66 THE SUBSECTION EU-CANINZ OF THE GENUS ROSA 
Sacer clipes, acuminate, usually Sige, rarely ilose, 
compound. 
quite entire on edges. Flowers flattish, reddish. Styles subin- 
cluded. Stigmas in a villous head. Fruit broadly ovate or sub- 
globose, Kavter red.” 
The e almost as many descriptions of R. duwmetorum as 
oe : authors to describe them, all of which descriptions differ 
s from one es just as ieee ens do, and it has 
a p ieenidatls list of synonyms. I have reproduced ‘four a the 
descriptions, viz. Thuillier’s, the caligieial author, supplemented by 
that of Déséglise, with a ong s and Wood’s to show the British 
inte: 
ing this epseies hun obtusi wyolie, and, I might add, from 
_ R. urbica, while some of ihe: forms run very near R. pesifelin Its 
1 in size, but less so in sha: authors 
n 
pecimens. In this respect they are intermediate between those 
of R. obtusifolia and RK. urbica, being less broad and less obtuse 
than in the former, and less acute than in the latter, but there are 
many exceptions on all sides. They are quite simply serrate, the 
teeth being usually rather broad, shallow, and subobtuse, so as to 
be Ave It will be noticed that Thuillier makes no mention 
of the serration. The prickles are less robust than eee of wrbica 
or sipactoke, sometimes they are quite weak and even absent 
altogether, especially on the petioles. The sapene character of 
their being binate under no leaves may be seen in some speci- 
mens, but so they may in other species throughout the whole 
subsection. The Lew tiles are usually more oe pectin or 
n in average obtust- 
duneles are ‘iirly einer f slightly pubescent but 
not Y eolictanney so, and this charact be found in other 
species. I have not, however, observed it either in wrbica or 
obtustfolia. Woods’s specimens show this hairiness well. All 
aoe indicate a fruit approaching to subglobose, but specimens 
show it to be more often ovoid, me even elongate so that in this 
organ He is nearer to f. wrbica than to RB. obtusifoia. I cannot 
se 
aa 
g 
“lo 
J 
eh 
ech remarks that there is tienes doubt as to what form 
should be designated R. dumet a om . name now 
appears to be used only in an aggregate e also remarks 
t herbarium specimens are not codons “distingaichable from 
B. obiusifolia Desv. 
