DICELLANDRA AND PILEONEURON. 487 
confined to the seeds and fruits, but extend to the early stages 
of the flowers, although they are not conspicuous in their outward 
appearance, and have therefore been overlooked. How far they, or 
some of them, are determined by the ultimate development of the 
fruit, so different in both groups, it is impossible to say in our pre- 
sent imperfect knowledge of the case and with the little material 
at our disposal; but it is characteristic that they affect just those 
parts which go to build up the fruit, whilst the petals and the 
stamens which play a more transient, though not less important, 
róle in the reproductive phase of the life of these plants, are, 
group compared with group, either alike (the petals) or exhibit 
differences (the androcium) which traverse the parallelism so 
evident in the development of the fruits of the Pheoneuron type. 
The differences in the flowers, which coincide so remarkably with 
those in the fruits, concern the ovary and the calyx which takes 
part in the formation of the fruit. The ground-plan of the ovary 
is the same in both groups, and agrees with that usually found in 
the 5-merous Melastomace: ; and the ovary itself is, in both, com- 
pletely surrounded and exceeded by the calyx-tube, which bears 
the petals and stamens on a circular rim situated somewhat 
above the top of the ovary. This is, in fact, a character common 
to all Melastomacex. But whilst the ovary of Pheoneuron is 
free with the exception of the lower third or fourth part, where 
it is completely connate with the calyx-tube (Pl. 19. fig. 19), it 
is in Dicellandra joined with the calyx-tube by narrow radial 
lamellae which run from the very top towards the base, in such 
a way that they divide the space between the ovary and the 
calyx-tube into ten pockets for the reception of the anthers in 
the bud (PI. 19. fig. 1). The pockets are alternately shorter and 
longer. Those which contain the longer anthers of the outer 
staminal whorl reach nearly to the base, the others about to the 
middle. I need not remind the reader that this is a very common 
arrangement in Melastomacem. . In Pheoneuron there are no 
such lamelle and no pockets. The ovary is quite free for about 
or ł of its length. Another difference concerns the top of the 
ovary. This bears in Dicellandra a crenulated, somewhat fleshy 
ring which surrounds the base of the style and is fringed with 
minute gland-tipped hairs (Pl. 19. fig. 4). No trace of a similar 
appendage is noticeable in Pheoneuron. It is probable that 
these differences are in correlation with the differences in 
the mode of insect fertilization. As to the calyx itself, the 
