SOME EOSES FROM DORSETSIIIKE 17 



the pistils projecting 3 mm. ; the sepals folded down ; no chanii^e in 

 the coU)ur of the foliag-e. In November, I received ripe liips, with 

 tough, leathery coat, cherry-red (Dauthenay, liepert. de Couleurs, 

 J)!. 1)1, tone •!<), 15 to 18 mm. long b}^ 9 to 11 wide, pedicels 18 to 

 20 mm. ; disc dark brown, projecting ])istils 2 to 8 mm. long. One 

 which I opened contained G well-formed akenes (6 mm. by 4) in 

 addition to as many imperfectly developed. 



When I came across these bushes, their ap})earance struck me as 

 utterly unlike anything I had seen before, I was as much startled as 

 when I first saw R. rubrifolia alive, and 1 have since been unable to 

 identify the species from the descriptions of British and Continental 

 authors or after a diligent search through the herbaria in the British 

 and Paris Museums, at Kew, and in the Rouy herbarium in Paris, to 

 whicli Prince lioland Bonaparte has kindly given me access. 



This Kose falls under the detlnition of the Sfylosce section of the 

 genus, whatever the value of the grou]^ may be as a natural associa- 

 tion. I have compared it with R. sfi/losa, of w^liich the Deseglise 

 herbarium contains an example, from the Dept. Vienne, sent by 

 Desvaux as absolutel}^ typical. In this example, the folioles, 5 or 7 

 in number, are also sessile, although Desvaux describes them as 

 petiolulate, and bear 15 to 18 teeth on each side, but they are very 

 haiiy above as well as beneath, and more attenuate at the base. Of 

 Ji. IcHcochj'oa, I have seen an authentic specimen from Poitiers 

 (Desvaux, April 181-3) at Kew ; it has broader folioles, pubescent on 

 the ribs onh^, and the flowers are said to emit a pronounced odour of 

 musk. The t^^De of the var. micropliyUa Kou}^, from L'Arboie, Isere, 

 seems to me to represent nothing more than a small-leaved form of 

 R. sti/losa ; folioles pubescent above, not glabrous as stated by Kouy, 

 with 18 to 22 serrije on each side ; sepals longer than the petals, which 

 are stated to be pink ; pedicels smooth. R. systyla, which I have 

 found at Studland, is a high bush similar in habit to the larger forms 

 of R. canina, with larger pink flowers, much larger leaves with 

 petiolulate folioles, projecting beyond the flowers, as in the specimens 

 of R. sfylom and R. leucocliroa figured by Desvaux as well as in all 

 our Roses of the R. canina and R. sti/losa groups ; my specimens 

 agree well with some in the Deseglise collection {R. fastigiata Bast.), 

 also with Christ's description in Rosen der ScJiweiz, as well as with a 

 specimen from the Basle Jura sent by him to Deseglise. 



I cannot suggest a hybrid origin for this remarkable form. The 

 only llose-bush growing quite near was, as I have said above, a 

 R. rithiginosu ; not far off there were bushes of R. canina which seem 

 to fall under the var. dumalis, with white flowers and mostly 7- 

 foliolate leaves. Bechstein's R. dumalis in the strictest sense has 

 rose flowers and 5-foliolate leaves on the flowering branches. 1 am 

 well acquainted with it in Belgium (Famenne), where, owing to its 

 more shiny and darker foliage and smaller flowers, I can usually 

 distinguish it at a distance from the typical R. canina {luletiana) 

 in hedges where the two grow side by side. White-flowered var. 

 dumalis also occur in the same district in Belgium, but are the 

 exception and not the rule as round Studland, where the half-open 

 petals are often yellow at the base. There is no R. arvensis in the 

 Jouhjsal of I3ot.ln'y, — \()l. 58. [Ja><'Uaei', 1920.1 c 



