Chap. XIII.] INHABITANTS OF OCBANIU ISLANDS. 179 



found nowhere else in the world) is often extremely 

 large. If we compare, for instance, the number of 

 endemic land-shells in Madeira, or of endemic birds in 

 the Galapagos Archipelago, with the number found on 

 any continent, and then compare the area of the island 

 with that of the continent, we shall see that this is true. 

 This fact might have been theoretically expected, for, 

 as already explained, species occasionally arriving after 

 long intervals of time in the new and isolated district, 

 and having to compete with new associates, would be 

 eminently liable to modification, and would often pro- 

 duce groups of modified descendants. But it by no 

 means follows that, because in an island nearly all the 

 species of one class are peculiar, those of another class,- 

 or of another section of the same class, are peculiar; 

 and this difference si'ems to depend partly on the spe- 

 cies which are not modified having immigrated in a 

 body, so that their mutual relations have not been much 

 disturbed; and partly on the frequent arrival of un- 

 modified immigrants from the mother-country, with 

 which the insular forms have intercrossed. It should 

 be borne in mind that the offspring of such crosses 

 would certainly gain in vigour; so that even an occasional 

 cross would produce more effect than might have been 

 anticipated. I will give a few illustrations of the fore- 

 going remarks: in the Galapagos Islands there are 36 

 land-birds; of these 31 (or perhaps 83) are peculiar, 

 whereas of the 11 marine birds only 3 are pecyliar; and 

 it is obvious that marine birds could arrive at these is- 

 lands much more easily and frequently than land-birds. 

 Bermuda, on the other hand, which lies at about the 

 same distance from North America as the Galapagos 

 Islands do from South America, and which has a very 

 37 



