410 C. H. DANFOETH 



On this assumption it might be expected, although it need 

 not necessarily follow, that germ cells carrying determiners 

 for two or more characters of selective value would be more 

 favorably circumstanced than those carrying only one such 

 determiner. The available data is not sufficient to throw much 

 light on this question, but so far as it goes it would seem to in- 

 dicate that such may be the case. For example, in 3-A the 

 combination brachj^dactyl-poly dactyl-white is represented by 

 sixteen individuals where the expectation is thirteen on the basis 

 of observed percentages and only six on the basis of expected 

 percentages. 



The question as to the time in gametogenesis at which the 

 selection takes place must also be left unsettled. The pur- 

 pose of these experiments was to try to select between mature 

 germ cells, and this seems to have been accomplished, but there 

 is also some indication that the effects of the treatments have 

 persisted for a period greater than the probable life of such 

 cells, ^Moreover, there is some evidence, too meager perhaps 

 to be given much weight, that ova, or their nuclei, are likewise 

 selected. This may mean that the alcohol is effective as a 

 selective agent as far back as the first maturation division and 

 possibly determines at that time which nucleus will remain 

 in the egg of the female or which will be the more effective 

 sperm produced by the male. In this connection may be re- 

 called the finding of Arlitt and Wells (loc. cit.) that, in the 

 rat at least, the alcohol affects stages of spermatogenesis in 

 the reverse order, attacking late stages first, and early stages 

 last. 



There is one interesting by-product of these experiments 

 which should perhaps be further emphasized, the indication 

 of individual idiosjnicracies in the transmission of traits. 

 Among breeders a belief in 'prepotency' has occasionally had 

 currency, but the possibility of this supposed phenomenon 

 being real is often discounted by geneticists. When the data 

 for these experiments were being tabulated it was observed 

 with some surprise that they might be interpreted as furnish- 

 ing evidence of something analogous to prepotency. It would 



