OF THE GERM CELLS OF METAZOA. 157 



and more spherical than the chromosomes. In the monaster stage, in favorable cases 

 where the chromatic elements are not too densely arranged, are seen fourteen larger 

 elements, the chromosomes proper, and two smaller ones regularly rounded in form, 

 which are the chromatin nucleoli (Figs. 2, 3, N. %}. Sometimes the chromosomes are 

 rounded, but since they frequently appear slightly elongate on pole view of the spindle, 

 their division in metakinesis must be an equational one. In the metakinesis all sixteen 

 elements, the fourteen chromosomes and the two chromatin nucleoli, are divided, so that 

 each daughter cell (first spermatocyte) receives sixteen elements. 



Thus there are two chromatin nucleoli in the spermatogoiiia, and the chromatin 

 nucleolus of the spermatocystes is not, as I had previously described, formed by a modifi- 

 cation of one of the fourteen chromosomes of the spermatocytes, but is derived from the 

 two of the spermatogoiiia. My error was perhaps excusable, since in restudying the 

 preparations which were used for my former paper I find that they are not suitably 

 stained to show the chromatin nucleoli in the spermatogoiiia. 



Growth period of the spermatocytes (anaphases of the last spermatogonic division, 

 synapsis, postsynapsis, telophase and rest). The fourteen chromosomes in each daughter 

 cell (first spermatocyte) pass toward the pole of the spindle and become irregular in 

 contour and form. Then each becomes longitudinally split (Figs. 411). This splitting 

 cannot be clearly seen in all preparations, and is by no means as clear as in Anasa and 

 certain other Hemiptera '.; the preparations of my former paper were too deeply stained to 

 show it. The split commences in the-early synapsis stage (Fig. 4) and is most marked in 

 the postsynapsis (Fig. 9), and is clearly a single longitudinal split. Never do the split 

 halves separate widely from .one another, as Paulmier found for Anasa, but always appeal- 

 to remain close together and approximately parallel ; at the most there is a divergence 

 only at the ends of the chromosomes. On deep staining the split may be easily over- 

 looked. The two chromatin nucleoli do not become loose in texture, retain their charac- 

 teristic red stain with saffranine, and join together in the early synapsis to form one 

 dumbbell-shaped (bivalent) one (N. 2, Figs. 4, 5, 8, 10) ; they do not become longitudi- 

 nally split like the chromosomes proper. In the early synapsis they are frequently very 

 irregular in form, as I showed in my previous paper, but the apparent fragmentation of 

 them which I then described a fragmentation of a single long one into two is not a 

 fragmentation at all, but a stage before the two have joined to form one bivalent one. 



Reduction in number of the chromosomes. In my earlier paper I showed that the 

 number of chromosomes is reduced one-half during the syiiapsis period i. e., long before 

 the maturation divisions. I then considered it probable that the reduction in number 

 was effected by a union of chromosomes end to end, but was unable to prove this point. 

 Since then I have been able to demonstrate that this numerical reduction is effected in 



