CH. xxir] GEOGRAPHICAL DISTRIBUTION 233 



forming perhaps 1-2 per cent, of the total of species of very 

 restricted area. 



Very many arguments against the old position have been 

 brought up above, e.g. on pp. 58, 81, 88-94, 141, 164-6, and 179. 

 No one has yet attempted to reply to any of these, which have 

 mostly been already published, but the position is obstinately 

 held, and the facts brought out by the study of Tertiary floras 

 are especially appealed to. These show that there are without 

 doubt, in the north temperate zone, a number of forms, perhaps 

 even as many as 600 to 1000, rather widely separated from their 

 nearest allies (when they have any such), and probably Tertiary 

 relics ; but it is not properly realised that these are a mere trifle 

 when compared to the local species that occur south of the tropic 

 of Cancer. Brazil alone has about 12,000 endemic species, usually 

 well localised; even the little island of Ceylon has nearly 250 

 species of the most localised distribution possible, almost half of 

 them occiu-ring each on one mountain top only, and it has nearly 

 800 whose area does not exceed 4000 square miles (63 x 63 m,). 



In view of the facts that have been brought up abo\'e, showing 

 the Avay in Avhich not onh' the areas occupied by endemics, but 

 those occupied by other species, are arranged in hollow curves, 

 and showing that this same type of arrangement also occurs in 

 the grouping of genera and families into sizes, the idea of relic 

 nature, or of special local adaptation (except in so far as this is 

 needful for all species, if they are to survive), must, it seems to 

 me, be abandoned for the great majority of cases, and the 

 mechanical explanation adopted in its stead, that area occupied 

 goes with age. Nearly all forms are to be looked upon as in- 

 creasing their area, and only a few, not most, as moribund. 



That this view is in all probability the right one to take of 

 the phenomena of dispersal is shown very clearly by the way in 

 which, accepting it, predictions as to distribution may be made, 

 and have as yet been uniformly successful (in almost a hundred 

 instances). Very strong evidence, and evidence based upon 

 definite facts, not upon a priori reasoning, is now required to 

 show that the hypothesis of Age and Area is unsound. 



But not only have we seen reason to accept Age and Area, but 

 also to accept the similarly "mechanical" hypothesis of Size and 

 Space (Chapter xii, p. 113), which asserts that Avhen one deals 

 with groups of allied genera the size of a genus depends largely 

 upon the area that it covers, i.e. ultimately upon its age. This 

 follows almost of itself when one has once accepted Age and 



