Relativistic Velocity Addition

Velocity is a derived quantity — a spatial displacement divided by a time interval — and special relativity does something violent to both halves of that ratio. Because a Lorentz boost mixes space into time and time into space, two observers in relative motion disagree about how far a particle moved and about how long the motion took, so they cannot possibly agree on the naive Galilean sum of velocities. The relativistic composition law is exactly the bookkeeping that reconciles their measurements, and its single most important consequence is a ceiling: no chain of subluminal boosts, however long, ever reaches the speed of light. Set up the standard configuration. A frame $S'$ glides at speed $v$ along the shared $x$ -axis of a frame $S$ . A particle that moves with velocity $\mathbf{u}'$ in $S'$ is seen to move with velocity $\mathbf{u}$ in $S$ , and the relation between the two follows by transforming the displacement differentials $\mathrm{d}t,\mathrm{d}x,\mathrm{d}y,\mathrm{d}z$ under the boost and taking ratios. The longitudinal component obeys the famous one-line formula; the transverse components pick up a factor of $1/\gamma$ that has no Newtonian analogue and that is the seed of relativistic aberration. The deepest structural fact is that this law is the $\tanh$ of something linear. Assign every boost a rapidity $\phi=\operatorname{artanh}(v/c)$ , and collinear boosts add their rapidities exactly the way rotations add their angles. Since $\tanh$ saturates at $1$ , adding rapidities forever leaves the velocity forever short of $c$ . The composition law is therefore not a new postulate but a theorem of the Lorentz transformations, equivalent to the invariance of the spacetime interval, and its fingerprint runs through particle kinematics, astrophysical jets, and the aberration of starlight.

Key takeaways

Galilean addition $u=u'+v$ is replaced by $u=\dfrac{u'+v}{1+u'v/c^{2}}$ , which keeps every composition of subluminal speeds strictly below $c$ .
The law is a direct consequence of the Lorentz transformation of displacement differentials — it already contains time dilation and the relativity of simultaneity in its denominator.
Set $u'=c$ and the formula returns $u=c$ for every boost $v$ : the invariance of light speed is built into the algebra, not added by hand.
In rapidity variables the rule becomes purely additive, $\phi_{\text{tot}}=\phi_1+\phi_2$ , because a boost is a hyperbolic rotation of spacetime.
Transverse velocity components change even though the boost never touches that direction — a $1/\gamma$ factor that produces aberration and the headlight effect.
Key idea: velocities do not add — their rapidities do. The longitudinal law $u=(u'+v)/(1+u'v/c^{2})$ is the velocity-space image of $\phi_{\text{tot}}=\phi_1+\phi_2$ , and because $\tanh$ saturates at $1$ , no finite chain of boosts ever attains $c$ .

The core insight

Velocities do not add — their rapidities do; the longitudinal composition $u=(u'+v)/(1+u'v/c^{2})$ is the $\tanh$ of the additive law $\phi_{\text{tot}}=\phi_1+\phi_2$ , so $c$ is an unreachable ceiling rather than an imposed limit.

Where this lives today

At the Large Hadron Collider the composition law is not an exotic correction but the everyday coordinate. Because collinear boosts add in rapidity, physicists label particles by their rapidity $y=\tfrac12\ln[(E+p_z c)/(E-p_z c)]$ — and its massless approximation, the pseudorapidity $\eta=-\ln\tan(\theta/2)$ — precisely so that a boost along the beam axis becomes a mere additive shift. Differences in rapidity, and therefore the shapes of distributions like $\mathrm{d}N/\mathrm{d}\eta$ , are invariant under longitudinal boosts, which is what lets a detector fixed in the lab frame report physics that would look identical in the colliding-parton frame. Every jet-clustering algorithm and every angular cut is written in this language. The same law polices the sky. Very-long-baseline interferometry routinely watches blazar and quasar jets — the plasma outflows of M87 and 3C 279 among them — appear to stream across the sky at several times $c$ . The superluminal motion is an illusion of light-travel-time geometry for a source aimed almost at us, giving an apparent transverse speed $\beta_{\text{app}}=\beta\sin\theta/(1-\beta\cos\theta)$ that can exceed unity while the true composed velocity stays strictly below $c$ , exactly as the closure theorem demands. Measuring $\beta_{\text{app}}$ across many sources is now a standard way to bound the real Lorentz factors of relativistic jets — a working astrophysical tool built directly on the fact that velocities compose, and cap, the way they do.

