Allen . — The Spermatogenesis of Poly trichum juniper inum. 281 
zoids in section at a much later stage, when what remains of the androcyte 
cytoplasm has been reduced to a very thin layer. 
It is not easy to determine in just what order the cytoplasm in 
different parts of the androcyte disappears. It is possible that the 
vacuole so commonly present in the older androcytes (Figs. 46, 47), 
which seems to be somewhat larger in still older cells (Figs. 48-50), and 
which contains the granule that I have identified tentatively with the 
percnosome, indicates the beginning of the final disappearance of the 
unused cytoplasm, and that succeeding stages in the process consist in 
the progressive growth of this vacuole. At the very late stage shown 
in Figs. 51 and 52, the diminishing cytoplasmic substance is represented 
by two masses — one, the larger, is attached to the anterior portion of the 
antherozoid ; a second, irregular in form and including the limosphere, is 
enclosed within the arc formed by the posterior end of the antherozoid. 
This latter cytoplasmic mass is best shown in Fig. 53, which represents 
the posterior part of an antherozoid in the same antheridium as those 
shown in Figs. 51 and 52. By the time of the maturity of the anthero- 
zoid (Figs. 54, 55), the anterior cytoplasmic mass has disappeared, but 
the posterior mass with the limosphere still remains. This persistent 
remnant of the unused cytoplasm is the cytoplasmic • vesicle ’ of many 
authors. 
The gradual swelling and softening of the walls separating the an- 
drocytes, which has been described as beginning very early — even before 
the division of the androcyte mother-cells — continue until the walls, as 
definite structures, entirely disappear. The androcytes now lie embedded 
in a substance, evidently derived from the disintegrated walls, which stains 
very faintly, often so faintly that it shows no stain at all except by contrast 
with the adjacent uncoloured portions of the preparation. After the walls 
are disorganized, each androcyte in an antheridium sometimes seems to be 
enclosed in an envelope (Figs. 41, 45, 50) composed of a material which 
is itself faintly stained but is distinct from, and evidently denser than, 
the substance which otherwise fills the spaces between the androcytes. 
Sometimes this special envelope is only faintly visible, and often it is 
quite indistinguishable. Whether or not it is always present but visible 
only in specially favourable cases must be left an open question. When 
present, the envelope conforms in general to the shape of the androcyte, 
being therefore lenticular in the later stages (Fig. 50) ; but it is turgid, and 
the space within it is by no means completely filled by the androcyte. 
In antheridia containing mature antherozoids (Figs. 54, 55) I have 
never been able to distinguish this envelope. Antheridia at this stage 
do, however, differ markedly with reference to the stainability of the 
substance in which the antherozoids are embedded. In the antheridium 
from which Fig. 55 was drawn, each coiled antherozoid lies in a clear space 
