a trickster in his old age, all for the love of money? The unreliability of Mr. von Osten's signs is good proof of their involuntary nature. Anyone who had seen him work with the horse could not have helped noticing that he certainly did not have complete control over the animal, and was not able, at a given moment, to make Hans perform a certain feat, as would have been the case if the process had been one of "training". Again and again Hans failed to make the right count. Before a large audience, one time, it took four tests to get him to tap properly up to 20, and in all four I could note clearly that it was Mr. von Osten who, by his involuntary premature movements, was the innocent cause of the failure. On another occasion, after Hans had done some beautiful work in fractions, in the presence of a large number of spectators, the master asked him the simple question: "Where is the numerator in a fraction?"--The answer was first: "to the left", and then, after a severe reprimand: "down" (below), and finally: "up" (above). He often made just such incorrect movements of the head. In the color-selecting tests the average of error was quite unpredictable. With an equal number of tests, on one day, half would be successful, on another, four fifths, on a third, one-tenth. Often Hans appeared to be "indisposed" for days at a time. The color tests would often end in expressions of rage on the part of Mr. von Osten and in consequence Hans would become startled and would then storm about the courtyard so that it was dangerous to try to approach him. Some may object that all this was mere comedy and that possibly Mr. von Osten prevented some of the tests from turning out successfully. But this objection is to be met by the statement that very often failure would occur just when it was particularly desirable to have the tests appear in a favorable light before a large and enthusiastic assemblage of visitors. After such failures he would be downcast on account of Hans's contrariness. It is also significant that Mr. von Osten's percentage of error, corresponds very closely with my percentage of error in the "non-voluntary" tests, (page 84f.), whereas he never was able to obtain the errorless results which I obtained in my "voluntary" experiments.