158 
J. R. Matthews. 
numerous violets which he determined as natural hybrids between 
certain species. These plants were taken into cultivation and 
found to be more or less sterile, but the small number of seeds 
obtained from cleistogamic flowers gave rise to plants showing 
segregation in regard to several features. “ It is certain,” according 
to Bateson, “ that segregation in countless instances plays a part 
in the constitution and maintenance of characters held by 
systematists to be diagnostic of species. One has only to glance 
over trays of birds’ skins, the portfolios of a herbarium, or drawers 
of butterflies and moths to discover abundant ‘ species * which are 
analytical varieties of others. The principles of heredity we trace 
in our experimental breeding are operating throughout the natural 
world of species.” 
If segregation occurs in the genus Rosa, and it is difficult to 
believe that it does not occur, we might expect to find a large 
number of visually distinct forms showing various combinations of 
Mendelian unit characters such as hairiness, leaf serration, 
glandularity, glaucousness, etc. Moss (1912) has pointed out that 
the numerous forms of the highly variable species Stellaria 
dilleniana Moench. may be regarded theoretically as various 
combinations of several unit characters such as glaucousness or 
greenness, large or small petals and many-flowered or few-flowered 
cymes. 
Before attempting to make a theoretical analysis of some of 
the British species of Rosa on the basis of a few separate 
characters such as have been mentioned, it will be useful first to 
indicate the groups into which the genus may be divided. For this 
purpose the following table, which presents the sections and groups 
more or less in accordance with the views generally held by 
systematists, has been drawn up. 
It will be seen from the dates of publication included in 
brackets in Table I that the term “ group ” generally, though not 
always, corresponds to one of the older aggregate species. The 
numerous sub-species and varieties of these aggregates which have 
since been described I shall attempt to analyse as far as is possible 
on the basis of a few characters commonly employed by 
systematists for the differentiation of these forms. 
Rosa arvensis Hudson, FI. Angl. 1762, is the British 
representative aggregate species of the Section Synstylae in which 
the styles are combined into a long slender column equalling the 
stamens. A number of forms (the term is used throughout in a 
