124 'Journal of Comparative Neurology and Psychology. 



it rose from straight below, midway between the forks. In the first 

 third, then, the position of the forks seems to have made Httle 

 difference with the results. 



In the second third, the percentages of right and wrong cases 

 were as before, 52 per cent, and 48 per cent. But 75 per cent, of 

 the right cases happened when the fish was on the same side as the 

 green fork, and only 25 per cent when it was on the opposite side; 

 while of the wrong cases 64 per cent, occurred w^hen the subject 

 was on the opposite side, 18 per cent, when it was on the same 

 side, and 18 per cent, w^hen it rose from below, between the two 

 forks. In the last third of the experiments, 81 per cent, were 

 right and 19 p^r cent. WTong. Of the right cases, 50 per cent, 

 were from the same side and 50 per cent, from the opposite side; 

 of the wrong cases, 50 per cent, were from the same side, 25 per 

 cent, from the opposite side and 25 per cent, from the middle. 

 In the previous experiments, where the food was in the red fork, 

 the association was pretty well established before the final method 

 was adopted, which allowed the place of the fish's entrance and the 

 position of the forks to be regularly varied. Of the few errors that 

 occurred after the adoption of this method, half were made when 

 the fish entered on the same side as the red fork, and half when it 

 entered from the opposite side; while, of the right cases, 50^ per 

 cent, happened when the fish was on the same and 49^ per cent, 

 when it was on the opposite side. It looks, then, as if the associa- 

 tion once formed was wholly independent of the subject's position, 

 but that, at a critical period, w^hen the animal was not strongly 

 drawn to either fork, the fact that it happened to come upon one 

 of them first was in some degree an influence leading it to bite at 

 that one. 



In summary, the experiments indicate the following conclusions : 

 I. Semotilus atromaculatus distinguishes red from green and 

 from blue pigments, the discrimination being independent of the 

 relative brightness of the colors. It must be borne in mind that 

 owing to the great physical complexity of pigment colors, the 

 existence of specific visual qualities in the fish's consciousness 

 cannot be inferred with as great certainty as if pure (homo- 

 geneous) colors had been used. 



2. An individual of this species is capable of forming with 

 considerable rapidity an "association" between the impulse to 



