Rose Forms os determined by their Cytological Behaviour. 169 
for disclosing the peculiarities characterizing the meiotic phase in Rosa. 
Examination of its somatic mitosis very soon brought us into contact with 
the unorthodox chromosome number of thirty-five. Although influenced, 
no doubt, by the apparent certainty of Strasburger’s count of thirty-two in 
other roses, and likewise by the improbability of the occurrence of pentaploid 
species, we were inclined to think that our counts were rendered unduly 
high by the presence of precociously dividing chromosomes, by their being 
cut accidentally by the microtome or split by other means. However, the 
somatic number is undoubtedly thirty- five, and Text- fig. 3, a f adequately 
represents the large number of somatic prophases from which we have 
made counts. 
In the meiotic stages, when direct comparisons are made with 
R. arvensis and R. pimpinellifolia at parallel points, whether preceding, 
during, or just subsequent to synapsis, no discrepancies are to be noted. 
There is the same gradual passage from- the reticulum to close synapsis 
yielding in its turn to an equally perfect hollow spireme. Notwithstanding 
this close resemblance here, in the later stages they disagree in almost 
every detail, and to determine the exact import of the striking differences 
manifested the outline of events during meiosis described for R. Sabini 
must be carefully borne in mind. Coupled with this reference to R. Sabini 
must be the fact that this rose is admitted to be of hybrid origin by every 
one competent to put forward an opinion. 
Returning now to R. coriifolia , after the formation of the loops from 
the spireme in the heterotype prophases, . we have, as with R. arvensis , 
a thickening and condensation of the chromosomes, which, however, are 
seen to be behaving dissimilarly. Some are clearly discernible as threads 
folding and interlacing on themselves, as the sides of the individual loops 
close up ; others are just as certainly isolated portions, severed, so to 
speak, from the Contracting spireme. In this fashion, we glean that the plan 
followed adheres rather to that of R. Sabini than to that of R. arvensis. 
Further, the view is strengthened in diakinesis, for the chromosomes in 
sight are very obviously partly bivalents and partly univalents. Nevertheless, 
although the bivalent combinations are decidedly in the minority, no perfect 
count can be made of the precise number of each type present, for the total 
number of chromosomes is too great. 
As the chromosomes pass to the spindle all this uncertainty passes 
away. In a regular manner, observed in one heterotype metaphase after 
another, the bivalents arrange themselves with mathematical precision 
immediately around the centre of the equatorial plate (see PL X, Fig. 30), 
and in doing so reveal themselves as being seven in number (see Text-fig. 
3,/). Next, the univalents approach and encircle them, in general with 
considerable regularity, giving beautifully dear plates, on which twenty- eight 
chromosomes lie visible, the seven inner bivalents standing out as larger 
