332 Mr. GRfFFiTH on the Root- Parasites referred to Rhizantheae, 



The styles of a compound pistillum may be themselves divided, as in Cordia, 

 some Ferbenacece, and many Euphorbiacece. In these instances, if the stig- 

 mata were taken as guides, the ovarium would be considered as composed of 

 twice the real number of parts. In such cases regard should be paid, in my 

 opinion, to the primary divisions or indications of division of the style, the 

 situation of the vascular fascicles and their relations with those of the ova- 

 rium, and also to the relative situation of the secondary divisions. This last, 

 which is very applicable to Cordia, is, I think, negatived by one species at least 

 of Artocarpus, in which however the opposition of the two stigmata may per- 

 haps not improperly be referred to mechanical causes. 



Further obscurities may arise from the stigmata, instead of having their 

 usual relations with the styles of a compound ovarium, being confined to a 

 part of these lower than usual, and from these partial stigmata coalescing, as 

 in many Apocynece, in which an annulus of stigmatic surface is exhibited 

 surmounted by an apiculus. Something of the same kind, though in a much 

 more obscure degree, is presented by the stigma of most Asclepiadece, whether 

 it be described as apiculatum or muticum. It is also to be met with in some 

 Meliacece and in Heliotr opium. 



The sources of obscurity affecting the stigma not unfrequently affect the 

 style, so that no absolute rule applicable to the style of a compound ovarium 

 drawn from its perfect state can be opposed to the speculations of the theoreti- 

 cal botanist regarding certain anomalies. In all such the examination must be 

 carried back to that early period when the disc or mass of cellular tissue, from 

 which the various parts are first moulded, presents the carpella in the shape 

 of so many distinct points. 



It rarely happens however that the obscuring causes, existing in the mature 

 flower, affect equally all the component parts of a pistillum, each of which 

 should be examined in detail. The examination should be extended to the 

 allied genera. Such rules applied to Punica go far enough to invalidate the 



the convolution becomes less we shall have corresponding varieties of elongated, linear or spathulate 

 stigmata. 



Although the convolution by which the style is formed is generally most complete in the direction of 

 the ovarium itself, yet in compound styles the reverse is sometimes the case, as in Gmelina, in which 

 the branches of the style are perfectly convolute, while the style itself presents one common canal. 



