138 BULLETIN OF THE 
division might take place transversely to the axis, this result could still 
be attained by a rotation of the nucleus, even when the tendency was for 
the nucleus to divide at right angles to the previous division. It is 
obvious that rotation would occasionally be apparent, provided it took 
place soon after division, and previous to the absorption of the proximal 
end of the connecting filament. I examined a large number of prepara- 
tions to find evidence of rotation, but I must admit that the evidence 
was slight, and hardly sufficient to establish the hypothesis which I had 
formulated. It is therefore put forth provisionally, in the hope that it 
may lead to further investigations in this line. 
The most striking instance of rotation was found in one of the quadri- 
nucleate cells (Fig. 33, nuclei a and b). It is evident that three nuclear 
divisions have taken place without any division of the cell, producing 
two, three, and four nuclei. The arrangement of nuclei makes it rea- 
sonably certain that the dower pair arose by division of one, and the 
upper pair by division of the other nucleus of the binuclear stage. 
Only under this supposition could the daughter nuclei of that stage 
have had the normal arrangement, to which all the neighboring cells 
rigidly conform. We further find, that, while the upper pair of nuclei 
has arisen by a division in the long axis of the cell, the lower pair has 
been produced by division in the transverse axis, and therefore in con- 
formity with the law previously stated (p. 136). One nucleus of each 
pair (a and b) retains a remnant of the connecting filament, which is 
directed, not toward the sister nucleus, but to a point 90° distant from 
it. This condition could have been brought about only by rotation of 
the nuclei, which in both cases has been through an are of 90°. 
In the serosæ from older embryos, the daughter nuclei almost inva- 
riably recede from each other in the course of division, The amount 
of recession is governed by the length of the cell (Fig. 15). In the 
younger membranes, as already stated, the constriction is deep and 
narrow, so that the nuclei not infrequently lie very near together 
(Fig. 13). In these young membranes, however, the nuclei are larger, 
and the cells are usually smaller, than in the old membranes. Since, 
moreover, the large binucleate cells of young membranes almost always 
have their nuclei symmetrically placed at the ends, it is probable that 
the nuclei gradually move apart after division, as the cell increases in 
size. 
It will be seen that my interpretation of the primary cause of the 
division of these nuclei agrees in part with the hypothesis advanced by 
Chun (’90) for the explanation of amitotic division in general, This is, 
