Force Models & Inertial Frames

A force is not something an object carries around. It is one side of an interaction between two objects, and it is specified only when both can be named: the force on the crate by the rope, the force on the ball by Earth. Contact forces — normal force, friction, tension, spring forces, drag — need touching, and at the atomic scale every one of them is electromagnetic. Long-range forces act across a gap — gravitational, electric, magnetic — and for most everyday objects near the ground only gravity matters. So a force inventory is two questions: what touches the object, and what reaches it without touching? A 0.145 kg baseball just after release is touched only by air and reached only by Earth, which pulls it down with 0.145 × 9.81 = 1.42 N. The pitcher's push ended when contact ended; there is no “force of the throw” travelling with the ball. Earth is pulled up by the ball in the same interaction, but pairing those two forces is the third law's business, and they never share a diagram.

Superposition is an experimental result, not a definition. Hook two calibrated spring scales to a ring and it responds exactly as it would to a single force equal to their vector sum, and each scale's reading still depends only on its own stretch, whatever else is attached. That licenses ΣF = F₁ + F₂ + …, with one term for each interaction on the chosen object, and the sum is done in components. Seen from above, two ropes drag a sledge: 120 N at 30° to the left of the x axis and 90 N at 45° to its right. Then ΣFₓ = 120 cos 30° + 90 cos 45° = 103.9 + 63.6 = 167.6 N and ΣF_y = 120 sin 30° − 90 sin 45° = 60.0 − 63.6 = −3.6 N, so together the ropes pull with 167.6 N at 1.2° to the right of the x axis. Adding magnitudes would have claimed 210 N. For a particle model the sum is all that matters; where on the body each force is applied starts to matter only when rotation or deformation is part of the question.

Superposition says how forces combine; a force law says how large each one is. The ideal spring, Fₓ = −k(x − x₀), shows the pattern. Its input is a fact about the pair — how far the spring is stretched beyond its natural length x₀ — and its parameter is measured: hang a 1.00 kg mass from the spring, see it settle 4.9 cm lower, and k = 9.81 N ÷ 0.049 m ≈ 200 N m⁻¹. Build a scale around that spring and every later force you read is measured through the same law. Each force law also has a range. The spring is linear only below its elastic limit. Weight mg is the near-surface limit of inverse-square gravitation, and 10 km up it already overstates the pull by 0.31%. Kinetic friction and drag are empirical fits, trustworthy only for the surfaces and speeds they were fitted to. None of them warns you when it is pushed outside its range; each simply returns a confident wrong number.

A frame of reference is a set of rulers and a clock attached to something — the ground, a lab bench, a moving carriage. It is inertial if an object with no net interaction keeps a constant velocity when measured in it; Newton's first law, the next topic, asserts that such frames exist. The test is operational. Take a puck on a level air table: the air film pushes up to balance Earth's pull and leaves almost no friction. Release it and track it. If it moves in a straight line at constant speed to within your resolution, the frame passes, at that resolution. Repeat the test in a car braking at 3.0 m s⁻² and the puck accelerates forward at 3.0 m s⁻² relative to the car with nothing pushing it: the frame, not a force, is responsible. No real frame passes perfectly, since Earth's rotation alone gives the ground an acceleration of a few hundredths of a metre per second squared. So “inertial” always carries a tolerance: a frame is inertial enough when its acceleration is small beside the accelerations you mean to measure.

Let frame S′ move at constant velocity V along the x axis of an inertial frame S, with the origins together at t = 0. One event then has x′ = x − Vt, and both frames share one clock, t′ = t. Differentiate once: v′ = v − V, so velocities depend on the frame. Differentiate again: a′ = a, because V is constant. Now look at what force laws take as inputs. A separation transforms as x_A′ − x_B′ = (x_A − Vt) − (x_B − Vt) = x_A − x_B, and a relative velocity cancels V in the same way. So S′ assigns every force the value S does, measures the same accelerations and writes the same laws. A ball dropped from 1.20 m inside a carriage gliding at a steady 30 m s⁻¹ lands 0.495 s later directly below the release point; seen from the platform it travels 14.8 m forward along a parabola. Paths, velocities and kinetic energies differ between the two frames. The 9.81 m s⁻² and the weight that causes it do not.

Newton's laws do not break at a line; their errors grow smoothly, and the working question is whether the error exceeds your uncertainty. Four checks cover it. The frame must be inertial to the precision you need. Speeds must be small beside c = 3.00 × 10⁸ m s⁻¹: momentum mv falls short of the relativistic value by the fraction 1 − √(1 − v²/c²) ≈ v²/(2c²), which is 3.3 × 10⁻¹⁰ for a satellite at 7.7 km s⁻¹, 0.50% at 0.10c and 13% at 0.50c. The object's quantum wavelength h/(mv) must be negligible beside every length that matters: 6.6 × 10⁻²⁵ m for a 1.0 mg dust grain drifting at 1.0 mm s⁻¹, but 3.3 × 10⁻¹⁰ m for an electron in an atom, larger than the atom itself. And gravity must be weak, GM/(rc²) ≪ 1: about 7 × 10⁻¹⁰ at Earth's surface, near 0.2 at a neutron star's. Pass all four, use every force law inside its range, and a Newtonian prediction is as good as its inputs.