466 BOTANICAL GAZETTE [june 



Up to and including the second contraction, a nucleolus is 

 usually present (figs. 17, 18), although some of the drawings fail 

 to show such, as it was either not included in that section, or it was 

 purposely left out, owing to obscuring too great a portion of the 

 other chromatin material. 



During second contraction (tig. 8), the characteristic radiating 

 loops extend from the tangled mass, the first sign of segmentation 

 being seen at the peripheral end of the loop, resulting in the free 

 ends being next to the nuclear membrane (figs. 19, 20). Allen 

 (1) has reported a similar observation in Lilium canadense, but 

 this is contrary to the reports of Mottier (21, 22, 23) and Beer 

 (2), these investigators describing the radiating loop as forming 

 the bivalent. With the loop segmenting at its outer bend, it 

 necessarily follows that either the bivalent is continuous at the 

 lower end, as stated in the description of this stage, or that seg- 

 mentation occurs at both ends, followed by a pairing of single 

 somatic chromosomes. During the period of segmentation and 

 formation of bivalents, the chromatin thread contracts, although 

 not suddenly (cf. figs. 18-24), resulting in 8 thick bivalents being 

 fairly evenly distributed within the nuclear cavity. 



All cytologists agree upon the point that in heterotypic mitosis 

 a bivalent consists of 2 somatic chromosomes, but concerning the 

 mode of formation there is a great difference of opinion. Allen 

 (1), Bergh (3, 4), Gregoire (16), Gregoire and Wygaert (17), 

 Overton (25, 26), Rosenberg (27), and Yamanouchi (32, 33), 

 who claim there is a pairing of somatic chromosomes or spirems 

 in early prophase, state that the bivalent is composed of 2 segments 

 that in the spirem were side by side, reduction therefore occurring 

 by the pairing of 2 spirems; while Beer (2), Farmer and Moore 

 (9), Farmer and Shove (10), and Mottier (21, 22, 23) demon- 

 strate that it is formed by the twisting about of 2 somatic chro- 

 mosomes that previously were end to end in the spirem, thus 

 causing a transverse segmentation to be responsible for the reduc- 

 tion. Allium tricoccum confirms this latter view, and if the series 

 lure given be followed, it will be seen that the arms b and <7 of 

 fig. 19 have not arisen from the separation of 2 approximated 

 spirems of a previous stage (fig. 18), yet doubtless these 2 will 



