466 BIIWS OF THE GA LAP A G OS ARCEIPELA G 0—RIDG WA Y. vol xix. 



(3) Geospiza. This heteromorphic genus is of very uncertain rela 

 tionsliii). SoTiie of the species resemble somewhat the Central Ameri- 

 can genus Cyanoloxia,^ but perhaps still more the West Indian genus 

 Melanospiza,- no other American Fringillidiie being enough like Geospiza 

 to suggest even distant kinship, unless it be the Central American and 

 South Ameru;an genus ()ryz<>horu,s, which, superlicially at least, recalls 

 silch Geospizce as G. magnirostris^ G. pachyrhyncha, and G.strenua in the 

 excessive shortness and thickness of the beak. The more slender-billed 

 species (formerly separated under the generic name Cactornis) have no 

 continental or West Indian prototy])e The only form closely resembling 

 them is the genus Cocornis, peculiar to Cocos Island, which is essentially 

 a small, slender-billed '■'■ G adorn is'''' with exactly the same sexual and 

 seasonal differences of plumage as the species of '■'•Cactornis'''' and Geo 

 spiza. It is a singular and most suggestive circumstance that the 

 peculiar departure from the normal Iringilline type, begun in the thicker- 

 billed '■'■Cactorni''' and carried, through a nicely graded transition, to its 

 extreme develoi)ment in Cocornis, should be in the direction of the 

 Hawaiian family mentioned under the head of Certhidea.^ 



(4) Camarhyncln(s. This is another heteromorphic genus, whose 

 variations of structure exactly parallel those of Geospiza^ the longest- 

 billed species [G.pallifhis) having been originally referred to " Cactornis." 

 I have been unable to tind a continental or West Indian genus that 

 could be of common origin with it, unless it be the West Indian genus 

 Pyrrhuhif/ra. Certainly the western Pernviaii genus Neorhynchus^ 

 the only one which has been mentioned as possibly related,^ is not its 

 prototype. Certain Hawaiian genera, by some writers referred to the 

 Fringillida' and by others considered to be thick-billed Dica'idic, 

 although very different from Camarhynehus in coloration, strongly sug- 



' Cyanoloxia, Bonaparte, Conspectus Avium, I, Aug. 15, 1850, p. 502, Type, by 

 elimination, Coccohoriis cyanoides, Lafresnaye. 



• Melano>iinza, Ridgway, new genus. Type, LoxigiUa richardsoni, Cory. 



Generic characters. —Similar to Geospiza, Gould (size of type intermediate between 

 G. fortis and G. fuliginosa), but tail relatively much longer and wing much more 

 rounded (first quill .shorter than seventh) ; mandible relatively broader basally ( basal 

 width cousi(leraV)ly exceeding length of gonys); culmen quite straight, and man- 

 dibular rami much narrower. Coloration: Adult male wholly deep black, except 

 legs and feet, wliich are brownish white. (P^emale and young unknown.) 



^This genus Covornis may possibly furnish the key to the derivation of the family 

 Coerebidif, since it shows unquestionable resemblance in form to the cbietiy West 

 Indian genus ('oereba{=^Certhiola, Sundevall). The close resemblance between the 

 adult male of Cocornis and the adult (the sexes being alike) of Coereha atrata is indeed 

 remarkable, so nuich so that there can be no doubt, in my mind, that the similarity 

 is something more than merely accidental. That Cocornis belongs on the fringillinc 

 side of the line, however, is proven by the fact that while the adult male is wholly 

 uniform black the adult female and the young male are varied with olive and rusty 

 above, while their under parts are conspicuously streaked with dusky on a yellowish 

 ground — exactly like tlie species of Geospiza; the sexes of Coereha, on the other hand, 

 being alike, and neither the adult nor young streaked beneath. 



^Salvin, Trans. Zool. Soc, Lond., IX, Ft. ix, 1876, p. 488. 



