THE QUARTERLY REVIEW OF BIOLOGY 



incredulous. One by one he checked over 

 the possible secondary criteria which 

 might be giving the animal the clue for 

 its choices. He was using an electric grid 

 to punish wrong choices (in addition to a 

 food reward), and it occurred to him that 

 the animal might in some way be able to 

 detect which of the two grids was elec- 

 trified in a given trial. He therefore ran a 

 control series with both grids electrified, 

 cutting off the current on the side of cor- 

 rect choice only after the choice was made. 

 The habit broke down immediately, the 

 animal now responding quite by chance 

 to the right or left side. It seemed that 

 the difference in potential between the 

 charged electrodes had caused a release of 

 ozone, in minute quantities, but evidently 

 sufficient to be the basis of the discrimi- 

 nating behavior. 



The first question we should try to 

 answer relates to the kinds of odors to 

 which dogs will respond. Binet and 

 Passy 00 claim that dogs respond to 

 animal odors but not to odors arising from 

 vegetable matter. Heitzenroeder (15) 

 using a Spitz obtained evidence leading 

 him to the conclusion that the threshold 

 for odorous mixtures of animal origin was 

 far lower in the dog than in man but 

 that the threshold for odors of plants and 

 perfume was lower in man than in the dog. 

 His method was that of introducing the 

 air carrying the odor by means of a funnel- 

 like arrangement over the dog's nose and 

 taking as indication of sensitivity a 

 sniffing movement or change in the breath- 

 ing rate. That lack of such overt response 

 would indicate lack of sensitivity is to be 

 seriously doubted. Seffrin (39) used 

 Heitzenroeder 's method and reached simi- 

 lar conclusions. Henning (16) recognizes 

 three possible explanations : (1) the olfac- 

 tory continuum shows prominent gaps or 

 breaks; (2.) all odors are sensed, but some 

 types have a disproportionately high 



threshold; (3) such motor responses as 

 sniffing are given, normally, only to 

 biologically important odors. Henning 

 apparently felt that the last possibility 

 was most likely the true one, for in his 

 study he used a different form of response. 

 Henning's principal method involved the 

 use of nine handkerchiefs scattered on the 

 floor of an empty room. One of these was 

 saturated with a definite odor. In an 

 adjoining room the dog was given the 

 opportunity of smelling the odor in 

 question and was then sent into the test 

 room to retrieve the scented handkerchief. 

 The position of this handkerchief was 

 varied from trial to trial. A Dobermann 

 and a fox terrier are said to have retrieved 

 without error the correct handkerchief 

 when the following odors were used in 

 intensities equalling the human threshold 

 (sub jectively determined) : vanillin, helio- 

 tropin, cumarin, oil of rose, oil of geran- 

 ium, jasmine, oil of lemon, peppermint, 

 wintergreen, and others. A greyhound 

 could not be trained to do this. This test 

 gave no evidence of discriminatory ability 

 but only of sensitivity, since only one odor 

 was used at a time. Further tests of a 

 rather informal nature indicated that the 

 dogs could discriminate between food odors 

 and flower odors. 



Buytendijk (4) presents data not only 

 on the lower threshold but on discrimina- 

 tory ability. He employed two methods, 

 principally. In the first of these, two 

 boxes were used, exactly similar and each 

 containing a biscuit. The lid of one was 

 fastened shut, while that of the other 

 could be opened. A small vessel contain- 

 ing a chemical, diluted in paraffin or 

 water, stood before each box, and it was 

 supposedly only by detecting the odors 

 that the dog could tell which box was 

 unlocked. Buytendijk apparently did not 

 take care to eliminate the experimenter 

 from the situation, and we cannot be sure 



