(374) 
thera. This weak development of the septa in the ovary of 
fevillea is still more significant when we consider the condi- 
tion found in some related genera (Alsomztra, Zanonia) in 
which the ovary, composed of three carpels, is unilocular 
and open to the exterior at the apex. The development of 
placental tissue which finally fills the locules in Fevillea 
occurs quite late and never to that degree found in members 
of the Melothrieae and Cucurbiteae. 
Among the Sicyoideae, Coccénza represents the condition 
closely approximating that found in the preceding cases so far 
as the structure of the ovary is concerned (see text-figure 3). 
The carpels unite to form the partitions, and the placentae in 
the same manner as in the Cucurbiteae. In the case of JZ- 
crampelts, however, only two carpels are apparent, each of 
which produces, in our species, but two ovules. The margins 
of the carpels form two longitudinal elevations which meet at 
the center and become flattened against one another as in the 
preceding cases, but instead of a row of ovules upon each 
flank there appears but one, making four in the entire ovary. 
While no evidence of more than two carpels has been found 
in any of the young ovaries of Micrampelis examined, nor 
of more than one ovule on the placental fold in any case, yet 
it seems probable that the ovary of this form is derived by 
the reduction of three carpels to two. We find in fact that 
this form as well as others is subject to some variation, and 
ovaries are frequently found in the mature state evidently 
composed of three carpels, inasmuch as they contain three 
locules and six seeds. But the prevailing number of locules 
is two. We are reminded of the peculiar apron-like char- 
acter of the placenta of Micrampelis in the development of 
that organ in Luffa and others of the Cucurbiteae, in which 
there is a downward prolongation of the placenta toward the 
base of the ovary for a considerable distance beyond its in- 
sertion. If in the case of Lufa but a single ovule were to 
arise on each flank of this downward prolongation, we should 
have a condition quite similar to that noticeable in the early 
stages of Micrampelzs. Inasmuch as we find at no stage in 
