380 



ANIMAL BIOLOGY 



gives a partial answer, since the continual selection of the best 

 animals for mating and the best plants for seed has been a profit- 

 able procedure. But it has long been known that after a certain 

 amount of selection has been practiced it may cease to be effec- 

 tive, and thenceforth serves chiefly to keep the character at the 

 higher level attained. (Fig. 242.) 



The crux of the matter is in regard to exactly what the varia- 

 tions are. Both modifications and recombinations are usually in- 

 cluded, and this mixture of non-heritable and heritable variations 



Fig. 242. — Schematic representation of the effect of selection from the 

 viewpoint of Galton's 'law of filial regression.' 1, Mode before selection; 

 2, 3, 4, new (successive) modes, the results of selections of individuals at 

 2', 3', W '. The mode has been shifted in the direction of selection (toward the 

 right). But there has been each time an amount of regression indicated by 

 the length of the arrows. 



is what makes confusion. If we rule out recombinations by inbreed- 

 ing or by self-fertilization of homozygous individuals, soon we 

 establish pure lines. Then the variations are all modifications 

 and selection is ineffectual with characters which are not inherited. 

 The importance of this point was discovered by Johannsen in 

 careful experiments on the inheritance of characters in single pure 

 lines of a brown variety of the common garden Bean. For example, 

 by keeping the progeny of each individual bean separate from that 

 of all the rest, he was able to isolate a number of pure lines which 

 differed in regard to the average weight of the beans. Thus selec- 

 tion resolved the bean population with which he began into its con- 

 stituent 'weight types,' or lines, each of which exhibited a charac- 

 teristic variability curve of its own with a mode departing more 

 or less from that of the population But when Johannsen selected 

 within a pure line (ruled out recombinations) nothing at all re- 

 sulted; he was unable to shift the mode because he was dealing 

 with non-heritable characters. In other words, selection sorts out 

 preexisting pure lines (lines with homogeneous germinal constitu- 



