M. Koidzumi 
wall of the bell. The axial rod of the nipple is devoid of the constriction, 
behind the anterior knob, seen in the former variety. The inner layer of the 
wall of the bell is thicker than in japonica. The deeply stained zone at the 
boundary of the inner and the outer layer of the bell is distinct and shows an 
appearance of a sheet of minute granules closely aggregated. The acuteness 
of the inclination of the wall of the bell and the proportionally large dimensions 
of the nucleus are associated with a difference in the shape of the mem¬ 
branous wall of the corbule. In the majority of cases, it assumes the shape 
of an inverted cone, or sometimes of a cylinder, touching the nucleus tan¬ 
gentially in the vicinity of its equator and stretching behind it. 
Dividing forms are occasionally met with in this variety. In the termites 
taken directly from their nest, or those kept in large vessels under conditions 
supposed to be natural for them, dividing forms are very seldom met with; 
and a large number of termites have to be examined before hitting upon any 
dividing forms. I happened, however, to come across a method of getting 
dividing forms with less difficulty: namely, by keeping the termites, for 
several days, in a small glass dish with a small quantity of mud, and a piece 
of cotton-wool to serve as food. Such conditions acting upon the host seem 
to induce the multiplication of the intestinal organisms. It is rather re¬ 
markable that a large number of dividing forms may be found in a single 
host, while not a single one is met with in another subjected to the same 
conditions. 
The division processes of the Italian species have been described by Foa 
(1904), and Kofoid and Swezy (1919) have described those of T. campanula. 
The division processes of the forms under consideration are of the same type 
as those described by these authors. 
Actual division takes place at first in the nipple. It is bisected longi¬ 
tudinally into two equal halves, and a rather thick strand (“fuso esterno" 
of Foa, and “ paradesmose ” of Kofoid and Swezy) makes its appearance at 
the base of the daughter nipples, keeping them in connexion. This strand is 
somewhat fibrous and is stained rather feebly by iron-haematoxylin. Before 
the above-mentioned division takes place, nuclear changes usually occur. In 
the resting stage the nucleus is of the same structure as that of the former 
variety; but in the variety formosana the chromatin masses are more often 
not distinctly separated from one another. The first indication of the nuclear 
division is the disappearance of the membranous structure of the corbule, and 
the scattering of the chromatin masses into the ground substance of the 
nucleus. There the chromatin masses appear swollen and broken into pieces 
(pi. X, fig. 6). The nuclear membrane becomes indistinct and the ground sub¬ 
stance becomes denser in its constitution. At about this time, the division of the 
nipple takes place. The connecting strand continues to be elongated and the 
daughter nipples are separated by degrees. The nucleus and the strand, which 
were originally separated by some distance, are now gradually brought nearer 
to one another until they come to be united as is shown in PI. X, figs. 7 and 8. 
