Vol. XX 

 1Q03 



1 Snodgrass, Geosfiza, Cocor/n's, and Cer-thidia. 4^3 



scarcely more than Coconiis does from Certhidia. The difference 

 is not nearly so great as that between the slender-billed and the 

 thick-billed forms of Geospiza itself. Hence, a study of the 

 characters of these three genera, is suggestive of a possible deri- 

 vation of Geospiza from Cocornis and of Cocornis from Certhidia. 

 This, however, would place Geospiza in the Mniotiltidas ! 



In the descriptions of the skulls following it will be shown that 

 the Geospiza skull departs widely from that of any ordinary Frin- 

 gillid species. The writer, however, does not possess enough 

 knowledge of comparative avian anatomy to venture any theory 

 on the correct classification of the three genera discussed, or on 

 their possible interrelationships. A few facts are set forth in the 

 hope that they may be of value to others. 



It is probably not impossible that Geospiza, Cocornis, and Cer- 

 thidia may be genetically related. Cocos Island is not very dis- 

 tant from the Galapagos, and an ancestral Geospiza could easily 

 have gotten there from the latter place. Moreover the climate 

 and flora are utterly different on Cocos and the Galapagos, and 

 the conditions look very unfavorable for convergent evolution. 

 Some authors have claimed a common geological origin for the 

 islands. In such a case we would look for a relationship at least 

 between Cocornis and Geospiza. 



A study of the plumage phases of the Geospizce shows that they 

 are most probably descended from a plain yellowish-olivaceous 

 bird. They advance through six stages from this to the entirely 

 black phase. The lowest forms never go beyond the third stage, 

 being in this plumage when adult. ^ Therefore, Cocornis must 

 have branched off from one of the higher groups, for it is black in 

 the adult stage. This conclusion is not endangered by the fact 

 that it is antagonistic to the assumption that the bill and skull of 

 Cocornis are intermediate between those of Certhidia and the 

 lowest Geospiza. There are four groups of Geospiza separable on 

 a color basis, and in each the evolution of the bill has been inde- 

 pendent. The lowest members of the higher groups have bills 

 more similar to the Certhidian bill than have those of the lowest 



1 Discussed in Papers from Hopkins-Stanford Galapagos Expedition, No. 

 — , Birds, Snodgrass and Heller (MS.), Proc. Wash. Acad. Sci., Vol. 



