Asymmetry and Acceleration in the Twin Paradox

Two clocks leave the same event and meet again at another. One coasted; the other went out four light-years, turned, and came home. At the reunion the coasting clock reads $10$ years and the travelling clock reads $6$ . Nobody disputes the numbers. What almost everybody disputes is why. The story you were told is that the traveller accelerated, and acceleration is absolute, and that is what broke the symmetry. Paul Langevin said exactly this in 1911, and it has been repeated ever since. It is wrong — not in its conclusion, but in its causal claim. Take the acceleration away entirely. Use three clocks that never change speed: one at home, one already coasting outward, one already coasting inward. Hand a reading across at the far point. The relayed total is still $6$ years. No engine fires; nobody feels a thing; the age gap is exactly what it was. So acceleration is not the cause. It is a symptom — the visible mark of a worldline that had to be assembled out of two inertial frames instead of one. The cause is geometry: proper time is a path length, and the traveller's path is shorter. And when the traveller tries to audit Earth's clock she finds $3.6$ years of Earth time on her two coasting legs and a $6.4$ -year hole at the turnaround. That hole is $2Lv/c^2$ — and it closes her books to the year.

Key takeaways

Signature $(+,-,-,-)$ : $\mathrm{d}s^2 = c^2\mathrm{d}t^2 - \mathrm{d}x^2$ , timelike $>0$ , and $c\,\mathrm{d}\tau = \mathrm{d}s$ .
The three-clock relay kills the acceleration story. Three inertial clocks, zero acceleration anywhere, and the relayed total is still $T/\gamma$ . What actually breaks the symmetry is that the traveller's worldline needs two inertial frames and the stayer's needs one.
The traveller's audit of Earth: $2L/(v\gamma^2)$ of Earth time on the legs, plus a turnaround jump of $2Lv/c^2$ — with no $\gamma$ . Books close exactly: $2L/(v\gamma^2) + 2Lv/c^2 = 2L/v$ .
The clock hypothesis — $\mathrm{d}\tau$ depends on instantaneous speed, not on acceleration — is why a hard turnaround and a soft one agree in the limit. CERN's storage-ring muons confirm it at $\sim10^{18}\,g$ .
Anchor: $\beta = 0.8$ , $L = 4$ ly $\Rightarrow$ Earth $10$ yr, traveller $6$ yr, jump $6.4$ yr, legs $1.8 + 1.8 = 3.6$ yr.
Key idea: acceleration is not the cause of the age gap, it is the receipt for it. Bending a worldline costs proper time; the bend is what an accelerometer feels, but the cost is set by the path, not by the sharpness of the corner.

The core insight

Elapsed proper time is the Minkowski arc length of a worldline, and in Minkowski geometry the straight worldline between two timelike-separated events is the longest one. Everything else is commentary. The traveller's path is bent, therefore shorter, therefore she is younger — and the bend can be replaced by a handover between two clocks that were never bent at all. Acceleration is how you notice you are on the short path. It is not the reason the path is short.

Where this lives today

The clock hypothesis is the load-bearing assumption of the whole resolution, and it is the one thing here that could conceivably have been false. It is not a consequence of the relativity postulates: nothing in them forbids a clock whose rate depends on $\mathrm{d}v/\mathrm{d}t$ . It has to be measured. The measurement exists, and it is spectacular. In the CERN muon storage ring, muons at $\gamma = 29.33$ ( $\beta = 0.99942$ ) circulate on a $7$ -metre-radius orbit. Circular motion at that $\gamma$ means a proper acceleration of $a = \gamma^2 v^2/R \approx 1.10\times10^{19}\ \mathrm{m/s^2}$ — about $1.1\times10^{18}\,g$ , an acceleration no macroscopic object will ever experience. If acceleration had any direct effect on clock rate, a muon is where it would show. Bailey and collaborators (1977) measured the circulating lifetime as $\tau^- = 64.368(29)\ \mu\text{s}$ against a proper lifetime of $\tau_0 = 2.1948(10)\ \mu\text{s}$ . The prediction is $\gamma\tau_0 = 29.33 \times 2.1948 = 64.37\ \mu\text{s}$ . The time-dilation factor is confirmed to a fractional error of $2\times10^{-3}$ , with no acceleration-dependent term at $10^{18}\,g$ . The muon is a real twin: it goes out, comes back, and is younger, and the storage ring is the turnaround. The everyday version is quieter but ubiquitous. GPS satellites are travelling twins in a weak gravitational field; their velocity contributes a special-relativistic slowing of $-7.2\ \mu\text{s/day}$ , which is over-compensated by a gravitational speeding of $+45.8\ \mu\text{s/day}$ for a net $+38.6\ \mu\text{s/day}$ that the system corrects for by design. Hafele and Keating (1971) flew caesium clocks east and west around the world and found the two directions disagreed — the asymmetry between them being the purest laboratory echo of the twin problem, because the eastbound and westbound aircraft have different speeds relative to the non-rotating frame while sharing a gravitational potential.

