122 Self-Incompatibility in Hermaphrodite Plants 
was determined by controlled self-pollinations involving over 30,000 
flower heads and about 600,000 individual flowers, show conclusively 
that self-compatible plants occur sporadically among the progeny of self- 
incompatible plants, and that self-incompatible plants continue to appear 
among the offspring of self-compatible plants. Either self-compatibility 
or self-incompatibility can arise from the other in a line of inbred or 
selfed progeny; in this sense the two conditions are reversible. Since 
this is the case, dominance or recessiveness of these characters cannot be 
adequately determined.. 
None the less there is a tendency to heritability in these characters. 
Self-fertile plants appear to constitute from 0 to 10°/, of the progeny of 
self-sterile parents. The proportion of such plants immediately increases 
in the J,; in this generation of the red-leaved Treviso numbering 
351 plants (Table IV), 30°/, of the plants were self-fertile to some 
degree. In other families the percentage was higher (Table VIII, 1918). 
The offspring of self-fertile plants are more likely to be self-fertile by a 
proportion of about 5 to 1. This proportion can be changed only slightly 
by the selection of parents of different degrees of self-fertility. 
Certain lines and families appear to maintain somewhat different 
grades of self-fertility. This is most marked in respect to the range of 
individual self-fertility. All families agree in general behaviour as to 
regression, byt some are more highly self-fertile than others. The 
character of self-compatibility is one of incomplete heredity; a self- 
fertilized line of descent does not breed true; in pedigreed lines of 
descent the characters self-fertility and self-sterility are reversible. 
From the standpoint of a factorial description of the results obtained 
in chicory, several points are of significance. The spontaneous occul- 
rence of self-compatible plants after several generations of self-sterile 
parentage suggests the phenomenon of mutation referable to single 
factors or to the recombination of modifying multiples. But submitted 
to the test of self-breeding the character of self-fertility does not breed 
true, hence any particular factors or combinations that may be assumed 
are not stable. The frequency distribution for fertilities suggests 
variation that is often interpreted in terms of multiple factors of quanti- 
tative and modifying values, and this is also suggested by the evidence 
that certain families exhibit somewhat different degrees of self-compati- 
bility. Yet in all families there is marked regression to self-sterility: 
Perhaps the most significant fact from the standpoint of hereditary 
analysis is that the proportion of self-fertile to self-sterile plants in the 
progenies of self-fertile plants seems to fluctuate about a 1; 1 ratio. This 
