ASSOCIATION AND COLOR DISCRIMINATION 



467 



Thus, although no method was at hand for determining the 

 intensity of a colored light, lights of varying intensity could be 

 passed through the monochromatic filters and the effect noted. 

 The following intensities of light were used: 4.9 candlemeters, 

 2.5 candlemeters, 1.4 candlemeters, 0.68 candlemeters. These 

 values were obtained by measuring the light when it was in the 

 box shining through the two glass plates inserted to protect the 

 gelatin filters. When the color filters were in position, the in- 



Fig. 4 Tin box in which light from an electric lamp, L, was passed through the 

 gelatin filters, F . P, P^ are glass plates inserted to prevent overheating the gela- 

 tin filters, F. During the experiments the box was covered with a tight lid 

 which is not shown in the diagram. 



tensity was greatly decreased. During these experiments the 

 translucent window shades were tightly drawn to increase the 

 effect of the colored lights, and the fish tanks were surrounded 

 by a black opaque screen (S) so that the fishes might not be in- 

 fluenced by the movements of the investigator. 



Experiments were first carried on with mudminnow no. 27 and 

 mudminnow no. 40. The light was first used full strength (4.9 

 cm.) with red no. 71 for paper and green no. 74 for food. Wlien 

 these fishes has shown by a perfect record that they had formed 



