IN THE FEATHERS OF BIRDS. 225 



by the leaves and branches of the succeeding feather. This 

 bristle is only found in the head- and gorget-feathers. The body- 

 feathers, where coloured, end as shown in text-fig. 2 c. 



From these arrangements it happens that the best display of 

 colour occurs when the body of the bird is in the nearly vertical 

 position it assumes when hovering. 



In the Sun-Bird each leaf has a curious rasp-like shape, and 

 the upper coloui'-producing surface is in the form of 3 or 4 

 convex plates, in each of which the virtual image of the sun or 

 other source of light appears as a coloured spot. 



The chief peculiarities of the metallic feathers of Birds of 

 Paradise are that the leaves are only developed on one side of the 

 branch, and are so disposed with reference to it as to be parallel to 

 the stem (text-fig. 3 b). The upper surface consists of more or less 

 rounded lobes, and the cross-section is shown in text-fig. 3 c. This, 

 so far as I have observed, is the rule in all the metallic feathers, 

 whether on the head, wings, or tail. 



In the Rifle-Bird, which is a near connection of the Birds of 

 Paradise, the leaves are only developed very slightly on that side 

 of the branch on which, in the true Birds of Paradise, they 

 are absent. 



In the Peacock and all other pheasant-like birds, the leaves 

 are of the form shown in text-fig. 4 a, b. The cross-section of the 

 leaves is comma-shaped and the whole structure is transversely 

 corrugated. The colour layer lies on the upper surface of the 

 "dot "of the comma, which thus presents a series of rounded 

 knobs to the light, each giving rise to a spot of colour. 



In the head-feathers of Ducks the colour is developed on the 

 leaves on botli sides of the bra.nch, but in the speculum, on one 

 side only, the uncoloured leaves lying below the coloured part of 

 the adjacent branch and serving to lock the two in position. 



As before stated, all these feather colours disappear when the 

 structure is compressed. For this test I place the feather 

 between a quartz plate and a plano-convex lens of the same 

 material, of a foot radius, these being mounted to fit on the stage 

 of a microscope. The only difficulty in applying this test is 

 to separate a suitable part of the coloured material on which to 

 operate. The result is a conclusive proof that the colours are 

 not due to any form of pigment, a,nd strong evidence that they are 

 due to interference. Most feathers are extremely impermeable 

 to fluids, but in certain cases (the Peacock for instance), when 

 immersion takes place, the colour changes at once to one of a 

 longer wave-length : blue becomes green, green yellow, and so 

 on. Where this happens it is evidence either that the feather is 

 to a certain extent permeable by the fluid used, or, more probably, 

 that the colour-production depends on some quality or grain of 

 the outside surface. 



The greater number of the metallic feathers which I have 

 experimented with show no change on immersion in any ordinary 



