The Relativity of Simultaneity

Newton opened the Principia with a clock. Absolute time, he wrote, flows equably without relation to anything external — one universal tick for the whole of creation, so that did these two things happen at the same moment? always has an answer, and the same answer for everybody. Two centuries of mechanics were built on that clock — Absolute Space and Absolute Time shows how much of Newton leans on it. In 1905 it broke. What broke it was not an experiment. It was a definition. To say that a flash on the Moon and a flash in your kitchen happened at the same instant, you must do something: send a signal, catch an echo, compare readings. The only signal whose speed everyone agrees on is light, and Einstein's second postulate fixes that speed at $c$ in every inertial frame. Follow the consequence and the universal "now" does not survive: two observers in relative motion, each performing an impeccable measurement, disagree about which distant events were simultaneous — and neither is mistaken. There is no fact for them to be mistaken about. Einstein made it vivid with a train: lightning strikes both ends at once, says the embankment, while the passenger riding the midpoint meets the two flashes at different times and concludes the strikes were not simultaneous. The argument needs nothing but distance = speed × time with the speed pinned at $c$ — and it goes all the way to a number. For a $100\ \mathrm{m}$ train at $\beta = 0.6$ the flashes reach her $312.7\ \mathrm{ns}$ apart, the front one four times sooner.

Key takeaways

The proof needs only the constancy of $c$ and never leaves the embankment frame: the rider at the train's midpoint runs into the front flash at $t_+ = L/[2(c+v)]$ and is chased down by the rear flash only at $t_- = L/[2(c-v)]$ — an exact gap $t_- - t_+ = Lv/(c^2 - v^2)$ .
At $L = 100\ \mathrm{m}$ , $\beta = 0.6$ : $416.9 - 104.2 = \mathbf{312.7\ ns}$ , the front flash arriving $(1+\beta)/(1-\beta) = \mathbf{4}$ times sooner — a number from a train and two bolts of lightning and nothing else.
Simultaneity is first order in $\beta$ ; the rate and length factors are second order, $1 - \sqrt{1-\beta^2} = \beta^2/2 + \cdots$ . Simultaneity beats them by $2/\beta$ — a factor of 7.2 million for a train at $300\ \mathrm{km/h}$ . That is why nearly every "paradox" in relativity is a simultaneity blunder.
Key idea: "now" is not a slice of the universe, it is a slice of a frame. Boost along the line joining two simultaneous events and one moves into the other's past. Nothing is illusory: two observers have made different, equally correct measurements of a quantity that was never absolute. Only pairs too far apart for light to connect can be reordered: causal order is protected by the very postulate that destroyed "now".

The core insight

"Now" is not a property of the universe; it is a property of a frame. Two events simultaneous in one frame and separated by $\Delta x$ along the motion acquire a time offset of order $v\,\Delta x/c^2$ for an observer moving at $v$ — first order in $v/c$ , while every other relativistic effect is second order.

Where this lives today

A GPS satellite moves at about $3.9\ \mathrm{km/s}$ and its signal is read thousands of kilometres away: $\beta = 1.3\times10^{-5}$ across a $\Delta x$ of $10^{7}\ \mathrm{m}$ , so $v\,\Delta x/c^{2}$ is tens of nanoseconds — tens of metres of range error once multiplied by $c$ . The system does not fight this by pretending simultaneity is absolute. It fixes one non-rotating coordinate time by fiat, refers every clock in the constellation to it, and corrects every one-way signal transfer explicitly (the engineering is GPS and Relativistic Timing Corrections (S7); the protocol beneath it is Einstein Synchronization of Clocks (1860)). The largest velocity-dependent relativistic term in the whole GPS budget is a simultaneity term, first order in $\beta$ , exactly as this article says it must be.

