THE ANGIOSPERMAE 1229 



carpels in this genus, there are three or four distinct series of ovules attached 

 to each carpel, each series being supplied by branches from corresponding 

 vascular bundles in the carpel wall. This seems to be evidence for the 

 occurrence of multiple placentae, though such cases are rare. 



The gynoecium of Papaver has several striking peculiarities, which have 

 been the subject of much discussion. The ovary is paracarpous, but, as is 

 often the case in other paracarpous genera, the carpels have closed loculi 

 at the base and apex. The partial septa which occupy the main part of the 

 ovary are really placentae, each being a duplex structure arising from the 

 lines of marginal fusion of the carpels. They thus equal the carpels in 

 number. As the crown of the ovary is approached the carpels separate by 

 the splitting of each placenta along its median plane. The surfaces thus 

 freed become stigmatic, so that each stigmatic ray on the crown of the 

 ovary is composed of the adjacent faces of two placentae and shows its 

 double nature clearly. 



The commissures where the carpel margins are joined, in paracarpous 

 ovaries, may be more or less distinctly marked by grooves. The marginal 

 vascular bundles may remain separate, forming a pair at each commissure, 

 or they may be united, as are the marginal bundles of the united petals in 

 some gamopetalous corollas. 



The vascular bundles which supply the placentae, whether marginal or 

 otherwise, are often markedly larger than the dorsal bundle, which has some- 

 times been urged against the foliar interpretation of the carpel, since the 

 dorsal bundle would represent the mid-rib of the carpel leaf and this is 

 normally the principal bundle in a foliage leaf. The enlargement of mar- 

 ginal bundles where these are connected to the placentae follows an obvious 

 physiological necessity and carries no morphological weight. Moreover 

 Arber has shown that a similar reversal of importance between mid-rib 

 and lateral bundles may occur even in foliage leaves. 



Finally we must refer to the peculiar type known as free-central 

 placentation, in which the ovules are distributed over the surface of a 

 dome or column of tissue which rises up from the base of the loculus in 

 some paracarpous ovaries. It is a constant feature of the Primulaceae and 

 of many Caryophyllaceae and a few other families such as Lentibulariaceae 

 and Santalaceae, as well as certain genera such as Dionaea (Droseraceae). 

 Formerly this was attributed to an upgrowth of the receptacle on which the 

 (axial) ovules were borne, in contradistinction from the carpel-borne ovules 

 of the majority of genera. There seems good reason to believe that the axis 

 does sometimes take part in the formation of these placentae, but that if 

 present, it forms only the central core and is surrounded by the adnate 

 adaxial portions of the carpels, on which the ovules are formed. It is prob- 

 ably a reduced condition, since in the Silenoideae (Caryophyllaceae), the 

 free state is only reached ontogenetically by the detachment of the placental 

 column from the carpel walls to which it is originally joined by septa. In 

 Melandriiim dioicimi the ovary remains septate at the base, but the placenta is 

 free above and, although originally joined to the top of the ovary, breaks 



