120 NATURAL HISTORY OF PLANTS. 



classifications' been placed among Maffnoliacp^ or Dilleniacets. The 

 difficulty is here the same as with Erythroapermum, sometimes 

 referred to the Bixacece, sometimes to Berberidacea, according as 

 more stress has been laid on the position and form of the placenta, 

 or on the other whorls of the flower and their symmetry. In this 

 respect, again, the DilleNiacece touch the Cancllacece, which we include 

 in MagnoUacea. Dillenia, Wormia, and other analogous genera come 

 very near Magnolia by their leaves with dilated petioles, membranous 

 and stipuliform on the edges," while by the arrangement of their 

 gynjcceum, they recall that of Illicium and Drinip. The number of 

 parts of the flower excepted, Billema and Wormia are, we may say, 

 far more like Magiwlia than like most Dillcniacea of the CandoUea 

 group. The way the indefinite stamens are inserted on the 

 receptacle of the flower, the very position of the flower at the end of 

 the branch, and even the absence of a sacciform membranous ari? to 

 the seeds of true Dillenias, are features which would have rendered 

 it impossible to place these in a different order from that of Lirio- 

 dendron or Talauma, if the structure of the gyna^ceum, apparently 

 so different, had not been taken into account. But we have shown* 

 that the carpels of Wormia and the analogous genera are really free 

 like those of Magnolia^ not united into an ovary whose cells are 

 separated from one another by simple dissepiments ; while the styles 

 are distinct from one another towards the base, and are joined only 

 from a certain point, to diverge afresh in the stigmatiferous portion. 

 Thus Wormia and Davilla serve as a passage towards Acti?iidia, which 

 we cannot remove from them, and which also resembles Sauraja so 

 much as to have been placed with it and Stachgurus in a separate 

 tribe of the order Ternstroimiacea.^ It has been shown, too, how 

 through this last order the Dilleniacece are indirectly allied to Ericinca, 

 Ehetiacca' and Pillosporacea. We might also point out some more 

 distant relations between Schumacheria and certain Dipterocarpcte,^ 



' E«peciully in tl.osc of Jcssikc {Gen., 2R1), * Sur I'Organimtion Floraff d'un W.imiin 



DE CAMiOLLK(iVo</r., i. 7'J), K.NDUCUKU((;en., dea Sei/cMltn, Adansouia. vii. 343. In thw 



n. 4734), au;. nicnioir it is pmvcd that tliiro is at every iige u 



' Seo Adansouia, vi. 271. considcnible cavity in the intervals between tlio 



* A cliunuUr of so iiiuch iniporUmce in the ovurien, and tliut the styles unite ubovc this xpuce. 



eyes of severui authorn, that it \x, for inHtunce, » Sec p 10J», note 2. 



one of the reiiwjnH which lin» determined the in- " ]{. H., (;,„., is J. 



tnxluctionof Cronsutuma into the order JJillmi- 1 KH|)f('i:iily in the venation of the leaves «u>d 



acea rather tlmu JiaHuuculacea:. the unilateral arrnngeuient of the flowers. 



