39 
viduals of these two forms, the pollen must be taken either from the 
staminate tree to the hermaphrodite flower or from the hermaphro- 
dite flower to the pistillate tree. That is to say, the seventh or eighth 
type of pollination must be employed. There is objection to the 
seventh on the ground that it employs a male parent, many of whose 
characters are unknown, because it bears no fruit, and further because 
the hermaphroditic character would probably be lost. In this latter 
particular the experiments have not been carried far enough to 
determine the sex characters of the offspring of such a cross for the 
reason that the first attempts to impregnate the ovary of the herma- 
phrodite flower with pollen from the male tree resulted in repeated 
failures, for reasons which have been referred to (see p. 38). Since 
the hermaphrodite is apparently a modified form of the male, or 
vice versa, it would not be surprising to find that sex results in this 
cross would correspond rather closely with those which Shull has 
found in Lychnis dioca. Should that prove to be true, no herma- 
phrodites would result from such a cross, unless it be as an occasional 
mutant. 
If this be the case there would remain the other alternative, viz, 
to use the hermaphrodite flower as the male parent, applying its 
pollen to a pistillate tree of known characters. No experiments of 
this kind have been completed in which a pistillate tree of purely 
dioecious origin has been used. A similar experiment in which a 
pistillate tree from hermaphrodite stock was used as the female 
parent is interesting. It is possible that results would not be differ- 
ent had the pistillate tree been of pure dioecious stock. Seeds were 
collected from a chance fruit of unknown origin and planted as No. 
2087. The fruit was chosen purely for its flavor and other pleasing 
qualities, and not with a view to the specific experiments which later 
developed. Seventeen trees were planted out when too young to 
exhibit sex characters. These developed into 12 pistillate trees, 4 
hermaphrodite, and 1 staminate. When they came into bearing No. 
20S7 :3 9 was crossed with No. 2087 : 17 $ . The F t from this cross was 
planted under accession No. 2491. At the time the trees were exam- 
ined there were 5 pure pistillate individuals, 9 with more or less 
tendency to hermaphroditic characters, and 1 staminate. The latter 
produced only staminate flowers. Of the 9, there was one normal 
hermaphrodite of the elongata form, and another the same except in 
the case of two fruits. The remainder exhibited flowers and fruits 
of varied form, some with the stamens reduced to five in number, and 
producing fruit of the pentandria form. In others certain of the 
carpels failed to develop in the fruit, due to defective stigmas, thus 
giving rise to gibbous fruits. The stigmas and the anthers were both 
inclined to irregularity of position, anthers being found on the carpels. 
