amined under the microscope. The change in color with chang¬ 
ing angle of incidence is that corresponding to interference colors, 
and is much greater than with substances like magenta, showing 
selective reflection. In fact, the change of color with changing 
angle of incidence is practically negligible with magenta and the 
other pigments of this type. The natural conclusion is, therefore, 
that th-fe iridescent colors are not due to selective reflection. 
The colors of iridescent feathers belong to Newton’s series of 
interference colors, being caused by a laminated or plate-like 
structure of the barbules, where the color originates. These 
barbules present broad, flat surfaces forming an approximately 
plane surface which exhibits the colors with great brilliancy. 
The cross section of the barbule shows a central pith or core 
about two microns thick, which appears to be of somewhat 
fibrous or granular character. Enveloping this core are three 
thin layers or laminae, of equal and uniform thickness, and 
seemingly in contact. The color originates in these three layers, 
each about OA/j. thick, which make the barbule act as a multi¬ 
ple thin film. The laminated outer layer of the barbules is 
particularly characteristic, and was not found on the nonmetallic 
feathers of the same bird where these were available for examina¬ 
tion. In all cases a definite structure is associated with metallic 
luster. The barbules are always broad and flattened, distinctly 
segmented, with blunt ends, and generally contain much more 
dark pigment than adjacent noniridescent barbules of the same 
feather. Altum 4 says that “the metallic luster is always asso¬ 
ciated with the peculiar, hard, and smooth texture. A blind man 
could easily pick out all the birds with metallic luster in a collec¬ 
tion of any size.” The relation between structure and metallic 
luster is so close that Mr. Mason was able to pick out from the 
collection of drawings of barbules in Chandler’s paper 5 those 
which were of the metallic type. Comparison with the descrip¬ 
tion of the coloring of the feather showed very few errors, even 
though the drawings cover almost all species and include many 
feathers not possessing any noniridescent barbules for comparison. 
The thickness of these films is about the value calculated for 
the interference colors of the order observed in feathers, though 
the whole structure is too small for very accurate measurements. 
In all the iridescent feathers examined, the colors appear to lie 
in the upper second or third orders of the series, which include the 
most brilliant of the interference colors—yellow, and red in the 
upper second order, and dull purple, blue, green, yellow, and 
red in the third order. Since we have no good method of de¬ 
termining exactly the order of interference and isolated color re¬ 
flected from a thin film, this is based on matching the colors of 
feathers with those of artificial films, the colors of which are of 
known order. 
With increasing angle of incidence the color usually changes 
through two or three of the colors of the above series. The 
change is frequently from the blue to the red end of the spectrum, 
but this is not necessarily the case. The pigeon, for instance, has 
4 J. Ornilholcgie, 2 (1854), 19. 
6 University of California Publications, Zoology, 13, No. 11 (1914-16). 
( 2 ) 
