296 Papers from the Marine Biological Laboratory at Tortugas. 



lay on the bottom. They were not even snapped at or closely examined ; 

 in fact they excited no visible interest. Formalin normals (uncolored) 

 were then taken directly from 2 per cent formalin, rinsed, and thrown while 

 still saturated with formalin. They were taken at once, and about 20 were 

 thus taken. Additional reds and some yellows were then offered, but re- 

 mained practically unnoticed. The stimulus of an immediately preceding 

 feeding on white did not result in the taking of red or yellow. 



The behavior of the snappers showed quite conclusively their avoidance 

 of red, which was even more marked than immediately after the establish- 

 ment of the red-unpalatability association. At that time and also on July 

 27 an occasional red fish was taken, but on August 8 they remained not only 

 untaken, but appeared to excite no interest. 



Although the snappers refused red on August 8 they also refused yel- 

 low, while they took blue less readily than before. The warning mean- 

 ing conveyed thus appears to have been transferred from the red to the 

 yellow, and to a less extent to the blue. Probably atherinas of all colors 

 would have been avoided, and possibly in proportion to the likeness of the 

 respective colors to red. Nevertheless formalin normal atherinas (white) 

 were taken, apparently as readily as ever. 



The facts stated in this section seem to me to warrant the- conclusion that 

 the red-unpalatability association established on July ig zvas still effective 

 on August 8. The associatii'e uienior\ of the snapper has at least this 

 duration. I left the island on my return north on that day. 



Vlll. RESULTS OF FEEDING THE LABORATORY COLONY OF GRAY 

 SNAPPERS ON CONSPICUOUSLY COLORED CORAL-REEF FISH. 



In the preceding sections of this paper evidence is given to show that 

 the gray snapper distinguishes colors, forms with great rapidity associations 

 involving color discrimination, and retains these associations for a consid- 

 erable time. It has been further shown that it is possible, in a short time, 

 to establish in the gra\' snapper a warning-color association, of such a sort 

 that its natural prey is protected from attacks when artificially warningly 

 colored. All this lends support to the theory of warning coloration as 

 applied to coral-reef fishes, yet it remains to be learned by experiment 

 whether any of these fish are actually protected from the attacks of the gray 

 snapper by the assumed combination of conspicuous coloration with un- 

 palatability. To test this assumption as many as possible of the coral-reef 

 fishes were collected and fed to the Laboratory colony of snappers. These 

 fish were usually thrown living from the dock, so that they fell with a splash 

 into the water near the snappers. A few were slipped in quietl}- near the 

 shore, so that, as they swam seaward, their approach to the snappers was 

 more normal. A few of the fish were dead when offered to the snappers, 

 but most of them were very active and made every effort to escape. A few 

 individuals of Adnbefduf marginatus were rendered immobile by pithing. 



