CRITERIA. 39 



shape differs slightly between the species, but this can not be expressed with sufficient 

 definiteness to render the character of much descriptive value. The central flowers 

 are always regularly 5-lobed. The marginal corollas, on the other hand, are so highly 

 modified and there is so much variation even in plants which are otherwise almost 

 identical, that their use is of but shght importance in taxonomic work. 



The color of the flowers is sometimes used as the sole criterion for the recognition of 

 so-called species, quite contrary to usage in connection with other flowering plants. 

 At the best, color can be looked upon only as indicating strains or races when forms are 

 found to differ by this feature alone. In some cases color is associated with the amount 

 and quality of the light. Thus, A. calif ornica normally has yellowish or brownish flowers, 

 but along the desert borders they are commonly red. Similarly, the flowers of A. 

 norvegica are pale yellow in the Rocky Mountains, but Arctic forms are frequently 

 pink or red. Rydberg has made use of color in retaining A . purshiana as distinct from 

 A. gnaphalodes (N. Am. Fl. 34:273, 1916) and numerous observations confirm his state- 

 ment that the latter form, with dark-brown or purplish corollas, occurs only to the east 

 of the Rocky Mountains. West of this area only the yellow-flowered "purshiana" is 

 found, but unfortunately for the correlation with geographic distribution, this occurs 

 also at times within the area assigned to the purple-flowered form (see further on, p. 86). 



The futility of attempting to use flower-color as a specific or even a subspecific criterion 

 in the field is evidenced by studies made on Coal Creek, in the Teton Mountains of 

 Wyoming (samples preserved at herbarium. University of California, 11444, 11445). 

 Here, at 2,500 meters altitude, there grows an abundance of A. vulgaris flodmani, the 

 plants forming colonies of considerable extent. In one of these the corollas were all pure 

 greenish-yellow or at the most edged with pale purple, but neighboring colonies were all 

 deep purple. While this difference was not great, it was positive and each plant could 

 be placed, on close examination, in either the yellow or the purple group. Therefore, if 

 this character is to be used for the separation of species in one part of the genus, it should 

 be appUed throughout, and this would result in breaking up flodmani, already too close 

 to discolor, into two subspecies and assigning these to identical habitats. A similar 

 division would presumably be necessary for many other species and subspecies. Such 

 procedure, while possibly desirable after a close genetic analysis and after a method 

 has been devised for classifying these minor forms, is neither necessary nor advisable 

 at the present time. The pubescence of the corolla will be referred to later (p. 43). 



Number and reduction of flowers in the head. — The reduction or complete suppression 

 of certain types of flowers in the head was used by Besser as the chief basis for his classi- 

 fication into sections, and it is still the most important character available for this pur- 

 pose. All species of Artemisia may be classed as either heterogamous or homogamous. 

 In the former the head consists of an outer circle of ray-flowers and an inner one of disk- 

 flowers, these two kinds of flowers being rather easily distinguished by reference to the 

 plates. Homogamy has been brought about through a complete elimination of ray- 

 flowers. This is regularly the case in the section Seriphidium. That the reverse has not 

 been the direction of evolution— that is, that heterogamy has not arisen from homog- 

 amy — is evidenced by a consideration of the fact that such an assumption would 

 necessitate the corollary of the development of ray-flowers in Artemisia. This in turn 

 would lead either to the assumption that the Artemisias were the progenitors of all 

 of the radiate Compositae or to the polygenetic origin of this character. If further 

 evidence is needed it is seen in the nature of the ray-corolla when this is present. It is 

 always an apparently functionless structure with an irregular orifice and is plainly a 

 vestige of the regular corolla and not so modified as to give evidence of being an incipient 

 ligulate corolla. A connection between heterogamy and homogamy is apparently found 



