6 ROBERT H. BOWEN 
bipartite or dumb-bell shaped,! the two parts being typically 
equal in size. Meves figures much the same thing, though he 
described the formation as a curved rod. This same division of 
the proximal centriole occurs in Plethodon (figs. 7 and 10), the sep- 
aration of the two parts being sometimes very marked. ‘To one of 
the halves the axial filament is attached. When the nucleus begins 
to draw out, this division of the proximal centriole is apparently 
lost in many cases—sometimes, no doubt, by the recombination 
or fusion of the two parts and at other times by the orientation 
of the centrioles. At the same time, the part to which the axial 
filament is attached increases very rapidly in size, while the other 
part remains relatively small and inconspicuous. However, 
with a little searching, examples can be found at any of the 
intermediate stages in which the two parts of the proximal cen- 
triole can be clearly separated (figs. 18, 14, and 15). That the 
two bodies in these cases are actually the equivalent of the appar- 
ently single proximal centrioles of other cases can be demon- 
strated by comparison of adjacent spermatids in which the parts 
can or cannot be separated (figs. 138 and 14). It seems probable, 
as Professor Wilson has suggested to me, that in McGregor’s 
preparations the large centriole was extracted more than the 
smaller one, giving rise to the interpretation already noted. The 
1 It should be pointed out that the origin of this bipartite condition has not 
yet been adequately worked out. I have followed the usual custom in considering 
the two parts as formed by the division of the proximal centriole, my own prepa- 
rations not permitting a detailed study of the very early spermatid centrioles. 
Doctor Wilson has, however, called my attention to the fact that generally in 
the differentiation of the sperm a ring arises from only a portion of the distal 
centriole, while the remainder later comes into close relation with the proximal 
centriole. It is possible that something of the same kind obtains in the case of 
the urodele. Thus we might consider the portion of the so-called ‘proximal’ 
centriole to which the axial filament is directly connected as the central part of 
the ring centriole, while the other portion would actually be the proximal centriole 
itself. This would explain the apparent transfer of the insertion of the axial 
filament from the distal to the proximal centriole as required by McGregor’s 
description, and would fall in line with the usual accounts of sperm formation 
which derive the axial filament from the distal centriole. Further study of this 
possibility is contemplated. However, in view of the uncertainty attaching to 
these speculations, I have thought it best to follow the traditional accounts for 
the present purely as a matter of convenience in description. 
