THE SPERMATOGENESIS OF LEPJDOSIEEN PAEADOXA. 11 
safely assume that the chromatin threads are all pointed 
to that part of the cytoplasm in which is embedded the centro- 
some, or, at any rate, the kinetic centre of the cell, though I 
could not be sure of this structure at this stage in Lepido- 
siren. 
PI. 2, fig. 12, shows a zygotene stage (rather more advanced 
than PI. 1, figs. 10 and 11), seen from the pole. In this 
nucleus the free ends of the fusing pairs are many of them 
embedded in a rather faintly staining coagulum, a condition 
which we meet with again, though not so pronounced, in 
PI. 2, fig. 14. 
In PL 1, figs. 10 and 11, it is well seen how widely 
diverging the conjugating leptotene threads are as they pass 
away from the nuclear pole — a condition hard to reconcile 
with the view that we are here dealing with a re-fusion of 
temporarily separated daughter-halves of the chromosomes. 
This view meets with still further difficulties when we trace 
back the fusing threads of the zygotene stage to the leptotene 
threads of the preceding stage, which, as we saw, are not 
arranged in any paired way at all. 
Some authors have adduced in support of the theory of 
parallel conjugation the fact that no threads of an inter- 
mediate thickness between leptotene and pachytene filaments 
can be found. Others, having found that in the forms studied 
by them such intermediates are present, have considered 
this fact as strong evidence against the existence of this 
mode of conjugation. In Lepidosiren it is certainly true 
that it is not always possible to tell whether a given chroma- 
tin thread is of the leptotene or pachytene order, but this is 
no evidence against the view of the formation of the pachytene 
threads by fusion of pairs of leptotene filaments. From the 
moment of their first appearance the threads are continually 
shortening and thickening, audit is probable that conjugation 
does not always take place at precisely the same stage of 
contraction. Moreover, the intertwining and stickiness of the 
threads seems to cause them to contract unevenly, so that 
the same filament may be thicker in one part where its 
