154 BOTANICAL GAZETTE | MARCH 
gestion of such an occurrence is shown in fg. 27, and even here 
the usual cell has two nuclei. This preparation looks as if two 
potential macrospores might be developing one above the other. 
GERMINATION OF THE MACROSPORE. 
A typical nucellus just before the division of the primary 
nucleus of the fertile macrospore is represented in fig. 74. 
There is a tier of three tapetal cells and one potential macro- 
spore, the latter already somewhat crowded by the growing fer- 
tile macrospore. The nucleus of the fertile macrospore is 
accompanied by the structures known as centrosomes. No 
attempt was made to investigate these bodies, but they were 
noticed in two other preparations. In fg. 77, which shows a 
portion of a nucellus, there is a tier of four tapetal cells and two 
potential macrospores. The first division of the primary nucleus 
of the fertile macrospore was observed in about forty cases. 
The most frequent appearance is that shown in fig. 79, which 
has a central strand of protoplasm traversing a vacuole and con- 
necting the daughter nuclei. The vacuole may be absent, as 
in fig. 18. The spindle in the first division of the primary 
nucleus is parallel with the long axis of the macrospore, the 
only exception observed being that shown in fig. 2z. In the 
second division several mitotic figures were found. The spindles 
at the micropylar end were always transverse to the long axis 
of the macrospore, while those of the antipodal end were 
always longitudinal. fig. 20 might fairly represent all the cases 
examined. In fig. 76 the nuclei have a slightly different posi- 
tion, but it must be remembered that the nuclei in a germinat- 
ing macrospore gradually change their position. A peculiar four- 
celled stage is shown in fig. 22, where the position of the nuclei 
and the size of the nucleus at the micropylar end would seem 
to indicate that after the first division the micropylar nucleus 
had failed to divide, while the nucleus at the antipodal end had 
divided and one of the resulting nuclei had divided again. The 
next division, giving four nuclei at each end of the sac, was 
observed in only two instances, neither of which was very satis- 
