Rose Forms as determined by their Cy to logical Behaviour. 165 
diakinesis (figured on PL X, Fig. 29) both^types are determinable with the 
greatest possible ease. 
Compared with similar stages in other plants, subsequent events follow 
an anomalous course, which, however, in the light of extensive studies in 
Rosa ) is found to be more or less characteristic of the genus. In assembling 
on the equatorial plate for the heterotype division, the 14 bivalents mass 
themselves round the central point of the spindle with the utmost regularity. 
Later the univalents follow with less exactness and arrange themselves in 
a complete ring round the paired individuals (Text-fig. 2, b). In good 
sections showing horizontal plates, the total number of chromosomes is 28, 
which agrees perfectly with the somatic count of 42, since 14 bivalents 4- 
14 univalents yield a total of 42. 
The anaphase, too, is noteworthy, for it occurs in two distinct phases, 
one involving the bivalents and the later one the univalents. Often enough 
the former is almost over (PL X, Fig. 33), and even sometimes merging 
into the telophase, before the univalents split and attempt to move. It 
would almost appear that the force urging the chromosomes had spent 
itself with the passage of the bivalents, since the split halves of the remainder 
lag, wander, and occasionally get lost. However, in the majority of cases, 
the greater portion of the split halves unite in the telophase with the 
14 whole chromosomes already awaiting them, to constitute one daughter 
nucleus. Still a few of the laggards mass themselves apart into groups of 
varying size, and finally develop into micronuclei (see PL X, Fig. 34). 
An interesting feature in connexion with the heterotype spindle, in this 
form, is its -close proximity to the periphery of the mother-cell, as is well 
shown on PL X, Fig. 34. 
Very soon after the short interkinetic period preparation is made for 
the homotype division. The fact that many of the split halves of the 
univalents are included in the two major nuclei becomes very apparent 
here, for counts on good homotype plates (PL X, Fig. 38), differing widely 
in number it is true, but often reaching the twenties, can be made. Once 
again events move abnormally. As before, the chromosomes descended 
from the original 14 bivalents travel first to the centre of the plate 
followed by a varying number of halves. The former divide precociously 
and pass to the telophase with such rapidity that before many of the 
splitting halves can approach them the daughter nuclei may be reconstructed. 
Thus there is a general tendency for the major nuclei, representing the 
genuine nuclei of the tetrad, to be built up from 14 chromosomes. 
Curiously enough, the dividing halves derived from the univalents seen in 
the original diakinesis may also show a more or less pronounced inclination 
to keep together, although they may .diverge. When they do act as a unit, 
the ‘ tetrad ’ contains eight major nuclei and may therefore more appro- 
priately be termed an 4 octad ’. On the other hand, if they separate, when 
