THE COLORATION OF REEF FISHES 559 



that through the operation of natural selection, in view of their 

 undemonstrated possession of disagreeable characters, there has 

 been developed in five species of one habit, and ^vithout any 

 direct relation to it, a single type of warning coloration which 

 has not reappeared once among many warningly colored species 

 of other habits. That this hypothetically conspicuous color 

 should be that one of all possible shades which first loses its 

 distinctive quality in dim light, in which the animals that bear 

 it habitually live, is an additional difficulty. 



The five species under discussion represent three families of two 

 ■widely separated groups, and even the two families of the same 

 group are not closely related (Jordan and Evermann, '96, p. 

 1237). Their common color does not, therefore, seem to be due 

 to common descent, unless it is held that they resemble a very 

 ancient ancestral type, from which all related families have di- 

 verged. Even so the question remains, why types of one habit 

 should have been stable, or persistent, while others underwent 

 modification. There are apparently only two possible solutions 

 to the problem: either the environment through some direct 

 action controls the pigmentation of these creatures, or red, under 

 the conditions in which they live, possesses selective value 

 hitherto unrecognized. In any case the correlation of red color 

 with a definite habit renders it highly desirable to have detailed 

 information regarding the habits of fishes in general in order to 

 determine whether this is simply a sporadic instance, or a strik- 

 ing example of conformity to a general law. 



The knowledge desired is the answer to the question, where and 

 how the species investigated normally spend their time. One 

 must narrow the inquiry, however, so effort has been concen- 

 trated upon the determination of three sets of facts. These 

 seem to be of fundamental importance and may be ascertained 

 with comparatively slight difficulty. When the investigation is 

 completed, one should be able to say when the different species 

 normally feed, where they are usually found by day, and at what 

 level they habitually swim. 



Nocturnal feeders are little given to diurnal wandering, differ- 

 ing decidedly in this respect from those that feed during daylight. 



