OKTMANN : THE CKAWFISHES OF THE STATE OF PENNSYLVANIA. 441 



of the Susqueliiuiiia. If traii-sport were at all jnubaltle we sliould expect to fiml 

 that it liad taken place here, as well as in the region of Lake Erie. 



Further, and this is the most important olyection to the transport theory, while 

 C. obscurus has invaded the Lake drainage, not only in Pennsylvania, hut also in 

 New York (Genessee River), in no case has the opposite taken place namely, that 

 C. propinquus has invaded the Ohio drainage. If the cro.«sing of the divide were 

 due to passive transport, the same cause should have acted in hotli directions; but 

 C propinquvs is entirely absent from the Ohio system. 



The latter objection holds good also with reference tcj another assumption, that 



C. obscurus may have crossed into the lake drainage by the aid of the old canal 

 which connected the Beaver River with Lake Erie (Erie extension of Beaver canal). 

 This canal (see Jenkins, 1903, p. 288, 289) was in i)art used as early as 1834, and 

 was completed in 1844 ; it was abandoned in 1871, and it cannot be denied that by 

 it C. obscurus might have been able to reach the Erie drainage. I would not liesi- 

 tate to accept this as correct if it were not for the fact that C. propinquus has not g(jne 

 in the opposite direction.'"" Precisel}' in the region of this old canal my collections 

 are very complete, and are supplemented by those of others (Messrs. O. E. Jennings, 



D. C Hughes, and W. R. McCouuell), so that I am positive about the absence of 

 C. j>ropinqims. 



On the other hand, we have seen that the specimens of C obscurus from the 

 tributaries of the lake seem to approach more closely those of Beaver River than 

 those of French Oreek. This would be in favor of the canal-theory, the canal run- 

 ning from Newcastle by the way of Shenango River to Conneaut Creek (Jenkins, 

 /. c), while French Creek was not so closely connected with it (although there was 

 a "French Creek feeder"). The aljsence of C p/-o^>/?i7»«.5 in the lieaver drainage 

 may be due to the fact that in Erie County, the canal was not so closely connected 

 with the streams running to the lake, and that thus the lake species could not get 

 into the canal ; or else C. propinqutis being the weaker species of the two could not 

 make any headway against the more vigorous C. obscurus. 



There remains another theory, namely, that the migration of C. obscurus into 

 Conneaut and Elk Creeks is due to stream-piracy. The latter has undoubtedly 

 taken place in this region in Postglacial times. The Postglacial divide between 

 Lake Erie and the Ohio was formed originally by moraines t)f the late Wisconsin 

 stage (Lake escarpment morainic system. See Leverett, 1902, PI. 18; also Carll. 

 1880, PI. 1) or by higher elevated parts of the non-morainic drift lying immediately 



^ It should, however, he home in mind that the dischnrpe of tlie wnter from the cnnal %vaa downward toward the 

 lake and thus that migration might in that direction have heen easier than in the opposite. — Editor. 



