A. ARGENTEA, 285 



2. A. CAPUT-MEDUSAE Eastwood, Proc. Calif. Acad. II, 6:316, plate 46, 1896. — The form or state of A. argen- 

 tea typica in which the faces of most of the fruiting bracts are covered with flat, horny, acuminate, and often 

 twisted processes, and the bracts themselves plainly stalked. Such plants are especially common in northern 

 New Me.xico, but field studies indicate that there is no dividing-line between this form and typica. Tiie vari- 

 ability in the appendage character, also in the length of the stalks to the bracts, and in the distribution of the 

 flowers is now well known (see, for example, plates 43 and 44). Similar but not identical bracts are sometimes 

 encountered also in subspecies expansa. The type locality of capid-medusae is near Recapture Creek, along the 

 banks of the San Juan River, southwestern Utah. 



3. A. EXPANSA Watson, Proc. Am. Acad. 9:116, 1874. — A. argentea expansa. 



4. A. EXPANSA var. mohavensis Jones, Contr. West. Bot. 11:20, 1903. — Differs from genuine A. argentea 

 expansa only in the smaller and therefore sharper teeth along the margins of the practically sessile fruiting 

 bracts. Principally Californian, although the other form occurs there also. Standley has added the smaller 

 size of the bracts as a criterion (N. Am. Fl. 21 :47, 1916). All of these differences are due to a reduction in the 

 width of the herbaceous border in many Californian plants, so that the teeth are smaller and therefore seemingly 

 more acute. These far-western specimens have bracts which are also more uniformly sessile or subsessile and 

 with a less frequent tendency toward the formation of appendages. But as to dentation, the genuine expansa 

 type of bract is also reproduced in western California on plants in which other bracts have only the narrow, 

 finely dentate border (e. g., Redondo, Piemeisel 3706, US), and conversely, the mohaven.ns type sometimes occurs 

 on plants from New Mexico (Mesilla Valley, October 5, 1889, Wooton, US), together with bracts of typical ex- 

 pa7isa. These conditions are illustrated in the accompanying plates and text-figures. As to the difference in 

 the length of the stalk, it need only be said that the large series of specimens from Texas and New Mexico now 

 in herbaria shows sessile about as often as stalked bracts, although in the caput-medusae type they are usually 

 long-stalked, with ordinary sessile ones on the same plant (e. g., Archer County, Texas, Reverchon 823, US). 

 Type locality of variety mohavensis, Mojave region to San Bernardino, California. 



Fio. 36. — Fruiting bracta of minor variations of Atriplex argentea expansa: a, 5, e, from western California {Piemeiael S708 

 US) ; d, e, /, from Mesilla Valley, New Mexico (Wooton US) ; a, h, i, from western Texas (Earle 631 NY). All X 2. 

 (See under minor variation 4.) 



5. A. EXPANSA TRINERVATA Macbride, Contr. Gray Herb. 53: 9, 1918. — Based upon A. trinervata, which see. 



6. A. HiLLMANi Standley, N. Am. Fl. 21 :48, 1916. — Based upon A. argentea hillmani, which see. 



7. A. MOHAVENSIS Standley, 1. c, 47. — Based upon A. expansa mohavensis, which see. 



8. A. NODOSA Greene, Pittonia 1:40, 1887. — A. argentea expansa. An insect-stung monstrosity, according 

 to Jepson (Fl. Calif. 436, 1914). A fragment of the type is preserved at the University of California. It 

 seems to be a common form of expansa in which the long-stemmed bracts are covered with prominent append- 

 ages, as in capul-medusae (No. 2 of this list). 



9. A. RYDBERGi Standley, 1. c, 47. — A striking variation which perhaps should be treated as a distinct sub- 

 species, but too httle known and the distinguishing characters too subject to fluctuation to warrant a final dis- 

 posal at this time. The leaves are all cuneate or acute at the base. Such leaves sometimes occur in true 

 argentea, but always mixed in with a larger number with broad bases. The staminate glomerules are in slender 

 interrupted spikes, these 2 to 6 cm. long. Such spikes are unusual in A. argentea typica. They occur, however, 

 in some specimens, notably one from near Grand Junction, Colorado {H all 11045). In this the leaves are exactly 

 intermediate (fig. 37, d, e, f, g). Long, staminate spikes develop also in subspecies expansa (Westminster, 

 Orange County, California, June 20, 1896, McClalchie, UC). The type specimen of rydbergi, which came from 

 southeastern Utah, is young and but few of the lower bracts are well formed. These are compressed, flabelli- 

 form, the margins with 4 or 5 coarse teeth, the sides either smooth or slightly muricate. They suggest A. 

 powelli in shape, but lack the broad entire summit of that species and the leaves have an entirely different 

 venation. (See fig. 37.) 



10. A. TRINERVATA Jcpsou, Pittoda 2:305, 1892. — A. argentea expansa, but the large leaves irregularly 

 toothed and strongly 3-nerved from the base, the bracts also strongly nerved. The large size of the leaves is 

 associated with vigor of growth and doubtless is responsible in turn for the dentations, which are seldom pro- 

 nounced and never regular. All plants of expansa show a triple nerving of the leaves. The prominence of the 

 nerves is exceedingly variable and can not be correlated with the other characters enumerated. Macbride 

 (Contr. Gray Herb. 53:9, 1918) recently has attempted to sustain trinervata in varietal rank, calling dis- 

 tributional considerations to its support. But contrary to his assumption, it does not replace typical expansa 



