BY HEBER A. LONGMAN. 39 



should he traced hack to one isolated niutation ; in fact, 

 many ohservations on the origin of mutations and vari- 

 ations to-day are opposed to that view, for such often 

 arise in profusion and not in isolation. 



Have we not undouhted evidence that the paths of 

 variation in the past, even under the iron restriction of 

 natural selection, have been radial I It seems to me 

 that radial diagrams, such as those used by H. W. Conn, 

 represent the courses of evolution more accurately than 

 simple dichotomous branching ; and, to be logical, one 

 must assume that such diagrazns should be primarily radial, 

 and secondarily polychotomous. 8uch diagrams, although 

 necessarily tentative, are surely an appropriate expression 

 of the labyrinthine processes of evolution. 