The physical picture

Stand on an embankment. Two bolts of lightning strike, one at each end of a passing train, and they strike together — you know they did, because you stand exactly midway between the scorch marks they left on the rails and both flashes reached you at the same instant. Equal distances at equal speed take equal time. That is not an inference; it is a measurement. Now the passenger. She rides the train's exact midpoint, level with you at the instant of the strikes, and keeps going: toward where the front bolt struck, away from where the rear one struck. The front flash has less ground to cover, so she meets it first — and you watch it happen. Here is where Newton's clock dies. She is entitled to reason exactly as you did. The scorch marks are on her train, one at each end, and she is at the middle, equidistant from both — she can pace it out afterwards with a tape measure and no clock at all. By the second postulate the light from each mark came to her at $c$ in her frame: equal distance, equal speed, equal transit time. She got the front flash first, so the front bolt struck first. She is not confused and not wrong. You are both right, and you disagree.

Figure

Einstein's train in the embankment frame. The bolts strike both ends together at t = 0; the rider then meets the front flash at 104.2 ns and is caught by the rear flash only at 416.9 ns, while the motionless embankment observer meets both at 166.8 ns. β = 0.6, L = 100 m.

Figure

PUNCHLINE — the whole argument as one position–time plot in the embankment frame. Light rays run at 45°. The rider's line (β = 0.6) cuts the front flash at ct = 31.25 m and the rear at ct = 125 m: ratio exactly 4:1, gap 312.7 ns. The embankment observer's vertical line cuts both at ct = 50 m.

The core equations

Everything here is squeezed out of one sentence — light covers ground at $c$ in this frame, whatever the source was doing — applied in the embankment frame and nowhere else. Put the strike points at $x = \pm L/2$ , both bolts falling at $t = 0$ ; the rider is at $x = 0$ then and $x = vt$ after. The front flash runs leftward from $+L/2$ , so she and it close head-on at $c + v$ ; the rear flash runs rightward from $-L/2$ and must close a gap opening at $c - v$ .

t_+ = \frac{L}{2(c+v)}, \qquad t_- = \frac{L}{2(c-v)} \qquad (1)

where: $t_+$ — embankment time at which the front flash reaches the rider; $t_-$ — the same for the rear flash; $L$ — separation of the strike points in the embankment frame; $v$ — the train's speed; $c$ — the speed of light, the same for both. Physical reading: these are not relativistic formulas at all, just the arithmetic of a closing gap. Relativity's only contribution is the refusal to write $c \pm v$ for the speed of the light — the light goes at $c$ ; the rider supplies the extra $\pm v$ . Subtract, and the whole effect appears in one line — exact, with no expansion and nothing borrowed. Take the ratio instead and $L$ and $c$ both cancel.

t_- - t_+ = \frac{Lv}{c^{2} - v^{2}} = \frac{L}{c}\cdot\frac{\beta}{1 - \beta^{2}}, \qquad \frac{t_-}{t_+} = \frac{c+v}{c-v} = \frac{1+\beta}{1-\beta} \qquad (2)

where: $\beta \equiv v/c$ . Physical reading: the gap is linear in $v$ at small $\beta$ — reverse the train and it changes sign, which already says there is no fact of the matter about which bolt fell first. It also grows with $L$ : the effect is speed times separation — invisible on a bench, unavoidable across a satellite constellation. At $\beta = 0.6$ the ratio is $1.6/0.4 = 4$ , exactly. Finally, the result that organises the rest of special relativity. Two clocks the crew synchronized by their own light signals sit at the train's ends, a proper distance $L_0$ apart, and Derivation B gets their disagreement with no machinery at all.

\Delta t_{\text{clocks}} = \frac{L_0 v}{c^{2}}\Big(1 + O(\beta^{2})\Big), \qquad 1 - \sqrt{1-\beta^{2}} = \frac{\beta^{2}}{2} + \frac{\beta^{4}}{8} + \cdots \qquad (3)

where: $\Delta t_{\text{clocks}}$ — how far the rear clock reads ahead of the front one at a single embankment instant ("leading clocks lag"); $L_0$ — the crew's own measured train length; $\sqrt{1-\beta^2}$ — the rate-and-length factor derived in The Light Clock and Time Dilation (S1) and Length Contraction (S1), and not needed here. Physical reading: on a system whose light-crossing time is $L/c$ , the simultaneity offset is a fraction $\beta$ of it while the second-order effects are a fraction $\beta^2/2$ . Simultaneity beats them by $2/\beta$ , which at ordinary speeds is enormous.

