168 GENUS CHRYSOTHAMNUS. 



related with their arrangement in vertical rows. When the rows are obscure the bracts 

 are likely to be flattish, without a distinct keel, and usually obtuse unless appendaged; 

 when the rows are well-defined the bracts are folded or at least incurved from the midrib, 

 which thus forms a sort of keel, and the bract is more or less boat-shaped and usually acute. 

 The former is characteristic of most of the Typici, while the latter is best seen in C. nau- 

 seosus and especially in the Pulchelli. 



The texture of the bract escapes every attempt at its characterization, except when 

 herbaceous tips or thickenings appear. A definite herbaceous tip, or appendage, is 

 present only in C. parryi, where it is common for some of the outer bracts to be drawn 

 out at the apex into a slender greenish appendage. While this is useful as a specific 

 character when considered as a tendency, it can not be applied as a constant one. Cer- 

 tain specimens referable here by all other characters have the tip poorly developed or 

 entirely wanting in some heads. The herbaceous thickening referred to is even less reli- 

 able than the prolonged apex. It is usually found as a minute rounded mark and is 

 subapical. In using this character it must be borne in mind that except in one species 

 the spot is distinctly green only in fresh material. In dried specimens it fades to a dull 

 brown. It has been used as a specific trait in the segregation of at least four proposed 

 species. In C. teretifolius and C. vaseyi it is connected with other and more im- 

 portant characteristics which it supplements very well. In teretifolius especially, the 

 decidedly thickened spot is noticeably green, even in herbarium specimens. Quite differ- 

 ent is the case of elegans. This is so close to certain other forms that the presence of 

 the subapical spot, which is here never sharply defined and often very obscure or even 

 wanting, can not save it from reduction to subordinate rank under C. viscidiflorus. Like- 

 wise in the fourth case, namely linifoHus, a reduction to a subspecies of viscidiflorus is 

 necessitated by the discovery of intermediate forms. 



The size of the involucre and the number of its bracts have been investigated as to 

 their value as differentiating criteria. The breadth of the head is a poor character, since 

 it depends largely upon the state of maturity and the manner of taking the measurement. 

 Furthermore, it varies somewhat with the number of flowers. The height is more defi- 

 nite and can sometimes be used to advantage, but only in dealing with very unlike species 

 or when only subspecific distinctions are looked for. When a large series of measurements 

 is taken, these are found to present an annoying range of variation, sometimes even 

 within a single taxonomic subspecies. This is indicated for the subspecies of C. parryi 

 in table 22 (p. 207). A similar tabulation but with less difference between the extremes 

 could be presented for the subspecies of C. viscidiflorus. The most that can be said of 

 the results of such studies of these two species is that there is a tendency toward either 

 short or long involucres in certain subspecies, but that the overlapping is too pronounced 

 to warrant the use of this criterion for specific segregation. Somewhat different is the 

 case of C. nauseosus, as will be seen by reference to table 23 (p. 230). In this species the 

 involucre has the remarkable range in length of from 6 to 12 mm. Furthermore, pairs of 

 subspecies can be selected in which there is no overlapping, for example, gnaphalodes 

 with an involucre length of 6 to 8.2 mm. and bernardinus with one of 10 to 12 mm. While 

 this would furnish a specific character as between these two forms, its use for that pur- 

 pose must be abandoned because of other subspecies which form intergrading series 

 between them; for example, those in which the length runs 6.5 to 8 mm., 6 to 9 mm., 

 7.5 to 10 mm., and 10 to 12 mm., respectively. Furthermore, the remarkable variation 

 exhibited within certain subspecies, as brought out in the table, is a warning to proceed 

 cautiously in the use of this character. When the bracts are in sharply defined vertical 

 rows the total number usually varies by fives. This necessarily follows from the fact 

 that there are five rows and that in any particular head the number of bracts in any one 

 row is the same as in any other. But since the number of bracts per row varies within 



