aqueous extracts of skipjack flesh (several experiments) or tunny 

 blood (one experiment). In all cases the dialyzate gave a positive 

 response when tested (Nos„ 39, 46, 53, 57, 90, 118, 123), showing 

 that the attractant consists (in part, at least) of a small molecule, 

 rather than a large molecule such as a protein. In no case, however, 

 was complete separation attained by dialysis. In most of the experi- 

 ments the residue (portion remaining in the tube) gave a stronger 

 response than the dialyzate (compare Nos, 39 and 40; 52 and 53; 57 

 and 58; 90 and 91) despite several changes of the distilled water sur- 

 rounding the tube. 



Precipitation of the proteins by dilute (10-percent) hydro- 

 chloric acid was next attempted. After the centrifugate of an aqueous 

 extract of skipjack flesh was treated with acid and centrifuged, both 

 the neutralized centrifugate (No. 68) and the neutralized residue 

 (Noo 69) gave positive responses, showing incomplete separation of 

 the attractant. 



As the above results indicated that the attractant was not 

 a protein, a series of pond experiments was undertaken in which the 

 proteins were precipitated by heatings On heating the centrifuged 

 aqueous extract of skipjack flesh to 69° C. , a precipitate appeared. 

 When this was removed by centrifuging, the centrifugate was still 

 attractive (No^ 76), When the centrifuged aqueous extract was boiled 

 for 30 minutes, resulting in a heavy precipitate, both the filtrate in 

 one experiment (No. 77) and the centrifugate in another (No. 78) were 

 still attractive to the fish. However, separation was not complete, 

 as is shown by the weaker positive responses obtained with the washed 

 residue from centrifuging (compare Nos, 75 and 76; Nos. 78 and 79), 



It is uncertain why the filtered extract of boiled tuna flesh 

 failed to give a positive response in the earlier tank experiments 

 NoSo 5, 9, 11, 12, and 21)o Although the behavior of the fish in the 

 tank was not too satisfactory, it seems certain that had a response 

 occurred, it would have been noted or measured in one of the five 

 experimentSo It is unlikely that the attractant was destroyed by 

 boiling in view of the results reported in the preceding paragraphs and 

 others to be presented later. There remains the possibility that the 

 attractant was precipitated with the soluble proteins or otherwise 

 adsorbed with the insoluble proteins during the process of boiling the 

 flesh; the residue, which may have contained the attractant, was not 

 tested in the five experinnents conducted with the boiled flesh. 



Chemical tests were run on the clear centrifugate obtained 

 after boiling of the extract. The Biuret test was positive, showing 

 the presence of the amide link. The Xanthoproteic test wa,s positive, 



38 



