CHROMOSOMES AND SEX-DETERMINATION 403 



seven (or the outer twenty-eight) and nineteen with the 

 converse arrangement. In the bisexual family 14.35, in which 

 15 o^o^ and 19 $? were reared from forty-six eggs, four eggs were 

 found in which the inner spindle had twenty-seven, the outer 

 twenty-eight chromosomes, and four with the converse arrange- 

 ment (all ' good ' counts including both spindles of each egg), 

 so it does not appear that the all-female families have twenty- 

 seven in the inner spindle with any greater frequency than in 

 bisexual families of the same stock. 



It seems evident from the facts given above that the deter- 

 mination of sex in the fifty-five-chromosome strain of Abraxas 

 grossulariata does not depend on the passage of the odd 

 chromosome to one or other pole of the first polar division. 

 At the same time, since females of this strain have fifty-five 

 chromosomes in their diploid nuclei and males have fifty-six, 

 a chromosome must be eliminated at some stage from those 

 eggs in which twenty-eight travel to the inner pole of the first 

 polar spindle. Attempts to find a chromosome which does not 

 divide in the second maturation division have not been success- 

 ful, and it seems clear that the elimination does not occur at 

 that stage. Only two possibilities remain : either a chromo- 

 some is eliminated at some division after fertilization — pre- 

 sumably the first segmentation division, or the odd chromo- 

 some must degenerate so that the twenty-eight chromosomes 

 present in about half the eggs at the inner pole of the first 

 polar spindle are reduced to twenty-seven by the degeneration 

 of one of them. Neither possibility seems likely on general 

 grounds, but there are some facts which make the hypothesis 

 of the degeneration of a chromosome less entirely improbable 

 than would appear at first sight. These, will be discussed in 

 the next section. With regard to the hypothesis of the elimina- 

 tion of a chromosome in the first segmentation division, I can 

 only say that I have not succeeded in obtaining figures in which 

 the chromosomes in this division can be accurately counted ; 

 in the few segmentation divisions present in my material the 

 chromosomes tend to become aggregated into small groups, 

 apparently of two or three, so that counts give numbers not 



