Sources of error include the effect of parallax in determining when 

 the fish is exactly below the lines, deviation from a path exactly 

 perpendicular to the lines ^ and deviation from a hbrizontal plane. 

 However f, variation induced by these sources of error is small com« 

 pared with the variability from trial to triaL The results of obser- 

 vations under control and experimental conditions are summarized 

 in table 7o 



Table 7o "-Summary of data on speed of swim- 

 ming under various experimental 

 conditions 



The mean speed of swimming increased beyond that of 

 the controls when the tunny were stimulated by lures^ and it in- 

 creased still more when they were under the combined stimulus of 

 lures ajid extractc The mean difference between control and lures is 

 not significant statistically (P ~ O0O8L However, that between lures 

 and lures plus extract^ and also that between control and lures plus 

 extract are both highly significant fP < 0, 01 in each case)„ The in- 

 creased rate of swimming of the tunn/ when stimulated by the extract 

 was obvious to the observer. The actual mean swimming rate under 

 these conditions (3„84 feet per second) is equivalent to 2,6 nniles per 

 hour. Of course^ bursts of speed several tinnes this magnitude were 

 frequently observed, 



j'eeding Activity 



Feeding activity may be measured by the number of 

 passes (bitings or attempted bitingst at the lures both when used 

 alone and when used in conjunction with extract. 



Table 8 shows the nnean number of passes at the lures 

 per IS'-minute period on 5 successive days The grand means for 

 lures and for lures plus extract are respectively 14 and 40 passes 



73 