Where the result comes from

Derivation AEinstein's train, made exact, in the embankment frame alone

The textbook version stops at "and so the passenger says the bolts were not simultaneous" — true, but it leaves the impression that the argument is a hand-wave, a story told before the real algebra arrives with the Lorentz transformation. It is not. The whole thing carries through to an exact number using only the second postulate and the definition of speed, in one frame, transforming nothing — and it then draws the boundary that keeps cause before effect.

Step 1. Set the stage in the embankment frame and stay there. Bolts at $t = 0$ , $x = \mp L/2$ ; rider at $x_{\text{rider}} = vt$ ; flashes at $x_{\text{front}} = L/2 - ct$ and $x_{\text{rear}} = -L/2 + ct$ . Why: that last line is the second postulate doing its only job. The bolts struck a moving train, but the light does not care — its speed in this frame is $c$ , not $c \pm v$ .

Step 2. Meet the front flash: $vt = L/2 - ct$ gives $t_+ = L/[2(c+v)]$ . With $L = 100\ \mathrm{m}$ , $\beta = 0.6$ , $c = 2.998\times10^{8}\ \mathrm{m/s}$ : $t_+ = \mathbf{104.2\ ns}$ , by which time light has run $31.25\ \mathrm{m}$ and she $18.75\ \mathrm{m}$ . Why: she and the flash approach each other, so their separation shrinks at $c + v$ — a rate of closing, not a speed. Nothing moves faster than $c$ .

Step 3. Be caught by the rear flash: $vt = -L/2 + ct$ gives $t_- = L/[2(c-v)] = \mathbf{416.9\ ns}$ , with $ct_- = 125\ \mathrm{m}$ . Subtract: $t_- - t_+ = Lv/(c^2-v^2) = \mathbf{312.7\ ns}$ ; divide: $t_-/t_+ = 1.6/0.4 = \mathbf{4}$ exactly. Why: the rear flash chases a receding target and closes at only $c - v$ — four times slower than the head-on approach.

Step 4. Now let her reason. The scorch marks are burned into her train's ends, so in her frame they are equidistant from her — checkable with a tape measure and no clock. P2 says the light from each came to her at $c$ in her frame: equal distance, equal speed, equal transit. She got the front flash first, so the front bolt struck first. Why: load-bearing, and it uses P2 twice — once in the embankment frame for the reception times, once in hers to license the inference. Drop the postulate and the argument collapses.

Step 5. Flip the train. Send it at $-v$ and every sign reverses: she meets the rear flash first and concludes the rear bolt struck first. Why: a quantity whose sign you can choose by choosing your speed is not a property of the events. That is what "no fact of the matter" means, stated as arithmetic.

Step 6. Now bound the damage. If a boost can move an event into your past, can it move an effect before its cause? It cannot, and the guarantee comes from the same postulate. Take a pulse emitted at $E$ , absorbed at $R$ : in any frame P2 says it travels at $c$ , so it covers $|\Delta x'| \geq 0$ at speed $c$ and arrives having taken $\Delta t' = |\Delta x'|/c \geq 0$ . No boost makes that negative. Why: the machinery that destroyed the universal "now" is the machinery that protects causality. Relativity does not merely tolerate causal order; it is the first theory in which causal order is a theorem.

Step 7. State the boundary: a pair can be reordered exactly when no signal at $c$ or below could have run between them, $|\Delta x| > c|\Delta t|$ — spacelike separation, whose invariant machinery belongs to The Invariant Spacetime Interval (S1). Why: the bolts satisfy it with room to spare — light needs $333.6\ \mathrm{ns}$ to cross $100\ \mathrm{m}$ and they fell together, so the train could reorder them. Make them $500\ \mathrm{ns}$ apart and $c|\Delta t| = 150\ \mathrm{m} > 100\ \mathrm{m}$ : no train below $c$ can reverse them. *Verified in verify_aq-sr-relativity-of-simultaneity.py: $t_\pm$ solved symbolically from the meeting conditions, the difference simplified to $Lv/(c^2-v^2)$ and the ratio to $(1+\beta)/(1-\beta)$ ; numerically $104.24\ \mathrm{ns}$ , $416.94\ \mathrm{ns}$ , gap $312.71\ \mathrm{ns}$ , ratio exactly $4$ , $ct_\pm = 31.25\ \mathrm{m}$ and $125\ \mathrm{m}$ ; the $v \to -v$ sign flip; the $333.6\ \mathrm{ns}$ crossing time; and the criterion $|\Delta x| > c|\Delta t|$ evaluated for the spacelike pair, the timelike pair and the light-connected boundary case.*

