86 
ON THE CLASSIFICATION OF BIBDS. 
length of their secondary and tertial quills. In another 
fissirostral genus, Collars, which includes the short- 
billed rollers, there is an evident falling off of the typi- 
cal perfection ; some of the primaries anil secondaries 
are indeed notched, yet the first quill feather is shorter 
than the second and third ; but in the genuine rollers 
tliese notches disappear, the primaries are much shorter, 
and the second, third, and forth quills are nearly of 
equal length. We have not been an eye-witness to the 
manner of flight of either of the two last-named genera, 
nor does any author appear to have noticed them. 
There can be no question, however, that the rollers not 
only are fissirostral birds, that is to say, taking their 
food on the wing, but that they immediately and natu- 
rally follow the short-biUed rollers ; the structure of 
their mouth and feet completely disproving the assertion 
of some authors, that they hve, like magpies, by search- 
ing for insects upon the ground. It will be unnecessary 
to notice the different representations of the fissirostral 
type of wing, seeing that it is generally combined in 
such types with other characters, too intricate to be 
well understood by the student. 
(790 FALCATED, or sword-shaped form of 
wing, is only possessed by the numerous family of hum- 
ming-birds (Jfff. 43.). It differs from that last described 
by the two or three first primaries being curved towards 
their ends, which are rather suddenly rounded off, while 
all the others which succeed them are pointed. The 
primaries are certainly developed even more than in the 
swallows ; a circumstance which is decidedly favourable 
to the idea that these bird are the swiftest. The whole 
strength of the wing, it may almost be said, is thrown 
into these feathers, for the shortest, which is of course 
