NILSSON'S DISCOVERY 73 



degree of dissimilarity. The cases of such mixed progeny 

 were rare enough to be considered as the consequence of 

 the selection of such hybrids, and special experiments have 

 since given sufficient proof of the truth of this assertion. 

 Leaving these hybrids aside, the cultures of 1893 advanced 

 the importance of the selection of single individuals as the 

 one reliable source of purity, to the rank of an experimentally 

 estabHshed fact. 



From this fact it could further be deduced that a repeated 

 selection would be unnecessary. The next generation 

 might be expected to be as pure and as true to the type as 

 the first. IMoreover, the uniformity was such as to make 

 another selection simply impossible. All the differences 

 which formerly afforded the material for selection had dis- 

 appeared from these new strains. They were observed to 

 exist among the separate cultures, and allowed a compari- 

 son of these in exactly the same sense as they formerly had 

 made possible a choice in the fields. But within each cul- 

 ture no other differences were seen than those unavoidable 

 degrees in development which result from the differences 

 in location between the central plants and those of the border, 

 or between accidentally crowded or locally favored individ- 

 uals, and the average of the group. 



These observations led to the estabhshment of the second 

 great principle, that of the sufficiency of the one initial 

 choice. After that, the newly isolated type has only to be 

 multipUed and to be kept free from accidental admixtures. 

 On this point, the Svalof method agrees with the principles 

 observed by Le Couteur and Shirreff, who, likewise, did not 

 repeat their selection. For the industrial side of the work, 

 this principle has a high value. In the beginning, it was 

 feared that the reduction of the commencement of a race to 

 one single head might protract its multiplication so as to re- 

 quire more generations to reach the quantity necessary for 



