1 Q9?l 



PAYSON STUDY OF THELYPODIUM AND ITS IMMEDIATE ALLIES 249 



phyllus in which the seeds are narrowly winged and in C. cali- 

 jornicus in which the cotyledons are deeply trifid. 



The flowers.— The flower parts in Caulanthus offer more points 

 of taxonomic and phylogenetic interest than they do for most of 

 the genera of this family. The sepals may be nearly equal as in 

 the majority of the species, or one pair may be definitely longer 

 than the other, as in C. Coulteri. In many species they are more 

 or less saccate, this character reaching its extreme in C. calif orn- 

 icus. Species are now admitted to this genus with reduced, nearly 

 flat sepals, and in one species, C. anceps, they are distinctly 

 spreading. 



The petals are equally diverse. In the great majority of the 

 species they are narrow and crisped, with little differentiation 

 between blade and claw. Petals of this type are usually channeled 

 and curved outwards at the apex. In the past this type of petal 

 was considered a generic character of Caulanthus, and species that 

 did not show it were not included within the genus. In the pres- 

 ent work, however, it has seemed impossible to exclude species 

 with petals having plane blades and short claws. 



In color the sepals and petals vary from purple to yellow, or 

 the petals may be green in some species or individuals. An in- 

 teresting change in color is noticed in the sepals of C. Coulteri, 

 C. Lemmonii, and C. calif omicus as development proceeds. In the 

 bud these organs are a deep purple but as the flower develops the 

 color becomes paler until in the old blossom the purple color is 

 scarcely evident. 



The stamens of C. inflatus, C. Coulteri, C. Lemmonii, and C. 

 calif omicus are of particular interest because of the frequent ten- 

 dency of the two pairs of longer stamens to be more or less united 



TV.i« oharnotpr serves not onlv to confirm the 



by the filaments 



f C. califomicus with Coulteri but 



dence that the four long stamens have developed from two by 

 multiplication of the anthers and filaments. The present cases 

 would be considered partial reversions to an ancestral condition. 



The phylogenetic sequence of the different types of floral struc- 

 ture becomes evident only by a correlation of these characters 

 with the characters of the leaves, trichomes, pods, etc. Since 

 the most primitive species possess a closed and slightly saccate 

 calyx and narrow, crisped and channeled petals, it has been as- 

 sumed that these types are primitive and that from them the 

 other tvDes have been evolved. It would seem that the flat, 



