Statement of the Problem. 263 



The gaudily-painted fish occur on the inner reefs of the Tortugas, but 

 the " brilHant colors of the rainbow," " the flowers and fruit '" I am unable 

 to see in their environment. The brilliant fish are therefore conspicuous 

 against a dull and nearly uniform background. I know of nothing in the 

 environment which they resemble even remotel}'. 



Striking sexual differences have not been noted in these fishes by syste- 

 matists, and I have observed none. I am unable to tell the sex by any ex- 

 ternal character. Sexual selection seems therefore to be excluded as a possi- 

 ble explanation, although it is to be expected that when we know the 

 mating habits of the fish we shall find that they include displays of color. 



The theory of warning color often applied to certain of these fishes, most 

 recently by Bristol (1903), seems then to aflford the most plausible explana- 

 tion of their conspicuousness. This theory attempts to account for con- 

 spicuousness, more particularly in insects, by its association with some dis- 

 agreeable quality. The vertebrate enemies of the conspicuously colored 

 animal are believed to be warned of its disagreeableness by its color. They 

 thus learn to avoid it. In this way, in each generation, the most conspicu- 

 ous are preserved and through this selection conspicuous coloration is be- 

 lieved to have been perfected. The theory thus attaches to warning colora- 

 tion a biological significance a present function while at the same time 

 it affords an historical explanation of the coloration, by asserting that it has 

 been perfected through past selection of the functionally best adapted. 



The present paper embodies the results of a search for the biological 

 function of the brilliant colors and striking color-patterns of conspicuous 

 coral-reef fishes. If it appear that a warning function does not at present 

 attach to this conspicuousness, then it becomes extremely improbable that it 

 could have arisen in connection with such a function. 



The theory of warning colors embodies certain fundamental assvmip- 

 tions which it seemed possible, in the particular case in hand, to test experi- 

 mentally. 



(al As stated by Beddard (1892, p. 155), "The theory of warning 

 color implies not a special recollection of any type of insect, but a general 

 association of bright colors with poisonous or dangerous qualities." A like 

 idea is expressed by Poulton (1887). After pointing out the few colors 

 and limited number of patterns among warningly-colored insects, he says : 

 "It is to be noted that advantage would accrue in the greater thorough- 

 ness of the education, no less than by shortening the process, for a few 

 colors with a few simple patterns would be remembered more easily than 

 a larger number with a separate pattern in nearly every species." Again, 

 following Meldola (1882), he says: "All the conspicuous and dangerous 

 or distasteful species in any country will be found to share between them 

 a few strongly contrasted colors, arranged in few and simple patterns again 

 and again repeated." 



