28 BOTANICAL GAZETTE foxy 
in a number of Cosmos species is the rostrate achenes. Cosmos 
bipinnatus Cav., C. parviflorus H.B.K., C. caudatus H.B.K., C. sul- 
phureus Cav., and C. ocellatus Greenm. are among those species dis- 
playing this character in a marked degree. A study of such species 
shows that the rostrate achenes are accompanied in almost every 
case by two other characters; namely, some shade of red in the 
ligules and the appearance of the interior involucre in the somewhat 
immature heads, suggesting the conspicuous inner involucre found 
so commonly in species of Coreopsis. But there are a few species 
of Cosmos in which the mature achenes tend to be erostrate. Thus, 
Cosmos crithmifolius H.B.K. and C. linearifolius (Schz. Bip.) 
Hemsl., in the many specimens that I have seen, fail almost uni- 
formly to exhibit achenes swollen below and distinctly long-rostrate 
above as in C. bipimnatus. Yet in color of ligules and character 
of involucre they harmonize perfectly with Cosmos. While 
neither of these two characters is absolutely diagnostic, their 
simultaneous occurrence, coupled with a tendency of the central 
achenes at maturity to be elongate, even though indistinctly 
rostrate, shows both species to be true Cosmos beyond all question, 
and not Bidens. 
Hemstey (Biol. Centr. Amer. 2: 203. - 3882), in dealing with 
the Compositae of Mexico, very correctly considered these two 
species as belonging to Cosmos. But “Bidens Seemannii,” a 
species so identical generically with Cosmos crithmifolius that _ 
Hemstey himself erroneously referred to it Parry and Palmer 
485 (true C. crithmifolius), he retained as Bidens. At a later date 
Asa Gray suspected Ghiesbreght 264 of being “Bidens Seemanii”’ 
and stated that, if it was, the name should become Cosmos See- 
mannii eee Amer. Acad. =* 16. 1883). In Gray Herbarium, 
the ¢ breght | tudied by Gray is still preserved in 
S pee Ss ideniical with Seemann 2or4 (in Herb. — 
Kew), thus ction Gray’s supposition. It is accompanied 
bya letter to Gray fron HEMSLEY, peters must have been written _ a 
: Foren 1881 and Beha: ‘later than citi Ube « dates. of eae oe 
