Inverse Lorentz Transformation and Reciprocity

Two spacecraft drift past each other in empty space. There is no wind, no landmark, no bank of the river — nothing that could tell either crew which of them is "the one moving". They agree on their relative speed. The only thing they disagree about is its direction: to me you are receding at $+v$ , and to you I am receding at $-v$ . That looks like a triviality, and every textbook slides past it in half a sentence: "the inverse transformation is obtained by replacing $v$ with $-v$ ." But it is not a triviality, and it is not a definition. It is a theorem, and it has to be proved from the forward transformation, because nothing in the algebra of $x' = \gamma(x - vt)$ says in advance what the origin of $S$ is doing when viewed from $S'$ . You have to go and look. When you do, you find that it recedes at exactly $-v$ — the same magnitude, no factor of $\gamma$ , no correction term — and only then are you licensed to invert the boost by flipping a sign. The stakes are higher than the tidiness of an algebra exercise. Reciprocity is what makes the Lorentz transformations a group: $\Lambda(v)\Lambda(-v) = I$ , so every boost has an inverse that is itself a boost. And it is what forces $\gamma$ to be an even function of $v$ . Suppose it were not — suppose I found your clocks slowed by $1.5$ while you found mine slowed by only $1.04$ . Both accounts can be made consistent with the constancy of light. What they cannot be made consistent with is the principle of relativity, because now there is a fact of the matter about which of us is really moving: whichever of us has the bigger factor. The ether would be back, wearing a false moustache.

Key takeaways

The inverse Lorentz transformation is the forward one with the single substitution $v \to -v$ : $ct = \gamma(ct' + \beta x')$ , $x = \gamma(x' + \beta\,ct')$ .
This is not an assumption. Track the origin of $S$ (the line $x = 0$ ) through the forward transformation and it comes out moving at $\mathrm{d}x'/\mathrm{d}t' = -v$ in $S'$ — reciprocity is a theorem, proved from the boost you already have.
Matrix form: $\Lambda(v)\Lambda(-v) = I$ and $\det\Lambda = +1$ , both resting on the single identity $\gamma^2(1-\beta^2) = 1$ . Boosts along one axis form a group, and they close: $\Lambda(\beta_1)\Lambda(\beta_2) = \Lambda(\beta_1 \oplus \beta_2)$ with $\oplus$ the velocity-composition law.
$v \to -v$ is not the same instruction as "swap primed and unprimed". You must do both, and the reason that works is precisely reciprocity.
The relativity principle forces $\gamma$ to be even, $\gamma(v) = \gamma(-v)$ . Light-speed constancy alone only gives $\gamma_+\gamma_-(1-\beta^2) = 1$ — a whole hyperbola of possibilities. Every point on it except the symmetric one hands you a preferred frame.
Key idea: the only physical difference between two inertial observers is the sign of their relative velocity. Reciprocity is that statement turned into an algebraic identity, and the identity is what makes the Lorentz transformations a group — which is what guarantees that "each sees the other's clock run slow" is a symmetry, not a contradiction.

The core insight

Reciprocity is the demand that the map from me to you and the map from you to me differ only by the direction of the relative velocity — same $\gamma$ , same functional form. It is equivalent to $\Lambda(-v) = \Lambda(v)^{-1}$ , and it is the algebraic content of "no inertial frame is preferred". Drop it and you can still satisfy the constancy of light; what you cannot do is stop an experiment from finding out which of us is really at rest.

Where this lives today

Reciprocity is the reason a particle physicist may work in whichever frame is convenient and trust the answer. Every collider analysis boosts back and forth between the laboratory frame and the centre-of-momentum frame dozens of times per event, and each of those round trips is an application of $\Lambda(-v)\Lambda(v) = I$ . If the inverse boost were anything other than the $v \to -v$ boost, a rest-frame quantity — a particle's rest mass, its proper lifetime, its decay angular distribution — could not be recovered unambiguously from lab-frame data, and invariant-mass reconstruction would collapse. When a detector reports a $Z$ boson at $91.19\ \mathrm{GeV}$ , it is reporting a number obtained by boosting measured four-momenta into a frame nobody occupied, and trusting the group to bring it back. The evenness of $\gamma$ is also, quietly, an experimental claim, and it has been tested. A hypothetical odd-in- $v$ term would produce a directional dependence in clock rates and in the speed of light, and modern rotating cryogenic optical-resonator experiments of the Michelson–Morley and Kennedy–Thorndike type bound such anisotropies at $\Delta c/c \lesssim 10^{-18}$ . Reciprocity is not merely elegant. It is one of the most precisely verified symmetries in physics.

