24 EXPERIMENTS WITH DROSOPHILA AMPELOPHILA. 



Condition 7. From all we know of the inheritance of fluctuating 

 variations, the mating of one grade with another will give variations 

 about a mid-grade. If it takes a certain grade or strength of the germ- 

 inal factor to produce a somatic effect and we mate a low grade (somati- 

 cally normal) with one just sufficient to produce somatic abnormalities, 

 the variations about the mid-grade will, in most cases, be too low to 

 produce visible abnormalities. In other words, normality will dominate 

 over abnormality. If, however, the parents are such that the grades 

 of the offspring are about that required to produce somatic effects, the 

 dominance will be imperfect. The abnormality will appear to be obey- 

 ing "the spirit of the Mendelian law/' but it will naturally "pay little 

 attention to the letter of the simple law or any of the modifying clauses, ' ' 

 especially in the latter generations, where the abnormal strain had the 

 germinal factor much strengthened by selection. 



It seems to me, then, that if we accept the notion of some specific 

 factor in the germ which brings about the details of the characters of the 

 soma the facts here discussed may be considered to be the result of the 

 action of a factor present in all germs. The strength of this factor 

 varies, and when of a certain strength produces certain visible effects. 

 The partial dominance of normal over abnormal is due to the mean con- 

 dition of the factor in the offspring of 



(flies with factor strong enough to produce extra veins) x (flies with factor weak) 



usually being below the strength required to produce abnormality. 



I have not taken up the resemblance of the behavior of these abnor- 

 malities to that of ' ' ever-sporting varieties, ' ' because I feel that classing 

 these as such would not be a step toward an explanation. It would 

 merely be naming the difficulty. It is also not the intention to imply 

 that this hypothesis would apply to those other cases which are trouble- 

 some from a Mendelian standpoint and to which the principle of vary- 

 ing potency of Mendelian determiners has been applied (for example, 

 Davenport, 1910). It seems not only possible but probable that many 

 apparently non-Mendelian cases may be explained as a combination of 

 alternative and blending inheritance (Lutz, 1908). But a simpler and 

 more probable explanation of these data, provided we accept the some- 

 what dubious "germinal factor " idea, seems to be that we are dealing 

 here solely with a fluctuating character the strength of the abnormality- 

 producing factor and that the study of its inheritance is made difficult, 

 if not largely impossible, by the fact that only in the upper part of its 

 range can we judge of the relative values of this variable, for in the 

 lower part its effects are invisible. 



