Oscillation and restoring forces

Observation

A plucked ruler, a child on a swing, a mass bobbing on a spring, and a pendulum released from the side all share a behavior that is easy to watch and less easy to explain. Displaced from rest, each one comes back, then keeps going past rest, out the other side, and back again, many times, before it finally settles. Being pushed away from equilibrium does not merely pull the system home. It pulls it home with enough vigor to carry it clean through and out the far side.

The question this entry answers: what property of a force makes a system overshoot and repeat, rather than drift back once and stop?

Variables

Fix what changes and what does not. The displacement x is the distance from the equilibrium position, the place where the net force is zero; it is the measured quantity and it changes sign as the system passes through the middle. The restoring force F is what the system exerts to return, and observation shows it is not constant: it grows as x grows and always points opposite to x. The mass m resists acceleration. The stiffness, written k, is the controlled property of the particular spring, string, or pendulum length, fixed for a given setup. Time t is the independent variable. What stays invariant across a cycle, once energy loss is set aside, is the total mechanical energy.

Model

The simplest relationship consistent with "grows with x" and "points opposite to x" is direct proportionality:

F = -kx

The constant k is positive, and the negative sign carries the entire physical claim: whichever way x points, F points the other way. This is the linear restoring law. It is not a coincidence that a spring and a pendulum, physically unrelated, obey the same form, and the next section shows why the linear case is the generic one rather than a special one.

Derivation

First, why the linear law is generic. Any smooth restoring force about a stable equilibrium can be written as its Taylor expansion in x. At equilibrium the force is zero, so the constant term vanishes; the first surviving term is proportional to x, and for small enough displacement it dominates every higher term. So F = -kx is not a property peculiar to ideal springs. It is the leading behavior of almost any restoring force near its resting point, which is why oscillation is everywhere.

Combine the linear law with Newton's second law, F = ma:

ma = -kx, so a = -(k/m)x

Acceleration is proportional to displacement and opposite in sign. That single statement is the signature of the whole motion class. The full solution of this equation of motion, treated as a second-order differential equation, is developed in the entry on second-order dynamics; here the physics matters more than the machinery, so take the solution and verify it:

x(t) = A cos(omega t + phi), with omega = sqrt(k/m)

Differentiate twice and the result is -(k/m) times x, which is the equation of motion, so the proposed solution holds. A is the amplitude, phi marks where in the cycle the clock started, and omega is the angular frequency. The period, the time for one full repetition, is T = 2 pi / omega = 2 pi sqrt(m/k).

Now the physical account of why it repeats. At maximum displacement the restoring force is largest but the velocity is zero: the system is turning around. Accelerating back toward equilibrium it gains speed, so it arrives at the middle, where the force is zero, moving at its fastest, with nothing acting to hold it there. Inertia forbids it from stopping at equilibrium; it overshoots. On the far side the displacement reverses sign, so the force reverses too and decelerates it symmetrically until it halts at the opposite extreme, and the exchange runs in reverse. Nothing in the mechanism contains a stopping condition. One consequence deserves stating because it is not intuitive: A does not appear in T. In this linear regime a small swing and a large swing of the same pendulum take the same time, a property called isochronism.

Worked example: the pendulum

A pendulum seems unlike a spring, yet it reduces to the same equation. The restoring force on a bob displaced by a small angle is the component of gravity along the arc, which for small angles is proportional to the displacement itself. Carrying that through gives omega = sqrt(g / L), where L is the length and g the gravitational acceleration, so T = 2 pi sqrt(L / g). The mass has cancelled out entirely. A pendulum's period depends on its length and on gravity, and not at all on how heavy the bob is, which is a claim worth testing rather than believing on authority.

Limits and boundary conditions

The linear law F = -kx is the leading Taylor term, accurate only while displacement stays small relative to the system's own scale: a shallow pendulum swing, a spring within its elastic range. Push far enough and the next term in the expansion stops being negligible; the restoring force curves away from proportionality, the period begins to depend on amplitude, and isochronism visibly fails. Galileo's report of amplitude-independent swing is a small-angle claim, exact only in that limit. This derivation also assumes no energy loss. Every real oscillator sheds energy each cycle to air resistance, internal friction, or radiated sound; that is damping, a distinct mechanism handled where this entry hands off to it, not a correction folded in here.

Common mistakes

Three errors are common enough to name. First, treating the restoring force and the force that eventually stops the motion as one thing. The restoring force creates the back-and-forth and, in the ideal case, never permanently stops anything; damping is a separate, dissipative effect. Second, assuming every oscillator's period depends on mass the same way. It does for a spring, where omega = sqrt(k/m) falls as mass rises, but for a pendulum the mass cancels and only length and gravity set the period. Third, expecting a larger push to change the period. In the linear regime the period is independent of amplitude, which is exactly the surprising part.

Build with it

Build one oscillator, a pendulum or a spring-and-mass, and hit a target period of 1.00 second within 2 percent. Predict first: for a pendulum solve L = g (T / 2 pi) squared, which for T = 1.00 s and g = 9.81 m/s squared gives about 0.248 m; for a spring, measure the static stretch under a known mass to get k, then solve for the mass that makes 2 pi sqrt(m/k) equal one second. Then measure: time twenty full cycles and divide by twenty, which averages out reaction-time error. Success is a measured period inside the 2 percent band. If you miss, diagnose it before adjusting: name whether the miss points to a mismeasured length or stiffness, a swing angle large enough to break the small-angle assumption, or friction bleeding amplitude away during the count.