I go CARL SKOTTSBERG 



agree very well with the type; only, the involucral scales are said to be »sub 

 2 — 3-nervatae» (Proc. Amer. Acad. V. 137) and I find them i-nerved (fig. 32 i). 

 The achenes have a corona of 4 — 5 long bristles (fig. 32 k). These bristles are 

 present in all the achenes, while there is no trace of them in A. crassipes. On 

 the other hand, the glands described above seem to be absent. The plant 

 found by the writer in Tierra del Fuego and named linearifolia (K. Sv. Vet.- 

 Akad. Handl. LVI [1916], 318, Taf. 23, Fig. 27) differs from the ordinary one 

 in its much more compact growth and in the much shorter, more canaliculate 

 leaves. Perhaps it ought to be separated from linearifolia, but the coroniform 

 pappus is the same. 



It is, however, possible that A. crassipes is not confined to Masafuera. 

 In Herb. Kew I found an unnamed specimen, labelled »Otway Harbour, Gulf 

 of Penas, Patagonia. Coll. Mr. MoSELEY Jan. 1876. Reed June 1876s. It is 

 very like my plant; the leaves are a trifle narrower, the scape not at all in- 

 crassate, in spite of the achenes being ripe, but some scapes in crassipes are 

 less elongated and incrassate than the rest. The scales (fig. 32 1) have the 

 same general outline and there are a few glands of the same kind. The achenes 

 are more distinctly clavate, without any beak at all (fig. 32 m). At present, I 

 distinguish this as A. Moseleyi nov. spec. Further material is required to decide 

 whether it is better to bring it to A. crassipes. 



Area of distribution: Endemic in Masafuera. 



Robinsonia DC. 



DECANDOLLE, Prodr. VI, established three sections: Symphyochaeta (R. 

 macrocephala), Eleutherochaeta (R. Guyana and thurifera) and Eleuthero- 

 lepis (R. gracilis). In several respects R. macrocephala differs so much from 

 the other species that the most modern taxonomists would reserve the name 

 Robinsonia for this species alone. According to my opinion, we loose more 

 than we gain by splitting up an isolated genus like this, the species of which 

 are, without the slightest doubt, genetically related to each other but lacking 

 closer connection with all other genera. That, in many orders, apparently 

 small characters must be, and have always been, used to distinguish the genera, 

 is one thing; but to make this a rule is wrong: this mechanical and spiritless 

 splitting-up process may, perhaps, satisfy the ambition of the authors of all 

 these »new genera* but it has become a burden to all other botanists. Thus, 

 I retain Robinsonia in its old sense, distinguishing two subgenera, Symphyo- 

 chaeta, comprising R. macrocephala, and Eleutherochaeta to receive all the 

 other species, which have free pappus setae and deciduous corolla. I propose 

 to subdivide the latter subgenus into two sections, Symphyolepis with soldered 

 (R. Gayana and thurifera) and Eleutherolepis with free involucral scales 

 (R. evenia, Masafuerae, gracilis). There are some other differences between 

 these sections. In Symphyolepis the base of the pappus forms a conspicuous 

 coronula and the style of the J 1 flowers is shortly bifid with acute penicillate 

 branches; in Eleutherolepis the pappus setae are free almost to their very base, 

 and the style is deeply bifid with linear-truncate branches. Of the three species 



