110 Mr. A. R. Wallace on the Genus Pitta. 



happens that it is on the continent that the species have the 

 widest range, though the varieties of physical condition in India, 

 from the Himalayas to Ceylon, must certainly be greater than from 

 island to island in the Archipelago. But those slight modifica- 

 tions which tend to bring a species into more exact harmony 

 with surrounding conditions can be accumulated and rendered 

 constant by " natural selection " in an island where intercrossing 

 with the forms of other districts is impossible ; while on a conti- 

 nent the same mode of action will be very often neutralized by 

 the intermingling of the various forms which must occasionally 

 come in contact with each other, except where the habits of the 

 animal are much opposed to locomotion. It is an interesting 

 confirmation of this theory that the only species of Pitta which 

 presents any well-marked varieties is that which has the widest 

 range. Two or three forms of P. bengalensis have been described 

 as distinct species ; but it is found that these forms are unstable 

 and graduate into each other. We have here an evident tendency 

 to produce distinct forms, which intercrossing continually pre- 

 vents ; but if continental India were broken up into three or four 

 large islands (a change which the southern extremity of Asia has 

 already undergone), we can hardly doubt but that a form spe- 

 cially adapted to the conditions, physical and organic, of each 

 island would be developed by natural agencies from the variable 

 material that we know already exists there. This segregation 

 has already taken place to a remarkable extent in the Archi- 

 pelago. Generally speaking, each island or little group of islands 

 has its peculiar species distinct from those of the islands that 

 surround it. Some of these cases of localized species are among 

 the niost extraordinary known. The little island of Banda, 

 hardly more than a mile across, has a species peculiar to it. 

 Ternate, a mere volcanic satellite of Gilolo, and not more than 

 ten miles from it, has a Pitta all to itself, though closely allied 

 to the distinct species which inhabits the large islands of Gilolo 

 and Batchian. The small rugged metalliferous island of Banca, 

 between Sumatra and Borneo (but so close to the former island 

 as to seem only a detached fragment of it), has actually two 

 species peculiar to itself; while, what is still more strange, the 

 two allied species of which they seem to be modifications (P. cya- 



