26 D. B. CASTEEL 



vary enormously, often independently of the failure or success 

 of its choices. It is true that first attempts are usually deliberate 

 and that imsuccessful turtles may be rendered inert through 

 continued punishment, but aside from these considerations the 

 above statement is correct. Without attempting to make too 

 detailed a classification, it may be said that some of the follow- 

 ing factors determine the time consumed by a turtle in making 

 its trips from entrance room to box: 



(i) temperament of the individual, which would remain 

 fairly constant but would differ greatly if different 

 individuals were compared; 



(2) physiological state, varying for the individual during 



a series of tests and dependent on general health, 

 unavoidable fluctuations in temperature and amount 

 of food taken, though efforts were made to equalize 

 feedings ; 



(3) temporary mental state, varying for the individual 



during a series of tests, as sulkiness and inertness 

 caused by an unfavorable physiological state or by 

 undue sensitiveness to punishment. 



For the above reasons the time records are deemed of little 

 value and are not given in the published tables though they 

 were taken in all the experiments. 



SUMMARY AND CONCLUSIONS 



Turtles of the species Chrysemys marginata were tested for 

 their ability to discriminate between, (a) black and white; 

 (b)' two patterns of different shape; (c) two series of parallel 

 lines of equal width but running in different directions; (d) two 

 series of parallel lines of different width running in the same 

 direction. 



Associations were established by the use of food and electric 

 boxes except in two experiments where the electric shock was 

 not used as punishment for failure. The apparatus employed 

 was similar in construction to the electric-box apparatus of 

 Yerkes, though considerably modified for adaptation to the 

 habits of turtles. 



Negative results only were obtained in the experiments for 

 pattern discrimination. It is felt that these results are not 



