INTRODUCTION. XXXIX 
as already observed, justly be regarded as appertaining to any one of the great 
primary regions of vegetation. The characteristic endemic shrubby Composite* and 
Lobeliacee are most nearly related to American forms; Perrottetia is a Mexican and 
Colombian genus, and the pomaceous Osteomeles anthyllidifolia is a member of a genus 
all the other species of which are Andine. Nama is otherwise restricted to America; 
and Hillebrand regards the American Lythrum maritimum, Daucus pusillus, and Aster 
divaricatus (== A. evilis) as indigenous. Prominent among the Australian types and 
common in all the islands according to Hillebrand are :—Metrosideros polymorpha, an 
exceedingly variable tree or shrub scattered throughout Polynesia, eastward to Pitcairn 
Island, and Acacia koa, already alluded to. The Australian genera Cyathodes and 
Exocarpus are also represented. Cyrtandra, of which there are thirty-two species 
endemic in the Sandwich Islands, is more Malayan in character, and many other such 
relationships exist, besides other more remote ones in the highest mountain flora, 
which includes such forms as Luzula campestris, Rumex, Silene, Ranunculus, Drosera 
longifolia, Hydrocotyle interrupta, Fragaria chilensis, Vicia, Vaccinium, Aster, and 
Artemisia. 
Added to the foregoing elements is a sea-shore element consisting almost exclusively 
of species having a very wide range in the Old World; many of them from the eastern 
coast of Africa or the Mascarene Islands and India to N. Australia, the Marquesas 
Islands and Easter Island. Further, the vegetation of the small and remote coral islands 
is entirely of this character. . 
Besides the phyllodineous Acacia above alluded to, it has long been known that 
there were two or three other noteworthy outlying Australian types in Madagascar ; 
but no important addition to these has been made by recent explorations. One or two 
species of Hibdertia (Dilleniacez), and two or three of Rulingia (Sterculiacee-Lasio- 
petalee); and the genus Adansonia is represented by one endemic species in Mada- 
gascar, one in tropical Africa, and one in North Australia. These widely-sundered 
areas of distribution of closely-allied forms have given rise to much speculation, and it 
is very problematical whether satisfactory paleontological evidence will ever be forth- 
coming which will account for the existing distribution of plants. 
BoraNIcaL DIVISION OF THE EARTH INTO PRimaRY REGIONS. 
From the data adduced in the preceding paragraphs, and numerous more familiar 
facts which it is unnecessary to repeat, it is clear that a system of botanical geography 
should be based upon a small number of primary regions, similar in many respects to 
Sclater and Wallace’s zoological regions. It has been shown where the botanical 
regions do not even approximately coincide with the zoological regions, and to some 
® See Bentham in Journ. Linn. Soc. xiii. p. 555. 
+ The Madagascar plant, long supposed to be a species of Evocarpus, as only foliage was known, is Neobaronia, 
Baker, a new genus of Leguminose, of which flowering specimens of two species have been collected by Mr. Baron. 
