STABILITY OF THE GENE 295 



merit — are selected for the second orange trial and for 

 the control in light and dark, we are dealing already with 

 a more responsive type, a selected group, and these are 

 expected to again respond in the next generation, as they 

 do in fact. 



G=3,73 



Normal 

 Coloring 



Light 



Dark 



Fig. 154. 



Diagram illustrating the results of selection of dark and light 

 pupae of the cabbage butterfly. (After Diirken.) 



Therefore, unless the material is genetically homo- 

 geneous at the start or unless other controls are used, the 

 evidence fails signally to establish the inherited effect of 

 the environment. 



The same error runs through nearly all the work of 

 this sort that has been done. Modern geneties, if it had 

 accomplished nothing more, would have justified itself 

 in showing the worthlessness of such evidence. 



We may pass now to a group of cases in some of which 

 it seems probable that the germ-cells themselves have 

 been directly injured by special treatment, and that the 

 injured germ-material is transmitted to later genera- 

 tions. Owing to this injury, malformations may appear in 

 successive generations. This means that the treatment 



