EEACTIONS OF FISHES TO ATMOSPHERIC GASES 257 



of this fish to the various factors. As shown by the chart, 

 Hybopsis clearly avoids water which has lost some of its salts, 

 carbon dioxide and nitrogen (p. 240), and reacts a great deal 

 more vigorously to water which has also lost most of its oxygen. 

 However the reaction to an oxygen gradient in boiled water is 

 less definite than to tap against boiled water. The reactions to 

 carbon dioxide are all definite, being most so to carbon dioxide 

 in boiled water. 



Table 20 not only gives relative vigor of reaction when read 

 from right to left but enables one to compare species by reading 

 from top to bottom, in so far as the various factors have been 

 worked. Notropis and Abramis stand second after Hybopsis. 

 The data are far too incomplete and the experiments too few in 

 number to justify general comparison. The column at the ex- 

 treme right suggests the reactions of the different species to 

 combinations of oxygen and carbon dioxide. The amounts of 

 carbon dioxide used were much higher than the animals com- 

 monly encounter in nature, so that these figures could not be 

 used for ecological ratings and comparisons even if there were 

 enough experiments to justify the attempt. The ratings on the 

 basis of a gradient of 5 cc. of carbon dioxide per liter given in 

 the column to the left of the last would come nearer to rating 

 the fishes according to their distribution in clear, well-aerated 

 water and stagnant and foul waters. The reactions to oxygen 

 are clearly greater than the rating of accidental preference for 

 one end or other of the control, while the reactions to the effect 

 of boiUng, with oxygen added to balance, are variable and the 

 rating is less than the errors of the controls for Lepomis, Notropis 

 and Catostomus. 



When compared with the habitat preference data of Forbes 

 and Richardson ('08, pp. 79-85), we are unable to see definite 

 correlations between the occurrence of fishes with reference to 

 size of stream or pond; or with reference to current or kind of 

 bottom. However, it appears significant that Hybopsis, which 

 proved to be the most sensitive fish in these experiments, is 

 shown to have the most limited habitat preference of any of the 

 fishes studied. Since the environment of fishes is a complex of 

 many factors we cannot expect correlation to be possible until 

 experimental study has been carried much farther. 



