Rose Forms as determined by their Cytologic at Behaviour . 179 
Proceeding, let us submit the contents of the table to detailed analysis. 
The occurrence of diploid, tetraploid, and hexaploid species within the same 
genus would, at first sight, call for little comment, for intensive work like 
that of Tahara on Chrysanthemum has revealed in that genus comparable 
numbers based on 9, Chrysanthemum carinatum having 9 haploid, C. leucan- 
themum 18, C. morifolium 27, C. Decaisneanum 36, chromosomes and so 
on. On the contrary the pentaploid microgenes, comprising fully ninety 
per cent, of our roses, are quite unexpected, and in view of the very obvious 
difficulties in the way of their producing a regular haploid number equivalent 
to half the diploid, their presence and chromosome complement must be 
regarded as quite anomalous. Nevertheless, a critical study of their micro- 
spore development affords some clue to their evolution. However, devia- 
tions from the normal do not end with pentaploid roses ; when the contents 
of the column labelled tetraploid are viewed in the light of the above 
descriptions they present themselves as a very heterogeneous assemblage. 
There not only do we encounter forms in which the course of pollen forma- 
tion is quite as even as in Rosa arvensis upon which we found our compari- 
sons, but in addition microgenes, appertaining to the Villosae, are included 
with their cytology in complete harmony with that of the pentaploid roses. 
For the purposes of this discussion the Villosae may thus be considered with 
the other subsections of the Caninae, i. e. the Eucaninae, Afzelianae, and the 
like. Similarly, the hybrid between pimpinellifolia and coriifolia may be 
classed with the hexaploid hybrid. Furthermore, although the chromo- 
somes behaving more or less regularly are twenty- eight in the last two 
plants, and fourteen in the pentaploid species and the Villosae, both groups 
may be treated as making one whole, since no significant discrepancies 
exist between the two sets ; whatever explains the untypical meiotic pheno- 
mena of the one explains it in the other. 
We are now in a position to take a comprehensive view of the 
cytological features emerging from our work. Almost immediately we 
discover that our local roses conform to two types. Type I, including very 
few species, pursues a course perfectly normal in all its details. Differing 
therefore in no wise from other phanerogams at the same stage it requires 
no further treatment than the above. On the other hand, Type II, com- 
prising the vast majority of our roses, whilst offering differences in detail in 
its various members, follows the same general but untypical plan throughout. 
In it we have displayed a heterotype division, equatorial as far as the bulk 
of its chromosomes is concerned, but reductional with a further fixed 
proportion, generally fourteen, but sometimes twenty-eight in number. 
As is obvious, three recognized hybrids are arranged under Type II, and, 
as far as behaviour goes, they cannot be severed from their companions. 
The pairing of their homologous chromosomes in preparation for the hetero- 
type division, and the number doing so successfully, are too suggestive 
N 2 
