SPECIES BY INFINITESIMAL VARIATIOlJS. 13 



appear to show, Namunakuli in the south-east. This being 

 so, one must suppose that they have been evolved by muta- 

 tions rather than by natural selection of infinitesimal variations , 

 and this would also explain why most of them are so rare, the 

 mutations not having proved specially useful, and why they 

 affect mountain tops, the conditions there being perhaps 

 sufficiently different to cause a tendency towards mutation. 

 In general they have characters which are, so far as one can 

 conceive, useless in the struggle for existence; they occur 

 in places where that struggle cannot have been very keen, or 

 between very large numbers; they often occur alongside of 

 their most nearly alhed species , and very often the differences 

 in character are such as can hardly conceivably have arisen 

 by the selection of infinitesimal variations. 



But, now, to the whole of this argument it may be objected 

 that though mutation may serve very well to explain things 

 in the notoriously peculiar case of Ceylon, which though not 

 an oceanic island has nearly 30 percent, of endemic species, it 

 will not serve for other countries. Let us therefore consider some 

 of these. We may begin with Mauritius, and take a few exam- 

 ples from Baker's " Flora." The bulk of the endemics first 

 mentioned in the list are not accompanied by any widespread 

 species, and we may therefore pass them by, only asking why 

 their peculiarities should fit them specially for Mauritius. In 

 Calophyllum we come to the first instance of an endemic species , 

 C. parviflorum, Hving beside the almost cosmopolitan species 

 of the Eastern Tropics , C. Inophyllum. The two are very much 

 alike, but the latter has a globose fruit, the former an oblong- 

 rostrate one. The latter lives in the beach forests, the former 

 more inland. Now, is it to be supposed that the shape of the 

 fruit can have any effect upon the life of the species sufficient 

 to account for its being evolved by natural selection of infini- 

 tesimal differences, though it may be correlated with some 

 internal character fitting it for life more inland ? 



Why should the awned carpels of Sida glutinosa fit it for 

 life alongside of the nearly alhed cosmopolitan S. humilis ? 



