If we would set this spinning top into motion, it would not fall, even if its axis would not be oriented perpendicular to the floor. Instead, its axis would change its orientation slowly. The spinning motion seems to stabilize the gyroscope, just as the moving bicycle is sort of stabilized by its turning wheels. This sounds simple and familiar, but can this really be grasped by intuition immediately?

I do not think so – otherwise it would not have taken us 2000 years to get over Aristotle’s assumptions on motion and rest. And simple experiments demonstrated in science shows would not baffle us – such as the motion of a helium balloon in an accelerating car.

The standard text-book explanation goes like this: There is gravity, as we assume that the spinning top is not supported in its center of gravity. Thus there is a torque. The gyroscope is whirling, thus it has angular momentum. A torque corresponds to a change in angular momentum, analogous to a force resulting in a change of (linear) momentum. The torque vector is perpendicular to gravity and to the axis of the gyroscope. Thus the change in angular momentum is always perpendicular to the current angular momentum vector and the tip of the spinning top moves in a circle. The angular momentum vectors changes all the time – not in length, but in direction – which is called precession.

As Richard Feynman pointed out in his Physics Lectures, this explanation constitutes rather mathematical step-by-step instructions than a real explanation. We do not see immediately why the spinning top precesses instead of falling to the ground.

Our skepticism is justified: The text-book explanation does not fully expound the dynamics of the systems and explain what really happens – in the very moment the spinning top starts to move. It rather refers to a self-consistent solution: If the gyroscope would already precess in a circle, that circular movement is consistent with the torque. As everybody in his right mind (R. Feynman) would assume, it actually might fall a bit if it is released.

Generally, the tip of the gyroscope keeps tracing out a wavy or loopy path, which is called nutation.

If the spinning top nutates / starts falling, it looses potential energy. This has to compensated by an increase in rotational energy, the velocity of the tip of the gyroscope is not a constant. (Note that the total angular momentum of the gyroscope is composed of contributions from the fast spinning motion and the slow precession). The tip of the gyroscope moves on a curved trajectory bending upwards, which finally leads to overshooting the average height.

Friction can make the wobbling decay and finally turn the trajectory into the simple-text-book-path. This simulation allows for turning on friction (which is also equivalent to Feynman’s explanation).

An excellent explanation can be found in this remarkable paper (related to the simulation): The gyroscope is set into rotational motion while still supported. When “gravity is suddenly turned on” by removing the support, the additional vertical component of the angular momentum – due to to precession – is suddenly turned on. The point is that the initial angular momentum is parallel to the symmetry axis of the gyroscope, and the axis starts from velocity zero. The total angular momentum – still parallel to the symmetry axis – is the sum of the one related to precession and the one related to the gyroscope’s fast movement. So the latter is not parallel to the axis any more: The tip of the axis starts tracing out the loopy path (nutation) when it precesses. Only if we tune the angular frequency carefully before we release the spinning top, the text-book solution can be obtained. In this case precession is really maintained by the torque.

So do we understand the gyroscope intuitively now? A deep understanding of angular momentum and torque is a pre-requisite in my point of view. On principle, all of classical mechanics can be derived from Newton’s laws, so the notions of force and momentum should be sufficient. Nevertheless, without introducing angular momentum, there is no way to explain the motion of the gyroscope briefly.

Why do we need “torque” in general? Such concepts are shortcuts that allow for a concise description, but they also reveal the underlying symmetry or essential aspects of a problem. You could describe the dynamics of a rigid body by considering the motion of all little pieces the body is composed of. But since it is rigid, actually two points would be sufficient. You can select any two points or basically any set of independent coordinates – 6 independent numbers.

The preferred choice is: 3 numbers – such as Cartesian co-ordinates, x,y,z – describing the motion of the center of gravity and 3 numbers describing the rotation of the body. You need two numbers to denote the direction of the axis about which to rotate (similar to two longitude and latitude to describe a point on a sphere), and one number to denote the angle – how much you rotate. You could also describe any rotation in terms of the components discussed for the gyroscope: precession, nutation and internal rotation.

Then Newton’s equation of motion for the rigid body can be re-written as a law of motion for the center of gravity (Force equal change of momentum of the center of gravity) and a law for two new properties of the system: the torque equals the change of the angular momentum. Actually, this equation defines what these properties really are. Checking the definitions that have evolved from the law of motion we conclude that the angular momentum is linear momentum times the lever arm, and the torque is force times lever arm. But these definitions as such would not make sense if they would not have been generated by the reformulation of Newton’s law.

I think we sometimes adopt or memorize definitions carelessly and consider this learning because these definitions are required by standards / semi-legal requirements and used within a specific community of experts. But there is no shortcut and no replacement of understanding by rote learning.

I believe you need to keep the whole entangled web of relations between fundamental laws in mind, but it is hard to restrict the scope. We could now advance from gyroscope and angular momentum to the deeper connections between symmetries and conservation laws. In order not get stuck in these philosophical musings all the time – and do something useful (e.g. as an engineer), you need to be able to switch to shut-up-and-calculate-mode (‘Shut up and calculate’ is often attributed to Richard Feynman, but I could not find an authoritative confirmation).