38 CONTRIBUTIONS TO PAL^ONTOLOGY 



faunas but is contradictory to the reputed Upper Cretaceous age of the Cannonball 

 marine member of the Lance, which corresponds to the Tullock member in its 

 position above the Hell Creek. 



Other formations in the Rocky Mountain region whose floras are shown to 

 be comparable in character and age to that of the lower Medicine Bow formation 

 are the so-called "Laramie" of Black Buttes, Wyoming and part of the Denver- 

 Dawson beds and of the Vermejo-Raton sequence of Colorado. 



The extensive Fort Union flora of the Rocky Mountain region is considered 

 to have little in common with that of the lower Medicine Bow. The often-quoted 

 statement that the Fort Union and Lance floras are essentially similar in aspect is 

 shown to be not strictly true. 



SYSTEMATIC PAL^OBOTANY 

 GENERAL PROCEDURE 



In view of the preponderance of leaves in nearly all occurrences of fossU plant 

 remains, it is only natural that the governing principle for taxonomic work is 

 that the leaves of trees and shrubs are sufficiently well-defined, diagnostic, and 

 constant within genera and species to make identification possible. There are, 

 of course, exceptional cases such as the modern Sassafras officinale Nees and 

 Ebermaier, whose leaf form varies normally from entire to simply lobate to doubly 

 lobate. Fossil specimens of such variable leaf forms of a single species might 

 conceivably lead to their identification as three distinct species. Were it possible 

 to obtain a large suite of such specimens, however, showing intergradations be- 

 tween the leaf forms, they would properly be designated a single species. Modern 

 palEeobotanic technique clearly recognizes leaf variabihty within certain limits of 

 both modern and fossU species. In a thousand leaves from a single modern tree 

 there may be found a dozen variations in size or shape or other leaf characters. 

 Systematically arranged, however, these variants usually are constant in other 

 characters, such as margin, venation, or character of the base or tip. In deahng 

 with fossU specimens the normal variants of a single species can likewise be recog- 

 nized, provided that a sufficient number of specimens are coUected and studied. 

 To achieve the most rehable identifications, therefore, it is desirable to obtain as 

 large a suite of specimens of each fossil leaf type as is practicable. 



In the taxonomic study of the Medicine Bow floras the coUections were first 

 sorted in the field into provisional species, based on similarity of leaf forms. From 

 each suite of provisional species a representative suite of variants was next selected, 

 if available, and shipped for laboratory study. Each selected specimen was then 

 compared and contrasted with pubhshed descriptions and illustrations of the 

 numerous Upper Cretaceous and early Tertiary species of North America. Where 

 simUarities scemed unquestionable the specimens were compared with the type 

 specimens, nearly all of which are in the coUections of the U. S. National Museum. 

 In the larger suites of provisional species, it lias often been possible to sliow that 

 certain leaf types previously regarded as distinct species were merely variants of 

 another species. In the case of Ficus planicostata, for example, I was faced with 



