334 SNODGRASS AND HELLER 



of the bill is ten millimeters and the greatest depth thirteen millime- 

 ters. One adult male has a deep groove on each side of the culmen 

 running from the nostril, parallel with the curvature of the culmen, to 

 the tomium, exactly as does one of the specimens of G. c. profinqua 

 from Tower. Some of the others have less distinct grooves. 



Hence in the shape of its bill this species might be related in three 

 different directions with the G.fuliginosa series through G. debili- 

 rostris, with the G. scandens series (both of these being very close), 

 and finally, but not so closely, with G. conirostris through G. c. pro- 

 pinqua. 



The adult males differ from the other species of Geospiza in having 

 the pale marginal parts of the under tail coverts of a very decided rusty 

 or even chestnut color. This is the only character by which the form can 

 be specifically separated from either G. debilirostris or G. scandens. 



We have no females which are surely adults. All of the female 

 specimens in the collection have yellowish bills, or yellowish with 

 dusky at the base and at the tip. They are plain brown above with 

 the feathers edged with buff ; below heavily streaked with dark brown 

 except on the middle of the abdomen, which is plain buffy whitish. 

 All of these have prominent wide rusty almost chestnut edgings to the 

 middle and greater wing coverts. 



Two other clearly young specimens, one a female and the other a 

 male, having purely yellow bills, have the spots below mostly confined 

 to the breast and the region in front of it. The abdomen is whitish 

 in the middle, strongly shaded with buff on the sides and on the flanks. 

 The wing coverts have bright chestnut borders. 



Two adult males of G. conirostris propinqua^ including the speci- 

 men of this species from Culpepper, have a slight tinge of chestnut on 

 the under tail coverts. 



The fact of the similarity of shape between the bills of G. septen- 

 trionalis and G. conirostris propinqua, the occasional occurrence of 

 grooves on the sides of the upper mandible in each, and the exceptional 

 presence of a chestnut color on the under tail coverts of the latter spe- 

 cies — a marked characteristic of the former — might be taken as evi- 

 dence of a derivation of G. conirostris direct from G. septentrionalis. 

 But since the bills of some specimens of G. conirostris propinqua 

 can almost be duplicated by bills of G. scandens rothschildi, which 

 stands at the top of the G. scandens series, and since the dark color of 

 the adult females and young in G. conirostris is simply the maximum 

 of the tendency shown by the whole G. scandens series, we think it 

 most logical to regard G. co7iirostris as following naturally G. scan- 



