CONSTANCE: SYSTEMATICS OF THE ANGIOSPERMS 435 



Thus, embryological characters are stated to demand exclusion of Lennoaceae 

 from Ericales (Copeland, 1935a) and of Polypremmn from Rubiaceae (Moore, 

 1948). They may point to possible affinities which need, however, to be checked 

 by evidence of other kinds, as the relationship of Adoxa to Samhucus (Fager- 

 lind, 1944), of Empetraceae to Ericales (Samuelsson, 1913), and of Podoste- 

 monaceae to Crassulaceae (Mauritzon, 1933). Such characters can be especially 

 helpful when thrown into the scales to decide between supposedly conflicting 

 relationships, as Cactaceae to Aizoaceae and Portulaceae rather than to Passi- 

 floraceae or Loasaceae (Neumann, 1935; Mauritzon, 1934), and Compositae to 

 Calyceraceae and Dipsacaceae rather than to Cucurbitaceae or Campanulaceae 

 (Poddubnaja-Arnoldi, 1931; Schnarf, 1931, 1933). They can perhaps be of 

 greatest service in testing tlie naturalness of taxonomie groups and in revising 

 internal arrangements. Venkateswarlu (1952) verifies the distinctness as a 

 family of Lecythidaceae, and Falser (1951, 1952) establishes the basic agree- 

 ment between data from the megagametophyte and from floral anatomy in An- 

 dromedeae of Ericaceae. Within Plumbaginaceae, Baker (1948) found that 

 pollen morphology, embryo-sac type, and apparently chromosome numbers could 

 be correlated to distinguish the tribes Staticeae and Plumbaginae. On the basis 

 of embryological features, and making use as well of interesting cytological 

 peculiarities (McKelvey and Sax, 1933; Whitaker, 1934; Granick, 1944), Wun- 

 derlich (1950) has found that the Agavaceae of Hutchinson — although the basic 

 Yucca-Agave relationship is valid — contain diverse elements, perhaps too vari- 

 ous to be retained wdthin the same family. Because more embryological data 

 are known for them than for any other family, Cave's (1948, 1953) application 

 of Maheshwari's embryological criteria to the delimitation of subfamilies, tribes, 

 and genera of Liliaceae suggests how rewarding this approach can be at its best. 

 It also underlines, however, how far we still have to go in the systematic accu- 

 mulation of information before we can hope for similarly profltable exploitation 

 of embryological data in other groups. 



Typological series, based on the initial stages of embryological development, 

 have been proposed by several authors ( Johansen, 1950) but, because they have 

 rarely been utilized taxonomically, these will not be discussed here. 



Fruit and Seed 



Fruits, especially, and seeds have long provided taxonomists with important 

 systematic characters, but there have been few serious attempts to work out a 

 consistent evolutionary scheme for either category of structure. Obviously, the 

 nature of the fruit depends primarily upon the constitution of the gynoecium 

 of the flower, that of the seed upon the ovule. Several authors have stressed the 

 idea that seed disperal is a critical phase in the life cycle of the plant, and one 

 during which the forces of natural selection have a maximum opportunity to 

 exert their effect (Corner, 1949; Stebbins, 1951). Stebbins has stressed the im- 

 portant consideration tliat selection operates on combinations of characters, so 

 that what types (primitive or advanced) of vegetative and rein-oductive struc- 

 tures will be found in successful association depends upon the habitat in which 

 the plant lives, and the various agents of pollination and seed dissemination 

 available to it. Bews and Corner regarded seeds with short viability and little 



