338 BOTANICAL GAZETTE [NOVEMBER 
The first four of these families were produced by self-fertilizing 
4 individuals of pedigree number o8115, and the rest by self- 
fertilizing 6 individuals of number o8119. Most of these families 
were too small to show obvious differences in the genetic compo- 
sition of the different parent plants, or between them and the 
hermaphrodite mutants tested | 
2 ae under case II. The small size 
of the families is due to the 
comparatively poor develop- 
ment of the ovaries and stig- 
(Mutant) O Xsut mas in most hermaphrodites, 
and the consequent difficulty 
| of securing large quantities 
of seeds by self-fertilization. 
OXset Most of the attempts to self- 
fertilize the hermaphrodites 
resulted in failure, and only 
in a small proportion were any 
2 oO o seeds produced. The total 
(Mutant) result agrees with results 
ie a 3 secured from the observation 
Pi® 3-— Model pedigree for case 111 of larger families, and it is 
fair to assume that the rela- 
tively large differences shown by some of. these families are not 
significant because of the smallness of the progenies. This con- 
clusion will be fully justified I believe, when it is observed under 
case IV that the very same plants, which produced the somewhat 
variable progenies shown above, gave uniform results when they 
were crossed with an unrelated female. 
CASE IV 
WHEN HERMAPHRODITES FROM THE PROGENY OF A SELF-FERTILIZED HERMAPHRODITE 
ANT ARE CROSSED WITH AN UNRELATED FEMALE 
The families 09133 to 09142, inclusive, resulted from pollinat- 
ing different flowers of a single female, 08114(4), with the pollen from 
to different hermaphrodites taken consecutively in family 08115; 
and the remaining 19 families were produced by pollinating the 
same female, 08114(4), with pollen from rg different hermaphrodites 
