A series of experiments was designed to tes' the hy^ 

 pothesis that the phosphate ionp assumed to be an integral part of 

 the attractanto was made available at a high or low pH through hy = 

 drolysiSo and that this was then precipitated ((and losti as magne- 

 sium ammonium phosphate by reaction with the seawater when the 

 test substance was introducedo Bigeye flesh was extracted with 

 alcohol and the filtrate was adjusted to pH 3 with hydrochloric acid 

 to make the phosphate available for later removalo The filtrate was 

 divided into sut parts,^ each treated differently as shown in figure 

 160 Since there was a strong response to that part subjected to the 

 most drastic method of removing the phosphate (No,; 303)o nannelyj 

 adjusting to pH l^, boiling 15 nnmuteSj adding Oo ^ go magnesium 

 chloride a adjusting to pH 9„ standing overnightj and then filtering 

 and testings, the hypothesis that attraction is associated with labile 

 or easily removed phosphate was discarded as untenablco 



Finallyj, a series of experiments was designed to in- 

 vestigate the variability of the data iwith only ',wo tunny left in the 

 pond)j the effects of pH, and the effects of time lag between pre- 

 paration of the material and testingo This serieSj conducted over 

 a 3=>week period in April 1952 (No. 313 et s^e^o )» has already been 

 discussed from the point of view of variability of the datao As may 

 be Calculated from table 3 fusing eight complete replicates! the 

 mean (observational; scores were lo8„ i„ \^ and lo3 at pH s of 2, 

 1, and lOj, respec tivelyo Although the results are suggestive of 

 decreased response at the two extremes of pHp the differences wheaa 

 tested with either the observational or the quantitative data are not 

 significant statisticallyo Moreover^ the relative superiority of the 

 acid and base ajfe reversed as compared with the preliminary re- 

 sults given in a preceding series with skipjack viscera extracto 



The preparations were made on a Saturday and were 

 tested in two replicates on the following Tuesday or Thursday by 

 one obseiverj and agaJi". in two more replicites on the following 

 Saturday by a second observer^ Thus the material tested by one 

 observer was prepared for a shorter period of time than that tested 

 by the second observero The datSi, although incomplete for reasons 

 discussed previously;, yield 12 paired comparisons which may be 

 segregated as follows according to time duration and pH; 



1 



"Tr-— — - 





pH 2 

 pH 7 

 pH 10 



Shorter time 

 (Observer 1^ 



Longer time 

 (Observer 2) 



2ol 

 lo4 

 lo3 



lo7 

 2ol 

 lo2 



Number of 

 comparisons 



55 