The physical picture

Walk forward at $u'$ inside a train that rolls at $v$ and Newton promises the ground sees you at $u'+v$ . Relativity intrudes twice. The ground observer disagrees with the train observer about how far your stride carried you (length contraction) and about how long it lasted (time dilation), and — most subtly — about whether the start and end of your stride even happened at comparable instants (relativity of simultaneity). The denominator $1+u'v/c^{2}$ is precisely the simultaneity correction: it is the factor that reconciles the two observers' clocks, and it grows exactly fast enough to prevent the sum from ever crossing $c$ . The first figure makes the ceiling literal. In Galilean bookkeeping two speeds of $0.8c$ pile up to $1.6c$ , sailing straight through the light barrier into a forbidden region. The relativistic law folds that same pair down to $0.976c$ — closer to the wall, but on the legal side of it. The speed limit is not a rule stapled on afterward; it is what the formula does.

Figure

Adding 0.8c to 0.8c: the Galilean sum 1.6c crashes through the light barrier into the forbidden zone, while the relativistic composition lands at 0.976c, just short of the ceiling c

Why the law never crosses the wall is clearest in the variable that boosts actually add: the rapidity $\phi=\operatorname{artanh}(v/c)$ . Map every speed onto its rapidity and collinear boosts simply sum their $\phi$ — the hyperbolic-angle analogue of stacking rotations. But the observable velocity is $u=c\tanh\phi$ , and $\tanh$ flattens toward $1$ no matter how large its argument grows. You may pour rapidity in without bound; the velocity only creeps asymptotically toward the light cone. The second figure plots this saturation and marks the real numbers: a rapidity of $\phi=1.10$ is $0.8c$ , and doubling it to $\phi=2.20$ (two $0.8c$ boosts stacked) yields $0.976c$ , not $1.6c$ .

Figure

Velocity saturates as rapidity grows: u = c·tanh φ climbs toward the ceiling u = c but never touches it — φ = 1.10 gives 0.8c, and stacking two such boosts (φ = 2.20) gives 0.976c, not 1.6c

The core equations

The whole subject is one differential ratio taken in two frames. Because a boost along $x$ transforms $\mathrm{d}x$ and $\mathrm{d}t$ together but leaves $\mathrm{d}y$ and $\mathrm{d}z$ alone, the velocity component along the boost obeys a different rule from the components across it. Start with the longitudinal law, the workhorse of the whole article.

u_x = \frac{u'_x + v}{1 + \dfrac{u'_x\, v}{c^{2}}} \qquad (1)

where: $u_x$ — the particle's $x$ -velocity measured in $S$ ; $u'_x$ — its $x$ -velocity measured in $S'$ ; $v$ — the speed of $S'$ relative to $S$ along $x$ ; $c$ — the invariant speed of light. The numerator is the Galilean guess; the denominator is the relativistic correction that keeps $|u_x|<c$ and reduces to $1$ when both speeds are tiny. A velocity perpendicular to the boost does not survive untouched, even though no boost acts along $y$ . The reason is entirely in the clock: the transverse displacement is unchanged but the elapsed time is dilated and simultaneity-shifted, so the ratio changes.

u_y = \frac{u'_y}{\gamma\left(1 + \dfrac{u'_x\, v}{c^{2}}\right)},\qquad \gamma=\frac{1}{\sqrt{1-v^{2}/c^{2}}} \qquad (2)

where: $u_y$ — transverse velocity in $S$ ; $u'_y$ — transverse velocity in $S'$ ; $\gamma$ — the boost Lorentz factor. The extra $1/\gamma$ has no Newtonian counterpart; it is what tilts the direction of a moving particle's velocity between frames, i.e. aberration. For a boost in an arbitrary direction the two rules fuse into a single vector statement, splitting $\mathbf{u}'$ into parts along and across $\mathbf{v}$ .

