306 The Pedigree Families. 



There is no reason to suppose that the first record of 

 a new species corresponds with its first appearance. I 

 found 0. elliptica and O. spatJiidata the first year that I 

 visited the place (in 1886) ; I found O. lata the year 

 after. It is very probable indeed that these and other 

 forms have also appeared in previous years either as 

 young plants or as seeds, 



I am convinced that the mutations of our Evening 

 Primrose were already in full swing when I began my 

 observations and experiments, and that I did not catch 

 it in the act of beginning to give rise to new species. 

 I simply discovered how the new forms, which though 

 in a latent condition were already there, came to light 

 from time to time. 



The beginning of the process of the origin of a new 

 species will escape observation so long as it is impossible 

 to induce mutations at will. And it is not likel}^ that we 

 shall be able to do that for a long time. 



The fact that two subspecies already existed at Hil- 

 versum in 1886 in full development speaks strongly in 

 favor of the view that 0. Lamarckiana was at that time 

 already in a mutable phase. The two species I am re- 

 ferring to are the O. laevifolia and O. hrevistylis. In 

 1886 I found two examples of the latter in flower; the 

 former I found in the same year as rosettes which flow- 

 ered in 1887 and provided the seed for the laevifolia- 

 family described above. Inasmuch as these two new 

 species had not been observed anywhere else it is reason- 

 able to believe that they arose on the spot. This view is 

 supported by the circumstance that I first found only 

 a few examples of each of the two species, each in a little 

 group to itself; 0. laevifolia in the northwest, and O. 

 hrevistylis at the northeast corner of the field. Whether 



