l (fig. 152), and the resulting later stages were
double-tailed (fig. 153). Figure 156 shows how the spindle-substance
goes into the tail and gradually disappears as the tail lengthens.

The centrosome is evidently applied to the nuclear membrane, as in
_Stenopelmatus_, and the middle-piece is developed in connection with
it, as in figures 156-157, 154-155, 158-160. The element _x_ of the
spermatids gradually disappears (figs. 150, 159). An acrosome develops
at the anterior end, the head condenses and lengthens, and we have the
ripe spermatozoon (fig. 161). The tail is very long and is shown only in
part.

Of the forms studied, _Blattella_ alone has many degenerate spermatozoa.
Some follicles have none, others a number varying perhaps from
one-fourth to three-fourths of the whole number. No evidence of
degeneracy was detected among the young spermatids up to the stage shown
in figures 154-155, where a few like figure 162 were found. Most of the
degenerate forms occur among the nearly ripe spermatozoa or in the
sperm-ducts. Such are shown in figures 163 to 168. The chromatin is
strangely broken up into irregular clumps, and probably no two of these
degenerate sperm-heads can be found which are alike. The tails are
always imperfect. The distribution and varying numbers of these
degenerate spermatozoa make it impossible to interpret their condition
as due to the absence of the accessory chromosome, as Miss Wallace does
in the spider. The only probable explanation, it seems to me, is
imperfect mitosis. Cases where more or less chromatin is left behind in
the cytoplasm, especially in the first spermatocyte mitosis, are very
common, and such cases as those shown in figures 149 and 150 are not
rare. The giant cells, so far as I have been able to trace them, do not
develop into spermatozoa.

The most important points are:

(1) The presence of the element _x_ in the spermatogonia, closely
associated with the nucleolus.

(2) The uneven number of chromatin elements in the metaphase of
spermatogonial divisions.

(3) The connection of the element _x_ with the spireme up to the stage
where the spireme segments to form the bivalent chromosomes.

(4) The varied character of the tetrads, showing the first spermatocyte
division to be a reducing division in the sense that it separates whole
chromosomes.

(5) The fact that the element _x_ fails to divide in the first
maturation division, does divide in the second, but can not be