45S BOTANICAL GAZETTE [june 



(fig. 18). The loops represent somatic chromosomes end to end 

 which separate at the outer bend; that is, at the curve away from 

 the entangled mass (figs. 19, 20, 21) rather than at the lower bend. 

 Each radiating loop, therefore, does not represent the bivalents 

 as seen in figs. 21.V, y, 22. Instead, one arm of a loop pairs with 

 and twists about an arm of a neighboring loop, this being 

 plainly evident in the material. Fig. 19 illustrates such a condi- 

 tion, for here a and a' were formerly a continuous loop, likewise 

 b and b'; but, following separation, a and b twist about one another, 

 then opposite ends in all probability form a loop, resulting in a+b 

 forming a bivalent, as seen in figs. 2ix, y, 22, 23. 



The bivalents of the aggregation continue to crowd together 

 until only paired free ends extending from a dense mass can be 

 recognized (fig. 20). When these segments loosen up, the chro- 

 mosomes are found to be growing shorter and thicker. In case 

 a misunderstanding should arise concerning the transition from 

 fig. 21 to fig. 22, separate chromosomes have been drawn (figs. 21.V, 

 2iv). After the bivalents become dissociated and well distributed, 

 eight are readily counted, each one representing 2 somatic chro- 

 mosomes that formerly were end to end, although now they are 

 twisted about each other. At times the 2 members of the bivalent 

 are still continuous, that is, forming a loop; while in other cases 

 they are not, causing the bivalent to be open at both ends. In 

 either case it is not due to the splitting or to a separation of the 

 halves of the spirem thread, since that spirem thread is composed 

 of 16 somatic chromosomes with an end to end arrangement. 



FROM THIRD CONTRACTION TO DAUGHTER NUCLEUS 



About the time the bivalents are more or less distributed within 

 the nuclear cavity, the fibers, which stain blue with Flemming's 

 triple stain, appear outside the nuclear membrane (figs. 22, 23), 

 running parallel to it, the ends extending out into the cytoplasm 

 into which they merge (figs. 23, 24). While these are increasing 

 in number and the nuclear membrane is disappearing, a peculiar 

 behavior of the chromosomes has been noticed which seems to have 

 a definite relationship to phases in the development of the spindle. 

 The nuclear membrane slowly and unevenly disappears (fig. 23), 



