428 FRANCIS B. SUMNER 



would seem that the fishes should behave in such a tank much the 

 same as upon an ordinary white bottom of opaque material illu- 

 minated from above. As a matter of fact, however, only one-'^ out 

 of seven specimens which were used actually became paler, even 

 afterastayofseveral(inonecaselO) days in this tank. All the other 

 specimens either become darker, if they had previously been pale, 

 or remained dark, if taken from a dark bottom. An examination 

 of the actual conditions of illumination showed the reason for 

 this apparent anomaly. In order to see an illuminated surface 

 below, it was necessary that one should look directly downward, 

 i.e., in a direction nearly perpendicular to the glass. The fish, 

 however, necessarily viewed the bottom at a very obtuse angle, 

 and perceived, not the illuminated surface of the mirror below, but 

 a surface of total reflection, where the glass bottom met the air 

 This surface itself acted as a mirror, reflecting the darkened 

 walls of the tank, rising above it. What the fish actually saw, 

 therefore, was a dark surface, not a bright one. This could be 

 verified by the observer himself by viewing the bottom at a proper 

 angle. 2^ 



12. A few tests were made with glass strips painted wdth rather 

 bright shades of yellow and red.^*^ The two fishes used (no. 12 

 and no. 4) were ones which had, under some other conditions, 

 shown a high degree of adaptability. One of these was kept for 

 five days upon the red and an equal length of time upon the yellow, 

 without any approach to the color of the background. The other 

 was kept for 10 days upon the red, with equally negative results, 



^* This one later showed various other anomalies in its reactions. 



-' This same optical principle was responsible for another anomaly in my results 

 which at first baffled me. I had, in the beginning, prepared my black and white 

 patterns with paper and cardboard, mounted between two strips of glass, which 

 included, of course, a film of air. Viewed from above, the pattern was perfectly 

 clear and distinct, as seen through the water; but beyond a certain angle, one saw 

 nothing but a shining mirror. I was surprized by the complete failure of the fishes 

 to adjust themselves to these patterns, until I chanced to view one of the latter, 

 at an appropriate angle, through the side of the jar. For further references to this 

 point see p. 478, below. 



^^ The red was of a shade closely approaching the no. 61 of Klincksieck and 

 Valette, the yellow of a shade similar to no. 226. 



