Hybridism and Classification in the Genus Rosa. 169 
probability of the presence of various combinations of Mendelian 
unit characters in the several groups than to give an extensive and 
involved analysis on the basis of a large number of separate 
characters. The argument throughout is, of course, entirely 
hypothetical and experimental results are required to establish the 
hypothesis. It is impossible to over-emphasise the need for 
cultural work, so intensive has the study of British plants become, 
and, as already pointed out, morphological methods alone are 
insufficient for the complete solution of the intricate problems 
involved. Yet, it is remarkable how little has been done to 
discover experimentally the genetic relationships within our 
numerous “ critical species ” and genera. The work of Trow (1912) 
on Senecio vulgaris Linn, clearly points to the need for similar studies 
on other aggregate species. 
Almquist, whose work has already been mentioned, makes a 
very interesting attempt in his account of Rosa , in Lindman’s 
Svensk Fanerogamflora (1918) to refer the micro-forms of the genus 
ultimately to certain characteristic types which are based on the 
following characters:—glaucousness, greenness, glabrousness and 
hairiness. “ Species ” with glaucous leaflets are referred to gl. 
(glauca) if they are glabrous, to gif. (glauciformis ) if the leaflets 
are hairy. Green-leafleted “ species ” are referred to vir. ( virens) 
if the leaflets are glabrous, to virf. (virentiformis) if hairy. This 
plan is applied throughout the eight sections and thirty-one 
species-types into which the genus has been divided. Thus, any 
rose whose sectional and type characters have been determined 
finds a place in Almquist’s scheme according to whether it is a gl., 
gl/., vir., or virf. form. This method of segregation has involved 
the creation by Almquist of a large number of new names, 
especially in the Afzeliana (R. glauca X coriijolia) and Caninae 
sections, where combinations of these characters are apparently 
frequent. But it is the scheme itself which claims one’s attention 
and interest. Almquist contends that the chief characters in 
classifying roses are the colour and consistence of the leaves, the 
form of the leaflets and the shape and direction of the teeth. He 
pays little attention to the direction of the sepals, while biserration, 
the development of glands, etc., are regarded as modifications of 
primary types. There is much in favour of Almquist’s views 
and I think his method must result at least in bringing together 
similar-looking plants. R. lutetiana and R. dumetorum, for 
example, find a place side by side in the Caninae section under 
species-type cuneatula, the former a vir. form, the latter a virf. 
