636 Fraser . — The Behaviour of the Chromatin in 
relation. At other points isolated loops occur with markedly similar sides, 
united by a swollen dot (Figs. 6, 8). A little later straight segments of the 
spireme may be found lying parallel one to another (Fig. 8), or again curved 
filaments connected by their longer limbs while their shorter limbs lie free 
(Fig. 9). These and other figures, in which the associated strands become 
intertwined and more closely approximated, are found after the loosening 
of synapsis is complete (Fig. 11) and, still more frequently, throughout the 
second contraction phase (Figs. 12-16), until the appearance of the mature 
gemini. As development proceeds, one end of a double loop is frequently 
set free (Fig. 13). The long twisted structures thus produced are 
the precursors of the figure-of-eight-shaped gemini on the heterotype 
spindle. 
The earliest recognizable approximations thus occur immediately on 
the loosening of synapsis, when portions of the spireme (which is not 
necessarily continuous) are first set free from the reticulum. The threads, 
paired and unpaired, soon extend through the whole of the nuclear area, 
but the increasing association between separate filaments no doubt entails 
some pulling about and rearrangement of the whole, till, as pairing becomes 
more frequent and the thread begins to thicken, the chromatin is drawn into 
the loose, irregular knot which constitutes the second contraction. The 
chromosomes, or their precursors on the spireme, run out to the limits of the 
nuclear vacuole but merge into a central tangle, so that each chromosome 
is pretty sure to lie in contact with several others, and attractions between 
allelomorphs as yet unpaired are thus facilitated. A similar arrangement 
has been observed by several investigators, and its interest lies in the com- 
paratively simple opportunity it affords for the association between homo- 
logous paternal and maternal segments. 
The spireme of the second contraction breaks up to form seven gemini ; 
some of these are at first of considerable length (PI. XLIV, Fig. 16), and are 
twisted hither and thither in the nuclear areas ; they undergo the usual 
shortening and thickening (Fig. 17), and, by the time that they are 
arranged on the spindle, they are relatively small and stout, and of more 
or less uniform size (Fig. 18). 
The Nuclear Membrane and Nucleolus. 
Until the formation of the gemini the nuclear membrane shows no 
marked alteration in appearance, while the cytoplasm, which is at first 
a uniform, finely granular mass, increases in density around the nucleus 
and leaves the periphery of the cell relatively empty. By these changes, 
and by the subsequent rounding off of the pollen mother-cell, the succession 
of stages in the nuclear history can be roughly gauged. 
As the gemini shorten and thicken, the cytoplasm encroaches on the 
nuclear area and the limiting layer loses its definite character, till, on the 
