74 HARRY MILES JOHNSON 



wood and at times were given the freedom of a lawn co\'ering 

 several acres for an hour or more at a time. On i6 April, 191 2 

 Dogs 7 and 8 while blind escaped with Dog 6, a normal male 

 of the same litter and chased a half -grown cat together. The 

 latter took refuge in a corner of a porch and was caught and 

 killed by Dog 7. It was not possible in the time left to the 

 experimenter to make a test of the vision of Dog 7, the animal 

 which after the operation seemed most nearly normal. It can- 

 not be ventured how much or how little vision either of these 

 animals had. Inasmuch as their records while blind compare 

 very favorably w^ith that of the normal dogs I have only nega- 

 tive and nconclusive evidence to offer on the question which 

 was the primary object of this investigation. What data I 

 have gathered seem to show only this positive fact : that the 

 dog is capable of learning to make complicated adjustments 

 and of performing a sui-prising number of "instinctive" reac- 

 tions perfectly without the aid of vision. This fact remains 

 established no matter what later tests may show regarding the 

 dog's ability to make visual discriminations. 



RETENTION 



Sixty days after the third problem had been learned by each 

 dog, a test was made of the retention of the learning of the 

 first three boxes. The results are not shown because there was 

 absolutely no significant variation among the records of the 

 different animals. None showed a loss of accuracy of over 10% 

 on the first day's work in the retention test, and in only one 

 case was over three days' work necessary to regain the loss. 

 This exception was Dog 2 on box i. In learning the problem 

 the an'mal had used first her paw, then her nose, to depress the 

 latch. She settled on the nose-method. In the retention-tests 

 the paw-method reappeared and persisted for six days before 

 the nose -method was completely re-establidied. The time re- 

 cfuired for opening the box by either method was about the 

 same — not over one second. 



CONTROL TESTS 



Two control tests were made on Dogs i, 2 and 5. The first 

 was for the purpose of observing what disturbance resulted from 

 requiring the dogs to perform their work in total darkness. Two 