Go deeper · Derivation B · 2 minWhy simultaneity is first order in beta and everything else is second

Derivation BWhy simultaneity is first order in beta and everything else is second

Students are told that time dilation and length contraction are the famous consequences of relativity, and simultaneity a curiosity mentioned in passing. The truth is the reverse: at any speed you will ever meet, the simultaneity offset is vastly larger than either, because it enters at first order in $\beta$ while they enter at second. The goal is to prove that ordering — and to get the offset formula honestly, without the Lorentz transformation.

Step 1. Let the crew synchronize their end clocks the only way they can — a flash at the train's midpoint, each clock started when the flash arrives. In their frame this is impeccable (equal distances at speed $c$ ), and it is exactly the rule Einstein Synchronization of Clocks (1860) sets out. Now watch that flash from the embankment: it leaves the midpoint at $t = 0$ , the rear runs toward it and the front away, so the rear clock starts at $t_R = (L/2)/(c+v)$ and the front at $t_F = (L/2)/(c-v)$ . Why: equations (1) again, with "rider" and "flash" swapped. The train argument and the synchronization argument are the same geometry read backwards — which is why the two subjects are one subject.

Step 2. Subtract: the front clock is started later, by $t_F - t_R = Lv/(c^2-v^2)$ , and reads behind the rear one ever after. The leading clock lags. Why: both were zeroed on arrival of the flash, so a clock started later carries a smaller reading forever. The offset is frozen in at synchronization and never goes away.

Step 3. Convert the delay into a difference of readings. Whatever rate $\rho$ the moving clocks tick at — a number the embankment must get from S1, not from us — the reading gap is $\rho \cdot Lv/(c^2-v^2)$ . But $\rho = \sqrt{1-\beta^2} = 1 + O(\beta^2)$ , and $L$ differs from the crew's own $L_0$ only at $O(\beta^2)$ too. Both are second order, so to first order $\Delta t_{\text{clocks}} = L_0 v/c^2$ . Why: we have not derived the rate factor and do not need to — two second-order corrections cannot contaminate a first-order answer, so the leading term is ours free and clear.

Step 4. Establish the ordering. Expand: $1 - \sqrt{1-\beta^2} = \beta^2/2 + \beta^4/8 + \cdots$ , so the fractional rate-slowing and length-shortening are each $\beta^2/2$ . Against the system's own light-crossing time $L/c$ , the simultaneity offset is $(L_0v/c^2)/(L/c) = \beta$ , so the ratio is $\beta/(\beta^2/2) = 2/\beta$ . At $300\ \mathrm{km/h}$ , $\beta = 2.78\times10^{-7}$ and $2/\beta = \mathbf{7.2\times10^{6}}$ . Why: the only natural yardstick a system of size $L$ owns is the time light needs to cross it. This is why the paradox chapter of every relativity book is really a simultaneity chapter — reach for $\sqrt{1-\beta^2}$ before asking whose "now" you are using, and you have already lost the leading term. *Verified in verify_aq-sr-relativity-of-simultaneity.py: $t_F - t_R$ derived symbolically and shown identical to Derivation A's gap; $Lv/(c^2-v^2) = (L/c)\beta + O(\beta^3)$ ; the series $1-\sqrt{1-\beta^2} = \beta^2/2 + \beta^4/8 + O(\beta^6)$ with zero linear term; and $2/\beta = 7.195\times10^{6}$ at $300\ \mathrm{km/h}$ .*

