INTRODUCTION. O 



the many relicts of artificial systems were eliminated by Luerssen, Engler, and Bessey, 

 and it is perhaps not too much to say that the modified Besseyan system, which has 

 drawn upon many others, is as natural as the present state of knowledge permits. Such 

 systems have dealt chiefly with the relationships of orders and famihes, however, and 

 are often very unsatisfactory in their treatment of genera and species. 



Relationship is the very essence of classification. This axiom is universally accepted 

 in theory, but too often ignored in practice, especially by the segregator. It is not suffi- 

 cient that the relationship be felt, but it must also be shown. This can not be done by 

 raising the sections of a genus to generic rank, or the variads of a species to specific 

 rank, and this fact constitutes the most serious indictment against such practices. 

 Moreover, while it is generally recognized that orders should be arranged and families 

 grouped in such manner as to reveal their relationship, it is not usually realized that 

 a similar relation should obtain between the genus and its sections, and between a species 

 and its variads. This principle is later discussed in detail, and here it will suffice to 

 point out its basic importance for a consistently natural system. 



Evolution and phylogeny. — Evolution is the process and phylogeny the record of descent. 

 Phylogeny is thus the measure of relationship, and is to be expressed in terms of com- 

 munity of ancestry; hence, if relationship is to express evolution adequately, it must 

 take account of each change, from the branch to the variad. It must arrange these 

 in their proper phylogenetic order, and denote them by such terms as will clearly indicate 

 their fixed phylogenetic position. This has long been done for the most part, but the 

 fact has too often been overlooked that the sequence of changes is fixed in practically 

 all regions of intense evolution. The genus exhibits lines of evolution expressed in 

 the sections, and the species shows more recent changes marked by variads. It serves 

 no useful purpose, but merely obscures evolution and phylogeny to treat the sections of 

 Astragalus, Gentiana, or Pinus as though they were genera, or the numerous variads of 

 Artemisia vulgaris or Chrysothamnus nauseosus as species. A natural classification 

 must maintain as well as reveal the different degrees of relationship as expressions of 

 different stages of evolution, and it can do this most accurately with genera, sections, 

 species, and variads, where the Unes of evolution are still in a condition conducive to 

 experimental study. 



The evidences of evolution are the materials of phylogeny, and these have been almost 

 exclusively morphological up to the present. Experimental studies, such as those of 

 Bonnier (1895, 1920), Jordan (1873), and the geneticists, have been few, and for the 

 most part without taxonomic objectives. Such work serves to prove that, while it is 

 convenient to employ morphological observations, it is neither necessary nor desirable 

 to do so without supplementing and checking them by means of statistical and experi- 

 mental methods. Since the limits of orders and families are determined by genera, and 

 the limits of genera by species, the whole problem resolves itself into a statistical and 

 experimental study of species and their evolution. Pending the general adoption of 

 such methods, morphology will continue to be employed, but it should be with due recog- 

 nition of the fact that both criteria and results are tentative and must be confirmed by 

 objective methods. While paleontology has often made brilliant use of morphology in 

 connection with sequence in tracing the course of evolution, its criteria for genera and 

 species have necessarily been drawn from existing units, and are similarly susceptible 

 to refinement as a consequence of statistical and experimental studies. 



Synthesis and analysis. — The prevailing practice in the segregation of genera and 

 species is chiefly detached analysis, with little consideration of relationship and prac- 

 tically none of evolution. Superficially, this practice would seem to follow the course 

 of evolution, since the latter is primarily a matter of differentiation. As a matter of 

 fact this is usually expressed in a vast number of individuals scattered over a wide 



