24 HEREDITY IN RABBITS, RATS, AND MICE. 



mates. Since this is the direct effect upon the English character of 

 one dose of the self race, it might be supposed that tico doses would have 

 a greater effect, so that if it were possible to lift the English character 

 bodily out of the English race and siuround it with the complete resid- 

 ual heredity of the self race, an effect perhaps twice as great as that 

 actually observed in the cross might be expected. Accordingly an 

 advance of between 2 and 2| grades may be attributed to the residual 

 heredity of the self race. Theoretically, if this residuum consisted of 

 a number of independent factors, then full effect would be secured 

 upon breeding with each other the highest-grade individuals of the 

 cross-bred race, repeating this process generation after generation 

 until each factor was present in a homozygous state. This is sub- 

 stantially the procedure which has been followed in the 5 full genera- 

 tions over which the experiment has extended. The advance realized 

 amounts to about 2f grades. 



Allowing for the fact that one of our arbitrary "grades" may not 

 have the same genetic value as another, it seems probable that we 

 have secured something more than the effect of the residual heredity 

 of the self race employed in the original cross. This may have resulted 

 either from a process of elimination from the heredity complex of 

 factors which tended to lower the grade of the English character or 

 from change in the hereditj' complex by some other process than addi- 

 tion or subtraction of factors — for example, by change in factors. 



The important fact which this experiment demonstrates is the same 

 as that shown in the selection expermient with rats, that the single 

 characters which serve to identify our domestic races of animals 

 and which give value to them, e-\-en though they conform with every 

 criterion of unifactorial Mendelian heredity in transmission, do nev- 

 ertheless vary through minute gradations. By reason of the fact that 

 the residual heredity affects such characters, a cross into an unrelated 

 race can not be made, except with the possibility, or usually with the 

 probabiUty, that the character or characters in question will be thereby 

 modified. This fact was formerly expressed in the statement that 

 "contamination" of unit-characters frequently follows upon cross- 

 breeding — a form of statement, however, which was challenged by 

 those who maintained that the gametes were "pure." Subsequent 

 investigation has sho'wai beyond question not only that unit-characters 

 are frequently greatly modified by crosses, but also that they can be 

 modified bj' selection alone unattended by crossing. 



Those who formerly maintained the doctrine of gametic purity now 

 shifted their ground, and while admitting that unit-characters might 

 change, insisted that single factors or "genes" could not change. 

 This is the doctrine of pure genes which Morgan has made so familiar. 

 This doctrine it is diflficult either to prove or disprove. Pragmatically 

 speaking, it is of small consequence, since it is admitted (1) that single 



