MODIFICATION OF THE BEHAVIOR OF FISHES 19 



reaction. The control fishes avoided the end corresponding to 

 the carbon dioxide end of the experiment as clearly as did those 

 in the experimental tank. When they were confined in the 

 avoided end the apparent preference for the other end was 

 broken in the control, but only strengthened in the experiment. 

 In the control of Expt. 58, Abramis established an apparent 

 preference for the end avoided in Expt. 84. Since Abramis does 

 not rest on the bottom, this reaction might be thought to be 

 parallel to the resting of other fishes (chart 1, Expt. 10), were it 

 not for the fact they that choose one end after visiting both 

 many times, while others usually come to rest after the dash 

 which follows their being poured into the tank. The behavior 

 of Abramis in these controls was similar in many respects to the 

 avoidance of carbon dioxide and acid given by this fish and others 

 in the experiments. That is Abramis sometimes reacted positive- 

 ly to one end when tap water was running into the two ends from 

 the same pipe at the same rate. The fishes do not appear to 

 have any special tendency to rest near objects. They may 

 sense the current at the end, but we have no evidence that 

 such is the case. 



These apparent preferences of Abramis demand further ex- 

 perimentation for their analysis. From the work so far they 

 apparently do not belong in the same category as the rest of the 

 reactions described here and may be due entirely to associative 

 memory. Although if this be true the associated elements are 

 at present unknown. 



A general summary of the data on modification is contained 

 in tables 3 and 4. In table 4, the various species are arranged 

 roughly in the order of their sensitiveness. Later work by 

 Mr. M. M. Wells has shown that the data in columns three and 

 four are not accurate, because the time to loss of correlation 

 or until "staggering" occurs is rather indefinite and difficult to 

 determine. He uses the time until death. However, the data 

 represent relative sensitiveness in a general way. The time 

 required to accelerate the respiratory movements in low oxygen 

 is likewise difficult to determine, but the data presented are more 

 reliable than that in the two preceding columns. The remaining 

 data are concerned with the number of trials before evidence of 

 modification was given by turning. This is a definite criterion 

 and does not seem open to serious criticism. Reading 6, 7, 8 



