212 C. SKOTTSBERG 



Die Rliitcnkopfe sind bei Robinsonia erheblich zahlreicher und kleiner, kurz und breit 

 glockenformig, bei unserer Pflanze aber schmalglockig. 



Wiirden diese Sippen demselben pflanzengeographischen Gebiete angehoren, so 

 wiirde man sie sicher nicht generisch trennen konnen. Was hier aber wesentlich dazu 

 zwingt, ist, dass die Zweihiiiisigkeit, die ja hier die wesentliche Ubereinstimmung be- 

 dingt, bei den Kompositen zu den verschiedensten Malen in den verschiedensten und 

 auch in denselben (iru})pen entvvickelt wiirde und daher keineswegs als Kriterium 

 fur eine generische Verwandtschaft angesehen werden kann. 



To this I shall remark that nobody would think of uniting Rohhisonia and 

 Rlicti}wdeudron, because they differ in essential floral characters. Brachionostylmn 

 is, to judge from the description and plate XC: B (only ? known), closely related 

 to Robi)iso7iia but not to RJietinodendron. The mode of growth is not quite the 

 same, for in Robinsonia (as well as in Rheti7iodendroii) 2 or 3 innovations are 

 developed, and none of them continues the mother axis. Besides, Brachionosiylimi 

 has petiolate, penninerved leaves of a type current in dicotyledons and widely 

 different from the sessile, linear-lanceolate, broadly clasping and parallel-veined 

 leaves of Robinsonia. 



gayana Dene and thurifera Dene are a pair of related species. 



evenia Phil, and Masafuerae Skottsb. form another pair. 



gracilis Dene stands more isolated. 

 Symphyochaeta (DC.) Skottsb. See 2^g. 785. 



macrocephala (Dene) Skottsb. 

 Rhetinodendron Aleisn. Berterii (Dene) Hemsl. See above. 



Centaurodendron dracaenoides Johow. Not, as has been supposed, just an ar- 

 boreous Centaurca, but differing materially in flower structure [24J). 

 Yunqiiea Tenzii Skottsb. See 2jy. 163. 



Dendroseris C. Don. An isolated genus, perhaps distantly related to TJiavmo- 

 sevis Phil, from Desventuradas Is., but probably not to Fiichia Hook. fil. Ben- 

 lllAM stated (20.480) that the achenes of Dendroseris differ from those of the 

 Cichorioideae in general, and that also the involucre and the habit are different. 

 With regard to the achenes this is certainly true of Dendroseris s. str., and also 

 of PJioenicoscris, whereas they are of a rather normal type in Rea and Hespe- 

 roscris. All are devoid of paleae, which are present in Tlianinoseris and Fitchia, 

 a distinctive feature pointed out by 1^p:ntiiam, but the awned achenes of /7/f///^ 

 differ from those of all the other genera mentioned. In spite of the profound 

 discrepances in sucli im[)ortant characteristics as inflorescence, involucre, recep- 

 tacle and stigma Tlianinoseris has more in common with the Dendroseris as- 

 semblage than with other genera. 



macrophylla \). Don, macrantha (Bert, ex Dene) Skottsb., margitiata (Bert, 

 ex Dene) I look, et y\rn. and litoralis Skottsb., all lumped together by JOIIOW, are 

 well-marked s{)ecics; comp. 22^1.201-204. 

 Phoenicoseris Skottsb., 2./9. 787. 



pinnata (Bert, ex Dene) Skottsb. and fter^er/ana (Dene) Skottsb. (Masatierra) 

 are closely related but quite distinct; regia Skottsb. (Masafuera) differs from both 

 in leaf shape, but fertile specimens have not been found. 



