756 ON PETEELS PBOM THE KEEMADEC ISLANDS. [Bec. 5, 



Petrels unless it be the somewhat similar relation in colour between 

 Diomedea regia and Diomedea exulans ; but these two closely 

 related species are said to breed on different islands and at slightly- 

 different times \ Macgillivray was the first to call attention to 

 this peculiarity in the birds of the Kermadec Islands (Zoologist, 

 1860, p, 7133), where he found incubating birds in both phases of 

 plumage. More recently Mr. Osbert Salvin has discussed the 

 question in Eowley's ' Ornithological Miscellany,' vol. i. p. 254. 

 He says that the colouring is not dependent either on sex or on 

 age, and that the colouring of the first plumage is retained thi'ough 

 life, both of which statements the present paper tends to confirm. 

 He considers, however, that ihe two phases of plumage do not 

 indicate different species, or even different varieties, but he thinks 

 they are probably due to a kind of dimorphism. The word 

 dimorphism has been used in zoology in several different senses, 

 but it always denotes that the two forms are related to each other 

 either as twin brothers and sisters, or else that there is an alter- 

 nation of generations between them. As Petrels lay only one egg 

 in a season, the first relation cannot hold here, so that — if it is 

 really a case of dimorphism — eggs laid by bicolour parents should 

 sometimes hatch out unicolour offspring and vice versa, of which 

 there is no evidence at present. Also, if it be truly a case of 

 dimorphism, there ought to be no intermediate varieties between 

 the two forms, whereas these intermediate varieties seem to be 

 sufficiently common to have led some ornithologists to the opinion 

 that the two forms are merely adult and young of the same species. 

 If we reject the idea of dimorphism as improbable, and that of 

 changes due to age as disproved, we have three different hypotheses 

 to choose from to explain the facts : — 



1. Two distinct species, sometimes producing hybrids. 



2. One excessively variable species, one form producing, or 



partially produciug in an irregular way, the other. 



3. Two species developed by ordinary variation going on for a 



long time, while the intermediate forms have not become 

 extinct. 



An examination of the breeding-ground would probably enable 

 us to decide which of these hypotheses is the correct one. If the 

 first is correct, the young shoidd always either closely resemble the 

 parents, or be a distinct hybrid two parents of which had different 

 styles of plumage. If the second is correct, then any variety 

 might produce any other and the offspring should generally be 

 different from the parents. While if it be the third hypothesis 

 which is true, then each form should produce yoimg very like 

 themselves, and the two parents should always resemble each 

 other. For my part I strongly incline to the last hypothesis, 

 which is more in accordance with what we know in other cases 



* Biiller, Trans. N. Z. Inst voL_xxiii. p. 230 and vol. xxiv. p. 68. 



