32 bulletin: museum of comparative zoology. 



plished, in arthropods at least, by one cross and one longitudinal division. 

 It is generally assumed that, by one of these divisions, — the transverse 

 division, — this reduction is accomplished through the breaking apart of 

 the chromosomes at the point at which they were united in the preced- 

 ing synapsis. Now, as the accessory chromosome is not obtained by the 

 union of two spermatogonial chromosomes, this reduction division is not 

 necessary for it and does not take place. For these reasons, while the 

 ordinary chromosomes are each composed of four parts, i. e. are tetrads, 

 this modified chromosome is made up of but two parts, i. e. is a dyad. 

 Furthermore, it is logically to be expected that the accessory chromosome, 

 being dyad in its nature, should take part in only one of the succeeding 

 divisions. This peculiarity has indeed been observed by many investi- 

 gators of insect spermatogenesis, and several hypotheses more or less 

 supported by observed facts are offered in explanation of it. These theo- 

 ries I shall mention later, in my discussion of the pertinent literature. 



While these important changes have been taking place in the centro- 

 some and in the nuclear structures, the cytoplasm has remained in prac- 

 tically the same condition in which we left it in the very early prophase. 

 It is reticular in structure, but the network is obscured by the fact 

 that the interstitial substance no longer remains uncolored, but retains 

 the stain with considerable strength. This increased staining power of 

 the hyaloplasm I believe to be caused by the dissolving of the archo- 

 plasm and its mixing with the liquid or viscid substance which fills all 

 the inter-reticular spaces. 



When the cell has reached the stage shown in Figures 29 and 30, the 

 nuclear membrane begins to disintegrate. This is first apparent in the 

 parts nearest the centrosomes. The membrane apparently becomes 

 granular, and the fibres of the nuclear network begin to arrange them- 

 selves in lines which converge toward the centrosome (Figs. 29, 30). 

 The astral rays developed outside the nucleus now seem to penetrate 

 into the nuclear space (Fig. 31). In some cases these rays seem to 

 unite with those formed by the liniu fibres, while in others the two sets 

 seem to remain distinct. I think it very probable from these observa- 

 tions that the linin reticulum gives rise to the mantle fibres, while the 

 astral rays upon the side of the centrosome nearest the nucleus form 

 those fibres of the spindle which do not connect with the chromosomes, 

 i. e. the spindle fibres. As the nuclear membrane continues to disin- 

 tegrate, the centrosomes move apart and thus draw the converging linin 

 threads out into a spindle-shaped figure (Fig. 31). By this same move- 

 ment the chromosomes are also drawn into a position between the centro- 



