

representing the lower, older stage, but having specialized in bill and tongue. In my 

 former paper I pointed to these same Ccerebidse as the possible ancestral relations 

 of the typical Drepanididae. It is, I think, significant that I should now, in a round- 

 about way, again be led towards this same family of birds. It was my fault that I did 

 not take the Tanagridse into our confidence ; it would have been a small step only 

 from the Ccerebidse, but I was biased by the watchword "Finches" or not Finches. 



Let us now sum up. Concerning the Tanagridse, they are neither the descendants 

 nor are they the direct ancestors of the thick-billed Sandwich Island birds, but they 

 come very near them. The origin of the Drepanididae we do not know ; in my former 

 paper I hinted at the Ccerebidae. I still fail to see any valid reasons against such a 

 descent : on the contrary, it seems now a little more probable. 



The Sandwich Islands have received their characteristic bird population from the 

 south-east, as an offshoot of the Columbian fauna (Drepanididae sensu latiore), and 

 from the south-west — Australian Meliphagida?, namely, Acrulocercus and Chcdoptila, 

 whose near relation, Leptomis, lives in the Fiji Islands; the Flycatcher [Chasiempis 

 sandvicensis) has its nearest relation, Ch. dimidiata, in Rarotonga, and the Thrush-like 

 Phasornis points also towards the south-west. 



