STUDIES OF THE DEVELOPMENT OF PEPEROMIA HISPIDULA 385 
the single individual embryo sac thus formed as a composite female 
gametophyte. This sort of composite young gametophyte has been 
observed in Avena fatua (Cannon, 1900) and in Crucianella (Lloyd, 
1902), with the difference, however, that in these forms 3 of the mega- 
spore nuclei degenerate, sometimes after one division each, while 
the fourth gives rise to all of the 8 nuclei of the mature embryo 
sac. Lloyd pointed out the significance of this association, of the first 
haploid nuclei from a megaspore mother cell, in one cell and spoke 
of this cell as ''an individual by coalescence?" Coulter (1908) applied 
this same criterion, the chromosome number, to the interpretation 
of all embryo sacs in which the first nuclei produced in the ovule by 
the reduction division are included, either primarily or secondarily 
in a single protoplast. According to this view the embryo sacs not 
only of Peperomia, Gunner a and the Penaeaceae, but also those of 
Lilium, Piper and many others where the "megaspore mother cell 
develops directly into the embryo sac, " are to be regarded as composite 
sacs. The demonstration of a reduction process and of evanescent 
megaspore walls in several species of Peperomia by Brown (1908) 
and Fisher (1914) and the especially clear case of tetrad formation 
in P. hispidula show that this genus furnishes several of the 
clearest instances of the formation of a composite mature embryo 
sac. The possibility of other interpretations of the observed phe- 
nomena which imply that chromosome reduction may occur at points 
in the life history, other than the divisions giving rise to the megaspore 
nuclei, have been suggested by Atkinson (1901), Brown (1908) and 
Ernst (1908, p. 27). The study of P. hispidula, however, together 
with that of other species by Brown and Fisher make it very clear 
that, in this genus, a genuine megaspore-formation is associated with 
the reduction division, and make it altogether probable that the 
same is true in Piper, Lilium, and in many other similar cases. 
The 2 further divisions of each of the 4 megaspore nuclei are not 
followed by cell walls, nor even by cell plates. These facts suggest 
the homology of these divisions to the first steps of megaspore germi- 
nation in ordinary angiosperms (see Brown, 1908, p. 453). Though 
it must be recalled that cell plates are formed in the first division of 
the megaspore nucleus of Lilium, Clintonia and perhaps others (R. W. 
Smith, 191 1, p. 216). The 16 nuclei resulting from the 2 divisions 
of the megaspores are usually grouped in distinct quartets, one of 
them at the micropylar end of the embryo sac. Two of the nuclei 
