L. DONCASTKH 7 



other lacticolor females there were 56. The females in which I first 

 found 55 belonged to a family descended from unisexual stock, and it 

 occurred to me that possibly the absence of one chrouiosome might be 

 correlated with the tendency to produce only female offspring. In the 

 autumn of 1912, therefore, I dissected all the larvae which attained a 

 sufficient size, and preserved the ovaries and testes ^ The sexes are 

 quite easily recognisable on dissection when the larvae are less than 

 half-grown, and the oogonial and spermatogonial divisions are more 

 numerous and clearer at that stage than later. In male larvae sperma- 

 tocyte divisions occur from about half-grown larvae onwards. The 

 methods used were the same as those described in my former pajKii-s ; 

 in counting chromosomes every figure has been drawn twice, at 

 intervals of a day or usually more, in order to get inde})endent con- 

 firmation of each count. 



Of the families obtained from females of unisexual broods only two 

 provided any considerable number of larvae which grew to a sufficient 

 size, 7.^. S, from a gross. $ ex. '11.^6 x gross. ^ ex. '11. 11, and '12. 25, 

 from gross. % ex. '11, 16xlact. ^J ex. '11. 14. From 'I'ii. 8, 88 larvae were 

 dissected, every one a female ; from 12. 25, 14 were dissected, 10 {/*, 4 J . 

 Several other families yielded from two to six larvae, some of which 

 will be referred to below. Since in normal fixmilies the larvae which 

 feed up in the autumn are much more often males than females, it may 

 be taken as highly probable that '12. <9 is a unisexual family ; where 

 only six or fewer females were dissected, it is possible that some males 

 will appear later. 



Of the 33 larvae (all females) of '12. 8 which were dissected and 

 their ovaries fixed, only 14 provided figures of oogonial divisions which 

 could be counted with any accuracy. Of these, nine provided altogether 

 eighteen figures in which 55 chromosomes could be counted with almost 

 complete confidence, and of these two or three especially are so perfectly 

 clear that no doubt about the number is possible. In a number of 

 other figures in these ovaries 55 is the most probable number, but in 

 counting chromosomes, when their number is large, a few really trust- 

 worthy figures are worth much more than many rather doubtful ones. 

 In addition, out of the fourteen larvae which yielded good or fair 

 figures, two showed four figures in which only 54 can be seen (none of 

 them first-class figures), and three showed four figures in which no 



1 The majority of grossulariata larvae cease feeding when quite small in the late 

 summer, and hibernate at that stage. A small and varying number feed up and pupate 

 in the autumn. All the latter were dissected, usually when about half-grown. 



