142 GENUS AHTEMISIA. 



Abrotanum, and that the connection is represented in America by A. bigelovi, a rare 

 southern species of the latter group. On the Seriphidium side of the line, the nearest 

 approach to bigelovi is found in A. tridentata. Except in the character indicative of the ' 

 section, the two species are so nearly alike that they are frequently mistaken for each 

 other. Of especial significance, as indicating their phyletic connection, is the occurrence 

 in each of the unique tridentate character of the leaves. This occurs nowhere else in 

 the genus and certainly its appearance in these two species, so closely similar in most 

 other characters, is more than a coincidence. The conclusion therefore is that, since 

 the section Seriphidium is derived from Abrotanum and the connection is best represented 

 on the side of the former by the species tridentata, this species inevitably must be con- 

 sidered as the most primitive of the group to which it belongs. Its connections with 

 the other members of the section will be taken up as these are reached. 



In considering the subspecies of A. tridentata an annoying nomenclatorial difficulty is 

 encountered. It so happens that, if the rule of priority is extended to include position 

 on the page, the name of the species must be changed to A. trifida Nuttall and the com- 

 bination of each subspecies must be changed to correspond. This follows from the fact 

 that trifida and tridentata were first published on the same page (Trans. Am. Phil. Soc. II, 

 7:398, 1841) and that the description of the former precedes that of the latter. But 

 still another name must be chosen by those who adhere to the American Rules (and they 

 are the only ones strictly bound to follow the rule of priority of position), for Nuttall's 

 A. trifida is antedated by A. trifida Turczaninov (Bull. Soc. Nat. Mosc. 5:196, 1832), 

 a species now universally considered as belonging to Tanacetum. The first available 

 name under these rules would be A. arbuscula, since this was published by Nuttall 

 on the same page as trifida and tridentata and placed so as to follow the former but to 

 precede the latter. It is partly because of these several possible applications of the rules 

 with the inevitable shifting of names that the International Code is here followed and 

 no attention paid to priority of position. By this method a choice is allowed between 

 arbuscula, trifida, and tridentata as the name for the collective species. Since the form 

 described by Nuttall as A. tridentata is by far the most common and widespread and 

 since this name is much better known than either of the others, it is here selected. In 

 the narrow original sense A. tridentata Nuttall here becomes A. tridentata typica. 



Of all the subspecies of A. tridentata perhaps none comes nearer to representing the 

 ancestral type than does typica itself. This conclusion follows not so much because of 

 its great abundance over large areas of territory, as because of its obvious connection with 

 A. bigelovi, which in turn represents the still more primitive group from which the species 

 probably has arisen. No other is so much like bigelovi in habit, foliage, and involucre. 

 Furthermore, each of the other subspecies exhibits characters or distributional pecu- 

 liarities indicative of more recently derived forms. As is so often the case when dealing 

 with small taxonomic units, the subspecific characters are found to group themselves 

 into a bewildering array of combinations, so that one is perplexed to know which forms 

 to accept as of this rank. Previous treatments have merely selected the more striking 

 variations and assigned to them either specific or varietal standing. As a method this 

 can scarcely be improved upon, since a recognition of all of the combinations of minute 

 characters is obviously impracticable. It should be pointed out, however, that the 

 number of possible combinations is exceedingly great and these are becoming more and 

 more numerous in herbaria as the result of more extended collections. This is one reason 

 for bringing all of the 7 subspecies here described into one species, although some of them 

 have not been heretofore considered as a part of A. tridentata. The various combina- 

 tions that have thus far received names, as well as some that have not, are indicated in 

 table 13, which, however, does not include all of the minor peculiarities of some of the 

 forms. The subspecies parishi is omitted, since it differs from all others by a constant 



