1228 A TEXTBOOK OF THEORETICAL BOTANY 



scattered over the septa and bear a great number of small ovules. (See Fig. 



The case of Biitomus, where ovulation is distributed all over the lateral 

 walls of the carpels, which correspond to septa, but not on the outer, 

 dorsal walls, has already been mentioned, but here we are really dealing 

 with practically free carpels, slightly coherent only at their bases. In the 

 related genus Hydrocharis, however, the ovary is inferior and syncarpous 

 and although ovulation is much reduced as compared with Butomus, the 

 same scattered, parietal placentation exists. 



When we come to deal with paracarpous gynoecia we find that placenta- 

 tion may be either marginal or truly parietal. Many examples of unilocular 

 paracarpous ovaries have placentae described as parietal which are really 

 marginal, the infolded margins of the carpels being only slightly, if at all, 

 developed. All intermediate stages can be found between completely 

 syncarpous ovaries and those in which the lateral carpel walls are absent, 

 the intermediates having all degrees of imperfect septation. Such incom- 

 plete carpels are in fact open, the margins of each individual carpel being 

 separated and never united. In this respect they might be considered more 

 primitive than closed carpels, but it is unlikely that the condition has any 

 phylogenetic significance. The extreme instance of the open condition is 

 found in Reseda, which is in no other way primitive, the w^hole paracarpous 

 ovary remaining open at the top. The ontogeny of paracarpous ovaries 

 shows that the receptacular apex is broad and that the carpel rudiments 

 arise on its flanks, leaving a central area of the receptacle which is not 

 absorbed into the carpel structure, as usually happens in the more closely 

 knit syncarpous ovaries. Yet it is not unusual to find septa at the base, 

 even in ovaries which are typically paracarpous in their upper portions, 

 making contact with each other by means of the residual apex tissue and 

 thus forming a plurilocular lower portion, while diminishing at a higher level 

 so that they no longer make contact and may be no more than ridges on the 

 inside of the ovary wall. Some ovaries of this kind may also (Capparidaceae) 

 possess a septate upper portion, where the septa again make contact 

 with each other, due, in part, to the narrowing of the whole ovary towards 



its summit. 



Truly parietal, that is dorsal, attachment of the ovules occurs in para- 

 carpous ovaries more frequently than in syncarpous types. A conspi- 

 cuous case is Begonia, where the three angles of the ovary have been 

 clearly shown to be the joined carpel margins, with the ovules on dorsal 

 placentae. The genera of Lardizabalaceae afford other examples. The 

 carpels of Orobanche have each two such parietal placentae. This decidedly 

 unusual circumstance might be interpreted as a modified form of marginal 

 placentation, the placentae having been secondarily displaced from the mor- 

 phological margins, or else false margins have extended beyond the original 

 fertile margins, in a manner analogous to the false indusium at the leaf 

 margins in Pteridium. Such interpretations may apply here but they can 

 scarcely apply in Gentiana. Assuming the orthodox view that there are two 