Origins · 2 minPaul Langevin and how this was discovered

Paul Langevin

1872–1946 · Paris

Paul Langevin (1872–1946), professor at the Collège de France, gave the travelling twin its first vivid statement. In a lecture at the Bologna philosophy congress on 10 April 1911, published that July as "L'évolution de l'espace et du temps" in Scientia X, 31–54, he described a traveller who spends one year of her own time flying out at $\gamma = 100$ ( $\beta = 0.99995$ ), turns, and comes back. She has aged two years. Earth has aged two hundred. It is still, after a century, the best advertisement the subject has. Langevin then explained the asymmetry — and got the explanation wrong, in a way that has outlived him. He called it "another example of the absolute character of acceleration": only the traveller, he wrote, undergoes an acceleration that changes the direction of her velocity, and that is the cause. The conclusion was right, the diagnosis was not. Max von Laue, in December 1911 and again in 1912–13, gave the correct account: the traveller's worldline rests in at least two inertial frames while the stayer's rests in one, and among all timelike paths between two events the straight one carries maximal proper time. Von Laue's is the argument this article proves; Langevin's is the one people still repeat. The portrait belongs to Langevin because he asked the question that mattered — and because the most instructive thing about the twin paradox is that the man who invented it got its mechanism wrong.

History

1905

— Einstein derives the moving-clock result in "Zur Elektrodynamik bewegter Körper": a clock carried around a closed curve lags a stay-at-home clock by ½tv²/c².

1911

— Paul Langevin, in Scientia X, 31–54, poses the travelling twin (γ = 100: two years aboard, two hundred on Earth) and attributes the asymmetry to the absolute character of acceleration.

1911

— Max von Laue, in December, gives the geometric resolution: the traveller rests in two inertial frames, the stayer in one, and the straight worldline maximises proper time.

1918

— Einstein, in the Dialog über Einwände gegen die Relativitätstheorie, Die Naturwissenschaften 6, 697, re-derives the result in the traveller's own frame using the equivalence principle.

1956

— the Dingle controversy: Herbert Dingle argues for years, in print, that the situation is symmetric and relativity is inconsistent. It is not, and it is not.

1971

— Hafele and Keating fly caesium clocks east and west around the world; the two directions disagree, as the theory requires.

1977

— Bailey and collaborators measure muon lifetimes at γ = 29.33 in the CERN storage ring under a proper acceleration of about 10¹⁸ g, confirming time dilation to 2 parts in 1000 with no acceleration-dependent term.

The physical picture

Start from the setup owned by `aq-sr-twin-paradox`, and take its numbers as given: coasting speed $\beta = v/c = 0.8$ , so $\gamma = 5/3$ ; turnaround point $L = 4$ light-years from Earth. Earth's clock runs $T = 2L/v = 10$ years. The traveller's runs $T/\gamma = 6$ years. That article proves it geometrically; this one asks what broke. The seductive wrong answer runs: "relativity says all motion is relative, so each twin should see the other as younger, so the situation is symmetric, so we need something to break the tie — and the tie-breaker is that only one twin felt a shove." Two of those clauses are false. Relativity does not say all motion is relative; it says the laws are the same in all inertial frames, and only one twin is in one. And the tie-breaker is not the shove. Here is the cleanest way to see that. Delete the traveller. Replace her with a relay.

Clock O sits on Earth, inertial, reading Earth time. Clock A is already coasting outward at $0.8c$ and happens to pass Earth at $t = 0$ ; it is zeroed at that moment. Clock B is already coasting inward at $0.8c$ , on a path that will bring it past the turnaround point at exactly the right moment; when it passes Clock A there, it is set to whatever A reads. B carries that reading home and hands it to O. No clock in this experiment ever changes speed. No accelerometer anywhere reads anything but zero. And the number B delivers to Earth is $6$ years, against Earth's $10$ . The age gap survives the complete removal of acceleration. What did not survive is the single frame. Clock A's frame and Clock B's frame are different inertial frames, and the relay is the moment we switch between them. That switch — not any force — is what the traveller's engine is for. The engine exists to move a physical body from A's frame into B's frame. The proper-time bookkeeping does not care how you get there.

