Color Discrimination in the Gray Snapper. 



379 



first. There is no reason to believe that those of one color sink more fre- 

 quently than those of the other, so that the sinking introduces no constant 

 error. It does increase the difficulty of making an accurate record, but here 

 again errors doubtless occur as frequently with one color as with the other. 

 The greater discrimination apparent in tables from experiments with formalin 

 fish I believe, then, to be an expression of the reality, rather than of a difter- 

 ence in accuracy of method. Table 8 shows the total number of blue and 

 white atherinas taken by 100 gray snappers in each order, when thrown in 

 lots of 10 5 blue and 5 white during 26 successive trials ; 260 fish thrown ; 

 5 remained untaken ; July 20, 1905. 



Table 8. 



Experiment ^o: Blue and white color discrimination. This experiment 

 was performed four times (July 19, 20, 21, 22). but the results were in- 

 variable, so that only the series of July 20 and 21 need be considered. In 

 each series the trials were preceded by a feeding of fresh atherinas and 

 were continued so long as the snappers took the food readily. The series 

 followed one another from time to time during the day. as the appetite of 

 the snappers permitted. 



The results of the four series of July 20 are plotted in table 9. The 

 data were first arranged in a table of the form of table 8, which combines 

 the first two series of July 20. The number of times that blue and white 

 fish were taken in each order was then calculated from the table in per- 

 centages of the ma.ximum number of times that it might be taken in that 

 order, that is, in percentages of the total number of trials involved in each 

 series of trials of the experiment. The diagram (table 9) shows in percent- 

 ages the frequency with which a colony of 100 gray snappers took blue or 

 white atherinas in each order in 4 series of trials including 460 atherinas. 

 The atherinas were thrown in lots of 10 5 white and 5 blue together. The 

 percentage of fish of each color taken in each order is calculated for each 

 series separately on the basis of the total number of times that fish of each 

 color might have been taken in each order, i. e., on the basis of the number 

 of trials in each series. This basis is in series I, 7 ; in series II, 19; in series 

 III. 5; in series IV, 15. 



The solid line represents the white fish, the broken line the blue fish. 

 The divisions on the ordinate at the left represent percentages ; while in 



