A paper on "Artificial Atavism," by Hugo de Vries, Director of the Botanical 

 Gardens, Amsterdam, was read by D. T. MacDougal, of the New York Botanical 

 Gardens. 



ON ARTIFICIAL ATAVISM 



By Hugo de Vries. 



Crossing is a means of analyzing compound characters. It is also a 

 means of coinbining the elements of such characters, and of building up the 

 original type out of its components. In some cases the compound nature of a 

 character may be evident, in others it is not, and in most cases it is as yet 

 doubtful. So it is clear that a combination gained by crossing may assume the 

 aspect of something quite new, and this will be nearly always the case where 

 it is not possible to discern the exact relation of the "new" character to 

 those of the parents used for the cross. 



If now this "new" character happens to have been present in some of the 

 ancestors of the crossed types, it wall resemble a reversion to this lost 

 feature, and, provided no other explanation offers itself, it will easily be 

 taken for an example of atavism. Looking at this inference from another 

 point of view we are led to suppose that perhaps many cases registered now 

 as atavism caused by crossing may probably originate in this way. 



In other words, we may expect in all cases, where a compound character 

 has been lost in the course of evolution, but where its components still exist 

 in separate species or varieties, that it will be possible to rebuild the old 

 characters by combining the partial ones by means of crossing. Such a com- 

 bination would evidently deserve the title of artificial atavism. Of course, I do 

 not assert that all atavism is to be explained in this way, or that all crossing 

 of the constituents of compound characters will have the expected result. I 

 only think that in a number of instances the now existing difficulties may be 

 overcome by this method. 



Taking a special instance, the colors of flowers present themselves in the 

 first rank. A great many of our garden flowers show large numbers of 

 varities and sub-varieties that exhibit different colors. In the case of peren- 

 nial plants, which are sold as plants and not as seed, it is evident that each 

 degree in the fluctuating variability of a color may give a distinct so-called 

 variety, as even this degree is constant enough when propagated by buds. But 

 in the case of annuals and biennials, and even with all such species as used 

 to be sold as seeds, this possibility disappears, and only such features as arc 

 transmitted in the course of generations may be used as good differences be- 

 tween the varieties of commerce. Each such variety has its own character, 

 which remains constant and recurs in each succeeding generation, even if it 

 is subject to much fluctuating and individual variability. 



