770 THOS. H. MONTGOMERY, JR. 
of it, the central pole being uppermost and the distal lowermost 
in the drawing. When the first maturation spindle is established 
each centriole pair occupies one pole of the spindle and lies deep 
within the cytoplasm (figs. 87-89, 93-97). At the close of this 
mitosis the two centrioles of each pair, without dividing, move 
apart from each other, as described in detail in my paper of 
1898, to constitute the poles of the second maturation spindle 
(figs. 102, 103). In this second mitosis, accordingly, each spindle 
pole has but a single centriole. 
During the anaphases of the second maturation division 
(figs. 104, 106-111) the centrioles become exceedingly small so 
that it is difficult to see them. But in each early spermatid a 
minute centriole may be seen at the central pole of the cell (¢, 
figs. 112, 113) in contact with both cell and nuclear membrane. 
To each centriole is attached a short flagelliim that projects from 
the cell surface. The centriole then moves with its flagellum 
around the nucleus to the distal end of the latter, figs. 114 to 120, 
exhibiting successive stages of this movement. As the centriole 
passes along the nuclear surface it rapidly increases in size, and 
its flagellum grows in thickness and length. Thus the centriole 
comes to lie against that part of the nuclear membrane which 
lacks the chromatin envelope, at the base of a small pit formed by 
an invagination of the cell wall. Such spermatids are distinctly 
bilateral, with the central pole occupied by the sphere (Sp), 
the distal by the mitochondrial body, and the centriole in a plane 
separating 4 right and left side; figs. 118 and 110 show spermatids 
from lateral view, and fig. 119 one from dorsal or ventral aspect. 
As the mitochondrial mass elongates and becomes divided length- 
wise (figs. 121-123) the flagellum come to extend from the cen- 
triole backwards between the mitochondrial moieties to project 
free at the distal end of the cell. I could not determine whether 
the flagellum thereby sinks into the cytoplasm, or whether it 
becomes placed within a longitudinal groove which closes later: 
but the latter is more probable, judging from the conditions 
shown in figs. 118 and 120, which indicate a groove along that 
side of the cell nearest the mitochondrial mass. So soon as the 
flagellum comes to lie between the halves of the mitochondrial 
