260 ARNOTT ON SOUTH AFRICAN PLANTS. 



tube; but by immersion and maceration in hot water, previous 

 to examination, the ovary usually becomes detached, carrying 

 with it the lower half of the disk, to the upper edge of which 

 the stamens and petals are attached ; thus the difference 

 between Raspallia (if, as I think, the fruit is dioecious) and 

 Berardia, is weakened, and the principal character must 

 depend on the free petals of the former, and the gamopeta- 

 lous corolla of the latter; I therefore remove Be?\ phylicoides 

 to Raspallia. Thamnea and Audouinia are separated by 

 Bronfjniart, the one beinij said to have a 3-celled and the 

 other a 1-celled ovary ; of Thamnea I have seen no specimen, 

 l.ut 1 am inclined to suspect, from an examination of Audou- 

 inia, that Brongniart may have overlooked the dissepiments, 

 and that it does not essentially differ from the latter, except 

 in having five instead of three cells, which is here of little 

 importance ; that other botanists entertain a similar opinion 

 I may perhaps be allowed to infer, from having received a 

 specimen of Audouinia capitata from my friend Mr Ben- 

 tham, under the name of Thamnea multijiora. Brunia has 

 been divided by Brongniart into two sections, one of which 

 has been separated by Ecklon and Zeyher under Thunberg's 

 name of Beckea ; but their B. virgafa, with the habit of 

 Brunia^ has tlie character of Beckea, and is left by Ecklon 

 and Zeyher in their restricted Brunia, with which it does not 

 agree in the structure of the ffower; it is therefore pre- 

 ferable again to unite them. I shall here give a clavis analy- 

 tica of the genera of the order : — 



I. Calyx 5-cleft. 

 A. styles 2, or 1 divided to the middle ; ovary 2-celled. 

 Fruit indehiscent, 1 -seeded : petals not clawed . ,1. Brunia.* 



Fruit dicoccous. 



Ovary 2-ovuled, 



Petals free, sometimes convolute, . 2. Maspallia. 



Petals cohering into a tube at the base, 3. Berardia. 



Ovary 4-ovuled, petals free, convolute, . . 4. Linconia. 



* Brunia however has not the fruit always truly indehiscent : in one 

 species I examined, it splits in a septicidal manner, the cells gaping at the 

 apex like a coccus. 



