THE STUDY OF THE EVOLUTION OF THE FLOWER. 295 



ovule, as exemplified by Cornus and Selwingia respectively, may prove a useful 

 character, the sympathetic development of the ovary in relation to the form of ovule 

 having evidently evolved independently in each genus during the period of abortion 

 through which the ovary has conceivably passed. 



Phylogeny of the Hamameltdace^.— Robert Brown* considered that the 

 Hamamelidaceae possessed affinities with Bruniacese on the one hand and the 

 CornaccEe on the other. Affinity with the Bruniacese and Saxifragaceae is more 

 nearly consistent with well-balanced modern opinion. 



Lindley f placed Altingia and Liquidamha^^ between the Betulaceae and Salicacese. 

 Hallier |, in 1905, reverts to this affinity, conceiving the descent of the Betulacece from 

 the Hamamelidaceae. This plausible supposition is not supported by the embr;yology, 

 and the endosperm as a crucial character has been too lightly regarded. L. S. Gibbs § 

 suggests that the Alsinoidese illustrate an intermediate stage in the development of the 

 exalbuminous seed by a progressive reduction in the functions of the nucellus ; and 

 Pechoutre's || investigations point to a similar tendency in the Rosaceae. But even 

 if the evolution of the exalbuminous seed has occurred in several families, it scarcely 

 follows, especially when the embryology is different, that the Betulaceae are virtually 

 exalbuminous Hamamelidoidese. The families are probably convergent, with separate 



and distinct lines of origin. 



The Hamamelidaceae, whilst exhibiting diverse characters of floral organization, form 

 nevertheless a fairly homogeneous group. It is difficult to subdivide the family into 

 well-defined sections, and every criterion hitherto used has failed. The characters are in 

 harmony with Saxifragacean characteristics, and the Hamamelidaceae pass presumably 

 into a pro-Saxifragacean plexus forming an older line originating perhaps from more 

 than one progenitor. 



Phylogeny of the Caphifoliace^. — The Caprifoliacese are remarkable, owing to a 

 singular correspondence in several important and deep-seated characters — the epigynous 

 ovary, the biseriate ovules, the evolution of an axially suspended seed owing to abortion, 

 and the tangentially orientated and similarly organized ovule. Here, again, the 

 characters that have been adopted to separate the genera into groups or even families 

 have proved futile, except in the case of Adoxa. 



I>iervilla temptingly recalls Poli/osma, but the resemblance reveals divergencies from 

 the Caprif oliace^e rather than a possible prototype ; the differences, however, are not 

 incompatible with affinity on the Saxifragacean side. 



Lindley % suggested a possible connection between Samlmcus and Hydrangea, partly 



* Eobert 15rown, ' Characters and Descriptions of Three new Species of Pkuta ' (Londoo, 18 J 8). 



t J. Lindley, Yeg. King. 2nd ed. (1847) 253. 

 t Hans Hallier in New Phyt. iv. (1905) 151. 

 § L. S. Gibbs in Ann. Bot. xxi. (1907) 51. 



II F. Pechoutre in Ann. Sci. Nat., Bot. S"'^ ser. xvi. (1902) 156. 

 % J. Lindley, Yeg. King. 767. 

 SECOND SERIES. — BOTANY, VOL. VIII. 



2x 



