XV] Segregation and Species 285 



But this Is not enough. We must eventually go further; 

 and, supposing such tests to be applicable on a compre- 

 hensive scale to great numbers of natural forms, we must 

 ask whether the result of such an investigation will show 

 first that certain kinds of differences seQreo-ate and that certain 

 other kinds do not segregate; and secondly whether we shall 

 then recognize that it is to the non-segregating that the 

 conception of species attaches with the greater propriety. 

 Since of the non-segregating characters we know as yet 

 almost nothing, I am loth to attempt an answer, but I 

 cannot imagine upon what evidence anyone would rely 

 who should maintain that the answer must be affirmative. 

 De Vries has, as it seems to me, incautiously, defined with 

 some strictness the differences between varietal and specific 

 distinctions, declaring that it is the property of varietal 

 characters alone to exhibit Mendelian heredity (295, &c.). 

 Though I agree with him in perceiving that genetic research 

 may ultimately provide some approach to a valid distinction 

 between species and variety, I am reluctant to accept any 

 evidence yet attained as an adequate basis for so vast a 

 generalisation. Of the consequences of specific crosses — 

 in the stricter sense — little is known, and no case has been 

 fully explored. Even as to the results of crosses between 

 petites especes differing in several characters, present 

 information is most imperfect. Such cases as that of the 

 Violets, mentioned above, must be thoroughly investigated 

 by critical methods; and when a good number of these 

 examples taken from an ample range of types have been 

 submitted to factorial analysis it wqll perhaps be possible to 

 come to a more confident decision. But before any decision 

 at all is pronounced or even contemplated, the laws which 

 govern the Incidence of sterility must be most carefully 

 determined. 



Feeling thus the Impossibility of now defining the 

 segregating from the non-segregating, I am unable to 

 follow de Vries in the further step which he has taken (299) 

 in assigning a definite physiological reason for the difierence 

 between these classes. His suggestion is that in Mendelian 

 heredity there is a process, spoken of as "bisexual," in 

 which each determining factor i^Anlage) meets a corre- 

 sponding opponent; while in the other, or "unisexual" case. 



