368 Gates. 
increased or diminished amount of anthocyan production which is 
inherited, the materials of the germ plasm which determine the 
production of the substances above-mentioned, must have undergone 
some quantitative readjustment, which caunot at present be further 
analyzed. 
Even white varieties, which are frequently supposed to have lost 
a factor for anthocyan production from their germ plasm, frequently 
show that such an hypothesis is too drastic by producing a few flowers 
or a few petals which are pigmented. I have observed this to be the 
case in white varieties of Chrysanthemums, and many other instances 
are on record. A definite factor lost from a race is irrecoverable, and 
a white race in which this has occurred should on this conception 
never be able to produce any anthocyan. But if this pigment some- 
times appears then the white race was not due to a factor irrevocably 
lost, but as already suggested, to such a quantitative readjustment of 
the substances of the germ plasm, that the processes leading to an- 
thocyan formation are ordinarily diverted into other channels. In the 
same way, many other cases of Mendelian color inheritance may be 
explained on a quantitative biochemical basis. This interpretation, 
while assuming, in accordance with the evidence, that quantitative 
differences may be inherited (a matter which requires further analysis 
for a full explanation) is yet sufficiently elastic to permit an explanation 
of the occasional reversions and variations which are not readily ex- 
plainable on the basis of a definite factor hypothesis. 
These studies on pigment variability are evidently in complete 
harmony with JOHANNSEN’S concept of pure lines, or genotypes and 
phaenotypes (1909). They show that differences in degree of pigmen- 
tation may or may not be strictly inherited, and that the only way 
to determine whether two plants belong to the same genotype is by a 
study and comparison of their offspring grown under similar conditions. 
The relation between the mutations in Oenothera and the pure lines 
of JOHANNSEN and JENNINGS (1910), will not be considered here. 
The series of crosses made with O. rubricalyx indicate that its 
hereditary behavior is the same as that of O. rubrinervis and other 
mutants from O. Lamarckiana, and the results of a given cross can 
therefore be correctly predicted. One important point which is not 
yet decided is whether O. rubrinervis =< O.rubricalyx will give dominance 
to O. rubricalyx in the F,, or whether there will be alternative inheri- 
tance, as in O. rubricalyx >< O. Lamarckiana. From the other behavior 
of O. rubricalyx the latter is the more probable result, in which case 
