58 HARRY MILES JOHNSON 



greatly disturbed by this procedure ; some apparently beeame 

 frantic; the movements of others were inhibited probably 

 from "fright." Still others, more quiet, became satisfied after 

 making a few unsuccessful attempts to escape, and remained 

 quiet for a long time. The defects n this method are obvious, 

 and the advantages of the one adopted for this work will become 

 more apparent when the animals' behavior is discussed. 



After the animal had been fed from the open box for a w^eek 

 or more, one day's work was done with the door of the problem- 

 box wedged open by only about two inches from the jamb. 

 The dog had to force it open in some way in order to get the 

 food. This was found necessary in the preliminary experiments 

 because the animals would not try to open the box when it 

 was latched. After sniffing and scratching lightly about the 

 corner they would lie down for hours. This behavior lasted 

 through three successive days and the experimenter believed 

 that prolonging the animals' hunger could produce only harm- 

 ful results. After one day of feeding from the partly closed 

 box every animal cjuickly attacked the problem of opening the 

 latched door. 



Time was taken at each trial after the box was first latched. 

 The record was of the number of seconds required for the animal 

 to obtain the food after it had crossed the dotted line from the 

 glass door to the corner of the problem-box. This point was 

 chosen rather than the wooden door, because Dogs i and 2 

 were much hea\-ier and slower of mo\'ement than any of the 

 other dogs, and required one to two seconds more time to reach 

 the problem-box from the wooden door, although their move- 

 ments in opening the boxes w^ere as swift and as accurate as 

 were those of any of the other dogs. There was one disad- 

 vantage, however, in that it was difficult to tell within one-half 

 second or so when the dog had crossed this line if the work 

 was done in total darkness. Time was taken with an ordinary 

 stop-watch. This records to 0.2 seconds. Fractional readings 

 less than 0.5 seconds were disregarded; those greater than 

 0.5 second were counted one second. Some experimenters 

 attempt to take time -readings in this way as closely as 0.2 

 seconds, and consider the fractions at full value in computing 

 their results. In such work as this, however, the writer be- 

 lieves such a record shows an apparent accuracy which is not 

 real. The experimenter's reaction-time will show considerable 



