THE SPERMATOGENESIS OF LFPIDOSIKEN PARADOXA. 19 
witli certainty, but soon this makes its appearance, and at the 
same time the chromosomes begin to pair again to form the 
rings, some modification of which nearly all of them assume 
on the equatorial plate. This secondary pairing of the chromo- 
somes, which takes place in exactly the reverse way to the 
previous separation, is illustrated by PL 4, figs. 25-28, which 
also show the mode of development of the spindle figure. It 
will hardly be necessary to say that the second pairing unites 
the same chromosomes that conjugated in the first instance; 
this is shown clearly by the equality of the members of each 
pair. 1 
In PI. 4, fig. 25 the spindle axes are only separated 
through an angle of about 45°. Of the thirty-eight chromo- 
somes, twenty-six are still separate univalents, but the remain- 
ing twelve have joined to form six bivalents. In the case of three 
of them (61-53) pairing is complete, and they form closed rings. 
In the case of the other three (64-66) they have at present 
united at one end only, forming bivalent quadripartite 
bands. Presently the open bivalents so formed will bend 
round and form closed rings also. 
This nucleus, like those in PL 4, figs. 26 and 27, is in a 
thick celloidin section, and was carefully drawn from both 
sides to verify all the chromosomes. In fig. 26 are shown 
the two drawings of another nucleus obtained in this way. 
In both figures all the chromosomes are shown, but those 
in the upper optical section are shaded, and those in the 
lower half shown in outline only. 
In this nucleus, although the spindles are further separated, 
pairing has not proceeded so far as in PL 4, fig. 25. The 
thirty-eight chromosomes can be plainly counted, and of 
these only two have completely joined to form a ring (ft). 
There are four couples in an early stage of pairing — Nos. 1 
1 The equality is only approximate, but the differences in size are 
slight enough to be accounted for by the condensation having proceeded 
at slightly unequal rates and by slight irregularities in outline. At any 
rate, in the case of the large pair, these are so much greater than any 
others that there is generally no possibility of mistaking them. 
