Relative Motion and Galilean Transformations

A fighter jet approaches an aircraft carrier at 240 km/h — a speed that would crumple any landing gear. So the carrier runs. Steaming into the wind at 56 km/h, it drags the runway out from under the plane, and the jet touches down at a survivable 184 km/h relative to the deck. Nothing about the aircraft changed; only the question "how fast?" got a better answer. Speed is not a property of a thing. It is a relationship between a thing and whoever is measuring it. The Galilean transformation is the precise dictionary between such observers: it translates the space-and-time coordinates recorded in one reference frame into those of another moving at constant velocity relative to it, under the classical assumption that a single universal clock ticks for everyone. From it flows the everyday rule that velocities add — the ball thrown forward on a moving train travels, for the platform, at train speed plus throw speed — and a far deeper one: that accelerations do not change at all. Position is a matter of opinion between frames; velocity is a matter of opinion; acceleration is a matter of fact. Because forces in classical mechanics depend only on relative positions and velocities, Newton's $\mathbf{F} = m\mathbf{a}$ reads identically in every uniformly moving laboratory. No mechanical experiment done inside a smoothly cruising ship can reveal that it is moving — Galileo's principle of relativity, four centuries old and still the working assumption of every physics problem you will ever solve at everyday speeds. This article builds the dictionary, proves what it preserves, shows you how to exploit it — the frame-hopping tricks that crack river crossings and collisions — and marks precisely where it fails, at the doorstep of Einstein.

Key takeaways

A Galilean transformation relates the coordinates of an event recorded in two inertial frames moving at constant relative velocity, leaving time absolute: $t' = t$ .
Positions transform as $\mathbf{r}' = \mathbf{r} - \mathbf{V}t$ , velocities add linearly ( $\mathbf{v}' = \mathbf{v} - \mathbf{V}$ ), and accelerations are identical ( $\mathbf{a}' = \mathbf{a}$ ).
Because acceleration is invariant and forces depend only on relative positions and velocities, Newton's second law $\mathbf{F} = m\mathbf{a}$ holds unchanged in every inertial frame — the Galilean principle of relativity.
The full transformation family has 10 parameters: 3 boosts, 3 rotations, 3 spatial translations, 1 time translation. Accelerating or rotating frames are not in the family.
Frame-hopping is a computational superpower: pick the frame where the problem is simplest (the water frame, the center-of-mass frame), solve, and boost back.
Galilean kinematics is the low-speed limit ( $V \ll c$ ) of the Lorentz transformation; absolute time is its defining approximation, with corrections of order $(V/c)^2$ .
Key idea: switching to a frame moving at constant velocity subtracts a fixed velocity from every velocity but leaves acceleration — and therefore Newton's laws — completely unchanged.

Where this lives today

Every flight you take runs on equation (2). A jet's flight management system continuously solves the wind triangle — airspeed vector plus wind vector equals ground-track vector — to hold a course, exactly the river-crossing geometry scaled to 900 km/h; the crab angle a landing aircraft holds in a crosswind is $\sin\theta = V_{\mathrm{wind}}/v_{\mathrm{air}}$ , the pilot's version of aiming upstream. Spacecraft rendezvous is the same discipline with higher stakes: a capsule and the ISS both circle Earth at 7.7 km/s, but docking is engineered entirely in the station's frame, where the closing speed is a few centimetres per second — frame choice turning a hypersonic encounter into a slow-motion parking maneuver. The center-of-mass boost of worked example 2 is the daily bread of collision physics, from billiard-ball mechanics to molecular-beam chemistry to particle physics, where usable collision energy is always quoted in the center-of-momentum frame. And the boundary of the Galilean world is now a consumer product: GPS satellites move fast enough ( $(V/c)^2 \approx 1.7\times10^{-10}$ at 3.9 km/s) and sit high enough in Earth's gravity that their clocks drift some 38 microseconds per day relative to yours — navigation would err by kilometres daily if the system assumed Galileo's absolute time. Your phone quietly runs the corrections every time it finds you.

The physical picture

Ride a smoothly cruising train and drop a coin. It falls straight down and lands at your feet — exactly as it would in your kitchen. Now let a friend on the platform watch the same drop through the window. She sees the coin trace a stretched parabola: it was moving forward at 30 m/s when you released it, kept that forward velocity all the way down (nothing horizontal acted on it), and landed at your feet only because your feet were moving forward at the same 30 m/s. One event, two honest descriptions. Neither is wrong; they differ by exactly the train's velocity, and that difference is the entire content of the Galilean boost.