Limiting cases

Case	Result	Meaning
Galilean limit, $c \to \infty$ ( $\beta \to 0$ )	$t_\pm \to 0$ , gap $\to 0$	Both flashes arrive together: simultaneity is absolute again and the universal clock of Absolute Space and Absolute Time is restored. Absolute time is not false — it is the $\beta \to 0$ limit.
Same place, $\Delta x = 0$	offset $= 0$ for every $v$	Only distant simultaneity is relative: collisions and coincidence circuits mean the same thing in every frame.
Reverse the boost, $v \to -v$	gap changes sign	The bolts swap places in the ordering. A quantity whose sign you may choose is not a property of the events.
Ultra-relativistic, $\beta \to 1$	$t_- \to \infty$ , ratio diverges	The rear flash can barely catch her.
First versus second order	offset $/(L/c) = \beta$ against $1 - \sqrt{1-\beta^2} = \beta^2/2$	Simultaneity outweighs the rate and length factors by $2/\beta$ — millions, at everyday speeds.
Light-connected pair, $\vert\Delta x\vert = c\vert\Delta t\vert$	order never reverses	Causal order is protected by P2 itself. Only spacelike pairs are reorderable.

Common trapThe relativity of simultaneity is NOT the delay of light reaching your eyes — it is what survives after every observer has corrected for that delay

The train picture invites the objection: "she just sees the front flash first because she moved toward it." True, and not the point. Both observers do correct for light-travel time — that is precisely Step 4 of Derivation A. She subtracts the transit time from each mark to her own position, using her own frame's distances and speed $c$ , and having subtracted them she still concludes the bolts were not simultaneous. Correcting for signal delay is bookkeeping every astronomer does before breakfast; the disagreement is what remains afterwards. And a frame is not an eye: it is a network of clocks, each recording events where it stands, so no travel-time correction is ever applied at all (see Frames, Observers, and Networks of Clocks).

Worked examples

1 · Einstein's train, to the nanosecond

The strike points are $L = 100\ \mathrm{m}$ apart in the embankment frame; the train runs at $\beta = 0.6$ . Find both reception times, the gap and the ratio — and check the embankment observer sees no gap.

Step 1. Front flash. She and it close head-on at $c + v = 1.6c$ across the initial half-separation of $50\ \mathrm{m}$ : $t_+ = 50/(1.6 \times 2.998\times10^{8}) = \mathbf{104.2\ ns}$ . Light covered $31.25\ \mathrm{m}$ , she $18.75\ \mathrm{m}$ — together, the $50\ \mathrm{m}$ they had to close.

Step 2. Rear flash. The light must make up $50\ \mathrm{m}$ against a target running away, so the gap closes at only $c - v = 0.4c$ : $t_- = 50/(0.4 \times 2.998\times10^{8}) = \mathbf{416.9\ ns}$ , with $ct_- = 125\ \mathrm{m}$ against her $75\ \mathrm{m}$ .

Step 3. Gap and ratio. $t_- - t_+ = \mathbf{312.7\ ns}$ , which is $Lv/(c^2-v^2) = (100)(0.6)/[(2.998\times10^{8})(0.64)]$ to the last digit; and $t_-/t_+ = \mathbf{4.000}$ , the clean form $(1+\beta)/(1-\beta)$ with $L$ and $c$ cancelled out of existence.

Step 4. The control. The embankment observer never moves, so each flash covers $50\ \mathrm{m}$ at $c$ and both reach him at $\mathbf{166.8\ ns}$ — together, as they must. Two observers coincident at $t = 0$ ; their accounts of the universe differ by $312.7$ ns. The gap is not an error. It is the answer.

2 · The train you can actually catch

A commuter train is $L_0 = 200\ \mathrm{m}$ long and runs at $300\ \mathrm{km/h}$ . Its crew synchronize a clock at each end by the midpoint-flash rule. By how much do they disagree, judged from the platform, and how does that compare with the rate at which they run slow?

Step 1. $300\ \mathrm{km/h} = 83.33\ \mathrm{m/s}$ , so $\beta = 2.78\times10^{-7}$ — small, but the question is what power of it survives. From equation (3), $\Delta t_{\text{clocks}} = L_0 v/c^2 = (200)(83.33)/(8.988\times10^{16}) = \mathbf{0.185\ ps}$ , the rear clock reading ahead. In that time light travels $\mathbf{55.6\ \mu m}$ , about the width of a human hair: invisible to a wristwatch, a comfortable measurement for an optical clock resolving fractional frequency shifts near $10^{-18}$ .

