82 PROCEEJJIIVGS OF THE 



one of the somatic cells, by which division a line of descendants 

 originates, lacking a Mendelian factor. This was carefully in- 

 vestigated in the potato by Evans, who concludes that all cases 

 of bud-variation observed by liim are caused by loss of a factor, 

 wliich in normal condition mendelizes. Furthermore, Cramer, 

 in bis important study of the literature of bud-variation, a quarto 

 of 474 pages, states that he is unaware of a single case of such a 

 variation being due to the gain of a factor. 



I have long thought that there was a case of bud-variation 

 known in an undoubtedly i)ure line — namely, that described by 

 Johannsen in his bean-cultures; but the discovery of Nilsson- 

 Ehle that a supposed pure line can be pure in respect to one 

 character and hybrid in respect to another, shows that this case 

 also fails to furnish proof. Anyhow, even if bud-variation by 

 loss of a factor occurred in a pure species, it would not be fatal 

 to my theory. My standpoint, that with the possible exception 

 of mutation through loss of a factor species are stable, now 

 forces us to ask, How do new species arise? The answer is, 

 I think, by 



Crossing. 



We all know that a cross between two so-called varieties gives 

 a hybrid, which segregates in the next generation in a number 

 of forms, of which some at least are stable and can have other 

 properties than the varieties originally crossed. Consequently, 

 there is abundant proof that so-called new varieties arise by 

 crossing. Inasmuch as I have defined such stable varieties as 

 micro-species, and as Jordan has shown that the Linnean species 

 essentially is nothing but a collection of such micro-species and 

 their hybrids, I could conclude witliout further ado that species 

 originate by crossing, were it not that there is a general feeling 

 among biologists that between a variety and a micro-species thei*e 

 yet is a difference. If this were only based on a feeling or on an 

 assumption in regard to the possible behaviour of factors, we 

 could let it pass as outside the possibility of proof or refutation, 

 but this, thanks to de Vries, we cannot do. De Vries has claimed 

 that it must be possible to judge, from its behaviour after crossing, 

 whether a certain form is a species or variety. Species after this 

 contention must give stable varieties segregating hybrids. 



The difficulty of the test lies in the so frequent sterility of 

 micro-species referred to different Linnean species, which should, 

 if possible, be used for such a test. Happily, Prof. Baur in Berlin 

 has found that many species (in the old sense) of Antirrhinum are 

 fertile inter se. He gave me, very liberally, the seeds of some o£ 

 these hybrids, which allowed me to study the progeny of the 

 crosses Antirrhinum glutinosiim by A. majus and A. sempervirens 

 by A. majus. I have raised great numbers of these, up to 16,500 

 in one year, and think I am justified in concluding that, far from 

 being stable, these hybrids, on the contrary, show an enormous 

 segregation. If the time permits, I will demonstrate to you at the 





