i2o8 A TEXTBOOK OF THEORETICAL BOTANY 



tains are the ventral carpel-traces, not axial bundles. This is, as has been 

 said, the general rule; but there are exceptions, especially in the families 

 Oxalidaceae and Caryophyllaceae, where the summit of the floral axis is 

 relatively broad and is not completely utilized in carpel formation. In 

 such cases the distinct floral axis persists upwards for some distance in the 

 centre of the compound ovary and is distinguishable by its own ring of 

 vascular bundles, the carpellary margins being fused around it. In unilo- 

 cular compound ovaries, where the carpel margins do not meet at the centre, 

 the floral axis does not rise above the base of the ovary. 



The well-known gynoecial structure in Pelargonium and Geranium 

 presents peculiarities. It is commonly stated that the carpels, which are 

 mechanically detached when ripe, are joined by their styles to an axial 

 column. The floral axis is, however, only present in the lowest part of this 

 column, at the level of the ovaries, and the upper part is formed entirely 

 by the union of five prolongations of the carpels. These are not true styles, 

 but sterile extensions of the ovaries, each with its separate loculus. The 

 outer walls of these elongations are highly sclerotic and break away from 

 the adaxial walls when the fruits are ripe, leaving the latter united in a 

 central column. 



The foregoing considerations do not, of course, apply to apocarpous 

 gynoecia, particularly when the carpels are numerous or spirally arranged, 

 for in such flowers the axis is prolonged to the top of the gynoecium and all 

 the carpels are separately and laterally attached to it. Their attachment is 

 generally sessile, but a carpellary stalk or carpophore is sometimes deve- 

 loped, which may elongate, in the fruiting stage {e.g., in many Leguminosae), 

 to more than a centimetre long. Carpels which are at first united but later 

 separate, as in Malvaceae, are also attached directly to a central axial column. 

 The carpel margins, whether fused together or free, are generally the 

 fertile portions on which the ovules are borne. They are usually to some 

 extent swollen and consist of soft, thin-walled tissues, which form the 

 placentae. The name is borrowed from mammalian embryology, but there 

 is no true analogy with the placenta in that group, since the ovules are 

 organically part of the parent plant. Only rarely are ovules borne super- 

 ficially on the carpel walls, the Nymphaeaceae and Butomaceae being the 

 chief examples. In these families the ovules are borne all over the inner 

 surface of the carpels, except for the region of the mid-ribs. The Poppies 

 appear to show a similar superficial distribution, but the septa, which do not 

 meet centrally, are not, in this genus, the lateral walls of the carpels, but 

 are expansions of the fused carpel margins, i.e., they are really elongated 

 placentae, as ihay be seen bv comparison with the allied genus Meconopsis 

 (Fig. 1 178). 



We have already, on p. 1202, discussed briefly the difference between 

 the syncarpous and the paracarpous ovary. Paracarpous ovaries are those in 

 which the carpels are united only by their margins, and in which therefore 

 there are no septa. It has been suggested by Troll that the paracarpous 

 state may have evolved in the following way (see Fig. 1061). In most com- 



