HABIT FORMATION IN THE DOG 35 



ratus was moved into another room, and the experimenter re- 

 mained entirely outside the animal-room until after the animal 

 had reacted and been admitted to the proper food-box. The 

 animal was observed through the center window shown in Fig. i. 



Besides the elimination of the secondary criteria mentioned 

 above, it is also desirable to eliminate another possible source 

 of disturbance, namely the lapse of attention which may ensue 

 if the animal's reaction is delayed. This may be accomplished 

 by giving the stimulus very shortly — sa}' within a second, 

 before the animal's choice is to be made. The optimal time 

 may vary with different animals, but it should be possible to 

 control it. In using the stimulus-cage described above the 

 stimulus may be presented at any time desired before the ani- 

 mal reaches the turn into alleys D and D'. 



The criterion of discrimination should not be less than a 

 perfect record, maintained through at least three days. The 

 arbitrary standard of 95 perfect trials in the last 100, which 

 some investigatoi-s of other kinds of discriinination have 

 adopted, may be allowable ; but certainly a poorer record 

 will not suffice. The higher standard is preferable to this. 



The final condition which must be fulfilled is the control- 

 lability of the stimulus. It is well kno^yn that musical tones 

 differ not only in pitch, l)ut also in intensit}', timljre, non-musical 

 concomitant noise, and perhaps in localization. These other 

 characteristics must be controlled if we are to make the assump- 

 tion legitimately that the subject is discriminating on the basis 

 of pitch alone. In addition to this desideratum, it is also desir- 

 able that the conditions under which one experimenter works 

 may be reproducible by another, that results may be comparable. 



At the present time no satisfactory means is known of 

 measuring the intensity of sound. It may be A'aried quite 

 widely, however, in electrically actuated tuning forks and in 

 l:)lown pipes and bottle whistles ; the latter vary in pitch with 

 the quantity of air admitted. The method of varying the inten- 

 sity of the stimulus in forks will be described later in this section. 



The best means of controlling timbre is by the use of instru- 

 ments which will give tones as nearly pure as possible. Reed 

 instruments cannot be controlled in this respect. Tuning forks 

 weakly actuated give tones as nearl}- pure as can be had, al- 

 though even in these a high anharmonic partial can usually 



