EVOLUTION OF THE COLORS OF BIRDS. 99 



color only in so far as it is seen as such, being, from a 

 mechanical point of view, nothing but molecular vibra- 

 tions. In spite of this fact, some colors appear to have 

 been brought about from some relation existing between 

 the being possessing them and some other percipient 

 being or beings, while other colors are purely adventi- 

 tious. It is indeed difficult in many cases to distinguish 

 between the two kinds, although there are some instances 

 in which opinion is practically unanimous. For exam- 

 ple, no one can deny that certain animals are protect- 

 ively colored; but in asserting that an animal is pro- 

 tectively colored, the implication is necessarily made 

 that there are or have been other animals in the same 

 environment which would see it if it were colored other- 

 wise. Accordingly, in these protectively colored species, 

 the colors are not adventitious but have been produced 

 with the view of not being seen. Warning colors and 

 recognition marks, on the other hand, have been pro- 

 duced for the purpose of being seen. If, then, these 

 three great classes of colors in animals — protective, 

 warning, and recognition — have been produced in order 

 that the animal possessing them may be seen or not 

 seen, as the case may be, is it logical to assume that 

 the colors which appeal to us as beautiful, although it is 

 impossible, in the majority of cases, to draw any sharp 

 line between them and such as may be more or less of 

 use for protection or recognition — is it logical to assume 

 that these beautiful colors have been produced without 

 any reference to some other individual perceiving them? 

 Why, in this one instance of the colors which above all 

 others attract the human eye, and, apparently, the eye 

 of all the higher animals, should we assume that they 

 have been produced without any reference to their fellow 

 beings which are continually seeing them? 



Having stated at some length the leading arguments 



