HABIT FORMATION IN THE DOG 29 



left in going to the food-box, the reaction was nearly always 

 correct ; but if she started by a middle path toward the boxes, 

 which were near together, the reaction was incorrect as often, 

 at least, as connect. It was usually possible for the experimenter 

 to predict from the animal's breathing when she was about to 

 become indifferent. This stage was observed sometimes in the 

 beginning of a series, sometimes onl}^ about the middle, and 

 sometimes near the end. It could hardly be interpreted there- 

 fore, as fatigue, especially since the animal would always 

 make some choice very quickly. The "delayed reaction" factor 



