THE STUDY OF THE EVOLUTION OF THE FLOWER. 301 



case of very few families, and he claims polyphyly for at least three genera of the 

 Caprifoliacese and eight of the Cornacese, although in the latter family some genera 

 have remained closely associated for a considerable period. Now very many families 

 comprise a certain number of obscure forms, and subsequent analysis may determine 

 not only that these cannot be included in any existing family, being the sole repre- 

 sentatives of extinct lines, but even the apparently stable members of the family may 

 show evidence of parallel development and convergence.^ 



Allowance must also be made in phylogenetic systems for cases where differently 

 organized forms may possess probable affinity, such as, for instance, JPolyosma and 

 Dlervilla. but the suffs^ested ori<dn of one from the other must be comi)atible in 



j-gV^Ol^^ "*"^0 



all respects with morphological structure and evolutionary methods established by 

 experiment ; thus, relatives of Folyosma should possess the power of advance towards 

 the condition in Diervilla, both with respect to gamopetaly and the number and nature 



of the ovules. 



It is a long-established custom to regard certain characters as more constant and 



dependable than others, conceiving that the ovule, for example, would be less liable to 

 change than more superficial characters. According to De Vries's view of indiscriminate 

 mutation, however, the ovule would be equally liaole to change, but since mutation 

 is a rare phenomenon, difficult to perceive. Even then, however, the ovule would pass 

 through relatively fewer mutations than the floral organization, so that a certain form 

 of ovule might remain constant for a considerable range of floral and vegetative change. 

 Hallier reverts to an indiscriminate use of characters without due regard to relative 

 value ; thus Garnja is retained in the Cornacese owing to the discovery of aucubin by 

 Herissey and Lebas « in this genus, although a criterion of this kind is of doubtful 

 value unless supported by considerable research. Hallier f has attempted that which 

 cannot be accomplished from the evidence yet accumulated, and the inevitable result 

 of this finds expression in the extreme instability of his system. 



Knowledge of the phylogeny of Angiosperms can only be truly advanced by the 

 detailed morphological and experimental investigation of many more families, and then, 

 but not till then, can Engler's system be replaced by a greater scheme more nearly 

 approximating to natural relationships. In such a scheme the barrier between the 

 Polypetalse and Sympetal^e will doubtless be removed, and the various families, upon 

 reconstruction, will be arranged in plexuses according to their probable origin and 

 relationship. It is possible to forecast the lines upon which this will be attempted, but 

 the effort lies beyond the scope of this present work. 



I desire to acknowledge my indebtedness to Mr. C E. Jones, B.Sc, F.L.S., for many 

 friendly suggestions ; to Dr. A. B. Rendle, E.R.S., for valuable help and criticism ; and, 

 lastly, to Professor J. B. Farmer, E.R.S., for his continuous advice and encoun 

 throughout the investigation. 



o 



* Herissey and Lebas in Bot. Centralbl. civii. (1911) 176, 



t H. HaUier in Beihefte Bot. Centralbl. xiiii. 2 (1908) 81. and Arch. Neerl. d. Sci. Ex. et Xat. sc'r. 3, Bd. i 



146(1912). 



