20 
W. E. AGAE. 
and 3, 5 and 6, 7 and 8, 9 and 10. None of these are as yet 
in actual contact, but they are in each case united by one 
or two linin threads. With one exception none of the 
other univalents are connected by visible bridges. It is an 
interesting question whether the linin thread which appears 
between two chromosomes shortly before they pair is a new 
formation, or only a becoming visible of a previously existing 
structure. Everything points to the connection being a new 
formation. In PL 3, fig. 24, no connections are visible, and 
in order to see whether they could be brought into view by 
over-staining the cover-slip was taken off and the section 
densely re-stained with Heidenhain’s haematoxylin. This 
process, however, did not result in showing any linin connec- 
tions between any of the chromosomes. The frequent wide 
separation of the two members of a pair is also against their 
being really connected all the time by an invisible thread. 
The linin bridges seem to act as contractible fibres to pull 
the chromosomes together. Heidenhain considers linin to be 
a specifically contractile substance. In thisv figure Nos. 2 
and 4 are probably the large chromosomes still unpaired. 
In PL 4, fig. 27, the spindle apices have nearly reached 
opposite poles, and pairing is almost complete. There are in 
fact only two unpaired univalents (ttl, u2). Of the bivalents 
fifteen are completed rings or “ tetrad ” modifications of rings, 
and three are joined by one point of contact only. One of 
these (Z) is the pair of large chromosomes. 
In Pl. 4, fig. 28, the spindle is practically ready, and 
pairing is complete. Only a few of the chromosomes are 
fig ured to show the characteristic shapes of the bivalents 
and their great difference in size. The large pair, now a 
complete ring, is shown at Z. Some of the gemini are typical 
“ tetrads,” the appearance being due to the pronounced 
transverse constrictions and close lateral approximation of 
the two univalents composing them. 
As will be seen by comparing Pl. 4, figs. 25 and 26, the 
process of pairing does not always proceed exactly pari passu 
with the separation of the spindles. Indeed, sometimes 
