NILSSON'S DISCOVERY 63 



in exceptional cases a real improvement had been obtained. 

 But, of course, usually only successful instances are pub- 

 lished, and concerning the remainder ordinarily no evidence 

 is at hand. 



In Svalof, however, where numerous experiments of the 

 same k : nd, but with different varieties of cereals, were con- 

 ducted side by side, the fact could not escape observation. 

 Soon, the idea suggested itself that if success is an exception, 

 the principle involved in the method can hardly be a valuable 

 one, at least, not one on which the breeder may confidently 

 rely. 



An instance may be given. As such, I choose the Chev- 

 alier barley, which is one of the most valuable kinds for the 

 brewers. In Sweden, it has the defect of being often ex- 

 posed to lodging or lying down. The culms are too weak, 

 and are thrown down by wind and rain shortly before the 

 time of the harvest. Great losses are usually the conse- 

 quence, and it was considered a distinct necessity to breed a 

 Chevalier barley with culms stiff enough to resist these evils, 

 even on the hard soils of the middle parts of Sweden. Here 

 the culture of this variety had been tried on a large scale, 

 but was given up in consequence of the defect alluded to. 

 In order to introduce it anew, and to restore the brewer's 

 industry to its former degree of development, the Chevalier 

 barley had to be improved and adapted to the circumstances. 

 This demand seemed the more proper, since elsewhere and 

 especially in Germany, this barley had attained the height 

 of its renown exactly at that time. There could not be any 

 doubt that it was the best kind for brewing purposes. On 

 this account, it was cultivated at Svalof on a large scale, and 

 with all possible care. It was submitted to repeated selec- 

 tion with the distinct purpose of giving it stifTer culms. The 

 results, however, did not correspond with the expectations. 

 The harvest remained comparatively small, and the quality 



