DERIVATION OF THE FLORA AND FAUNA 265 



to a series of forms, probably ^ood species, rejKjrted from subaiitarctic America, 

 Falkland, Tristan da Cunha, Australia, etc. and possibly descending from an 

 old Antarctic stock. 



Peiiicltya: 13 species, 8 Mexico and Centr. America to Chile, south to I^\iegia 

 and Falkland, i Galapagos Is., 2 in Tasmania and 2 in New Zealand; /'. rigida 

 is linked to Andean species, but very distinct {2f;2). The genus is more diversi- 

 fied in the American sector, and this seems to be where it originated, having 

 reached New Zealand across the Antarctic, if not with (Janl/Zicria, well developed 

 in New Zealand, descending from a common Antarcto-tertiary stock. 



EmpctruDi is a bipolar genus, the family most likely of boreal origin. Con- 

 cerning /:". rubyum see 2./9. 781. 



Of the 7 DichoJidra species 5 are neotropical, i endemic in New Zealand 

 and 1). repens (inch sericed) spread round the world. It is common on the coast 

 of Chile and possibly adventitious in Juan Fernandez. The occurrence of an en- 

 demic species in New Zealand suggests that Antarctica witnessed part of its 

 history. 



Calystegia. About 25 species have been described, scattered over the globe, 

 C. scpinm sensu lat. reported from America, Eurasia, N. Africa, Australia, New 

 Zealand, Easter I., etc. and evidently very easily naturalized. The plant found 

 on Masafuera and also on the mainland was described as C. Hantelniajnn Phil, 

 and later identified with tiigiirioruiii from New Zealand; see 24^. 783. If this is 

 correct, C. tugiiriorum offers one of the very few cases of a species restricted to 

 Chile and New Zealand, but even if they are kept apart, they present a remark- 

 able case of disjunction. 



Selkirkia Berteroi, the only representative of Boraginaceae, so richly devel- 

 oped on the mainland, was regarded as an isolated, Old Pacific type {22y. 2,1,224. 

 593) until Johnston {148) showed that it comes close to Hackelia and diff"ers from 

 this principally by its arboreous habit. With Urtica fernandeziana, the species of 

 Eryngium etc. I refer it to the neotropical element. 



Rhaphithajnnus venustus is of neotropical ancestry (see above p. 208); the 

 second species is common in Centr. and S. Chile. 



Briquet placed Cuuiinia in the Stachyoideae-Menthinae next to Oreosphacus 

 Phil., a shrub of the high Cordillera in the boundary region between Chile and 

 Argentina, but this genus has a schizocarp of four nuts. It has also been com- 

 pared with Bystropogon L'Herit. (Canary Is., Colomb.-Peru, different sections). Ep- 

 LING (using the fancy name JoJiowid) referred Cummia to Prasioideae, an ancient 

 group showing great disjunctions: Prasiiun Mediterranean, Stenogyne, Phyllostegia 

 and Haplosiacliys in Hawaii, Bostrychanthus and Govipliostemma in Asia; Cuminia 

 differs in the shape of the corolla (see above p. 208), but even so it seems to repre- 

 sent a palaeotropic element in the island flora. 



Solamnn fernajidezianum is a distinct species of indubitable neotropical and 

 Andean ancestry. With regard to .S. robvisonianuni, see above p. 200. 



Nicotiana cordifolia has, according to GOODSPEED [112. 347), its closest resem- 

 blance, in flower structure as well as in general habit, to N. Rai))iondii. Crosses 



