THE COLORATION OF REEF FISHES 545 



between the two phases experimentally induced being probably 

 greater than any consciously observed upon the reef up to that 

 time. 



Following the clue afforded by the experiment it gradually 

 appeared that where the character of the bottom is uniform over 

 large areas, there is close agreement in color between individuals 

 of certain changeable species, or, in other words, that definite 

 color phases are correlated with simple environments of different 

 kinds. With species that will gather about food provided for 

 them it soon became clear that particular phases may be dem- 

 onstrated at will by leading the fishes from one place to another 

 appropriately chosen. The color phases of others may eventually 

 be seen succeeding one another in the same way, if one under- 

 takes the patient observation of uncontrolled specimens passing 

 from one type of bottom to another. 



The first method has been applied to five species, and it appears 

 that their color changes are definite, and that they involve the 

 repetition upon the fish of the dominant color of the environ- 

 ment, either uniformly, except for counter,shading, or in dis- 

 tinct spots, stripes or bands. The second method has been 

 followed with four species whose color changes may be predicted 

 with accuracy, and are marked by similar convergence toward 

 the color of the background that induces them. 



Finally, with fishes of small or medium size it is possible to 

 test for and to demonstrate adaptive color changes under labora- 

 tory conditions by using combinations of colors occurring nor- 

 mally in the environment of the animals experimented upon. 

 For example, the bottom of a tank 28 inches long and 13 inches 

 wide and deep, inside measurement, was covered to a depth of 

 several inches with coarse sand and fragments of corals and 

 shells from the beach. In the sand at one end over half the bot- 

 tom turtle grass (Thalassia testudinum) was planted until a 

 dense green covering had been secured. The other end was left 

 bare. Fishes that had been seined upon reef flats covered more 

 or less thickly with the same kind of plants, with bare spaces 

 appearing at intervals among them, were then placed in the 

 tank, and it was discovered that a number changed regularly 



