154 
J. R. Matthews. 
descriptions nor with one another. If the creation of new names 
continues at the present rate and according to the present system 
or want of system, the time will soon arrive when it will be 
impossible to say more of any member of the genus than that it is 
a rose. To make a determination and to give a binomial will 
become the work of a few specialists and there need be little 
hesitation in predicting that no two specialists will agree. 
There is, no doubt, a strong tendency to place an ever finer 
construction on the idea of species whenever a genus is submitted 
to critical investigation and it is contended by many that the 
Jordanian practice of splitting the “ Linnean species” into 
innumerable forms or micro-species has, in many instances, been 
carried to an extreme. Whatever the advantages or disadvantages 
of this practice may be, it brings us face to face with the problem 
of deciding what forms are to be regarded as worthy of specific 
rank. On this question, in so far as it concerns the genus Rosa, a 
unanimous opinion is not likely to be forthcoming, for “ species ” 
of Rosa must, in the present state of our knowledge, rest largely on 
differences which are in themselves artificial, and the chief difficulty 
is to decide what are really important and what are unimportant 
differences. It is gradually being realised that external morphology 
alone is inadequate for the solution of such problems, yet it is 
remarkable that still very few “ critical species ” among British 
plants have been subjected to the test of experimental investigation. 
Only by culture, combined as far as possible with cytological study, 
will it become possible, I think, to determine finally the genetic 
relationships of the numerous micro-species into which old, well- 
known species like R. canina Linn, have been split. 
Regarding purely morphological investigation, it may be of 
interest to refer here to the work of Parmentier (1898), who 
endeavoured to throw some light on the problem by following the 
anatomical method. This worker, after an exhaustive enquiry into 
the detailed structure of the plants, divided the genus into primary 
species, morphological species, morphological sub-species and 
secondary forms. But this does not appear to have produced a 
better or more natural classification than that adopted by 
systematists. We find, for example, that while R. canina L. ranks 
as a primary species, R. lutetiana Lem., R. dumetorum Thuill. and 
R. coriifolia Pr. belong to the third category and R. glauca Vill. is 
a mere secondary form. Whatever the value to be attached to 
these forms of the aggregate R. canina , systematic students are 