The physical picture

Start with the thing that must be checked rather than assumed. The forward transformation tells you what happens to an event: give it $(t, x)$ in my frame and it returns $(t', x')$ in yours. It does not, on the face of it, tell you anything about how I am moving as seen from your frame. That is a separate question, and the answer has to be extracted. So extract it. I am at rest at the origin of $S$ , which means my worldline is the set of events with $x = 0$ , for every $t$ . Feed that line — and nothing else — into the forward boost: $ct' = \gamma(ct - \beta \cdot 0) = \gamma\,ct$ and $x' = \gamma(0 - \beta\,ct) = -\gamma\beta\,ct$ . Divide the second by the first and the $\gamma$ 's and the $t$ 's vanish together, leaving $x'/(ct') = -\beta$ . In your frame, I am at $x' = -v t'$ : I recede at exactly $v$ , in the opposite direction. Notice how easily this could have failed. Nothing in the structure of a linear map guarantees that the $\gamma$ cancels, or that the magnitude comes out equal — and in a theory with a preferred frame it would not. Reciprocity is a fact about the Lorentz transformation, discovered by inspection; once you have it, you may invert the boost by flipping a sign, and not before.

Figure

The round trip. An event at (ct, x) = (10, 4) m is pushed into a frame moving at β = 0.6 (γ = 1.25), where it reads (9.5, −2.5) m — a completely different pair of numbers, in a frame where the event is now on the NEGATIVE x-axis. Push it back with the SAME transformation and v → −v, and it lands exactly on (10, 4) again. Λ(v)Λ(−v) = I, to the last digit

The core equations

The forward boost is derived in The Lorentz Transformation Equations and taken as given. Writing $\beta = v/c$ and $\gamma = 1/\sqrt{1-\beta^2}$ , and putting time in metres so both coordinates carry the same units:

ct' = \gamma\,(ct - \beta x), \qquad x' = \gamma\,(x - \beta\,ct) \qquad (1)

where: $(ct, x)$ — an event in $S$ ; $(ct', x')$ — the same event in $S'$ , which moves at $+v$ along the $x$ -axis of $S$ . Physical reading: one event, two descriptions. Nothing here says anything about how $S$ looks from $S'$ ; that is the question this article answers.

ct = \gamma\,(ct' + \beta x'), \qquad x = \gamma\,(x' + \beta\,ct') \qquad (2)

where: the inverse. Physical reading: it is equation (1) with the primes exchanged and $\beta \to -\beta$ . Both operations, together. The exchange of primes says "now I am asking the mirror-image question"; the sign flip says "and in the mirror, the velocity points the other way". Doing only one of the two is the commonest error in the subject. In matrix form the structure becomes visible, and this is where the article earns its keep.

\Lambda(\beta) = \gamma\begin{pmatrix} 1 & -\beta \\ -\beta & 1 \end{pmatrix}, \qquad \Lambda(\beta)\,\Lambda(-\beta) = I, \qquad \det\Lambda(\beta) = +1 \qquad (3)

where: $\Lambda$ acts on the column $(ct, x)^{\mathsf T}$ . Physical reading: the inverse of a boost is a boost — not some other kind of transformation, not a boost plus a correction. That closure is what makes the set of boosts a group, and the group is the formal statement that no frame is special. $\det\Lambda = +1$ says a boost preserves spacetime volume and does not reflect: it cannot turn a right-handed frame into a left-handed one, or the future into the past. The group does more than invert. It closes on itself under composition, and the composition law is exactly relativistic velocity addition.

\Lambda(\beta_1)\,\Lambda(\beta_2) = \Lambda\!\left(\beta_1 \oplus \beta_2\right), \qquad \beta_1 \oplus \beta_2 = \frac{\beta_1 + \beta_2}{1 + \beta_1\beta_2} \qquad (4)

where: $\oplus$ — the velocity-composition law, owned by Relativistic Velocity Addition. Physical reading: two boosts in a row are a single boost, at the composed velocity — never a boost plus a rotation, provided both are along the same axis. Setting $\beta_2 = -\beta_1$ gives $\beta_1 \oplus (-\beta_1) = 0$ , so $\Lambda(\beta_1)\Lambda(-\beta_1) = \Lambda(0) = I$ : reciprocity is closure, evaluated at the inverse element. Why composition becomes plain addition in the rapidity variable is developed in Rapidity and the Hyperbolic Form of Boosts, and is not re-derived here. Finally, the identity on which all of the above rests, because everything cancels through it:

\gamma^{2}\left(1 - \beta^{2}\right) = 1 \qquad (5)

where: the definition of $\gamma$ , rearranged. Physical reading: this is the engine of every cancellation in the article. It is why $\Lambda(v)\Lambda(-v) = I$ , why $\det\Lambda = 1$ , and why the algebraic inversion of (1) closes at all. At $\beta = 0.6$ : $\gamma^2 = 1.5625$ and $1 - \beta^2 = 0.64$ , and $1.5625 \times 0.64 = 1$ exactly.

