L. DONCASTER 9 



were not so much smaller than the normal chromosomes of the species. 

 Such specks are not infrequent on the edge of the spindle, and probably 

 have no connexion with chromosomes. In this female, therefore, it 

 must be concluded that only 5.5 normal chromosomes were present, 

 unless possibly a very much reduced 56th is represented by one of the 

 specks mentioned. 



All these facts taken together lead to the conclusion that in Abraxas 

 gross id ariata the commonest chromosome number in the feuiale is 56, 

 but that in some strains one of these may be very much reduced or 

 absent. There is no necessary connexion between the number 55 and 

 the tendency to produce only female offspring, since all individuals 

 of the female-producing strain have 55, whether they belong to uni- 

 sexual or to bisexual broods. The nature of the mechanism which 

 results in unisexual families will be dealt with in connexion with the 

 account of the maturation divisions (below. Section d). 



c. Spermatogenesis of the strain which has 55 oogoniul chromosomes 

 in the female. 



In my first paper on the spermatogenesis of Abraxas grossulariata, 

 I described 28 chromosomes in both first and second spermatocyte 

 divisions ; I found no spermatogonial figures which could be counted 

 quite accurately. The testes then examined belonged to individuals 

 of the normal form with 56 chromosomes in both sexes, but it was 

 clearly a point of importance to find out whether the same number 

 occurs in males of the strain in which the female has 55. 



I have examined a number of testes of broods '12.3, '12.18, '12.25, 

 '12.26, '12.31, all of which families are either known to have females 

 with 55, or are descended from the same stock. In all of the numerous 

 first sjjermatocyte figures counted there are clearly 28 chromosomes', 

 but this would probably be the case whether the spermatogonial 

 number were 56 or 55. The number in the secondary spermatocyte 

 divisions is somewhat less certain, but there is practically no doubt 

 that it is also 28 in all cases. In testes of the broods mentioned, be- 

 longing to the 55-chromosome strain, I have counted 41 good secondary 

 spermatocyte equatorial plates ; twenty-nine of these have quite clearly 

 28, and twelve have 26 distinct chromosomes and a pair of chromosomes 

 in contact. In males of two broods ('08.10, '12.7) not belonging to the 

 55-chromosome strain, the condition with two chromosomes in contact 



' The only exception I have seen is in one figure in a testis of a male of '12.7, which 

 does not helong to the 55-chromosome strain. This single figure has 27, but many other 

 figures in the same testis have 28. 



