A. NORVEGICA. 63 



on morphological characters alone. The degree of difference in the size of the head and 

 the amount of overlapping is indicated in table 4, where these features are brought out 

 in the columns headed "Involucre." The size of the head is also proportional to the 

 number of flowers. While this number will not serve to differentiate the two forms, 

 because of the amount of variation, a decided tendency towards the higher numbers is 

 seen in the specimens from the Old World. 



While the Scandinavian plants exhibit but little variation among themselves, the 

 American branch yields a considerable number of minor derivatives, a majority of 

 which are ecologic, at least in origin. These differ from one another principally in habit, 

 type of inflorescence, size of heads, and especially in the amount of pubescence. Some 

 have been given specific names, and these are here treated under the heading "Minor 

 variations." 



In addition to the two subspecies already discussed, there are three which are alike in 

 that all have undergone a reduction in the inflorescence. This has been accompanied 

 by a dwarfing in the plant itself and the pubescence usually is not only more copious, 

 but has taken on a close, canescent, often silky character. The subspecies are hetero- 

 phylla, globularia, and glomerata. They are restricted to northern North America and 

 the extreme easterly part of Siberia. This group (fig. 6) is indicated as having been 

 evolved before the segregation of saxaiilis, but there is no evidence other than that 

 obtained from the leaves to substantiate this arrangement. On the basis of geographic 

 distribution, it is simpler to assume that the branch is an offshoot from saxatilis, in 

 which the leaf-blades have retained the short, rounded outline sometimes found in that 

 subspecies. The least amount of modification has taken place in subspecies heterophylla. 

 In this the inflorescence has been reduced from the usual open panicle to a spike, and 

 when this is much shortened it becomes almost capituliform, as in some of the specimens 

 of the type collection of A. richardsoniana, here referred to heterophylla. The subspecies 

 is widely distributed in northwestern America, which accounts for the presence of 

 several intergrading forms. 



The final stage in the reduction of the inflorescence is reached by globularia and 

 glomerata. Here the heads are always in dense terminal clusters. In these northern 

 derivatives the foliage, as well as the stem and inflorescence, fails to reach full develop- 

 ment and the pubescence is not so white, appressed, and silvery as in heterophylla. 

 While probably of ecologic origin, either from this or from some similar subspecies, 

 their best characters seem now to be fixed, as indicated by the absence of a completely 

 intergrading series. Much more copious collections from the far north are needed before 

 such points can be finally determined. The subspecies globularia and glomerata are 

 closer in general aspect than any other pair within the species, but on the basis of the 

 statistics presented in table 4 they are the most distinct. The total absence of over- 

 lapping in some of the characters tabulated might seem good evidence of their specific 

 segregation. However, it should be pointed out that the differences are all of one 

 order, namely, the smaller number of flowers and their smaller size in glomerata, with 

 the inevitable result of a reduction in the size of the involucre. This character, even 

 when taken in connection with the tendency towards a more simply cut foliage, can 

 scarcely be of specific value, especially when so few collections are available for study. 

 Moreover, heterophylla is also to be taken into account. By injecting its measurements 

 into the series a nearly complete gradation is obtained, even from the small number 

 of specimens tabulated. The additional criterion of pubescent corollas in glomerata 

 as contrasted with glabrous or only glandular-granuliferous ones in globularia has been 

 used by Besser (DeCandoUe, Prodr. 6:116, 1837) and others, but the unreliability of this 

 character, which has been demonstrated in the discussion on criteria, is here evidenced 

 by the nearly entire absence of hairs on the corollas of the three collections of glomerata 