Figure

PUNCHLINE — the traveller's audit of Earth's clock, in years, for β = 0.8 and L = 4 ly. Her two coasting legs credit Earth with 1.8 + 1.8 = 3.6 years. Her turnaround swings her line of simultaneity from the shallow outbound slice to the steep inbound one, and Earth's clock leaps from 1.8 yr to 8.2 yr — a jump of 6.4 = 2Lv/c², with NO γ. Books: 3.6 + 6.4 = 10.0, exactly Earth's elapsed time. Insert a γ and the jump becomes 10.67, the books total 14.27, and 4.27 years of Earth history appear from nowhere

The core equations

Proper time is an integral along a path, and that single fact carries the whole subject.

\tau[\mathcal{C}] \;=\; \int_{\mathcal{C}}\mathrm{d}\tau \;=\; \int_{t_A}^{t_B}\sqrt{1-\frac{v^2(t)}{c^2}}\;\mathrm{d}t \qquad (1)

where: $\mathcal{C}$ is a timelike worldline from event $A$ to event $B$ and $v(t)$ is its instantaneous coordinate speed. Physical reading: $\tau$ is a functional of the path, not a function of the endpoints. Two worldlines joining the same two events may return different numbers, and nothing is paradoxical about that — no more than two roads between two cities having different lengths. Note what does not appear in the integrand: $\mathrm{d}v/\mathrm{d}t$ . Acceleration is absent from the formula. Because the integrand never exceeds $1$ , and equals $1$ only where $v=0$ :

\tau_{\text{bent}} \;<\; \tau_{\text{straight}} \;=\; T, \qquad\text{with equality only for the inertial path} \qquad (2)

where: $T$ is the coordinate time between the two events in the frame where the straight worldline is at rest. Physical reading: the inertial worldline maximises proper time — the reverse of the Euclidean triangle inequality, and the reason a detour in spacetime costs you age instead of buying you distance. For the standard two-leg trip at constant coasting speed, (1) collapses:

\tau_{\text{trav}} = \frac{T}{\gamma} = \frac{2L}{v\gamma}, \qquad T = \frac{2L}{v}, \qquad \gamma = \frac{1}{\sqrt{1-\beta^2}} \qquad (3)

where: $L$ is the Earth-frame distance to the turnaround. Physical reading: at $\beta=0.8$ , $L=4$ ly this is Earth $10$ yr, traveller $6$ yr. Recalled from `aq-sr-twin-paradox`; this article does not re-derive it. And the equation this article owns — the traveller's audit of Earth, and the one the textbooks most often mistype:

\boxed{\;\underbrace{\frac{2L}{v\gamma^{2}}}_{\text{her two legs}} \;+\; \underbrace{\frac{2Lv}{c^{2}}}_{\text{turnaround jump}} \;=\; \underbrace{\frac{2L}{v}}_{\text{Earth's total}}\;} \qquad (4)

where: every symbol is measured in the Earth frame. Physical reading: the jump is $2Lv/c^2$ with no $\gamma$ factor, and (4) is an identity — $\frac{2L}{v}\!\left(\frac{1}{\gamma^2} + \beta^2\right) = \frac{2L}{v}$ , because $1/\gamma^2 = 1-\beta^2$ . It closes to the year. Multiply the jump by $\gamma$ , as several standard references do, and the identity fails: at $\beta=0.8,\,L=4$ ly you get $3.6 + 10.67 = 14.27$ years against Earth's actual $10$ .

Where the asymmetry actually lives

Derivation AThe three-clock relay: the age gap with zero acceleration

The claim to be established is the strongest available refutation of the acceleration story, and it is worth stating sharply before proving it: there exists an experiment containing no acceleration whatever, in which three permanently inertial clocks reproduce the twin-paradox age gap exactly. If that is true then acceleration cannot be the cause of the gap, because you can remove a cause and keep the effect only if it was never the cause. What follows constructs the experiment and computes its output.

Step 1. Lay down the three clocks. O rests on Earth at $x=0$ forever. A moves at $+v$ forever, and is zeroed as it passes Earth at $t=0$ . B moves at $-v$ forever, on a worldline arranged to cross A's at the far point $x = L$ . All three are freely falling; all three accelerometers read zero for all time, before, during and after. Why: the entire content of the argument is that we never once switch on an engine. Everything else is arithmetic.

