But now, curiously enough, if you were to apply a sidewise pressure at _A_, pushing to the left (as we view the diagram) to help on the motion of precession, the obstinate apparatus will cease altogether to move in that direction and the point _A_ will begin to rise instead, the frame _B A C_ rotating on its axis _B C_. This rise of the axis _O A_ will take place even though the downward pressure is continued. You have disturbed the equilibrium of the top--unbalanced it--and it must seek a new position. Contrariwise, if you would have the point A moved to the right, you must push it upward; if you would have it go down, you must push it to the right. This seems rather weird behavior, but if you will note the direction of the arrow on the wheel you will see a certain method in it. It will appear that in each case the force you apply has been carried round a corner, as it were, by the whirling disc, and made to act at right angles to the direction of its application. This change of direction of a force applied is strictly comparable to the change effected by the familiar device known as a pulley. With that device, to be sure, a pull instead of a push is used, but this is a distinction without a difference, for pushing and pulling are only opposite views of the same thing. Possibly this suggested explanation of the action of the gyrostat may not seem very satisfactory, but the facts are perfectly clear, and if you will bear them steadily in mind you will readily be able to understand the Brennan gyroscope, as you otherwise cannot possibly hope to do. You have only to recall that pushing down at _A_ causes motion (called "precession") to the left, and pushing up at _A_, motion to the right; and that in order to make _A_ either rise or fall, you must "accelerate precession" by pushing to the left or to the right, respectively. But you must understand further, that when, through the application of any of these disturbing forces, you have forced the axis _O A_ into a new position, it will tend to maintain that new position, having no propensity whatever to return to its original position. It is quite as stably in equilibrium with its axis pointing upward as when in the position shown in the diagram. One position is quite like another to it; but having accepted a position it resents any change whatsoever. Now we are prepared to understand the Brennan gyroscope, which consists essentially of two such gyrostats as that shown in our diagram _A_, set into the frame of the car on the axis _D E_, their wheels revolving in opposite directions and their outer frames so linked together that when one turns in one direction on its axis _D E_, the other must turn in the opposite direction. As the sole object of having two of the gyroscopes is to facilitate the going around curves, we may for the moment neglect the second one, and consider the action of only one of the pair.