432 NATURAL HTSTOBY OF PLANTS. 



insertion, like the Pseonies ; that Crossosoma, whether we make it 

 belong to Ranuncidacea, or its next-door neighbour BiUeniacca, has an 

 unmistakeabty concave receptacle, and that in this same order Dilleni- 

 ace(B, there is one Hibbertia with the receptacle of PofentiUa, though 

 it is quite inseparable from the other Hibberfias, and at one time 

 placed among Bosacea under the name of Warburtonia. 



There are two other orders, so closely allied with Rosacea, so 

 little distinguished by any absolute character whatever, that their 

 separation from it must be regarded as a matter of pure convention, 

 these are Saxifragacece and Legimiinosm . 



As regards the former we do not refer to the commonest type, repre- 

 sented by the Saxifrages themselves and the allied herbaceous genera ; 

 the knowledge of these types, more widely diffused than that of the 

 peripheral genera of the natural order, has led most authors to leave 

 Saxifragacece among the group of orders with parietal placentation, 

 and there we too shall leave them ; for in a linear series it is impos- 

 sible to consider simultaneously all affinities and characters of im- 

 portance. But apart from the fact that most natural orders whose 

 flowers have gynseceums with several free cai-pels, may also include 

 genera, which though otherwise quite similar, have their carpels 

 united edge to edge into a single ovary with several parietal pla- 

 centas •} we have in the tribe Cmioniece genera whose carpels are 

 free, or nearly free, and flowers altogether formed like those of 

 several SpireetB ; the fruit, inflorescence and habit, are the same on 

 both sides, so that it will thus be understood how certain genera 

 have been placed under different names in BaxifragacecB and Bomcea 

 indifferently.^ True, there is one way, regarded as entirely infal- 

 lible till quite recently, to distinguish the two orders, when we can 

 examine the structure of the seeds ; those of the former order having 

 been considered as invariably albuminous, of the latter invariably exal- 

 buminous. But now unfortunately this is lost as a distinctive cha- 

 racter, for we know many Bosacece whose embryo is surrounded by a 

 more or less abundant layer of perisperm, as we have seen in the 



' Sucli as Monodora in Anonacea, Berheri- Itosacerr, uiuler the respective names of Adeni- 



dopsis in Berheridacere, CaneUew in Magno- lema and Jloteia. Luetkea or Eriogyna, whicn 



liacere, &c. (See pp. 119, 159, IGfi, 239, 255). is really only a member of Spireetc had also 



2 We need only recall the fact that Nfillia ^been considered a genus of the order Saxi- 



has been classed among Saxifrm/acett, and As- fmijaceee. 

 tube was for some time regarded as belonging to 