Reciprocity as a theorem

Derivation AReciprocity is proved, not assumed — the velocity of S in S′

It is tempting to declare reciprocity by fiat: "by symmetry, if $S'$ moves at $+v$ in $S$ then $S$ moves at $-v$ in $S'$ ." That is a plausible-sounding sentence, and it is also an unpaid promissory note. The forward transformation is a specific map with a specific $\gamma$ in it, and whether the origin of $S$ comes out moving at $-v$ , or at $-v/\gamma$ , or at $-\gamma v$ , is a question of fact, to be settled by computing. Only the forward boost, equation (1), is used below — no symmetry is assumed anywhere.

Step 1. Identify the worldline of $S$ 's spatial origin: the set of events with $x = 0$ , for all $t$ . This is what "at rest at the origin of $S$ " means — a definition, not a physical claim. Why: we need an object whose motion in $S$ is known exactly, so that whatever we find in $S'$ is attributable to the transformation and nothing else.

Step 2. Feed $x = 0$ into the forward boost. The two equations collapse to $ct' = \gamma\,(ct - 0) = \gamma\,ct$ and $x' = \gamma\,(0 - \beta\,ct) = -\gamma\beta\,ct$ . Why: $\gamma$ appears in both, which is exactly why it is about to disappear — a fact that could not have been anticipated without doing the calculation.

Step 3. Eliminate $t$ by dividing: $\dfrac{x'}{ct'} = \dfrac{-\gamma\beta\,ct}{\gamma\,ct} = -\beta$ . Hence $x' = -\beta\,ct' = -v\,t'$ , and differentiating, $\dfrac{\mathrm{d}x'}{\mathrm{d}t'} = -v$ . Why: the $\gamma$ 's cancel identically, so the speed of $S$ in $S'$ is $v$ exactly — not $\gamma v$ , not $v/\gamma$ . The magnitudes match, and only the sign differs. That is reciprocity, and it has now been derived.

Step 4. Draw the licence. Because the only physical difference between the two frames is the sign of the relative velocity, the principle of relativity says the transformation from $S'$ to $S$ must have the same functional form as the transformation from $S$ to $S'$ , with $v \to -v$ . Substituting into (1) gives exactly (2). Why: this is the payoff, and the order of reasoning matters. Reciprocity licenses the $v \to -v$ rule; the rule does not define reciprocity. *Verified in verify_aq-sr-inverse-lorentz-transformation-reciprocity.py: $\mathrm{d}x'/\mathrm{d}t' = -v$ is obtained symbolically from the forward boost with $x = 0$ , and is confirmed to be neither $-\gamma v$ nor $-v/\gamma$ ; the brute-force algebraic inversion of (1) is carried out and returns $(ct, x)$ identically in $\beta$ ; and $\Lambda(\beta)^{-1} = \Lambda(-\beta)$ is checked as a symbolic matrix identity.*

Go deeper · Derivation B · 2 minThe group: Λ(v)Λ(−v) = I, det Λ = +1, and closure under composition

Derivation BThe group: Λ(v)Λ(−v) = I, det Λ = +1, and closure under composition

Reciprocity is one identity in a family, and the family is what matters. The boosts along a fixed axis form a one-parameter group: they have an identity ( $\beta = 0$ ), every element has an inverse which is itself a boost, and the composition of any two is again a boost. That last property — closure — is what makes the inverse rule non-trivial. If two boosts could compose into something other than a boost, then "the inverse is the $v \to -v$ boost" would be a lucky accident rather than a structural fact, and no observer could be sure that undoing a boost was even possible with the tools available.

Step 1. Multiply the matrices. With $\Lambda(\beta) = \gamma\begin{pmatrix}1 & -\beta\\ -\beta & 1\end{pmatrix}$ and $\Lambda(-\beta) = \gamma\begin{pmatrix}1 & \beta\\ \beta & 1\end{pmatrix}$ , the product has diagonal entries $\gamma^2(1 - \beta^2)$ and off-diagonal entries $\gamma^2(\beta - \beta) = 0$ . By (5) the diagonal is $1$ , so $\Lambda(\beta)\Lambda(-\beta) = I$ . Why: the off-diagonal cancellation is exact and requires nothing; the diagonal one requires precisely the definition of $\gamma$ . That is the whole of the inverse rule, in one line of arithmetic.

Step 2. Take the determinant: $\det\Lambda = \gamma^2(1 \cdot 1 - \beta^2) = \gamma^2(1-\beta^2) = +1$ . Why: a proper Lorentz transformation preserves the spacetime volume element and does not reflect. Had the determinant been $-1$ , the boost would have swapped past for future or left for right, and no continuous change of velocity can do that.

