NILSSON'S DISCOVERY 99 



which he selected a goodly number of these aberrant types 

 must of course have been represented, since he selected only 

 those which caught his eye by some striking and useful 

 difference from the main type. Of course, he sought for 

 ears of one and the same ideal type, having a large number 

 of big kernels. But notwithstanding tliis, his handful of 

 ears must have belonged to more than one elementary 

 species, the real value of which could be judged only in 

 their progeny. Among these units of his selection some 

 must have been better yielders than others and the subse- 

 quent selection of his twenty years of pedigree-culture must 

 slowly but surely have eliminated the units of minor worth. 

 This would result in the end in a complete isolation of the 

 best one of all the types, wliich he originally but unconscious- 

 ly selected and mixed. 



Or in other words, Rimpau's pedigree culture was started 

 as a mixture of a number of excellent types, and his yearly 

 selection has gradually reduced this number, until he had 

 isolated and purified the very best one among them. This 

 point was, of course, only unconsciously reached, but then 

 it must have made his rye independent of all further real 

 selection, reducing the process to the care of excluding vi- 

 cinism. 



If this explanation of Rimpau's process is true, it of 

 course holds good for all similar cases of slow and gradual 

 improvement of agricultural plants by selection. Thereby 

 it would deprive the theory of the origin of species by slight 

 and continuous changes of its last support in the realm of 

 the vegetable kingdom. 



It remains to be shown that the new facts give sufficient 

 proof of the accuracy of this suggestion. These facts may 

 be grouped under three heads. First, the general occur- 

 rence of elementary species and their constancy. Secondly, 

 a comparison of the value of fluctuating variability and 



