272 THE METALLIC COLORS OF FEATHERS 



As has been stated before in this paper, these measurements are only approx- 

 imate estimates. The complex individual variations in the color-producing struc- 

 tures make attempts at greater precision of no value in a paper of this nature. 



We may sum up these observations as follows: (1) when the sum of the angles 

 of incidence and reflection is less than about 90 and the incident rays make an angle 

 of at least 48 with the plane of the feather, a brilliant green is obtained; (2) when 

 the sum of these angles is greater than 100 and less than 140, purple effects are 

 obtained; (3) when the sum of the angles is greater than 140, no metallic colors 

 appear. 



The refraction-prism hypothesis of Gadow ('82) requires that the angle or series 

 of angles just described be more than 180, a physical impossibility here, for the fol- 

 lowing reasons: (1) either the incident or the reflected ("refracted") rays would 

 be intercepted by the contiguous barbules or by the barb; (2) metallic colors do 

 not appear unless this angle is more than 140. "Refracted" rays that could pos- 

 sibly reach the eye would require a source for the incident rays ventral to the feather, 

 whereas the illumination is necessarily dorsal. 



5. The Nature of the Metallic Colors. Even if the surface of the barbule were 

 covered with ridges or striae, it is inconceivable that they could produce these metallic- 

 color phenomena. Reflection gratings give nothing comparable to these effects; they 

 require special conditions of illumination with parallel rays of light passed through a 

 narrow slit, whereas the feather is iridescent in diffused daylight. Having satisfied 

 myself that the metallic colors under consideration are not reflecting-grating effects 

 produced by stride, ridges, knobs, or pits on the barbule surface, I was led to favor the 

 thin-plate hypothesis of Altum ('54, '54 a ) and Briicke ('61). The outer transparent 

 layer (Figs. 11, 12, tu.) is thin enough to produce interference colors, being not more 

 than three half green-wave lengths in thickness. Furthermore, it is very uniform 

 in thickness and has a comparatively smooth surface, as I have pointed out before 

 in this paper. The constant occurrence of pigment granules of a spherical shape 

 only in the barbules giving metallic colors, however, seemed too significant to war- 

 rant accepting the thin-plate hypothesis without qualification. 



At first I had regarded the function of the pigment simply that of an absorbing 

 background. I mounted a feather on a pad of paper and placed the mount on the 

 stage of a microscope. By means of a mirror, strong sunlight was reflected from 

 the side on the feather, and portions of single barbules were examined with a Leitz 

 No. 7 dry objective. The view obtained was exceedingly suggestive. There was 

 no uniform glow of diffracted light such as is ordinarily obtained from thin plates. 

 On the contrary, there appeared a large number of small glowing dots in a black field. 



