BTJCERID*. COnVIN^. 
95 
internal structure of this member been fully ascertained. 
Lastly, the feet are generally so very short, as to appear 
calculated only for perching. United to the scansorial 
birds by moans of the toucans, they would seem to re- 
present the Ranorea, but the structure of their feet, more 
imperfect than any of the familes in this order, forbids 
the supposition. This opinion we had long entertained 
from theory, but it has recently been confirmed by a 
singular fact in their economy, communicated by an 
officer long resident in India. It seems that all the 
species of Rucbtos he has met with in a live state, are 
constantly in the habit of throwing their food up in the 
air, and catching it before it is swallowed ! It is im- 
possible to imagine a more beautiful insipient develope- 
ment of the fissirostral economy than is manifested by 
this propensity — a propensity which, divested of its 
analogical relations, would be perfectly incomprehen- 
sible, seeing that the birds are not obliged (like swallows, 
and all the true Fmiroatres) to catch their food in the 
air in the first instance. 
(109.) The CoBvinJS, or crows, appear to he that 
family nearest alhed to the last, although the intervening 
forms are few. The genus Frigilus, in fact, is the 
only representative we at present know of that subfamily 
which intervenes between the Buceridte and the Corvirue. 
The whole family, in short, has never yet been analysed ; 
so that the leading divisions alone can yet he made out 
or stated with any degree of certainty. The little value 
tliat can he attached to speculations on the rank of the 
present genera, founded upon mere sgntheais, will best 
appear by looking to those artificial arrangements that 
place the short-legged rollers close to the long-legged 
and powerfully constructed grakle {Gracula religioaa), 
— two genera, moreover, which analysis has convinced 
us do not belong to this family. Nothing, in short, is 
more easy than to divide a group like this into three, 
five, seven, or any other given number ; but the divi- 
sions must always be considered as temporary, nntil 
confirmed by analysis. We have not yet carried our 
investigations so far as to lay before the reader an 
