the Tubinares in the North Atlantic Islands. 453 



island^ but not very abundant ; lays in November/' What 

 more likely than that this so-called T. melanog aster — an in- 

 habitant of the Australian Seas and south Indian Ocean — is 

 in reality Oceanodroma castro (Harcourt) breeding in St. 

 Helena in Novetnber. 



General Conclusions. 



The very extraordinary range of this species has excited 

 the curiosity of every naturalist who has turned his attention 

 to the geographical distribution of the Procellariidse. In 

 the north Atlantic it is a remarkable fact that up to the 

 present there is no reliable record of its having bred on any 

 of the islands of the Canary Archipelago, although it breeds 

 in every other group of the north Atlantic islands from the 

 Azores to the Cape Verde Islands, and as far south as 

 St. Helena. 



In the groups with which this paper deals it will be seen 

 by referring to the appended schedule that Oceanodroma 

 castro may be found engaged in nesting duties in every 

 month of the year. 



It appears, therefore, that this Petrel has two main 

 breeding-seasons — at any rate, in the Madeira Group, which 

 are the only islands upon which consecutive notes have been 

 made extending throughout the entire year and, in this case, 

 for a number of years. The only alternative which presents 

 itself is that O. castro breeds indiscriminately "whenever 

 the spirit moves \i" which is a theory I cannot bring myself 

 to believe ! 



Undoubtedlv the bird is somewhat erratic in its breedine- 

 season, which accounts for the prolonged period in which 

 eggs and young birds may be found. 



In studying this problem we must bear in mind two im- 

 portant factors : 



1. The breeding-season varies considerably iu each 

 separate group of islands;, and in some cases even in 

 each individual island of the same group, and because 

 we find the birds breeding in the Desertas in June, it 



