464 FRANCIS B. SUMNER 



this purpose, I have employed two methods, that of staining the 

 skin of the fish and that of covering it with a cloth, stitched along 

 the margin of the body.^- At first I made a full-length 'swimming 

 suit' for the animal, but this did not seem necessary, since from 

 the position of its eyes, only the anterior portion of the body can 

 fall within its range of vision. Accordingly, a mask only was 

 employed in my later experiments, apertures being made for the 

 eyes. 



Now I have devoted considerable time and trouble to experi- 

 ments of this sort. Eighteen fishes were provided with cloth 

 masks or coverings for the body, and 8 others were stained in 

 various ways.''^ Fully satisfactory tests were, however, found to 

 be difficult, if not, indeed, impracticable. It was, for example, 

 very hard to cover the head completely with cloth, and at the same 

 time permit of an unobstructed view for the eyes. In the earlier 

 experiments, the body alone was covered, leaving the head, or 

 most of it, exposed. The results from such are, I think, wholly in- 

 conclusive. Stains, unless rubbed in with considerable force, were 

 found to affect only the mucus covering the body, and to be 

 removed with the discharge of this secretion. 



Moreover, all of the methods employed were open to one seri- 

 ous objection: they injured and sooner or later killed the fish. 

 Under such circumstances, it would be expected that disturbances 

 of the normal reactions should occur, and such was indeed the 

 case. A merely negative result in any instance, i.e., the failure 

 to respond to a given stimulus, cannot therefore be regarded as of 

 great importance. We cannot, on the other hand, deny the sig- 

 nificance of any positive results which were obtained, if the experi- 

 ments were otherwise above criticism. 



Suppose, now, that the anterior parts of a dark flounder be 

 covered with a white mask, and that the animal be placed in a 

 white tank. The fish would see itself as white. According to the 

 hypothesis we are testing, there would seem to be no reason for 

 change. The converse experiment might be performed with a pale 



'2 This suggestion of covering the fish with a cloth I owe to Professor Parker. 

 '^ Potassium permanganate and silver nitrate were the stains chiefly used. Both 

 imparted a very dark shade to the skin. A white stain proved to be impracticable. 



