300 
AMERICAN JOURNAL OF BOTANY 
[Vol. 9, 
a region where there is room for four cells to lie side by side across the 
anther not more than three will be found. During presynapsis and the 
early stages of synapsis this large amount of intercellular space is filled with 
a colloidal matrix which disappears near the middle of the synaptic period 
and leaves the mother cells free within the anther during the remainder 
of the reduction divisions. Just after the disappearance of the colloidal 
matrix the mother-cell walls begin to thicken. Whether or not the material 
of the colloidal matrix contributes to the thickening of the wall, or whether 
the water released by dissociation of the intercellular colloid is subsequently 
adsorbed by the gelatinized cell wall of the mother cells, has not been deter- 
mined. It may be that there is no relation between the disappearance of 
the gel between the cells and the thickening of the cell wall. 
A peculiarity of the thickening of the walls of the mother cells of Nelumho 
lutea is the high degree of inequality in the thickening on the various sides 
of the same cell. It is usually found that the sides of the cells toward the 
ends of the anthers are much more thickened than those toward the lateral 
surfaces. By the time the stage of interkinesis between the heterotypic 
and homoeotypic divisions is reached, it is found in some cases that the 
thickening of the wall on each of the two ends is equal to at least one half 
of the diameter of the cell lumen, whereas the walls on the lateral sides are 
scarcely thickened at all. Another feature noted about the anther of this 
species is that the tapetum usually remains living and intact throughout 
the reduction process. It consists commonly of very large cells, in some 
cases almost as large as the pollen mother cells. They may contain more 
than one nucleus, and mitotic figures are of common occurrence within them. 
In Magnolia, as previously reported (7), the different stages of the 
reduction divisions may be found in the same anther, and there seemed to 
be no special arrangement or order of progression from one part of the 
anther to others. In Sisyrinchium (8), a very definite graded series exists 
from one end of the anther to the other. Either the distal or the proximal 
end of the anther may be more advanced, but there is a regular succession 
of stages, the whole range of stages in any anther not being very great. 
In Nelumho lutea an intermediate condition exists. There is a succession 
of stages from one part of the anther to another, but it is not always from 
one end of the anther to the other. Furthermore, the range of stages found 
within a given anther is frequently much larger than in Sisyrinchium. 
Cases were found in which the tetranucleate condition was found at one 
end and the microspore stage at the other, with a series showing quadri- 
partition in successive degrees of advancement between. In other cases 
diakinesis was found at one end of the pollen chamber and the tetranucleate 
condition at the other. In some cases both ends of the anther are more 
advanced than the middle. The middle portion may show interkinesis 
while the ends are in the tetranucleate stage; or the reverse maybe true, 
with tetranucleate cells in the middle and interkinesis or even diakinesis at 
