Rose Forms as determined by their Cytological Behaviour. 177 
later in the ordinary fashion into an apparently continuous spireme. When 
this breaks up its units soon reveal themselves as of two distinct types : uni- 
valent rods and twisted bivalent loops. This peculiarity was first observed 
in Rosa omissa, and afforded the clue which led to the discovery of the par- 
tial reduction in pentaploid and tetraploid alike (PI. X, Fig. 28 a). At the 
inception of the spindle the bivalents are sharply marked off from the uni- 
valents by their early passage to the equator, and even by separating ere the 
latter reach it. Here we think it necessary to remark that, in spite of this, 
equatorial plates with all the chromosomes practically in one plane are much 
more frequent in tetraploid than in pentaploid roses. 
The bivalents diverge to the poles first and form a kind of cap, whilst 
the others divide and keep together in twos and three to pass slowly towards 
them. In the Villosae nearly all, and indeed sometimes actually all, attain 
m 
'R.coeruie* T*.vn»Ui S 
An 
Ms 
If 
■R-Omissa 
#♦ •• 
Ife 
$ 
R &uber»eci-A 
OJV; m S 
* * to* 
'R.Omrv&Sa. 
% ♦ 
R.coerulea 
’R.t^cjUis 
e 
'R/St^erATf-dV. 
"R.Suberecta 
Text-fig. 5. Section Villosae. Somatic and heterotype metaphases, x 3,000. 
that goal, and are included in the daughter nuclei ; occasionally one or two 
may be left out and yield micronuclei. 
The homotype divisions begin in much the same way as the heterotype. 
Usually, however, only the seven descendants of the bivalents group them- 
selves at the poles, for the others delay their splitting so long that they fail 
to join them, thereby necessitating the formation of accessory nuclei. If 
this section manage to keep together, the extra nuclei may have as many 
chromosomes as, or even more than the normal nuclei formed at the cap, 
but this occurs only rarely. In general we are inclined to think that nearly 
always they contain less, and only microspores derived from the cap are 
capable of growth. The cytoplasm of the multinucleate body (better 
described as an octad) then segments more or less evenly into eight parts, 
each containing one or more nuclei. 
As demanded by the chromosome counts made during meiosis, every 
Villosa microgene gave a somatic count of twenty -eight (Text- fig. 5, a-e), 
N 
