70 THE STRUCTURE OF FLOWERS. 



arising from the basal portions of the carpellarj leaves to 

 reacli the centre of the ovarian cbamber, and be there fused 

 together into a solid mass, we should obtain the apparently 

 axial structure of Primulacece, Santalacece, etc., with the few 

 or numerous ovules hasipetal in order of development, cor- 

 responding to the centrifugal order in Dioncea and the 

 ascending order in Drosera. 



The probability that this is the correct view is supported 

 by a case I have met with in which the carpels 

 of Primula sinensis were dissociated, and more 

 or less foliaceous with rudimentary ovules, not 

 only along the margins, but with several borne 

 on heel-like processes,* which extended towards 

 the centre of the ovary, as represented in Fig. 18. 

 Anatomical investigations entirely corrobo- 

 ^.^ ~^T^ rate the carpellary nature of the central placenta 

 of Primulacece. The circle of cords, usually ten 

 in number, which pass up the column to nourish the ovules 

 are originally separated from the sides of the sepaline by 

 radial chorisis, and become superposed to the sepals ; the 

 dorsal cords (about ten) having also parted company from 

 the five sepaline and five petaline. The latter, however, do 

 not give rise to any placentary cords ; hence there are really 

 five carpels superposed to the sepals. 



With regard to the position of the spiral vessels, they 

 are not oriented as if axial, but are completely embedded 

 in the phloem, and consequently central. Moreover, the 

 cords in section are circular in form, and not wedge-shaped. 

 The central (if not external) position of the tracheae and the 

 circular lorm of the cords are both eminently characteristic 



* Vau Tieghem, though once regarding the central placenta as axial 

 (Recherches sur la Structure du Pistil, 1868), has more recently arrived 

 at the same conclusion as myself (Traiie de Bot., 1884). 



