Although their reactions were noted, they were not measured because 

 of the unsuitable swinnnning behavior. At times they would circle slow- 

 ly for hours at one end of the pond; at other times they would circle 

 for long periods at one end, move along one wall to the other end, amd 

 repeat the performance; only rarely would they move regularly up amd 

 down the length of the pond. 



Under ideal control conditions, the tunny would swim up 

 arid down the pond in one or two schools at a leisurely but regular 

 pace, occasionally pausing to circle or "play" for a few seconds at 

 the ends„ A trip down the pond ajid back usually required about 3 

 minutes. When joined by the yellowfin, sometimes the faster tunny 

 ar.d sometimes the slower yellowfin acted as pacennakers, thus vary- 

 ing the swimming speed. At times, either of their own volition or 

 attracted by the yellowfin, the tunny would spend long periods of time, 

 sometimes an hour or nnore, at one end of the pond„ When this oc- 

 curred, experiments could not be conducted for the test material 

 might be diluted below the threshold for response before the fish en- 

 tered the attraction area. Several experiments were spoiled when, 

 following uniform control conditions and the introduction of the test 

 substance, the fish failed to return to the area within the required 

 time. Even if they entered the area for the first time during the 

 second or third test period, the quantitative data would not be com- 

 parable with those obtained during normal behavior. 



Measurement of Response 



From the sources of variation discussed so far - weather, 

 variation in testing procedure, and erratic behavior of the fish during 

 eit;her control or test conditions - the difficulty of obtaining consistent 

 quantitative data will be apparent. Moreover, the timing and counting 

 did not measure the more obvious characteristics of a connplex posi- 

 tive reaction - surfacing, splashing, and feeding activity. On one 

 occasion the response was recorded on motion picture film, a method 

 which might lead to a better quantitative measure. However, this 

 method was not feasible in regular testing because of the great varia- 

 tion in visibility of the fish (depending on weather, turbidity of the 

 water, height of the tide, and depth of swimnning of the fish) and also 

 because of the prohibitive cost of the thousands of feet of filnn that 

 would be necessary to record the many experiments which were con- 

 ducted, 



A series of pond experiments (Nos. 94 to 100, inclusive) 

 has been chosen to illustrate the nature of the data and the problenns 

 of statistical amalysis and interpretation under average conditions of 

 observation and behavior of the fish. These seven experiments were 

 undertaken in succession throughout the day of August 26, 1952, four 



18 



