1882.] DR. H. GADOW ON THE COLOUR OF FEATHERS. 417 



position A ; if we turn the bird to position B it will, let us say, look 

 green ; and halfway between B and C this bird will assume a blue 

 colour, which again passes into violet before appearing black again 

 in })Osition C. If the bird begins with bronzy red, it will change 

 through golden green to pure green, then through bluish green to 

 blue and violet. There is not a single feather which, if moved from 

 B to A, changes from green to blue, in other words from the violet 

 to the red end of the spectrum. Thus we are able to predict into 

 which colour a bird can change if we know its colour in position B. 

 Thus a blue feather can only become purple or violet ; a green one 

 has more changes, and a golden-green one still more. It is, 

 however, very rare that a feather changes through more than half 

 of the spectrum ; a coppery-red feather will generally cease with 

 green ; a violet feather cannot change at all, except into black, since 

 beyond the violet there are no visible colours in the spectrum. 



Another important fact is that metallic feathers can appear in any 

 colour which is represented in the spectrum, but not in any which, 

 like brown or grey, are not spectral colours. All these circumstances 

 induce me to explain the changeable metallic colours as prismatic ; 

 and in order to prove this we have to examine the feathers for their 

 prismatic structure. In any metallic feather the metallic colour is 

 confined to the radii which are entirely devoid of cilise, and cousist 

 of a series of variously shaped compartments which overlap one 

 another like tiles of a roof. 



Fig. 1, Plate XXVIII., represents a barbule of a violet feather of 

 ^thopyga ; it consists of about fifteen compartments, each 0"03 

 mm. long and 0"03 mm. broad, and each forming a plane. Fig. 9 a, 

 Plate XXVII., shows part of a barbule oi Nectarinia famosa, brilliant 

 green in position B, blue in C, black in A; each compartment was 

 0*027 mm. long and 0"015 mm. broad. If turned on the edge and 

 looked upon sideways, they look like fig. 9 b ; each compartment is 

 convex-concave, with the convex side lying uppermost turned towards 

 the light. In Jacamar, coppery red in positions B and A, green in C, 

 they measure 0'040 by 0"018 mm. Every one of these compart- 

 ments is surrounded by a transparent horny coat, the thickness of 

 which varies from O'OOOBd mm. {Sturnus) and 0*00 15 mm. (Jaca- 

 mar), 0-0012 {JEthopyga) to 0*0022 mm. ( Galbula tomhacea). The 

 surface of this coat is either perfectly smooth and polished, like in 

 Nectarinia famosa, or, in violet feathers, contains very fine longitu- 

 dinal ridges, or, as in Jacamar, it showed very fine and numerous 

 little dots. Below this transparent and apparently colourless sheath 

 lies brownish to black pigment evenly dispersed. 



Now, then, let us take a transverse section of such an arrangement 

 (S S being the transparent sheath, P the pigment), and put it in a po- 

 sition that the light falls upon it from L (fig. 2, p. 418). A very small 

 part of the orbit of a circle or any curve may be treated as a straight 

 line ; we thus may regard this sheath as consisting of a number of 

 small prisms. We know that a prism in sucb a position, with the 

 top or one edge directed upwards, breaks the light in such a way as 

 to produce a spectrum on the side furthest from the light, with the 



