BIRDS OF NORTH AND MIDDLE AMERICA. 491 



recognized to consist of two or more genera instead of one, tlien, to 

 be consistent, Camar^hynchus must also be divided into two genera. 



A ver}^ careful consideration of all the facts in the case, as they 

 appear to me, compels me to conclude that the safest course is to 

 regard the species of ^' Cactornis'''' as simply more slender-billed 

 Geo^pha'^ the degree of departure from the tjqiical Goespizine bill 

 being largely a specific character. This is a conclusion which I regret 

 haying to adopt, for I regard large genera as a nuisance, and would be 

 glad to haye the slighest excuse for keeping Geospiza to its old limits. 



As to the relationships of the group to which the generic name 

 CamarTiynchus Gould has been generally applied, while admitting the 

 difEculty of formulating strongly-marked structural characters, I ney- 

 ertheless cannot follow Messrs. Rothschild and Hartert ^ in referring 

 the latter to Geospiza., for reasons explained under the head of Camar- 

 hynchus^ on page 476 of the present work. 



As a matter of conyenience, I haye separated the species into two 

 groups, which correspond in their limits with Cactornis and Geospiza 

 as usuall}' recognized;^ but how slight and unsatisfactory a basis this 

 diyision rests upon ma}" be seen b}" reference to the characters given 

 in the first part of the following '"key to the species." 



Owing to the gradual transition from one form to another, and the 

 almost perfect resemblance between them in coloration, I have found 

 it impossible to construct an analytical ' ' key " to the species after the 

 usual plan, liut have drawn up the following as an aid to their more 

 readjMdentirication. I am prevented from making the "key'"' more 

 satisfactory 1)\' the circumstance that I have no specimens of G. 

 assunilis and G. Ixirringtonl for comparison with the other so-called 

 CactorvL and am therefore unable to give comparative measurements 

 of these forms alongside of G. fatlgata and G. ahlngdoni. 



In another respect I am, much to my regret, compelled to dissent 

 from the conclusions of Messrs. Rothschild and Hartert. In their 

 most valuable work, already referred to, these gentlemen say^ that 

 in the case of certain genera of Galapagos Passeres (the present one 

 among the number) their ''material has generally left very little 

 doubt" whether the}' should "treat a form as a species or subspecies." 

 Nevertheless, their decisions in this respect are not in all cases satis- 

 factory, and in one or two cases almost certainly erroneous.* When 

 we consider that in this genus, as here restricted, the plumage is 

 alwa^^s practically and in the case of a majority of the forms identic- 



1 Novitates Zoologies, vi, 1899, 152-154. 



'^ With this difference: Cactornis pallida Sclater and Salvin 1 have transferred to 

 Camarhynchns, to which it is certainly far more nearh^ related (see p. 476). 



^Novit. ZooL, vi, 1899, 142. 



*For example, their reference of Geospiza bretirostris to G. co)tirostris, as a sub- 

 species of the latter, G. brevirostris being a true " Cactornis." 



