374 Gates and Rees . — A Cytological Study of 
observed in various plants and animals for many years. Their significance 
as a possible basis for genetic crossing over will be discussed later, and 
important differences from the chiasmatypy described by Janssens ( 1909 ) 
will be pointed out. Actual proof that this twisting of the homologous 
chromosomes about each other is followed in any case by an exchange of 
segments is extremely difficult to get. Inspection of the figures shows 
that the wrapping of the chromosomes about each other is much more 
intimate in some cases than others. Quite probably in many cases no 
exchange of segments follows, but such cases as Figs. 33 and 49 appear to 
furnish definite evidence in favour of the occurrence of an interchange 
o 
of loops. 
During this period the nuclear membrane is often extremely thin or 
quite invisible (Figs. 18-23). That the chromosome pairs differ constantly 
in length in these early stages of their formation, as in the later stages, is 
clearly shown by such figures as 23 and 28. This differentiation in length 
makes it possible to show that the pairing is between homologous 
chromosomes of similar length, and therefore of maternal and paternal 
origin respectively. 
Diakinesis. 
The chromosome pairs, lying almost invariably side by side, and 
usually more or less twisted about each other, undergo progressive 
condensation to form the definitive chromosomes. During the later 
stages of this condensation the chromosomes become very compact, and 
finally (Figs. 37-48) appear as straight or curved rods of different lengths, 
usually showing no indication of their bivalent nature, except in some 
cases by a fork at one end. It is possible that the swelling produced by 
the fixing fluid may obliterate the line of separation between the two halves 
of a bivalent chromosome more completely than in the living condition, 
but in any case the relationship between the two* halves is exceedingly 
close at this time. Clear indications of the twisted condition are often 
observed until the condensation is almost complete (Figs. 33-36). But 
the only trace of the double nature of the definitive bivalent chromosomes 
(Figs. 37-48) is a faint longitudinal split or a fork at one end. 
The nine bivalent chromosomes form a graded series which can be 
arranged in a general way in three groups, three of maximum length (four 
or five times as long as broad), three of intermediate length (about two 
or three times as long as broad), and three very short and almost 
cubical. 
Frequently one of the first three is longer than the other two, and 
the remainder form a series which can only roughly be divided into 
groups. The curved shape of the chromosomes also frequently renders 
impossible the determination of their exact length. Hence the variations 
