BEACHES OF LAKE DANA. 771 



powerful river in the slowly falling waters of the Otisco Valley. The gorge heads 4 

 miles northwest of Marcellus, and a mile west of Sheppards Settlement, on limestone, 

 with an elevation at the intake, as given by Dr. Gilbert, of 812 feet. Here, at the head 

 of the eroded "gulf" (the only local name), the drift and shale are removed down to 

 the hard limestone rock over considerable area. It is evident that an enormous vol- 

 ume of water escaped at this point. This was the water of Lake Warren, which found 

 here an outlet lower than its old one westward across Michigan to the Mississippi. 

 Its western flow, that had been sustained perhaps some thousands of years, was, by 

 the removal of the ice dam in this region, slowly reversed and shifted to the east 

 toward the Mohawk-Hudson. This was the end of Lake Warren proper. For the 

 similar bod}' of water, but with falling surface and diminishing area, which found 

 lower and lower outlets eastward along the ice front, we can have no specific name, 

 but, using a generic term, maj' speak of it as the hyper-Iroquois w.iters. 



The Warren overflow into the Otisco Valle}'' would have been quickly checked if 

 some eastward outlet were not provided. This is found in another rock gorge, which 

 we call the Cedarvale channel, that leads southeast from Marcellus to the Onondaga 

 Valley. A great part of the excavation of this gorge was done pari passu with the 

 cutting of the gulf and by the same water, but the initial height of the outlet must 

 have been less than the height of Lake W^arren. 



Theoretically the Warren waters entered this region with an elevation of more 

 than 880 feet. Evidence of erosion at near this level appears in a cliff on the west 

 side of the valley, 1 mile south of the intake, which has the appearance of stream 

 cutting. This is near Mud Pond. On the east side of the channel, at the east-and- 

 west road, one-half mile below the intake, is a gravel plain at 880 ± feet, which is 

 probably a flood plain of the early river. 



THE WITHDRAWAL FROM LAKE WARREN TO LAKE ONTARIO. 



LAKE DANA (LAKE LUNDY?) 



In falling- from the level of Lake Warren to that of Lake Iroquois the 

 glacial waters appear to have made brief halts at several levels below the 

 lower beach of Lake Warren. These halting places are indicated b}' weak 

 beaches. Of these temporary lake levels probably the most important is 

 one which has recently been given the name Lake Dana by Fairchild.' In 

 explanation of the halting at this level Fairchild has suggested the great 

 resistance to erosion offered by the limestone which underlies the Cedarvale 

 channel near Marcellus, it being thought that the Cedarvale channel was 

 utilized by the falling waters down to that level. The shore of this lake 

 has as yet been but partly traced, the most important section being on the 

 west side of the valley of Seneca Lake in the vicinity of Geneva, N. Y. 

 It has there been given the name Geneva because of this relationship to 



1 Bull. Geol. Soc. America, Vol. X, 1899, pp. 56-57. 