Step 2. Find the crossing event. A's worldline is $x = vt$ , so it is at $x=L$ at Earth time $t_1 = L/v$ . B must be there too, so B's worldline is $x = L - v(t - L/v) = 2L - vt$ , which reaches Earth at $t = 2L/v = T$ . Why: B is not "sent" anywhere — it was always coasting inward, and we simply chose the one whose path happens to meet A's at the right event. No agency, no force.

Step 3. Read A at the crossing. A is inertial with speed $v$ , so between Earth-passage and the crossing it accumulates $\tau_A = t_1/\gamma = L/(v\gamma)$ . With $\beta=0.8$ , $L=4$ ly: $t_1 = 5$ yr and $\tau_A = 5/(5/3) = 3$ yr. Why: this is ordinary time dilation for a single inertial clock, applied once, in one frame. Nothing controversial has happened yet.

Step 4. Hand the number across. At the crossing event, B is set to read whatever A reads: $3$ yr. This is a local comparison of two clocks at the same place — the one kind of clock comparison that every frame agrees on. Why: no simultaneity convention is used, no signal is sent over a distance, and no clock is disturbed. B's hands are set, and B's motion is untouched.

Step 5. Read B at Earth. B is inertial with speed $v$ , and takes Earth time $t_1 = L/v = 5$ yr to get home, accumulating another $3$ yr. So B arrives reading $3 + 3 = \boxed{6\ \text{yr}}$ , while O reads $T = 10$ yr. Why: the relayed total is exactly $T/\gamma$ , the same $6$ years the accelerating traveller would have logged. The gap is intact and the acceleration is gone.

Step 6. Name what was actually removed and what was not. Nothing accelerated — but the record carried home was assembled from two different inertial frames, A's and B's, spliced at the crossing. Earth's record was assembled from one. That is the asymmetry, and it is the only one present. Why: this is the precise sense in which the situation is not symmetric. It is a statement about how many inertial frames a worldline needs, and it survives the deletion of every force in the problem. *Verified numerically and symbolically in verify_aq-sr-twin-paradox-asymmetry.py: the crossing event $(t_1,L) = (5\ \text{yr}, 4\ \text{ly})$ ; $\tau_A = \tau_B = L/(v\gamma) = 3$ yr; the relayed total $2L/(v\gamma) = 6$ yr $= T/\gamma$ identically in $\beta$ and $L$ ; and every accelerometer reading zero throughout.*

Go deeper · Derivation B · 2 minThe turnaround jump is 2Lv/c-squared, and it carries no gamma

Derivation BThe turnaround jump is 2Lv/c-squared, and it carries no gamma

Now put the traveller back and let her audit Earth's clock from her own point of view. She will find that her two coasting legs account for only part of Earth's elapsed time, and that the missing part appears in a single lurch at the turnaround. The size of that lurch is the most frequently mistyped number in this entire topic — two of the three standard textbook treatments the base corpus draws on print it with a spurious factor of $\gamma$ — so we will derive it, check it against an identity that cannot be fudged, and then show precisely what the $\gamma$ error costs.

Step 1. Fix the turnaround event. In Earth coordinates it is $E = (t, x) = (L/v,\; L) = (5\ \text{yr},\ 4\ \text{ly})$ . Why: everything below is a question about which Earth event the traveller calls "now" as she passes through $E$ — a question whose answer depends on her velocity, and which therefore has two answers.

Step 2. Ask the outbound frame. A frame moving at $+v$ slices spacetime along $t' = \text{const}$ , i.e. along $t - vx/c^2 = \text{const}$ (recalled from `aq-sr-relativity-of-simultaneity-s1`). Through $E$ that slice is $t - \beta x/c \cdot c = 5 - 0.8(4) = 1.8$ , so at $x=0$ it hits Earth at $t = L/v - vL/c^2 = 5 - 3.2 = 1.8$ yr. Why: just before she turns, the traveller's "now on Earth" is Earth-time $1.8$ yr — which is exactly $\tau_{\text{out}}/\gamma = 3/(5/3) = 1.8$ , the ordinary time-dilated reckoning. The two routes agree, as they must.

Step 3. Ask the inbound frame. A frame moving at $-v$ slices along $t + vx/c^2 = \text{const}$ . Through $E$ : at $x=0$ it hits Earth at $t = L/v + vL/c^2 = 5 + 3.2 = 8.2$ yr. Why: an instant later, by her own clock, the same traveller's "now on Earth" is $8.2$ yr. She has not moved; she has re-aimed her definition of simultaneity.

