Special Relativity Postulates and the Lorentz Transformation

In 1905 physics held two treasures that could not both be right. Newtonian mechanics came with a symmetry — the Galilean transformation $x'=x-vt,\ t'=t$ — under which its laws keep their form in every inertial frame. Maxwell's electrodynamics predicted electromagnetic waves travelling at one fixed speed $c$ , with no reference to who is measuring. Boost yourself after a light beam and Galileo says it should slow down; Maxwell's equations say it cannot. The Michelson–Morley experiment sided with Maxwell: no motion of the Earth through any "aether" was detectable, to extraordinary precision. Einstein's resolution was not to patch either theory but to replace the transformation itself. From two economical postulates — no inertial frame is privileged, and light moves at $c$ in all of them — the map between inertial coordinates is forced, almost uniquely, to be the Lorentz transformation. Space and time coordinates mix. Simultaneity becomes a frame-dependent bookkeeping choice. And a single quadratic quantity, the spacetime interval $s^2=c^2t^2-x^2$ , takes over the invariant role that distance played in Euclidean geometry: boosts are revealed as hyperbolic rotations. For a mechanics course this article is the gateway of the relativity chapter. Everything downstream — time dilation and length contraction, four-vectors, four-momentum, and the chapter's payoff, the relativistic Lagrangian — is unpacking of the one matrix derived here. We keep the treatment compact and mechanical: postulates, the derivation, the geometry, the limits, and the numbers.

Key takeaways

Special relativity rests on two postulates: physical laws take the same form in every inertial frame, and the vacuum speed of light $c$ is the same in every inertial frame, regardless of source or observer motion.
The postulates force the Galilean transformation to be replaced by the Lorentz boost $x'=\gamma(x-vt)$ , $t'=\gamma(t-vx/c^2)$ , with Lorentz factor $\gamma=(1-v^2/c^2)^{-1/2}$ .
The boost is exactly the linear map that preserves the spacetime interval $s^2=c^2t^2-x^2$ — a hyperbolic rotation of the $(ct,x)$ plane through a rapidity $\phi$ with $\tanh\phi=v/c$ .
Relativity of simultaneity, time dilation, length contraction, and the velocity-addition law $u'=(u-v)/(1-uv/c^2)$ are all consequences of this one matrix; rapidities, unlike velocities, simply add.
For $v\ll c$ the boost collapses smoothly to Galileo's $x'=x-vt$ , $t'=t$ : Newtonian mechanics is the infinite-signal-speed limit of relativity.
Key idea: Demanding that the speed of light be invariant is equivalent to demanding that a single quadratic form — the spacetime interval $s^2=c^2t^2-x^2$ — be invariant; the linear maps that preserve it are the Lorentz transformations, hyperbolic rotations of spacetime.

Where this lives today

Every collider analysis is Lorentz kinematics on an industrial scale. At the LHC, protons circulate with $\gamma\approx7460$ ( $E=7\ \mathrm{TeV}$ ), and analysis frameworks quote every jet and muon in rapidity (or its massless cousin pseudorapidity $\eta$ ) precisely because rapidity differences are boost-invariant along the beam axis — Derivation C is baked into the coordinate system of every detector at CERN. Time dilation is likewise an engineering input: muon beams for the 2026-era muon-collider design studies only exist as usable beams because $\gamma\tau$ stretches a 2.2 μs lifetime into transportable milliseconds. The transformation is also, quietly, consumer infrastructure. GPS satellite clocks run fast by about $+45\ \mu\mathrm{s}/\mathrm{day}$ gravitationally but slow by $-7\ \mu\mathrm{s}/\mathrm{day}$ from special-relativistic time dilation at orbital speed $3.9\ \mathrm{km/s}$ ; both corrections are programmed into the clock rates before launch, and without the Lorentz part alone, positions would drift by roughly two kilometres per day. Interferometric gravitational-wave detectors, pulsar-timing arrays, and the optical-clock networks now testing Lorentz invariance to parts in $10^{20}$ all live inside the same symmetry group $O(3,1)$ defined by equation (8).

The physical picture

Think of a boost as a rotation, but with hyperbolas instead of circles. An ordinary rotation mixes $x$ and $y$ while keeping $x^2+y^2$ fixed; a Lorentz boost mixes $ct$ and $x$ while keeping $c^2t^2-x^2$ fixed. Because the preserved quadratic form has a minus sign, the "angle" is a rapidity $\phi$ , and $\cosh\phi,\ \sinh\phi$ replace $\cos\theta,\ \sin\theta$ . Everything strange about relativity is this one sign.

