AN EXTRA DYAD AND EXTRA TETRAD IN CAMNULA 419 
in other individuals anywhere along the germ line. Even in the 
four individuals considered it is possible to account for the in- 
constant chromosome numbers by assuming that the variations 
arose during spermatogonial kineses. But in the latter case it is 
necessary to assume also that the supernumerary frequently 
fails to divide normally within one individual. The data so far 
collected indicate that the element is quite regular in its be- 
havior at mitosis. In only four first spermatocyte cells out of 
some hundreds examined was non-disjunction of the two halves 
of the supernumerary detected. The explanation of the vary- 
ing counts that has been offered is believed to be the simplest 
one consistent with the facts, assuming for animals 950 and 2525 
only one abnormal division of the supernumerary, and for 980 
and 2511 but two such divisions. Nevertheless, I am well aware 
that further work on the members of this genus may invalidate 
some of the conclusions which have resulted from this preliminary 
study. 
D. Accumulation or piling up of supernumeraries 
As Carothers (717) points out, in populations where unpaired 
extra chromosomes occur, on account of their free segregation in 
the maturation divisions, a piling up of these elements is to be 
expected. But since she never found more than two in any indi- 
vidual in Trimerotropis or Circotettix, she concludes that there 
must be some method of elimination. 
Stevens (12 b) found that the supernumeraries in Diabrotica 
might accumulate to a certain extent, for she observed five in one 
individual. Wilson (’09 b) also found as many as five in one indi- 
vidual in Metapodius, and these extra elements in Metapodius 
were seemingly homologues, although they segregated freely in 
the maturation division. 
In Drosophila XXY individuals are viable, while XXX _ indi- 
viduals are not (Bridges, ’16). 
I have examined some fifteen animals from the camnulan insu- 
lar population of Puget Sound and have never observed more than 
three supernumeraries in a cell. Since some of the atypical ani- 
