' 



Although it is, as a rule, not an agreeable task to acknowledge one's self in the 

 wrong, I have in this case derived a good deal of pleasure from my renewed and more 

 extensive investigations, since — if our conjoint conclusions are correct — they have 

 revealed one of the most remarkable instances of convergent analogies between what 

 are generally called Fringillidse and some of the Drepanididse. 



Nobody has as yet been able to diagnose any family of the Fringilliformes. Certainly 

 the Drepanididse, after the addition of the thick-billed genera, defy any real diagnosis, 

 except perhaps that they are nine-quilled Oscines, which are confined to the Sandwich 

 Islands. This is perhaps a step in advance. 



I firmly believe that in time to come we shall more and more frequently have to 

 admit geographical distribution as a diagnostic character not only of species and genera 

 but even of larger groups. 



I am inclined to accept the central portion of Dr. Sharpe's scheme of nine-quilled 

 Passeres {cf. Cat. Birds, x. p. 2), but modify it slightly as follows: — 



Ccerebidae Tanagridse. 



Drepanididse. Fringillidse. 



Which, translated into the apparently very exact, but really mystical and still all-in-the- 

 clouds parlance of the phylogenist, means that there was once an undefined stock of 

 generalized Ccerebine and Tanagrine birds, whence have sprung as two independent 

 offshoots the Drepanididse and Fringillidse. The more numerous of these " families " 

 has specialized more in the direction of seed-eaters, while the other very small family 

 underwent the necessity of adapting itself to peculiar insular conditions, and either 

 specialized as insect-eaters (probing the insects out of cracks, not catching them on the 

 wing), or in a roundabout way became as much graminivorous and thick-billed as 

 many of the Fringillidse and Tanagridse. 



In talking of these " families " we are apt to forget, or rather we never appreciate, 

 the solemn fact that, strictly speaking, all the Oscines together are of the rank of one 

 family only ! The greatest differences between the so-called families of Oscines are in 

 reality of very small value ; and when we are discussing, as in the present case, the 

 morphological difference between what should be termed subsections of subfamilies, 

 we have about arrived at the end of our tether, or rather perceptive insight. Of 

 course there are differences between them, larger and more stable than those between, 

 for instance, various species of Paroaria, which are " striking " enough, but we do not 

 know them ! In fact we have to be grateful for small mercies. 



We cannot, as said before, define either Drepanididse or Fringillidse, Coerebidse or 

 Tanagridse — that means to say, we have no single character, nor a combination of 

 features, which apply to all the species of each family. This concerns the pterylosis, 

 namely distribution of feathers, fluffy nature of the feathers of the lower back, relative 

 length and shape of the primaries and of the tail, the pattern of colour ; it applies 



