QUANTITY 121 



improve the yield unless larger and larger grains and 

 particular plant characters are always associated with 

 larger crops. Even in its most drastic application, which 

 consists, say, in selecting the very largest plants with 

 the largest heads and retaining only the very largest 

 grains, this method may not succeed, for it may be 

 that none of these characters is associated with larger 

 yield. 



It was early observed at Svalof that none of the 

 varieties tested was absolutely pure but that all in- 

 cluded sub-varieties which were distinguishable by 

 slight botanical differences. At first this was not 

 regarded as important, but, when a few plots which 

 were known to have been descended from single plants 

 were found to be uniform, and uniformity was thus 

 shown to be attainable, it was realized that the method 

 by which improvement had been sought would have to 

 be modified, and pure-breeding kinds only submitted to 

 investigation. Every sound experimenter knows that 

 reliable results are not to be expected with uneven 

 materials. 



Professor Hjalmar Nilsson, who was now director, 

 decided to adopt what is known as the " single ear " 

 method : a method bv which two British farmers, 

 Shirreff and Le Couteur, had achieved remarkable 

 results early in the nineteenth century. By this 

 method, the descendants of a single plant are kept 

 separate till stock enough has been accumulated to tell 

 whether the line or variety so selected is good or bad. 

 About the same time the idea that botanical characters 

 are indications of yield was abandoned at Svalof and 

 the indications of the weighing machine adopted as the 

 onlv reliable criterion. 