Figure

One dropped coin, two true stories — straight fall on the train, a parabola from the platform

Think of the transformation as sliding the origin at a constant rate. The moving observer's coordinate grid glides over the stationary one, so any position reading picks up a correction $-\mathbf{V}t$ that grows linearly in time. Differentiate once: every velocity is shifted by the constant $-\mathbf{V}$ . Differentiate again: the correction is gone — a constant differentiates to zero — and acceleration comes out identical in both frames. That one cancellation is the keystone of the whole subject. It is why the coin behaves the same on a parked and a cruising train, why you can pour coffee on a jet at 900 km/h, and why no experiment sealed inside a smooth-riding cabin can measure the cabin's speed.

Figure

The bookkeeping — the train has carried its origin a head start of Vt, so readings differ by exactly that

One warning before the algebra: everything here presumes the relative velocity is constant. The moment the second frame brakes, accelerates, or spins, the clean cancellation fails, free objects appear to swerve with no cause, and you have left the Galilean family for the land of fictitious forces. The transformation below is a treaty between inertial frames only.

The dictionary and what it preserves

Start with the treaty itself. Let frame $S'$ (the train) translate at constant velocity $\mathbf{V}$ relative to frame $S$ (the ground), origins coinciding at $t = 0$ , and let both observers label the same events. Classical physics makes one silent, enormous assumption — that a single clock serves them both — and with it the dictionary is a single line:

t' = t, \qquad \mathbf{r}' = \mathbf{r} - \mathbf{V}t \qquad (1)

where: $t, t'$ — time recorded in $S$ and $S'$ (equal: absolute time); $\mathbf{r}, \mathbf{r}'$ — the position of one event as measured in each frame; $\mathbf{V}$ — the constant velocity of $S'$ relative to $S$ . Read it as pure bookkeeping: by time $t$ the train has carried its origin a distance $\mathbf{V}t$ down the track, so the train's reading is the ground's reading minus that head start. The innocent-looking clause $t' = t$ is a genuine physical claim — one clock fits all — and it is exactly the clause Einstein would later have to break. Differentiate equation (1) once with respect to the shared time and the boost passes to velocities. This is the rule you have used since childhood without naming it — walkway speeds add to walking speeds, headwinds subtract:

\mathbf{v}' = \mathbf{v} - \mathbf{V} \qquad (2)

where: $\mathbf{v} = \mathrm{d}\mathbf{r}/\mathrm{d}t$ — the particle's velocity in $S$ ; $\mathbf{v}'$ — its velocity in $S'$ . Equivalently $\mathbf{v} = \mathbf{v}' + \mathbf{V}$ : ground velocity is velocity-in-the-frame plus velocity-of-the-frame. This is Galilean velocity addition, and it is a full vector statement — the river crossing below lives entirely in its perpendicular components. Differentiate once more and the story changes character completely. The constant $\mathbf{V}$ has nothing left to contribute:

\mathbf{a}' = \mathbf{a} \qquad (3)

where: $\mathbf{a} = \mathrm{d}^2\mathbf{r}/\mathrm{d}t^2$ — acceleration in $S$ ; $\mathbf{a}'$ — acceleration in $S'$ , identical. Velocity is an opinion that differs from frame to frame; acceleration is a fact they all share. Equation (3) is why the dropped coin behaves the same on any smoothly moving vehicle, and it is the hinge on which the next result — the invariance of Newton's law itself — turns. Because classical forces (gravity between bodies, spring tensions, friction, contact pushes) depend only on relative positions and relative velocities — both of which the boost leaves untouched, since the $\mathbf{V}t$ and $\mathbf{V}$ shifts cancel in every difference — the force on a body is the same in both frames. Combined with equation (3) and the invariance of mass, the second law survives the translation word for word:

\mathbf{F}' = m\mathbf{a}' = m\mathbf{a} = \mathbf{F} \qquad (4)

where: $\mathbf{F}$ — the net force, computed from the same physics in either frame; $m$ — mass, an invariant scalar in classical mechanics. This is the Galilean principle of relativity in equation form: the laws of mechanics are identical in all inertial frames, so no mechanical experiment performed inside a closed, uniformly moving laboratory can detect the motion. There is no experiment that finds the "truly at rest" frame, because the equations refuse to name one. The boost is one member of a larger family. You may also rotate your axes, shift your origin, and reset your clock — none of it touches the physics. The complete dictionary between classical inertial observers is the 10-parameter Galilean transformation:

