b 



A TEIBE OF EHIZOPHOEACEJ:. 67 



and LoganiacecB, near which Lindley places them in his * Vegetable 

 Kingdom.' The free petals and their aestivation, the stamens 

 inserted on a perigynous disk, not in the tube of an essentially 

 gamopetalous corolla, the placentation and other points, appear to 

 me to remove them far away from any of the orders between 

 which the Logcmiacece form the connecting link. 



The number of genera of Legnotidece now known is nine ; 

 but as they most of them consist of only one or two species, and 

 some are but imperfectly known, it is probable that a better ac- 

 quaintance with the flora of their chief geographical area — across 

 tropical Africa, through the Mascarene Islands, Ceylon and the 

 Indian Archipelago to tropical Australia, — may enable their future 

 consolidation into natural groups. Of these genera, three, Caralliay 

 Fellacalyx and Haplopetalum, have the seed-bearing part of the 

 ovary and fruit completely adnate or inferior ; three, Gynotroches, 

 Crossostyles and AnstrutJieria, have the ovary superior, but attached 

 by a broad base to the broadly turbinate calyx-tube; and the 

 remaining three, Blepharistemma, Dactylopetalum and Cassipoureay 

 have the ovary much less dilated at the base, quite free within a 

 campanulate or ovoid calyx, as in LytJiracece, 



Caeallia, 



This genus, originally established by Eoxburgh, has been more 

 accurately described in detail by Blume, Mus. Bot. vol. i. p. 128. 

 He shows that although the fruit is often by abortion one-celled 

 and one-seeded, the ovary is divided into 4 or rarely 3 or 5 cells 

 with 2 pendulous ovules in each; not one-celled, as stated in 

 Arnott's observations, ' Ann. Nat. Hist.* i. p. 371, in consequence 

 probably of Roxburgh's having so described one of his species. 

 Blume also first described the albuminous seeds of Carallia with 

 a curved embryo, which I have verified in different specimens of 

 C. integerriina, and in the O. Icmcecefolia also, as far as I could 

 tell in a not quite ripe fruit. 



With regard to the species of Carallia, they have evidently been 

 much multiplied. There is one very common one, extending from 

 Ceylon and the Indian Peninsula to Khasiya, China, the Indian 

 Archipelago and north-west Australia. The leaves in the Cin- 

 galese specimens are often all obovate and very obtuse ; in some 

 Chinese and Philippine Island ones, narrow-oblong and acumi- 

 nate ; in the majority of specimens from the greater part of the 

 area, oval-elliptical, with or without a short acumen ; but in each 

 district these forms appear to pass gradually one into the other, 



r2 