\mathbf{u} = \frac{1}{1+\dfrac{\mathbf{u}'\cdot\mathbf{v}}{c^{2}}}\left[\frac{\mathbf{u}'}{\gamma} + \mathbf{v} + \frac{1}{c^{2}}\,\frac{\gamma}{\gamma+1}(\mathbf{u}'\cdot\mathbf{v})\,\mathbf{v}\right] \qquad (3)

where: $\mathbf{u}$ — velocity in $S$ ; $\mathbf{u}'$ — velocity in $S'$ ; $\mathbf{v}$ — the boost velocity of $S'$ relative to $S$ . The awkward third term is what makes composition non-commutative: swap $\mathbf{u}'$ and $\mathbf{v}$ and you get a different vector, differing by a rotation (the Wigner rotation). The mess of equation (3) dissolves the moment you switch to rapidity. Define it through $\tanh$ and collinear composition becomes ordinary addition.

v = c\tanh\phi,\qquad \phi = \operatorname{artanh}\!\left(\frac{v}{c}\right),\qquad \phi_{\text{tot}} = \phi_1 + \phi_2 \qquad (4)

where: $\phi$ — the rapidity, the hyperbolic boost angle; $\gamma=\cosh\phi$ and $\gamma v/c=\sinh\phi$ . Rapidity runs over the whole real line while velocity is trapped in $(-c,c)$ ; the map between them is $\tanh$ , and its saturation is the speed limit. Feed a photon into the longitudinal law and the second postulate falls out as an identity — light is measured at $c$ in every frame.

u'_x = c \;\Longrightarrow\; u_x = \frac{c+v}{1+v/c} = c \quad \text{for all } v \qquad (5)

where: the result is independent of $v$ . This is not an assumption imported into the derivation; it is a fixed point of the composition map, the algebraic face of "you cannot catch light". Applying the law to the direction of a light ray rather than its speed gives the aberration formula — the tilt of the stars as the observer's velocity changes.

\cos\theta = \frac{\cos\theta' + \beta}{1 + \beta\cos\theta'},\qquad \beta=\frac{v}{c} \qquad (6)

where: $\theta$ — the ray's angle to the boost axis in $S$ ; $\theta'$ — the same angle in $S'$ ; $\beta$ — the dimensionless boost speed. As $\beta\to1$ every ray swings forward and the sky's light concentrates into a cone ahead of the observer — the headlight effect.

Derivation AThe composition law from the Lorentz transformation

The goal is to derive both the longitudinal and the transverse rules from a single source — the differential form of the Lorentz boost — and to show along the way exactly where the celebrated denominator comes from and why the transverse component carries a $1/\gamma$ that the longitudinal one does not. Everything reduces to writing the four spacetime displacements in $S$ in terms of those in $S'$ and taking the ratio $\mathrm{d}x/\mathrm{d}t$ (or $\mathrm{d}y/\mathrm{d}t$ ) that defines a velocity, so the law is a theorem, not a new postulate.

Step 1. Write the inverse boost on infinitesimal displacements: $\mathrm{d}t=\gamma(\mathrm{d}t'+v\,\mathrm{d}x'/c^{2})$ , $\mathrm{d}x=\gamma(\mathrm{d}x'+v\,\mathrm{d}t')$ , $\mathrm{d}y=\mathrm{d}y'$ , $\mathrm{d}z=\mathrm{d}z'$ . Why: the Lorentz transformation is linear with constant coefficients, so it acts on differentials identically to how it acts on coordinates — this makes the derivation exact, not a small-displacement approximation.

