SPERMATOGENESIS IN AMPHIBIA AND INSECTS 5 
long after the stage shown by McGregor in his figure 32, in which 
it has supposedly been incorporated with the proximal centriole 
to form the middle-piece of the sperm. Its ultimate fate is 
unknown, preparations of late stages in sperm formation not yet 
having been successfully impregnated. Meanwhile the proximal 
centriole (in particular) increases progressively in size and stain- 
ing capacity (compare figs. 11, 12, and 18) and becomes applied 
very closely to the nuclear membrane. At first it is spherical, 
but as the head elongates and becomes narrower, the sphere 
begins to draw out (fig. 14) and it becomes eventually a long rod 
(fig. 19)—the (centrosomal) middle-piece of the sperm. In this 
process I have been quite unable to find any evidence of an 
actual penetration of the nuclear wall by the ‘middle-piece,’ as 
described by both Meves and McGregor. The centriole seems 
rather to be merely in close contact with the nucleus, as is the 
case in many other animal sperms. 
With respect to the origin of the middle-piece, I am, therefore, 
in general agreement with Meves. The unusual topographical 
relation existing between the Golgi remnant and the centrioles 
was doubtless the cause of McGregor’s error in the derivation of 
the middle-piece, coupled also with the poor staining conditions 
at this period. There is also a possibility that some large, darkly 
staining granules of doubtful relations, which occur in the sper- 
matids, may have been confused by him as derivatives of the 
cast-off ‘sphere.’ 
Finally, the conditions in the completed middle-piece described 
by McGregor remain to be explained. Since, according to this 
author, in the formation of the middle-piece the proximal cen- 
triole merely becomes embedded in a mass of sphere material, 
it should be possible to separate these two components by dif- 
ferential staining. This was in fact apparently accomplished 
(McGregor’s fig. 32), the sphere material staining less heavily 
than the centriole. My own observations of the proximal cen- 
triole (so-called) suggest, however, a quite different interpre- 
tation of this appearance. As was noted by McGregor, the 
proximal centriole in the earlier spermatid stages often appears 
