THE SPERMATOGENESIS OF LEPIDOSIREX PARADOXA. 21 
nuclei are found in which pairing is complete by the time the 
spindles have separated through a very small angle only. In 
general, however, the less the time that has elapsed since 
the disappearance of the nuclear membrane the larger the 
proportion of univalents, and the nearer the spindle is to 
completion the greater the number of bivalents. 
'the following is a summary of the behaviour of the 
chromosomes up to this stage. They appear out of the resting 
nucleus as the leptotene threads. These are intimately 
united in pairs along their whole length in zygonema. In 
strepsinema they separate again, but for a time remain attached 
to each other by their ends to form long rings. These rings 
open, first at one point of contact of the conjugants, then at 
the other, and the somatic number of univalent chromosomes 
is again obtained. Each univalent as it shortens becomes 
transversely constricted. After the dissolution of the nuclear 
membrane and appearance of the achromatic figure the 
chromosomes that were previously united pair again. This 
pairing takes place by a process exactly the reverse of 
the previous separation. First, a linin thread appears con- 
necting a pair of corresponding chromosomes; then these 
come into contact by one end to form a rod-shaped (quadri- 
partite) bivalent ; then the rod bends round on itself to form 
a closed ring* or modification of a ring. 
In the metaphase of the first meiotic division one member 
of each bivalent goes to one daughter-nucleus, the other to 
the other. 
The chromosomes on the equatorial plate present the same 
shapes as shown in PI. 4, fig. 28. In PL 4, fig. 29, is seen 
a polar view of this stage, to show the nineteen gemini. 
Pi. 4, fig. 30 shows an early anaplmse, to show the mode of 
separation of the chromosomes. The large pair is well seen. 
None of the chromosomes as yet show the longitudinal split 
preparatory for the second division, which is so often 
observed at this stage. 
The Second Meiotic Division. — There is no resting- 
stage between the two divisions. In PI. 4, fig. 31 we see 
