10 



STELLA B. VINCENT 



an average of 14.9 errors the first trial while those on the black- 

 white maze made only half as many, 7.5. 



Notice the difference in the initial height of the error curves 

 of the two mazes (Figs. 2 and 3). The error curve of the black- 

 white maze is not like any error curve made for a normal maze 

 and it is a direct expression of the accuracy. The time curve 

 which accompanies it, as has been said, is low because of the 

 fewer errors and not because of more rapid speed. 



There is a great decrease in total as well as in beginning errors 

 from the mark set by the normal maze. The latter maze has to 

 its credit an average of 66.6 errors per animal or 1.48 per trial 

 for each rat while the black-white maze gives only 35.6 errors 



Fig. 6. Graph showing the point at which the animals made their first trial with- 

 out error. Each vertical bar represents an animal.' The black represents 

 those of the black-white maze. This is superimposed upon the white, which 

 represents the animals in the normal maze. 



per animal or .8 per trial. The accuracy is nearly twice as great 

 in the black-white maze. 



' One would naturally expect then, what really is the case, 

 that the error curve for the black-white maze (Fig. 3) would 

 reach its lowest level much sooner than the error curve for the 

 other maze — that the automatism would be more quickly estab- 

 lished. The figures show that the rats on the black- white maze 

 made their average first trip without error on the 4.2 trial but 

 the average for this trip on the normal maze was 8.5. If we 

 take the first ten trials and compare them we find that the rats 

 in the black- white maze have an average of 4.2 perfect trips to 

 their credit while the normal maze has only half that number, 2. 

 The graph seen in Fig. 6 is made from the records of 20 animals 



