GEOUPS OF SPECIES. 



17 



a course without explaining fully, as is done in the notes following the individual specific 

 descriptions, the reasons for adopting the view that is here published. The no* 

 application of the canon affects, but only slightly, the numbers and proportions in the 

 distribution tables so far as the European province is concerned, since for it the recent 

 clear and careful account of the European species by Herr Steininger has be n closely 

 followed, and since that author separates as species forms that are capable of different ion 

 only by characters that a wider study of Siberian, Chinese, and Himalayan ones shows 

 to be variable within the limits of the genus. For example, P. lusitanica is speci6cally 

 separable from P. sylvatica only by relative characters, and by the present canon can 

 only be a variety of the latter; and P. Frid 'erici- Aug us I i ', which Caruel 1 declares to be the 

 same species as P. petiolaris, is kept apart by Steininger 2 (who cannot understand why 

 this reduction should be proposed) only by two or three vague differences of habit 

 and by the different colour of corolla, which last, though certainly a striking character, 

 has not always a really diagnostic value, as the evidence of P. me g alantlui and some other 

 species very distinctly shows. 



But, as Sir J. D. Hooker has clearly shown, 3 species are by no means units of 

 equal value. This is especially true of the " species " of Pedicularis, however rigid be 

 the canons applied in denning them. Clearly species that differ from each other so little 

 as does P. carnosa from P. corymbosa, or P. Clarkei from P. Prainiana, or P. polyphylla 

 from P. tenuisecta— the first of each of these pairs has a beaked, the other a beakless 

 corolla, while the members of each pair agree so closely as to habit, calyx, capsule, and 

 seeds that the verbal description of all these characters has in both to be ide 

 are more closely related to each other than species like P. Oedcri and P. pyenantha, 

 which with very similar corolline characters differ in habit so fundamentally that the 

 first has alternate leaves with a continuous spiral centrifugal inflorescence, the other 

 opposite leaves with an interrupted verticillate centripetal spike. 



In Pedicularis, as in Thalictrum, in Rubus and in Saxifrage this arrangement of the 

 separable forms in evidently natural groups is a characteristic feature. And it is probable 

 that these groups, rather than the forms here termed " species," constitute the natural 

 units of the genus. It is therefore of some importance to recognise ar ^ " ' " " - 

 of these groups, since otherwise a system of classification might be adopted that m its 

 application divorced naturally allied forms. The evidence as to the composition of these 



i « » • • mnnTT ,. OU np^« the converse of that which enables us to define 



natural " groups" is in many respects tne cou 



. „ x i " ^„™ +Tiof flo-ree m the greatest possible number of 



" species." In such groups forms that agree i g r ....... 



, i t^ x ~*iw».. /> n«pniipntlv the characters that prove least reliable m 



characters are brought together, Consequently , . .. . . L . . v , u 



d define the 



ax. xu„f oro nf orpatest value in distinguishing the 

 fie diagnosis are those that are ot greate I * j 



Hah 



d 



r i ^oropfprs- and it is as indicative of affinity bet 

 psule now give useful characters, dim » / 



allied' species rather than as diagnostic of separable ones that seed-characters are chiefly 



I fta/niAl in Parlatore* Flor 



Steinioger in CTft-n. «- **»* *** «"«* " *£ 



a gnosis 



it 



Frider 



ootstock with stouter roots 



never 



bracts .hich are longer than the calyx, by the colon, of the ««*••<*£ 



he spike, by its differently- shaped 

 ter adds : — " It is very difficult to 



Friderici'Juqusti." But for those 



who have e 

 conceive how wi 



specifically. 



w it is possible for Uuraei w P™F- ' ' Himalayan forms, the difficulty will be to 



xamined large suites of undoubtedly conspecinc b»ben» 



ically. . . Outftne* o/ Me Distribution of Arctic Plants, Tra 



Hooker : Introductory Essay to Flora of lasmama, V . 



Zi#*. Soc, vol. xxiii, p. 279. 



Ann Roy Bot. Gard Cai.c .-uita, Vol. Ill- 



