108 JOSEPH U. YARBROUGH 



with each animal on each of the three boxes. Now one box 

 was chosen first and now another. For convenience these data 

 will be recorded in two separate tables (X and XI), the first 

 containing data recorded on the three box experiments, and the 

 second, those obtained when only two boxes were used. 



TABLE X 



Three Box Experiments Total 



reactions 



Order of response abc acb ab ac made 



Jim 7 3 22 1 33 



Bess 7 1 22 4 34 



Order of response bac bca ba be 



Jim 10 1 2 



Bess 2 7 3 12 



Order of response cab cba ca cb 



Jim 8 6 3 23 40 



Bess 3 4 15 22 



Table X analyzes all incorrect responses made on the three 

 box experiments, here included, and gives the relative number 

 of times each subject followed the different possible orders. The 

 number of errors made beginning with boxes a and c is about 

 equal with both animals, while the number beginning with b 

 is very low. In fact, as the table shows, Jim only made 2 errors 

 when b was selected first. Bess, however, made 12 such errors, 

 or about one-half the number she made after selecting c first. 

 Of the 40 times Jim selected c first, he selected b next 29, or 

 72% of the time. And of the 33 times he selected a first, 29, 

 or 87% of the time b was the next box selected. When a was 

 selected first by Bess, she chose b next 29 times out of 34, or 

 85% of the time. And when she selected c first, she chose b 

 next 18 times out of 22, or 81% of the time. It may be con- 

 cluded then, that whenever the reaction began at the end of 

 the row of boxes, i.e., a or c, the tendency was to take the boxes 

 in order until the solution was reached. Only three times in 

 320 trials did Jim go to the same box twice in the same trial. 

 (These are listed in table XII as ' persistent errors.") The 

 form of this position habit was c b c b a, and was repeated three 

 times within 20 trials. Bess returned to the same box in the 

 same trial only one time, and the order of the boxes chosen 

 was babe. One further thing to be noted is that Jim made 

 25 " three place errors," responses where the animal tests each 



