110 VINNIE C. HICKS AND H. A. CARR 



tion and the rational ability of the organism. This does not 

 necessarily mean that no causal relation obtains between rational 

 ability and the initial progress in learning. If any causal rela- 

 tion obtains, it is indirect and obscured by other factors. One 

 can merely conclude from these facts that one can not infer 

 the relative intelligent ability of any two groups of organisms 

 from this particular characteristic of their learning curves. 



3. During the rapid initial spurt, the rate of elimination is 

 not uniform. During some trials the elimination is more rapid 

 than at other times. Certain differences between the groups 

 exist in this respect. In every case the most rapid elimination 

 occurs for the children during the 1—3 trials. For the adults 

 this period occurs later in the learning process, between the 

 3-5 or 5-7 trials. The rats again occupy an intermediate posi- 

 tion, though the curve is more similar in this respect to that 

 of the adults than to that of the children. The most rapid 

 elimination occurs generally from the 3-5 trials, though some- 

 times during the second and third trials. Again no correlation 

 obtains between the phenomenon and the relative intelligence 

 of the groups. 



4. The children's curves invariably exhibit some indication 

 of the plateau phenomenon from the third to the seventh trials. 

 Some slight indication of plateaus are exhibited by the adults, 

 but only for the error and distance values. The rats' curves 

 give the least indication of this characteristic. 



IV. COMPARISON OF TIME, ERROR AND DISTANCE ELIMINATION 



The relative rate of elimination of time, error, and distance 

 for any group might be shown by plotting the percentage values 

 of Table IV. The facts can better be exhibited by the following 

 plan. In comparing time and error, for example, the number 

 of errors for each trial is divided by the number of minutes for 

 the corresponding trial. This gives a series of values representing 

 the number of errors made per unit of time for the successive 

 trials. The ratios between these values are determined with 

 the initial value regarded as unity, and these ratios are plotted 

 in a curve. Any rise in this curve means that for that period 

 the time values have been eliminated faster than those for 

 error. A fall of the curve means that the errors have been 

 decreased more rapidly than the time. The degree of differ- 



