CONSTANCE: SYSTEMATICS OF THE ANGIOSPERMS 443 



le type herbace" (1952, p. 179). Finally, McNair finds data from the distribu- 

 tion of fats and other chemical substances to convince him that herbs may have 

 been derived from trees. 



The present weight of evidence suggests to me that, in general, modern 

 herbaceous types may have been derived from more primitive woody ones, and 

 that the opposite derivation appears to be liighly unlikely. This does not pre- 

 clude the likelihood that some apparently "woody" members of predominantly 

 herbaceous families may not have arisen secondarily (Cotton, 1944). Neither 

 does it rule out the possibility that herbaceous primitive angiosperms existed, 

 but a demonstration of both their existence and of their archaic character is 

 still awaited. 



Status of "the Amentiferae" 



A major point at issue between existing schemes of classification hinges on 

 the interpretation of the woody, catkin-bearing, largely anemophilous dicoty- 

 ledons producing predominantly unisexual flowers with no or poorly developed 

 perianth, and often exhibiting "breech-fertilization," or chalazogamy. The ar- 

 rangements of Engler and Wettstein, emphasizing the similarity of these so- 

 called "Amentiferae" to living Gnetales and Coniferales in inflorescence, embry- 

 ology, and mode of pollination, treated them as having a direct origin from 

 gymnospermous types. Foreswearing a strictly phylogenetic arrangement, Rendle 

 (1925) grouped his orders by grade of differentiation in floral structure, and 

 commenced his system with six orders of "Amentiferae." Bessey, Hallier, Arber 

 and Parkin, Parkin (1952), Hutchinson, Eames, Schaffner, "Wieland, Mez, Wode- 

 house, Copeland, Lawrence (1952), and Puri, among others, construe this group 

 of dicots as an artificial aggregation of highly specialized families, whose fea- 

 tures of apparent simplicity owe largely to what Stebbins (1950, 1951) considers 

 a general evolutionary trend toward reduction and fusion in angiosperm flowers, 

 presumably as a consequence of the major shift in agency of pollination. A 

 number of authors, encountering difficulty in deriving this group from bisexual 

 forms or in deriving predominantly bisexual types, like Ranales, from unisexual 

 ones, have suggested a biphyletic or polyphyletic origin for the angiosperms as 

 a whole (Karsten, 1918; Sprague; Campbell, 1928; Davy, 1937; Gunderson, Fa- 

 gerlind, Lam, Metcalfe and Chalk, Suessenguth and Merxmiiller). 



Engler seems to have included in "the Amentiferae" — although no such 

 group-name is employed — the following eighteen families: Casuarinaceae, Sau- 

 ruraceae, Piperaceae, Hydrostachyruaceae, Saliceae, Garryaceae, Myricaceae, 

 Balanopsidaceae, Leitneriaceae, Juglandaceae, Julianiaceae, Batidaceae, Betula- 

 ceae, Fagaceae, Ulmaceae, Ehoipteleaceae, Moraceae, and Urticaceae. ^yettstein 

 regarded Casuarinaceae, Garryaceae, Salicaceae, Batidaceae, Moraceae, Cannabi- 

 naceae, Ulmaceae, Urticaceae, and Piperaceae as groups evincing the same "mor- 

 phological stage" as the remaining amentiferous families, but having no, or no 

 clear, genetic connection with them. Rendle recognized an amentiferous grouping 

 comprising Salicaceae, Garryaceae, Myricaceae, Juglandaceae, Julianiaceae, Betu- 

 laceae, Fagaceae, and Casuarinaceae; these he regarded as "isolated remnants of 

 relatively ancient groups which have no descendants among the more highly 

 developed orders of our present-day flora" (1925, p. 41). Hjelmqvist conceived 

 of Casuarinaceae as near but not in "the Amentiferae," which he restricted to 



