were present the reaction was similar to that described in the fore- 

 going paragraph, but was less violent and less spectacular. Even 

 with a larger number of fish, it could vary in its intensity from the 

 very strong reaction described above to a slight sensing of the mate- 

 rial - a mere swerving in the area as the fish passed through. Often 

 only one or two (e, g. a and d above) of the five reaction components 

 were evident. Also, of course, many of the substances which were 

 tested gave no noticeable or measurable response, 



PROBLEMS IN TESTING 



From the foregoing account of the methods, procedures, 

 and response one might assume that precise, clear-cut conclusions 

 could be drawn from each test. This was not the case. There were 

 several uncontrolled variables which affected the results and which 

 nnade it difficult, and often impossible, to obtain reasonably consist- 

 ent and comparable data. 



Weather 



In the tank, when the sky was overcast the fish were 

 difficult to see and hence to time and count. Several experiments 

 v/ere conducted under unfavorable weather conditions, thus introduc- 

 ing an additional source of error in the quantitative data. Testing 

 during rain squalls was usually avoided, for the rain rippled the sur- 

 face and made observation very difficult. Occasionally, however, 

 rain occurred while experinnents were in progress, thus not only 

 changing the conditions for observation of the fish during test as com- 

 pared with control periods, but also providing a distraction which 

 might affect the reaction of the fish. 



These same difficulties were encountered to even a 

 greater extent in the pond, where the water was deep>er, and the fish 

 were difficult to see from the tower when swimming deep. Occasion- 

 ally, visibility was greatly reduced by southeast "kona" winds, which 

 not only created a turbidity but also, blowing lengthwise down the 

 pond, created ripples on the surface. Often, to avoid wasting a whole 

 day, experiments were conducted under such conditions. Although 

 the resulting data were not strictly comparable with those obtained 

 during fair weather, a positive response during test periods could 

 usually be detected as, when reacting, the fish tended to swim at the 

 surface where they could be seen even under unfavorable weather 

 conditions. 



16 



