SPERMATOGENESIS IN AMPHIBIA AND INSECTS r 
variability in staining behavior of the centrioles in Plethodon 
leads me to look upon this as a very plausible explanation. 
In the late stages this small centriole seems to fuse with the 
free end of the much elongated middle-piece proper, from one 
end of which the axial filament arises (compare fig. 15). The 
function of this small centriole is not yet understood, but a pos- 
sible explanation has suggested itself, which I hope eventually 
to prove or disprove definitely. It will be recalled that the tail 
of the urodele sperm bears a long, undulating membrane, the 
base of which is inserted along the axial filament of the tail, 
while the free edge is marked by a definite filament somewhat 
smaller in size than the main tail filament. The axial filament 
proper undoubtedly arises from the large centrosomal middle- 
piece. The exact origin of the membrane filament has not, 
however, been satisfactorily explained. I would like to suggest, 
on the basis of observations already made, the possibility of the 
origin of this filament from the second, eventually much the 
smaller, of the original halves of the proximal (?) centriole. Un- 
fortunately, this filament is so delicate that its free portion can 
be made out in its earlier stages only with some difficulty, while 
the insertion can never be satisfactorily followed. It frequently 
happens, however, that the developing sperms are cut obliquely 
through the base of the head, and in such sections one can some- 
times see what appears to be a fine filament arising from the 
small centriole and passing backward through the ring centriole 
(figs. 16 and 17). (See also fig. 18, in which the ring was not 
included in the section.) Should further study bear out such an 
interpretation, the origin of the membrane filament would fall 
in line with the facts already known as to the origin in general 
of vibratile filaments. It is possible, nevertheless, that the 
small centriole plays a réle in the late stages during the drawing 
out of the ring centriole; for I have constantly observed a small 
darkly staining mass at the base of the tail filament which seems 
to originate from the centrosomal middle-piece (fig. 19). This 
mass subsequently moves out along the filament toward the free 
end of the elongating ring centriole, but its fate and function are 
alike unknown. 
