TERNSTRCEMIACEffl. 255 



V. Pellicerie.e. — Flowers enveloped in two large bracts with 

 imbricated corolla and calyx. Androceum isostemonous ; ovary of 

 which only one cell is fertile and uniovulate. — -Trees with alternate un- 

 symmetrical leaves and solitary subsessile terminal flowers. — (1 genus.) 



VI. Marcgravie^e. — Corolla imbricate, with the parts free or 

 united below for a great distance in a sort of cap which is detached 

 at the base. Stamens definite or indefinite in number. Anthers sub- 

 basifixed. Fruit with thick pericarp, indehiscent or rarely partly 

 dehiscent. Seeds small and numerous, exalbuminous ; embryo fleshy, 

 straight or lightly fornicate, with cotyledons shorter than the radicle. 

 Epiphytal or sarmentose shrubs with terminal inflorescence in short 

 racemes or umbels, with 1 -flowered bracts often transformed into 

 ascidia, &c. — (3 genera.) 



VII. Caryocare^e. 1 — Corolla imbricate, with petals free or ad- 

 hering and falling together as a cap. Ovary cells uniovulate. Fruit 

 indehiscent. Seed exalbuminous, macropodal, with much-developed 

 radicle, inflexed or spirally rolled. — Trees or shrubs with compound 

 digitate leaves and flowers disposed in terminal racemes. — (2 genera.) 



In Ternstroemiacece, as in all families formed " by concatenation," 

 there are few characters which are absolute, and those of one series 

 pass easily to those of another. The features of organization, which 

 vary but little in a natural group, and suffice even to distinguish one 

 family from another, are not constant in a given series. Thus from 

 one genus to another we see the corolla gamopetalous or polypetalous, 

 the stamens definite or indefinite in number, hypogynous or peri- 

 gynous, the fruit dry or fleshy, the seeds with or without albumen, 

 the leaves alternate or opposite stipulate exstipulate. 2 Hence arises 

 a great difficulty in separating distinctly Ternstroemiacece from the 

 numerous neighbouring groups with which it presents affinities. 

 According to all authors it is very nearly allied to Tiliacece, and in 

 general very badly distinguished from it except in the prsefloration of 

 the calyx, which is always imbricated in Ternstrcemiaceae and generally 

 valvate in Tiliacece? Dipferocarpacete, which differ distinctly from 

 the latter by the very decided imbrication of the sepals, become thus 



1 Rhizololece DC, Prodr., i. 599. — Endl., 3 We know, however, that the imbrication of 

 Gen., 1075, Ord. 231. — Rhizobolacece Lindl., the calyx is very decided in certain species of 

 Veg. Kinffd.,398, Ord. 143. Sloanea, gene) ally inseparably from those with 



2 When these organs exist in Ternstroemiacea valvate calyx (on the value of this character see 

 they are always but very slightly developed. Adansonia, x. 190). 



