440 BOTANICAL GAZETTE [DECEMBER 
the small vacuoles in the central region coalesce to form a large 
central one (fig. 6), the cytoplasm becoming spread out as a parietal 
layer which is thickest at the ends of the sac where the radius of 
curvature is small. The two nuclei lie in these two regions. These 
nuclei soon divide simultaneously, giving rise to the four-nucleate 
sac (fig. 7). In this division and in the preceding one no traces 
of cell plates could be distinguished on the spindle fibers. After 
considerable enlargement of the embryo sac, the third division 
occurs, the two spindles formed in each end varying in position. 
Usually they lie approximately at right angles to each other; in 
such cases they may be equidistant from the end of the sac (fig. 8), 
or one may lie at a little distance from the other along the lateral 
wall. Cell plates appear on the fibers of all four spindles, so that 
the resulting eight nuclei are separated in the usual manner. In 
the micropylar end the transverse spindle gives rise to the two syner- 
gids, while the longitudinal one forms the egg and polar nucleus. 
In the chalazal end three antipodal cells and a polar nucleus are 
formed in an exactly similar manner (fig. 9). The egg and synergids 
increase in size, and the two polar nuclei approach each other and 
fuse (fig. 10). 
Exceptionally the two spindles in the chalazal end of the sac, 
instead of lying at right angles, come to lie more or less parallel to 
each other and usually to the longitudinal axis of the sac (fig. 11). 
As division proceeds they may become coalesced, forming one 
large spindle instead of two ordinary ones (figs. 12-14). The con- 
clusion that such is the explanation of the large spindle shown in fig. 
12 is supported by the fact that the plate of chromosomes could be 
seen by focusing to be made up of two groups of approximately 
equal size, and that altogether their number was plainly larger 
than that in either of the micropylar spindles. The same was true 
of the corresponding spindle of fig. 13. The division of this double 
spindle may keep pace with that of the micropylar ones (fig. 12); 
or it may be delayed as shown by figs. 13 and 14. In fig. 14 the 
wall on the fibers of the chalazal spindle is only slightly younset 
than those of the micropylar ones, showing but a slight delay. 
In fig. 13 the delay has been greater, the chromosomes of the chala- 
zal spindle having very recently separated, while in the micro- 
