268 C. SKOTTSBERG 



The distribution of Lagowphora clearly testifies to its Antarctic origin: 3 

 Magellanian species, one of them also on Tristan da Cunha and one on Masafuera, 

 7 in New Zealand, 2 in Australia, one of these north to the Philippines, i in Fiji 

 and 3 in Hawaii, the four Pacific ones connected with L. Bil/ardieri Cdiss. (Austral.), 

 whereas the Magellanian species point toward L. puiuila Cheesein. 2iX\d. petiolata 

 Hook. fil. (New Zealand). 



In Iiyii:;tyo}i we meet again with a large boreal genus strongly represented 

 along the Andes and ranging south to Patagonia, P^iegia and P'alkland; 135 species 

 are reported from Kurasia, inch India, 345 from North America, 35 from Centr. 

 America and the W. Indies, 95 from S. America, 10 from Africa, 8 from Australia 

 and a single species from Oceania (Rapa), E. rapensis F. B. H. Brown, and this is 

 compared with our island species. To judge from the description it has the same 

 habit as these, a shrub about 3 dm tall with the leaves in terminal rosettes. In 

 I lawaii the genus is represented by l^ctra)nolopiu7)i, which is very close to Erigeroit\ 

 \'lKKii.\ri'KR [22g. 182) suggested that the Juan P^rnandez species came near the 

 Hawaiian /:. lepidotus Less., which is now referred to Tetraniolopmnt, and that we 

 ought to look for relatives among the Andine and Mexican forms. Among our 

 island species li. nipicola differs much in habit from the rest, but all belong together 

 and most likely represent a special branch of the Andine Erigeroii flora which, in 

 its turn, comes from a boreal stock. 



To what I said above about GnaphaEiuj>i spicifornie nothing can be added 

 at present. Chile is well provided with poorly limited species related to G. pur- 

 pur ciiui L.; most of them are badly known. Their boreal parentage can hardly 

 be doubted. 



Ahrotajiclla resembles LagejiopJiora in its distribution, but is absent from Oce- 

 ania; tiie majority inhabits New Zealand with its subantarctic islands (9 sp.), i is 

 found in X'ictoria, 2 in Tasmania, i in New Guinea and 5 in S. America (W. 

 Patagonia to h\iegia and P'alkland); in addition, one is found on Rodriguez I. 

 Tiie genus is, as it were, tricentric and its Antarcto-tertiary character indisputable. 



The concentration of isolated arborescent Compositae in the Pacific was em- 

 |)hasized by Bkn'IIIAM; tiie main grouj)S of the family are represented among them^ 

 and the accunuilation of endemic genera in Hawaii and Juan P^rnandez has led 

 to nuich discussion. Tiic enormous development and differentiation in the family 

 on American soil is an undeniable fact, but simply to derive the endemic Pacific 

 genera from America as did (ill'l'V [121) does not seem [)ossible. Speculations (comp. 

 f. inst. Sl'/i (111:1.1, 2r()) led back to Antarctica, but not until BracJiioioslyluin was 

 discovered in .\ew (iuinea and found to be nearly related to Robinsoiiia o'i ]\x^Vi 

 l^>rnandez were we able to stand on tolerably firm ground. 



With regard to Geiitaurodoidroji, to which J//';/*//^'^^'? seems to be related, the 

 situation is different. Coitaurca and all the genera of Centaureinae belong to the 

 Old World with the exce})tion of a single species in N. America and a few in 

 the Andes, south to Centr. Cliile. Tlie grou{) ranges over luirope with a strong- 

 hold in the Mediterranean, the Orient, P'.thiopia and through Centr. Asia to Japan. 

 The Chilean species of Centaurca belong to the Plectocephali, but Ceiitauroden- 

 droji differs not only from this section but from all in sex distribution (24J). It 



