NOo 94 and No, 96; although the observed reaction is judged the 

 same, X, the quantitative data diverge widely (although not signi- 

 ficantly according to statistical tests)„ As will be indicated later, 

 the observer's judgement of the strength of a reaction tended to 

 change with time and with decreased reaxtivity of the fish. Thus 

 a reaction which might be classed as X in August would likely be 

 classed as XX or XXX in ApriL 



Variation in Response 



The strength of the response to an attractive substance, 

 whether measured by the quantitative data or estimated by visual 

 observation, varied greatly from test to test both within and be- 

 tween daySo Throughout the mass of data there are numerous 

 examples of this„ For exannple in experiments NoSc 67, 72, 80, 

 97, 106, 111, 117b and 173, which were all conducted by one ob- 

 server using the same material (250 ml„ stock extract), the res- 

 ponse varied from 1 to 3 (X to XXX) on the observational scale. 



At times it was suspected that the variation in response 

 was associated with either the frequency of testing or the accunnu- 

 lation of test products. These factors could readily influence the 

 results obtained with fish in the tank, for it had both a small 

 volume and a small r<»te of flow of salt water. Thus, following 

 four expe rinnents in which relatively large quantities of test sub- 

 stances were used (Nos, 59 to 62|, there was no response to a 

 relatively large volume (1, 000 ml,) of stock extract (No, 63) in the 

 fifth experiment of the day, -Afte r a rest period of 3 days in which 

 no tests were conducted, much snnaller volumes (250 ml, ) of stock 

 extract gave strong responses (Nos, 81 and 82),, In the pond, at 

 times there appeared to be a gradual increase in activrity under con- 

 trol conditions and a gradual decrease in response to attractive 

 substances when several experiments were conducted in succession 

 during a day (fig, 5). This may be attributed to several possible 

 causes including chance or random variation, an increasing con- 

 centration of materials in the pondj, or aji increasing threshold of 

 response under repeated stmnul^itiono When a response was pro- 

 tracted beyond the ordinary tmne limit, usually the observer 

 would wait for 15 to 30 minutes before starting the next experiment 

 to allow the fish to resume normal control activity. To sonne 

 extent this would overcome the trends which are noted above. 



In April 1953 a series of experiments (No, 313 ^t_se q, ) 

 was undertaken to determine if there were differences in response 

 to materials maintained at three different hydrogen ion concentra- 

 tions (Ti"-pH 2, T2 = -pH 7, T3»-pH 10), but also to investigate 



24 



