150 
SEXUAL SELECTION: BIRDS. 
Pakt II. 
smaller and less perfect than on the other feathers, 
with the upper parts of the external black rings de- 
ficient, as in the case just mentioned. The imperfec- 
tion here seems to be connected with the fact that the 
spots on this feather shew less tendency than usual to 
become confluent into stripes ; on the contrary, they are 
often broken up into smaller spots, so that two or three 
rows run down to each ocellus. 
We have now seen that a perfect series can be fol- 
lowed, from two almost simple spots, at first quite 
distinct from each other, to one of the wonderful ball- 
and-socket ornaments. Mr. Gould, who kindly gave me 
some of these feathers, fully agrees with me in the com- 
pleteness of the gradation. It is obvious that the stages 
in development exhibited by the feathers on the same 
bird do not at all necessarily shesv us the steps which 
have been passed through by the extinct progenitors of 
the species ; but they probably give us the clue to the 
actual steps, and they at least prove to demonstra- 
tion that a gradation is possible. Bearing in mind 
how carefully the male Argus pheasant displays his 
plumes before the female, as well as the many facts 
rendering it probable that female birds prefer the more 
attractive males, no one who admits the agency of 
sexual selection, will deny that a simple dark spot with 
some fulvous shading might be converted, through 
the approximation and modification of the adjoining- 
spots, together with some slight increase of colour, 
into one of the so-called elliptic ornaments. These 
latter ornaments have been shewn to many persons, 
and all have admitted that they are extremely pretty, 
some thinking them even more beautiful than the 
ball-and-socket ocelli. As the secondary plumes be- 
came lengthened through sexual selection, and as 
the elliptic ornaments increased in diameter, their 
