208 H. M. JOHNSON 



bull-terrier, are shown on page 352 of that report. When the 

 width of each stripe on the "positive" field was 5.2 mm., and the 

 distance from the eye was 60 cm., the animal chose the positive 

 field 36 times in 50 presentations or "trials". In the last 20 

 trials, 18 choices were correct. Believing at the time that he 

 had learned the problem, I reduced the width of the stripes. 

 On examining his subsequent daily records with these, I became 

 convinced that the improvement was accidental, and due to a 

 change in the animal's "position-preference" as a time when his 

 right-left order of choosing happened to fit the order of presen- 

 tation quite well. When the width of each stripe was 3.9 mm. 

 the animal chose correctly in 127 cases out of 200; when it was 

 reduced to 3.72 mm., 112 out of 175 choices were correct. 7 



These results did not show anything conclusive regarding the 

 dog, except that working near the limit of the instrument I had 

 not made a stimulus-difference large enough to be effective in 

 50% of the cases. (This condition would be met when 75% of 

 a large number of responses were correct.) 



The general method followed in these experiments had yielded 

 quite definite and positive results in the work on the monkey 

 and the chickens. The individual stripes composing the posi- 

 tive system had subtended a very large visual angle throughout 

 all the w T ork on the dog. The failure to obtain positive results 

 on the dog raised several questions regarding the work. 



Early in the training the animal had learned to choose the 

 correct food-box by testing the electrodes at the entrance-alleys 

 for electrical charge. When I eliminated this factor he showed 

 great disturbance. It seemed advisable to use another indi- 

 vidual, which had not been thus disturbed during the training. 

 I selected a pure-bred female beagle-hound, four months old, 

 for this purpose. Beginning May 26, 1914, I gave her ten trials 

 daily for 90 days, using the method described in the work on Dog 

 1, the monkey and the chickens. She showed no consistent pref- 

 erence at any time for the positive field, each stripe on which was 

 5.2 mm. wide. The minimal distance between test-field and the 

 eye at which a choice could be made was 50 cm. A detailed re- 

 port of her work would be without value in view of facts which 

 we later ascertained, and which I shall mention below. 



7 After the original report was prepared I resumed the work at this stimulus- 

 value, and obtained a lower percentage of correct choices over 150 additional trials. 



