384 
DUNCAN S. JOHNSON 
perhaps, on the form of the ovule (see Johnson, 1900, p. 2; Brown, 
1908, p. 453; Stephens, 19096, p. 383; Brown and Sharp, 191 1, p. 446; 
Fisher, 1914, pp. 421-430). That the form of the ovule and embryo 
sac and the arrangement of the contents of the latter may be related 
to each other and may vary considerably in the same plant is shown by 
this Peperomia itself. Thus when the 4-nucleate embryo sac is about 
isodimensional the nuclei are arranged at the corners of the tetra- 
hedron, while in other cases where the embryo sac is twice as long as 
broad the nuclei are in 2 subterminal pairs (fig. 72.) Of the external 
and internal causes determining the form of the ovule and the asso- 
ciated form and arrangement of the embryo sac we know almost 
nothing (Brown and Sharp, 191 1, p. 446). 
As for the phylogenetic origin of this tetrad arrangement of the 
megaspores of P. hispidula it seems clear that it is a peculiarity de- 
veloped within the genus, for the allied and probably more primitive 
genera Piper, Heckeria, Saururus, etc., have the megaspores or megas- 
pore nuclei arranged in a row (see Johnson, 1900&, 1910; Fisher, 1914, 
p. 156). No adequate evidence has been offered to show that this 
megaspore tetrad has been derived directly from a more primitive 
ancestor of which this arrangement was typical, as it is of the mega- 
spores of the higher pteridophytes. The same can be said of those 
other atypical angiosperms, having tetrahedrally arranged megaspore 
nuclei not separated by walls, such as Gunnera (Schnegg, 1902; Ernst, 
1908; Samuels, 1912), the Penaeaceae (Stephens, 1909a) and certain 
Euphorbias (Modilewski, 1909). In none of these has the tetrahedral 
arrangement of the megaspore nuclei been shown to be primitive, and 
in each case this arrangement has been found within a rather rounded 
embryo sac. 
4. Embryo Sac. — Under this head will be considered the disap- 
pearance of the megaspore walls, the further division of the megaspore 
nuclei, the formation of the egg apparatus, of the endosperm nucleus, 
and of the occasional peripheral cells. 
The separating walls of the megaspores are delicate and do not 
persist long after their formation from the cell plates of the second 
division. As they disappear the cytoplasm again becomes continuous, 
and a large central vacuole replaces the series of numerous small periph- 
eral vacuoles, characteristic of the earlier phases of development. 
This single continuous mass of protoplasm with its 4 nuclei must un- 
doubtedly be regarded as the product of fusion of 4 megaspores, and 
