STUDIES OF THE DEVELOPMENT OF PEPEROMIA HISPIDULA 389 
nation of the seed the endosperm, with the enclosed embryo, bursts 
out df the seed-coat and fruit and finally is itself ruptured by the rad- 
icle of the growing embryo. The tips of the cotyledons remain for a 
long time embedded in the cap of endosperm, which persists between 
them and the perisperm. From this position of the endosperm it must 
evidently serve as a conveyor of the stored starch from the perisperm 
to the cotyledons. This location of the stored starch in the sporangial 
tissue is by no means a very primitive feature among seed plants. It 
is found in no pteridophyte and in no gymnosperm, except the bare 
remnant of nucellus left in the seed of Torreya. Among angiosperms, 
both dicotyledons and monocotyledons, perisperm occurs only in 
forms that on other grounds have been considered decidedly specialized 
(Johnson, 1902, p. 337). 
The persistence of the endosperm as a nurse for the embryo during 
its development and germination is characteristic of all the Piperales 
thus far studied and also of certain perisperm-containing monocoty- 
ledons (Johnson, 19006, 1902). In the case of the orchids and the 
Podostemaceae, as described by Magnus (191 3), endosperm is wanting 
and thus, as Magnus points out, the nutritive function of the female 
gametophyte is completely lost. In these two families this gameto- 
phyte serves simply to mature a functional egg apparatus. Magnus 
suggests that this disappearance of the endosperm allows a more rapid 
transfer of material to the embryo directly from the parent sporophyte, 
which is aided by a haustorium of the suspensor. It is not evident to 
the writer why this should allow more rapid nutrition of the embryo, 
nor is it proven that the embryos of these forms actually are developed 
more rapidly, in proportion to their size, than those of forms with 
endosperm. In the series of seed plants showing different degrees of 
development and functioning of the diploid generation, beginning with 
the gymnosperms and Ricinus and ending with the orchids and Podo- 
stemaceae, it is evident that Saururus, Peperomia and Canna show the 
next to the last step in the reduction of the endosperm, both in relative 
bulk and in variety of functions. Magnus suggests that such forms 
without endosperm, the Podostemaceae for example, may become the 
progenitors of entirely new races of seed plants. The haploid gener- 
ation of these hypothetical plants might, we can imagine, come to con- 
sist merely of •one of the four (or possibly three) haploid nuclei re- 
sulting from the reduction division of the nucleus of the megaspore 
mother-celL It is also quite possible to conceive that one of these 
