402 BULLETIN No. 127. [August, 



been questioned. All of our conclusions, however, have been based 

 upon the supposition that the data obtained in experiments with 

 fluctuations, were obtained from homogeneous material. Johann- 

 sen's (58 a, b) work has thrown into considerable doubt the homo- 

 geneity of natural populations. He has, moreover, concluded that 

 the selection of fluctuations has nothing to do with the improvement 

 of a race. Probably no other conclusion of recent times is so im- 

 portant to plant breeders. The work should certainly be duplicated 

 along as many lines as possible; for its corroboration would not 

 only sound the death knell of methods of improvements by the se- 

 lection of partial fluctuations, but would entirely change our con- 

 ception of procedure in other breeding operations. 



Johannsen's experiments were made upon typical fluctuating 

 characters, such as weight and length of seeds. The plants used 

 were species, like beans, that could be self-fertilized during suc- 

 cessive generations. All of the descendants of a single plant, aris- 

 ing by self-fertilization, he speaks of as a "pure line." The mem- 

 bers of a pure line were distributed normally around a modal or 

 type value in the case of each character considered. Likewise, all 

 seeds from plants of the same variety, made up of a large number 

 of pure lines, showed a normal variability. Some of the modal 

 values of the pure lines were very close to the modal value of the 

 variety, while in other pure lines the modes were quite different 

 from it. When any individual, differing widely from the mean 

 value of its pure line, was selected for propagation, its offspring 

 showed almost a complete regression to the type of its particular 

 line ; but showed no regression whatever to the type of the variety. 

 He concludes, then, that a natural variety consists of a larger 

 or smaller number of distinct types, each type having a distinct 

 modal value for 'particular fluctuating character. These distinct 

 types he calls "biotypes." Only by mutation or some rearrange- 

 ment of characters can a pure line come to contain more than one 

 biotype. If such a phenomenon takes place, the new biotype dm 

 be isolated, and remains true until another mutation or rearrange- 

 ment takes place. It is quite clear that the only role of selection is 

 to more or less completely isolate the different biotypes of a variety. 



Johannsen leaves out of consideration forms of vegetative prop- 

 agation; though for what reason I cannot understand. Tuber re- 

 productions in potatoes is a form of reproduction in a pure line. If 

 these conclusions are wholly warranted, and Johannsen's work is 

 extremely careful, no improvement can be made by selecting plus 

 fluctuations in potatoes, except upon the intervention of mutative 