Figure

Rotation versus boost — an ordinary rotation slides points along circles keeping x squared plus y squared fixed, while a Lorentz boost slides events along hyperbolas keeping the interval fixed; the 45-degree light-cone asymptotes are shared by every observer

The constancy of $c$ is geometrically the statement that light rays travel along the $45^\circ$ lines $x=\pm ct$ — the asymptotes of the invariant hyperbolas. Every observer's light cone is the same cone. A boost slides events along their hyperbola but can never push a point across its own light cone, which is why the time order of causally connectable events is absolute even though almost everything else is negotiable. Simultaneity is a choice of slicing. An inertial observer calls a set of events "now" if they share her time coordinate — a horizontal line in her spacetime diagram. A boosted observer's "now" is a tilted line. The two slicings disagree, and that single disagreement is the root of every apparent paradox in the subject.

Figure

Two observers slice spacetime differently — the boosted frame's time axis is the moving worldline and its now-slices tilt toward the light cone; events A and B simultaneous for the rest frame lie on different now-slices of the moving frame

The Galilean transformation is the $c\to\infty$ limit. When the invariant speed is infinite, the light cones flatten into a single absolute-simultaneity plane, the hyperbolas degenerate into horizontal lines, and time becomes universal. Newtonian mechanics — the whole edifice built from the Lagrangian Function and Hamilton's Principle onward — is special relativity with an infinite signal speed.

The core equations

Everything in this chapter flows from one linear map. For frames in standard configuration — $S'$ moving at speed $v$ along the common $+x$ axis, origins coinciding at $t=t'=0$ — the coordinates of one event are related by the Lorentz boost:

x' = \gamma\,(x - vt),\qquad t' = \gamma\left(t - \frac{v}{c^2}\,x\right),\qquad y'=y,\quad z'=z \qquad (1)

where: $x,t$ — coordinates of an event in frame $S$ ; $x',t'$ — coordinates of the same event in $S'$ ; $v$ — relative frame velocity; $c$ — vacuum speed of light; $\gamma$ — the Lorentz factor of equation (2). Two features deserve a hard look. The spatial equation is Galileo's $x-vt$ stretched by $\gamma$ ; the radical novelty is the $-vx/c^2$ in the time equation, which makes where an event happens feed into when it happens — simultaneity becomes relative. Derivation A shows these coefficients are forced by the two postulates; nothing about them is adjustable. The stretching factor appears so often it gets its own symbol. Define

\gamma \equiv \frac{1}{\sqrt{1-\beta^2}},\qquad \beta \equiv \frac{v}{c} \qquad (2)

where: $\beta$ — dimensionless speed, $0\le\beta<1$ for material frames; $\gamma$ — Lorentz factor, equal to $1$ at rest, growing monotonically, and diverging as $\beta\to1$ . Useful anchors: $\beta=0.6$ gives $\gamma=1.25$ ; $\beta=0.8$ gives $\gamma=5/3$ ; $\beta=0.998$ (a cosmic-ray muon) gives $\gamma\approx16$ . The divergence at $\beta=1$ is the mathematical form of "you cannot boost to the speed of light." What survives the transformation unchanged? Not lengths, not durations — but one specific combination of both. For an event at $(t,x,y,z)$ the spacetime interval from the origin,

s^2 = c^2t^2 - x^2 - y^2 - z^2 = c^2t'^2 - x'^2 - y'^2 - z'^2 \qquad (3)

where: $s^2$ — squared interval, the same number in every inertial frame (Derivation B verifies this by direct substitution). The relative minus sign between time and space is the entire geometry of relativity: $s^2>0$ (timelike) means a clock could travel between the events, $s^2=0$ (null) means only light connects them, $s^2<0$ (spacelike) means no signal can. The interval does for spacetime what $x^2+y^2+z^2$ does for Euclidean space. Because the boost is linear, it earns a matrix. In the $(ct,x)$ block,

\begin{pmatrix} ct' \\ x' \end{pmatrix} = \begin{pmatrix} \gamma & -\gamma\beta \\ -\gamma\beta & \gamma \end{pmatrix} \begin{pmatrix} ct \\ x \end{pmatrix} \qquad (4)