Step 2. Compare with the second-order effects. The rate deficit is $1 - \sqrt{1-\beta^2} = \beta^2/2 = 3.86\times10^{-14}$ . Against the train's own light-crossing time $L_0/c = 667\ \mathrm{ns}$ , the simultaneity offset is a fraction $\beta = 2.78\times10^{-7}$ — a ratio of $2/\beta = \mathbf{7.2\ million}$ . On this train, "the clocks disagree about now" is seven million times bigger than "the clocks run slow" — and it is the second effect everyone has heard of.

Physics of the portrait

The portrait is an idea, not a person: it has no single author, and its authors did not agree about what it meant. Poincaré had the operational half of it seven years before Einstein. What he would not do — and this is the whole difference — was let the definition bite.

Origins · 2 minRelativity of Simultaneity and how this was discovered

Relativity of Simultaneity

1898–1905 · Poincaré's convention · Einstein's postulate · the end of absolute time

The idea arrives in two acts, the second a change of nerve rather than of mathematics. Act one: Henri Poincaré (1854–1912). In La Mesure du Temps (Revue de métaphysique et de morale, 1898) he dismantles the intuition — we have, he says, no direct intuition of the simultaneity of distant events — and shows that a distant clock can only be set by signals, on the convention that light's outbound and return trips take equal time. In 1900 he reads Lorentz's mathematical fiction of "local time" as the physical reading of clocks synchronized in exactly that way. Every ingredient is on the table. Act two: Albert Einstein (1879–1955). In §§1–2 of "Zur Elektrodynamik bewegter Körper" (1905) he adopts the same definition and does what Poincaré would not: he deletes the ether, deletes the true time the local time was supposed to approximate, and lets the definition stand alone. He demonstrated it with a moving rod carrying a clock at each end; the lightning-and-train story came twelve years later, in the popular Über die spezielle und die allgemeine Relativitätstheorie (Vieweg, 1917). Poincaré, who died in 1912, never took the step. The mathematics he had written; the courage was Einstein's.

History

1687

— Newton's Principia posits absolute time, "flowing equably without relation to anything external", and with it a universal "now" — the assumption this article destroys (Absolute Space and Absolute Time).

1898

— Poincaré, La Mesure du Temps: we have no intuition of distant simultaneity, and clocks can only be set by light signals, on the convention that light takes equal time each way.

1900

— Poincaré gives Lorentz's "local time" an operational reading: it is what clocks synchronized by light signals actually read.

1905

— Einstein, "Zur Elektrodynamik bewegter Körper", §§1–2: the synchronization definition, then the demonstration — with a moving rod carrying clocks at both ends — that simultaneity is frame-relative.

1917

— Einstein's popular book Über die spezielle und die allgemeine Relativitätstheorie (Vieweg, Braunschweig) introduces the embankment-and-train story. It is not in the 1905 paper and never was.

1924

— Eddington shows that a clock carried slowly between two points agrees with light-signal synchronization, sharpening rather than settling how much of "now" is convention.

Going further

Texts. Taylor and Wheeler, Spacetime Physics, 2nd ed., Ch. 3 · Rindler, Introduction to Special Relativity, 2nd ed., §2 · Griffiths, Introduction to Electrodynamics, 4th ed., §12.1.2 · Einstein, Annalen der Physik 17 (1905), §§1–2 · Einstein, Über die spezielle und die allgemeine Relativitätstheorie (1917), §9.

