370 Gates. 
variations are more numerous than the minus variations, while in 
O. rubrinervis the reverse is the case. This race breeds true to this 
condition, thus showing a constant inherited tendency to slightly 
greater anthocyan production than in O. rubrinervis. 
3. The range of variation in O. rubrinervis forms an absolutely 
continuous series and I have found no evidence of inheritance of 
different degrees of pigmentation within this series. An extreme plus 
variant, however, which I have called O. rubricalyx, has a greatly 
increased capacity for pigment production, as shown by its red hypan- 
thium and red on the ventral surface of the rosette leaf petioles, as 
well as in other parts of the plant. There are no intermediates con- 
necting this condition with the extreme of pigment production in 
O. rubrinervis, so that, when in flower, the two forms can always be 
distinguished at a glance. O. rubricalyx fluctuates about a new mode 
in regard to anthocyan production, except for certain individuals 
(about 25% in some cases) which revert to O. rubrinervis. 
4. On account of these reversions in O. rubricalyx, which happen 
in the first and in all later generations, its origin cannot be attributed 
to the loss of a “factor” or an inhibitor or other substance from the 
germ plasm. The change has been a positive one just as it appears 
to be. The Mendelian presence-absence hypothesis, commonly used 
to explain the numerous cases of Mendelian color inheritance in plants 
and animals, will not apply here. The difference between O. rubricalyx 
and O. rubrinervis is instead a purely quantitative one, O. rubri- 
calyx having originated through a quantitative readjustment of the 
materials of the germ plasm leading to the formation of the substances 
which determine anthocyan formation as a product of the plant’s 
metabolism. This hypothesis is rendered necessary by the fact that 
these quantitative differences in capacity for anthocyan production 
are strictly inherited, notwithstanding the well-known fact that this 
character is subject to wide fluctuations owing to environmental 
conditions. It is probable that many cases of Mendelian color in- 
heritance are to be accounted for as the result of similar heritable 
quantitative differences, rather than by the hypothesis of the presence 
or absence of certain factors in the organisms. This point of view 
was first expressed in another publication (GATES, 1910). 
5. These studies in quantitative variability are in accord with 
JOHANNSEN’S conception of phaenotypes and genotypes or pure lines, 
6. A series of crosses between O. rubricalyx and other forms 
indicate that its behavior is similar to that of O. rubrinervis. Thus 
