6o PROCEEDINGS OF THE 



a new system of L-Iassificalion einhracing tlie best features of both 

 scliools. Tliere is no ;^airisaviiig tlie fact lliat Bentliaiu and 

 Hooker's system, which was modelled on that of De Candoile, is 

 out of (Uite, but two or three of its main features may justify 

 ujaiutenance. These are: — (1) The retention of the lianalian 

 families at the cuiumencement of the Angiosperms ; (2) the 

 ])hicing of the JJicoLyledons as a \\hole in front of the Monocoty- 

 ledons; and perhaps (3) the retention of the time-honoured 

 triple division of the Dicotyledons founded on the corolla. In 

 the Kanales we not only have a preponderance of primitive 

 floral features from the standpoint of the strobilus theory, but 

 also indications of most of the main modifications of tiie tiower, 

 which became characteristic for other groups. As the Monocoty- 

 ledons are now generally considered to have sprung from 

 Dicotyledonous ancestors, they should certainly follow, and not 

 jn-ecede, as Engler has them, the Dicotyledons. The series, Poly- 

 l)etala), Sympetaloe, and Apetahe or their equivalents, may not as 

 yet have outlived their usefulness as convenient sub-divisions of 

 the Dicotyledons, provided we guard against attributing to them 

 any monophyletic signification. The Poiypetala) may be viewed 

 as polyphyletic from tiie Kanales or Pro-Kanaies if preferred, and 

 the kiympetahc are doubtless so from the Polypet:d;e. The 

 retention of the Apetala; would only be justifiable on the grounds 

 of our inability to connect such forms with Polypetalous cohorts. 

 Instead of forcing relationships it might be convenient as a 

 temporary measure to keep a third series for such families, 

 whether it be called Apetahe, lnconi[)leta3, Monochlamydene, or by 

 some other less committal name. 



De Candoile was the first; to commence a linear sequence with 

 Kanalian families, and Jientham and Hooker followed suit. These 

 systematists failed, however, to perceive in this any ])hylogeiietic 

 significance*. Systematic botany was, and still is to someextent 

 dominated by the idea of the 5-whorled ]ientanierous flower con- 

 stituting the ground-plan of the majority of Dicotyledonous 

 flowers ; and the principles of doubling, splitting, and branching 

 have been too freely invoked to account for members in the 

 whorls greater than five. By adopting the lianalian theory there 

 is no need to press the matter in this fashion. Unless the balance 

 of evidence is strongly ou the other side, it is more natural to 

 assume a sign of primitiveness in many membered whorls. Let 

 it. be clearly understood that there is uo desire to infer that there 

 have been no cases of incrensein floral members by splitting and the 

 like. I withliohl judgment, only emphasising the necessity of re- 

 stud} ing all such apparent instances by tht^ help of the strobilus 

 theory. Many obscure points in floral struciure, in my opinion 



* See in tliis fomu'ction a ieder dated May l.Sth, lUO". from Sir J. D. 

 Hooker to ])r. Newell Arber, reproduced in 'l^ife and Letters of Sir J. 1). 

 Hooker' by Leonard Hnxley, Lonilon, 1918, vol. ii. p. 22. 



