that an attractive substance would lure the fish to the rubber inlet 

 and hold them there,, but this proved to be the case only with highly- 

 attractive substances o With more weakly attractive substances there 

 appeared tc' be an increase m cruising speed and a tendency to circle 

 closer to the inlet o Accordingly., the reactions were measured in two 

 ways; (l) by determining the time required for the fish to make 10 

 passes across an imaginar-y line (AC in fig, 2) drawn across the tank 

 from the inlet, witli the fish travelling in either direction (when 

 the fish is circling normally this vd.ll be equivalent to the time of 

 5 complete revolutions) ^ and (2) by counting the number of passes 

 (out of 10) made across the line AB in the neer or west half of the 

 tank (whea the fish is circling normally there ild be 5 passes 

 across AB and 5 across BC)o 'Therefore j perception of the stimulus 

 would be indicated by a decrease in the time of 10 passes from that 

 of normalj representing an increase in cruising speed and/or a de- 

 crease in size of the svamming circleo Attraction would be indica- 

 ted by an increase beyond 5 in the number of passes across ABj 

 representing one or more circles completely within the near (west) 

 half of the tank,. In one case, as indicated laterj the substance 

 was siphoned in at C and the number of passes across BC was countedo 



It was assuraedj and later established; that the reactions would 

 vary with the state of hunger of the fisho To minimize variation 

 from this factor j the tuna were starved for 2ii hours before the start 

 of an experiments However j, the state of hunger induced a factor of 

 alertness which could easily result in grave error ss at the approach 

 of an observer the fish became excited, expecting food, and circled 

 close to the observer at increased speed for a considerable periodo 

 Tests were not started until the fish became accustomed to the pre- 

 sence of the observer,, and a "normal" cniising speed was resumedo As 

 the so-called "normal" cruising speed varied from day to day, it was 

 necessary tc establish its value before each experiments Itiis was 

 done by repeating the timing and counting of the passes until approxi- 

 mately constant values were obtained, vdth a half -minute interval 

 between successive tests,. The substance would then be introduced and 

 the timing test would be continued, still with half -minute intervals 

 between thera_, until the normal values Mere again approximated o 



As for a long time only one yellowfin was present., many of the 

 results were obtained on this one fisho liVhen more tunas were intro- 

 duced the schooling instinct became apparent. When tv/o yellowfin and 

 five tunny were present, the two species tended to school separately 

 and to exhibit different reaction patterns » After one yellowfin died, 

 the other joined the tunny, and although a slower swimmer, it attenpt- 

 ed to keep up with the schoolo It showed, therefore, an increased 

 speed of reaction as compar-ed with the results obtained when it was 

 the sole resident of the tank. After four of the five tunny had 

 died, the single tunny schooled with the single, larger yellowfin 



11 



