1909] GATES—CH ROMOSOMES IN OENOTHERA 187 
chromosome bivalents are mutually repelled. It is true that in the 
early telophase the chromosomes form a close group, so that they 
certainly cannot be repelled at that time, but may be attracted. 
However, the medium in which bodies float frequently changes their 
qualities of attraction and repulsion, and it appears that the repul- 
sion first develops after the appearance of the karyolymph in which 
the chromosomes float. The facts all suggest that the chromosome 
bivalents mutually repel each other at this time, while the halves of 
these are held together, probably by attraction. 
The studies of Witson (37) and others on insect chromosomes 
show that there are selective attractions between certain chromo- 
somes at the time of synapsis or at some other period of meiosis. 
These of course require something more specific than electromag- 
netic forces to explain them. 
Fig. 22 shows one of the nuclei of a pollen mother cell in the pro- 
phase of the homotypic mitosis. Eleven chromosomes are present, 
_ having the same bivalent structure as in the telophase of the previous 
mitosis. Both nuclei always go through the various stages of the 
second mitosis simultaneously. The method of spindle formation 
has not been studied with great care, but corresponds with the Gladi- 
olus type of multipolar spindle formation (LAWSON 19, 21). There 
is no indication of an intra-nuclear network, the spindle being wholly 
extra-nuclear in origin. A portion of the weft of fibers surrounding 
the nuclear membrane remains im situ when the latter disappears, 
forming a close meshwork, against which the chromosomes lie. The 
fibers then rearrange themselves; the fibrillae from the cones already 
formed come in and become attached to the chromosomes; finally 
the spindle becomes bipolar by the rearrangement of the fibers forming 
the cones. 
The numerous papers on spindle formation in angiosperms need 
not be cited here. Among the more recent critical studies of this 
structure may be mentioned that of BERGHS (2), who insists that the 
distinction between kinoplasm and trophoplasm will not hold; that 
the spindle results simply from the gradual orientation of the material 
of the cytoplasmic reticulum, and returns to a reticulum after mitosis. 
Sometimes spindle formation begins on one side of the nucleus, 
as in fig. 23. In such cases the cones of the multipolar spindle may 
