FROM THE SIDES OF THE NECK OF THE DOMESTIC PIGEON. 269 



cavities are closely packed with spherical granules of melanin pigment which I frequently 

 find lying loose outside of the barbule cell cavities in cross-sections of the feather germ. 

 There is no indication of fusion, and the granules are of pretty uniform size. Though 

 I have examined during the past three years many sections of fully cornified bar- 

 bules containing the ordinary rod-shaped melanin granules, I have never found the 

 granules lying loose and detached from the barbule in this way. They are usually 

 grouped together in irregular masses and if not actually fused, they are so closely 

 packed as to make it impossible in many cases to distinguish individual granules. 

 Furthermore, they seem to be actually imbedded in the horny mass of the barbules 

 which possesses no conspicuous cell cavities. In these barbules giving metallic colors, 

 however, the spherical granules do not appear to be imbedded at all, but are simply 

 packed in a chamber once occupied by the cytoplasm of the cell just as balls are packed 

 in a box. 



In the ontogeny of these barbules, they receive rod-shaped granules of melanin 

 from typical pigment-cells or chromatophores in the manner described by myself 

 (Strong, :.02 a ) and others. During the differentiation of the barbule cells, however, 

 these granules are metamorphosed from the rod-like shape to a spherical form. In 

 a single cross-section of the feather germ, one sees sections of both metallic and non- 

 metallic barbules. Sections of non-metallic barbules occur in the dorsal half of the 

 feather-germ section, and sections of metallic-colored barbules in the ventral half. 

 The rod-like shape is retained permanently in the former, whereas in the latter the 

 spherical form is assumed. The barbule walls approach each other in the ventral 

 half of the barbule until they are separated by an interval less than the diameter 

 of a pigment sphere (Figs. 11, 12). This explains the very slight pigmentation of 

 the ventral halves of the barbules. 



I have not been able to find stria>, ridges, pits, or knobs sufficiently numerous 

 on the surface of the outer transparent layer to account for the observed colors by 

 diffraction. There are sometimes suggestions of irregular strise which I believe, how- 

 ever, are usually optical effects only. Moreover, if these appearances are actual striae, 

 they do not occur frequently enough to produce the strong metallic colors of these 

 feathers. A study of many cross-sections reveals no observable irregularities in 

 the surface of the barbules, though views of the entire barbule show slight elevations 

 here and there. 



3. The Nature of the Color-producing Structures. The complex details in struc- 

 ture that we have just noticed make interpretations of physical relations difficult. 

 Instead of a simple prism of glass, a soap-bubble film, or an artificial diffraction 

 grating with its precise regularity in detail, we have here many reflecting sur- 



