If you have taken the trouble to follow carefully the description just given, you will be prepared to understand the anomalies of action of the gyrocar; for example, why its side rises when a weight is placed on it; why it leans toward the wind, and why it leans to the inner and not to the outer side of the track in rounding a curve. The substance of the explanation is that the greater the force brought to bear on the roller R1 by the segment of the car that strikes against it, the stronger its precession, and hence the more powerful its lift. The oscillations and counter-oscillations thus brought about continue to operate powerfully on the roller R1 so long as the weight of the car is out of balance; and balance is restored only when the heavier side of the car rises, bringing the centre of gravity over the track, just as a man carrying a weight on the right shoulder leans toward the left, and vice-versa. Thus, when the gyrocar has a heavy weight on one side, or encounters a strong wind, it may lean far over, but still be perfectly and securely balanced, the gyroscopes finally remaining quiescent in their new position until some new disturbance is applied. It remains to be said, however, that there is another element introduced when the car rounds a curve. To understand this, we must revert to the action of the Foucault gyrostat, as illustrated in diagram 1. If you held such a gyrostat in your hand in the upright position in which it is shown in the diagram, and whirled it about, the axis _O A_ would of course maintain a fixed direction so long as the gyrostat was free to revolve on the axis _D E_. But if you prevented such revolution, as by clutching the spindle _E_ firmly, and then whirled the gyrostat about at arm's length, the axis _O A_ would at once be forced to take an upright position. If your hand whirled to the right, the point _A_ would rise; if your hand whirled to the left, the point _A_ would go down; the principle determining this motion in either case being that the direction of whirl of the gyroscope must correspond to the direction of curve given to the apparatus as a whole by the motion of your arm.