THE PEINCIPLE OF COHESION. 75 



then becomes a cup, wliicli finally contracts above to 

 form tlie style, just as in Primulacece. It is, therefore, 

 unilocular, while a circle of ovules appears on a thick ring 

 of tissue within the base of the ovary. Other circles of 

 ovules appear concentrically and centrifugally. It might be 

 questioned, therefore, whether the ring which carries them 

 were axial or not. I think, however, the same interpreta- 

 tion will apply here as elsewhere ; that is to say, the ovules 

 arise from the place where the bases of the carpels would 

 have appeared had they been differentiated out of the axis. 



In the allied genus Drosera the placentas are strictly 

 parietal, and the ovules, commencing to emerge half-way up 

 the wall, appear successively, both upwards and downwards. 

 1^0 w, as they are centrifugal in Bioncca (corresponding to 

 the upivard development in Drosera), it looks as if only a 

 portion of the upper half of the carpels were really repre- 

 sented at all. 



In this genus there is a barren central space within the 

 ring of ovules, perhaps representing the termination of the axis. 



That the basal portion only of syncarpous pistils should 

 bear ovules is common enough, and the pla- 

 centas often swell out there to form bosses 

 which we may reasonably conceive as coa- 

 lescing to form the continuous ring character- 

 istic of Dionoia. Thus^ce?- illustrates how each 

 of the two carpels gives rise to two globular 

 protuberances on which the ovules are borne 

 (Fig. 17). Aneviiopsis, as figured by Payer, 

 has a confluent protuberance bearing several 

 basifugal ovules. Similar multiovular bosses Fig. n —Carpels of 

 occur in Solaneoe and Scroplndarinece, giving 

 the characteristic dumb-bell shape in a transverse section. 



Now, if we imagine these swollen ovuliferous placentas 



