574 THE AMERICAN NATURALIST. [Vor. XXXIX. 
This remarkable proportion of forms derived from the hybrids 
AB, i. e, A, AB, and B in the ratio 1:2: 1 can only be explained 
on the assumption that the germ cells of the hybrid are pure 
with respect to the characters of either one or the other of the 
parents. The gametes from the hybrid, with the pure charac- 
ters of either A or B and approximately equal in number, may 
unite with one another in three possible combinations AA, AB, 
or BB forming three types of offspring, one pure A, another 
mixed AB, and the last pure B. By the law of chance the pro- 
portions of these combinations (AA, AB, ànd BB) in a simple 
case will be in the ratio 1:2:1. This assumption of the purity 
of the germ cells of hybrids has been found to conform with the 
facts ina number of simple experiments where two characters 
such as A and B were sharply contrasted. When one of the 
characters in the hybrid is dominant and the other recessive the 
ratio can be expressed as D: DR: R as 1:2: 1 which is merely 
a substitution of D and R for the characters A and B. 
There are of course many factors which tend to modify the 
ratios as stated above and complicate the results. Thus the 
normal number of gametes may be of varying vigor and mortal- 
ity so that there will be proportionately more or less of one type 
of fusion than is called for by the law of chance. Sometimes 
the characters of the parents remain evenly balanced in the 
hybrid and refuse to split up in the succeeding generations, 
remaining in a stable union in the germ cells produced by the 
hybrid. Such conditions prove exceptions both to the law of 
dominance and to that of purity of the germ cells. From these 
exceptions and particularly the last it is difficult to believe that 
any large proportion of the germ cells is absolutely pure, i. e., 
bearing only the pure characters of one parent or the other. 
However, there is much evidence from our knowledge of the 
distribution of the chromosomes from one generation to the next, 
that certain relations are possible in the separation of germ plasm 
which approximate the ratios of Mendel's law and while rarely 
giving absolutely pure germ cells nevertheless do make possible 
a large proportion of re/atively pure cells. 
Let us examine now the chromosome history as a possible 
physical basis for the Mendelian principles. Such considerations 
