494 



THE DESCENT OF MAN, 



these ornaments should have been formed throngli the selec- 

 tion of many successive variations, not one of which was 

 originally intended to produce the ball-and-socket effect, 

 seems as incredible Jis that one of RaplieaFs Madonnas 

 should have been formed by the selection of chance daubs 



of paint made by a long 

 ABC succession of young art- 

 ists, not one of whom 

 intended at first to draw 

 the human figure. In 

 order to discover liow 

 the ocelli have been 

 developed we cannot 

 look to a long line of 

 progenitors nor to many 

 closely allied forms, for 

 such do not now exist. 

 But fortunately the 

 several feathers on the 

 wing suffice to give us 

 a clew to the problem, 

 and they prove to dem- 

 onstration that a gra- 

 dation is at lojist pos- 

 sible from a mere spot 

 to a finished ball-and- 

 socket ocellus. 



The wing - feathers, 

 bearing the ocelli, are 

 covered with dark 

 stripes (fig. 57) or 

 with rows of dark six)ts 

 (fig. 59), each stripe or 

 ' of ^ 



the 

 outer side of the shaft 



to'the left of the shaft, has been cut oSr * ^ ^O OUC of the OCclH. 



The spots are gen- 

 erally elongated in a line transverse to the row in which 

 they stand. They often become confluent either in the line 

 of the row and then they form a longitudinal stripe or 

 transversely, that is, with the spots in the adjoining rows, 

 and then they form transverse stripes. A spot sometimes 



Flff. 57. Part o{ secondary winpr-feather of . ^e K\\f\i& 



Arjfus pheasant, showing two perfect ocelli, ^'^'" '^'^ ^ptno 



a and 6. A, B, C, D, etc., are dark stripes obllQUelv down 

 running obliquely down, each to an ocellus. * - - 



