418 MITCHEL CARROLL 
class, three in the twenty-five class, and only one in the twenty- 
three class were found. 
Individuals 980 and 2511 gave diploid counts ranging from 
twenty-three to twenty-six. (The fact that the twenty-four- 
chromosome class was not found in 2511 is hardly significant in 
such a small number of counts.) Assuming that these individuals 
each inherited the extra chromosome in duplicate, the numbers 
twenty-four and twenty-six can be accounted for by the non- 
disjunction of the two halves of one member of this pair at any 
of the earlier cleavages of the germ cells. To derive the twenty- 
three-chromosome complexes it would then be necessary to 
assume a recurrence of the non-disjunction process at a suc- 
ceeding mitosis in the twenty-four-chromosome daughter cells. 
If this is the way the variations in the complex originated the 
number of follicles belonging, respectively, in the twenty-three-, 
twenty-four-, twenty-five-, and twenty-six-chromosome classes 
should be in the proportion 1, 2, 9,4. The chromosome counts 
recorded for these two individuals can also be derived from an 
inherited unpaired supernumerary by assuming two successive 
abnormal divisions of the extra element. In this case the expec- 
tation is that the class of follicles will give the proportion 4, 9, 
2,1. Asa matter of fact, in individual 980 the ratio is 2, 2, 8, 1, 
and in 2511 it is 6, 0,4, 2. But the number of follicles in which 
counts were obtained is relatively so small that not much signifi- 
cance can be attached to the ratios. 
It is perhaps best not to speculate concerning conditions in 
individual 2526. Variation was discovered in only two prophase 
cysts, and in these the counts are uncertain. I am inclined to 
believe, however, that the mutation causing the variation in this 
animal must have occurred further along the germ tract than in 
the case of the other four aberrant individuals. For the count 
is constant for twenty follicles (all which contained metaphase 
figures). 
Other interpretations of the phenomena recorded in this paper 
are, of course, possible. It is quite possible, for instance, that 
the numerical mutations that have been interpreted as occurring 
in four individuals prior to the formation of the testis may occur 
