70 ISLAND LIFE 



find them now isolated in remote parts of the globe, the 

 natm^al inference is that the family of which they are 

 fragments once had an area embracing the countries in 

 which they are found. Yet this simple and very obvious 

 explanation has rarely been adopted by naturalists, who 

 have instead imagined changes of land and sea to afford a 

 direct passage from the one fragment to the other. If 

 there were no cosmopolitan or very wide-spread families 

 still existing, or even if such cases were rare, there would 

 be some justification for such a proceeding ; but as about 

 one-fourth of the existing families of land mammalia have 

 a range extending to at least three or four continents, while 

 many which are now represented by disconnected genera 

 are known to have occupied intervening lands or to have 

 had an almost continuous distribution in tertiary times, 

 all the presumptions are in favour of the former continuity 

 of the group. We have also in many cases direct evidence 

 that this former continuity was effected by means of exist- 

 ing continents, while in no single case has it been shown 

 that such a continuity was impossible, and that it either 

 was or must have been effected by means of continents now 

 sunk beneath the ocean. 



Concluding Remarks. — When writing on the subject of 

 distribution it usually seems to have been forgotten that 

 the theory of evolution absolutely necessitates the former 

 existence of a whole series of extinct genera filling up the 

 gap betAveen the isolated genera which in many cases now 

 alone exist ; while it is almost an axiom of " natural selec- 

 tion " that such numerous forms of one type could only 

 have been developed in a wide area and under varied 

 conditions, implying a great lapse of time. In our 

 succeeding chapters we shall show that the kno^A^ and 

 l^robable changes of sea and land, the knoAMi changes of 

 climate, and the actual powers of dispersal of the different 

 groups of animals, were such as would have enabled all the 

 now disconnected gi'oups to have once formed parts of a 

 continuous series. Proofs of such former continuity are 

 continually being obtained by the discovery of allied extinct 

 forms in intervening lands, but the extreme imperfection 

 of the o-eolooical record as reo^ards land animals renders it 



