THE AXGIOSFERMAE 1211 



of a fertile appendage. Its axial nature explains the paucity of branch veins 

 coming from the dorsal bundle. The carpel walls are supposed to represent 

 a pair of seed-bearing cupules united by their margins, in short a female 

 shoot of Caytonialean type, reduced to two opposite cupules now^ concres- 

 cent. Union in this way would bring their stigmatiferous margins, supposing 

 the cupules to have been comparable with those of Caytonia (see Volume 

 III), to near the base on the ventral side of the carpel. Thomas suggests 

 that this was the primitive position of the stigma and that its rise tow^ards 

 the carpel apex has been due to natural selection favouring greater acces- 

 sibility of the stigma in crowded gynoecia. 



To sum up this theoretical discussion we may say that the classical 

 concept of the carpel, as a lateral, leaf-like appendage analogous to a sporo- 

 phyll, with the necessary reservations as to the previous history of such 

 appendages, is still a useful working hypothesis. 



The three traces which the carpel receives from the receptacle are 

 separate from the base. The median trace may originate at a lower level 

 in the receptacle than the other two, and when it enters the carpel it folloW'S 

 a Straight course upward into the style. Its course is marked, as a rule, by 

 an external ridge or angle called the dorsal suture. The other traces diverge 

 to the two margins of the carpel wall, which they follow- upward, giving off 

 branches to the ovules. Being the supply channels to the ovules they are 

 often stronger than the dorsal bundles. They remain separate, although 

 the parenchymatous tissues of the carpel margins become united during 

 development. Even in a syncarpous ovar}' they usually maintain their 

 separate identity, but in a paracarpous ovar\' they generally unite with the 

 marginal bundles of the neighbouring carpels. They may, or may not, 

 pass into the style. In addition, the lateral walls are supplied by a number 

 of lateral branches from the three traces and in a large carpel the branches 

 may form a network connecting the main bundles. It is very notable, 

 however, that the majority of these branches originate from the marginal 

 bundles and few, if any, from the dorsal bundle. This is the direct opposite 

 of the mode of branching of veins in the leaf, which is away from the mid-rib, 

 not towards it. Whether this difference is attributable to the superior 

 physiological importance of the marginal bundles in the carpel, or whether 

 it is a relic of the previously three-lobed condition of the carpel, cannot be 

 decided, but the fact itself affords support to those who deny that the carpel 

 is a simple leaf. 



The orientation of tissues in the dorsal bundle is normal, that is to say 

 the xylem is adaxial, but in a compound ovary the orientation of the bundles 

 at the points corresponding to the inturned margins of the carpels is 

 inverted, as would be expected if the carpel is an infolded structure. This 

 inversion distinguishes the carpel bundles from those of the floral receptacle, 

 if this is present in the centre of the ovary axis, since the receptacular 

 bundles preserve their normal orientation (Fig. 1180). 



A theon,' has been put forward by Saunders, under the name of " Carpel 

 Polymorphism ", to the effect that there are three distinct types of carpel 



