34 NEW YORK STATE MUSEUM 



the channels are existing features, and so much more abundant, 

 widely distributed and conspicuous that they can be appreciated 

 better than the other evidences of the extinct and now invisible 

 lakes which produced them. 



These glacial water bodies can be divided, somewhat arbitrarily, 

 into two classes, according to the direction of overflow, (i) The 

 earlier lakes which poured their outflow across the divide into 

 drainage outside the Erie basin. This class of lakes produced 

 the channels described briefly on pages 15-18. (2) The later local 

 lakes which found escape for their surplus waters past the ice front 

 and across the intervalley ridges, but wholly within the Erie basin. 

 These lakes, which often constituted tributary series, made the 

 hundreds of channels which have been briefly described in the 

 preceding pages. 



With outlets to southern drainage 



All the larger lakes of this class existed in territory which has 

 not yet been surveyed for the State topographic map; and as the 

 stud}'- of lakes and water levels requires partictilarly a knowledge 

 of altitudes which are now known in this territory only along a few 

 lines of railroads, the investigation is made under a disadvantage. 

 Without the topographic map the amount of labor required to 

 secure full data is out of all proportion to the value of the results, 

 and hence the subject is discussed here only in a general and pro- 

 visional way, and subject to future correction and amplification, 

 when the topographic sheets of Cattaraugus and Wyoming counties 

 are available. The map, plate i, will show the general course of 

 the main water-parting between Erian and Alleganian waters ,and 

 the passes or cols across this line. 



At the extreme west end of the State one or more small lakes may 

 have been held in the north-sloping valleys with outlets across the 

 divide to French Creek. Other examples of this class of lakes may 

 be located on plates 2, 3, by the outlet channels across the main 

 divide. The two larger ones occupied the highest portions of the 

 Canadaway and Walnut valleys. 



All the other notable lakes of this class lie in either Cattaraugus 

 or Wyoming counties. The earliest primitive lakes must have been 

 along the southern side of the Cattaraugus valley, in the heads of 

 the valleys which incline northward. The lowest pass available 

 for each valley was compelled to carry some overflow for a time, 

 but possibly leaving little evidence of the flow. (The word "time" 

 is used in a geologic sense and as relative, and may cover 10 years, or 

 100 years, or 500 years, or even more. We have no way of estimat- 



