408 MITCHEL CARROLL 
in the late prophases and forms a tetrad like a typical euchromo- 
some pair. This tetrad divides in metaphase like any other 
tetrad; the two halves divide again in the second mitosis, as in 
the case of other pairs. When the supernumerary is present in 
triplicate in a first spermatocyte cell, two of the extra dyads 
synapse to form a tetrad, the other behaves like a tyipeal un-. 
paired supernumerary and exhibits no tendency to form a hexad. 
Occasionally a supernumerary pair apparently fails to synapse. 
When this happens the free segregation of the two homologous 
dyads, and the fact that one of the latter may divide in the first 
mitosis, explains certain second spermatocyte counts, not in 
agreement with the rest of the cells in the cyst and follicle. With 
respect to the supernumeraries and the accessory, six (with pos- 
sibly rarely eight) classes of spermatozoa may be formed in the 
testis of one individual. 
DISCUSSION 
A. Extra elements and numerical variations in the complex in 
other material 
I have referred to variations in the chromosome number, due 
to the presence of one or more supernumeraries, among the indi- 
viduals of a species reported in the Hemiptera by Wilson (’07 a, 
07 b, ’09 a, ’09 b, ’10), in the Diptera by Bridges (’138, 714, 716), 
in the Coleoptera by Stevens (’08, ’12 b), in the Orthoptera by 
Stevens (712 a), Carothers (17), McClung (’17), Robertson (17), 
and Wenrich (17). These observers stress the constancy of the 
chromosome count within the individual. Wilson (’09 b, p. 185) 
alone mentions having observed in a few cells a numerical dis- 
crepancy. With this exception, no variation for the cells of the 
individual have been reported by the above authors. On the con- 
trary, all, including Wilson, emphasize the fact that the zygotic 
complex apparently maintains itself from the time of fertilization 
to the first naturation division, and that there is no real numeri- 
cal-variation (with the above exception) in the first spermatocyte 
complexes of the individual. | 
Supernumeraries occur, too, in plant material. They have 
been reported for the Oenotheras by Gates and Thomas (’14), 