Step 2. Form the longitudinal ratio and divide top and bottom by $\mathrm{d}t'$ : $u_x=\mathrm{d}x/\mathrm{d}t=(\mathrm{d}x'+v\,\mathrm{d}t')/(\mathrm{d}t'+v\,\mathrm{d}x'/c^{2})=(u'_x+v)/(1+u'_x v/c^{2})$ . Why: the common factor $\gamma$ cancels between numerator and denominator, and what remains — the denominator $1+u'_x v/c^{2}$ — is the imprint of relativity of simultaneity, the term $v\,\mathrm{d}x'/c^{2}$ that a moving clock adds to the elapsed time.

Step 3. Form the transverse ratio: $u_y=\mathrm{d}y/\mathrm{d}t=\mathrm{d}y'/[\gamma(\mathrm{d}t'+v\,\mathrm{d}x'/c^{2})]=u'_y/[\gamma(1+u'_x v/c^{2})]$ . Why: now $\mathrm{d}y=\mathrm{d}y'$ contributes no compensating $\gamma$ in the numerator, so the $\gamma$ from the dilated denominator survives — a transverse speed is genuinely reduced between frames, which is the origin of aberration.

Step 4. Check the null cone: substitute $u'^{2}_x+u'^{2}_y+u'^{2}_z=c^{2}$ into the three composed components and confirm $u^{2}_x+u^{2}_y+u^{2}_z=c^{2}$ . Why: a light ray must remain a light ray in every frame, so the map has to send the sphere $|\mathbf{u}'|=c$ to itself — the fact that it does is equivalent to the invariance of the spacetime interval, and it verifies the whole construction.

Go deeper · Derivation B · 1 minThe speed of light as an unattainable ceiling

Derivation BThe speed of light as an unattainable ceiling

The goal is to prove, with nothing but algebra, that composing any two speeds strictly below $c$ produces a speed strictly below $c$ , so that $c$ is an unattainable supremum reached only when one of the inputs is already light. Working in dimensionless variables $a=u'/c$ and $b=v/c$ , each confined to the open interval $(-1,1)$ , we show the composed value stays inside the same interval by writing the "distance to the ceiling" as a manifestly positive product.

Step 1. Compute the deficit below $c$ : $1-\dfrac{a+b}{1+ab}=\dfrac{1+ab-a-b}{1+ab}=\dfrac{(1-a)(1-b)}{1+ab}$ . Why: factoring the numerator into $(1-a)(1-b)$ turns an opaque inequality into a question about the signs of three simple quantities.

Step 2. Read off the signs. With $a<1$ and $b<1$ both factors $(1-a)$ and $(1-b)$ are positive, and with $a,b>-1$ the denominator $1+ab>0$ . Why: a positive deficit means $\dfrac{a+b}{1+ab}<1$ — the composed speed is strictly less than $c$ , for every legal pair of inputs.

Step 3. Bound from below the same way: $1+\dfrac{a+b}{1+ab}=\dfrac{(1+a)(1+b)}{1+ab}>0$ , so the composed value exceeds $-1$ . Why: combined with Step 2 this traps the result inside $(-1,1)$ , i.e. the composed speed magnitude is below $c$ in both directions — the interval is closed under composition.

Step 4. Locate equality: setting $a=1$ gives $\dfrac{1+b}{1+b}=1$ . Why: the ceiling is touched only when an input already equals $c$ — the photon case of equation (5) — which is why a massive particle, always starting below $c$ , can never be boosted up to it no matter how many times you add velocity.

Go deeper · Derivation C · 1 minRapidity, and why boosts add like angles

Derivation CRapidity, and why boosts add like angles

The goal is to expose the hidden linearity behind the nonlinear composition law by showing that a one-dimensional boost is a hyperbolic rotation, parameterized by a single additive angle. Once boosts are written as matrices $\Lambda(\phi)$ built from $\cosh\phi$ and $\sinh\phi$ , composing two boosts is just multiplying two such matrices, and the hyperbolic addition formulas collapse the product into a single boost whose rapidity is the sum — with the velocity law reappearing as the $\tanh$ of that sum.

