90 GENUS ARTEMISIA. 



reconstructed. It should not be inferred from these remarks, however, that the accom- 

 panying chart is presented with confidence as to its accuracy. Rather is it to be looked 

 upon as a first attempt to unravel the tangled skein of relationships in this complex 

 species, and to arrange into some sort of natural sequence the numerous taxonomic units 

 bequeathed by the descriptive botanists. Many changes in grouping, and especially in 

 direction, will doubtless be found necessary as further evidence is brought to bear upon 

 this problem. 



The most primitive of all of the subspecies of A. vulgaris is probably tilesi. This 

 has large heads with numerous flowers, a tendency towards twice-pinnatifid leaves, and 

 a northern distribution. The cumulative evidence indicates that this group of American 

 Artemisias had its origin far to the north and that the forms have been multiplied as 

 the stock spread to the south. If the opposite were assumed, namely, that the southern 

 forms are the older, while the northern ones are the derived strains, then serious difficul- 

 ties would be encountered in interpreting the origin of the character differences. In the 

 present case it seems fairly certain that the large-headed forms like tilesi, which occur 

 only in the north, have given rise to the more southerly forms with fewer flowers to the 

 head and with other reductions. 



Assuming that the subspecies Wesi was early established in the north, it is not unlikely 

 that it represents a still more primitive type which ranged far to the east and west in 

 arctic and subarctic regions. One of the derivatives of this form, or perhaps of tilesi 

 itseK, is subspecies typica. This nomenclatorial type of the species is native in north- 

 central Asia and is well known in Europe as an introduced plant. It differs from tilesi in 

 the usually reduced size of the heads and in the reduced numbers of flowers in the indi- 

 vidual head, as indicated in table 6, but it retains the character of twice-dissected leaves. 

 At the base of the petiole are 2 stipule-like lobes, an almost unique feature in the genus, 

 but these structures are occasionally found also in tilesi (Dutch Harbor, Unalaska, Van 

 Dyke 101, Gr), as indicated in figure 10. Additional connecting forms may be expected 

 as the flora of Siberia and Alaska become better known. There are also frequent sug- 

 gestions of a connection between typica and some of the other West- American subspecies, 

 notably heterophylla. The similarity in the foliage is indicated in figure 7, p. 74. Even the 

 stipule-like basal lobes are sometimes present in specimens referable only to heterophylla 

 (northern Idaho, Sandberg 822, US) and they are rather common in A. v. discolor} It is 

 believed, however, that these similarities are due to the persistence of ancestral traits 

 rather than to any immediate phyletic connection. From these considerations it is con- 

 cluded that typica is an offshoot from the original stock best represented by tilesi and that 

 it is not directly connected with any other American form. It is thus a western diver- 

 gence which has reached our territory only as an introduction by way of Europe, and for 

 this reason it is indicated on the chart as lying entirely to one side of the other subspecies. 

 It should be noted, however, that typica is sometimes considered as native in the North- 

 east, and that if introduced it reached America a long time ago (Fernald, Rhodora 2: 135, 

 1900). Notwithstanding its close connection with tilesi, as shown above, there is a wider 

 phylogenetic gap between these two subspecies than is usually the case among series of 

 forms as here set forth. 



Another subspecies is conceived as diverging from tilesi, but in a different direction. 

 This is candicans, of the northwestern United States and adjacent Canada, where its range 

 meets that of tilesi. Apparently it has given rise to none of the others, although there 

 is some evidence of a connection, especially with gnaphalodes. The subspecies candicans 

 scarcely differs from tilesi, except in the presence of a copious white tomentum on the 

 upper as well as on the lower surface of the leaves. This appearance of paired subspecies, 



' The nature of the false stipules in Artemisia vulgaris and other plants has been investigated^by Schiller (Sitzb. Akad. 

 Wiss. Wien, 112:793-819, 1903). 



