SIZE AND FORM PERCEPTION 89 



after which the diameter of the variable was changed by incre- 

 ments of i mm. The perfect responses continued almost as 

 regularly as the variable was increased. Finally the behavior 

 of the chick indicated that it was discriminating between 

 o 28+ and o 27+ (see record of November 23 and 24). 



A o 28+ — g 27+ discrimination was incredible, for the human 

 eye could scarcely detect a difference between a 28+ sq. cm. 

 circle and a 23+ sq. cm. circle. The only thing to do was to 

 test the chick on o 28+ — o 28+ discrimination. The result was 

 a perfect series. (See record 1, November 25.) Then I began 

 to look for the cue on which the chick relied. At last I noticed 

 a small crack where the outside extension of the electric box 

 was joined to the experiment box. A similar crack appeared 

 on both sides in corresponding positions, hence it seemed that 

 in this factor there could be no clue by means of which the 

 chick was guided in its choice of the right compartment. 



Close inspection, however, proved that in these two minute 

 cracks lay the cue by means of which the chick had been dis- 

 criminating. Where the rabbeted edges of the shifter and 

 tracks rubbed, there was a bright edge, and wherever the shifter 

 rested, this bright surface was covered. Thus when the shifter 

 was at the left, the right end of the track was uncovered, and 

 from this uncovered part was reflected some of the light rays 

 from the upper illumination. A small portion of this reflected 

 light could enter the crack on the left side. Now, when the 

 standard circle was presented at the left, it happened that the 

 shifter was moved to the right ; when the standard appeared at 

 the right, the shifter stood at the left. As a result, the crack 

 that was illuminated was always the one where the standard 

 circle was displayed; the other crack, being reached by no 

 reflected light, was always dark. 



Noticing this slight variation in the condition of illumination 

 in the two compartments, and desiring to ascertain by what 

 clue the chick was choosing, I closed the small cracks through 

 which the light was reflected. The effect of this change appears 

 in the. records following the first on November 25. The perfect 

 reactions abruptly stopped with the closing of the cracks; out 

 of 15 tests there were only six correct responses. As appears 

 in the subsequent part of table 2, I continued the experiment, 

 with the cracks stopped, by reducing the area of the variable to 