Step 3. Compose two boosts along the same axis: $\Lambda(\beta_1)\Lambda(\beta_2) = \gamma_1\gamma_2 \begin{pmatrix}1 + \beta_1\beta_2 & -(\beta_1+\beta_2) \\ -(\beta_1+\beta_2) & 1 + \beta_1\beta_2\end{pmatrix}$ . Factor out $(1 + \beta_1\beta_2)$ : the matrix becomes $\gamma_1\gamma_2(1+\beta_1\beta_2)\begin{pmatrix}1 & -\beta_3 \\ -\beta_3 & 1\end{pmatrix}$ with $\beta_3 = \dfrac{\beta_1+\beta_2}{1+\beta_1\beta_2}$ . Why: the composed matrix has the shape of a boost. What remains is to check that its prefactor is the right $\gamma$ — because a boost matrix is not just any symmetric matrix; its prefactor is tied to its $\beta$ .

Step 4. Check the prefactor. One shows that $\gamma_1\gamma_2(1 + \beta_1\beta_2) = 1/\sqrt{1 - \beta_3^2} = \gamma_3$ . Hence $\Lambda(\beta_1)\Lambda(\beta_2) = \Lambda(\beta_3) = \Lambda(\beta_1 \oplus \beta_2)$ : the product of two boosts is a boost, at the relativistically composed velocity. Why: closure, and with it the group. And note the corollary — putting $\beta_2 = -\beta_1$ gives $\beta_3 = 0$ and recovers Step 1, so reciprocity is not an extra fact but the identity element of the composition law. *Verified in verify_aq-sr-inverse-lorentz-transformation-reciprocity.py: $\Lambda(\beta)\Lambda(-\beta) - I$ simplifies to the zero matrix symbolically, in both orders, and $\Lambda(\beta)^{-1} = \Lambda(-\beta)$ ; $\det\Lambda(\beta) = 1$ identically; the composed matrix is shown symbolically to have the boost shape of Step 3, the prefactor claim $[\gamma_1\gamma_2(1+\beta_1\beta_2)]^2(1-\beta_3^2) = 1$ is a polynomial identity, and closure is then confirmed entrywise on a $13\times13$ grid of $(\beta_1,\beta_2)$ .*

Go deeper · Derivation C · 2 minThe relativity principle forces γ to be even — and what breaks if it is not

Derivation CThe relativity principle forces γ to be even — and what breaks if it is not

Here is the sharp result, and it is the one this article owns. Everyone knows that $\gamma = 1/\sqrt{1-\beta^2}$ depends only on $\beta^2$ and is therefore unchanged by $v \to -v$ . What is far less often noticed is that this evenness is not implied by the constancy of light. Allow the forward and inverse maps to carry different factors, $\gamma_+$ and $\gamma_-$ , and the light postulate still goes through — it merely constrains their product. A whole one-parameter family of non-reciprocal theories survives, every one of them consistent with $c$ being the same in both frames. What kills them is the principle of relativity.

Step 1. Postulate a linear map with an unknown factor: $x' = \gamma_+(x - vt)$ for $S \to S'$ , and allow the inverse to have the same form with $v \to -v$ but a possibly different factor: $x = \gamma_-(x' + v t')$ . So far this is weaker than reciprocity — the two frames are permitted to disagree about the size of the effect, though not its structure. Why: we are deliberately keeping the weakest assumption that still lets us write anything down, so as to find out exactly which postulate does the remaining work.

Step 2. Impose the light postulate. A pulse leaving the common origin satisfies $x = ct$ in $S$ and must satisfy $x' = ct'$ in $S'$ . Substituting $x = ct$ into the forward map gives $x' = \gamma_+ t\,(c - v)$ ; substituting $x' = ct'$ into the inverse map gives $ct = \gamma_-\,t'(c + v)$ , so $t' = ct/[\gamma_-(c+v)]$ . Why: the light postulate is being used as a constraint on the pair $(\gamma_+, \gamma_-)$ , not as a definition of either one.

Step 3. Demand consistency: $x' = ct'$ requires $\gamma_+ t\,(c-v) = c \cdot \dfrac{ct}{\gamma_-(c+v)}$ , which rearranges to

\gamma_+\gamma_-\,(c^2 - v^2) = c^2, \qquad\text{that is}\qquad \gamma_+\gamma_-\left(1 - \beta^2\right) = 1.

Why: and there is the trouble. This is one equation for two unknowns. The constancy of the speed of light does not determine $\gamma_+$ and $\gamma_-$ separately — it only pins their product to $\gamma^2 = 1/(1-\beta^2)$ . Every point on that hyperbola is a viable, light-postulate-respecting theory.