Provenance. The reception-time argument is Einstein's own, made quantitative. Anchors, all recomputed in verify_aq-sr-relativity-of-simultaneity.py: $c = 2.998\times10^{8}\ \mathrm{m/s}$ ; $L = 100\ \mathrm{m}$ at $\beta = 0.6$ giving $104.24\ \mathrm{ns}$ , $416.94\ \mathrm{ns}$ , gap $312.71\ \mathrm{ns}$ , ratio exactly $4$ , light paths $31.25$ and $125\ \mathrm{m}$ , embankment reception $166.8\ \mathrm{ns}$ ; the commuter train ( $200\ \mathrm{m}$ , $300\ \mathrm{km/h}$ , $\beta = 2.78\times10^{-7}$ ) giving $L_0v/c^2 = 0.185\ \mathrm{ps} = 55.6\ \mathrm{\mu m}$ against $\beta^2/2 = 3.86\times10^{-14}$ and $2/\beta = 7.2\times10^{6}$ ; the series $1 - \sqrt{1-\beta^2} = \beta^2/2 + \beta^4/8 + \cdots$ ; and the $333.6\ \mathrm{ns}$ crossing time of $100\ \mathrm{m}$ . Scope note: the coordinate form of the result is read off the Lorentz transformation and is not derived here — see Derivation of the Lorentz Transformation (S1) and The Lorentz Transformation Equations (S1). Everything above is first order in $\beta$ , where nothing from S1 is owed.

Builds on. The Constancy of the Speed of Light · Einstein's Two Postulates · Inertial Frames of Reference · Absolute Space and Absolute Time. Leads to. Einstein Synchronization of Clocks · The Conventionality of Simultaneity · The Light Clock and Time Dilation · Length Contraction.

Isn't this just about when the light gets to her? Correct for travel time and the problem goes away.

Backwards — correcting for travel time is part of her argument, not an escape from it. She takes the two reception events, subtracts the transit time from each mark to her position (measuring those distances in her own frame, with light at $c$ ), and finds the emission times still differ. Both observers do the correction; both do it right; the disagreement is what is left over. Better still, do away with eyes: build a frame as Frames, Observers, and Networks of Clocks does, with a clock at every point recording events where it stands. No travel time is involved at all, and the disagreement is unchanged.

If they disagree about the order of the bolts, which one is actually right?

Neither, and the question has no answer — not because we are ignorant but because there is nothing to be right about. The bolts are $100\ \mathrm{m}$ apart and fell together, so neither could have influenced the other. Their order is not a fact observers estimate with varying success; it is a label that comes with a frame, like "left" and "right". Where there is something to be right about — a cause and its effect, or any two events a signal could connect — every observer agrees, and Steps 6–7 of Derivation A show why P2 forces them to.

Could a fast enough boost let me kill the cause after seeing the effect?

No — the postulate that broke simultaneity is the one that protects causality. If a signal ran from $E$ to $R$ then in every inertial frame it moves at $c$ or slower, so it takes a non-negative time to cover a non-negative distance and $R$ comes after $E$ . No boost flips this. The events whose order can be flipped are exactly the ones no signal could connect, $|\Delta x| > c|\Delta t|$ — and reordering those changes nothing, since neither could have influenced the other. Relativity does not weaken causality; it makes it a theorem.

Is the 312.7 ns gap also what the passenger measures between the two bolts?

No, and the distinction matters. The $312.7\ \mathrm{ns}$ is the interval, in the embankment frame, between two events at the rider's own position: the arrival of each flash. What the rider says about the interval between the strikes themselves is a different question, and answering it in closed form means transforming coordinates — the business of the Lorentz transformation, which belongs to S1. Einstein's point is that you need none of that machinery to establish the fact that the bolts were not simultaneous for her.

If simultaneity is bigger than time dilation, why does every popular account lead with time dilation?

Because time dilation has a headline and simultaneity has a subordinate clause. But look at the scalings: the offset between the ends of a system is first order in $\beta$ , the rate deficit second order, so at $300\ \mathrm{km/h}$ the first beats the second by $7.2$ million. Every classic paradox — the pole and the barn, the twins, the rod that seems to outrun light — is resolved by finding a dropped simultaneity term, never by re-deriving $\sqrt{1-\beta^2}$ more carefully.

Did Einstein really use a train?

He did, but later than you have been told — this is one of the most reliably mis-sourced anecdotes in physics. The 1905 paper does not mention a train. Its §2 uses a rigid rod moving along the $x$ -axis with a clock at each end, and shows that observers riding the rod judge those clocks unsynchronized while stationary observers judge them synchronized. The lightning, the embankment and the railway carriage first appear twelve years later, in the popular book Über die spezielle und die allgemeine Relativitätstheorie (Vieweg, Braunschweig, 1917). If you see the train credited to the 1905 paper, the source is repeating a legend.