10 



MISC. PUBLICATION 200, U. S. DEPT. OF AGRICULTURE 



among flowering plants, is probably 

 the largest in the number of indi- 

 viduals and is one of the most widely- 

 distributed. Some genera, such as the 

 bluegrasses (Poa), the bromegrasses 

 (Bromus), and the immense genus 

 PaJiicum, contain numerous often 

 closely allied species. Some genera 

 contain but a few species or only 

 one. 



When an attempt is made to 

 classify a group of related variable 

 species the question always arises 

 whether there are several closely re- 

 lated but distinct species or a few 

 distinct species, each of which shows 

 great variation. It is but natural that 

 botanists should differ in their con- 

 clusions. This explains in part the 

 different classifications of the same 

 group given by botanists of different 

 periods or even of the same period. 

 A satisfactory classification depends 

 upon the study of abundant material 

 both in the field and in the herbarium. 

 By observation in the field one learns 

 the range of variability of a species, 

 while in the herbarium one can com- 

 pare plants from different localities, 

 interpreting the dried specimens in 

 the light of field experience. 



In the classification of variable 

 species it is found convenient some- 

 times to separate variants as varieties. 

 A variety comprises those individuals 

 of a species that show a definite 

 tendency to vary in a certain direc- 

 tion, but which are connected with 

 the species by rather numerous inter- 

 grades. Sometimes a variety is found- 

 ed on a single variation which is 

 distinct but trivial, for example, 

 pubescent specimens of a glabrous 

 species. A variation supported by a 

 distinct geographical range or even by 

 a distinct habitat is given greater 

 weight than is a variation found in 

 a few individuals growing among 

 plants of the typical form. 



The study of a vast amount of 

 material in field and herbarium 

 during some 40 years has resulted in 

 the recognition of relatively few va- 

 rieties, the intergrades proving to be 



more numerous than fairly clear-cut 

 variants. Well-marked varieties are 

 given a separate paragraph in the 

 text, but are not usually given in 

 the keys. Less well-marked varieties 

 are given in the paragraph with the 

 species. Many additional forms are 

 indicated in a descriptive statement 

 without being formally recognized as 

 species or varieties. For example, 

 under Digitaria gracillima appears, 

 "A tall plant with * * * has been 

 called D. bakeri (Nash) Fernald"; 

 and under Eriochloa michauxii, "a 

 form with * * * has been described 

 as E. mollis var. longifolia Vasey." 



The arrangement of the genera in 

 this manual is, in general, from the 

 simple to the complex. It is, of course, 

 impossible to arrange all the genera 

 in linear sequence and at the same 

 time represent a gradual increase in 

 complexity because plants have not 

 developed in a single line, but have 

 diverged in all directions, their rela- 

 tionships being a complex network. 

 The highest genus of one tribe may 

 be much more complex than the 

 lowest genus of the next tribe above. 

 On the average the Bambuseae seem 

 to be the most primitive and the 

 Tripsaceae the most complex. A 

 grass with a spikelet consisting of 

 glumes and several florets, the lem- 

 mas and glumes being similar and 

 resembling bracts, is a primitive form. 

 Grasses with spikelets in which the 

 parts are reduced, enlarged, or much 

 differentiated, are derived or complex 

 forms. Derived forms may be simple 

 from the reduction of parts and yet 

 not be primitive. In the main the 

 genera of grasses fall readily into a 

 few large groups or tribes, but several 

 genera of uncertain affinities are, for 

 convenience, placed in the recognized 

 tribes on artificial characters, with 

 the hope that further study and 

 exploration will bring to light their 

 true relationships. 



The grasses of the world (some 600 

 genera) have been grouped into 14 

 tribes, all of which are represented 

 in the United States. 