\mathbf{r}' = R\,\mathbf{r} - \mathbf{V}t - \mathbf{d}, \qquad t' = t - \tau \qquad (5)

where: $R$ — a constant rotation matrix (3 parameters); $\mathbf{V}$ — the boost velocity (3 parameters); $\mathbf{d}$ — a spatial translation of the origin (3 parameters); $\tau$ — a time offset (1 parameter). What survives every member of this family — the quantities all classical observers agree on — are the distance between simultaneous events, $|\mathbf{r}_2 - \mathbf{r}_1|$ , and the time interval between any two events, $t_2 - t_1$ . Note what is not on the list: positions, velocities, kinetic energies, momenta. All of those are frame-dependent bookkeeping; the invariants are the physics.

Which relation answers which question

you want	use	notes
translate one event's coordinates	$\mathbf{r}' = \mathbf{r} - \mathbf{V}t$ , eq. (1)	origins coincide at $t=0$ ; add rotations via eq. (5)
compose velocities (boat, wind, walkway)	$\mathbf{v} = \mathbf{v}' + \mathbf{V}$ , eq. (2)	vector addition — draw the triangle
check a frame change is harmless	$\mathbf{a}' = \mathbf{a}$ , eq. (3)	only for constant $\mathbf{V}$
justify solving in a convenient frame	eq. (4)	same laws, any inertial frame — pick the easy one
know when Galileo fails	$(V/c)^2$ correction, Derivation C	jets: $\sim 10^{-12}$ ; GPS satellites: $\sim 10^{-10}$ and it matters

Derivation AAcceleration is invariant, and Newton's second law rides along

We are deriving the two invariance statements, equations (3) and (4), that make all inertial frames dynamically equivalent — the mathematical content of Galileo's ship argument. The strategy is the shortest in mechanics: differentiate the boost (1) twice, watch the frame velocity die under the second derivative, then check separately that the force is also frame-independent, because $\mathbf{F}' = m\mathbf{a}'$ is only worth something if both sides transform the same way. The step most treatments skate over is that force check — it is where the physics (forces depend on relative configurations, not absolute positions) actually enters; the calculus alone proves nothing about dynamics.

Step 1. Write the boost with its shared clock: $\mathbf{r}'(t) = \mathbf{r}(t) - \mathbf{V}t$ , with $t' = t$ and $\mathbf{V}$ constant. Why: because the clocks agree, $\mathrm{d}/\mathrm{d}t' = \mathrm{d}/\mathrm{d}t$ — both observers can differentiate with respect to the same variable, which is the quiet workhorse of every step below.

Step 2. Differentiate once:

\mathbf{v}' = \frac{\mathrm{d}\mathbf{r}'}{\mathrm{d}t} = \mathbf{v} - \mathbf{V}

Why: the derivative of the linear term $\mathbf{V}t$ is the constant $\mathbf{V}$ — velocities disagree between the frames, and must, by exactly the frame velocity.

Step 3. Differentiate again:

\mathbf{a}' = \frac{\mathrm{d}\mathbf{v}'}{\mathrm{d}t} = \frac{\mathrm{d}\mathbf{v}}{\mathrm{d}t} - \frac{\mathrm{d}\mathbf{V}}{\mathrm{d}t} = \mathbf{a}

Why: the disagreement between the frames is constant, and a constant cannot survive a derivative. This is equation (3) — and notice the proof consumes the constancy of $\mathbf{V}$ completely: let $\mathbf{V}$ depend on time and an extra term $-\dot{\mathbf{V}}$ appears, which is precisely the fictitious-force acceleration of non-inertial frames.

Step 4. Check the force. Classical interactions depend on relative separations and relative velocities: gravity on $\mathbf{r}_i - \mathbf{r}_j$ , springs on their extension, friction and drag on velocity differences between surfaces or through a medium. Under the boost, $\mathbf{r}_i' - \mathbf{r}_j' = \mathbf{r}_i - \mathbf{r}_j$ (the $\mathbf{V}t$ shifts cancel) and $\mathbf{v}_i' - \mathbf{v}_j' = \mathbf{v}_i - \mathbf{v}_j$ (the $\mathbf{V}$ shifts cancel). Why this step is not optional: a hypothetical force law depending on absolute velocity would ruin the theorem even though $\mathbf{a}' = \mathbf{a}$ still held — form-invariance of dynamics needs both an invariant acceleration and an invariant force.

