

6 



operculum of the typical Drepanididse from Fringillid conditions is absolutely 

 impossible. 

 IV. All the birds in question are related to each other as one group, Drepanididse ; 

 but out of them have sprung the Fringillidse. Morphologically this assumption 

 is possible from a general point of view, but it does not at all work satisfactorily 

 in detail. First, we cannot assume that a family like the Fringillidaj has 

 spread from a volcanic group of oceanic islands all over the world, with the 

 very exception of the bulk of the Pacific islands and of the Australian region. 

 Secondly, Pseudonestor, ChLoridojps, Loxioides, and Bhodacanthis have by their 

 bills, reached a stage more exaggerated than that of most of the typical thick- 

 billed true Fringillidse, while they differ from the geographically nearest rather 

 thick-billed Clitoris in the pattern of colour, general moulding of the beak, and, 

 above all, in the configuration of the narial region. It is equally futile, as said 

 before, to connect them with Coccothraustina. 



Lastly, the remarks concerning Chrysomitridops and Loxojps militate against this 

 assumption (No. IV.). 



Consequently there remains only assumption No. I., namely, that the thick-billed 

 birds of the Sandwich Islands are modified Drepanidid-se, which by convergence, by 

 adaptation to similar habits, have developed features which we are accustomed to 

 associate with typical Fringillidse. 



Although we have now disposed of the Fringillidse, the whole question of the 

 affinities of the thick-billed genera cannot be considered as threshed out until we have 

 substituted the Tanagridse for Fringillidse in the four hypotheses examined above. 



Terminal forms of the Tanagridse are Orchesticus, Saltator, Tcmagra. Comparison 

 between them and the Sandwich Island genera is rather favourable. The resemblance 

 in the whole narial configuration and the build of the bill is striking. The stout-billed 

 Tanagridse have, in fact, modified their bills and nostrils in exactly such a way as, 

 anatomically speaking, we expect these parts to be modified when starting from a 

 condition like that which is represented by Coerebidse and the slender and short-billed 

 Drepanididse. This is one of the very points which excludes any direct relationship 

 of the thick-billed islanders with Fringillidse. 



In the general build and in the pattern of colour Loxio'ides and Bhodacanthis 

 resemble certain Tanagridse — say, for instance, Saltator, Orchesticus, and Pitylus — much 

 more than they do any of the Fringillidse. The Tanagridse seem to be devoid of a 

 pronounced crop, they possess only a slight dilatation ; but the same applies also to 

 Embernagra among the Fringillidse, and thus this once cherished character is weakened. 

 On the other hand, the Tanagridse differ strongly from the Sandwich Island birds in 

 the structure of the palate (especially their broad maxillo-palatine processes and the 

 partly separated palatines) and the tongue, which is fleshy, rather bifid, and ends in 

 two cone-shaped horny caps. 



However, the Tanagridse are in all probability related to the Coerebidse, the latter 



