416 MITCHEL CARROLL 
one member of a pair are not found within the cyst, when the cyst 
arrives at the first spermatocyte stage it should have less than the 
typical number of cells. 
The variations in the second spermatocyte complexes have 
been shown to result from the normal free segregation of the ac- 
cessory and the supernumeraries present in the first spermatocytes 
or from the occasional aberrant behavior of chromosomes in the 
first spermatocyte mitosis. Exclusive of the second spermato- 
cytes, counts were obtained in some seventy-four cysts in the five 
animals with extra chromosomes. In sevén of these cysts (early 
secondary spermatogonia) only one count was obtained in each 
cyst. In each of the other sixty-seven cysts from two to forty- 
three counts were made. In only two cells (table 3, A, 1; 
table 2, K, 13) were complexes found which might be considered 
exceptions to the statement that the numerical organization of 
the chromosome group is constant for the cyst. It was shown, 
too, in the analysis of the counts, that the fundamental numeri- 
cal organization of the complex is constant for the follicle prior 
to the second spermatocyte generation. 
The majority, if not all, variations in the chromosome group 
observed within the testis must have arisen, then, prior to the 
formation of the follicles. Nor, if there is only one primordial 
germ cell in Camnula, could they have arisen at the separation 
of this cell from the soma. They originated somewhere between 
_ the first cleavage of the primordial cell and the formation of the 
testis. Were it possible to secure accurate counts in all of the 
fifty or sixty follicles composing the camnulan testis, the par- 
ticular cleavages at which the mutations took place might be 
determined. But this, unfortunately, cannot be done. 
No complexes with an unequal pair or lacking one member of 
a normal pair, and no first spermatocyte cysts lacking a con- 
siderable proportion of the typical number of sixty-four cells 
were found in the atypical individuals. Hence it is probable 
that the numerical variations did not originate in the fragmenta- 
tion or abnormal behavior of a member of the ordinary chromo- 
some complex. They seem to be rather the result of the occa- 
sional aberrant behavior of the extra elements during mitosis 
