406 MITCHEL CARROLL, 
gated freely and might have gone into the same spermatid. 
According to the counts given in table 3 (follicle A), another 
supernumerary dyad should have been found in the complex 
illustrated in figure 108. What became of this chromosome I 
do not know, but the complex cannot be accepted (in the absence 
of other unquestionable instances) as a case of variation within 
a first spermatocyte cyst. For the missing element, as some- 
times happens, might have been pulled out of place by the micro- 
tome knife. 
The instances of occasional aberrant behavior of the super- 
numeraries just given may explain the three exceptional com- 
plexes found in cyst 5 of 2511 (figs. 65 and 67) and the one in 
cyst 6 of 2525. For these second spermatocyte groups of eleven 
and fourteen chromosomes, in cysts where only twelve and 
thirteen groups should occur, would result from non-disjunction 
of the extra pair in the first maturation division. 
The two unsynapsed dyads of the extra tetrad, too, shown in 
figures 85 and 136, as well as the pair in a similar complex which 
is not shown in the drawings, furnish added proof that the dyads 
of this tetrad are homologues of the unpaired supernumerary. 
They have the same size, form, fiber attachment, and position 
on the spindle. 
E. Classes of spermatozoa in individuals with a varying number of 
supernumeraries 
Since the supernumary within one individual (tables 1 and 2) 
may be either unpaired, paired, or present in triplicate in different 
follicles, six classes of spermatozoa may be formed in the testis of 
such an animal. If we denote the euchromosomes by ‘e,’ the 
accessory by ‘z,’ and the supernumeraries by ‘s,’ these classes are: 
lle 
lle+2 
lle+s 
lle+2s 
llet+e+-+s 
lle+2z+2s 