Step 4. Subtract. The Earth clock's reading leaps by

\Delta t_{\text{jump}} \;=\; \left(\frac{L}{v}+\frac{vL}{c^2}\right)-\left(\frac{L}{v}-\frac{vL}{c^2}\right) \;=\; \boxed{\frac{2Lv}{c^{2}}} \;=\; 2(0.8)(4) = 6.4\ \text{yr}.

Why: no $\gamma$ appears, and none can, because the $L/v$ terms cancel and what is left is twice a single $vL/c^2$ . The quantity is a <strong>reading</strong> of a clock at rest in the Earth frame, and Earth-clock readings are Earth coordinate time.

Step 5. Close the books, which is the check that cannot be fudged. Her two legs credit Earth with $\tau_{\text{trav}}/\gamma = (2L/(v\gamma))/\gamma = 2L/(v\gamma^2) = 10/(25/9) = 3.6$ yr. Add the jump:

\frac{2L}{v\gamma^{2}} + \frac{2Lv}{c^{2}} = \frac{2L}{v}\left(\frac{1}{\gamma^{2}}+\beta^{2}\right) = \frac{2L}{v}\left(1-\beta^{2}+\beta^{2}\right) = \frac{2L}{v} = T.

Numerically $3.6 + 6.4 = 10.0$ yr, on the nose. Why: this is an identity in $\beta$ and $L$ , not a coincidence at one speed. It is the audit balancing, and it is what a correct jump formula must do.

Step 6. Watch the $\gamma$ error break it. Substitute the commonly printed $2Lv\gamma/c^2 = 6.4 \times 5/3 = 10.67$ yr and the audit reads $3.6 + 10.67 = 14.27$ yr against Earth's actual $10$ — an invented $4.27$ years, a $43\%$ overshoot. The general failure is $\frac{2L}{v}(1-\beta^2+\gamma\beta^2) \ne \frac{2L}{v}$ for any $\beta \ne 0$ . Why: a formula that violates a conservation identity is not "approximately right"; it is wrong, and this is how you catch it without consulting an authority. *Verified symbolically in verify_aq-sr-twin-paradox-asymmetry.py: the two simultaneity slices through $E$ hitting Earth at $L/v \mp vL/c^2$ ; $\Delta t_{\text{jump}} = 2Lv/c^2$ with sympy simplifying the $\gamma$ -free form; the identity $2L/(v\gamma^2) + 2Lv/c^2 - 2L/v \equiv 0$ ; and the $\gamma$ -contaminated version failing by $2Lv(\gamma-1)/c^2$ , which is $4.27$ yr at the anchor.*

Go deeper · Derivation C · 2 minThe clock hypothesis, and what a real turnaround costs

Derivation CThe clock hypothesis, and what a real turnaround costs

An instantaneous turnaround is a fiction: it demands infinite proper acceleration. So one must ask whether the idealised answer survives a real burn, and whether the acceleration itself adds some further, unaccounted slowing on top of what the speed profile already implies. The assumption that it does not is called the clock hypothesis, and it is not a theorem of special relativity — it is a physical assumption about clocks, which is why it must be tested. Here we make it precise, compute an exactly solvable finite-acceleration trip, and watch it converge.

Step 1. State the hypothesis. An ideal clock's rate depends only on its instantaneous speed: $\mathrm{d}\tau = \sqrt{1-v^2(t)/c^2}\,\mathrm{d}t$ , with no term in $\mathrm{d}v/\mathrm{d}t$ or any higher derivative. Why: this is exactly the statement that equation (1) is the whole story. If a clock had an acceleration-dependent rate, (1) would need a correction term and the twin answer would depend on how hard she turned.

Step 2. Build a real turnaround. Replace the corner with a burn of constant proper acceleration $a$ that reverses the rapidity from $+\varphi$ to $-\varphi$ , where $\varphi = \operatorname{artanh}\beta = \operatorname{artanh}(0.8) = \tfrac12\ln 9 = 1.098612$ . Rapidity is linear in proper time under constant proper acceleration (recalled from `aq-sr-rapidity-additive-velocity`), so the burn lasts $\Delta\tau_b = 2c\varphi/a$ and, since $\mathrm{d}t = \cosh\varphi\,\mathrm{d}\tau$ , occupies Earth time $\Delta t_b = (2c/a)\sinh\varphi = (2c/a)\gamma\beta$ . Why: everything about the burn is fixed by $\varphi$ and $a$ alone — a genuinely exactly-solvable turnaround, not a numerical approximation.

