STUDIES ON INSECT SPERMATOGENESIS. 325 



phases, a condition which seems to be connected purely with the 

 spatial relations inside the cell. Thus by reference to Fig. u it 

 will be seen that in the prophases the nucleus and the cell wall are 

 closely approximated adjacent to the centriole, so that the ends 

 of the threads are automatically located in the same optical 

 plane. When the nuclear membrane breaks down, however, the 

 spindle region is relatively small in comparison, and the mito- 

 chondria tend to spread out in the extra space, though they never 

 encroach on the spindle area proper. So it happens that th'e 

 striking effect of polarization is somewhat obscured, but it is per- 

 fectly clear that the orientation remains fundamentally unaltered. 

 The chromosomes now divide and the daughter plates begin to 

 draw apart (Fig. 16), while the cell itself elongates very con- 

 siderably, so that the distance between the centrioles is increased. 

 The result of these movements upon the mitochondria is twofold. 

 In the first place the polarized ends of the threads are straight- 

 ened out along the spindle and in the second place they are drawn 

 along with the diverging centrioles. As an obvious result the 

 opposite ends of the threads are passively drawn out of the equa- 

 torial tangle, and in an early anaphase (Fig. 16) the threads, now 

 arranged in lines more or less parallel to the spindle, conie -to 

 form a sort of " palisade," or sheath, encircling the mitotic figure. 

 The constriction of the cell wall meanwhile develops rapidly and 

 soon comes in contact with the mitochondrial palisade. The 

 threads, however, show no tendency to divide autonomously or to 

 be divided mechanically, and as the constriction deepens they are 

 carried inward by the advancing furrow. Thus the palisade as 

 a whole is constricted very markedly in the equatorial plane, while 

 the opposite ends flare widely, leaving an open space in which lie 

 the daughter chromosome groups. Finally, the constriction 

 reaches the spindle, which, by its tendency to persist as a connec- 

 tion between the daughter secondary spermatocytes, appears to 

 be of a rather firm consistency. The mitochondrial threads 

 which happen to lie immediately in the path of the constriction 

 are thus at last caught between the spindle and the cell wall and 

 seem to be severed mechanically into two parts (Fig. 17). That 

 the threads are actually cut in two, not merely withdrawn into 

 the daughter cells, is made practically certain by the whole series 



