(370 ) 
ous. That these may be in some sense an accidental feature, 
due to some process in the preparation of the material, is pos- 
sible, though this was the only place where they were found, 
and the same methods were used in the preparation of the 
material of Cyclanthera that were used in other cases. No 
starch was observed in this or later stages. 
After fertilization the endosperm nuclei increase in size and 
multiply rapidly (jg. 765), and the mass as a whole expands 
in the hollow form. 
PART III. DISCUSSION AND CONCLUSIONS 
The facts set forth in regard to the organogeny may be 
summarized briefly as follows: The pistillate flower begins 
as a lateral outgrowth, the apex of which becomes depressed 
and then concave. ‘The sepals arise first around the margin 
of the torus and the petals follow on an inner circle, the mem- 
bers alternating with the outer series. Within the circle of 
petal-rudiments the carpels arise and become extended upward 
to form the style. The placenta arises as a longitudinal ridge 
upon which the ovules are borne laterally, except in the case 
of S%cyos, in which the ovule is pendent from the top of the 
ovary. 
Of those forms whose development has been studied and 
described above, it appears that Fevzllea is the most primi- 
tive. It is easy to see how all the others may have been de- 
rived from sucha form. Although, as Miller™ suggests, the 
ovary of the Cucurbitaceae may have consisted originally of 
five carpels, the simplicity of floral construction in Fevzllea, 
in both staminate and pistillate flowers, points to the possible 
course of evolution of the ovary of this family. With- 
out entering at present upon a discussion of the difficult 
problem of the morphology of the androecium, some facts 
have been observed in the ontogeny of the ovary which may 
be discussed in this connection. 
In the three tribes, Fevilleae, Melothrieae and Cucurbiteae, 
the prevailing type of ovary is the tricarpellary form, and 
