Alt. VI.] Herrick, Taste in Fishes. 95 



presence of the observer (which required several months of 

 training) they began to show evidence of visual recognition of 

 a moving bait, if very near them, and provided they had just 

 previously been fed with the same food in the same way. 

 They never under any circumstances notice visually a still bait 

 and their recognition of a moving bait is at best very imperfect 

 and only an occasional occurrence. 



Upon putting a concealed bait in a tank with the fishes I 

 found no evidence that they are able to locate it by the sense 

 of smell or otherwise from a distance, provided the water is 

 still. If, however, they swim near enough to the capsule con- 

 taining the bait (beef liver, cheese, etc.) to pass the barblets 

 into the strong diffusion currents emanating directly from the 

 bait, it is located instantly. The reactions here are essentially 

 like those by which the tom cod localizes a concealed bait, 

 though I have not completed the experiment by extirpation of 

 the nose to determine what part, if any, is played by the sense 

 of smell. So far as my experiments have gone, these fishes 

 will not locate a concealed bait in still water until they pass 

 within 5 cm. of it. 



In running water, however, the case is quite different. I 

 constructed a long, narrow tank, so arranged that a slow stream 

 of water can pass through it from end to end. By covering 

 the lower end of the tank and illuminating moderately the up- 

 per end, it can be so arranged that the negative phototaxis will 

 counteract any positive rheotaxis and the fishes will remain in 

 the lower end of the tank. If now liver or other strong bait is 

 placed above them, the fishes will promptly swim up the cur- 

 rent and locate the meat. 



The experiments seem to indicate that concealed food can- 

 not be located by these fishes from a distance in quiet water 

 (cf. Nagel, '94); but that if the fish passes within a few centi- 

 meters of it the diffused juices are recognized and the food lo- 

 cated promptly. In running water, however, the fish will fol- 

 low the diffused juices up the stream for considerable distances 

 and so find the food — a fact well known to every fisherman. 

 Tactile sensations are clearly not involved ; it lies between the 



