482 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1918. 



difficult to secure the result desired. But one's trials mount very 

 rapidly with failing illumination, shyness of the subjects, their 

 activity, and, above all, with the roughness of the water. The camera 

 box is large and somewhat awkward to handle. It requires both hands 

 and knees to steady and manipulate it. In the meantime while one is 

 using it the heavy hood, left to its own devices, except insofar as it 

 may be steadied against the camera by pressure of the forehead, rides 

 upright or otherwise according to conditions. The long rolling swell 

 pushes one now a step or two forward, now back too far to accomplish 

 anything, and meanwhile the image in the mirror, never overbright, 

 goes into complete eclipse, as one's breath condenses upon the window 

 of the hood. Then whatever advantage of position may have been 

 gained for the moment has to be abandoned ; the glass must be flooded 

 with water and the attempt renewed from the beginning. 



The camera readily records such differences in regard to their 

 habits, as appear in the case of the fishes shown in plates 1 and 2. 

 The former, as has been said, are nocturnal ; the latter, diurnal reef 

 rangers. The one class, as one might infer from the picture, is 

 marked in general by its comparative inertia when undisturbed ; the 

 other is equally characterized by its restless activity. 



Each of these primary groups includes a great variety of species, 

 and is capable of subdivision upon the basis of differences in distri- 

 bution and behavior. , Some fishes sedulously avoid the light, and in 

 proportion to their numbers are rarely seen. Others swarm about 

 the coral beds or stacks, and are not found at all upon the open reef. 

 Some are found chiefly, or solely, on sandy bottoms; others among 

 particular sorts of marine vegetation. Some haunt the bottom; 

 some, the surface ; some, again, the intermediate depths ; while others 

 may swim at any level. Now follows a fact of great significance: 

 When the fishes are grouped naturally upon the basis of their micro- 

 geographical distribution, so to speak, it appears that the range of 

 variation in color is much less upon the average within each of the 

 subordinate groups than it would be in a group of the same size 

 selected at random from the local fish fauna as a whole. In addition 

 the colors dominant in the different groups are correlated in general 

 in a definite way with those of the preferred haunts of the group, 

 and tend to repeat them. This seems to indicate almost as clearly as 

 anything can that the colors of the fishes in question are upon the 

 whole what they should be in order to make them inconspicuous in 

 their normal surroundings. 



This is quite at variance with the conclusion naturally drawn from 

 the appearance together of many species of different colors in one 

 circumscribed region. But, with few exceptions, readily recognized 

 as such, the places where species congregate most freely are foci 

 where many simpler environments meet, or even overlie one another. 