where: the symmetric $2\times2$ matrix is the boost acting on the time–space plane. Its determinant is $\gamma^2-\gamma^2\beta^2=\gamma^2(1-\beta^2)=1$ : boosts preserve area in the $(ct,x)$ plane and never reverse orientation. The inverse is the same matrix with $\beta\to-\beta$ — exactly what the principle of relativity demands, since $S$ moves at $-v$ relative to $S'$ . The matrix in (4) begs to be parametrized like a rotation. Setting $\gamma=\cosh\phi$ and $\gamma\beta=\sinh\phi$ — consistent because $\cosh^2\phi-\sinh^2\phi=1$ automatically enforces $\gamma^2(1-\beta^2)=1$ — turns the boost into

ct' = ct\,\cosh\phi - x\,\sinh\phi,\qquad x' = -\,ct\,\sinh\phi + x\,\cosh\phi,\qquad \tanh\phi = \beta \qquad (5)

where: $\phi$ — the rapidity, the hyperbolic angle of the boost. Rapidity is the additive parameter: composing collinear boosts adds rapidities, $\phi_{13}=\phi_{12}+\phi_{23}$ (Derivation C), while velocities combine nonlinearly. Accelerator physicists quote beam energies in rapidity for precisely this reason. Differentiating the boost gives the transformation of velocities. If an object moves at $u_x=\mathrm{d}x/\mathrm{d}t$ in $S$ , then in $S'$ ,

u_x' = \frac{u_x - v}{1 - u_x v/c^2} \qquad (6)

where: $u_x'$ — the object's velocity measured in $S'$ . The numerator is Galileo's answer; the denominator is the relativistic correction that keeps every subluminal speed subluminal. Feed in $u_x=c$ and the formula returns $u_x'=c$ identically — the second postulate, now a theorem of the transformation. For $u_xv\ll c^2$ the denominator is $1$ and ordinary velocity subtraction returns. Finally, apply the time equation of (1) to two events simultaneous in $S$ ( $\Delta t=0$ ) but separated by $\Delta x$ :

\Delta t' = -\,\gamma\,\frac{v}{c^2}\,\Delta x \qquad (7)

where: $\Delta t'$ — their time separation in $S'$ : not zero. The event ahead in the direction of motion happens earlier in $S'$ . This is the relativity of simultaneity, the most commonly dropped — and most load-bearing — consequence of (1); the warning box below and the twin-rod puzzles of the next article both turn on it. The full symmetry structure, stated once for orientation and developed properly in the four-vector article: with the Minkowski metric $\eta_{\mu\nu}=\mathrm{diag}(+1,-1,-1,-1)$ , the Lorentz transformations are all matrices $\Lambda$ obeying

\Lambda^{\mu}{}_{\alpha}\,\Lambda^{\nu}{}_{\beta}\,\eta_{\mu\nu} = \eta_{\alpha\beta} \qquad (8)

where: $\Lambda^{\mu}{}_{\alpha}$ — a general Lorentz matrix (boosts plus spatial rotations), and the condition says $\Lambda$ preserves the metric — the exact analogue of $R^TR=\mathbb{1}$ for rotations. These matrices form the Lorentz group $O(3,1)$ ; adding spacetime translations gives the Poincaré group, the symmetry algebra on which relativistic mechanics and field theory are built.

You want to…	Use	Watch for
Transform an event's coordinates	boost (1) or matrix (4)	standard configuration; origins synced
Transform a velocity	addition law (6)	it is $u$ and $v$ , not two frame velocities blindly
Compose two boosts	rapidities add, eq. (5)	velocities do NOT add
Check causal character	interval (3)	sign convention $(+,-,-,-)$ fixed here
Compare clock rates / rod lengths	next article's $\Delta t=\gamma\Delta\tau$ , $L=L_0/\gamma$	identify whose time is proper

Derivation ALorentz transformation from the two postulates

We derive the boost (1) from the postulates alone, to show that no freedom remains once light speed is invariant. The strategy: homogeneity of space and time forces the map to be linear, so we write the most general linear ansatz for motion along $x$ ; the definition of relative velocity, the principle of relativity (no privileged frame), and the invariance of $c$ then fix all four coefficients. Watch how each physical postulate kills exactly one unknown — and how the last step produces $\gamma$ as a derived quantity, not an assumption.

Step 1. Write the general linear ansatz for a boost along $x$ :

x' = A\,x + B\,t,\qquad t' = D\,x + E\,t

Why linear: homogeneity means no event is special — the transformation must look the same around every origin, and only linear maps do that. Equivalently, force-free particles move on straight worldlines in both frames, and only linear maps send every straight line to a straight line.

