NATURAL ORDERS. 119 



Of RIIIZOPJIORE/E/ as I have formerly proposed to [«7 

 limit it, namely, to llliizopliora, l^niguicra, and Carallia, 

 the collection contains only one i)lant, which is a 

 species of Khizophora, the Mangrove of the lower part of 

 the river, and probably of the whole line of coast, but very 

 different both from that of America, and from those either 

 of India or of other equinoctial countries that have been 

 described. There is, however, a plant in the collection 

 which, though not strictly belonging to this order, suggests 

 a few remarks on its affinities. 



I referred CaraJVur to Rhizophorcas, from its agreement 

 with them in habit, and in the structure of its flower. It 

 is still uncertain whether its reniform seed is destitute of 

 albumen ; the absence of which, however, does not seem 

 necessary to establish its affinity with the other genera of 

 this order ; for plants having the same remarkable eco- 

 nomy in the germination of the embryo as that of Rhizo- 

 phora, may belong to families which either have or are 

 destitute of albumen. 



The plant referred to from Congo may be considered as 

 a new species of Legnotk having its petals less divided 

 than those of the original species of that genus, and each 

 cell of its ovarium containing only two pendulous ovula. 

 The genus Legnoiis agrees with Carallia in habit, especially 

 in having opposite leaves with intermediate stipules ; in 

 the valvular aestivation of its calyx, and in several other 

 points of structure of its flower. It differs in its divided 

 petals ; in its greater number of stamina, disposed, how- 

 ever, in a simple series ; and in its ovarium not cohering 

 with the calyx. It is therefore still more nearly related 

 to BicJiceia of M. du Petit Thouars,^ from which perhaps 

 it may not be generically distinct. The propriety of asso- 

 ciating Carallia'* with Rhizophoreac is not perhaps likely to 

 be disputed ; and its affinity to Legnotis, especially to the 

 species from Congo, appears very probable. It would 

 seem, therefore, that we have already a series of structures 



1 Flinders' s Voy. 2, p. 5i9. (Ante, p. 20.) " Roxburgh. Coromnnd. 3,/;. S, t. 211. 



^ Nov. Gen. Madarjasc. n. 84. 



* Or Earraldeia, iJu Petit Tliouars, Nov. Gen. Madagasc. n. 82. 



