1908] BROWN—PEPEROMIA 453 
tion of evanescent walls in the first and second divisions seem to indi- 
cate that the sac is composed of the descendants of the nuclei of four 
megaspores and that the primary embryo sac nucleus is a mother cell 
and not a megaspore nucleus. 
If the walls corresponded to those of prothallial cells, we should 
expect to find them in the third division, but here not even a cell plate 
was seen. Besides this, the nearest phylogenetic relatives in which 
the first divisions of a megaspore result in a cellular structure are 
found among the leptosporangiate Filicales, where the heterospory 
is supposed to be of rather late origin, and it does not seem probable 
that Peperomia has reverted to the characters of an ancestor as remote 
as one in which we would find the first divisions of the megaspore 
giving rise to a cellular structure. 
This position is strengthened when we consider the four-nucleate 
stage of the Mexican species. Here the nucleus which is cut off is 
considerably larger and surrounded by much denser protoplasm than 
the other three. The resemblance to the four megaspores of the 
ordinary angiosperm is quite striking. 
The presence of the extra nuclei in the mature sac is in harmony 
with the view that these nuclei are the descendants of four megaspore 
nuclei. 
The nuclei of the four-nucleate sac of Peperomia have the same 
position with reference to each other as the nuclei of a tetrad of fern 
spores or of a tetrad of microspores of a spermatophyte. This posi- 
tion is not always apparent when the nuclei are dividing, and as the 
sac is somewhat rounded it may be that this arrangement is a mechan- 
ical response to the physiological conditions. 
That four potential megaspore nuclei may be included in a single 
cell has been shown by CANNON (’00) for Avena fatua. Here the four 
megaspore nuclei may or may not be separated by cell walls, but in 
either case three degenerate and the other forms the embryo sac. 
A similar phenomenon is reported by SmrrH (’98) for Eichhornia. 
The case of Crucianella (LLoyp 702) is interesting in this con- 
nection. Here the four megaspores are not separated by walls, but 
indications of plates are found on the spindles in the divisions of the 
megaspore mother cell nucleus. According to Lioyp, the four 
Megaspores are physiologically and morphologically similar. Each 
