234 VICTOR E. SHELFORD AND W. C. ALLEE 



which was followed by resting in the low for two minutes. 

 In experiment 53 Notropis showed similar alternations of inva- 

 sions and avoidances which took place more rapidly because 

 the fishes moved back and forth more rapidly. Notropis 

 (experiment 86, chart 1) invaded the high concentration only 

 once after the first minute. In the controls for all the experi- 

 ments, the fishes moved back and forth with more or less regu- 

 larity and on the average with little apparent preference for 

 either end. 



An inspection of table 6 and the graphs (pp. 230-233), shows 

 that activity is much greater in the experiments where low con- 

 centrations were used, as indicated by the number of crossings 

 of the center. The activity where high concentrations were used 

 was variable. The percentage of time in the low half of the ex- 

 periment is in all cases much greater than in the high. Time per- 

 centage and number of turnings back in the gradient portion are 

 the best indications of reaction. The number of turnings back 

 from the higher concentration was greatest in all cases. In the 

 controls, the number of turnings was usually nearly equal. 



The reactions of all the species of fish experimented upon are 

 quite similar although they belong to several taxonomic groups. 

 The data in table 6, if averaged, show an almost equal number 

 of crossings of the center in the experiment and control, but 

 there is a marked time preference for the low carbon dioxide 

 half of the experimental tank and a nearly equal division of 

 time between the two ends of the control. The average number 

 of turnings back from the higher concentration is thirteen as 

 opposed to one from the lower. The average turnings in the 

 controls are three from each half. 



Preliminary experiments were tried with Schilbeodes, Ethe- 

 ostoma, Boleosoma, and Hadropterus. While the experiments 

 were not carried far enough to give results of definite value they 

 suggest that these swift water fishes which probably encounter 

 very little carbon dioxide, may react less definitely to it than 

 the fishes which live more often in the presence of carbon dioxide. 



Table 7 shows the reactions of Abramis from the low oxygen 

 stock and Hybopsis to a carbon dioxide gradient in boiled water 



