48 
SPERMATOGENESIS OP NORMAL 
Oils cells for union at fertilization. If a spermatozoon and an 
egg containing characteristics of the same species unite, then 
the reversion will be to that species; if a sperm-cell containing 
the characteristics of one species happens to unite with an 
ovum containing characteristics of the other species, then the 
offspring will be of the mixed type again. By the law of prob¬ 
ability the latter will be the more prevalent occurrence, because 
there are four combinations possible, and two of the four would 
result in the production of mixed offspring, while only one 
combination could result in a return to one of the ancestral 
species. 
From the fact that in cases of apparently complete return to 
one parent type, characteristics of the other parent may never¬ 
theless crop out from time to time in succeeding generations, 
it is evident that all of the germ-cells are not absolutely “pure.’’ 
The occasional inequalities in the division of individual chromo¬ 
somes, as already mentioned may account for this fact. Not 
infrequently in division more than half of one chromosome goes 
to one cell; thus, what might otherwise have been a pure cell 
will contain a few characters of the foreign species and these 
may appear later in offspring as variations of the intermediate 
type. Thus the actual number of intermediate forms will 
doubtless be augmented somewhat beyond the proportion in¬ 
dicated in the probability of combination where all the mature 
germ cells are pure. It is also possible that such mixed cells as 
result at one end of a tripolar spindle will introduce yet another 
factor for variation though tripolar spindles are too few in most 
fertile hybrids for this element to enter to any great extent. 
It is probable that the irregular distributions of chromatin, 
where such occur, have more to do with such succeeding off¬ 
spring as show variation, and less with those which return to 
the specific types. In the latter case the chromatin of each 
species has frequently remained entirely distinct (in marhed 
hybrids), or has normally separated again at the sundering of 
the bivalent chromosomes (in mild crosses) into the two orig¬ 
inal plasmas. Thus it is very obvious that most of the 
variation seen in the offspring of fertile hybrids is due to the 
union again of two “pure” germ-cells, each of which repre¬ 
sents a different one of the original parent species, but it is also 