Step 2. The origin of $S'$ ( $x'=0$ ) is the point moving at $v$ in $S$ , i.e. the line $x=vt$ . Substituting:

0 = A\,vt + B\,t \ \Rightarrow\ B = -A\,v \ \Rightarrow\ x' = A\,(x - vt)

Rename the remaining coefficient $A\equiv\gamma$ , to be determined.

Step 3. Invoke the principle of relativity: $S$ moves at $-v$ relative to $S'$ , and neither frame is privileged, so the inverse map must have the identical form with $v\to-v$ and the same stretch factor:

x = \gamma\,(x' + v\,t')

Why the same gamma: if the forward and inverse stretches differed, the two frames would be physically distinguishable — one could say absolutely which frame was "really" moving — contradicting postulate 1 (isotropy of space rules out a sign asymmetry as well).

Step 4. Impose the second postulate. A light pulse leaving the common origin at $t=t'=0$ obeys $x=ct$ in $S$ and must obey $x'=ct'$ in $S'$ . Substitute both into the Step-2 and Step-3 relations:

ct' = x' = \gamma(c-v)\,t,\qquad ct = x = \gamma(c+v)\,\frac{x'}{c}\cdot\frac{c\,t'}{x'} \ \Rightarrow\ ct = \gamma\,(c+v)\,t'

Step 5. Multiply the two boxed relations: $c^2\,t\,t' = \gamma^2(c-v)(c+v)\,t\,t'$ , so

\gamma^2\,(c^2 - v^2) = c^2 \ \Rightarrow\ \gamma = \frac{1}{\sqrt{1-v^2/c^2}}

Step 6. Recover the time equation by eliminating $x'$ between Steps 2 and 3: $x = \gamma\left(\gamma(x-vt) + vt'\right)$ , solve for $t'$ :

t' = \frac{x - \gamma^2(x-vt)}{\gamma v} = \gamma\left(t - \frac{v}{c^2}\,x\right)

Why the algebra collapses so neatly: $1-\gamma^2 = -\gamma^2 v^2/c^2$ , which is where the $v/c^2$ coefficient — the relativity of simultaneity — is born. *Verified: as $c\to\infty$ , $\gamma\to1$ and (1) reduces to Galileo's $x'=x-vt,\ t'=t$ ; substituting $x=ct$ returns $x'=ct'$ exactly; matches Morin Ch. 11 and Landau–Lifshitz CTF §4.*

Go deeper · Derivation B · 1 minInterval invariance and the boost as a hyperbolic rotation

Derivation BInterval invariance and the boost as a hyperbolic rotation

We verify that the boost preserves the interval $c^2t^2-x^2$ by direct substitution, then show the preserved-hyperbola structure means every boost is a hyperbolic rotation. The calculation is short but the payoff is the entire geometric picture: it converts "the speed of light is invariant" into "spacetime has an indefinite metric," and it identifies the natural additive boost parameter. Keep track of how the cross terms cancel — that cancellation is not luck but the signature of an orthogonal-type transformation.

Step 1. Substitute the boost into $c^2t'^2 - x'^2$ :

c^2t'^2 - x'^2 = \gamma^2\left(ct - \tfrac{v}{c}x\right)^2 - \gamma^2\left(x - vt\right)^2

Step 2. Expand both squares:

= \gamma^2\left[c^2t^2 - 2vtx + \tfrac{v^2}{c^2}x^2 - x^2 + 2vtx - v^2t^2\right]

Why the cross terms cancel: the time equation carries $-vx/c^2$ and the space equation carries $-vt$ with matched coefficients; this pairing is precisely what "preserving a quadratic form" requires, the Minkowski analogue of $R^TR=\mathbb{1}$ .

Step 3. Group the surviving terms and factor:

= \gamma^2\left[(c^2-v^2)\,t^2 - \left(1-\tfrac{v^2}{c^2}\right)x^2\right] = \gamma^2\left(1-\beta^2\right)\left(c^2t^2 - x^2\right) = c^2t^2 - x^2

since $\gamma^2(1-\beta^2)=1$ by definition. The interval is invariant.

Step 4. Geometry: events with fixed $s^2$ lie on hyperbolas $c^2t^2-x^2=\mathrm{const}$ , and boosts slide events along them. Parametrize with a hyperbolic angle by setting

\cosh\phi \equiv \gamma,\qquad \sinh\phi \equiv \gamma\beta

which is consistent because $\cosh^2\phi-\sinh^2\phi = \gamma^2(1-\beta^2) = 1$ holds identically, and gives $\tanh\phi=\beta$ .

