370 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1915. 



introduced, there will be a profusion of forms, utterly unlike each 

 other, distinct also from the original parents. Many of these can 

 be bred true, and if found wild would certainly be described as good 

 species. Confronted by the difficulty I have put before you, and 

 contemplating such amazing polymorphism in the second generation 

 from a cross in Antirrhirvmn, Lotsy has lately with great courage 

 suggested to us that all variation may be due to such crossing. I do 

 not disguise my sympathy with this effort. After the blind compla- 

 cency of conventional evolutionists it is refreshing to meet so frank an 

 acknowledgment of the hardness of the problem. Lotsy's utterance 

 will at least do something to expose the artificiality of systematic 

 zoology and botany. Whatever might or might not be revealed by 

 experimental breeding, it is certain that without such tests we are 

 merely guessing when we profess to distinguish specific limits and 

 to declare that this is a species and that a variety. The only defin- 

 able unit in classification is the homozygous form which breeds true. 

 When we presume to say that such and such differences are trivial 

 and such others valid, we are commonly embai'king on a course for 

 Avhich there is no physiological warrant. Who could have foreseen 

 that the apple and the pear — so like each other that their botanical 

 differences are evasive — could not be crossed together, though species 

 of Antlrrhtnwm so totally unlike each other as majus and molle can 

 be hybridized, as Baur has shown, without a sign of impaired fer- 

 tility ? Jordan was perfectly right. The true-breeding forms which 

 he distinguished in such multitudes are real entities, though the great 

 systematists, dispensing w^ith such laborious analysis, have pooled 

 them into arbitrary Linnean species, for the convenience of collectors 

 and for the simplification of catalogues. Such pragmatical consid- 

 erations may mean much in the museum, but with them the student 

 of the physiology of variation has nothing to do. These " little 

 species," finely cut, true breeding, and innumerable mongrels between 

 them, are what he fiiMs when he examines any so-called variable type. 

 On analysis the semblance of variability disappears, and the illusion 

 is shown to be due to segregation and recombination of series of 

 factors on predetermined lines. As soon as the " little species " are 

 separated out they are found to be fixed. In face of such a result 

 w^e may well ask with Lotsy, Is there such a thing as spontaneous 

 variation anywhere ? His answer is that there is not. 



Abandoning the attempt to show that positive factors can be added 

 to the original stock, we have further to confess that v.'e can not 

 often actually prove variation by loss of factor to be a real phenom- 

 enon. Lotsj'^ doubts whether even this phenomenon occurs. The 

 sole source of variation, in his view, is crossing. But here I think 

 he is on unsafe ground. When a well-established variety like " Crim- 

 son King" Primula, bred by Messrs. Sutton in thousands of in- 



