274 Allen. — - The Spermatogenesis of Poly trie hum juniperinum. 
pressed against the blepharoplast and the plasma membrane in this region. 
Fig. 19 represents a cell whose nucleus has assumed the peripheral 
position fairly early ; in other respects, however, this cell seems to be in 
about the same stage of development as are those shown in Figs. 17 and 
18, whose nuclei are still centrally located. 
The next conspicuous change to be noted in the developing androcyte 
is a division of the limosphere. The division is preceded by an elongation 
of the limosphere in a direction perpendicular to the anterior portion of 
the blepharoplast, to which the limosphere now seems, at least in most 
cases, to be attached (Figs. 21, 22). Then follows a constriction (Figs. 23, 
24), and finally a division (Figs. 25, 26) into two bodies which usually differ 
noticeably in size. The smaller of the two, hereafter referred to as the 
apical body , remains in contact with the anterior end of the blepharoplast ; 
the larger one, since it is of approximately the size, and retains essentially 
the appearance, of the limosphere before its division, will still be spoken 
of as the limosphere. It moves away from the anterior end of the 
blepharoplast (Fig. 26) and soon comes to lie in contact with the nucleus 
(PL XVI, Fig. 27). The apparently divided limosphere shown in Fig. 20 
(PI. XV) represents a different condition ; the androcyte from which this 
drawing was made is the only one I have found at so early a stage which 
contained a double (or constricted) limosphere, although less marked 
irregularities in the outline of this body have sometimes been observed. 
It is possible, though I think hardly likely, that the condition in Fig. 20 
indicates a premature division of the limosphere. Something apparently 
equally removed from the usual course of events is shown in the lowermost 
androcyte of Fig. 24, whose limosphere seems to be dividing into four 
bodies instead of two. 
Sometimes, when the apical body and the limosphere are moving away 
from each other (Fig. 28, PL XVI), they seem to be connected by a definite 
fibre or strand. This has not been observed in the majority of cases, but 
in many of the androcytes in the antheridium from which Fig. 28 was 
drawn this connecting strand is plainly visible. If it is generally present 
at these stages, the staining methods used are not adequate invariably to 
demonstrate the fact. 
At about the time that the limosphere divides, the nucleus begins to 
elongate in a direction parallel to the length of the blepharoplast. The 
earliest evidences that I have found of the extension of the nucleus are in 
the androcytes shown in Figs. 22 and 23, PL XV. In each of these cases the 
nucleus is drawn out into a point that is directed towards the anterior end 
of the blepharoplast, the posterior part of the nucleus remaining rounded. 
The elongation of the nucleus seems usually to begin in this way; but 
there are exceptions to this rule, as in the androcyte shown in Fig. 25, 
whose nucleus is rounded anteriorly and somewhat drawn out at its 
