Studies on the Variability and Heritability of Pigmentation in Oenothera. 361 
In ıgro a total of 1166 plants were grown for these experiments, 
but since many of them remained rosettes in which O. rubrinervis and 
O. rubricalyx could not be distinguished with certainty, I shall not 
give here the numbers of individuals for each experiment, but shall 
only state probable results, which will require verification another year. 
In the first place, out of 12 plants in the F, from the original 
O. rubricalyx mutant (4 grown at Wood’s Hole and 8 at Chicago) one 
reverted to O. rubrinervis, showing that reversions occur in the F,. Two 
self-fertilized O. rubricalyx individuals of the F, both split in the F, 
into O. rubricalyx and O. rubrinervis in the ratio of about 3:1. Two 
of these O. rubrinervis reverts were selfed and produced in 1910 a 
total of 196 rosettes, all apparently O. rubrinervis. Only two plants, 
one from each culture, came into bloom, but they were both O. rubri- 
nervis. Hence it is reasonably certain that the O. rubrinervis reverts 
breed true, thus behaving like Mendelian extracted recessives. 
Similarly, three selfed O. rubricalyx individuals of the F, split in 
the F,. In the case of four other O. rubricalyx plants of the F,, the 
offspring all remained rosettes, so that it could not be determined 
with certainty whether any O. rubrinervis was present. But, from the 
evidence thus far, it seems probable that O. rubricalyx always produces 
reverts to O. rubrinervis in every generation, and that no pure-breeding 
individuals of O. rubricalyx will be found. The fact that the O. rubri- 
calyx mutant split in the F,, and that, so far as they have been 
tested, all the O. rubricalyx offspring continue to split in every 
generation, is not strictly in harmony with the Mendelian conception 
of dominance and heterozygy, although the 3:1 ratio obtained in the 
F, may not be without significance. If the conception of DE VRIES 
as to the origin of a mutant be applied, then the O. rubricalyx mutant 
originated from a cross between a O. rubrinervis germ cell and a 
O. rubricalyx germ cell and the latter is dominant. This leaves the 
absence of pure dominants (i. e. pure-breeding O. rubricalyx) in later 
generations to be accounted for, but it is possible they may yet 
be found. 
On the other hand, DE VRIES has shown, and my own crosses 
have confirmed, that O. Lamarckiana when crossed with certain of 
its mutants, such as O. rubrinervis, gives alternative inheritance, both 
O. Lamarckiana and O. rubrinervis appearing in the F,, and both 
breeding true in later generations. It is therefore equally probable 
that O. rubricalyx behaves in the same manner when crossed with 
O. rubrinervis, giving both types in the F,. An experiment was made 
