the same longitudinal slope as the valley drift proper as 

 to give good ground for suspicion that they wei'e formed at the margin of 

 the valleys, as suggested by Professor Dana. But if so, it is not certain 

 that they were formed at the margin of a great river filling the whole valley. 

 During the final melting, the ice in the valleys — if we may follow the anal- 

 ogies of ordinary glaciers flowing in valleys — might sometimes melt fastest 

 on the side next to the warmed hills. A stream would form in these mar- 

 ginal depressions, and the sediments deposited in them would now appear 

 as terraces. These narrow high-level terraces may therefore be of semi- 

 glacial origin, i. e., formed between the bare hills on the one side and the 

 ice of the valley on the other. 



SUMMARY. 



The channels of the rivers of valley drift time have been greatly 

 deepened and widened, partly by the direct action of the rivers upon the 

 valley drift which then filled up the lower parts of the larger valleys, 

 partly by the rains and by subterranean waters. In this process terraces 



