8 VICTOR E. SHELFORD AND W. C. ALLEE 



Chart I 



Showing modification of fish behavior by contact with boiled water, and with 

 boiled with acetic acid added and with boiled water with ammonia added. 



Distance between the small vertical lines adjoining the scales at the top cor- 

 responds to the length of the tank. The scales represent time in minutes divided 

 into ten-second periods. The horizontal distances in the graphs represent the 

 portion of the length of the tanks traversed by the fishes, the obliqueness of the line 

 as measured by vertical distance and the vertical portions of the graph lines rep- 

 resent respectively the time required to move the distance and the time spent in 

 resting or moving crosswise. The numbers above the generic names at tht top 

 represent the number of fishes used. When all or two of the fishes moved together 

 a solid line occurs. The movements of single individuals are shown by broken lines. 

 Double pointed arrows above the tracings indicate the distance occupied by 

 the gradients. When the concentration was high the entire tanks were gradient 

 and the secondary gradient is represented by the lighter double pointed arrows. 

 A corresponding portion of the controls is likewise indicated. The kind of water 

 introduced at the end indicated by the words "tap" and "boiled" the former being 

 the kind of water in which the stocks here discussed were kept. The oxygen con- 

 tent of the water is given in cc. per liter in the vertical wordings where the amount 

 of added solute is indicated also in cc. per liter excepting acetic acid which is given 

 in grams per liter. For statistical purposes (Shelford and Allee '13) the proportion 

 of time in the two halves, the number of turnings in the gradient are used as data. 



In experiment 10 the fishes are shown to have entered the boiled water three 

 times during the first two minutes, spending about one-half of their time there. 

 At the end of the two minutes, they began turning back occasionally. This con- 

 tinued until the end of ten minutes when turning became the rule with more or 

 less rhythmic entrance into the treated water. The experiment lasted forty minutes 

 but the remaining ten minutes showed nothing different. When emptied into 

 the control tank the fishes came to rest in one end and remained there for the first 

 twenty minutes, a common reaction when the water is like that from which they 

 were taken. At the end of the twenty minutes the fishes began moving back and 

 forth in a symmetrical manner. Experiment 13 shows the reactions of sun fishes 

 which are representative of the reactions of the fishes studied, to boiled water. 

 The graph of experiment 41 shows a reaction to acetic acid comparable to that 

 given to carbon dioxide. The reaction to ammonia might well be that of a control. 



Since three fishes were used it was not always possible to distinguish the different 

 individuals in the experiments and in working over the graphs. In the case of 

 the gregarious species, all three nearly always moved together. With non-gregarious 

 fishes it was nearly always possible to distinguish the different individuals in any 

 one group of invasions of the stimulating water. Individuality was lost only in 

 the periods of rest between invasions. After the graphs were made, the tracing of 

 each fish was followed with red, blue or green ink, the individuals being followed so 

 far as possible, but where individuals were lost during a period of rest each tracing 

 was continued without conscious reference to what the individuals had done pre- 

 viously. This was due to the fact that the tracings were made as a basis for calcu- 

 lating time spent in the two ends and before any discussion of modification was in 

 mind. The individuals are then, where not clearly distinguished, treated in a 

 chance fashion, and should show the same average result as the actual movement 

 of the individuals. The comparisons of species and factors are valid because such 

 errors as occur are present in all the experiments and we believe that any errors 

 arising from failures to distinguish individuals are of minor significance. 