Step 4. Look at a point that is not the symmetric one. Take $\beta = 0.6$ , so $\gamma_+\gamma_- = 1/0.64 = 1.5625$ . Choose $\gamma_+ = 1.5$ ; then $\gamma_- = 1.5625/1.5 = 1.0417$ . Light still travels at $c$ for both observers. But now $S$ finds $S'$ 's clocks slowed by a factor $1.5$ , while $S'$ finds $S$ 's clocks slowed by only $1.0417$ — and that is an experiment. The two crews compare their factors by radio, find them unequal, and conclude that the one with the larger factor is the one that is really moving. Why: a preferred frame has appeared out of nowhere. Not a preferred frame smuggled in as an assumption, but one manufactured by the asymmetry itself, and detectable by an experiment that both crews can perform.

Step 5. Invoke the relativity principle and close the case. It asserts that no experiment can distinguish inertial frames — so no such asymmetry may exist, and $\gamma_+ = \gamma_-$ . Combined with Step 3 this forces $\gamma_\pm^2\,(1-\beta^2) = 1$ , hence $\gamma = 1/\sqrt{1-\beta^2}$ : the familiar Lorentz factor, and an even function of $v$ . Why: so evenness of $\gamma$ is not a happy algebraic accident of the square root. It is the relativity principle, wearing an algebraic disguise — and it is the exact point at which the last trace of the ether is expelled from the theory. *Verified in verify_aq-sr-inverse-lorentz-transformation-reciprocity.py: the light-postulate constraint $\gamma_+\gamma_-(1-\beta^2) = 1$ is derived symbolically; setting $\gamma_+ = \gamma_-$ recovers $\gamma = 1/\sqrt{1-\beta^2}$ ; and the asymmetric point $(\gamma_+, \gamma_-) = (1.5,\ 1.0417)$ at $\beta = 0.6$ is checked to satisfy the constraint exactly while breaking the symmetry.*

Figure

The punchline. At β = 0.6 the light postulate does NOT fix γ — it fixes only the product γ₊γ₋ = 1/(1−β²) = 1.5625. Every point on that hyperbola is a theory in which light travels at c for both observers. Only ONE of them, the point where the curve meets the diagonal, gives the two observers the same factor. The relativity principle throws away all the rest

Limiting cases

Case	Result	Meaning
$\beta = 0$	$\Lambda = I$ , $\gamma = 1$	the identity element of the group; forward and inverse coincide
Galilean limit, $c \to \infty$	$x' = x - vt$ , $t' = t$	inverse is $x = x' + vt'$ , $t = t'$ — reciprocity survives, and is where the intuition comes from
$\beta_2 = -\beta_1$	$\Lambda(\beta_1)\Lambda(-\beta_1) = I$	reciprocity is closure evaluated at the inverse element
$\beta_1 = \beta_2 = \beta$	$\beta \oplus \beta = 2\beta/(1+\beta^2) < 1$	doubling a boost never reaches $c$ ; at $\beta = 0.6$ it gives $0.882$
Either $\beta_i = 1$	$\beta_1 \oplus 1 = 1$	light composes with anything to give light — $c$ is the fixed point of $\oplus$
$\beta \to 1$	$\gamma \to \infty$ , $\det\Lambda = 1$ still	the group structure never breaks, however violent the boost
$\gamma_+ \ne \gamma_-$	$\gamma_+\gamma_-(1-\beta^2) = 1$ still holds	light-postulate-compatible, relativity-principle-violating: a preferred frame

Common trapv → −v is NOT the same instruction as "swap the primes" — you must do both

The inverse transformation requires two operations, applied together: exchange primed and unprimed symbols, and flip the sign of $\beta$ . Students routinely do one and not the other, and both failure modes are fatal. Do only the sign flip, and you get $ct' = \gamma(ct + \beta x)$ , which is not the inverse of anything — it is a boost in the opposite direction applied to the same frame. Do only the prime swap, and you get $ct = \gamma(ct' - \beta x')$ , which composes with the forward boost to give $\Lambda(\beta)^2 = \Lambda(2\beta/(1+\beta^2)) \ne I$ — at $\beta = 0.6$ you land at $\beta = 0.882$ , $\gamma = 2.125$ , instead of coming home. The reason both operations are needed, and the reason doing both works, is the theorem of this article: reciprocity says that a change of sign and a change of viewpoint, made together, are the whole of the difference between the two frames.