Step 3. Note the burn's geometry. Since $\mathrm{d}x/\mathrm{d}\tau = c\sinh\varphi(\tau)$ and $\varphi$ runs symmetrically from $+\varphi$ through $0$ to $-\varphi$ , the ship's net displacement over the burn is zero: it coasts out, arcs over, and comes back to the same $x$ it began the burn at. Its farthest excursion is $x_{\max} = (c^2/a)(\cosh\varphi - 1)$ beyond that point. Why: this lets us hold the trip's far point fixed at $L = 4$ ly and simply shorten the coasting legs by $x_{\max}$ , so we are comparing like with like.

Step 4. Assemble the totals. With coasting distance $L - x_{\max}$ on each side,

T = \frac{2(L-x_{\max})}{v} + \Delta t_b, \qquad \tau_{\text{trav}} = \frac{2(L-x_{\max})}{v\gamma} + \Delta\tau_b .

Why: the coasting legs are pure inertial motion and the burn is exactly integrated. There is no approximation anywhere in the expression.

Step 5. Put numbers in. One $g$ corresponds to $a/c = 1.032295\ \text{yr}^{-1}$ . Then $\Delta\tau_b = 2.128\ \text{yr}$ , $\Delta t_b = 2.583\ \text{yr}$ , $x_{\max} = 0.6458$ ly, and the trip gives Earth $10.969$ yr, traveller $7.160$ yr — still a $3.81$ -year gap. At $100g$ : $\Delta\tau_b = 0.02128$ yr ( $7.8$ days), and Earth $10.010$ yr, traveller $6.012$ yr, against the idealised $10$ and $6$ . Why: a $1g$ turnaround is not a perturbation — it eats $2.1$ years of her $7.2$ — but the traveller is still younger, and as $a$ grows the answer slides smoothly onto the sharp-corner result. There is no discontinuity and no hidden acceleration term.

Step 6. Read the moral. The ratio $\tau/T$ goes $0.6527$ at $1g$ , $0.6006$ at $100g$ , $0.6000$ at $a\to\infty$ . The acceleration changed the answer only by changing the velocity profile — by making her spend $2.1$ years turning instead of zero. It contributed nothing of its own. Why: this is the clock hypothesis in action, and it is why the sharp-corner calculation, physically absurd though it is, gives the right answer. *Verified numerically in verify_aq-sr-twin-paradox-asymmetry.py: $\varphi = \operatorname{artanh}(0.8) = \tfrac12\ln 9$ ; $\Delta\tau_b$ , $\Delta t_b$ , $x_{\max}$ and the assembled $T,\tau$ at $1g$ and $100g$ to six figures; and the convergence $\tau/T \to 1/\gamma = 0.6$ as $a \to \infty$ .*

Figure

The three-clock relay — the twin paradox with the acceleration deleted. Clock A (outbound, always inertial) is zeroed at Earth and reads 3 yr at the crossing. Clock B (inbound, always inertial) is set to 3 yr there and reads 6 yr when it reaches Earth. Every accelerometer in the figure reads zero, for all time. Earth reads 10 yr. The 4-year gap is untouched — so acceleration was never what produced it

Limiting cases

Case	Result	Meaning
$v \to 0$	$\tau_{\text{trav}} \to T$ , jump $\to 0$	no trip, no gap, no paradox
$\beta \ll 1$	$T - \tau \approx \tfrac12\beta^2 T$	Einstein's 1905 $\tfrac12 tv^2/c^2$ ; at $\beta=10^{-4}$ it is $5$ ppb
$\beta \to 1$	$\tau_{\text{trav}} \to 0$	the null limit: a light-speed round trip logs no proper time at all
$L \to 0$	jump $\to 0$	the simultaneity gap is distance-driven; a turnaround next door produces none
$a \to \infty$	$\tau/T \to 1/\gamma$ exactly	the sharp corner is the limit of real burns, not a separate physics
$a = 1g$ , $L=4$ ly	$\tau/T = 0.6527$	a real turnaround is not a small correction — it costs $2.1$ yr of her life
Turnaround at $x=0$	still a gap, jump $=0$	she can loop back at home; the age gap is the path, not the jump

Common trapTwo offsets that look alike and are not

The turnaround jump is $2Lv/c^2$ — no $\gamma$ — because it is a difference of readings of a clock at rest in the Earth frame, and Earth-clock readings are Earth coordinate time. It is not the same object as the coordinate offset $\Delta t' = -\gamma v\Delta x/c^2$ between two events simultaneous in $S$ , which does carry $\gamma$ , and it is not the same object as the ignition offset $-\gamma vL/c^2$ in `aq-sr-bell-spaceship-paradox`, which also carries $\gamma$ and is legitimately there. The three have the same units and different meanings; the master key that tells them apart is in `aq-sr-relativity-of-simultaneity-s3`. If you cannot decide which you want, apply equation (4): only the $\gamma$ -free jump closes the books.

