THE COMPARATIVE PSYCHOLOGY OF COLOUR VISION 137 



were left transparent and a biscuit was placed within view in the home 

 box. Next, cards of medium grey were placed in S,., S:i, S4, and S5. 

 Finally, only one shutter contained a grey card. The dogs were thus 

 taught always to go to the opaque screen in order to get through into 

 the home box and receive the reward. 



The various test series were then commenced. (1) "Colour prefer- 

 ence series." Maximum blue, red, yellow and green cards were put in 

 four screens and the frequency with which each was selected was noted 

 in 70 tests. For one dog the numbers were red 32, blue 26, yellow 8, 

 green 4. The red and blue were much darker to the human eye than 

 the yellow and green. (2) " Approximate brightness value series." 



f- ^ ■ ^ ....■ v - J^ . '. ^ 'V — ■■■■"■■;^-^' 



fl t f I // 



?■* ^ i I // 



i s L I I * / 



ii . . . ■.-^. $t . . . .1. .....t^ > .. 4,/ ... 



Fig. 43. Elevation of experimental dog box. (E. M. Smith.) 



Red, blue, yellow, green, and 14 shades of grey were used. First, the 

 particular colour was presented simultaneously with the grey which was 

 most remote from it in brightness. When the animal had given proof 

 of its ability to discriminate between these two the next grey, differing 

 only slightly less in brightness, was substituted. Second, all the greys, 

 in the same order as before, were presented on the same day for dis- 

 crimination from the standard colour, each grey being exposed 5 times 

 in succession. Next, the colour and the greys were presented in 

 irregular order. It was found that in the case of the most intelligent 

 dog the average percentage of right reactions was much higher for green 

 and yellow than for blue and red : further the first colour tested, blue, 



