THE SPERMATOGENESIS OF Hysprip PEAS §23 
that it is not seldom difficult surely to determine the character of 
a given cell, To study the divisions from which the sporiferous 
tissue arises it is necessary to obtain very young anthers indeed, 
those which are merely slight protuberances on the primordium of 
the flower. The difficulties in the way of studying the anther are 
decidedly increased from the fact that the inflorescence consists of 
not more than four or five flowers, all usually in widely different 
stages of development. 
For the most part the nuclear divisions in the sporogenous 
cells of Fillbasket x Debarbieux are of the regular equational 
type, quite the same as those in the somatic cells of the plant. 
That is, fourteen segments are formed in the prophases and these 
become arranged in the equatorial plate and in metaphase are 
already split before the migration to the poles of the spindle. In 
anaphases the chromosomes are frequently L-shaped, although 
not so sharply bent, and are arranged haphazard, that is, as purely 
by. chance. In the telophases the chromosomes unite end to end 
‘and form a spireme; this becomes changed into a delicate reticu- 
lum in which all trace of the chromosomes as such disappears. 
In some of the later sporogenous divisions a departure from 
the usual appearance was noted as in the case of the pure form 
Fillbasket. In Fillbasket x Debarbieux the same curious associ- 
ation of chromosomes two and two was observed in a few cases. 
&. 6 shows an anaphase of the last sporogenous division ; the pair- 
ing was even more clearly marked in the preparation than in the 
sketch. The pairing was with a single exception observed in 
anaphase or early telophase only. The exception is shown by 
J. 5, which is an equatorial plate in probably the last division be- 
fore synapsis. Ihave not seen the association in early anaphase 
for the reason perhaps that the chromosomes are at that time 
under the influence of the spindle fibers. 
When the chromosomes reach the poles of the spindle they 
unite by their polar ends, and these appear to fuse together, mak- 
ing a more or less well-marked ring from which depend the chro- 
mosomes in pairs. This is shown very imperfectly by / 7, & and 
9. In no case did I see a spireme which I could surely say was 
continuous or which was made up of more than two elements ; 
this may well be because the particular stage was wanting in the 
