Antarctic Plants in Polynesia 303 



Fernandez but not on the coast of Chile. However, I fail to see 

 that there is any principal difference between these genera and, 

 for example, Lagenophora; Coprosma and Halorrhagis very 

 likely reached Juan Fernandez by way of land connected with 

 what is now southern Chile. The definitive feature of all these 

 occurrences is that the American representation is very weak as 

 compared with the west Pacific one. 



Another genus, Coriaria, might be quoted in this connection. 

 C. rusci folia used to be credited to both Chile and New Zealand, 

 but now seems to be confined to America and replaced in 

 New Zealand by C. sarmentosa, which extends to Warikauri 

 (Chatham), Kermadec, and Tahiti. Another species is credited 

 to tropical South America and New Zealand. When we consider 

 the distribution of the genus in the northern hemisphere, where 

 there are 4 species in an area extending from the Mediterranean 

 region across Himalaya to Japan, and the occurrence of Coriaria 

 in early Tertiary beds of Europe, the Antarctic nature of the 

 family becomes very doubtful. 



There are other genera for which an Antarctic-Polynesian dis- 

 tribution might be suggested. One of these is Agrostis, which has 

 a Marquesas species that comes close to a Magellanian one.^ 

 However, we do not know these species or their positions within 

 this largely Boreal genus well enough to discuss their possible 

 history in the south. 



If we leave the firmer ground of identical genera, some further 

 examples of Antarctic-Polynesian distribution may be quoted. 

 The Dianella group (Liliaceae) consists of 3 genera: Dianella, 

 extending from the Mascarene Islands over Indomalaya-Austra- 

 lasia-Polynesia, Stypandra in Australia, and Excremis in South 

 America. Metrosideros, an eminendy western Pacific genus, 

 reaches all high Pacific islands, has one stronghold in New 