Step 1. Parameterize a boost of the $(ct,x)$ plane by $\beta=\tanh\phi$ , so that $\gamma=\cosh\phi$ and $\gamma\beta=\sinh\phi$ , and the boost matrix is $\Lambda(\phi)=\begin{pmatrix}\cosh\phi & \sinh\phi\\ \sinh\phi & \cosh\phi\end{pmatrix}$ . Why: this is the hyperbolic twin of a rotation matrix — it preserves $c^{2}t^{2}-x^{2}$ exactly as a rotation preserves $x^{2}+y^{2}$ — so a boost is literally a rotation through an imaginary angle.

Step 2. Multiply two boosts and use $\cosh(\phi_1+\phi_2)=\cosh\phi_1\cosh\phi_2+\sinh\phi_1\sinh\phi_2$ and the matching sine identity to get $\Lambda(\phi_1)\Lambda(\phi_2)=\Lambda(\phi_1+\phi_2)$ . Why: the product of two boosts is a single boost whose rapidity is the plain sum $\phi_1+\phi_2$ — collinear boosts form a one-parameter additive group, the cleanest possible statement of "how velocities combine".

Step 3. Recover the velocity law by taking $\tanh$ of the sum: $\beta_{\text{tot}}=\tanh(\phi_1+\phi_2)=\dfrac{\tanh\phi_1+\tanh\phi_2}{1+\tanh\phi_1\tanh\phi_2}=\dfrac{\beta_1+\beta_2}{1+\beta_1\beta_2}$ . Why: equation (1) is nothing but the $\tanh$ addition identity in disguise, which is why the messy velocity rule hides a trivial additive one — and why $\beta_{\text{tot}}<1$ always, since $\tanh$ of any finite number is below $1$ .

Limiting cases

The one formula spans everyday intuition and the rigid relativistic ceiling; the table reads off its behaviour in the four regimes that matter.

Case	Result	Meaning
Galilean limit	$u \approx u' + v - \dfrac{u'v(u'+v)}{c^{2}}$	For $u',v\ll c$ the denominator $\to 1$ and Newton is recovered; the leading correction is of order $u'v/c^{2}$ , about $10^{-16}$ for a jet on a runway.
Light invariance	$u'=c \;\Rightarrow\; u=\dfrac{c+v}{1+v/c}=c$	Any speed composed with $c$ returns $c$ ; the second postulate is a fixed point of the map, holding for every boost $v$ .
Near-light saturation	$u'=v=0.9c \;\Rightarrow\; u=\dfrac{1.8c}{1.81}=0.9945c$	Two large subluminal speeds give a result nearer the wall but still below it — the ceiling is approached, never crossed.
Rapidity additivity	$\phi=\phi_1+\phi_2,\quad u=c\tanh\phi$	Collinear boosts form a one-parameter additive group; velocity is the bounded $\tanh$ shadow of the unbounded, freely-adding rapidity.

Common trapVelocity addition is one-dimensional, so you can always just plug speeds into (u'+v)/(1+u'v/c²)

This is the single most common error, and it fails in two distinct ways. First, a velocity perpendicular to the boost does not obey equation (1) at all — it is reduced by the factor $1/[\gamma(1+u'_x v/c^{2})]$ , which is why a particle's direction, not just its speed, changes between frames. Second, in two or three dimensions the composition is neither commutative nor associative: composing two non-collinear boosts yields a boost plus a rotation, the Wigner rotation, whose accumulated effect on a circulating electron is Thomas precession. The scalar formula is exact only when every velocity involved lies along the same line. Whenever directions differ, use the full vector law (3) or, better, rapidity plus an explicit rotation — and never treat $\mathbf{u}'+\mathbf{v}$ as if it had any frame-independent meaning.

Worked examples

1 · Stacking boosts, and why they never reach c (numeric)

Two spacecraft leave a station in opposite directions, each at $0.8c$ as measured by the station. (a) How fast does one crew measure the other to recede? (b) A three-stage rocket fires stages that each add $0.6c$ relative to the stage before; what is its final speed? Cross-check both with rapidity.

