564 



It will be noted that the sokition became less deadly as time passed, 

 yet the fishes placed in it on the fourth day were dead on the ninth 

 day. The explanation for the rapid death of the fish on the thirteenth 

 day is not clear. The last of the other fishes w^as removed on the 

 ninth day and the water was then not disturbed till the thirteenth day. 

 During these four intervening days the water seems to have gained 

 in toxicity. The tests were not carried further because the stock of 

 fishes was nearly exhausted. The solutions in the other jars all 

 showed the same remarkable retention of toxicity, but none of them 

 v/as left undisturbed for several days and then retested, as it was not 

 thought that such treatment would have any effect other than a further 

 gradual diminishing of the toxicity. 



It is quite evident that solutions of CO do not behave as one 

 might expect were the gas simply in solution in the water, and it is 

 hard to account for the tenacity with which these solutions maintain 

 their toxicity except by supposing that the gas forms some irreversible 

 or slowly reversible compound with the water itself, or with some 

 substance in solution or suspension in it. In any event, it is certain 

 that the addition of even minute amounts of CO to natural waters 

 introduces a serious menace to the life of the organisms therein. The 

 extremely toxic effect of very small concentrations of the gas, together 

 with the fact that w'ater once poisoned with it is slow- to resume its 

 normal condition, makes this gas a source of grave danger to aquatic 

 life wherever introduced into natural waters. 



Reaction Bxperiinents. — Shelford and Allee ('14) showed that 

 fresh-w\ater fishes are very sensitive to carbon dioxide in a gradient 

 and that they will turn back quite definitely from small concentrations 

 of the gas. I have shown further (Wells, '15) that fresh-water 

 fishes tend to select a concentration of carbon dioxide that for most 

 species varies between i and 6 c.c. COo per liter. Shelford ('14) has 

 pointed out that carbon dioxide may be used as an index to the suit- 

 ability of bodies of water for fishes. 



To determine whether or not fishes detect carbon monoxide and 

 react to it in a gradient, a series of thirty-five experiments was run in 

 which ten species of fishes were tried out. The amount of CO intro- 

 duced into the treated end of the gradient varied from .5 to i c.c. 

 per liter. Higher concentrations were tried, but they killed the fishes 

 so rapidly that no results could be obtained with them. With the 

 lower concentrations some very interesting results were obtained. 

 There w-as no indication upon the part of the fishes that they detected 

 the presence of the CO with any precision, and most of the records 

 (made by graphing the movements of the fishes to a time scale) show 



