442 Distribution of Tubinares in North Atlantic Islands. 



majority of Petrels and Shearwaters wliicli breed in the 

 north Atlantic Islands^ and which are hardly ever recorded 

 outside their own particular sphere until we draw near 

 to another breeding- station, perhaps at the other side of 

 the world. Even in such a comparatively small radius as 

 is included in the accompanying map, the distribution 

 of individual species is very striking. For instance, 

 0. castro and CE. m. fe(2 each breed in the Madeira 

 Group and in the Cape Verde Archipelago, but do not, so 

 far as we know, breed on the Canary Islands, which lie 

 between them. Another similar case is P. m. hypoleuca 

 breeding on the Salvage Islands and on the Cape Verde 

 Islands, but not on the Canary Group. Surely this points 

 to very circumscribed distribution at the present day, else 

 why should not P. m. hypoleuca have taken up its abode 

 on M.ontaiia Clara, an uninhabited island of the eastern 

 Canary Islands suited in every way to its requirements ? 

 The reason why so many of the Petrels inhabiting the Atlan- 

 tides have their nearest allies in the Pacific Ocean is certainly 

 hard to explain, particularly as there are no links or geogra- 

 phical subspecies in the regions between. Bulw^er's Petrel 

 (^B. bulweri) inhabits, in addition to the north-east Atlantic, 

 the Hawaiian Islands, and, as far as we know, the bird does 

 not breed anywhere in the western part of the Atlantic. 



The only possible explanation of the peculiar range of 

 Bulwer's and other Petrels whose different colonies are often 

 separated by vast areas, is to be looked for in the great 

 antiquity of the order to which they belong, and to the 

 enormous changes which have taken place in the distribution 

 of land and water on the surface of the earth. The question 

 is too complex to be more than alluded to in this paper, but 

 it seems to be a much neglected study. In all the compre- 

 hensive works dealing with " Petrels "" which have appeared, 

 I can find no attempt at an explanation of the present geo- 

 graphical distribution which many species enjoy. 



In former days the Petrel family must have had a very 

 extended range, which is yearlybecoming more circumscribed. 

 The birds which at one time ranged universally from the 

 north Atlantic to the south Pacific are now becoming 



