H. Onslow 59 



The result of pairing two types together is to produce nothing but 

 types, whether the parents are extracted from melanics or not, as follows : 



Type X Type \RR x RR]. 



Imagines 



Totals ... 2 269 



In family '1*1 E both parents were wild types; in family '18 F the 

 </* parent was an extracted type ; in families '18 and '18 P the 

 $ parents were extracted types. In this case again less than 1 °l^ 

 of melanic insects have been included accidentally^ The eggs are so 

 exceedingly small that it is extremely easy to transfer one inadvertently 

 to the wrong box in the hairs of the brush. 



The evidence from the above experiments, which include over 1,400 

 insects, shows that the melanic variety of T. comonaria behaves as a 

 simple Mendelian dominant with regard to the pale type form. More- 

 over the melanics are a perfectly definite class and show no tendency to 

 vary in either direction. 



In conclusion I should like to express my thanks to Mr L. W. Newman 

 for supplying me with the material, and some of the details mentioned 

 above, and to Miss Helen Moodie for her care of the larvae, to which 

 the size of the families is entirely due. 



^ (Note added Nov. 25, 1919.) Professor J. W. H. Harrison, whose work on this 

 species will be published shortly, tells me he has had occasional melanics from pure type 

 pairings, and vice versa, which could not be attributed to accident, as they occurred in • 

 single test broods. With the possible exception of the 1915 broods (p. 55) I have been 

 , unable to exclude the possibility of error. 



