CRITERIA. • 241 



to the plumule. These various positions are illustrated in the plates accompanying 

 the descriptions of the species. The position of the radicle, whether inferior or superior, 

 is utilized in the assembling of the species into two subgenera, Euatriplex and Obione, as 

 already discussed. 



In addition to the vertical seeds just described, there are also horizontal seeds in one 

 species, namely A. hortensis. In this species the vertical seeds are borne naked between 

 the bracts, whereas the horizontal seeds are produced only in those flowers in which a 

 calyx is present. According to Eichler (Bluthendiagr. 2:84, 1875), only the flowers with 

 horizontal seeds are a regular part of the inflorescence, the others occurring as adventi- 

 tious shoots. Furthermore, a few introduced species (patula, rosea, etc.) produce two 

 kinds of vertical seeds: (1), small black seeds with convex sides, the tip of the radicle 

 scarcely produced; and (2), larger, brownish seeds with flat sides and a shallow groove 

 near the margin, the radicle strongly produced. Only the latter are included in most 

 descriptions. Details as to the different kinds of seeds have been discussed by Collins 

 (Seeds of Commercial Saltbrushes, U. S. Dept. Agr. Div. Bot. Bull. 27, 1901). 



Additional seed characters doubtless would be very useful in the classification of Atri- 

 plexes, if consistently applied. The paucity of mature seeds in herbaria precludes the 

 extensive use of size and color as criteria until further field studies and collections can be 

 made. The samples preserved in seed laboratories are not very helpful in this connec- 

 tion, since they are but rarely accompanied by full specimens necessary to their accurate 

 identification. It is believed that size of seed will prove to be a more reliable criterion 

 than size of fruiting bracts, since the latter often depends largely upon the extent to 

 which such purely vegetative structures as free margins and appendages are developed. 

 The dimensions and color of the seeds as described in the present paper apply to these 

 bodies only in their fully mature condition. Much confusion would result from measure- 

 ments or colors determined from juvenile material. 



Flowers. — The flowers of Atriplex are of two sorts, staminate and pistillate. Thus far 

 the former have been scarcely utilized for taxonomic purposes, except as to their arrange- 

 ment, as will be described under inflorescence. The depth to which the calyx is lobed or 

 cleft is much greater in some species than in others, but there is also much variation within 

 single species. The number of calyx-lobes is sometimes given in descriptions, but it is of 

 doubtful value and never to be depended upon as a criterion unless checked by the exam- 

 ination of a large series of specimens. Field studies indicate that the staminate flowers 

 may be either 4-merous or 5-merous in the following species: patula, rosea, semibaccata, 

 pusilla, parishi, truncata, argentea, coronata, powelli, leucophylla, elegans, and leniiformis. 

 The variation holds even within single subspecies of these, and in most of them it has 

 been found on individual plants. On the other hand, only 5-merous flowers have been 

 found on bracteosa, although a large series from different localities was examined. Simi- 

 larly, graciliflora and saccaria yielded only 5-merous flowers, but only a limited number 

 of plants were studied. Usually the number of stamens is the same as the number of 

 calyx-lobes in the same flower. Cases of 5 stamens accompanied by a 4-merous calyx 

 have been found in truncata and leucophylla. The number of stamens and lobes is some- 

 times reduced to 3 each in elegans and probably also in other species. Cases of a double 

 calyx, that is, with 10 lobes, are not rare. The original number of calyx-lobes and sta- 

 mens doubtless was 5, but it would be unsafe to assume that when this number is found the 

 species in hand is primitive. This would lead to the placing of species like bracteosa at 

 the bottom of the phylogenetic tree, while by all other criteria they are classed as among 

 the most highly specialized forms. 



Reference has thus far been made only to staminate flowers. The pistillate flower is 

 so simple as to yield but little of value in addition to the pistil itself, which has been 

 described under the heading of seeds. Obviously the perianth is a mark of primitiveness 



