232 BOTANICAL GAZETTE [OCTOBER 
sufficiently complete to enable one to judge quite accurately as 
to the real nature of the centrosome-like body which occurs in 
the generative cells of these plants. In Ginkgo the centrosome- 
like body cannot be detected in the generative cell until two 
months after the germination of the pollen grain, when the gene- 
rative cell has greatly increased in size. It then appears asa 
very small round body in the cytoplasm between the nuclear 
membrane and the cell wall. It gradually increases in size, and 
a month later is found to have grown to three or four times its 
original diameter. It is thus evident that in Ginkgo these bodies 
originate in the cytoplasm of the generative cell. The early 
stages of development of the generative cell of Zamia have not 
been studied, and, therefore, in this case their first appearance 
has not been observed. It has, however, been determined that 
they increase in size here as in Ginkgo. In Zamia the centro- 
some-like body, as described in my previous papers, finally rup- 
tures during the division of the generative cell, and the mem- 
brane formed by its wall becomes greatly extended in length, 
ultimately forming a narrow band arranged in the form of a heli- 
coid spiral on one side of the cell. This band gives rise to the 
cilia which form the motile organs of the antherozoid. In fecun- 
dation, as shown above, this ciliiferous band developed from the 
centrosome-like body is left in the cytoplasm at the apex of the 
archegonium, while the nucleus wanders alone from that point 
to the oosphere. The ciliiferous band surely remains intact at the 
apex of the archegonium for some considerable time after fecun- 
dation, and is then gradually absorbed as the embryo develops. In 
the divisions of the oosphere immediately following fecundation, 
furthermore, no indication of any body resembling a centrosome 
could be found. Hirase, who studied the division of the oosphere 
of Ginkgo, was also unable to find any trace of a centrosome. 
It thus seems quite evident from the above facts that the bodies 
in question cannot be considered as true centrosomes. The 
mere fact that during a part of their existence they are situated 
approximately where a centrosome might be expected to occur 
if present, and for a considerable time during the resting condi- 
