412 Dr. Planchon on Meliantheae, 



stamens and petals with the quinary division of the calyx is equally striking 



in Diplerisma and Natalia among Melianthece, and in Paullinia, Serjania 



and other Sapindacece ; the cohesion of two of the calycine segments takes 



place in several Serjanice and Paullinice as well as in Natalia ; the excentri- 



city of the floral organs is the same in all these cases ; the disc is obliquely 



unequal and incomplete ; moreover, as the petals of many Sapindacece have 



on the inner side of their claws a lamelliform or crest-shaped appendage, so 



we find occasionally on the apex of the claws of the petals of Natalia Paulli- 



niotdes, Planch., some fleshy tubercles, which are evidently the rudiment of a 



corresponding crest ; and, lastly, as nothing can be more striking than the 



resemblance and general agreement of the capsule o( Diplerisma with that of 



Cardiospermum, so, on the other hand, the coriaceous capsule of Bersamece, 



with its arillate seeds, single in each cell, and ascending from the base of its 



inner angle, corresponds in all respects with the fruit of Paullinia. Thus the 



most important points of structure concur in proving the close connexion of 



Melianthece with Sapindacece. 



The analogy of floral organization which exists between Melianthus and 

 Polygala is too obvious to be totally neglected, although the balance is 

 rather against an immediate approximation of those plants. In both cases 

 we have a remarkably irregular calyx, cut into five segments ; the petals con- 

 nected by their claws, and the stamens in a quaternary proportion. But here, 

 as everywhere else, care must be taken not to confound floral analogy with 

 real signs of immediate connexion ; not to mistake parallelism of structure for 

 that direct tendency which habit, the true touchstone of afiinities, points out 

 more or less clearly from one natural group to another. Now, it is on the 

 combined suggestions of that general resemblance and of particular links of 

 connexion, such as that of Krameria with Janusia, of Securidaca with Acri- 

 docarpus and Hiptage, it is, I say, upon these grounds that I am inclined to 

 consider the affinity oi Polygalece with Malpighiacece as more close than that 

 of each of those orders separately with Sapindacece and Melianthece. Still, 

 however, these groups cannot be far removed from each other. 



Besides the four genera which rank naturally under the first of these groups, 

 another anomalous Cape genus may, according to Mr. Harvey's suggestion, 

 be conveniently placed near them. The plant I allude to is Aitonia, whose 



