2 7° Papers froin the Marine Biological Laboratory at Tortngas. 



a trial ; each experiment consists of a number of successive trials, often 

 divided into several series. 



In the preliminary trials the alherinas w ere taken so rapidly that it was im- 

 possible to keep an accurate record. It became further evident that normally 

 hungry snappers might distinguish food of different colors, but might yet, on 

 account of their hunger, fail to discriminate between one color and another. 

 To reduce the rate of feeding to such a point that an accurate record could 

 be kept and to make evident a possible power of color discrimination, it 

 became necessary to prolong the preliminary feeding until the appetite of 

 the snappers had been sufficiently dulled. The number of fresh atherinas 

 taken by a coIon\ of about lOO snappers was usually about loo during the 

 first minute. The preliminary feeding was continued until this rate had 

 been reduced to 15 or 20 per minute. The discrimination e.xperiments fol- 

 lowed immediately thereafter. When, as in 1907, the atherinas used in the 

 experiments had been preserved in formalin they were taken more slowly ; 

 the preliminary feeding was then not so long coiUinued and the discrimi- 

 nation experiments began as soon as the rate of feeding of the snappers was 

 slow enough to permit an accurate record to be made. 



The atherinas used in each discrimination experiment were of the same 

 average size. Those of the two colors were given a like taste by treatment 

 with acetic acid or formalin as described below. Thus errors due to size 

 or chemical properties are believed to have been eliminated. 



In each trial the 10 fish were taken together in the hand and thrown 

 by a single movement of the arm so that they fell upon the water spread 

 over an area of i or 2 square yards and intermingled at random. They 

 were seized by the snappers from beneath. The random intermingling of 

 the two colors in each trial, unlike in successive trials, is believed to have 

 eliminated errors of position in the horizontal plane. In the experiments 

 of 1907 all atherinas used were made to float (except in experiment 26, 

 q. v.), so that all lay on the water in one plane. Fish which sink are 

 brought nearer to the snappers and are taken more readily than those which 

 float. Thus if the atherinas of one color should sink more readily than 

 those of the other color there would be introduced an error of vertical posi- 

 tion. This is avoided by making all the atherinas float, so that all are 

 equally accessible to the snappers. 



COLOR DISCRIMINATION EXPERIMENTS (19O7). 



While Atherina was abundant at the Tortugas in June of each year of 

 my stay, it became scarce in July. On my arrival on July 8, 1907, the fish 

 were already so scarce that it was impossible to obtain on any one day the 

 large number needed to conduct discrimination experiments with fresh fish. 

 I had. therefore, to accumulate a supply of atherinas by preserving in forma- 

 lin those obtained each day. The discrimination experiments (except experi- 



