Meiotic Nuclear Divisions of Galtonia candicans . 741 
but in either case they closely resemble the univalent chromosomes when 
they have but newly separated after second contraction, and are still united 
at one end (PL LXII, Fig. 61). The univalent lengths of spireme may show 
every degree of union with their homologous pair. They may, as has been 
shown, be arranged end to end (PI. LXI, Fig. 44), and certainly at a slightly 
later stage they may be joined side by side (Figs. 52 and 53), and there 
may be every intervening degree of connexion between these two extremes. 
Or they may yet be entirely independent one of the other. Although the 
secondary union of the individual strands is ultimately achieved, each 
strand is primarily univalent, and it is only by a bending over and fusion of 
univalent lengths that the bivalent segments are formed. 
Thus it appears that the parallel portions of linin which entered the 
synaptic knot, and represented the condensing longitudinal halves of sections 
of somatic chromosomes, have, during synapsis, become concentrated and 
joined end to end to form lengths ofwhole somatic chromosomes. The concen¬ 
tration is not always complete, for the space between the longitudinal halves 
reappears in the post-synaptic stages as longitudinal fission. Further, that 
during synapsis the rearrangement may have been still more extensively 
elaborated, and that the homologous univalent lengths of spireme may 
have joined in pairs (Figs. 44 and 45). Overton ( 24 ) (1909) has stated that 
during synapsis there may not only be a conjugation of parental chromo¬ 
somes, but also ‘ an actual interchange of influence ’. 
As the loops and strands come out of synapsis they distribute them¬ 
selves throughout the nuclear cavity (Pdgs. 46 and 47). Their anastomosis 
and coiling is most intricate. Sometimes the spireme lengths appear to be 
joining end to end (Fig. 51). Sometimes for a distance they are united 
side by side (Fig. 49). The nucleus then moves to the centre of the cyto¬ 
plasm which surrounds it (Fig. 51). The nucleolus has generally by this 
time returned to the nuclear boundary (Figs. 46 and 47). There it may 
remain throughout the subsequent stages until its final dissolution at 
diakinesis (PI. LXII, Fig. 64), or it may be carried out by the spireme as it 
emerges from synapsis, and then take up a central position in the nucleus 
(PL LXI, Fig. 52). During the loosening of the knot it sometimes stains 
for a time chromatically (Figs. 46, 47, and 48), but when the ‘hollow-spireme ’ 
stage is reached, it once more becomes colourless (Figs. 49, 51, etc.). 
The nuclei of the ‘ hollow-spireme ’ stage again show every degree of 
variation. In some the loops and strands may consist of beads of chromatin 
(Fig. 47), in others they may be more ribbon-like (Fig. 49). Possibly this 
apparent difference in the composition of the spireme may be partly 
accounted for by the greater or less strain exerted on the nuclear contents. 
Again, the loops may lie more or less freely in the nuclear cavity (Fig. 47), 
or there may be an intricate anastomosis (Fig. 48). The anastomosis shows 
that the combinations necessary for the formation of the bivalent chromo- 
3 E 
