Mr. Gould is well known rarely to admit the existence of varieties, for he esteems very slight differences as specific; now he states[180] that near Bogota certain humming-birds belonging to the genus Cynanthus are divided into two or three races or varieties, which differ from each other in the colouring of the tail,—“some having the whole of the feathers blue, while others have the eight central ones tipped with beautiful green.” It does not appear that intermediate gradations have been observed in this or the following cases. In the males alone of one of the Australian parrakeets “the thighs in some are scarlet, in others grass-green.” In another parrakeet of the same country “some individuals have the band across the wing-coverts bright-yellow, while in others the same part is tinged with red.”[181] In the United States some few of the males of the Scarlet Tanager (_Tanagra rubra_) have “a beautiful transverse band of glowing red on the smaller wing-coverts;”[182] but this variation seems to be somewhat rare, so that its preservation through sexual selection would follow only under unusually favourable circumstances. In Bengal the Honey buzzard (_Pernis cristata_) has either a small rudimental crest on its head, or none at all; so slight a difference however would not have been worth notice, had not this same species possessed in Southern India “a well-marked occipital crest formed of several graduated feathers.”[183] The following case is in some respects more interesting. A pied variety of the raven, with the head, breast, abdomen, and parts of the wings and tail-feathers white, is confined to the Feroe Islands. It is not very rare there, for Graba saw during his visit from eight to ten living specimens. Although the characters of this variety are not quite constant, yet it has been named by several distinguished ornithologists as a distinct species. The fact of the pied birds being pursued and persecuted with much clamour by the other ravens of the island was the chief cause which led Brünnich to conclude that it was specifically distinct; but this is now known to be an error.[184]