24 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY. 
Views of the eight-cell stage are shown in Figures 15 to 18. 
From a cyto-mechanical standpoint, the third cleavage may be char- 
acterized as follows. The first division of the asters is along a line at 
right angles to the axis of the previous spindles, and indicates the posi- 
tion of the spindles for the next cleavage. These lie in the long axes 
of the cells, and the cell walls are formed in the position demanded by 
the principle of least surfaces. 
Fourth Cleavage. 
Immediately after the division of D” (Plate 2, Fig. 12), the asters in 
d^! and d^? begin to extend laterally, at right angles to the axis of the 
preceding spindles, and each becomes divided into the two asters for the 
following spindle. In d*', Figure 11, the two asters for the succeed- 
ing cleavage are still connected by a striate band. Figure 11 shows a 
ventral view of the same egg as Figure 12, the five-cell stage. The 
corresponding dorsal view of a slightly later stage is shown in Figure 
14. The asters in d^? are moving apart in the same manner as in d^", 
save that the line of separation is slightly oblique, the left aster being 
higher. 
In the same way the asters in the cells a*1—c41 and at? - c*? become 
constricted, and divide at right angles to the axis of the preceding 
spindles, The dividing asters in c^? are shown in Figures 14 (Plate 2) 
and 17 (Plate 3), and those of e*! in the latter figure. Views of the 
other four cells would show similar conditions. 
From the manner in which the asters separate in all of the eight 
cells, one would be led to expect that the next cleavage would be merid- 
ional, at right angles to the third cleavage. This expectation is 
strengthened by the fact that the lateral dimensions of the cells in 
which the asters lie are considerably greater (in the quadrants A, D, and 
O, at least) than the opposite measurements (Fig. 17). 
Jut in a slightly later stage it is observed that the line joining the 
asters in d*! has become oblique, like that joining those of d^? (as 
mentioned above). This oblique position of the asters in di is shown 
in the ventral view (Fig. 15). The left aster (right side of the figure) 
has become ventral, the right one dorsal. The sagittal section (Plate 1, 
Fig. 7) of a slightly later stage shows the completion of the rotation 
thus begun ; the line connecting the two asters and passing through the 
nucleus is now approximately dorso-ventral in direction. 
At the same time a similar rotation has taken place in the cell Ur 
but the position taken by the two asters is not the same as in d. One 
