2-4 



THE QUARTERLY REVIEW OF BIOLOGY 



which an observer with stop watch was 

 seated behind a screen. Mr. Herbert was 

 stationed along the wall outside the room 

 about six feet to the right of the open 

 doorway and hence completely out of 

 view of the objects located some 15 feet 

 inside the room on a line with the door. 

 The objects when retrieved were returned 

 before the next command by a third person 

 stationed outside the room with Mr. 

 Herbert, who distracted the dog in the 

 meantime by playing with him. The 

 dog always stood outside the room facing 

 his master when the commands were 

 given. He must then turn about, go to 

 the door, enter it and approach the objects 

 lying 15 feet ahead and retrieve the 

 proper object. The commands were often 

 repeated but great care was taken not to 

 repeat a command after the dog had 

 reached the threshold of the door. This 

 means, of course, that we have here, as in 

 the case of all type II situations, a delayed 

 reaction experiment and not a simple dis- 

 crimination or object-recognition set-up. 

 Inasmuch as Fellow moves about rather 

 deliberately the interval of delay between 

 the auditory signal (command) and the 

 essential response (picking up the proper 

 object) was quite considerable. The de- 

 lay was never less than 5 nor more than 35 

 seconds and in most cases was about 15 or 

 zo seconds, as can be seen from table z. 



A preliminary test, October 30, in which 

 at first ten and later five objects were 

 employed, led to the decision to use only 

 three objects in the test proper. A large 

 number of objects could not be placed in 

 the room so as to be equally accessible to 

 the dog on approaching them, and the 

 objects themselves naturally differed 

 greatly in size. The low visual acuity of 

 the dog for still objects has already been 

 commented upon and suggests the need 

 of ruling out any large differences in dis- 

 criminability of objects in making tests 



of this sort. Three of the following five 

 objects were used in each test: a man's 

 kid glove, a seven inch scrub brush, a 

 hotel key with tag attached, Fellow's 

 chain collar, and a package 4x5x1m. 

 in size. The three objects were placed 

 in a row about z feet apart and the order 

 kept constant during each of the three 

 test periods. 



The results of this series of tests (table 

 z) are rather encouraging in view of the 

 difficulty of the task. According to 

 chance iz of the 36 responses should be 

 successes and Z4 should be failures. Only 

 15 definite failures occurred as compared 

 with 16 clear cut successes. Three of the 

 five doubtful cases should probably be 

 counted correct since the proper object 

 was actually retrieved after a wrong object 

 had first been picked up and dropped. 

 The other two doubtful cases occurred in 

 connection with the return by the dog of 

 the wrong object brought on the previous 

 trial. He failed to drop the returned 

 object and came back with both it and 

 the correct one in his mouth. These 

 should probably be scored errors. On 

 this accounting 53 per cent of the responses 

 were correct as against an indication of 

 33^ per cent by chance. The general 

 arrangement of this test was more natural 

 than the earlier one in which Mr. Herbert 

 was concealed behind the screen, since the 

 dog now came in contact with his master 

 between trials. Fellow showed a slight 

 tendency to bring the same object that he 

 had brought the previous time, but this 

 does not account for many of the errors. 

 A more important factor was probably the 

 relative ease with which the various 

 objects could be picked up. Fellow has 

 lost two teeth on one side, which inter- 

 feres somewhat with the use of his mouth 

 in retrieving objects. He appeared to 

 avoid the package when it was lying down 

 flat, and therefore difficult to pick up, 



