248 E. P. CHURCHILL, JR. 



.25 hours. The time lessened from day to day though the 

 fish had to a certain extent to relearn the maze each day. If 

 a partition of wire netting with openings 5 mm. square were 

 used the fish swam at once through the passageway and to the 

 nest. She thought the fish saw the netting and the opening 

 through it and consequently was guided by the sense of sight. 

 From various experiments she concluded that the illumination, 

 distribution of light and shade, and the play of light on the 

 walls of the container guided the fish. The objects in and 

 about the maze were not made use of as data by the animals. 

 The faculty of "motor memory" may be possessed by the fish 

 but she thought it was a less important factor in solving the 

 maze than the sense of sight. She also worked extensively on 

 the discrimination and memory of colors and form in fish. 



In view of the fact that the behavior of fish in a maze has 

 received less emphasis than the other features of the work with 

 this animal, it seems of interest to record an attempt at a quan- 

 titative estimate of the ability of fish to learn to run a maze 

 of a moderate amount of difficulty. 



Goldfish {Carassius auratus) were used because they live 

 readily in aquaria and can subsist on the prepared food which 

 does not diffuse through the water. As food was the stimulus 

 used this latter point was of importance. Eight fish were used. 

 These varied in length from individuals measuring 4.5 cm. to 

 one of 5.5 cm. They were kept in individual numbered pens. 

 One trial a day was given. The food was cut into squares about 

 4 mm. across. One such square was used for each trial and this 

 was the only food the fish received until the next day. 



A diagram of the maze used is shown in Fig. 1. It was of 

 the same general plan as the one used by Thorndike and con- 

 sisted primarily of a tank with glass sides and ends and meas- 

 ured 66.25 cm. in length, 37.5 cm. in width and contained water 

 to the depth of 18 cm. The tank was divided into three com- 

 partments (1, 2, and 3, see Fig. 1) of approximately equal size 

 by means of two partitions. In case of two groups of fish the 

 partitions were constructed of wire netting, in the case of the 

 other group they were of unpainted wood. In each partition 

 was a rectangular opening 2.5 cm. square. In the case of the 

 partition between compartments No. 1 and No. 2 the opening 

 was situated at the surface of the water and within 5.6 cm. of 



