ASPECTS OF KINETIC EVOLUTION 39I 



but they seldom or never surpass it. The breeder concludes 

 that hybridizing is a mistake and finds that much more can be 

 accomplished by selection. This conclusion is quite correct if 

 he is dealing only with long-domesticated strains of plants and 

 animals, and if he wishes to obtain from them the greater ac- 

 centuation of some character already specialized by selection. 

 If the varieties are not too unlike, or too long selected, the result 

 of crossing will be to restore the more normal but less desirable 

 diversity. If the varieties crossed are somewhat more remote, 

 the diversities may balance each other into a somewhat uniform 

 intermediate average. Still longer selection may establish the 

 specialized characters as definitely alternative, in the Mendelian 

 sense, so that they do not combine again into a single hereditary 

 pattern, but separate regularly into the two original components, 

 as in the pea hybrids studied by Mendel, and the many other 

 instances discovered by more recent investigators. 



In none of these three cases or types of hybrids is there any 

 reason to expect an increase of characters beyond the range of 

 accentuation to be reached by selection ; they all involve, instead,' 

 a lessening of the amplitude of diversity obtainable through 

 selection. If the selective specialization of characters of a va- 

 riety were a true step in the evolution of the species, these kinds 

 of hybrids could be called reversions or retrogressions, since 

 they appear to go backward and undo the results of selection. 

 To call them reversions is very misleading, however, from the 

 evolutionary standpoint, since the closely selected type, however 

 useful, represents only a temporary and abnormal phenomenon, 

 a holding of the pendulum of variation to one side, instead of 

 permitting it to describe its normal vibrations, or to change its 

 general position and point of support. 



The simple analogy of the pendulum proves entirely in- 

 adequate as a means of illustrating the normal conditions and 

 requirements of true evolutionary advances of specific groups, 

 for we are not dealing then with vibrations of single characters, 

 but with a complicated network, a veritable fabric of descent 

 and of character-combinations. The pendulum analogy is ap- 

 propriate only for the single lines or narrow strands of descent 

 which selection separates from the web of the species, and 



Proc. Wash. Acad. Sci., February, 1907. 



