18 



DIAGNOSIS AND CLASSIFICATION. 



valuable. The component forms of these groups do not shade into each other in Pedicu* 

 laris so gradually as they do in the other genera alluded to, so that there is less tendency 

 on the part of authors to reduce species prematurely than there is to treat intermediate 

 forms, that are probably quite stable, as hybrids. 1 



Th 



•- 



ement of groups 



in higher subd 



■See t 



Maximowicz — is a still more difficult matter, but the 



of Bentham, Serie 

 difficulty of all 



of 



is 



experienced in determ 



o 



what are the approximately natural primary div 



Linnaeus 



d Willdenow 



ged their 



species according to characters derived from 



habit, while Steven, experiencing the inadequacy of such characters as the number of 



in defining five of his primary divisions, viz. Per sonata?. 



sp 



ed 



relied 



Faucidentes. Eostratw, Bicuspid 



and Edentulce, on charact 



derived from the structure 



of the corolla ; only one of his divisions, Verticillatw, being, as its name indicates, based 

 on a character derived from habit. It is somewhat curious that while the other 

 five tribes have been recast and modified so that only the outline of one {Rosiratce) 

 lias been retained, and not one of them even in name continues as a primary division, the 

 Verticillatce have continued in every subsequent revision of the genus to hold primary 

 rank and to retain the original or an equivalent name. 



The time appears to have now come for the abandonment of this last Stevenian 



primary division based on the constant possession or the reverse 

 of whorl ed or opposite leaves is a highly natural one, and gives rise to two subdivisions, 



section, for though the 



viz. 



(1) opposite-leaved species, forming 35*7 



r>er cent, of the whole, and 



(2) alternate-leaved species, for 



ming 



64*3 per cent, of the whole 



> 



this is only one step towards subdividing the genus 



> 



d we have to seek 



ture tl 



evide 



necessary to ensure further subdivision 



he most 



■ 



floral struc- 

 tural 



how 



striking 



character undoubtedly is the presence or absence of a beak to the corolla. When, 

 ever, we enquire what relationship this bears to a division based on phyllotaxy, we find 



it one of practical indifference 



the appended table shows 



Table IV.~ General relationship of phy Mo tax is to floral structure. 



leaves alternate 



» 



opposite 



Species with 



• • « 



• • • 



- • • 



• • • 



• • • 



Corollas beaked. 



45 per cent 



55 



M 





Corollas beakless. 



55 per cent. 



45 



19 



The 



by floral structure; and when on 



plane indicated by phy llo taxis is thus at 



ght 



gles to that indicated 



further subdividing the genus according to floral charac 



o 



ters it is found that there is not a single type of corolla amongst alternate-leaved species 



that 



without a parallel amongst opposite-leaved 



it become 



evident that phyll 



taxis and floral structure cannot both continue to be employed to indicate subdivisions of 



1 Steminger in UAlworm, Bot. Centralblatt, xxx, p. 22, who clearly shows the hybrid character of several separable 

 European form*, treats P atrorubens in this fashion, and considers it equal to P. incarnata X recutita. Thac it is 

 he intermediate form in that group to which the presumed parents belong is certain ; that it is a hybrid is I believe 

 doubtful. In this case, too, Herr Steininger does not advance any direct proof of his contention. 



