368 
DUNCAN S. JOHNSON 
1914, p. 143). It is quite possible that walls are not formed in all 
tetrads, though it is surely the rule. In figure 69, e. g., is shown a sac 
of large size in which the central vacuole has begun to form and in 
which the cell plates seem to be disappearing without giving rise to 
walls. From this first divergent step onward the development of 
this embryo sac differs greatly from that of other angiosperms, the 
chief peculiarity being that all the descendants of all 4 megaspore 
nuclei participate in forming the mature embryo sac. 
The next step toward the development of the ripe embryo sac is the 
formation of a mitotic spindle by each of the four free nuclei. These 
appear very soon after the disappearance of the megaspore walls and 
before any other change occurs in the sac except a slight growth in 
length, a separation of the 4 nuclei and the formation of a central 
vacuole (figs. 73, 74). The spindles are located in the peripheral 
portion of the cytoplasm and the axis usually lies parallel to the neigh- 
boring portion of the wall of the sac (figs. 75, 76). The axis of the 
spindle may have any position in relation to the axis of the sac itself. 
The eight nuclei resulting from this division were connected, in many 
of the cases seen, by the persistent fibers of the four spindles. No 
cell plates could be found, however, nor could evidence be obtained of 
the separation of these 8 nuclei by even temporary cell walls (figs. 76, 
77). This absence of plates, as has been pointed out by Brown (1908, 
p. 445), indicates that this is the equivalent of the free nuclear division 
found at the germination of the megaspore of most angiosperms. 
Smith (191 1, p. 216) finds however that a cell plate is formed at the 
first division of the embryo sac of Clintonia. The eight resting nuclei 
at this stage are each 8 to 10 ju in diameter. Each has a nucleolus 2 or 
3 in diameter and a wide-meshed, peripheral chromatin net not un- 
like that of the spore nuclei (fig. 77). During the division of the 
megaspore nuclei the sac increases in size and at the 8-nucleate stage 
has a width of 40 or 45 and a length of 60 jx or over. The micro- 
pylar end of the sac is narrowed to a pocket of less than half the 
diameter of the opposite end. In this micropylar pocket lie two 
nuclei derived from the nucleus of the micropylar spore of the tetrad, 
while just below the middle lie the three pairs of nuclei derived from 
the other three spores (fig. 77). The embryo sac at this stage has 
become more distinctly bipolar than in the four-nucleate stage, and it 
continues from this time on to the ripe seed to show this definite 
polarity. 
