and the Mcndelian hypothesis. 527 



considered ; these produced respectively 16, 22, 17 and 27 

 imagines, and all were black with the exception of a 

 single intermediate "■•• in the brood of 22. In this brood, 

 and also in that of 27, the paternal grandmother was 

 black, the maternal purple ; in the other two, the reverse 

 was the case ; in no instance was a male grandparent 

 known, as all were second generations from wild females. 

 It is interesting, too, that in one of the broods which 

 almost failed, the only four imagines that developed were 

 black, notwithstanding that both the parents (black) were 

 the offspring of a purple female. Practically therefore, 

 black X black could be depended upon to breed true, 

 whatever the ancestry. 



Black x Purple. — Five broods give the following 

 results: (1) 42 black, 27 purple; (2) 37 black, 82 purple, 

 4 intermediate ; (3) 36 black, 31 purple ; (4) 49 black, 

 47 purple; (5) 7 black, 10 purple. This gives a total of 

 171 black, 147 purple, 4 intermediate; or respectively 

 (omitting fractions) 53 per cent., 45 per cent., 1 per cent. 



The brood which shows the greatest disparity in the 

 representation of the two forms — numbered (1) above — 

 had the $ parent black and the $ purple, while in the 

 other four cases the sexes were reversed ; but I am not 

 inclined to attach much weight to this slight discrepancy. 

 The pedigree, so far as known, was as follows : (1) paternal 

 grandmother black, maternal purple ; (2) and (3) vice 

 versa; (4) ^ parent a wild purple specimen, $ parent 

 (black) the offspring of black $ and $, which, in their turn, 

 sprang of captured purple and black females respectively ; 

 (5) both parents (brother and sister) the offspring of a 

 captured purple female. 



Purple x Purple. — I reared nine of these broods, 

 mostly with somewhat complicated pedigree, as I was 

 especially interested in attempting to eliminate the black 

 element by selection ; that I failed in this attempt, the 

 following figures will show. Six of the broods were large 

 enough, mimericall}^, to be taken into account :-f (1) 52 

 purple, 21 black {$ parent purple, wild ; $ parent purple, 



* Real intermediates are, in my experience, of exceedingly rare 

 occurrence ; I have only reared 10 which can be so regarded, 

 amongst considerably over 1000 specimens — i. e. less than one per 

 cent. 



t Even the few specimens reared in the remaining three cases 

 showed, in each instance, at least one black specimen. 