Common trapΔt = γΔt′ and Δt′ = γΔt are BOTH true — and they are not about the same events

Written side by side, these two equations imply $\gamma^2 = 1$ and hence $v = 0$ , which is absurd. Books nonetheless print both, and both are correct, because they refer to different pairs of events — an ellipsis that has confused generations. The first, $\Delta t = \gamma\,\Delta t'$ , concerns two ticks of a clock at rest in $S'$ : those events have $\Delta x' = 0$ , and $S$ measures them with two spatially separated clocks of its own. The second, $\Delta t' = \gamma\,\Delta t$ , concerns two ticks of a different clock, one at rest in $S$ , with $\Delta x = 0$ , measured by two separated $S'$ -clocks. No single pair of events satisfies both. Each frame uses one moving clock and two stationary ones, and the two frames disagree about whether the other's stationary pair is synchronized at all. So reciprocity does not say "each clock is slower than the other", which would be nonsense. It says the rule each observer applies to the other is the same rule, with the same $\gamma$ — a symmetry, made consistent by the relativity of simultaneity. The bookkeeping is done in Consistency of Time Dilation and Length Contraction.

Worked examples

1 · Send a muon into its own rest frame, and bring it back

In the Frisch–Smith experiment a muon is created at the top of the flight path and detected $1907\ \mathrm{m}$ lower down, travelling at $\beta = 0.995$ . Take creation as the origin. Transform the detection event into the muon's rest frame, and then transform it back.

Step 1 — the event in the mountain frame. The flight path is $\Delta x = 1907\ \mathrm{m}$ (the detector-to-detector drop; the summit's $1917\ \mathrm{m}$ elevation is a different quantity). The flight time follows from the speed: $c\Delta t = 1907/0.995 = 1916.58\ \mathrm{m}$ , so the detection event is $(c\Delta t,\ \Delta x) = (1916.58,\ 1907)\ \mathrm{m}$ .

Step 2 — get $\gamma$ . $\gamma = 1/\sqrt{1 - 0.995^2} = 1/\sqrt{0.009975} = 10.013$ .

Step 3 — boost into the muon's frame. Using equation (1) with $\beta = 0.995$ :

c\Delta t' = 10.013\,(1916.58 - 0.995 \times 1907) = 10.013 \times 19.118 = \mathbf{191.42\ m},

\Delta x' = 10.013\,(1907 - 0.995 \times 1916.58) = 10.013 \times 0 = \mathbf{0}.

The muon does not move in its own frame — as it had better not. Its proper time is $191.42/c = 0.6385\ \mathrm{\mu s}$ .

Step 4 — bring it back with $v \to -v$ . Using equation (2):

c\Delta t = 10.013\,(191.42 + 0.995 \times 0) = 1916.58\ \mathrm{m} \;\checkmark, \qquad \Delta x = 10.013\,(0 + 0.995 \times 191.42) = 1907.0\ \mathrm{m} \;\checkmark.

Exactly back where we started, to every digit carried. That is $\Lambda(-\beta)\Lambda(\beta) = I$ , demonstrated on a real event from a real experiment.

Step 5 — read the reciprocity. In the mountain frame the muon moves at $+0.995c$ . In the muon's frame the mountain moves at $-0.995c$ : same speed, opposite sign, no $\gamma$ anywhere. The muon is entitled to regard itself as at rest with a $1907\ \mathrm{m}$ -thick atmosphere rushing up at it — contracted, in its frame, to $1907/10.013 = 190.5\ \mathrm{m}$ , which at $0.995c$ takes exactly $0.6385\ \mathrm{\mu s}$ to pass ✓, matching Step 3. Two frames, two stories, one muon.

2 · The group law, and the theory that light permits but relativity forbids

(a) Verify $\Lambda(0.6)\Lambda(-0.6) = I$ and $\det\Lambda(0.6) = 1$ explicitly. (b) At $\beta = 0.6$ , construct a non-reciprocal pair $(\gamma_+, \gamma_-)$ consistent with the constancy of light, and say what it predicts that the real world does not.

Step 1 — the matrices. $\gamma = 1/\sqrt{1-0.36} = 1.25$ and $\gamma\beta = 0.75$ , so

\Lambda(0.6) = \begin{pmatrix} 1.25 & -0.75 \\ -0.75 & 1.25\end{pmatrix}, \qquad \Lambda(-0.6) = \begin{pmatrix} 1.25 & 0.75 \\ 0.75 & 1.25\end{pmatrix}.

Step 2 — the product. Diagonal: $1.25^2 - 0.75^2 = 1.5625 - 0.5625 = 1$ . Off-diagonal: $(1.25)(0.75) - (0.75)(1.25) = 0$ . The product is $I$ exactly, and the determinant is the same arithmetic: $\det\Lambda(0.6) = 1.5625 - 0.5625 = 1$ ✓.

