422 ELEMENTARY SPECIES IN AGRICULTURE. 



to produce new forms. According to this theory species are not 

 changed into one another, but new forms arise sideways from the 

 old ones. The whole strain continues unchanged and only produces 

 from time to time single aberrant individuals. These are the real 

 sources of all progress, and experience has shown that in the main 

 their new characters are hereditary, and that their progeny remains 

 true to their new types even from its first appearance. 



In agricultural breeding-practice the production of new races is 

 an intricate problem. In many cases their relation to the theoretical 

 conceptions is quite clear, in others it is still surrounded with doubt. 

 In my book on the mutation theory I have explained how the obvious 

 facts agree with that idea, but it was at that moment impossible to 

 remove all doubts and so I purposed to return to these questions 

 another time (Mut. Th., I, p. 82). Five years have since elapsed 

 and new discoveries have been published which enable us to give a 

 far more complete analysis of the agricultural breeding processes. 

 Especially at the agricultural experiment station in southern Sweden 

 quite unsuspected facts relating to the variability of agricultural 

 crops have been brought to light. They are of a nature to throw over 

 all the old ideas concerning race amelioration and give proof that 

 the methods now generally in use in Europe are faulty as well from 

 a practical as from a scientific point of view. The director of that 

 station. Dr. Hjalmar Nilsson, has discovered that most of our 

 ordinary agricultural crops are not only composed of elementary 

 species, as was long known before him, but that each cultural 

 variety contains hundreds of sharply definite types. These are 

 widely distinct from one another as well in botanical characters as 

 in those properties which decide on their utility from the breeder's 

 point of view. Moreover, they differ so widely from one another 

 as to respond to almost all the requirements of the agricultural 

 practice. By simply searching among them, the proper type may 

 forthwith be found for almost each gap in practice. In this way 

 they are seen to afford almost unexhaustible material for selection. 



For to-day's theme I have chosen an application of these dis- 

 coveries of Nilsson to a criticism of the current views concerning the 

 bearing of agricultural breeding processes on the theory of evolution. 

 Formerly I gave the warning not to trust too much to these pro- 

 cesses and to make use, in scientific discussions, only of the most 

 simple and clear cases (Mut. Theory, I, p. 59). The new facts, now 

 at hand, go to prove that even the apparently simple methods of 

 selection have been far more complicated than their authors sus- 



