268 



GENUS ATRIPLEX. 



minute scales, may be either present or absent. These scale structures, or sepals, were 

 first noted by Watson (Proc. Am. Acad. 9:108, 1874), who thereupon promptly reduced 

 Endolepis to Atriplex (1. c, 110). 



When Standley reached this group in his revision for the North American Flora, he 

 attempted to sustain Endolepis on the basis of the perianth in the pistillate flowers, and 

 he did this by the daring method of splitting phyllostegia into two species, assigning those 

 plants without the perianth to Atriplex phyllostegia, while those with perianth were 

 referred to his new Endolepis covillei, a species especially set up to receive them. An 

 inconsistency resulting from this treatment was noted by Macbride (Contr. Gray Herb. 

 N. S. 53:11, 1918), who pointed out that in the type species of AinpZex {A. hortensis) some 

 of the pistillate flowers have a regular perianth, as called for in all modern descriptions, 

 including that of Standley. Macbride then accepted covillei as a species and transferred 



Via. 32. — Leaves of Atriplex phyllostegia to illustrate variation in shape of blade and length of petiole; also to show lack of 

 concomitance with flower characters: o, b, c, leaves from the type of the form called A. covillei, perianth present 

 in pistillate flowers of this plant; d, «, /, g, leaves from the type of the form called A. draconia, perianth wanting 

 in pistillate flowers; h, i, leaves as in genuine A. phyllostegia (Reno, Nevada, Curran, 185923 UC), perianth 

 present as in A. covillei. All X 0.8. 



it to Atriplex, apparently without investigating the value of the character upon which 

 it was based. As a result of the present study, covillei is found to be entirely devoid of 

 substantiating characters and it can not be retained even in the most subordinate rank. 

 An examination of the material in several herbaria shows conclusively that the perianth 

 scales may be either present or absent, even on a single plant, and that when present the 

 number varies from 5 to 1. The size also is variable, while the tendency toward com- 

 plete reduction is not characteristic of plants of any particular geographic area, and can 

 not be correlated with other characters. Most of these facts are perhaps sufficiently 

 well brought out in table 25, which, however, lists only a portion of the collections exam- 

 ined. The others all point in the same direction. The lack of concomitance between 

 width of leaf and the presence or absence of scales is incompletely shown by figure 32. 

 Further evidence is furnished by plants from Fresno County, California, {Kearney 21, 

 US), which although having comparatively narrow leaves, as in the type of covillei, have 

 no perianth in the pistillate flowers. This collection carries genuine phyllostegia into 

 middle California, while only covillei is accredited to this State in the North American 

 Flora, a peculiarity in distribution to be taken into account by those who would retain 

 the latter in specific rank. 



The conclusion, therefore, is that Atriplex phyllostegia is in a fluctuating condition as 

 far as the perianth of the pistillate flowers is concerned. This structure is to be looked 

 upon merely as a vestige, sometimes consisting of several much reduced sepals, but 