Step 5. Assemble: $\mathbf{F}' = \mathbf{F} = m\mathbf{a} = m\mathbf{a}'$ . The second law reads identically in $S'$ — equation (4). A physicist sealed in either frame writes the same equations, predicts the same experiments, and has no way to tell whose laboratory is "moving." Note: the failure mode for time-dependent $\mathbf{V}$ reproduces the $-m\dot{\mathbf{V}}$ fictitious force of non-inertial frames.

Go deeper · Derivation B · 2 minMomentum conservation survives the boost — the consistency check

Derivation BMomentum conservation survives the boost — the consistency check

The goal here is a consistency theorem the two previous results quietly promise: if total momentum is conserved in one inertial frame, it is conserved in every other, even though the value of the total momentum is different in each. This matters practically — it is the license for the center-of-mass trick used in the second worked example — and conceptually, because a conservation law that held for only one observer would smuggle an absolute rest frame in through the back door. The route is direct substitution of the velocity rule (2) into the total momentum, with one subtlety to watch: the proof needs the total mass to be conserved too, a fact so obvious in classical mechanics that it is easy to miss that the theorem depends on it.

Step 1. Let a closed system of particles $m_i$ have velocities $\mathbf{v}_i$ in $S$ , with total momentum $\mathbf{P} = \sum_i m_i \mathbf{v}_i$ conserved through some internal process. Why "closed" is doing work: momentum conservation follows from the absence of external forces, so all forces must be internal — collisions, springs, mutual gravity.

Step 2. Transform each velocity with equation (2) and sum:

\mathbf{P}' = \sum_i m_i \mathbf{v}_i' = \sum_i m_i (\mathbf{v}_i - \mathbf{V}) = \mathbf{P} - M\mathbf{V}

where: $M = \sum_i m_i$ — the total mass. Why the result is clean: the boost shifts the total momentum by the constant $M\mathbf{V}$ — the two observers disagree about the momentum's value by a fixed, time-independent amount.

Step 3. Compare before and after any internal process (a collision, an explosion): $\mathbf{P}'_{\text{after}} - \mathbf{P}'_{\text{before}} = (\mathbf{P}_{\text{after}} - M\mathbf{V}) - (\mathbf{P}_{\text{before}} - M\mathbf{V}) = \mathbf{0}$ . Why mass conservation sneaks in: the $M\mathbf{V}$ terms cancel only because $M$ is the same before and after — classical mechanics conserves mass, so the cancellation is exact.

Step 4. Track kinetic energy under the same substitution: $K' = \sum_i \tfrac12 m_i |\mathbf{v}_i - \mathbf{V}|^2 = K - \mathbf{P}\cdot\mathbf{V} + \tfrac12 M V^2$ . The value of $K$ is badly frame-dependent — but $\mathbf{P}$ , $M$ , and $\mathbf{V}$ are constants of the process, so the change obeys $\Delta K' = \Delta K$ . An elastic collision ( $\Delta K = 0$ ) is elastic for every observer. Why this soothes a famous confusion: energy bookkeeping differs between frames, but the conservation verdicts always agree, because kinetic energies in the two frames differ by a quantity that does not change.

Step 5. Conclude: conservation of momentum, of mass, and of energy are frame-independent statements, even though momentum and energy are frame-dependent numbers. You may therefore do the accounting in whichever inertial frame makes it trivial — which for collisions is the frame where $\mathbf{P}' = \mathbf{0}$ : boost with $\mathbf{V} = \mathbf{P}/M$ , the center-of-mass frame.

Go deeper · Derivation C · 2 minGalileo as the low-speed limit of Lorentz

Derivation CGalileo as the low-speed limit of Lorentz

We are deriving the precise sense in which everything above is an approximation — and how good the approximation is. Since 1905 the exact dictionary between inertial frames has been the Lorentz transformation, built on the experimental fact that the speed of light is the same for every observer. The strategy is to expand the Lorentz boost in the small parameter $V/c$ and watch the Galilean formulas emerge at leading order, with corrections of order $(V/c)^2$ . The term to watch is the one in the time equation: the Lorentz clock reading mixes in a position-dependent piece $Vx/c^2$ , and its disappearance as $c \to \infty$ is exactly the recovery of absolute time — the assumption we flagged at equation (1).

Step 1. Write the exact Lorentz boost along $x$ for frames with relative speed $V$ :

x' = \gamma\,(x - Vt), \qquad t' = \gamma\left(t - \frac{Vx}{c^2}\right), \qquad \gamma = \frac{1}{\sqrt{1 - V^2/c^2}}

Why this is the starting point: these formulas are forced on you by two postulates — the principle of relativity itself (which Galileo would applaud) plus the invariance of $c$ (which he never had reason to suspect).

