126 CG. CLIFFORD DOBELL. 
the spores become exceedingly hard and refringent (fig. 7) and 
refuse to stain in the slightest degree. 
Whilst the spore membrane has been forming, the sincere 
mediate part of the spiral has broken down completely, and 
as it degenerates, the protoplasm changes its staining reac- 
tions. At the beginning, the spiral was red, in a bluish cyto- 
plasm (figs. 18, 14, 16). As it degenerates, and diffuses 
through the protoplasm, it imparts to it a purplish (fig. 17), 
and finally a pink tinge (fig. 18). ‘The spores, which at first 
occupied the ends of the cell, now move towards one another 
somewhat, so that little polar caps of protoplasm are left at 
the extremities of the rod. In contradistinction to the 
rest of the protoplasm, these caps stain blue with Giemsa 
(fig. 18). The rods with their two terminal spores present 
a very characteristic appearance at this stage (figs. 3, 18), 
for they remain actively motile for some time longer, some- 
times for several hours. In the end, they come to rest, 
and the protoplasm degenerates (fig. 19), so that the two 
fully-formed spores alone remain, connected by the dead and 
broken remains of the organism. 
The fully formed spores, like those of B. biitschlii, vary 
somewhat in shape, and show a small cap-like eminence at 
one end (fig. 8). This, in B. biitschlii, is the pole at which 
germination occurs. I regret that I have not been able to 
watch the germination of the spores in B. flexilis. In size, 
the spores are about 5 uw X 2 uw, and hence cannot be easily 
mistaken for any other spores found in the same locality. 
I should point out that the figures of stained specimens 
have all been drawn from Giemsa preparations in which 
drying has taken place between staining and mounting. 
They are all, therefore, somewhat flattened out, and so appear 
broader than they really are (see especially figs. 9, 14, and 
17). This flattening of the organisms serves, however, to 
show their structure more clearly. And although the method 
is not to be recommended for the most accurate work, it 
gives very pretty results. I have used the method only in 
conjunction with others (see Appendix), but the results have 
