Hybridism and Classification in the Genus Rosa. 165 
a biserrate form, R. glauca var. denticulata Kell., has been recognised. 
R. subcollina (Chr.), as at present understood in this country, is a 
slightly larger group, and includes the type of Christ, which is 
uniserrate, and at least three biserrate forms, viz. R. subcorii/olia 
Barclay, R. ccesia var. incana Bor. and R. pruinosa Bak. It would 
be possible to make an analysis of the subcanina-subcollina series on 
the lines already followed for preceding groups, though a considerable 
number of combinations would stand without names. But since 
the groups have not been at all fully worked out, it does not follow 
that the combinations do not occur in nature. R . subcoriifolia 
Barclay is in itself a somewhat aggregate species. Its commonest 
form has subfoliar glands and smooth peduncles, and is thus an 
HBGs form, corresponding to var. Lintoni of the coriifolia group. 
But HBgs and HBGS combinations have also been met with, 
although they have not received distinctive names. 
I have, however, a suggestion to make regarding the sub - 
canina-subcollina series. R. subcanina is, in a number of features, 
intermediate between R. glauca and R. canina. Christ, in his 
description of the plant, remarks that it connects the two. In the 
possession of reflexed or downwardly spreading sepals it comes near 
R. canina t but in the woolly head of styles it bears a closer 
resemblance to R. glauca. In the length of the peduncles, the time 
of ripening of the fruit, and the persistence of the sepals it is more 
or less intermediate between R. glauca and R. canina and one is 
tempted to ask if R. subcanina may not be a hybrid between these two 
species. Again, R. subcollina as an aggregate is more or less inter¬ 
mediate in its characters between R. coriifolia and R. dumetorum. 
It holds an essentially similar position between these two aggregate 
species as R. subcanina does for glauca and canina , and one 
wonders if it, also, is not of hybrid origin. Experimental work is 
here extremely desirable. So far, ae I am aware, there is nothing 
in the geographical distribution of these forms against the theory 
here suggested. Both R. subcanina and R. subcollina are most 
frequent in Scotland, becoming much rarer in the north of England, 
and they are quite absent in the south. Their distribution thus 
follows very closely that of glauca and coriifolia. R. canina 
and R. dumetorum , are, of course, abundant and widespread. It is 
noteworthy, also, that the two suggested hybrids, while they have 
the same distribution as R. glauca and R. coriifolia , are rarer than 
either of these two species. 