Step 3 — the constraint that light imposes. From the derivation above, the light postulate alone gives only $\gamma_+\gamma_- = 1/0.64 = 1.5625$ . The true theory sits at the symmetric point $\gamma_+ = \gamma_- = 1.25$ , and indeed $1.25 \times 1.25 = 1.5625$ ✓.

Step 4 — build the rogue theory. Pick $\gamma_+ = 1.5$ . Then $\gamma_- = 1.5625/1.5 = \mathbf{1.0417}$ , and $1.5 \times 1.0417 \times 0.64 = 1.000$ ✓. This theory is fully consistent with the speed of light being $c$ in both frames. Nothing in Michelson–Morley touches it.

Step 5 — kill it. It predicts that $S$ sees $S'$ 's clocks running slow by $1.5$ , while $S'$ sees $S$ 's clocks running slow by only $1.0417$ . The two crews radio their measured factors to each other, find $1.5 \ne 1.0417$ , and immediately know which of them is "really" moving: the one whose partner reports the larger dilation. That is a preferred frame, detected from inside two inertial laboratories — precisely what the relativity principle forbids. The only survivor on the hyperbola $\gamma_+\gamma_- = 1.5625$ is the symmetric point, and reciprocity is what selects it.

Physics of the portrait

The $v \to -v$ rule appears in Einstein's 1905 paper as a step in the algebra, and in Poincaré's of the same year as something deeper. But the sharpest statement belongs to a paper almost nobody reads.

Origins · 1 minReciprocity and how this was discovered

Reciprocity

1905–1910 · Einstein's inverse · Poincaré names the group · Ignatowski shows reciprocity almost suffices

That the inverse boost is the forward boost with $v \to -v$ is stated in §3 of Einstein's "Zur Elektrodynamik bewegter Körper" (1905), where it is used to check that the transformation is self-consistent. In the same year Henri Poincaré, in "Sur la dynamique de l'électron", made the structural observation that turns the fact into a principle: the Lorentz transformations form a group — he coined that name for them — with $\Lambda(v)^{-1} = \Lambda(-v)$ as the inverse-element property and closure under composition as the group law, from which he deduced constraints on relativistic dynamics that Einstein reached by other means. The most remarkable development came in 1910, when Vladimir Ignatowski (Physikalische Zeitschrift 11, 972) asked how much of relativity survives if you delete the light postulate altogether and keep only the principle of relativity, the homogeneity of space and time, isotropy, and reciprocity. His answer astonished his contemporaries and still surprises students: almost all of it. Those assumptions alone force the transformation into the Lorentz form, with a single undetermined invariant speed — which experiment then identifies with $c$ . Relativity, it turns out, is not fundamentally a theory about light. It is a theory about symmetry, and reciprocity is one of the loadbearing beams.

History

1905

— Einstein, in §3 of "Zur Elektrodynamik bewegter Körper", obtains the inverse transformation by the substitution

v \to -v

and uses it as a consistency check on the forward one.

1905

— Poincaré, in "Sur la dynamique de l'électron", shows that the Lorentz transformations form a group, names them, and exploits closure and the inverse property to constrain relativistic dynamics.

1908

— Minkowski recasts boosts as hyperbolic rotations, making

\Lambda(-v) = \Lambda(v)^{-1}

geometrically obvious: it is a rotation through the negative angle.

1910

— Ignatowski shows the relativity principle, homogeneity, isotropy and reciprocity alone — with no light postulate — force the Lorentz form, leaving only an undetermined invariant speed for experiment to fix.

1911

— Frank and Rothe give a rigorous version of the Ignatowski argument, tightening the assumptions under which the group law is forced.

Going further

Texts. Taylor & Wheeler, Spacetime Physics, 2nd ed., Ch. 3 · Rindler, Introduction to Special Relativity, 2nd ed., §§2.6–2.8 · Griffiths, Introduction to Electrodynamics, 4th ed., §12.1 (eqs. 12.18–12.22). On the postulate-light route: W. v. Ignatowsky, Phys. Z. 11, 972 (1910); and for a modern treatment, J.-M. Lévy-Leblond, "One more derivation of the Lorentz transformation", Am. J. Phys. 44, 271 (1976).

Provenance. The reciprocity derivation follows Rindler §2.7 and Griffiths eqs. 12.19–12.22. The even- $\gamma$ theorem — that light-speed constancy gives only $\gamma_+\gamma_-(1-\beta^2) = 1$ , and that the relativity principle is what collapses that hyperbola to its symmetric point — is derived and machine-checked above; it is the article's owned result. Every number is recomputed in verify_aq-sr-inverse-lorentz-transformation-reciprocity.py: the $\beta = 0.6$ anchors, the round trip $(10, 4) \to (9.5, -2.5) \to (10, 4)\ \mathrm{m}$ , $\Lambda(0.6)\Lambda(-0.6) = I$ , $0.6 \oplus 0.6 = 0.882$ with $\gamma = 2.125$ , the rogue pair $(1.5,\ 1.0417)$ , and the Frisch–Smith muon round trip ( $(1916.58,\ 1907) \to (191.42,\ 0) \to (1916.58,\ 1907)\ \mathrm{m}$ , contracted path $190.5\ \mathrm{m}$ , proper time $0.6385\ \mathrm{\mu s}$ ).