Step 1. For part (a), work in the rest frame of ship A. The station moves at $0.8c$ relative to A, and ship B moves at $0.8c$ relative to the station in the same direction, so compose: $u=\dfrac{0.8c+0.8c}{1+(0.8)(0.8)}=\dfrac{1.6c}{1.64}=\mathbf{0.9756c}$ . The Galilean answer $1.6c$ is meaningless; the true relative speed sits below $c$ exactly as the closure theorem guarantees.

Step 2. Cross-check with rapidity. Each $0.8c$ boost is $\phi=\operatorname{artanh}(0.8)=1.0986$ . The two rapidities add: $\phi_{\text{tot}}=2.1972$ , and $\tanh(2.1972)=\mathbf{0.9756}$ , identical to Step 1. Rapidity turned a nonlinear composition into a one-line sum.

Step 3. For part (b), add three $0.6c$ boosts. Composing pairwise: $0.6c\oplus0.6c=\dfrac{1.2c}{1.36}=0.8824c$ , then $0.8824c\oplus0.6c=\dfrac{1.4824c}{1.5294}=\mathbf{0.9692c}$ . In rapidity this is trivial: $\phi=3\operatorname{artanh}(0.6)=3(0.6931)=2.0794$ , and $\tanh(2.0794)=0.9692$ — the same number.

Step 4. Push the pattern to its limit. To reach $0.99c$ you need $\phi=\operatorname{artanh}(0.99)=2.647$ , i.e. about $2.647/0.6931\approx\mathbf{3.8}$ stages of $0.6c$ ; to reach $c$ you would need infinite rapidity, hence infinitely many stages. Two $0.8c$ boosts give $0.976c$ ; three $0.6c$ boosts give $0.969c$ ; the ceiling $c$ demands infinite rapidity and is never reached.

2 · Fizeau's moving water and the Fresnel drag (numeric)

Light travels through water ( $n=1.33$ ) that itself flows at speed $v$ along the beam. (a) Derive the lab speed of the light from velocity addition and identify the drag coefficient. (b) Evaluate it for Fizeau's laboratory flow $v=7\ \mathrm{m/s}$ , and (c) for a hypothetical $v=0.5c$ to expose the limits of the first-order formula.

Step 1. In the water's rest frame light moves at $u'=c/n$ . Boosting to the lab where the water moves at $v$ , compose longitudinally: $u=\dfrac{c/n+v}{1+v/(nc)}$ . Expand for $v\ll c$ : $u\approx\dfrac{c}{n}+v\left(1-\dfrac{1}{n^{2}}\right)$ . The bracket $1-1/n^{2}$ is the Fresnel drag coefficient; the moving medium drags the light along, but only partially.

Step 2. Evaluate the coefficient for water: $1-1/n^{2}=1-1/1.7689=\mathbf{0.435}$ . So the water drags the light by only $43.5\%$ of its own speed — a prediction Fresnel made in 1818 to save the stationary-ether picture, and which velocity addition reproduces exactly as a first-order expansion.

Step 3. For Fizeau's flow $v=7\ \mathrm{m/s}$ the extra speed is $0.435\times7\approx\mathbf{3.0\ m/s}$ on top of $c/n=2.25\times10^{8}\ \mathrm{m/s}$ . That fractional change is $\sim10^{-8}$ , invisible to any speedometer — which is why Fizeau measured it interferometrically, as a fringe shift between beams sent with and against the current.

Step 4. For the extreme $v=0.5c$ the full formula gives $u=\dfrac{0.7519c+0.5c}{1+0.5\times0.7519}=\dfrac{1.2519c}{1.3760}=\mathbf{0.9098c}$ , whereas the first-order Fresnel estimate $c/n+0.5c(0.435)=0.969c$ overshoots badly. The Fresnel drag $1-1/n^{2}=0.435$ is only the linear term; at relativistic flow speeds the exact composition law (giving $0.910c$ , safely below $c$ ) must be used.

Physics of the portrait

Velocity composition is where relativity first bit hard: the moment you accept that light outruns you at $c$ no matter how fast you chase it, the whole Galilean law of adding speeds has to go, and something bounded must take its place.

