For the series of settings used in connection with problem 1, the reader is referred to page 18.  In the first setting, the doors numbered 1, 2, and 3, were opened.  As it happened, the animal when admitted to the reaction-chamber immediately chose box l.  Having received the reward of food, he was called back to D, and doors 8 and 9 having been raised in preparation for the next trial, he was again admitted to the reaction-chamber.  This time he quickly chose box 9 and was confined therein for thirty seconds.  On being released, he chose after an interval of four minutes, box 8, thus completing the trial.

As it is highly important, not only in connection with the present description of behavior, but also for subsequent comparison of the reactions of different types of organism in this experiment, to present the detailed records for each trial, tables have been constructed which offer in brief space the essential data for every trial in connection with a given problem.

Table 1 contains the results for Skirrl in problem 1.  It is constructed as follows:  the date of a series of trials appears in the first vertical column; the numbers (and number) of the trials for the series or date appear in column 2; the following ten columns present respectively the results of the trials for each of the ten settings.  Each number, in these results, designates a box entered.  At the extreme right of the table are three columns which indicate, first, the number of trials in which the right box was chosen first, column headed R; and second, the number of trials in which at least one incorrect choice occurred, column headed W. In the last column, the daily ratio of these first choices appears.

Taking the first line of table 1 below the explanatory headings, we note on April 19 ten trials, numbered 1 to 10, were given to Skirrl.  In trial 1, with setting 1, he chose correctly the first time, and the record is therefore simply 1.  In trial 2, setting 2, he incorrectly chose box 9, the first time.  At his next opportunity, he chose box 8, which was the right one.  The record therefore reads 9.8.  In trial 3, setting 3, he chose incorrectly twice before finally selecting the right box.  The record reads 6.7.3, and so on throughout the ten trials which constitute a series.  The summary for this series indicates three right and seven wrong first choices, that is, three cases in which the right box was entered first.  The ratio of right to wrong first choices is therefore 1 to 2.33.  Since the total number of doors open in the ten settings is thirty-five, and since in each of the ten settings one door is describable as the right door, the probable ratio, apart from the effects of training, of right to wrong first choices is 1 to 2.50.  It is evident, therefore, that Skirrl in his first series of trials closely approximated expectation in the number of mistakes.

TABLE 1