324 VICTOR E. SHELFORD AND EDWIN B. POWERS. 



Water differing as little as possible from that in which the 

 fishes usually live was used for control readings. Controls were 

 observed and the conditions in the two ends of these were the 

 same either because the water introduced at the two ends was 

 alike or because no water was run into either end (standing 

 water) . 



In the controls the fishes usually swam from end to end in a 

 rather symmetrical fashion, and thus by comparing these move- 

 ments with those occurring when the fishes encountered differ- 

 ences in water, we are able to determine the reactions of the 

 fishes to the differences. Various kinds of water were used at 

 one end as follows: (i) water with varying amounts of carbon 

 dioxide added; (2) water with oxygen added; (3) water with 

 hydrogen sulfide added; (4) fresh water. 



When the difference between the solutes at the two ends of the 

 tank was not great we found by chemical tests that the central 

 portion of the tank was a gradient between the characteristic 

 waters introduced at the two ends. Usually the end thirds were 

 essentially like the inflowing water. When the difference in con- 

 centration was great the region of the gradient was propor- 

 tionally longer and the ends with the inflowing concentrations 

 correspondingly shorter. When the difference in concentration 

 was very great the entire tank was gradient. For an experiment 

 a fish was placed in a dish containing enough water to barely 

 cover it and set above the tank. When all was in readiness 

 the fish was emptied into the center of the tank. Marks on the 

 sides divided the tank into thirds. The fish nearly always swam 

 back and forth, apparently exploring the tank. The movements 

 of the fish were recorded graphically as shown in Chart I. For 

 this purpose sheets of ruled paper were used. Four vertical 

 double rulings corresponded to the thirds and two ends of the 

 tank. Distance from right to left was taken to represent the 

 length of the tank, vertical distance to represent time and the 

 graphs drawn to scale. The width of the tank was ignored. 

 The graphs on the following pages are copies of the originals. 



Before or after the experiment, the headings of the sheets 

 were filled with data regarding the kind, size, and previous 

 history of the fish, the conditions in the tank, concentration of 



