170 
J. R. Matthews. 
The need for some such system in dealing with our British 
Roses is obvious, hut an entirely new classification should not he 
attempted until the support of experimental evidence is forth¬ 
coming. Cultural results will not be obtained without the expen¬ 
diture of much time and patience, but even if a single aggregate 
species like R. canina L. were submitted to systematic crossing 
and fully worked out along the lines of Mendelian research, we 
should obtain results, I think, which would form areal contribution 
to our knowledge of this most difficult genus. 
I do not wish, at present, to enter very far into the controversy 
as to how many names should he retained or how many more 
should be invented. If it be proved that elementary species arise 
and remain stable it may be desirable to name such forms as 
the natural units of classification, but in the present state 
of our knowledge, the creation of innumerable names in a genus 
like Rosa, seems to be devoid of all scientific principle and is 
much to be deprecated. I would at present strongly support any 
systematist who returned to the Linnean conception of the species, 
denoting its various combinations by some purely symbolical 
method. 
In conclusion I wish to mention that the present contribution 
has grown out of a correspondence with Mr. A. G. Tansley, F.R.S., 
on some of the questions that have here been discussed, and I 
gladly take this opportunity of expressing my thanks to him for 
the great interest he has shown in its preparation. 
Postscript. 
Since the above paper was written Colonel Wolley-Dod has 
made a further contribution to the study of the genus under the 
title “A Revised Arrangement of British Roses,” published in the 
Journal oj Botany , 1920. The author has reached the conclusion 
that “ most of the very detailed descriptions of D^sdglise, Ripart 
and others can hardly be other than those of an individual bush 
or specimen, which cannot be completely matched by any other,” 
and in consequence of his conviction of the worthlessness of nearly 
all the so-called species of these authors, he has reduced the 
number of names having full specific rank to eighteen. We find, 
for example, in the large subsection Eu-caninae only five species— 
R. lutetiana , R. Blondceana, R. dumetorum, R. Reuteri and R. 
tomentella. It is true the author still retains a formidable list 
of names under most of the species, but these are never accorded 
more than varietal rank and many are regarded simply as forms. 
