626 THE EYE IN EVOLUTION 



von Frisch (1912-25) also conducted an elaborate series of training 

 experiments on Phoxinus presenting food in grey and coloured tubes 

 or in association with grey or coloured papers. He found that his fish 

 readily learned always to seek the colour to which they had been 

 trained in preference to any shade of grey, even if the food were 

 omitted so that gustatory or olfactory clues were eliminated ; red and 

 yellow tended to be confused, but blue and green were not, either 

 between themselves or with red and yellow. This work seemed to 

 refute the conclusions of von Hess (1909-22) based, as we have seen, 

 on more doubtful evidence, and was corroborated on several species of 

 Teleosts by Goldsmith (1914), White (later Hineline) (1919-27), Reeves 

 (1919) and Hurst (1953) and on Phoxinus by Burkamp (1923), 

 Schiemenz (1924), Wolff (1925), Kiihn (1925) and Hamburger (1926). 

 It has been shown that once a food-relationship had been adequately 

 established with a particular colour, this colour is regularly sought by 

 the fish even when the factor of luminosity has been experimentally 

 eliminated, while Miss Reeves's experiments with a hue-discrimination 

 box with adequate controls can only be interpreted on the thesis that 

 the two species which she employed ^ appreciate hues as such. They 

 can be trained to go for food to a particular colour even when its 

 position and intensity are varied at random, and are not confused by 

 any other colour in any degree of brightness. 



This mass of experimental material suggests that the retina of 

 teleostean fishes contains a mechanism adequate to subserve colour 

 vision and the further conclusion would seem inescapable that these 

 fishes are possessed of a colour sense ; they appear to be able to 

 appreciate qualitative differences between the wave-bands appreciated 

 by us as red, yellow, green, blue, violet and the near ultra-violet (up 

 to 365 m/x and j^erhaps shorter, Merker, 1934-39). From the fact 

 that the most ready confusion exists between red and violet, it would 

 appear that their sensations may form a closed colour-circle. The 

 fact that they react to the human complementary mixtures of yellow 

 and blue, red and blue-green, orange and blue-violet, and so on, as 

 to white light suggests that their colour-system is closely akin to our 

 own (Hamburger, 1926, in Phoxinus, Beniuc, 1933, in the Siamese 

 fighting fish, Betta splendens). 



Beniuc's technique was ingenious. He trained the fighting fish to respond 

 positively to a grey disc and negatively to a slowly revolving disc of two 

 compleinentary colours in sectors yielding grey to the human eye in rapid 

 rotation ; when the speed of revolution produced 130 sector impressions per sec. 

 the fish responded positively as if to grey. At 90 imiaressions per sec. the fish 

 reacted negatively as to separate impressions — their fusion-frequency is therefore 

 much higher than that of man. 



^ The dace, Semotilus, and the sun-fish, Lepomis ; verified by Hurst (1953) on 

 the latter. 



Betta 



