POPPY FAMILY 567 



The primary division of the key in each of the above-mentioned accounts rests upon size 

 and character of the torus rim, number of stamens, and condition of cotyledons, with respect 

 to entirety. Secondary distinguisliing characters, as set forth in the key, are mainly those 

 relating to longevity, and to the vegetative parts and the two outer floral whorls (the two 

 inner floral whorls receiving but brief mention even in the diagnostic accounts). In the present 

 treatment an effort has been made to determine anew the actual value of these characters 

 (a) by comparison of a large niunber of herbarium specimens from the same locality and from 

 different localities, (li) by comparison of specimens taken from marked plants at different 

 times of the year, (c) by application of the diagnoses of Greene and Fedde to the specimens 

 quoted by them wherever these specimens were available, and (d) by the results of garden 

 cultures. 



1. Torus rim. — Characters of the torus rim, which is a conspicuous structure in the per- 

 ennials, are not satisfactory as differentiae, since this organ shows marked variation in size 

 and form. In the following specimens the flowers show variation in width of torus rim as 

 indicated: Dinsmores Ranch, opp. Buck Mt., Humboldt Co., Tracy 4138 (\'n to 1 line wide); 

 I't. Isabel, Contra Costa Co., Dax'tj, Apr., 1897 (Vo to 1 line wide) ; Twin Peaks, San Francisco, 

 K. Srandegee 14e (Vi to % line wide) ; Palo Alto, C. V. Baker 43 (Vi to % line wide) ; River- 

 side, Hall 3713 (% to I14 lines wide) ; Chinese Camp, Jepson 6316 (% to '% line wide). The 

 same specimens serve to show the increase in width of torus rim which often occurs with age, 

 the mature fruits showing rims of greater width than the buds or flowers. The great variation 

 in width of torus rim at difi'erent seasons of the year as shown by marked plants in natural 

 colonies proves also that this character is subject to seasonal variation: Vaea Valley, Jepson 

 4721, 5163. 6276, 6278, 6791, 6846. It is further to be pointed out tliat variations in the 

 torus rim are simply those of degree and do not involve any morphologic change in structure. 

 An examination of the descriptions of the torus in Greene's Revision of Esehseholtzia shows 

 certain discrepancies between the species charact^'rs as set forth in the diagnoses and the 

 species characters as set forth in the key. We note the heading in the key, "Outer margin of 

 torus not obscure, etc. ' ' Under that heading we find certain species whose diagnoses contain 

 a contrary or inconsistent statement as follows: E. douglasii Walp., "torus rim remarkably 

 narrow and inconspicuous" (Pitt. .5:230); E. shasteiisis Greene, "torus exactly funnelform, 

 the rim wholly inconspicuous, hardly surpassing the inner margin" (Pitt. 5:234); E. diversi- 

 loba Greene, "torus turbinate, its rim only small and insignificant" (Pitt. 5:256); E. 

 stramin-ea Greene, "torus verj' small, with rim greatly reduced, hardly more obvious than the 

 inner margin" (Pitt. 5:257). 



In the true annuals, however, the rimless torus is constant. The specimens we cite below 

 under the several species verify this generalization. 



2. Stamens. — In the great majority of perennial forms the number of stamens, about 

 24 to 44, is always indefinite. Greene 's first key division is characterized by ' ' Stamens mostly 

 40 or more. ' ' In certain specimens, unf|uestiouably perennial and therefore belonging under 

 this first division, flowers are found bearing fewer stamens (as few as 16 to 27): Simpsons 

 Ranch, Sweetwater Creek, Eldorado Co., K. Brandegee 23e; Mt. Vision, Marin Co.. Hall 8510; 

 clay hills north of Ocean View, K. Brandegee 1.5e; sand hills at Pt. Pinos, Monterey Co., 

 Heller 6539. Certain specimens whose basal portion has not been preserved, but whose broad 

 torus rim should place them in the group having "stamens mostly 40 or more" according to 

 Greene's Revision of Esehseholtzia, are found to have in some flowers as few as 17 to 27 

 stamens: Bodega Head, K. Brandegee lOe, lie; Salinas River near Castroville, Monterey Co., 

 K. Brandegee 12e, 13e. 



Also the following unnoted discrepancies occur in the "Revision" itself, that is, certain 

 species falling under the first division "stamens mostly 40 or more" are described as having 

 stamens fewer than 40: E. biriirmita Greene, "stamens few" (Pitt. 5:249); E. amhigua 

 Greene, "stamens about 12" (Pitt. 5:251); E. vcrnalis Greene, "stamens about 16" (Pitt. 

 5:259); E. peninsularis Greene, "stamens definitely 8" (Pitt. 5:360). Under the division 

 "Stamens fewer, in some definitely 16, 12, 8 or even 4," occurs the species E. glyptosperma 

 Greene. In this species we find as many as 27 stamens (Barstow, K. Brandegee 24c; Needles, 

 Huby Warner), a number greater than in any of the four segregates just not<'d which were 

 included by Greene in his other section. Greene him.self assigns ' ' 30 or more ' ' stamens to this 

 species (Pitt. 5:292). This character therefore does not by itself separate the genus into two 

 distinct groups and is both too variable and too indefinite to be of real value. 



3. Embryo. — Results obtained by the study of embryos as found in the seeds of a large 

 number of herbarium specimens indicate that the condition of the cotyledons with respect to 

 entirety, while not wholly invariable, is sufficiently so to render this character extremely 

 valuable. From the following specimens, which are unquestionably perennial, data are obtained 

 as recorded: Surf, K. Brandegee 7e (7 seeds examined, cotyledons all bifid); Surf, K. Brande- 

 gee 6e (10 seeds examined, 9 -with cotyledons bifid, the remaining one with one cotyledon bifid, 

 the other apparently entire) ; Seaside near Monterey, K. Brandegee 2e (6 seeds examined, 

 cotyledons all bifid) ; Stanford, Santa Clara Co., C. F. Baler 174 (3 seeds examined, cotyledons 



