250 Chamberlin and Leverett — Studies of the 



(Profile D, fig. 1). All but the last of these significant fea- 

 tures were noted by Mr. Carll. Having no opportunity to 

 examine the upper Conewango and Cattaraugus valleys he 

 suggested an outlet by the way of the Cassadaga valley. In 

 favor of the route here given we note (1) the greater width of 

 the Conewango valley (two miles or more), (2) its more direct 

 course to the lake basin (see accompanying map), and (3) the 

 presence of a much deeper buried channel. Three wells along 

 it penetrate drift to depths of 284, 314, and 330 feet respect- 

 ively without reaching rock, while a fourth at Yersailles, 

 seven miles from the lake, penetrates to a point 95 feet below 

 the lake level (aneroid) without reaching rock bottom, whereas 

 on the Cassadaga line the rock floor opens out to the Lake 

 Erie basin at an altitude of about 200 feet above the lake. 



The general configuration of the drainage features of the 

 region support this view of reversal as may be seen in part 

 by consulting the accompanying map of restored drainage 

 (map 2). The naturalness of this restored system lends sup- 

 port to the more positive evidence. 



We note in passing that on this line as well as other pre-gla- 

 cial drainage lines of this region the northward slope of the 

 rock floor is probably somewhat accentuated by a northward 

 differential depression. We would also note that rock exca- 

 vation by an interglacial stream may have occurred aloug this 

 line in the vicinity of the lake basin, it being not improbable 

 that an interglacial stream would have had its source some- 

 what south of the present divide, which is a moraine of late 

 glacial age. Both factors need to be eliminated in determin- 

 ing the rate of slope of the rock floor of pre-glacial times. 

 The present rate seems too great for the nature of the valley. 

 It may also be necessary to make allowance for excavation by 

 ice or by snbglacial water in the portion of the valley within 

 the glacial boundary, but such excavation could not have 

 occurred in the portion of the valley with north-sloping floor 

 that lies outside the glacial boundary, i. e. between Kinzua and 

 Steamburg. 



where on this and the three profiles subjoined, broken lines indicate not alterna- 

 tive but probable positions of the fluvial planes or deposits to which they apply. 



Profile B shows the erosion interval between the early glacial gravel and the 

 moraine-headed terrace on the Allegbeny near Tidioute, Pennsylvania, together 

 with the head of the early gravel train on Tidioute creek and the later gravel on 

 Brokenstraw creek. The Brokenstraw makes a detour to the east to join the 

 Allegheny. The detour is indicated by dotted lines, and the section is broken at 

 Garland where the stream turns eastward. 



Profile shows evidences of reversal of drainage on the Conewango valley. 

 On this and the succeeding profile the vertical lines beneath the line of the pres- 

 ent stream show the depth to which wells have reached. If no rock was encoun- 

 tered the sign (?) is introduced. 



Profile D shows the evidence for reversal from the Kinzua col northward along 

 the course of the old outlet. 



