BANUNGULACEJE. 71 



internal in the ascending ovule. However, no Dilleniad has yet been 

 observed in which the ovule, if solitary, is not ascending." 



The Bamincidacece are also closely allied to Berbcridacece, through 

 Poduphi/Uim and Jeffersonia. This last being also related to Papa- 

 veracecB by Sangidnaria, the Rammctdaceae come very near the Fapa- 

 veracecs, of which the organization of the pistil is alone different. 

 But we have shown' that, in spite of this difference, which is not 

 really great, we ought not logically to put Papaveracea and Ranua- 

 cidacece in distinct orders, while we do not separate Moiiodora from the 

 Anoriacece, or Berberidopsis from Lardlzahcdacece. We have also said' 

 that the Alismacece approach the Rammcidacece in every way, for 

 certain species of Alisma differ from some of the aquatic Rannn- 

 cidacecB in only one respect — ^the number of cotyledons in the embryo. 

 In our opinion, the conjunction of these two types, due to Adanson's 

 sagacity, " is most consonant with natural methods." 



Finally, the Rosacece, chiefly through Pofentdla, come far nearer 

 the Rannncidacece than is usually admitted. The insertion, a cha- 

 racter the value of which has been exaggerated, no longer separates 

 the two groups so clearly, since some of the Raimnculacece have been 

 demonstrated to be perigynous.^ The absence of albumen in the 

 Rosacece appears, on the contrary, to be a constant differentiating 

 character up till the present date. 



In fine, the relations of the Raiumc/dacca are multiple ; and if we 

 tried to represent them by arranging the different families allied to 

 this on a sort of map, we should have to put the Rammcidaceoi in 

 the centre, so that its frontiers would touch the Ddleniacea by Acro- 

 treiiia, the Berberidacece by Podophi/Uiim, the MaynoliacecB hyMi/osunts, 

 the lUiciecB by Knoioltonia, the Rosacea by Pceonia and Crossosonia, 

 the Papaueracece by Glaucidium, and AlismacecB by the aquatic 

 Ranunculi. 



We have to some extent indicated the geographical distribution' 



' AJansonia, iv. 39. soma, whose ]ilacc it is true is somewluit con- 



- See p. 64. A. L. be Jussieu {op. elf., testiblo, is very clearly jierigynous. 



235), also recalled these relations. ■• '"^ec p;eiK'rally for all conconiiuir ceoirniplii- 



3 It is quite certain that the perigyny of the cal distribution Di; Candoi.le's a,'o;iraph,e Hot. 



Pffionies is not well marked ; else it would have Eali. (1855) ; and for Kuropcan species, cspc- 



heen recognised long ago. Hut the concavity cially those , of the central plateau of '•Vi'ii'V. 



of the rcceptahle is not inore marked in several see fiECOQ, El. sur la Gcoij. Bui., iv. W2-'y-2b 



Rosacece. (Sec Adansonia, iii. 4G.) Crosso- (1855). 



