44, H. B. FANTHAM. 
In the usual method of longitudinal division the parts 
eradually separate (Pl. 2, fig. 17) by wriggling till they are 
finally 180° apart, but they frequently remain for a long time 
attached at one end with the free ends diverging about 30° 
to 40°, like the legs of a compass (Pl. 3, fig. 18), and 
executing very rapid movements the while. The unseparated 
end is often fastened to débris (text-fig. 10, and PI. 2, 
fie. 28), and the dividing forms remain in the same place 
during the process. 
On two occasions I watched the partially divided forms 
wriggling actively for about an hour, not always in unison, 
but usually so. It seems almost impossible to follow the 
whole process in life, for on each of the occasions that I had 
the opportunity the forms died before the completion of the 
division. Perrin also mentions this difficulty, for he watched 
a pair for forty minutes, and these died before the completion 
of division. 
The division of the chromatin masses takes place early, in 
the manner described in the preceding section when dis- 
cussing the nucleus and the significance of its various forms. 
Judging from stained preparations the order in division 
probably is: first the basal granules, next the membrane, 
then the chromatin masses, but the interval between the 
stages is often slight. | 
The nearly separated organisms remain attached only by 
the still unsplit end of the periplast of the parent, a vacuole 
appearing in the common portion of the periplast just before 
final separation. 
Regarding some of the phases of nuclear division mentioned 
by Perrin (8, 9), I was unable to trace his division of a karyo- 
some rod into bacilliform and dumb-bell shaped segments, 
followed by transverse and then longitudinal division. 
I have seen longitudinal division in life in 8. anodonte 
also, and this is the usual direction of fission in the Spiro- 
cheetes of Anodon (PI. 3, fig. 40). 
