386 Gates and Rees. — A Cy to logical Study of 
that the loops are radial or irregular in position instead of being polarized, 
and when the bouquet stage is accompanied by parasynapsis it of course 
differs also in the constitution of the individual loops. 
The complete loops separate from each other at the base in the 
segmentation of the spireme. Each bivalent chromosome is thus constituted 
from the two arms of a loop. This structure condenses greatly, and it 
appears that in some cases at least the torsion remains. If this account of 
the formation of the bivalents is true, and we can see no escape from it, 
then synizesis has no part in bringing about the pairing, and its significance 
as a unique physiological condition of the nucleus remains entirely obscure. 
In the later stages of diakinesis the nine bivalents ‘appear chiefly as 
rods of different lengths, with sometimes a straight line of fission between 
their two halves. In many cases even at this time the two halves are so 
closely associated that no line of separation between them can be observed. 
On the heterotypic spindle these bivalents become still further condensed, 
but Fig. 58 (PI. XVIII) clearly shows how in the heterotypic mitosis the 
longitudinal halves of the bivalents are gradually pulled apart, the 
separation beginning at one end and proceeding to the other. Whether 
this line of separation is the same as the line of approximation of the two 
arms of a postsynaptic loop forming respectively the paternal and maternal 
members of a bivalent chromosome, depends upon whether the twisting is 
undone again before the final condensation of the bivalents takes place. 
In any case this mitosis is essentially a reductional one, although if the 
separation is across the line of the twists it would only be reductional as 
regards the individual segments and not as regards whole chromosomes. 
This method of reduction differs from that in Oenothera (Gates, 1908) 
only in that in the latter the pachynema segments directly into a chain 
of chromosomes which are so short and stout that the members of each 
pair simply come to lie side by side after the chain segments into pairs 
connected at one end, since there is no opportunity for them to twist 
around each other. In Galtonia (Digby, 1910) and Primula (Digby, 1912) 
the course of events appears to be still more closely similar to that in 
Lactuca. We therefore regard the telosynaptic course of events as clearly 
established for a number of plants, but we see no necessity for regarding 
it as universal for the whole Plant Kingdom. Some years ago it was 
suggested (Gates, 1910) that both the telosynaptic and the parasynaptic 
methods of synapsis may occur, the latter perhaps more largely in forms 
with long thready chromosomes and the former in forms with short and 
stout chromosomes. Ten years ago the senior author had the opportunity 
of examining critically the beautiful preparations of Janssens (1905) of 
Batrachoseps , and was convinced, as Wilson (1912) and others have been 
since, that an approximation of filaments or univalent spiremes, accompanied 
by their polarized arrangement, takes place in this form. Janssens’s figures, 
