THE FLORA. 353 



these epochs with the more adequate records of Europe. For example, 

 Phragmites has not been found in America in any epoch since the Eocene, but 

 its abundance to-day makes its occurrence certain. This certainty is checked 

 by the fact that it is recorded for the Miocene in France, the Pliocene in 

 England, and the Pleistocene in England, Wales, Sweden, and Germany. 

 The case of Scirpus is even more striking, as it is recorded in America only 

 from the Eocene of Canada, but it is found in both the PUocene and Pleistocene 

 of Britain. 



The question of distribution in space, i. e., of the range of a species during 

 a period or epoch, is less clear. Under the former assumption of a uniform 

 climate for the Cretaceous and Tertiary, the inference was unavoidable that 

 genera and probably maJiy species ranged widely. The hypothesis here 

 advanced is that there has been considerable and often marked differentiation 

 of climate and climaxes since the Permian, and perhaps before. This would 

 suggest that ranges are to be determined by the record, but widely ranging 

 genera, such as Acer, Fagus, Populus, Pinus, Quercus, Ulmiis, and others, must 

 have found favorable habitats in regions not yet recorded. In the case of 

 Acer, it seems probable that it must have occurred in more areas in the Rocky 

 Mountain region during the Eocene than those represented by the Green 

 River, Evanston, and Fort Union formations. Similarly, Fagus, which is 

 recorded for the Green River and Fort Union areas of the central Rockies, 

 appears again abundantly in the Kenai of Alaska. It seems certain that it 

 must have also occurred in the vast area between. 



Inferences &om phylogeny.— The fundamental axiom of phylogeny that 

 famiUes and genera earher in the line of descent must have appeared before 

 those later in the line fvumshes an invaluable method of determining the 

 probable occurrence and distribution of genera not yet recorded. For the 

 great groups, such as the algae, fim^, bryophytes, and pteridophytes, this 

 method is simple and convincing. In the case of flowering plants, where it is 

 most important, it must be used with care, owing to the divergence of opinions 

 as to phylogeny. In the last two or three decades, systems of phylogeny have 

 begun to have more in common (Bancroft, 1914 : 1) and it seems probable 

 that the method of phylogenetic inference may be used with greater certainty 

 in the near futiu-e. Even at present it yields a large number of important 

 suggestions, especially if reproductive criteria are regarded as paramount 

 throughout (Bessey, 1897, 1914). Thus, in the monocotyledons, it seems 

 most probable that Alismcdes are primitive, and lAlicdes and Arales derived, 

 the one in the main line of modification in respect to insect pollination, the 

 other in a side-line responsive to wind pollination. A similar divergence 

 occurs from the lihes as a center, leading to Indoles and Orchidales in the one 

 direction and to Poales in the other. Thus, the presence of Arales would 

 imply the existence of Alismaceae for example, and that of sedges and grasses 

 the occurrence of liHes as well. The occurrence of Sagittaria would pre- 

 suppose that of Alisma, and the existence of Typha in the Dakota would imply 

 the earlier appearance of the somewhat more primitive Sparganium, which is 

 not recorded in America before the Eocene. The presence of Carex in the 

 Dakota points to the existence of Cyperus, Scirpus, and other genera with more 

 primitive flowers. In the case of grasses, the occurrence of Stipa in the Mio- 

 cene of Florissant indicates that the Festuceae, Aveneae and the simpler 