Step 2. Expand the Lorentz factor for $V \ll c$ : $\gamma = 1 + \tfrac12 V^2/c^2 + \mathcal{O}(V^4/c^4)$ . Why the expansion is safe: for any vehicle humans build, $V/c$ is tiny, so the series converges violently fast — the first correction term is the entire story of "relativistic effects" at low speed.

Step 3. Substitute into the space equation: $x' = (x - Vt)\left(1 + \tfrac12 V^2/c^2 + \dots\right) \to x - Vt$ . Why: to leading order $\gamma$ is simply 1, and the Galilean position rule, equation (1), reappears untouched.

Step 4. Substitute into the time equation: $t' = t - Vx/c^2 + \mathcal{O}(V^2/c^2) \to t$ . Why this is the deep step: the offending term $Vx/c^2$ says moving clocks disagree about simultaneity in proportion to their separation — the genuinely non-classical ingredient. It dies as $c \to \infty$ , and absolute time, the silent axiom of the whole Galilean edifice, is recovered as a limit rather than a truth.

Step 5. Put in numbers to calibrate your trust. A jet at $V = 300\ \mathrm{m/s}$ : $V/c = 10^{-6}$ , so corrections enter at $(V/c)^2 \sim 10^{-12}$ — one part in a trillion, hopelessly beyond any kitchen-scale measurement. A satellite at $7.7\ \mathrm{km/s}$ : $(V/c)^2 \approx 6.6\times10^{-10}$ — still tiny, yet GPS clocks accumulate exactly such effects into microseconds per day, which navigation must correct or drift by kilometres. Why the calibration matters: Galilean relativity is not wrong; it is a controlled approximation whose error budget you now know how to compute. *Note: using $c = 2.998\times10^8\ \mathrm{m/s}$ , the numerical checks give $(300/3\times10^8)^2 = 10^{-12}$ and $(7.7\times10^3/3\times10^8)^2 = 6.6\times10^{-10}$ , confirming the calibration figures quoted above.*

Limiting cases

Case	Result	Meaning
zero relative velocity	$\mathbf{V} = \mathbf{0} \Rightarrow \mathbf{r}' = \mathbf{r},\ \mathbf{v}' = \mathbf{v}$	the transformation reduces to the identity; both observers agree on everything
low speeds	$\gamma \to 1$ as $V/c \to 0$	Lorentz collapses to Galileo; corrections $\sim (V/c)^2 \approx 10^{-12}$ for a jet
accelerating frame (breakdown)	$\mathbf{V} = \mathbf{V}(t) \Rightarrow \mathbf{a}' = \mathbf{a} - \dot{\mathbf{V}}$	frame leaves the Galilean family; the fictitious force $-m\dot{\mathbf{V}}$ appears
rotating frame (breakdown)	$\mathbf{a}' = \mathbf{a} - 2\boldsymbol{\omega}\times\mathbf{v}' - \boldsymbol{\omega}\times(\boldsymbol{\omega}\times\mathbf{r}')$	rotation is outside the group; Coriolis and centrifugal terms appear
current faster than the boat	$V > u \Rightarrow \sin\theta = V/u$ has no solution	no aim angle prevents downstream drift — a hard geometric limit of velocity addition

Common trap“Velocities just add — for anything that moves”

Galilean velocity addition is a low-speed approximation, not a law of nature. Apply it to light and it fails outright: the beam from a moving train's headlamp travels at $c$ for the platform observer too, not $c + V$ — the exact composition rule is $u' = (u - V)/(1 - uV/c^2)$ , which protects $c$ for everyone. Two more traps hide closer to home. First, the transformation treaty covers constant-velocity frames only: the instant your frame brakes, turns, or spins, equations (1)–(4) are void and fictitious forces take over — a car rounding a bend at steady speed is already outside the family, because its velocity direction is changing. Second, do not promote frame-dependent numbers into physics: kinetic energy and momentum have different values in different frames (a parked truck carries enormous kinetic energy in the frame of a passing car), and only their conservation — not their values — is a frame-independent fact.

Worked examples

1 · Crossing the river (confidence builder)

A boat moves at $u = 5.0\ \mathrm{m/s}$ relative to the water; the river flows at $V = 3.0\ \mathrm{m/s}$ and is $w = 120\ \mathrm{m}$ wide. (a) At what angle upstream must the pilot aim to land directly opposite, and how long does the crossing take? (b) If instead she aims straight across, where does she land, and which crossing is faster?

