THE LEARNING OF A MAZE BY GOLDFISH 251 



on its eighth trial. Fish No. 3 was the best of the group. Its 

 last five trials, the forty-second to forty-sixth inclusive, never 

 exceeded one minute in length while the last three were of a 

 duration of thirty seconds each. The average physiological 

 limit of the group seemed to have been reached, at about the 

 thirty-sixth trial, though the habit bore up well for the remain- 

 ing trials. 



The members of this group apparently used the sense of 

 sight very little in running the maze. They would repeatedly 

 swim directly past the opening without passing through. The 

 . interstices of the wire netting were fairly large, three to every 

 2.5 cm., and it is doubtful if the fish discriminated by the sense 

 of sight between these and the openings intended for their 

 passage. As far as could be observed they found the passage- 

 way by playing about the netting and thrusting the head into 

 the interstices until the opening was hit upon, this being large 

 enough to allow their passage. They gradually learned to 

 explore more and more closely about the opening and thus 

 the time was reduced. Toward the last they swam fairly ac- 

 curately to the opening and passed through with very little 

 nosing about. This was especially true when they were passing 

 through the second partition. The passage from compartment 

 No. 1 to No. 2 usually required more time than that from No. 2 

 to No. 3. This was no doubt due to the fact that the fish had 

 to orient itself after being placed in compartment No. 1. The 

 fish of this group very seldom went back into the compartment 

 from which they had just come. 



In Group II two fish were used with the wire partitions pro- 

 vided with the black bands about the openings. The curve for 

 this group is shown in Fig. 3. Forty trials were given in this 

 case. The curve is very abrupt in its initial part. Neither fish 

 succeeded in running the maze the first day. The number, 405, 

 toward which the arrow points in Fig. 3, represents the average 

 time the fish were left in the maze during the unsuccessful at- 

 tempts. On the next day one fish ran the maze in 7 minutes 

 and never consumed more than 15 minutes at any later trial. 

 As far as could be seen the failure of the first trial was due to 

 fear on the part of the fish. They avoided the region of the 

 openings during the first trial. However, after the first few 

 trials the improvement was more rapid and the final achievement 



