STUDIES OF THE DEVELOPMENT OF PEPEROMIA HISPIDULA 373 
ficial chromatin reticulum, and, before its first division may have as 
many as 25 or 30 nucleoH. The latter differ greatly in number, size 
and shape which differences are evidently due to the fragmentation 
noted above of the 14 nucleoli contributed by the fusing parent nuclei 
(figs. 92, 95). The resting period of this nucleus is apparently not a 
very long one. Relatively few examples were found where the con- 
stituent nuclei were completely fused and the product so formed was 
still undivided. There were many scores of examples of fusing nuclei 
to a few dozens of those undivided or in division. 
The first spindle organized in this fusion nucleus is a very large 
one, with rather broad ends and very numerous chromosomes, which 
are crowded together at the equator to form a plate often 15 or 20 )u in 
diameter (figs. 98, 99). The axis of this spindle is usually approxi- 
mately transverse to the ovule and the dividing wall which is at once 
formed on its equatorial plate is nearly longitudinal to the ovule 
(figs. 98, 99, 102, 105). This wall, however, may make any angle 
with the sagittal plane of the ovule itself. The number of chromo- 
somes formed on this spindle and on the later ones of the endosperm 
is far greater than in any other mitoses of this species. They are, 
however, so crowded that it was impossible to count them with cer- 
tainty. In all cases where the number could be made out approxi- 
mately it was found to be far above a hundred (figs. 98, 99, 106). 
This makes it altogether probable that the 144 or 196 chromosomes of 
the 14 contributing nuclei retain their individuality in the resulting 
endosperm nucleus. The form of the chromosomes is at first rod-like, 
rather elongated and sometimes bent. Later they become shortened 
and rounded (figs. 98, 99). The two new nuclei, which are promptly 
organized about the poles of the spindle, are but slightly smaller than 
the endosperm nucleus. They are usually decidedly flattened in the 
plane of the new cell wall which is immediately formed between them. 
They are also often more or less lobed in the manner of the primary 
endosperm nucleus (figs. 102, 103). In fact the wall of the daughter 
nuclei is often so very irregular at first that in sections it looks as if 
torn or incomplete (figs. 102, 105). The chromatin net of these nuclei 
resembles that of the primary endosperm nucleus, except that it is 
more nearly homogeneous, in size of threads and meshes. During the 
first mitosis the nucleoli disappear (figs. 98, 99), but they soon reappear 
in the daughter nuclei, in numbers ranging up to a dozen or more each 
(figs. 102, 106). As in the later phases of the primary nucleus so here 
