176 ENDEMISM AND DISTRIBUTION: GENERA [pt. ii 



shows them arranged in this way, with many monotypes (or 

 genera with one species only), a fair number of dit3^pes, and a 

 tail of a few larger genera. When plotted graphically they con- 

 sequently form the hollow curve that we have begun to meet so 

 often in dealing with distribution (cf. fig. on p. 177). One 

 must make allowance, in considering the figures above given, for 

 the "lumping" that is practised in my Dictionary, especially at 

 the fives and tens. 



If one add up the grand total of 1582 endemic genera of all 

 the islands of the world, one finds that they show 1037/1 (1037 

 of one species), or 65 per cent, of the total, and 233/2, or 14-7 

 per cent., these two making up nearly four-fifths of the w^hole. 

 There are 104/3, or 6-5 per cent., 53/4, 49/5, and so on, the largest 

 endemic genus of islands being Oncostemoji with CO species. If 

 one take for comparison the endemic genera of Brazil, 533 in 

 number, one finds 334/1, or 62 per cent,, 91/2, or 15-2 per cent., 

 33/3, or 6-2 per cent., and so on, the largest having 50 species. 

 In both these cases the same type of result, showing a well- 

 marked hollow curve, is obtained, and one gets the same what- 

 ever region of the world one may try for endemic genera, e.g. any 

 of the other countries of South America, or South Africa (cf. the 

 first two and 4th and 7th curves in fig. on p. 177). 



It is worthy of notice that in these two instances, the islands 

 and Brazil, the percentages of genera of different sizes are much 

 the same, the monotypes for example being 65 per cent, in the 

 one, and 62 per cent, in the other. The islands, which actually 

 cover about two million square miles, would probably be nearly 

 equal to Brazil if the included seas were taken. The average 

 number of species per genus is also not unequal (islands 

 1582/3461, average 2-1; Brazil 533/1291, average 2-4). 



The endemics of mountains are also as a rule small genera, 

 though there are a fair mmiber of exceptions to this, but only 

 in the large mountain chains. In the Andes, for example, there 

 are Chaetanthera (30 species), Cinchona (40), Cristaria (30), Nas- 

 sauvia (50), Psammisia (35), Fuya (25), and many more of 

 smaller size. 



One may go on to deal with still larger floras, and find that 

 they are arranged in precisely the same way, so that the pheno- 

 mena shown by the endemic genera are exactly paralleled by 

 those shown by genera that occupy more area. If one take (as 

 usual from my Dictionary, in which uncertain fours are counted 

 as fives, etc.) the genera that are confined to single continents 



