THE STUDY OF THE EVOLUTION OE THE FLOWEE. 273 



biovulate ovaries, the ovules from their situation belong to different carpels, but the 

 manifestations of the second unit are confined to the additional ovule and a small 

 additional stigmatic lobe. Here, again, as in Samhucus, the dorsal portions of the 

 carpels have become a permanent portion of the ovary-wall and are non-variable. The 

 entity of the carpel is almost completely lost, so that the disappearance of the septa 

 must be regarded as only modifying the polymerous condition of the ovary ; the ovary 

 is not reduced to a single carpel. 



The polymeric hypothesis outlined above receives contributory support from the 

 evidence that has been deduced from studying the course taken by the vascular tissue in 

 the flower. The principal bundles are clearly composite — that is to say, they exist or 

 existed in relation to more than one organ, as they do, for instance, in Vlhiirnmn. Thus, 

 in the upper part of the ovary, the bundles connected with the petals branch. These 

 branches are comparable to the carpellary bundles, which traverse the style in Samhucus 

 and Cornus; whilst the remaining alternate hundles occupy the site of stemonal bundles, 

 although stamens are absent (when a stamen, normally present, is absent, as in Aralia, 

 Davidiay and Selwingiay the corresponding bundle is absent also). Hence each of the 

 eight bundles, in Aucuha, is partly of carpellary origin. If the carpellary element in 



alternately of " ventral " and " dorsal 



5> 



pondence with the bundles 



of Garrya, this would indicate a tetramerous ovary. But, on the other hand, the car- 

 pellary elements might have originated from a single carpel of the type found in Davidia 

 and the Araliacese, having several bundles. The evidence afforded by biovulate ovaries, 

 however, is opposed to the latter possibility, since the appearance of an additional unit 

 is not accompanied by a corresponding increase in the number of carpellary branches 

 in the style — an incn^ease which, according to Benjamin Clarke, happens with the 

 manifestation of an additional unit in Cornus. 



If these speculations are correct, the flower was originally hermaphrodite and the 

 ovary polymerous — multilocular in its lower, unilocular in its upper portion, as in 

 JLeycesteria. By analogy with the Caprifoliacese and Hamamelidaceae, but perhaps 

 prior to the attainment of epigyny, the loculi were pluriovulate — tbe ovules in the lower 

 portion axially, those in the upper portion parietally borne. Then the septa disappeared 

 and the ovules were numerically reduced, leaving only a single parietally suspended 

 ovule analogous to the single fertile ovule present in the loculi of Aralia and 

 Hamamelis. 



Now it is evident, from the facts that have been enumerated for both Cornus and 

 Aucuba, that considerable differences exist between them. First, the ovule of Cornus 



is suspended axially — 



for the CaT)rifoliaccae with terminal 



ovules; whereas, in Aucuha, it is borne upon a parietal placenta — a placeutation 

 apparently constant for the terminal ovules of the Hamamelidaceae. This difference is 

 one of fact, and does not depend upon the hypothesis of origin outlined above. 

 Secondly, the nucellus in Aucuha exhibits distinctive characters, and in its primary 

 bulk, definite " calotte " epidermal cap, and subsequent girth-extension, owing to 

 cell-multiplication, is closely comparable to that of the Rosaccae. 



2s2 