Builds on. The Lorentz Transformation Equations · The Lorentz Factor Gamma · The Two Postulates of Special Relativity · Relativity of Simultaneity. Leads to. Time Dilation · Length Contraction · Rapidity and the Hyperbolic Form of Boosts · Consistency of Time Dilation and Length Contraction.

Why isn't reciprocity just obvious? Of course we recede from each other at the same speed.

Because "of course" is doing all the work, and in a theory with a preferred frame it would be false. Imagine sound in still air: if you and I move through the air at different speeds, the relation between our descriptions is not symmetric — the air breaks the tie, and there is a fact of the matter about who is moving. Reciprocity is a property the Lorentz transformation has, not one every transformation must have. So take the forward boost, put $x = 0$ into it, and look. The $\gamma$ 's cancel and you get exactly $-v$ . Had they not cancelled — had the answer been $-\gamma v$ , say — we would have known which of us was really moving.

If each of us sees the other's clock run slow, who is actually younger when we meet?

The question is malformed as posed. Two inertial observers in relative motion meet at most once, and a single meeting gives no comparison of elapsed times — you must be co-located twice to compare wristwatches unambiguously, and two straight worldlines that cross once never cross again. To get a second meeting one of you must turn around, which means accelerating, which breaks the symmetry — and it is that broken symmetry, not any failure of reciprocity, that produces a definite answer. That is the twin paradox: a story about acceleration. While you are both inertial, reciprocity is exact, and the "who is younger" question has no content.

Can I get the inverse just by solving the two forward equations simultaneously?

Yes, and you should do it once to see that it works. Treat $ct' = \gamma(ct - \beta x)$ and $x' = \gamma(x - \beta\,ct)$ as two linear equations in the unknowns $ct$ and $x$ , and grind: the combination $\gamma(ct' + \beta x')$ expands to $\gamma^2(ct - \beta x) + \gamma^2\beta(x - \beta ct) = \gamma^2(1-\beta^2)\,ct$ , the $x$ terms cancelling identically, and $\gamma^2(1-\beta^2) = 1$ finishes it — you land on $ct$ . The same manipulation on $\gamma(x' + \beta\,ct')$ returns $x$ . So brute force recovers exactly the $v \to -v$ form. What it does not tell you is why the answer had to be a boost rather than some other map. Algebra confirms; structure explains.

What actually goes wrong if γ is not an even function of v?

A preferred frame appears, and it is detectable. The light postulate constrains only the product $\gamma_+\gamma_- = 1/(1-\beta^2)$ , so an asymmetric pair like $(1.5,\ 1.0417)$ at $\beta = 0.6$ is perfectly compatible with light travelling at $c$ for both observers. But the two crews can compare notes: I report your clocks slowed by $1.5$ , you report mine slowed by $1.0417$ , and the disagreement is itself a measurement. Whoever's partner reports the larger factor is the one that is "really" moving, and we have just built an ether-detector out of two clocks and a radio. The principle of relativity forbids exactly this, and it does so by forcing $\gamma_+ = \gamma_-$ — which is to say, by making $\gamma$ even.

Do boosts in different directions also compose into a single boost?

No, and this is the one place the tidy group picture of this article does not extend. Two boosts along the same axis compose into a single boost at $\beta_1 \oplus \beta_2$ — equation (4) — which is what makes the one-axis boosts a one-parameter group. Two boosts along different axes compose into a boost plus a rotation, the Thomas–Wigner rotation, so the boosts alone do not form a subgroup of the full Lorentz group. That non-commutativity has real consequences: the Thomas precession of an electron's spin in an atom is one of them. Everything in this article is restricted to a single common axis, where the composition closes cleanly.

Is the inverse transformation a different physical operation from the forward one?

No — and this is the deepest content of the group law. $\Lambda(-v)$ is a boost, of exactly the same kind as $\Lambda(v)$ , just in the other direction. The two frames are not related as "original" and "derived"; they are two members of a family, and each may regard itself as the original. That is what closure buys you: no operation is available to one observer and unavailable to the other. If the inverse of a boost were not a boost, one frame would possess a transformation the other could not perform, and the principle of relativity would be dead in the algebra before it ever reached the physics.