Worked examples

1 · The relay, done end to end

Three inertial clocks, $\beta = 0.8$ , $L = 4$ ly. No engines. What does the relayed clock read at Earth, and by how much does it disagree with Earth?

Step 1 — identify. Every clock is inertial, so equation (1) reduces to $\tau = t/\gamma$ on each segment; $\gamma = (1-0.64)^{-1/2} = 1/0.6 = 5/3$ .

Step 2 — apply, segment by segment. Clock A: Earth-time $L/v = 4/0.8 = 5$ yr to the crossing, so $\tau_A = 5 \times 0.6 = 3$ yr. Clock B, set to $3$ yr at the crossing: another Earth-time $5$ yr home, so another $3$ yr of its own. It arrives reading $6$ yr.

Step 3 — Earth. Clock O logs $T = 2L/v = 10$ yr.

Step 4 — the boundary number. The disagreement is $10 - 6 = \mathbf{4\ \text{years}}$ , and there is no acceleration anywhere in the experiment to blame it on. The same $4$ years appear if you use a real ship with a real engine, because equation (1) never asked about the engine. This is the article's whole point, in one number.

Step 5 — check the generality. Symbolically, the relay delivers $2\times\frac{L}{v\gamma} = \frac{2L}{v\gamma} = \frac{T}{\gamma}$ for any $\beta$ and $L$ . It is not an artefact of $0.8c$ .

2 · The traveller's audit, and the $\gamma$ trap

Same trip. The traveller keeps a running estimate of Earth's clock. Reconstruct her ledger and make it balance.

Step 1 — identify. Two inertial legs plus one frame change. On each leg she applies ordinary time dilation to Earth's clock; at the change she must account for the simultaneity shift.

Step 2 — the legs. She ages $3$ yr outbound. She judges Earth's clock to run slow by $\gamma$ , so she credits Earth with $3/\gamma = 3 \times 0.6 = 1.8$ yr. Same inbound: another $1.8$ yr. Total from the legs: $2L/(v\gamma^2) = 3.6$ yr.

Step 3 — the jump. Just before turning, her "now on Earth" is $L/v - vL/c^2 = 5 - 3.2 = 1.8$ yr. Just after, it is $5 + 3.2 = 8.2$ yr. The leap is $8.2 - 1.8 = 6.4$ yr $= 2Lv/c^2$ .

Step 4 — balance. $3.6 + 6.4 = 10.0$ yr $= T$ . ✓ The ledger closes exactly, and it closes for every $\beta$ because $1/\gamma^2 + \beta^2 = 1$ .

Step 5 — the boundary number, and the trap. Now use the widely-printed $2Lv\gamma/c^2$ . It gives $6.4 \times \tfrac53 = 10.67$ yr, so the ledger reads $3.6 + 10.67 = \mathbf{14.27\ \text{yr}}$ — Earth's clock has apparently run $4.27$ years longer than the $10$ years it actually ran. The books-close identity is the cheapest possible defence against this error, and it costs one line.

Going further

Texts. Rindler, Introduction to Special Relativity, 2nd ed., §7 (the clock paradox and the clock hypothesis) · Taylor & Wheeler, Spacetime Physics, 2nd ed., ch. 4 · Møller, The Theory of Relativity, §8 · Langevin, "L'évolution de l'espace et du temps", Scientia X, 31–54 (1911) · Bailey et al., "Measurements of relativistic time dilatation for positive and negative muons in a circular orbit", Nature 268, 301–305 (1977).

Provenance. Signature $(+,-,-,-)$ . Every number recomputed in verify_aq-sr-twin-paradox-asymmetry.py. Base defect repaired: the AllQFT base for this slug prints the turnaround jump as $2Lv\gamma/c^2$ ; the correct value is $2Lv/c^2$ , and the base's own derivation only reaches the right total because it silently substitutes the contracted distance $L/\gamma$ for the Earth-frame $L$ in the very next line. Both errors are fixed here, and the books-close identity (4) is printed so the reader can catch the next one.

