294 ^^^^ S^oUsberg 



affinities, as shown in the sketch map accompanying this essay 

 (p. 295), has given some interesting results. The position o£ 

 sect. Periastelia (Hawaii and the Marquesas) is o£ special im- 

 portance. From its geographical position, it seems natural to look 

 for its closest relatives in the southwestern Pacific. But the 

 Tahitian species belongs to a different subgenus. In New Cale- 

 donia and New Zealand we find members o£ the same subgenus, 

 but they form another section not nearly related to Periastelia. 

 We discover its nearest ally where we would not be likely to 

 look for it, in the Magellanic Micrastelia. Simply because this 

 occurs in America, we are not entitled to classify the Hawaiian 

 species as belonging to an "American" element. Micrastelia is 

 a stranger in the neotropical flora. 



Collospermum, which I segregated from Astelia, perhaps 

 should be mentioned here because it goes back to astelioid ances- 

 tors and is a Pacific genus of Antarctic affinity. It has 3 species 

 in New Zealand, i in Fiji, and i in Samoa. 



The distribution of Gunner a is significant. The Indomalayan- 

 Australasian sector is the home of 2 of its 6 subgenera; Pseudo- 

 gunner a, with a single Malayan species of wide range; and 

 Milligania, with 8 species in New Zealand, i in Tasmania, and 

 I in New Guinea. The African sector is inhabited by subgenus 

 Perpensum with 4 species, i in South Africa, i on Kilimanjaro 

 and Usambara, i in Abyssinia, and i in central Madagascar. 

 (These may be regarded as subspecies of one species, but, even 

 so, they have a long geologic history behind them.) The Amer- 

 ican sector is the richest. It has 3 subgenera: Misandra, with 3 

 species, ranging from the Falkland Islands and Tierra del Fuego 

 to the tropical Andes; the monotypical Osteni gunner a^ recently 

 discovered in Uruguay and probably less distant from the Neo- 

 Zelandic Milligania than from any other subgenus; and Pangue 