Step 5. Rewrite the boost matrix (4) with these entries:

\begin{pmatrix} ct' \\ x' \end{pmatrix} = \begin{pmatrix} \cosh\phi & -\sinh\phi \\ -\sinh\phi & \cosh\phi \end{pmatrix}\begin{pmatrix} ct \\ x \end{pmatrix}

Why this is called a rotation: replace $\cosh\to\cos$ , $\sinh\to\sin$ and this is a rotation matrix. The sign difference in the off-diagonal — both negative rather than opposite — is what preserves $c^2t^2-x^2$ instead of $c^2t^2+x^2$ . *Verified: numerical spot check at $\beta=0.6$ — the event $(ct,x)=(300\ \mathrm{m},240\ \mathrm{m})$ maps to $(195\ \mathrm{m},75\ \mathrm{m})$ , and $300^2-240^2 = 32400 = 195^2-75^2$ ; matches Taylor & Wheeler Ch. 3.*

Go deeper · Derivation C · 1 minComposition of collinear boosts and the additivity of rapidity

Derivation CComposition of collinear boosts and the additivity of rapidity

We compose two boosts along the same axis and show the result is again a boost whose rapidity is the sum of the two rapidities — from which the velocity-addition law (6) falls out as a corollary. The strategy is pure matrix multiplication in the rapidity form of Step 5 above, letting the hyperbolic angle-addition identities do all the work. This derivation explains why velocities refuse to add: they are the $\tanh$ of the thing that adds.

Step 1. Write the two boosts as hyperbolic rotations with rapidities $\phi_1$ and $\phi_2$ and multiply:

\Lambda(\phi_2)\,\Lambda(\phi_1) = \begin{pmatrix} \cosh\phi_2 & -\sinh\phi_2 \\ -\sinh\phi_2 & \cosh\phi_2 \end{pmatrix}\begin{pmatrix} \cosh\phi_1 & -\sinh\phi_1 \\ -\sinh\phi_1 & \cosh\phi_1 \end{pmatrix}

Step 2. Expand the product entries and apply the addition identities $\cosh(\phi_1+\phi_2)=\cosh\phi_1\cosh\phi_2+\sinh\phi_1\sinh\phi_2$ and $\sinh(\phi_1+\phi_2)=\sinh\phi_1\cosh\phi_2+\cosh\phi_1\sinh\phi_2$ :

\Lambda(\phi_2)\,\Lambda(\phi_1) = \begin{pmatrix} \cosh(\phi_1{+}\phi_2) & -\sinh(\phi_1{+}\phi_2) \\ -\sinh(\phi_1{+}\phi_2) & \cosh(\phi_1{+}\phi_2) \end{pmatrix} = \Lambda(\phi_1+\phi_2)

Why this matters structurally: collinear boosts form a one-parameter group with rapidity as the additive parameter — the exact analogue of coplanar rotation angles adding.

Step 3. Translate back to velocities using $\beta=\tanh\phi$ and the sum formula for $\tanh$ :

\beta_{12} = \tanh(\phi_1+\phi_2) = \frac{\tanh\phi_1 + \tanh\phi_2}{1 + \tanh\phi_1\tanh\phi_2} = \frac{\beta_1+\beta_2}{1+\beta_1\beta_2}

Step 4. Read off the physics: since $|\tanh|<1$ for every finite argument, $\beta_{12}<1$ whenever $\beta_1,\beta_2<1$ — no chain of boosts ever reaches $c$ . Setting $\beta_2=1$ gives $\beta_{12}=(\beta_1+1)/(1+\beta_1)=1$ : light stays light in every frame. Why non-collinear boosts are subtler: boosts along different axes do not commute, and their composition is a boost plus a spatial rotation (the Wigner rotation, source of Thomas precession) — a Special Relativity–track story. *Verified: $\beta_1=\beta_2=0.5$ gives $\beta_{12}=1.0/1.25=0.8$ , and via rapidities $\phi=2\,\mathrm{arctanh}\,0.5=1.0986$ , $\tanh(1.0986)=0.800$ — consistent; matches Goldstein Ch. 7.*

