70 PLANT-BREEDING 



of variability itself had to be reconsidered. It is rather a 

 state of polymorphy. The idea of continual changes can 

 hardly be connected with it. It is the existence of numer- 

 ous different forms, each of which is simple and almost in- 

 variable. The term should convey the idea of a mixture, 

 but hardly refers to actual changes in the constituent types. 



Nilsson began his work by selecting a considerable num- 

 ber of samples from the varieties on the fields of the station, 

 but in doing so, he still followed the prevailing method. 

 Each sample was sown on a separate field plot and progress 

 was tested in regard to purity and quality. Nearly a thou- 

 sand lots were cultivated, but the result was as unsatisfactory 

 as before. Everywhere the groups were seen to be hetero- 

 geneous and to consist of a more or less motley mixture of 

 types. The samples, however, had been chosen under the 

 assumption that they were uniform, and it had been expect- 

 ed that they would yield each a uniform progeny. This 

 however, was not the case, and doubts arose as to the re- 

 liability of the whole process of selection. If the progeny 

 does not correspond to the mother plants from which it is 

 derived, how can we tell that the next selection will produce 

 a generation with the desired qualities ? 



All reliability of the selection-principle seemed to dis- 

 appear, when, fortunately, an accidental observation was 

 made which at once changed the whole aspect of the ques- 



Ition. Some few cultures were discovered among the thou- 

 sands which bore only one type. They were as uniform as 

 the remainder were heterogeneous. Concerning the initial 

 choice of the samples, an elaborate record had been kept, 

 and this enabled Nilsson to discover the cause of the purity 

 of these exceptional cases. According to the accepted 

 method, each sample had consisted of a certain number of 

 ears, which were as similar to one another a's could be ex- 

 pected, and which were therefore, simply taken to belong 



