Besides the plaintiff, only the head nurse knew the exact particulars of the deed, for as soon as the former missed her money she immediately asked the head nurse to help her find it, thus the head nurse had been able to learn the smallest details, which naturally rendered the experiment still more difficult, for she was precisely the one most suspected. The conditions for the experiment were better for the others, since they knew nothing concerning the particulars of the deed, and some not even that a theft had been committed. As critical stimulus words I selected the name of the robbed nurse, plus the following words: cupboard, door, open, key, yesterday, banknote, gold, 70, 50, 20, money, watch, pocket-book, chain, silver, to hide, fur, dark reddish, leather, centimes, stencil, receipt, Dosenbach. Besides these words which referred directly to the deed, I took also the following, which had a special effective value: theft, to take, to steal, suspicion, blame, court, police, to lie, to fear, to discover, to arrest, innocent.
The objection is often made to the last species of words that they may produce a strong affective resentment even in innocent persons, and for that reason one cannot attribute to them any comparative value. Nevertheless, it may always be questioned whether the affective resentment of an innocent person will have the same effect on the association as that of a guilty one, and that question can only be authoritatively answered by experience. Until the contrary is demonstrated, I maintain that words of the above-mentioned type may profitably be used.
I distributed these critical words among twice as many indifferent stimulus words in such a manner that each critical word was followed by two indifferent ones. As a rule it is well to follow up the critical words by indifferent words in order that the action of the first may be clearly distinguished. But one may also follow up one critical word by another, especially if one wishes to bring into relief the action of the second. Thus I placed together "darkish red" and "leather," and "chain" and "silver."
After this preparatory work I undertook the experiment with the three above-mentioned nurses. Following the order of the experiment, I shall denote the friend of the head nurse by the letter A, the head nurse by B, and the nurse who attended to the cleaning of the room by C. As examinations of this kind can be rendered into a foreign tongue only with the greatest difficulty, I will content myself with presenting the general results, and with giving some examples. I first undertook the experiment with A, and judging by the circumstances she appeared only slightly moved. B was next examined; she showed marked excitement, her pulse being 120 per minute immediately after the experiment. The last to be examined was C. She was the most tranquil of the three; she displayed but little embarrassment, and only in the course of the experiment did it occur to her that she was suspected of stealing, a fact which manifestly disturbed her towards the end of the experiment.
The general impression from the examination spoke strongly against the head nurse B. It seemed to me that she evinced a very "suspicious," or I might almost say, "impudent" countenance. With the definite idea of finding in her the guilty one I set about adding up the results. You will see that I was wrong in my surmise and that the test proved my error.
One can make use of many special methods of computing, but they are not all equally good and equally exact. (One must always resort to calculation, as appearances are enormously deceptive.) The method which is most to be recommended is that of the probable average of the reaction time. It shows at a glance the difficulties which the person in the experiment had to overcome in the reaction.
The technique of this calculation is very simple. The probable average is the middle number of the various reaction times arranged in a series. The reaction times are, for example,[139] placed in the following manner: 5, 5, 5, 7, 7, 7, 7, 8, 9, 9, 9, 12, 13, 14. The number found in the middle (8) is the probable average of this series.
The probable averages of the reaction are: