362 



THE AMERICAN NATURALIST [Vol. XLVI 



buted in the cell, not localized in chromosomes, which 

 may well have an influence on size. 



But whatever our conclusion may be concerning the 

 theoretical explanation of size inheritance, the practical 

 manipulation of it must clearly be different from that of 

 color inheritance. All possible combinations of color 

 factors existing in two distinct races we can secure 

 within two generations by crossing. New conditions of 

 color we can often obtain by the slower process of selec- 

 tion, thus modifying existing color factors. Modification 

 is, I believe, often accelerated by crossing, quite apart 

 from the effect it has in bringing about recombination, 

 because it has a tendency to increase quantitative varia- 

 tion. 



Size is an unstable character, ever varying. Slow 

 changes in size can be effected by selection without any 

 crossing whatever. Change in size is made more rapid 

 by crossing, because variability is increased thereby. 

 If further increase in size is desired regardless of other 

 qualities two large races should be crossed and the larg- 

 est second-generation offspring should be selected. Pro- 

 gressive diminution in size should be sought in a similar 

 way, crossing the smallest breeds. 



If a medium-sized race is desired, it may be obtained 

 by crossing a large with a small race and inbreeding the 

 offspring. Physiological limitations undoubtedly would 

 prevent unlimited size variations either plus or minus, 

 yet when we consider what extreme differences exist 

 among dogs, as for example between "toy terriers" and 

 "great Danes," we can scarcely doubt that the limits of 

 possible size variation have not been approached in most 

 of our domesticated animals. 



