Darling: Sex in dioecious plants 185 



As growth continues more of these bodies come from the 

 nucleolus, often two at a time ; the exact number, however, could 

 not be determined because those coming off first lose their identity 

 before the later ones appear. It seems certain that these bodies 

 are coming from the nucleolus rather than going into it, for in the 

 same section the nuclei at the bottom of the anther, which con- 

 tains the youngest stages, have bud-like processes just appearing, 

 while farther up they are merely attached, and in the uppermost 

 parts they have migrated to the wall of the nucleus. This very 

 marked progressiveness in development in the same anther is of 

 prime importance in considering the successive stages. 



These chromatin bodies differ somewhat in size ; and as they 

 continue to appear the denser portions of the linin take the chro- 

 matin stain more deeply, and soon these nodes begin to disappear 

 and the thread becomes more or less continuous (fig. id). Owing 

 to the size of the thread at this stage, I could not determine whether 

 it was split or not ; it appears somewhat jagged or irregular in 

 places, often bead-like, as frequently noted by other observers at 



this stage (figs, jo, ii). 



The chromatin bodies usually cease to come off from the 

 nucleolus at about the time when the spireme thread is well formed 

 and takes the chromatin stain deeply. Synapsis now follow^s ver> 

 quickly and the thread always appears to be associated with the 

 nucleolus (fjg, 12). After this stage the thread unfolds and there 

 appear thickenings upon it ; often two threads may be seen to be 

 parallel and sometimes united for a part of their length. The 

 thickening thus formed becomes more condensed and finally the 

 thread segments ; successive stages show that this is the method 

 of formation of at least some of the chromosomes (figs. 15-18). 

 The details in the formation of the chromosomes could not be deter- 

 mined with accuracy; however, it appears from the study of many 

 preparations in this stage, that many if not all are formed by the 

 approximation and fusion of two threads. Eight chromosomes 

 are formed by the segmentation of the spireme thread (figs. 16- 



. There can be no doubt about the number thus formed ; for 

 in scores of uncut nuclei only eight could be detected during these 



successive stages. 



Just previous to the formation of the synaptic knot, a large 



