454 BOTANICAL GAZETTE [JUNE 
elongated and presents a most singular structure. The nucleus 
of the original cell evidently divides into two, and one of the 
daughter nuclei forms within the unbroken Hawtschicht of the 
mother cell a new and wholly distinct Hautschicht which delimits 
a cell lying entirely free within the mother cell and surrounded 
on all sides by a layer of protoplasm of nearly uniform thickness 
(figs. ra and 2.) The other daughter nucleus remains free within 
the Hautschicht of the mother cell, but is pressed to one side by 
the interior cell. The free nucleus usually occupies the side of 
the mother cell opposite the pollen grain. Both the interior 
cell and the mother cell are crowded full of small spherical starch 
grains (fig 2). The second cell (fig. 7 gc) which is probably to 
be compared with the generative cell in conifers, is much larger 
then the first or proximal cell above described, and is provided 
with two small spherical organs, situated at the opposite ends of 
the nucleus outside the nuclear wall, that somewhat resemble 
centrosomes (fig. 3). They are exactly spherical and stain deep 
red with safranin, in the Flemming triple stain process, much as 
the nucleoli. There is no appreciable difference in size between 
the spheres in the same cell, but those of different cells frequently 
vary several micromillimeters in diameter, ranging from 7 to 10 #. 
Many of the spheres at this stage appear entirely homogeneous, 
but i in some cases a number of small vacuoles have been observed 
in the interior (fig. 4). Numerous long slender threads of kino- 
plasm radiate in all directions from the spheres, and in many 
instances may be seen plainly to extend to the Hautschicht with 
which they frequently appear to unite. The threads of kinoplasm 
are rather coarse and are plainly visible without staining. W ith 
: sie oe, triple stain they are colored intensely blue. They 
: in many instances to be firmly fastened to the Hautschicht, 
. which: is drawn in: more or less where the thread is attached ( fig. 5)- 
This seeming indentation may, however, be caused by shrinkage 
oe. daring fixing: and Suocames as Pseieg seaanieg are usually 7 toro # from: 
ee th in most instances strongly indented on 
a the ¢ ends next to the spheres. Threads of kinoplasm run from 
—- = to the lez ar Wall, | I ic h at this a is — a a 
