percent during control conditions^ Z4o 5 percent when two lures were 

 usedp and 29o 5 percent when the lures plus extract were usedc A 

 rough test (using "t'M of signiiicajice of the difference of the paired 

 observations (neglecting the varying intrinsic accuracy of the mean 

 determinations for each day^s experiments" indicates a significant 

 difference between control and lures {mean difference^ 13,8 percent; 

 P < OoODj but not between lures and lures plus extract (mean 

 difference^ 4o 0; P - 0„2)o 



Table 5, ^-Summary of data on percentage time 

 spent in the iTtraction area under 

 various conditions of stimulation 



Fish^seeonds in Area 



The summation of the product of the number of fish and 

 time spent in the area is probably a better measure of activity than 

 the two measures discussed aboveo This is equivalent to summing 

 the areas under the histograms in the kymograph records^ and may 

 be expressed in fish= seconds per I5-niinute periods The results are 

 given for individual experiments in table 6o 



In comparison with the controlSj, it will be observed that 

 the number of fish^seconds is generally greater when the lures are 

 usedp and slightly greater still when the lures are combined with 

 extracto The grand means for controlp lures^ and lures plus ex- 

 tract are respectively 114,3p 352o8j and 364o8 fish-sec ondSo In 

 experiments with lures .^alonej which were repeated in succession on 

 the same d<ayp the measure tends to fluctuate rather widelvp showing 

 neither a consistent increase not decrease in responseo In experi>= 

 ments with lures plus extractj the measure tends to decrease with 

 repeated stimulation;, indicating a dulling of the response There are 

 five comparisons of controlp lureSp and lures plus extract involving 



71 



