442 BOTANICAL GAZETTE [DECEMBER 
walls may become quite distinct (fig. 19), though they vary some- 
what in position owing to the various planes in which the spindles 
may lie. In the figure shown they are transverse to the longitudinal 
axis of the sac, so that the four nuclei have a linear arrangement. 
Usually, though not always, these walls disappear very soon. In 
the event of their complete disappearance, there results a four- 
nucleate sac like that represented in fig. 20, which is essentially the 
same as one derived from a single megaspore, but is conspicuously 
larger. That the four nuclei here shown have resulted from fewer 
divisions from the nucleus of the archesporial cell than have those 
in four-nucleate sacs derived from one megaspore seems to be 
indicated by their relatively larger size (cf. figs. 7 and 20). A 
similar condition was pointed out in connection with the two- 
nucleate stage (cf. figs. 6 and 18). 
t a stage as late as the four-nucleate sac it becomes very 
difficult to determine whether degenerating cells at the micro- 
pylar end of the sac are present or not, so that it is unsafe to depend . 
too strongly upon them as a criterion, but after the examination 
of a large number of cases the present writers hold the view that 
the four nuclei, which, on account of their origin and the appear- 
ance of walls at the mitoses which give rise to them, are megaspore 
nuclei, and that these by one further division give rise to an eight- 
nucleate sac entirely similar to one derived from a single megaspore- 
Important evidence in this connection is afforded by the wall 
formed between the two chalazal (megaspore) nuclei, which often 
shows a tendency to persist. In fig. 21 it is still visible as a remnant 
during the division to form the eight nuclei of the sac, the four 
spindles in this case showing an unusual irregularity in distribution. 
Since one of the micropylar spindles was in an adjacent section, it 
was not possible to demonstrate the presence of a wall in that end. 
In other cases the wall persists for a longer time, giving rise to the 
condition shown in fig. 22. Here it is observed separating the two 
undivided chalazal nuclei, while at the micropylar end the next 
division has taken place, cell plates being evident on the spindle 
fibers. A somewhat later stage is represented in fig. 23- The 
persistence of the wall seems to result in a delay of the nuclear 
divisions (fig. 21), or in their suppression, as shown by figs. 22 and 23- 
