given opportunity to select the right box without aid from the experimenter.  But as a matter of fact, while the ratio of right to wrong first choices was 1 to .67 in the series preceding this change of method, it was 1 to 1.50 in the first series following its use.  There is no satisfactory evidence that Julius profited by this experience, though as a matter of fact he did succeed in making his best daily record, eight right to two wrong choices, on August 4, after 1190 trials.

The curve of learning for this problem has been plotted and is presented in figure 19.  It is of course incomplete and it is offered only to indicate the extreme irregularity in performance.

Problem 1a.  First at the Right End

It was decided on August 19 that the further continuation of the work of Julius on problem 2 was not worth while.  He had become much discouraged, and although willing to work for food, gave no indications whatever of improvement and seemed to have exhausted his methods.  It seemed wise instead of giving up work with him in the multiple-choice method to return to a form of problem 1.  We may designate it as problem 1a.  The right box is definable as the first at the right end of the series instead of the first at the left end as in the original problem 1.  It was thought possible that Julius might quickly solve this problem by a process similar to that used for problem 1.

Work was begun on problem 1a, August 20, and for six successive days two series of trials per day were given, the settings for which as well as the resulting choices are given in table 10.  Most notable in these results is the large number of cases in which Julius chose first the second box from the right end of the series, or in other words that box which had been the right one in problem 2.  Contrary to expectation, he showed no inclination to abandon this tendency to choose the second from the right end, and the ratio of right to wrong choices changed in the direction opposite from expectation, beginning with 1 to 4 and ending on the sixth day with 0 to 20.

It was obviously useless to continue the experiment further since Julius had given up his attempts to locate the right box in the first choice and was apparently satisfied to discover it by a process of trial and error.  He had, it would seem, satisfied himself that the problem was insoluble.  These results obtained in problem 1a constitute a most interesting comment on the effects of problem 2 on the orang utan.  Behavior similar to that which he developed well might have been obtained from a child of three to four years placed in a like situation and forced to strive, day after day, to solve a problem beyond its ideational capacity.

In many respects the most interesting and to the experimenter the most surprising result of this long series of observations with Julius was the lack of consistent improvement.  It seemed almost incredible that he should continue, day after day, to make incorrect choices in a particular setting while choosing correctly in some other setting which from the standpoint of the experimenter was not more difficult.