490 ADAPTATIONS TO PHOTIC QUALIT^ 



Horio then sought a pair of colors which would be as difficult for the 

 fish to tell apart as were the triangle and square. He found it, in violet 

 and blue. Trained positive to a violet disc versus a blue one and to a 

 white triangle versus a white square, then offered a violet square versus 

 a blue triangle, the fish went to the positive color rather than to the 

 positive form. Color thus seems to lie between brightness and form as 

 regards its attention value. 



When color and form were used in summation instead of at cross- 

 purposes, Horio obtained some unexpected results. Fishes partially 

 trained to a red or violet disc versus a blue one, and separately to a 

 white triangle versus a white square, made more accurate choices when 

 offered a red or violet triangle versus a blue square than when the 

 stimuli differed in only color or form. But when these independent 

 trainings to color and form were both complete, the fishes made more 

 errors on the combination stimuli than on the simple ones. Horio de- 

 cided that this must mean that the fishes had had time to develop 'red 

 disc = food' and 'white square = no food' associations, not merely 'red = 

 food' and 'square = no food' ones. Hence, the red triangle had a weaker 

 effect on them than either the red disc or the white triangle. 



Amphibians — Most of the Amphibia are nocturnal and secretive, and 

 it is difficult to see what color vision could do for them if they had it. 

 The common (ranid) frogs are arhythmic animals however, which might 

 have, and might benefit from, color vision. Except for one recent 

 Japanese report of work on a larval salamander, which the writer has 

 not been able to see, all of the efforts to find color vision in amphibians 

 have been made upon frogs. 



As early as 1900 it was established by Himstedt and Nagel that the 

 frog has a Purkinje phenomenon. Their technical tool was the electro- 

 retinogram, the record of retinal action-currents. Granit and his co- 

 workers, with similar but refined methods, have found the photopic and 

 scotopic maximally-effective wavelengths to be practically the same 

 (560mp, and 507m^) as those giving the peak brightnesses for the 

 human. Therman, in the same laboratory, found an increased electrical 

 response in blue light and a decreased response to red, in dark-adapt- 

 ation. Expansion of the retinal pigment in darkness by injections of 

 adrenalin failed to upset this relationship, casting further doubt — if any 

 were needed — upon Hess's interpretation of the Purkinje phenomenon 

 in fishes. 