Origins · 1 minEinstein and how this was discovered

Einstein

1879–1955 · special relativity and the composition of velocities, 1905

Albert Einstein derived the velocity-composition law in the fifth section of his 1905 paper "On the Electrodynamics of Moving Bodies", the same paper that introduced special relativity. He did not reach it by fitting data or patching the ether; he deduced it, treating the two postulates — the principle of relativity and the invariance of $c$ — as premises and working out what addition of velocities must therefore look like. The result, $u=(u'+v)/(1+u'v/c^{2})$ , delighted him precisely because substituting $u'=c$ returned $c$ : the formula proved its own consistency with the second postulate. Einstein noted the "peculiar law" that two subluminal speeds always compose to a subluminal speed; that the same law reproduces Fresnel's drag coefficient, and so explains Fizeau's 1851 measurement of light dragged by moving water, was drawn out by Max von Laue in 1907 — a result that had puzzled physicists for half a century and that the formula accounts for without any ether at all. The rapidity reformulation, which makes the law a simple addition of hyperbolic angles, came shortly after through Minkowski's 1908 geometry and Varićak's 1910 hyperbolic kinematics; the non-commutativity of non-collinear boosts, latent in Einstein's vector formula, surfaced two decades later in Thomas precession. What began as a "peculiar law" tucked into one section of one paper became the kinematic backbone of every particle collider and every relativistic jet.

History

1728

— Bradley reports stellar aberration, an early velocity-composition effect in the sky

1851

— Fizeau measures the partial dragging of light by moving water

1905

— Einstein derives the velocity-composition law in "On the Electrodynamics of Moving Bodies"

1908

— Minkowski recasts Lorentz boosts as hyperbolic rotations of spacetime

1910

— Varićak formulates relativistic velocity addition as hyperbolic (Lobachevsky) geometry

1926

— Thomas explains atomic fine structure via the rotation from non-collinear boosts

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §2.9 · Wald, General Relativity (1984), Ch. 1 · Carroll, Spacetime and Geometry (2004), §1.2–1.4 · Schutz, A First Course in General Relativity (2009), Ch. 2 · Hartle, Gravity (2003), Ch. 4 · Jackson, Classical Electrodynamics (1999), §11.4. Key numerical benchmarks: $0.8c\oplus0.8c=1.6/1.64\,c=0.9756c$ ; $0.9c\oplus0.9c=1.8/1.81\,c=0.9945c$ ; three $0.6c$ stages give $3\operatorname{artanh}(0.6)=2.079\Rightarrow\tanh=0.9692$ ; the Fresnel drag coefficient for water ( $n=1.33$ ) is $1-1/n^{2}=0.435$ . All values are exact rational or hyperbolic evaluations of the composition law; the only physical constant invoked is $c$ .

Prerequisites. Lorentz Transformations · Postulates of Special Relativity · Time Dilation · Relativity of Simultaneity. Leads to. Lorentz Boosts and Rapidity · Four-Vectors · Causal Structure and Light Cones · Proper Time.

Ask the tutor: Ask me to stack any chain of boosts in rapidity, or to work out the aberration angle and headlight cone for an observer at a speed you choose.

Why don't velocities just add the way Galileo said?

Because velocity is displacement over time, and relativity changes both. When you measure a particle's speed you divide the distance it covered by the time it took — but two observers in relative motion disagree about that distance (length contraction) and about that time (time dilation), and they even disagree about whether the endpoints of the motion were "simultaneous" in the same way. The Galilean sum $u'+v$ silently assumes everyone shares one universal clock and ruler. Once those are frame-dependent, the sum must be corrected by the denominator $1+u'v/c^{2}$ , which encodes exactly the disagreement about simultaneity. At everyday speeds that correction is around one part in $10^{16}$ , so Galileo's rule is superb — just not exact.

What is rapidity, and why is it worth the trouble?

