BILLENIACEJE. 119 



wards and inwards, wliile it looks outwards in all Ranunculacea with 

 ascending ovules as yet known. Nevertheless, by making use of these 

 characters we can only distinguish BanunculacecB and DilleniacecB 

 approximatively. But one fundamental difference, difficult however 

 to make out in the adult flower, has been established by the study of 

 their organogeny. The evolution of the androceum is centripetal in 

 BanunculacecB, but centrifugal in all Dilleniacea as yet observed.' 

 The DilleniacecB have, moreover, incontestible affinities with numerous 

 orders of plants with unilocular or plurilocular ovaries. The 

 Australian types analogous to Hihhertia and Candollea are evidently 

 allied to Cistinea" and the neighbouring orders, especially to Bixacece.^ 

 On this matter we have expressed our opinion^ that "the floral 

 organization of certain Bixacecs, as Mayna, Caiyotroche, &c., leads us 

 to think that the order DilleniacecB might well have representatives 

 scattered through several groups with one- celled ovaries and parietal 

 placentation, and that in these will perhaps some day be found types 

 bearing the same relation to Hihhertia or Tetracera that Monodora 

 bears to Anonacece, Berheridopsis and Erytlirospermum to Menisper- 

 macecs and Berheridecs, or Papaveracece to BanunculacecB." As the 

 right of Monodora to a place among Anonacece is no longer contested, 

 it is probable that the opinion of Miers,' who ranks Canellacece among 

 Winteracece, will, sustained as it is by such good arguments, be also 

 unreservedly accepted before long. Then it will not be forgotten 

 that, on the one hand, Bentham & Hooker'' have recently put 

 forth clearly the close affinities of Canella and Samyda. And as these 

 last are actually placed by the same authors in the same order with 

 the Banarece, formerly considered as inseparable from Bixacece, it 

 will be seen that to take into account all the affinities of a large 

 order consisting of Bixacea and Samydacece both,^ we should place 

 it at the same time near to the Canellacece (a part of Maynoliacece), 

 and to those types with parietal placentation that recall the 

 DilleniacecB in most of their characters. This would explain how it 

 is that Carpotroche, confounded with the true Maynas, has in many 



' Payeil, Traite d' Organogenie cwnparee de la ' Contributions, i. 122. 



Fleur, 233, t. U. ; Adansonia, iii. 129 ; vi. 266. « Gen., 795, 797. 



* Adanson., loc. cit. — Agaedh, Tlieor. Sys- ' This would be an order in which are united 



tern. Plant., 200. hypogynous and perigynous genera, as the form 



3 Planchon, Voy. de Linden, 3. of the receptacle may be indifferently convex or 



^ Adansonia, vi. 274. concave in very many natural orders. 



