tgrr] BROWN & SHARP—EPIPACTIS : 447 
from four megaspores. In Peperomia hispidula (JOHNSON '7) and 
in Gunnera (ERNST 5) the embryo sac is rounded at the four-nucleate 
stage and there is no polarity, but as development proceeds the 
sac elongates and polarity is produced. In Strelitzia (BROWN 3) 
there are four megaspores, each of which may germinate, but the 
three micropylar ones degenerate and the sac is always formed from 
the chalazal one. The three micropylar megaspores are not elon- 
gated and their nuclei do not show a polar arrangement. 
At the second division of the embryo sac of Epipactis the spindles 
are arranged so that the daughter nuclei are again evenly distributed 
in the cytoplasm. 
At the third division the spindles in both ends are usually 
arranged approximately at right angles to each other. This is 
of course usually the case in the ends of embryo sacs and in other 
rounded masses of cytoplasm, and would seem to be the way in 
which the spindles and resulting nuclei would be most evenly dis- 
tributed. In Epipactis, however, the chalazal end is sometimes 
narrow, and in this case the two spindles lie side by side. The 
simultaneous division of the nuclei and the production of an equal 
number in each end is probably connected with the similar condi- 
tions in the two ends. The number of nuclei is very likely due to 
some kern-plasma relation. In later stages the similarity of the 
two ends is destroyed and the nuclei take on quite different appear- 
ances. In Epipactis there is sometimes less cytoplasm in the chala- 
zal than in the micropylar end, and this is connected with a delay in 
the divisions in the chalazal end. 
STRASBURGER (16) has pointed out that the walls produced at 
the last division in a normal eight-nucleate sac are formed on the 
fibers connecting the nuclei, and that since one nucleus at each end 
is nearer the center than the other three, no wall is formed around 
it, thus leaving it free in the cytoplasm. He ascribes the fusion 
of these two polar nuclei to the fact that they have ceased develop- 
ing and are in the same cell cavity. Evidence strengthening this 
Position has been constantly accumulating and, as previously 
Pointed out (BRown 1), is quite striking in the case of the sixteen- 
nucleate sacs, where all of thesnuclei not cut off by walls fuse to 
form the endosperm nucleus. In Epipactis the polar nuclei are 