Rapidity $\phi=\operatorname{artanh}(v/c)$ is the "hyperbolic angle" of a boost: velocity is its $\tanh$ , so $v=c\tanh\phi$ . Its virtue is that collinear boosts add their rapidities, $\phi_{\text{tot}}=\phi_1+\phi_2$ , turning the awkward composition law into ordinary addition — precisely the way rotation angles add while their tangents don't. It also makes the speed limit obvious: rapidity ranges over the whole real line, but $\tanh$ squeezes it into $(-c,c)$ , so no finite rapidity ever reaches $c$ . Particle physicists live in this variable because a boost along the beam is just a constant shift in rapidity, leaving the shapes of measured distributions invariant.

If I chase a light beam at 0.99c, why does it still outrun me at c?

Set $u'=c$ in the composition law and the boost speed $v$ cancels completely: $u=(c+v)/(1+v/c)=c$ for every $v$ . So no matter how fast you chase, light recedes from you at exactly $c$ , never at $c$ minus your speed. This is not a coincidence of the algebra — it is the second postulate, the invariance of the speed of light, re-expressed as the statement that $c$ is a fixed point of the composition map. In rapidity language, light sits at infinite rapidity, and adding any finite rapidity to infinity leaves infinity: you can never close the gap.

Why does a sideways velocity change when the boost doesn't act on that direction?

Because velocity is a ratio, and the boost changes the denominator even when it leaves the numerator alone. A boost along $x$ genuinely does not touch the transverse displacement: $\mathrm{d}y=\mathrm{d}y'$ . But it does change the elapsed time, $\mathrm{d}t=\gamma(\mathrm{d}t'+v\,\mathrm{d}x'/c^{2})$ , through time dilation and the simultaneity shift. Since $u_y=\mathrm{d}y/\mathrm{d}t$ , the altered time reshapes the transverse velocity by the factor $1/[\gamma(1+u'_x v/c^{2})]$ . Physically this tilts the particle's direction of motion between frames, which is the mechanism of relativistic aberration — the reason a fast observer sees the stars crowd toward the direction of travel.

Are blazar jets really moving faster than light?

No — they only appear to. When a jet points almost at us and moves at, say, $0.98c$ , the light it emits later has only a short remaining distance to cover, so successive images arrive bunched in time. The sky-projected motion is compressed into a shorter apparent interval, faking a transverse speed $\beta_{\text{app}}=\beta\sin\theta/(1-\beta\cos\theta)$ that can exceed $1$ for small angles $\theta$ . The plasma's actual velocity, composed properly, stays strictly below $c$ , and no information or matter outruns light. This "superluminal motion" was first resolved in quasars in the 1970s and is now routinely mapped by VLBI; measuring $\beta_{\text{app}}$ is a standard way to estimate a jet's true Lorentz factor.

Is velocity composition commutative?

For collinear boosts, yes — $u'\oplus v=v\oplus u'$ , because it reduces to the symmetric one-line formula. But for boosts in different directions it is not. Composing a boost along $x$ with one along $y$ does not equal composing them in the opposite order; the two results differ by a rotation, the Wigner rotation. Equivalently, the product of two non-collinear boosts is not a pure boost but a boost combined with a rotation. This non-commutativity is not a technicality: accumulated around a closed path it produces Thomas precession, the relativistic twist that corrects the spin-orbit coupling in atomic fine structure by the famous factor of one-half.

How is the aberration of light connected to velocity addition?

Aberration is velocity addition applied to a photon's direction rather than its speed. Write the light ray in frame $S'$ as $u'_x=c\cos\theta'$ , $u'_y=c\sin\theta'$ , push both through the composition law, and form $\cos\theta=u_x/c$ ; the result is $\cos\theta=(\cos\theta'+\beta)/(1+\beta\cos\theta')$ . The speed stays $c$ in both frames — only the angle changes. As the observer's speed grows, every ray swings toward the forward direction, so the sky's light concentrates into a forward cone (the headlight effect), and the stars appear displaced toward the apex of one's motion. Bradley's 1728 stellar aberration is the small- $\beta$ limit of exactly this formula.