CRITERIA. 41 



parallel with specific characters. This is in A. senjavinensis, in which the heads are 

 always in a dense terminal cluster, as far as known. It should be noted, however, that 

 this type of inflorescence also occurs in certain subspecies of the very closely related 

 A. norvegica. 



Leaves.— The leaves in Artemisia are always alternate and simple, but in almost all 

 other respects they run through long series of variations. They are evidently plastic 

 structures. They range from entire to twice-pinnatifid, even in a single species. Terete 

 leaves occur among the American species only when this condition has been brought 

 about by the folding of the margins, as in abrotanum, filifolia, etc. It seems that these 

 tightly revolute leaves can be changed back to plane leaves by a change in the environ- 

 ment. For example, in A. calif arnica the leaves and their lobes are usually revolute and 

 terete, yet in vigorous young growth the lobes flatten out and become 1.5 to 2 mm. 

 broad. This is seen even in the northerly part of the range of the species, while to the 

 south, and especially on some of the islands off the coast of southern California, it is 

 very pronounced. Somewhat different seems to be the case of a very narrow-leaved 

 form of Artemisia cana. This species usually has leaves 2 to 4 mm. wide. It sometimes 

 happens that the plants growing with the normal form have leaves only 1 mm. wide, 

 and no intermediate forms appear. The exact reason for this is not known, but possibly 

 the narrow-leaved type is a mutation from the broader-leaved one. The venation of the 

 leaf is sometimes very prominent and is then an aid in the recognition of the species. 

 This is particularly true in the case of A. rigida. 



FlQ. 3. — Leaves of Artemisia vulgaris hcterophylla, to show variation in outline: a, leaf from plant growing on a moist 

 northerly slope (most leaves on this plant were of this type, with long, spreading lobes) ; b, leaf from plant of loose 

 soil fairly moist (the moderately spreading lobes are characteristic of the plant) ; c, d, e, /, g, common types of 

 leaves on southerly exposure. AJl from the Oakland Hills, California. All X 0.4. See also fig. 7, p. 74. 



The extent and nature of the lobing of leaves furnish useful criteria in a few cases. 

 Thus, in certain species the blade is uniformly divided into fiUform lobes, in others a 

 unique type of dentation, confined to the summit of the leaf, furnishes a useful clue as 

 to relationships. In most species, however, the variation is so extensive that leaf- 

 characters fail almost entirely, even for the recognition of subspecies and varieties. 

 All students of the genus have attempted to use them, but those most familiar with the 

 plants in the field are the least inclined to attach importance to cut of leaf. Thus, 

 Piper, in speaking of species here referred to A. vulgaris, expresses the opinion that leaf 

 contour is worthless as a character and that its use leads to an artificial disposition of 

 species and subspecies (Contr. U. S. Nat. Herb. 11:588-589, 1906). Evidence of a 

 similar nature is recorded very incompletely in figure 3, where seven different cuts of 



