64 



WALTER S. HUNTER AND JOS. U. YARBROUGH 



60-day test. These rats had received 300 trials in the 30-day 

 test followed by some intermediate training on h. c. If this 

 data were included in the curves, there would be no variation 

 in their essential relations. If anything the interference would 

 be more apparent. 



id 



* 3 f r <> 7 Z ? 



Figure 4. — Relative rates of error elimination in h. c. and in B. 



on rats 15, 23 and 7 



S 



Based 



The curves in figure 4 begin at essentially the same height 

 and go along together throughout the first six-tenths of the 

 learning. It is during the last four-tenths of the curves that 

 the B -curve remains markedly below that for h. c. (There is 

 no evidence that this was caused by fear.) The interference 

 of the two habits is seen here and in a comparison of the last 

 of h. c. and the first of B. In the average B is no more than 

 a new habit with these rats. Its curve begins no lower than 

 that for h. c. The details are further brought out in table 9, 

 which gives the correct responses in each 10 trials of the first 



