OF CENTRAL AFRICA. 273 



regard the ovariiiin as liaving in all cases only twoplaccntcC, 

 and tliercforc formed of two pistilla or carpclla. But to 

 this, Avliich is certainly the more usual number, there arc 

 many exceptions. These exceptions occur chiefly in the 

 genus Capparis, which, as it is at present constituted, 

 incUides sjjecies differing from each other in having an 

 ovarium with from two to eight placentae, and consecpiently 

 composed of an equal luunber of pistilla. Capi)aris spinosa 

 is the most decided instance of the increased nund)er of 

 placenta?, and this, as well as some other nearly related 

 species, are also remarkable in having septa subdividing the 

 placentae, and uniting in the centre of the comj)ouiid 

 ovarium. 



In the herbarium there arc three species of the genus 

 Cleome. Two of these, C. pentaphylla and arabica, are in 

 many respects well-known plants ; the third I believe to 

 be an undescribed species, but nearly related to mono- 

 phylla. 



If the very natural group, formed by the Linnean genus 

 Cleome, is not to be preserved entire, its subdivision must 

 be carried much further, and established on other grounds, 

 than has been done by M. De Candolle, whose genera and 

 sections appear to me to have been equally founded on 

 partial considerations. Thus, his Tolaimia, uniting all the 

 Cleomcs whose stamina exceed six, contains in its first 

 section, in addition to the species from which the genus 

 was formed, at least two sets of plants, having very little 

 affinity either with each other or with the original species, 

 whose only congener is placed in a second section. 



Gijnandroims also consists of two groups not very inti- 

 mately connected ; the first is composed of species belong- 

 ing to South America, and having the usual aestivation of 

 the family : the second, of which C. pentaphijUa may be 

 taken as the type, is chiefly African, and is readily distin- 

 guished by its very different aestivation, — the great pecu- 

 liarity of which consists in the petals not covering the 

 stamina at any ])eriod. To this mode of aestivation l::i 

 of petals, which has never before been noticed, though it 

 equally exists in Crateva and in Rescdacca', I shall apply 



18 