Step 1. Set up the frame composition, equation (2) rearranged: $\mathbf{v}_{\mathrm{ground}} = \mathbf{v}_{\mathrm{boat/water}} + \mathbf{V}_{\mathrm{water}}$ . Take $x$ downstream, $y$ across. Aiming at angle $\theta$ upstream of straight-across: $\mathbf{v}_{\mathrm{boat/water}} = (-u\sin\theta,\ u\cos\theta)$ .

Step 2. For a straight crossing the downstream component must vanish: $V - u\sin\theta = 0$ , so $\sin\theta = V/u = 3.0/5.0 = 0.60$ , giving $\boxed{\theta = 36.9°}$ upstream. Note the built-in limit: this only has a solution when $u > V$ — a slow boat in a fast current cannot land opposite, however cleverly aimed.

Step 3. The across-river speed is what remains: $v_y = u\cos\theta = \sqrt{u^2 - V^2} = \sqrt{25 - 9} = 4.0\ \mathrm{m/s}$ . Crossing time: $t = w/v_y = 120/4.0 = \boxed{30\ \mathrm{s}}$ . Aiming upstream spends part of your speed fighting the current.

Step 4. Aiming straight across instead: the full $5.0\ \mathrm{m/s}$ goes into crossing, so $t = 120/5.0 = \boxed{24\ \mathrm{s}}$ — faster! — but the current carries her downstream at $3.0\ \mathrm{m/s}$ for those 24 s, landing her $3.0 \times 24 = \boxed{72\ \mathrm{m}}$ below the target. Shortest time and shortest path are different optimizations; velocity addition lets you price each.

Figure

The river-crossing triangle — the ground velocity is the vector sum of boat-through-water and current

2 · The collision that solves itself — the center-of-mass frame (exam level)

A $2.0\ \mathrm{kg}$ cart moving at $6.0\ \mathrm{m/s}$ strikes a stationary $1.0\ \mathrm{kg}$ cart head-on, perfectly elastically. Find both final velocities — without solving simultaneous equations — by boosting to the center-of-mass frame and back.

Step 1. Find the boost velocity. The center-of-mass frame is the one where the total momentum vanishes; by Derivation B it moves at $V_{\mathrm{cm}} = P/M = (2.0 \times 6.0 + 1.0 \times 0)/(2.0 + 1.0) = 12/3 = 4.0\ \mathrm{m/s}$ .

Step 2. Boost into it with equation (2): $v_1' = 6.0 - 4.0 = +2.0\ \mathrm{m/s}$ ; $v_2' = 0 - 4.0 = -4.0\ \mathrm{m/s}$ . Check the frame is doing its job: $P' = 2.0(2.0) + 1.0(-4.0) = 0$ . ✓

Step 3. Collide. In the zero-momentum frame an elastic head-on collision has only one possible outcome: each velocity simply reverses. Momentum stays zero either way, and reversing preserves each speed, hence the kinetic energy. So $v_1' \to -2.0\ \mathrm{m/s}$ and $v_2' \to +4.0\ \mathrm{m/s}$ . This is the entire "calculation" — the frame did the algebra.

Step 4. Boost back to the lab by adding $V_{\mathrm{cm}}$ : $v_1 = -2.0 + 4.0 = \boxed{+2.0\ \mathrm{m/s}}$ , $v_2 = +4.0 + 4.0 = \boxed{+8.0\ \mathrm{m/s}}$ .

Step 5. Audit in the lab frame. Momentum: $2.0(2.0) + 1.0(8.0) = 12\ \mathrm{kg\,m/s}$ — matches the initial $2.0 \times 6.0 = 12$ . ✓ Kinetic energy: before, $\tfrac12(2.0)(36) = 36\ \mathrm{J}$ ; after, $\tfrac12(2.0)(4.0) + \tfrac12(1.0)(64) = 4.0 + 32 = 36\ \mathrm{J}$ . ✓ Both conservation laws hold in both frames — exactly what Derivation B guarantees — and the standard textbook formulas $v_1 = \frac{m_1 - m_2}{m_1 + m_2}v = \frac{1}{3}(6.0) = 2.0$ and $v_2 = \frac{2m_1}{m_1 + m_2}v = \frac{4}{3}(6.0) = 8.0$ confirm the answer they take a page of simultaneous algebra to produce.