Limiting cases

Case	Result	Meaning
Low velocity, $\beta\ll1$	$\gamma\approx1+\tfrac12\beta^2$ , $x'\to x-vt$ , $t'\to t$	Galileo recovered; the $\tfrac12\beta^2$ correction seeds the relativistic kinetic-energy expansion in the four-momentum article.
$c\to\infty$	boost $\to$ Galilean transformation exactly	Newtonian mechanics is relativity with infinite signal speed; absolute simultaneity restored.
Ultra-relativistic, $\beta\to1$	$\gamma\to\infty$ , $\phi\to\infty$	Axes fold onto the light cone; finite-energy frames never reach it.
Null interval, $s^2=0$	$x=\pm ct$ in every frame	Invariance of the light cone IS the second postulate, encoded geometrically.
Composition with light, $\beta_2=1$	$\beta_{12}=1$	Velocity addition saturates at $c$ : light is the fixed point of all boosts.
Perpendicular directions	$y'=y$ , $z'=z$	Transverse lengths never contract — only the boost direction mixes with time.

Common trapTime dilation is real but "moving clocks run slow" without the simultaneity term is a contradiction machine

The single most common error is applying $\Delta t=\gamma\Delta\tau$ or $L=L_0/\gamma$ while silently assuming both frames agree on what happens "at the same time." They never do: events simultaneous in $S$ are separated by $\Delta t'=-\gamma v\Delta x/c^2$ in $S'$ (eq. 7). Each observer says the other's clocks run slow, and both are right — because each is comparing one of their own clocks against a different pair of the other's clocks, synchronized by a different convention. Drop the $-vx/c^2$ term and you will "prove" contradictions all night. Keep it, and every paradox in the next article dissolves.

Worked examples

Example 1 — transforming an event and checking the invariant (confidence builder). In frame $S$ an event occurs at $x=240\ \mathrm{m}$ , $t=1.00\ \mu\mathrm{s}$ (so $ct=300\ \mathrm{m}$ , taking $c=3.00\times10^8\ \mathrm{m/s}$ ). Find its coordinates in a frame $S'$ moving at $v=0.6c$ along $+x$ , and verify the interval is unchanged.

Step 1. Lorentz factor: $\beta=0.6$ , so $\gamma = 1/\sqrt{1-0.36} = 1/\sqrt{0.64} = 1.25$ .

Step 2. Space coordinate: $vt = (1.8\times10^8)(1.0\times10^{-6}) = 180\ \mathrm{m}$ , so $x' = 1.25\,(240-180) = 75\ \mathrm{m}$ .

Step 3. Time coordinate, most cleanly in the $ct$ form: $ct' = \gamma(ct - \beta x) = 1.25\,(300 - 0.6\times240) = 1.25\times156 = 195\ \mathrm{m}$ , i.e. $t' = 0.65\ \mu\mathrm{s}$ .

Step 4. Invariant check: in $S$ , $s^2 = 300^2 - 240^2 = 90000-57600 = 32400\ \mathrm{m}^2$ ; in $S'$ , $195^2 - 75^2 = 38025 - 5625 = 32400\ \mathrm{m}^2$ . The event maps to $(ct',x')=(195\ \mathrm{m},\,75\ \mathrm{m})$ and the interval $s^2=32400\ \mathrm{m}^2$ ( $s=180\ \mathrm{m}$ , timelike) is identical in both frames.

Example 2 — a two-stage boost and a simultaneity trap (exam level). A ship leaves Earth at $0.8c$ . It launches a probe forward at $0.8c$ relative to the ship. (a) Find the probe's speed in the Earth frame, by velocity addition and again by rapidity. (b) Two beacons at rest on Earth's axis, $3.0\ \mathrm{km}$ apart, flash simultaneously in the Earth frame; find the time between flashes as measured by a frame moving at $0.6c$ along the beacon line.

Step 1. Velocity addition with $\beta_1=\beta_2=0.8$ : $\beta_{12} = (0.8+0.8)/(1+0.64) = 1.6/1.64 = 0.9756$ . The probe moves at $0.9756\,c$ — not $1.6c$ .

Step 2. Rapidity route: $\phi_1 = \mathrm{arctanh}\,0.8 = \tfrac12\ln\frac{1.8}{0.2} = \tfrac12\ln9 = 1.0986$ . Total $\phi = 2.1972$ , and $\tanh 2.1972 = (9-1/9)/(9+1/9) = (8.889)/(9.111) = 0.9756$ — same answer, one addition.

Step 3. The combined Lorentz factor, useful downstream: $\gamma_{12} = \cosh 2.1972 = \tfrac12(e^{2.1972}+e^{-2.1972}) = \tfrac12(9 + 0.1111) = 4.556$ . (Check: $1/\sqrt{1-0.9756^2} = 1/\sqrt{0.0482} = 4.56$ ✓.)

