XI V] CLEISTOGAMY, ETC. 131 



unfrequently actinomorphic, and such cases, coupled with 

 the fact that all the flowers of a species normally zygo- 

 morphic occasionally become actinomorphic (e.g. Linaria), 

 help us to infer that actinomorphy was the primitive 

 state. 



Again, many plants, Viola, Oxalis, Lamium amplexi- 

 caule, Salvia verbenacea, &c., bear cleistogamic flowers in 

 addition to the normal one, and there are reasons for 

 concluding that nutrition affects these matters, as also 

 cases where some flowers of a plant, which normally 

 produces typical <J flowers, may be habitually deficient 

 in the pistil or stamens and so be rendered more or less 

 imperfect by abortion. All such phenomena suggest that 

 the flower is still to a certain extent a plastic mechanism, 

 and we may conclude tliat natural selection has rendered 

 some of these polymorphic states more or less fixed, 

 because they have proved of advantage to the species 

 in cross-fertilisation e.g. the heterostyled flowers of 

 Primula, Lythrum, Oxalis, Linum, Erythroxylon, Hottonia, 

 Pulmonaria, Bouvardia, Mitchella, Fagopyrum,Pontederia, 

 Statice, Menyanthes, &c. ; the gynodicecious flowers of 

 Thymus, Nepeta, Alsineae, Scabiosa, Plantago, &c. ; the 

 gyncemonoecious flowers of Compositse, &c. ; and the 

 andromoncecious and androdioecious flowers of Veratrum 

 and Dryas respectively ; as well as the polygamous flowers 

 so common in species of Fraxinus, Rhus, Rhainnus, 

 Parietaria, and many other plants. 



It must not be supposed that floral diagrams and 

 floral formulae are confined in their application to the 

 higher flowering plants. As reference to the following 

 examples, Figs. 34 and 35, will show, they are especially 

 useful in elucidating the structures of the inflorescences 

 and minute inconspicuous flowers, met with in most of our 

 forest trees and shrubs. Eichler worked out the theoretical 



92 



