Closely related studies, finally, deal with a mistake which enters when the movement is reproduced from memory after a certain time. The exactitude of a simple arm movement seems to increase in the first ten seconds, then rapidly to decrease. The emotional attitude, too, is of importance for the reproduction of a movement. I trained myself in making definite extensor and flexor movements of the arm until I was able to reproduce them under normal conditions with great exactitude. In experiments extending over many months, which were carried on through the changing emotional attitudes of daily life, the exact measurement showed that both groups of movements became too large in states of excitement and too small in states of fatigue. But in a state of satisfaction and joy the extensor movement became too large, the flexor movement too small, and _vice versa_, in unpleasant emotional states the flexor movement was too strong and the extensor movement too weak.[34] We have a very careful investigation into the relations between rapidity of movement and exactitude.[35] The subjects had to perform a hand movement simultaneously with the beat of a metronome, the beats of which varied between 20 and 200 in the minute. In general the accuracy of the movement decreases as the rapidity increases, but the descent is not uniform. Motions in the rhythm of 40 to the minute were on the whole just as exact as those in the rhythm of 20, and, on the other hand movements in the rhythm of 200 almost as accurate as those of 140 to the minute. Thus we have a lower limit below which decrease of rapidity does not increase the accuracy any further, and an upper limit beyond which a further increase of rapidity brings no additional deterioration. The mistakes of the unskilled left hand increase still more rapidly than the number of movements. If the eyes are closed, the rapid movements are usually too long and the slow ones too short.