4 CORA D. REEVES 



iish should be avoided. In the experiments to be described 

 the fish have been kept under normal conditions and have 

 remained in the same containers during the tests and in 

 the intervals between them, often for months together. 

 The experiments have been so conducted as to reduce 

 fright behavior to a minimum. During three years the 

 fish have taken food regularly, have remained healthy, 

 have grown and have been at all times very tame. 



The apparatus described in the following section was 

 devised to meet the experimental conditions already out- 

 lined. These call for; 



(1) An experiment aquarium in which the fish could 

 live continuously without being taken from the water. 



(2) An experiment procedure which would not arouse 

 fear behavior. 



(3) Two large stimulus patches of either mixed light or 

 light of restricted and known wave-length, interchange- 

 able in position, one of which could be varied through a 

 wide range of measurable intensities. 



(4) Provision for offering food before one of the patches 

 of light, which thus becomes the positive stimulus in the 

 formation of an association. 



(5) Constant conditions in the general luminous and in 

 the aqueous environment of the fish. 



(6) The experimental procedure involves further a 

 method of equating the brightness of the patches of light 

 of different wave-length by means of the behavior of the 

 fish. 



B. Apparatus and Methods. 



1. General Description of the Apparatus 



For the experiments a black-lined, galvanized-iron 

 aquarium was used. Its ground plan is shown in figure .1. 

 Movable partitions separate it into three parts, referred 

 to as the stimulus, the discrimination, and the retention- 

 compartments, respectively (1, 2, and 3, in fig. 1). 



The removable partitions A and B separate these com- 

 partments as indicated by their positions in figure 1. In 

 partition A is a vertically sliding door, D, shown partly 



