444 FRANCIS B. SUMNER 



shade, a plate of glass was employed as a bottom, having an 

 area of 17 x 27 cm., and divided transversely into a black and a 

 white half. Fishes nos. 11 and 12 were used in these experiments. 

 Both were healthy and active. The former was, at the time, 

 adapted to a very pale bottom, the latter to the dark sand. The 

 fishes, one at a time, were placed upon this background, in the 

 neighborhood of the division between the white and black halves, 

 and in such a position that they could plainly see both. This 

 experiment was repeated a number of times with each fish, the 

 latter being left in some cases for more than an hour upon this 

 bottom. No preference was shown for one surface more than the 

 other. The fish commonly remained very near to where it was 

 placed, whether or not it was adapted to the surface which immedi- 

 ately surrounded it, and in no instance crossed over into the oppo- 

 site side. On the contrary, it happened that in more than one case, 

 the dark fish moved further back into the white area, and vice 

 versa. 



Each of these fishes was then placed upon bottoms of dark and 

 of white sand. No. 11 showed no disposition to burrow in either. 

 No. 12 covered itself very little with either sand, and with one 

 no more than with the other. 



Such experiments are of course not entirely conclusive, but, 

 when taken in connection with other observations, they at least 

 render it improbable that the fish exercises much selection in 

 respect to the shade of its background. The behavior of this 

 animal is thus not at all in accord with that of the decapod crus- 

 tacean Hippolyte varians, as described by Gamble and Keeble.^ 



The role of sight in these reactions 



Various previous observers^" have recorded that blind fishes failed 

 to undergo adaptive color changes, and it has been pointed out that 

 both specimens which have become blind through natural causes, 



^^ Quarterly Journal of Microscopical Science, 1900, p. 601. (See especially plates 

 32 and 33.) 



*' E.g. Pouchet, Mayerhofer and Secerov, in works already cited. 



