250 PROCELLARIIDiE. 



The exceedingly extensive range of many species of this family of birds 

 adds to the difficulties of their study when the evidently close specific relation- 

 ship between many of them is considered. But could we compare specimens 

 taken from hreeding -stations, much of our perplexity would, I believe, vanish, 

 and the slight differences observable in specimens captured at various points 

 on the high seas would at once assume a greater value, and definite laws of 

 geographical distribution would be found to prevail in these as in other birds. 

 It is on certain islands that the Procellariidse assemble in the breeding-season, 

 sometimes in countless numbers ; and after the duties of incubation and 

 rearing their young are accomplished these colonies disperse at large over a 

 vast tract of ocean, to assemble again the following year. Thus, then, for a 

 considerable portion of the year birds of closely allied species may be found 

 flying together ; but they separate to their respective breeding- quarters at 

 the proper season. 



From this it may be gathered that the fact of two or more closely allied 

 Petrels being found together on the open ocean is not by any means so strong 

 a proof of their specific identity as would be the case in most other birds. 

 It is to the uniformity or otherwise of birds when assembled at their breeding- 

 stations that characters of real specific value are to be traced. 



At present our knowledge is defective in this important respect ; but, 

 with some exceptions, I am disposed, at least in the genus CEstrelata, to 

 consider even slight differences of colour and dimensions as of specific value. 

 To some of these exceptions I shall refer under the species described below 

 as CE. arminjoniana and CE. trinitatis. 



I have gone carefully over the descriptions of each bird given in ' The 

 Ibis ' by Dr. Giglioli and Count Salvadori, and compared them with the 

 types themselves. In the descriptions given below I have adhered closely to 

 the original text, making the fewest possible changes. The measurements 

 have been retaken in Enolish inches and decimals. 



