78 



and become attached to the fibres by its middle points. Then, as 

 the middle part of each longitudinal half respectively, is drawn out 

 towards the pole, the whole assumes the character of the later stages 

 of a ring (Fig. 8) after the longitudinal fission has been completed. 

 It really represents merely another form of this (elliptical) type in 

 which the splitting of the ellipse at the two ends has happened early, 

 instead of only occurring as the final step in the separation of the 

 two daughter chromosomes. 



It will thus be seen that the normal evolution of the heterolype 

 chromosomes in the divisions of the pollen mother cells of plants, 

 corresponds to what has hitherto been taken as the normal heterotype 

 evolution of the chromosomes in animal spermatocytes. But as the 

 variations occurring in this process among plants fall naturally into 

 several well-marked types it was thought desirable that a compara- 

 tive study of the variations in the formation of animal heterotype 

 chromosomes should be made. Up to the present time we have found 

 the greatest variation among animal heterotype chromosomes to be 

 presented during the great heterotype division of the spermatocytes 

 of Triton. 



In this form, the nucleus previous to division presents the well- 

 known animal contraction figure (synapsis) which so far as we can 

 ascertain corresponds essentially with that witnessed before the divi- 

 sion of the pollen mother cells in Lilies (Fig. la). 



The chromatic filament becomes longitudinally split throughout 

 portions of its length, and these split portions gradually separate from 

 one another into the definite number of chromosomes characteristic 

 of the species. 



The longitudinal splitting of the segments (chromosomes) may, 

 as in the case of Lilies be partial or complete. It may, and this in 

 Triton is the usual type, form a more or less elongated opening in 

 a loop which remains closed at both ends as in (Fig. 2 a), or it may 

 extend quite out to one end but not to the other (Fig. 12 a). Finally, 

 in some cases, the split traverses the whole thread to both ends and 

 as was the case in plants, in this case the chromosome is both in 

 its origin and general appearance (at the time in question) indistinguish- 

 able from a normal homotype thread. 



Just as in the case of plants, the succeeding evolution is modi- 

 fied throughout its entire course by these initial variations. 



When the split does not extend to the ends of the filament and 

 the chromosome is a true closed loop, we found that, at any rate in 

 he majority of cases, the long axis of the chromosome at first lay 