Origins · 1 minGalileo Galilei and how this was discovered

Galileo Galilei

1564–1642 · Pisa, Padua, Florence

Galileo never wrote a transformation equation — the algebra that bears his name was formalized centuries later — but he owned the physics completely. In the Dialogue Concerning the Two Chief World Systems (1632) he asked readers to shut themselves below decks of a smoothly sailing ship with dripping bottles, swimming fish, and fluttering butterflies: every drop, dart, and leap unfolds exactly as in port, and no observation made inside can reveal the ship's motion. The argument was built for a purpose — to answer the anti-Copernican objection that a moving Earth would fling arrows and birds backward — and it worked: uniform shared motion is imperceptible, so the Earth can fly through space without anyone stumbling. That single thought experiment seeded the principle of relativity that Newton codified as Corollary V of the Principia and that Einstein, in 1905, refused to abandon even for light — rebuilding time itself to save it.

History

1632

— Galileo's Dialogue argues from the sailing ship that shared uniform motion is undetectable — the relativity principle, invented to defend a moving Earth.

1687

— Newton's Principia, Corollary V: the motions of bodies among themselves are the same whether the containing space is at rest or moves uniformly in a straight line.

1887

— Michelson and Morley fail to detect Earth's motion through the ether: light stubbornly refuses to obey Galilean velocity addition.

1905

— Einstein keeps Galileo's principle but adds the invariance of light speed; the Lorentz transformation replaces the Galilean one, and absolute time falls.

1909

— Minkowski's spacetime formulation makes the contrast precise: Galilean physics is the geometry of absolute time, relativity the geometry of spacetime.

Going further

References. Morin, Introduction to Classical Mechanics, Ch. 3 & 11 · Goldstein, Classical Mechanics 3rd ed., §1.1 and §7.1 · Landau & Lifshitz, Mechanics Vol. 1, §3 (Galileo's relativity principle) · Taylor, Classical Mechanics, Ch. 1 & 9 · Kleppner & Kolenkow, An Introduction to Mechanics, Ch. 4 (momentum and the center-of-mass frame). All equations follow from differentiating the boost $\mathbf{r}'=\mathbf{r}-\mathbf{V}t$ , yielding $\mathbf{v}'=\mathbf{v}-\mathbf{V}$ and $\mathbf{a}'=\mathbf{a}$ ; the Lorentz expansion gives $\gamma=1+\tfrac12 V^2/c^2$ with $c=2.998\times10^8$ m/s (exact defined value). Key numerical results: closing speed 60 m/s; jet $(V/c)^2\sim10^{-12}$ ; satellite $(V/c)^2\approx6.6\times10^{-10}$ at $V=7.7$ km/s. Worked-example numbers: river crossing $\theta = 36.9°$ , crossing at $4.0\ \mathrm{m/s}$ in 30 s vs 24 s with 72 m of drift; elastic collision $V_{\mathrm{cm}} = 4.0\ \mathrm{m/s}$ , final velocities 2.0 and 8.0 m/s, 36 J conserved.

Builds on: Position, Displacement, and Reference Frames · Velocity and Acceleration · Vectors in Kinematics and Coordinate Systems · Newton's First Law and Inertial Frames. Leads to: Newton's Second Law · Projectile Motion · Circular Motion and Centripetal Acceleration.

If velocities just add, what happens with the headlight beam of a moving train?

Galilean addition predicts platform observers should measure $c + V$ for the beam — and the experiment says they measure exactly $c$ . This is not a small discrepancy; it is the crack through which relativity entered. The honest composition rule is $u' = (u - V)/(1 - uV/c^2)$ , which reduces to the Galilean $u - V$ when both speeds are tiny compared to $c$ , but conspires to return $c$ whenever either input is $c$ . For the train's headlight the "error" of the Galilean rule is total; for the ball thrown on the same train it is parts in $10^{13}$ . The rule you use daily is the low-speed shadow of the real one — trustworthy precisely because you live slowly.

Kinetic energy is different in different frames — so is energy conservation violated for some observers?

No, and Derivation B shows why in one line. Under a boost, kinetic energy shifts by $-\mathbf{P}\cdot\mathbf{V} + \tfrac12 MV^2$ — but for a closed system $\mathbf{P}$ , $M$ , and $\mathbf{V}$ are all constants, so the shift is the same before and after any process. Values change; changes don't. An elastic collision conserves kinetic energy for every inertial observer, and a crash that dissipates 50 kJ into heat dissipates 50 kJ in every frame. The confusion usually comes from comparing energy values between frames (a parked truck has huge kinetic energy in the frame of a passing car — true, and harmless) instead of comparing conservation statements, which all frames share.

