160 MONTGOMERY A STUDY OP THE CHROMOSOMES 



less extent, according to the object, the stages which precede and those which succeed. 

 It results from this that the stages of mitosis of the growth period cannot be exactly 

 compared with those of other cells, and hence the terms "anaphase" and ' ' telophase " 

 can here have a significance only approximately similar to that of other mitoses. 



The mnfiirnfiijii divisions. In the early prophases the longitudinal splitting of the 

 chromosomes is well marked, clearer than in preceding stages (Figs. 13, 14). ^Each 

 chromosome is, as before the rest stage, clearly bivalent, formed of two longitudinally split 

 univalent chromosomes joined so as to make an angle together (Figs. 13-15), and at the 

 bend of the angle is a connecting linin thread. These forms of the bivalent chromo- 

 somes were clearly figured in my earlier paper, except that then I had overlooked the 

 longitudinal split. The chromosomes gradually become closer, shorter, with smoother out- 

 lines, the longitudinal split gradually becomes hidden, and the definitive chromosome with 

 the form of a dumbbell results (Figs. 16-18). In the definitive chromosome there is 

 usually no trace of the longitudinal split, except occasionally as a slight indentation at 

 the free end of a. univalent component. The constriction of the dumbbell marks the point 

 of union of two univalent chromosomes, which is effected by a linin band which gener- 

 ally never becomes quite hidden. 



In the late prophases, just before the disappearance of the nuclear membrane, and 

 when the centrosome pairs have reached opposite poles of the nuclear surface, a remark- 

 able condition of the linin threads is found (Fig. 17) ; it was also shown in Figs. 152 

 and 153 of my earlier paper. The linin, previously in the form of fibres or strands, 

 now takes the form of chains of small globules quite as Van Beneden (1883) had 

 figured for Ascctris. I cannot explain this condition, but I have found it always at this 

 stage, and at this stage only. 



In the first maturation division there are seven bivalent chromosomes and one bival- 

 ent chromatin nucleolus, and all these elements are divided transversely in metakiuesis, 

 being placed in the monaster stage so that their constrictions lie in the plane of the 

 equator. Fig. 18 shows a monaster stage with all these elements on lateral view, Fig. 

 19 on pole view ; this stage was accurately described by me in my former paper, so that 

 I have no additions to make to that description. Whole univalent chromosomes are 

 separated in the ensuing metakinesis, and the univalent components of the chromatin 

 nucleolus are also separated. 



When the daughter chromosomes separate in the anaphase a constriction or indenta- 

 tion appears on them (Figs. 193-201 of my preceding paper). This I am now able to 

 prove, in agreement with Paulmier's observations on Anosa, is the reappearance of the 

 longitudinal split ; this indentation or constriction becomes placed in the equatorial plane 

 of the monaster stage of the second maturation division, so that the latter division divides 