Step 4. Simultaneity part with $\beta=0.6$ , $\gamma=1.25$ , $\Delta x = 3000\ \mathrm{m}$ , $\Delta t = 0$ : equation (7) gives $\Delta t' = -\gamma\beta\,\Delta x/c = -1.25\times0.6\times3000/(3\times10^8) = -7.5\times10^{-6}\ \mathrm{s}$ . (a) The probe's Earth-frame speed is $0.9756\,c$ with $\gamma=4.56$ ; (b) the flashes are $7.5\ \mu\mathrm{s}$ apart in the moving frame, the forward beacon flashing first.

Origins · 1 minHendrik Lorentz and how this was discovered

Hendrik Lorentz

1853–1928 · Leiden University

Lorentz was the reigning master of electromagnetic theory in the decades before relativity, and the transformation bears his name because he found it first — as a property of Maxwell's equations. Seeking to explain Michelson–Morley within aether theory, he discovered (1892–1904, partly with FitzGerald's contraction hypothesis) that Maxwell's equations keep their form under exactly the coordinate change of equation (1), introducing a "local time" $t-vx/c^2$ he regarded as a mathematical convenience. Einstein's 1905 leap was to read those formulas as the actual structure of space and time, derivable from two postulates with no aether at all. Lorentz, gracious about being superseded, chaired the 1911 Solvay Conference and remained the elder statesman of physics; Einstein called him "the greatest and noblest man of our times." The first Dutch Nobel laureate in physics (1902, with Zeeman), he spent his last years on both electron theory and, characteristically Dutch, the flood-control calculations for the Zuiderzee enclosure dam.

History

1865

— Maxwell's equations predict electromagnetic waves at a fixed speed c, silently incompatible with Galilean relativity

1887

— Michelson and Morley find no motion of Earth through the aether, to a few parts in a billion in fringe shift

1889

— FitzGerald, then independently Lorentz (1892), propose motion-induced length contraction to save the aether

1904

— Lorentz publishes the full transformation under which Maxwell's equations are form-invariant; Poincaré names it after him

1905

— Einstein derives the same transformation from two postulates, discarding the aether; simultaneity becomes relative

1908

— Minkowski recasts the theory as the geometry of a four-dimensional spacetime with an invariant interval

Going further

References. Morin, Introduction to Classical Mechanics, Ch. 11 · Goldstein, Classical Mechanics (3rd ed.), Ch. 7 · Landau & Lifshitz, The Classical Theory of Fields (Vol. 2), §§1–4 · Taylor, Classical Mechanics, Ch. 15 · Taylor & Wheeler, Spacetime Physics

Provenance. Postulates and boost derivation follow Einstein (1905) as presented in Morin Ch. 11 and Landau-Lifshitz CTF §§1-4. Verifier should: (1) confirm $\gamma^2(1-\beta^2)=1$ and interval invariance algebraically; (2) check the Example-1 numbers ( $\gamma(0.6)=1.25$ ; $300^2-240^2=195^2-75^2=32400$ ); (3) check velocity addition $0.8\oplus0.8=0.9756$ and $\cosh(2\,\mathrm{arctanh}\,0.8)=4.56$ ; (4) muon and GPS figures ( $\tau=2.2\,\mu$ s, $\gamma\approx16$ at $\beta=0.998$ ; $-7\,\mu$ s/day at 3.9 km/s) against Taylor & Wheeler. No nonstandard numbers introduced.

Scope note. This chapter gives mechanics' compact introduction. The full story — paradox taxonomy, relativistic optics and aberration, Wigner rotation and the general Lorentz group — lives in the Special Relativity track (migrates later as its own subject).

Builds on: The One-Dimensional Wave Equation · Galilean relativity and inertial frames Leads to: Relativistic Kinematics: Time Dilation, Length Contraction, and Velocity Addition · Four-Vectors and Minkowski Spacetime

Both observers say the other's clock runs slow. How can that possibly be consistent?

Because they are not making the same comparison. To rate a single moving clock you must compare it against two of your own clocks at different places — and those two clocks are synchronized by your convention, not the other observer's. The moving observer rejects your synchronization (equation 7 quantifies the disagreement), so she is entirely consistent in finding your clocks slow by the same factor. The symmetric situation only breaks if someone turns around: then one worldline is non-inertial, the comparison becomes one clock against one clock, and the elapsed proper times genuinely differ — that is the twin analysis of the next article.

