OCTOBEB 16, 1908] 



SCIENCE 



531 



is soTitlieastward, and half way between these 

 two points, it receives a tributary valley from 

 the north, this valley in its upper part being 

 occupied by Conesus Lake. The junction of 

 this valley with the Canaseraga Valley is more 

 or less obscured by drift, but can be readily 

 traced. The former stream of this valley evi- 

 dently discharged southward into the Cana- 

 seraga, and if the course of that stream was 

 northward, we have here a swinging of the 

 waters of the preglacial Conesus through an 

 angle of 135 degrees or more. Since the 

 present northern end of Conesus Lake is only 

 five miles in a direct line from the Cana- 

 seraga Valley at the village of Avon, a point 

 twenty miles north of, and therefore, on the 

 supposition of northward drainage, down 

 stream of, the point where the ancient Conesus 

 joined the Canaseraga, it is difficult to under- 

 stand why a branch of the main stream from 

 Avon did not capture and reverse the drain- 

 age, especially since the modem drainage is 

 along that line, and the rocks there are softer 

 than farther south. On the other hand, the 

 hjrpothesis of a southward drainage, in pre- 

 glacial times, of the Canaseraga and its tribu- 

 taries meets with no such difficulties, and the 

 form of the system is a perfectly normal one. 

 South of Dansville the Canaseraga Valley is 

 filled with drift, the top of which has a nearly 

 constant level between 1,300 and 1,400 feet 

 above sea-level. Since the elevation of the 

 valley bottom at Dansville is 700 feet, the 

 depth of the drift is between 500 and 600 

 feet. This is mostly local, and its character- 

 istics indicate that a glacier occupied the 

 valley to Dansville, and the discharge from 

 it built up the drift deposits in the valley to 

 the south. This valley, the southward con- 

 tinuation of the valley at Dansville, is fully 

 as broad and ancient as the valley at Dans- 

 ville. Without change in character, this val- 

 ley continues south past Bums and Hornell 

 and so along the present Canisteo Valley to 

 the Susquehanna. Any one who has traveled 

 through this region on the Erie railroad, 

 must have noted the well-developed and con- 

 tinuous character of these valleys. If there 

 ever was a divide here between a northward- 

 and a southward-flowing stream, as contended 



by the advocates of a former northward 

 drainage, this divide was in a valley as broad, 

 as deep and as open as any portion of the 

 valleys of these two streams for fifty miles 

 to the north and to the south of this divide. 

 Of course one might account for the character 

 of these valleys, by assuming that the entire 

 drainage system of the Susquehanna, inclu- 

 ding that stream as well, discharged northward 

 into the Ontario Valley, but I doubt if any 

 of the advocates of northward drainage would 

 go to such an extreme. The non-existence of 

 divides in these valleys, such as should be 

 expected in a normal drainage system, even 

 though it were developed on an ancient pene- 

 plain surface, as we have good evidence to 

 believe was the case with the drainage sys- 

 tem under discussion, this absence of divides 

 is in itself almost sufficient to condemn the 

 hypothesis of northward drainage. 



South of Dansville, where the ancient valley 

 changes from a southeasterly direction to one 

 a little west of south, the Wayland Valley, of 

 similar width and ancient character, comes in 

 from the east. Five miles east of the junc- 

 tion the Wayland Valley receives the Spring- 

 water Valley as a tributary valley from the 

 north, opposite the village of Wayland. 

 Seven miles north of Wayland, this ancient 

 Springwater Valley forks, the two prongs be- 

 ing occupied by Hemlock and Canadice lakes, 

 respectively. Here we have another drain- 

 age system, of the southward-draining type, 

 which joins the Canaseraga near Dansville. 

 Such a southward uniting of all the valleys 

 is wholly inexplicable on the hypothesis of 

 northward drainage during its development. 

 The Wayland Valley continues eastward to 

 North Cohocton, without change of character, 

 and with an average width of about a mile. 

 Here it receives another ancient tributary 

 from the northeast. This also forks, and in 

 the western branch lies Canandaigua Lake. 

 The valley of Honeoye Lake comes in here 

 from the north as another ancient tributary 

 valley. The points of junction of these val- 

 leys near Naples are obscured by heavy drift 

 deposits, but they are not difficult to trace. 

 They are well shown by the outcrops of the 

 strata delineated on the geological map of the 



