TEREIilXTIIACEJ':. 201 



of scales or liairs on the staminal filaments.^ The Riituceœ of the 

 series Zanthoxyleœ, with the carpels united iu a plurilocular ovary, 

 and which have been called Toddalieœ^ differ from Burscrcœ in having 

 leaves without glaudularpunctuation, a diplostemonoiis audroccum, and 

 an exalbnminous embryo. The species oî Amy ris, which, we know, are 

 lîhtaceœ, were formerly connected with Burscrcœ , again proving very 

 close affinities. Slightly more removed from the Burscrcœ are the 

 EiiphorhifKccv, having ho\\'ever, like them, unisexual flowers, pro- 

 vided with a corolla, a diplostemonous aucb'oceum and ovary cells 

 whose two ovules are descendent with the micropyle exterior and 

 superior. But these Euphorhiaccce are distinguished iu such a case, 

 either by the absence of a balsamic juice, or by non-compound leaves, 

 the presence of an obturator above the micropyle, or the existence of 

 albumen." I3y the Spondieœ and Anacardicœ, this family so closely 

 ajijproaches the Sapindaccœ, that it becomes difficult to separate them 

 distinctly when the flower of the latter is not irregular, and the disk 

 exterior to the androceum. But the ovule of the Anacurdiccc, with 

 its known peculiarities, is characteristic of this group, particularly 

 when it is suj)portcd by the ascendent funicle, of which we have seen 

 so many examples in the description of types. The Sapindacccc often 

 have an aril, aciu'led embryo, and usually from two to flve cells in the 

 ovary ; which does not exist in the Anacardicœ, and is observed on 

 the contrary in certain Spondlcce. But in these the ovule is always 

 descendent, the disk interior to the androceum, and the flowers 

 regular. When their carpels are independent, at least mostly, as 

 in Bmhanania and Sjwndias, they thus become very analogous 

 to Suhia and the Connaraceœ, but the former has petals and sta- 

 mens superposed to the sepals and biovulate ovaries, the latter ortho- 

 tropous and ascendent ovules, and carpels dry and generally dehiscent. 

 There remain the numerous affinities of the Muppieœ and l'hytocrencœ 

 in the first place with the O/aciiicœ, amongst which some still range 

 them, and which are, we think, more apparent than real, for they are 

 always separated very distinctly ^ by not having the stamens opposi- 

 petalous by the placenta being parietal, biovulate, the ovary unilocular 



' The (liplostemonoua androceum has also ^ These four characters are, it is true, rarely 



been mentioned as characteristic of the Bitisereœ, found united in any plant belonging to this 



but we know, that amongst them, TiUjniiochlamns family, but they are never all wanting at once. 



is isostemonous, and that a large number of ^ H. Bx. in Adaiisuiiia, xi, 203. 

 Sutaceie are diplostemonous. 



