Helen Dean King 365 
the same diameter as the sperm-head, but it is sharply marked off from 
it and shows not the slightest affinity for chromatin stains. In rare 
instances, as shown in Fig. 71, a slender fibre extends through the 
entire middle-piece. This fibre is evidently the connection between the 
centrosome at the posterior end of the middle-piece and the centrosome 
imbedded in the end of the sperm-head which has persisted through all 
stages in the development of the spermatid. As a rule, the connection 
between the centrosomes is broken after the stage of Fig. 65, and in the 
older spermatids and in the spermatozoa the middle-piece appears per- 
fectly homogeneous. 
The tail of the spermatozodn is very long and it is composed of two 
filaments which are connected by a thin, transparent, undulating mem- 
brane. The axial-filament, which grows out of the outer centrosome, 
is somewhat thicker than the marginal-filament and it extends some 
distance beyond the latter to form the end-piece of the tail. 
GENERAL DISCUSSION. 
As long ago as 1887, Flemming (15) described typical tetrad groups 
in the testes of Salamandra, although he considered these structures to 
be “anomalies” and not normal stages in the development of the 
spermatozoa. Later vom Rath (46, 47) maintained that tetrads are 
normally present in the testes of Rana as well as of Salamandra, and that 
one of the maturation divisions is a transverse or reduction division in 
the Weismannian sense. With but few exceptions, all of the investiga- 
tors who have recently worked on the spermatogenesis of amphibians 
have insisted that both of the maturation divisions are equatorial, and 
that normally tetrad groups are not present in the spermatocytes. This 
later work on amphibians, and the seemingly exhaustive studies of 
Brauer (5), Boveri (3), and Hertwig (22) on the germ-cells of Ascaris, 
together with the work of Strasburger (56, 57) and other botanists on 
the flowering plants, has given strong support to the view that reduction 
in the germ-cells can be effected either by a double longitudinal division 
or by one transverse and one longitudinal division, the division of the 
chromatin substance being the main thing and the manner of its achieve- 
ment quite secondary. The most recent studies on spermatogenesis, 
and ovogenesis, and experiments made by Boveri, Morgan, and others, 
have seemed to show that the chromosomes maintain their individuality 
from one generation to the next. If this is true, then in the maturation 
mitoses, reduction must in all cases be brought about by one transverse 