Why does a coin dropped on a moving train land at my feet instead of behind me?

Because at the moment of release the coin is moving forward at the train's speed — it is part of the train's world, not the ground's. Letting go removes the horizontal support, not the horizontal velocity; with no horizontal force acting during the fall, the coin keeps its 30 m/s forward motion (that is Newton's first law) while gravity accelerates it downward. Your feet also keep moving forward at 30 m/s, so coin and feet advance in lockstep and reunite. The platform observer sees a parabola; you see a vertical drop; both are correct, differing by exactly the boost $\mathbf{V}t$ . The coin would land behind you only if the train accelerated during the fall — which is precisely the test that distinguishes inertial frames from non-inertial ones.

Is there any experiment that decides which frame is really at rest?

Within mechanics — and, after Einstein, within all of physics — no. Equation (4) says the laws of motion are identical in every inertial frame, so any experiment you run in a sealed laboratory comes out the same whether the lab is "parked" or gliding at 200 km/h. This is not a technological limitation but a structural fact about the laws: they contain relative positions, relative velocities, and accelerations, and none of those can encode an absolute velocity. The nineteenth century bet otherwise — the ether was supposed to be the truly-at-rest frame, detectable by light — and Michelson and Morley's null result in 1887 closed that door. "At rest" is a convention you choose for convenience, like choosing where to put the origin.

Velocity is relative — so when a frame accelerates, what is it accelerating relative to?

This is one of the sharpest questions in classical physics. Operationally the answer is: relative to any inertial frame — and all inertial frames agree on the acceleration's value, by equation (3), so no ambiguity arises. You do not even need to look outside: an accelerometer (a mass on a spring) reads frame acceleration directly, with the blinds drawn. Velocity is undetectable from inside; acceleration announces itself. Why the universe singles out acceleration as absolute while velocity stays relative is a deeper matter — Newton credited absolute space, Mach argued inertia refers to the distant matter of the universe, and Einstein's general relativity ultimately reframed the question. For working mechanics, the operational answer is all you need.

Can I use Galilean velocity addition for sound? It feels like the same physics as the river

Yes — and that is exactly what makes sound different from light. Sound propagates through a medium, air, at about 343 m/s relative to the air; the air frame is a genuine, physically preferred frame for acoustics, and all compositions are honestly Galilean. A tailwind really does speed sound along the ground; a supersonic jet really does outrun its own noise. This is also why the classical Doppler formulas are asymmetric between a moving source and a moving observer — the two motions relate differently to the medium. Light has no medium, no preferred frame, and the same speed for everyone, which is why its Doppler effect depends only on relative velocity and needs relativity to derive.

A car turning a corner at constant speed — is its frame inertial? The speedometer never moves

No. "Constant velocity" is a vector condition, and the turning car fails it: the direction of its velocity changes, so it accelerates (centripetally, at $v^2/r$ ), and equations (1)–(4) do not apply to its frame. Ride inside and physics confirms it — loose objects slide across the seat with no real force pushing them, the fictitious centrifugal effect of a non-inertial frame. The Galilean family is strictly the constant-speed-and-constant-direction club. A useful habit: before boosting into any frame, ask what an accelerometer bolted to it would read. Zero — welcome to the family, transform away. Nonzero — you need the machinery of fictitious forces, not the Galilean dictionary.

How fast is too fast for Galileo? Where do I actually need the Lorentz transformation?

Budget it with the number from Derivation C: relativistic corrections enter at order $(V/c)^2$ . A car at 30 m/s: $10^{-14}$ . A jet at 300 m/s: $10^{-12}$ . A satellite at 7.7 km/s: $6.6\times10^{-10}$ . If your measurement cannot resolve one part in a billion, Galileo serves you perfectly at any speed humans travel. The interesting middle ground is precision timekeeping: GPS clocks are stable to parts in $10^{13}$ , so those "negligible" $10^{-10}$ effects accumulate to about 38 microseconds per day — kilometres of position error — and must be corrected. As a rule of thumb: below about $0.01c$ (3,000 km/s!) Galilean mechanics is fine for percent-level work; particle physics, astrophysical jets, and atomic clocks live beyond it.

Relative Motion and Galilean Transformations

Where this lives today

The physical picture

The dictionary and what it preserves

Which relation answers which question

Limiting cases

Worked examples

1 · Crossing the river (confidence builder)

2 · The collision that solves itself — the center-of-mass frame (exam level)

Going further

Before this

Where this leads