198 CHARLES A. COBURN 



crows' intelligence. A new form of reaction now developed. 

 When either of the crows had made an incorrect choice and the 

 exit door was opened, showing a dark exit-box, instead of enter- 

 ing as they hitherto had done, they would whirl about and 

 quickly go to the other exit and there wait, even for five or ten 

 minutes, until the door was opened. This behavior naturally 

 tended to lower the percentage of correct choices. 



The experimenter first tried to overcome this difficulty by 

 having the exit-box illuminated until they had entered it. Crow 

 no. 2 would always enter the box under these conditions, but 

 crow no. 1, after a few trials, refused to enter either box unless 

 there was a bit of food in view. 



To meet this difficulty, the gates, described on page 188, were 

 constructed. When the crow entered the wrong stimulus- 

 chamber, the exit door was opened and at the same moment 

 the gate between that chamber and the discrimination-chamber 

 was dropped, thus preventing the crow from escaping to the 

 other exit. The dropping of the gate tended to frighten them 

 somewhat, so they always quickly entered the exit-box, which 

 was again darkened as in the early experiments. The effect 

 of this improvement in the apparatus on the behavior of the 

 crows appears in the results of Table 4. 



The crows had been given one hundred and six tests for their 

 ability to distinguish a 6 centimeter circle from an 8.081 centi- 

 meter triangle. During these trials no appreciable increase in 

 the percentage of correct choices had been made. Immediately 

 after the gates were brought into use, improvement commenced 

 and thereafter the majority of the choices were correct. Crow 

 no. 2 did not make quite as high a percentage of correct reac- 

 tions as did crow no. 1. This was probably because no. 2 seemed 

 to be more frightened by the dropping of the gate. If an incor- 

 rect choice was made early in a series, there was a tendency, 

 on the part of no. 2, to avoid that stimulus-chamber during the 

 remainder of that series. 



The 6 centimeter circle, the 8.081 centimeter triangle, the 

 5.317 centimeter square, and the 3.29 centimeter hexagon are 

 of equal area. The last thirty tests were with figures unequal 

 in size. The 6 centimeter and the 9 centimeter circles each 

 possess a greater area than the 3 centimeter hexagon, whose area, 

 in turn, is almost twice as great as that of the 3 centimeter circle. 



