no 



integrate and pass into the reticulum although their outlines 

 can be dimly made out even after the reticulum is well formed 

 and the nucleoli have reappeared (Fig. i6). 



5. The centrosoine . 



It is extremely difficult to stain, and correspondingly hard to 

 find the centrosomes in this material ; even at the spindle-poles 

 its identification is not easy. I was able, however, to make it out 

 in two different stages (Figs. 11, 12 and 13). One of these was 

 in the mitosis of the primary sporocyte, the other in that of the 

 secondary sporocyte. In the first of these the centrosome at the 

 spindle-poles was double (Figs. 11 and 12), in the second it was 

 single (Fig. 13). 



II. Tetrad-formation. 



In cases where reduction in the Weismann sense is actually 

 known to take place, there have been wide variations in the ac- 

 counts of the process. It was first described by a former pupil 

 of Weismann's, Ishikawa, who did not find tetrads and who held 

 that reduction in the copepod Diaptonius is accomplished by the 

 separation of entire chromosomes. This result is entirely contra- 

 dictory to the more recent results obtained by the subsequent 

 study of Diaptonius and other copepods. 



Vom Rath, apparently the first to correctly interpret the for- 

 mation of tetrads, gave a different account of reduction in the in- 

 sect Gryllotalpa. He found that the spireme is double before it 

 breaks up and that there are half as many of the double segments 

 as there are chromosomes in the somatic cells. The halves of the 

 double segment separate, except at the ends, and a ring is formed. 

 Later the tetrads arise by concentration of the chromatin at four 

 points of the ring, a method by which the four parts of the tetrad 

 originate by longitudinal division represented by the original divi- 

 sion of the spireme, and by a transverse division. 



Hacker gave still another description of tetrad-formation in 

 Cyclops stvemius. His details have been denied by RUckert, who, 

 however, accepted the general results and agreed with him that 

 the tetrad is formed as in Gryllotalpa by a longitudinal and a 

 transverse division of the original spireme segments. RUckert 

 (1893 and 1894) has found two modes of tetrad formation, each 



