264 Papers from the Marine Biological Laboratory at Tortugas. 



(b) The theory of warning color assumes that the enemies of warningly- 

 colored species are capable of discriminating colors. 



(c) The theory assumes that the enemies of warningly-colored species 

 are able to form associations between the conspicuous colors and patterns 

 of their prey and their disagreeable qualities. 



(d) The theory assumes that the associations thus formed between con- 

 spicuousness and disagreeable qualities are indefinitely retained (associative 

 memory). 



It seemed possible to test these four assumptions by experiments on the 

 predaceous fish that occur about the coral reefs or along the nearby shore. 

 Do certain colors in themselves convey a warning so that prey showing 

 these colors is avoided ? Have the predaceous fish color vision ? Do they 

 form associations ? Have they associative memory ? Experimental evidence 

 on these points is presented in sections IV, V, and VI of this paper. 



Should the predaceous fish be found to have the qualities assumed by 

 the theory, the warning color explanation of the conspicuousness of coral- 

 reef fishes becomes more probable. But it would still need to be learned 

 by feeding e.xperiments whether any of the species of conspicuous fish are 

 avoided, whether any show that combination of conspicuousness with quali- 

 ties sufficiently disagreeable to render them relatively free from the attacks 

 of predaceous fish under the conditions that normally prevail. Experiments 

 involving the feeding of conspicuous coral-reef fish to predaceous forms are 

 described in section VII. 



Through the courtesy of Dr. A. G. Alayer, Director of the Marine Bio- 

 logical Station of the Carnegie Institution of Washington, I have been 

 able to spend parts of the months of June and July at the station during 

 the seasons of 1905 and 1907, while acting as assistant of the U. S. 

 Bureau of Fisheries. I am indebted to the U. S. Fish Commissioner, the 

 Hon. George M. Bowers, for permission to publish the results of my work. 



III. GENERAL METHOD EMPLOYED. 



The gray snapper, Lutiaims griscus (Linnsus), was chosen as the sub- 

 ject of the experiments. This, the commonest predaceous fish of the region. 

 averages 12 to 15 inches in length. At the Tortugas it is found about the 

 inner reefs, but occurs also along the shore of Loggerhead Key wherever 

 there is shelter. Nearly all of the individuals along the shore were, during 

 my stay, aggregated in three colonies. The largest of these colonies was 

 under the Laboratory dock and consisted in the summer of 1907 of from 150 

 to 175 individuals, but was smaller in 1905. The two other colonies were 

 found under the two docks belonging to the lighthouse, one on the east side 

 of the island and the other on the west. Each was about a quarter of 

 a mile distant from the Laboratory colony. The personnel of each of these 

 colonies seemed to be fairly constant. During the day individuals rarely 



