234 MB. G. BENTHAM ON ETJPHORBIACEJE. 



Two genera nearly allied to Mallotus form a fourth group, dis- 

 tinguished by the anthers, of which each cell is divided into two 

 locelli, as in Endospermum, though these locelli are not separately 

 globular as in Hernardia and Hassharlia. In Cleidion, a genus of 7 

 species dispersed over the tropical regions of the New as well as the 

 Old World, the 2 cells are lateral with the locelli superposed. In 

 Macaranga {Mappa of Jussieu), a genus of about SO species, chiefly 

 from the Indo- Australian region, but extending on the one hand to 

 tropical Africa, and on the other to the South-Pacific islands, the 

 character is not quite so constant, and the locelli not quite so 

 regular. In the great majority of species the anthers open in 4 

 nearly equal valves ; but sometimes two of the valves are smaller, 

 or there are only three or even two ; but the genus may be more 

 or less distinguished from Mallotus either by the inflorescence or 

 by the bracts, or by some floral characters. The species are dis- 

 tributed into three principal groups or sections. In the typical 

 Macaranga the stamens are indefinite, the anthers always 4- 

 valved, and the pistil almost always reduced to a single carpel. 

 In the section to which Mueller has limited the name of Mappa, 

 the stamens are also indefinite, the anthers usually 4-valved, but 

 a few 3-valved ones are occasionally mixed in with them, and 

 the ovary is usually 2-celled, very rarely 3-celled ; and in this 

 section should be included Mueller's DimorpJiantliera, two very 

 dissimilar species, one of which has 3-valved anthers mixed 

 with the 4-valved ones, whilst of the other the male flowers are 

 unknown. The third section, PacTiystemon, might easily be 

 restored as a distinct genus : the inflorescence is somewhat dif- 

 ferent ; there are but 1 or 2 stamens, all with 3-celled anthers ; 

 and the ovary is 4- to 6-celled, succeeded by a very flat-topped 

 capsule : it comprises six species. Mappa capensis, from South 

 Africa, transferred by Mueller to Mallotus, is as it were interme- 

 diate between the two genera. The inflorescence, the female 

 monocarpellary pistil, and the fruit are quite those of Euma- 



caranga ; but there are only 2 or 3 stamens, much like those of 

 Pachystemon, but apparently always with 2-celled anthers. 



The next group, also of only two genera, is remarkable for 

 the large number of anthers, sometimes above a thousand, 

 crowded on repeatedly branched filaments. 1. Ricinus, a single 

 but very comprehensive genus, probably of African origin, but 

 now established in so many tropical and subtropical countries, 

 and varying to such a degree, that several species have been dis- 



