260 CONTRIBUTIONS FROM THE NATIONAL HERBARIUM. 



disputed we have Costilla ulei, C. fallax, and C. australis; that is to 

 say, the known forms of the strictly South American group, with the 

 leaves more or less rounded at the base. 



Costilla ulei is not yet well known, but its fruits present such well 

 defined peculiarities as to segregate it widely from sill other studied 

 species. Each receptacle contains only 3 to 5 elongate nutlets sur- 

 rounded by a soft pulp; the leaves also appear to be smaller than in 

 any other species. 



Costilla fallax is another very aberrant type on account of its dry 

 fruits completely detached from each other, as are the female flowers. 

 The leaves are almost glabrous and shiny above, and covered beneath 

 with very short, soft hairs. 



Costilla australis offers a unique character in the shape of its style 

 and stigmas. These latter are short and broad and completely 

 exserted. The female flowers seem also to be free to the base, with 

 an urceolate perianth. This type is very imperfectly known, but its 

 localization at the southern extremity of the generic area makes it 

 quite probable that it is a well defined species. 



We could include among those strongly individualized types my 

 Costilla daguensis, but I shall refrain from doing so at present on 

 account of its possible identity with the preceding species. The only 

 obvious difference between the two consists in the shape of the style 

 and stigmas, and the character of these organs as described by Hems- 

 ley is so much out of the usual line that I am compelled to suspect an 

 error of observation or a casual anomaly in the specimens investigated 

 at Kew. 



The Central American forms have large leaves, distinctly cordate 

 at the base, and they are so uniform in their general aspect that up to 

 the present they have been considered as all belonging to one species, 

 Castillo elastic//. Observation, however, has shown that the floral 

 parts, as well as the fruits, present constant and definite differences. 



In a first subdivision, including Costilla lactifua and C. costarirana, 

 the primary male inflorescences are always borne on very short and 

 thick peduncles (pis. 27, 31). This character, which we know to 

 be really constant in both species, is in itself quite sufficient to segre- 

 gate them from the forms of the remaining group in which the same 1 

 stipes are relatively long and slender (pi. 37). On the other hand, 

 there can be no doubt as to the characters that distinguish C. lactifua 

 from C. costaricana. The complemental inflorescences are unlike; the 

 style is short, thick, and bristly, with longstigmns in the former, while 

 the same organ is long, smooth, and with tongue-like, reduced stigmas 

 in the latter. In the first, further, the drupes are not completely fused 

 together in the pulp, as they are in the latter, and the number of 

 seeds in each receptacle is less. 