Scope. The proper-time (geodesic) resolution and the trip's setup are owned by `aq-sr-twin-paradox` and recalled, not re-derived. The signal-counting resolution is `aq-sr-twin-paradox-doppler-resolution`. The simultaneity machinery is `aq-sr-relativity-of-simultaneity-s3`, which recalls the derivation from `aq-sr-relativity-of-simultaneity-s1`. Rapidity under constant proper acceleration is `aq-sr-rapidity-additive-velocity`.

Builds on. The Twin Paradox · Proper Time · Relativity of Simultaneity. Leads to. Doppler-Shift Resolution of the Twin Paradox · Thomas–Wigner Rotation · Born Rigidity.

If acceleration is not the cause, why does the traveller always turn out to be the one who accelerated?

Because in a connected trip, accelerating is the only way to get from one inertial frame into another. Acceleration is necessary for a single body to bend its worldline — but the age gap is caused by the bend, not by the bending. The relay proves the distinction: split the traveller into two inertial clocks and you get the bend (a kinked worldline in spacetime) with no acceleration (no single body ever changes speed). The gap tracks the kink.

Does the traveller really see Earth's clock leap forward by 6.4 years?

She does not see anything of the sort — light delays mean nothing jumps in her telescope, and the Doppler account in `aq-sr-twin-paradox-doppler-resolution` shows a perfectly smooth view. The jump is in what she calculates Earth's clock to read "right now", and "right now" is a frame-dependent bookkeeping choice that she changes at the turnaround. The lesson is that the jump is an artefact of a convention, which is precisely why the Doppler resolution — which uses no simultaneity convention at all — can dispense with it entirely and get the same $6$ and $10$ .

Could a clock exist whose rate does depend on acceleration?

Nothing in the postulates forbids it, which is why the clock hypothesis is an assumption and not a theorem. Any real clock has internal structure, and a sufficiently violent acceleration will simply destroy it — a pendulum clock in a centrifuge is not measuring proper time, it is breaking. The claim is about ideal clocks, and it is tested by finding a clock robust enough to survive: the muon, whose decay is governed by the weak interaction and whose "mechanism" is essentially structureless, survives $10^{18}g$ and keeps perfect proper time.

Is the turnaround jump physically real, or an accounting fiction?

It is a real feature of a particular accounting scheme, and it is not a physical event. The number $6.4$ years is exactly and unambiguously defined once you say "the traveller's instantaneous inertial rest frame" — but the choice to use that frame's simultaneity is a convention, and different (equally valid) conventions give different-looking narratives with identical predictions. What is not conventional: Earth's clock reads $10$ at the reunion, hers reads $6$ . Every scheme must reproduce those two numbers, and the ones that do not are simply wrong.

The traveller ages 7.16 years with a 1g turnaround instead of 6. Is the idealised answer therefore useless?

No — it is exactly right for the trip it describes, which is a trip with an infinitely sharp corner. The $1g$ trip is a different trip: it takes longer (Earth $10.97$ yr instead of $10$ ) and spends $2.13$ years of her life turning. Both are computed from the same equation (1), and neither contains an acceleration term. The idealisation is a modelling choice, not a physics approximation, and its accuracy is controlled entirely by how small $c\varphi/a$ is compared with $L/v$ .

Von Laue got the right explanation in 1911. Why is Langevin's wrong one still taught?

Partly because Langevin's is memorable and von Laue's requires a Minkowski diagram. Mostly, though, because Langevin's story is nearly right — acceleration really is what an accelerometer measures, it really is invariant, and it really does mark the traveller out. The error is subtle: it mistakes an infallible indicator of the non-geodesic path for the cause of the age gap. The three-clock relay is the cleanest way to expose the difference, because it deletes the indicator and keeps the effect.

Can the traveller turn around without ever leaving an inertial frame?

Not as a single body — but she does not have to be a single body. That is the relay. If what you care about is the reading that arrives back at Earth, the reading can be handed from clock to clock, and no carrier ever leaves its frame. What the relay makes explicit is that "the traveller" was never a physical prerequisite; the age gap belongs to the worldline, and a worldline can be assembled from pieces.

Asymmetry and Acceleration in the Twin Paradox

Where this lives today

The physical picture

The core equations

Where the asymmetry actually lives

Limiting cases

Worked examples

1 · The relay, done end to end

2 · The traveller's audit, and the $\gamma$ trap

Going further

Before this

Where this leads

Asymmetry and Acceleration in the Twin Paradox

Where this lives today

The physical picture

The core equations

Where the asymmetry actually lives

Limiting cases

Worked examples

1 · The relay, done end to end

2 · The traveller's audit, and the γ\gammaγ trap

Going further

Before this

Where this leads

2 · The traveller's audit, and the $\gamma$ trap