Is the Lorentz contraction something that physically happens to a rod, or just an artifact of measurement?

Neither phrasing survives scrutiny. The rod's proper length — measured in its rest frame — never changes, so nothing "happens" to it. But the frame-dependent length is not an illusion either: it is the honest answer to a well-posed operational question, the distance between simultaneous positions of the two ends, and simultaneity itself is frame-dependent. The cleanest language: length, like the time between ticks, is a projection of an invariant four-dimensional object onto a particular frame's slicing, just as the width of a tilted book's shadow depends on the light's direction while the book stays rigid.

Why is the transformation required to be linear? Could a nonlinear map also preserve the speed of light?

Nonlinear maps preserving light speed exist (conformal transformations), but they violate homogeneity: a nonlinear transformation treats some spacetime points differently from others, so identical experiments performed at different places or times would transform differently. Operationally, a free particle must move uniformly in every inertial frame — straight worldlines must map to straight worldlines, globally — and the only invertible maps sending all lines to lines are affine; with a common origin, linear. That is why Derivation A can begin with the four-coefficient ansatz without loss of generality.

Where exactly does the Galilean transformation go wrong, since it seems so obviously true?

It encodes an assumption so intuitive nobody noticed it: that simultaneity is absolute ( $t'=t$ ), equivalent to signals of unbounded speed. All everyday experience involves $\beta\lesssim10^{-6}$ , where the corrections of order $\beta^2$ ( $\gamma-1\approx5\times10^{-13}$ for a jet plane) are beneath perception, so intuition trained at low speed reports Galileo as exact. The failure is not in logic but in extrapolation. Maxwell's equations were the first piece of physics honest at high speed, which is why the conflict surfaced through electromagnetism rather than mechanics.

Does the invariant speed have to be the speed of light? What is so special about light?

Nothing, and this is worth internalizing. Relativity's postulates really say: there exists one invariant speed $c$ , finite or infinite. Homogeneity, isotropy and the relativity principle alone force the transformation to be either Galilean ( $c\to\infty$ ) or Lorentzian with some invariant speed; experiment then selects finite $c$ and measures its value. Light matters historically because photons are massless, so they travel at the invariant speed. If the photon had a tiny mass, light would move slightly below $c$ and every equation of this article would survive untouched — $c$ is a property of spacetime, not of light.

If nothing can go faster than c, what forbids me from boosting to c by composing many small boosts?

Rapidity bookkeeping answers this instantly. Each rocket stage adds a fixed rapidity increment $\Delta\phi=\mathrm{arctanh}(\Delta v/c)$ , and rapidities add without bound — but the speed is $\beta=\tanh\phi$ , which approaches 1 asymptotically and never reaches it for any finite $\phi$ . Physically, each successive boost is applied in an already-moving frame, and velocity addition (6) shrinks its lab-frame effect by the factor $(1-u v/c^2)$ -ish denominators. The four-momentum article restates this as energetics: $\gamma\to\infty$ means infinite energy would be needed for the last, unreachable step.

Do Lorentz transformations apply if my frame is accelerating?

The transformation relates inertial frames only — that is in the postulates. But special relativity handles accelerating objects perfectly well: at each instant an accelerating particle has a momentarily comoving inertial frame, and stitching these together (an integral over boosts) is how proper time along an arbitrary worldline is defined in the four-vector article. What acceleration breaks is only the global frame: an accelerated observer cannot cover spacetime with one rigid grid of synchronized clocks. Gravity is the separate case — locally it is indistinguishable from acceleration, and promoting that observation to a principle leads out of special relativity entirely, toward general relativity.

Is the gamma factor ever measured directly, or is all of this inference from light experiments?

Directly, daily, and to high precision. Storage-ring experiments compare the decay rate of circulating muons against muons at rest: at $\gamma=29.3$ (CERN's classic g−2 ring), lifetimes stretched from $2.20\ \mu\mathrm{s}$ to $64.4\ \mu\mathrm{s}$ , matching $\gamma\tau$ to within experimental error of a part in a thousand. Ives–Stilwell-type experiments read the time-dilation Doppler shift in fast ion beams at the $10^{-8}$ level. And every accelerator's beam-transport system would simply steer wrong if $\gamma$ deviated from $(1-\beta^2)^{-1/2}$ — a several-TeV proton is engineering-grade evidence, renewed every day the LHC runs.