THE QUARTERLY REVIEW OF BIOLOGY 



tirely over his boots, soles and all. 

 The dog was unable to trail him when he 

 wore his boots so covered. After walking 

 some distance a small bit of the paper wore 

 off the heel of one of the boots. When 

 the dog, which had been led along the trail 

 far behind in the meantime, reached the 

 point where the heel had worn through 

 and was thus making contact with the 

 ground it picked up the trail at once and 

 soon caught up with Romanes. This 

 little incident hints at the extreme deli- 

 cacy of the stimulus which is apparently 

 effective in trailing behavior. A charac- 

 teristic of trailing which has never been 

 adequately explained, as Johnson has 

 pointed out (zi), is that a dog when 

 coming upon a trail at right angles seems 

 able to detect almost at once the direction 

 taken by the animal or man making the 

 trail. It seems impossible to imagine 

 what the cue can be to such behavior. 



Romanes' dog could apparently dis- 

 criminate by odor his boots from other 

 boots. A more thoroughgoing test of 

 the ability of a dog to select objects be- 

 longing to a specific individual has re- 

 cently been made by Lohner (z8). In 

 this case not trailing, but retrieving was 

 the behavior employed to demonstrate 

 the ability to distinguish between the 

 odor traces left by various individuals. 

 But one subject, a two year old female 

 German Shepherd dog, was used, and in 

 every case the object to be retrieved was 

 one which had been handled by her keeper. 

 The objects in question were small pieces 

 of pine wood. One of these was just held 

 in the keeper's hand, and then placed, 

 with forceps, with ten to twenty other 

 pieces which had not been so handled. 

 The dog was allowed to sniff her keeper's 

 hand and was then told to get his piece of 

 wood. She was successful in every case. 

 An attempt was made to discover to what 

 extent it was necessary to handle the pieces. 



Apparently the shortest duration neces- 

 sary to their identification was z seconds 

 and the smallest surface was that touched 

 by the finger tip. Washing the hands of 

 the keeper in alcohol before touching the 

 wood did not interfere with the identi- 

 fication. Even though four or five pieces 

 of wood handled by others than her 

 keeper were introduced among the dummy 

 pieces, the dog appeared able to select her 

 keeper's piece. At no time was the dog 

 asked to retrieve pieces held by anyone 

 but her keeper. Apparently these ex- 

 periments were conducted extremely care- 

 fully, but, unfortunately, it is not stated 

 whether or not the keeper, or other 

 observers with knowledge of the correct 

 object, were within the visual range of 

 the dog during the selection. If this was 

 the case (and mention would probably 

 have been made of it were it not) the 

 results could scarcely be accepted. 



It may be, however, that a dog is cap- 

 able not only of following human and 

 animal trails but of selecting objects which 

 belong to, or have been handled by its 

 master. Whether this ability to identify 

 the odor traces of an individual person 

 extends to other than the dog's own mas- 

 ter has not been established. Nor has it 

 been proven that, upon being given the 

 hat or some other article belonging to a 

 stranger, the dog can select that stranger 

 from among a group of persons. In spite 

 of the fact that the dog's ability along 

 these lines is not definitely known, evi- 

 dence which depends directly upon such 

 ability is admitted by the courts of many 

 states of this country and also in certain 

 foreign lands. Much more of the validity 

 of such evidence should be known if 

 convictions are to be made on the basis 

 thereof. Realizing this, certain German 

 police officials, notably Police Lieutenant 

 K. Most, with the cooperation of Pro- 

 fessor Pfungst conducted a series of tests a 



