THE FACTORIAL HYPOTHESIS 267 
invariability of the factors involved. When, how- 
ever, these factors were sorted out so that strains 
became homozygous, some variability probably due 
to evironic differences still remained. That is, in 
addition to the variation due to recombination it 
has been found that even in pure races “unit char- 
acters”? vary.. Why, then, it may be asked, do not 
the factors that produce them vary also? 
Johannsen’s work on material of a kind suitable 
to give a definite answer to this question and by 
methods that have not been questioned, has brought 
out clearly certain facts. only vaguely stated before. 
In a population of beans he found that each bean 
gave rise by self-fertilization to what he called a pure 
line. Each of the original beans proved to be homo- 
zygous for all of the factors: involved. This was 
probably due to self-fertilization through many genera- 
tions, a process that automatically produces homo- 
zygous lines. The weights of the descendants of any 
given bean gave a curve of frequency which was 
different from that of the whole population (Fig. 62). 
Within the group derived from one bean, however, it 
was found that any bean, whether heavier or lighter 
than the others, gave a curve exactly like the curve 
of the line from which it came. Evidently then the 
size differences within these pure lines are not inherited. 
They must be due to the environment of the plant, or 
to the position of the bean in the pod, etc.; in other 
words to conditions that are extrinsic to the germ 
plasm. Here is a demonstration that the factors 
do not vary, but give identical results in successive 
