182 



THE AMERICAN NATURALIST [Vol. L 



ing" in a self fertilizing population) that allow of the forma- 

 tion and perpetuation of "little species," side by side and yet 

 quite distinct, within highly variable taxonomic species such as 

 the dandelion. These same characteristics of self fertilizing 

 populations furnish much of the material which plant breeders 

 use. Following Vilmorin, they find it necessary only to isolate 

 and propagate by themselves the variations which spontaneously 

 arise. The task of the breeder who is dealing with a continu- 

 ally cross-breeding organism is much more complex. He often 

 finds it necessary first to inbreed his stock, in order to learn 

 what potential variations it contains, or, if one prefers so to 

 express it, in order to induce variability, though this form of 

 statement is not strictly accurate. Such inbreeding of a natu- 

 rally cross-breeding organism often causes temporary loss of 

 vigor, as notably in the case of maize, and frequently in domes- 

 tic animals. But when the desired variations have been isolated, 

 vigor can usually be recovered by increasing the stock to such 

 an extent that matings become possible within the race and yet 

 not involving union of closely related individuals. 



Notwithstanding the utility of inbreeding in securing varia- 

 tions, there are important sources of variability which are found 

 in cross-breeding alone. Supposing that under inbreeding 

 variation has already occurred in different directions and the 

 original condition has been wholly lost, it is often possible to 

 recover it again by crossing. This is the familiar phenomenon 

 of reversion upon crossing. It is also possible by crossing to 

 combine in one race variations which have occurred separately 

 in different races, a thing which would be impossible under con- 

 tinuous inbreeding. But a certain amount of inbreeding must 

 usually in such cases follow up the cross-breeding in order to 

 isolate and make secure the combinations desired. 



It is not wise, therefore, unduly to exalt either inbreeding or 

 cross-breeding as evolutionary processes or tools of the breeder. 

 Each has its utility at the proper time and place. They are like 

 pick and shovel, each supplementing the work of the other. 



The question is worth considering in this connection — what 

 effect will inbreeding and cross-breeding respectively have on 

 the variability of single characters. This is a question to which 

 I have given considerable attention for several years and the 

 answer to it is, I think, becoming clear. A single character 

 which Mendelizes has its variability increased by crossing. Some 

 explain this as due to actual modification of the unit character 



