Special Relativity & Spacetime · Guide · glidepathscience.org
GuideSpecial Relativity & Spacetime
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round, in the order a textbook takes them. Part 1 prepares the Intro track and Part 2 the Advanced track; each takes about a quarter of an hour to read. Each section ends with a worked example that is not one of the sim’s own problems.
Part 1 · The essentials
Math you’ll need: algebra, square roots and the Pythagorean theorem; no calculus.
One speed for everyone
Throw a ball forward at 20 m/s from a train moving at 30 m/s, and someone on the platform sees the ball move at 50 m/s. Speeds add: that rule, Galileo’s, works so well in everyday life that it seems like common sense. Light breaks it. However fast its source moves, and however fast the person measuring it moves, light in a vacuum travels at the same speed, $c$ = 299,792,458 m/s, about 30 cm in a nanosecond.
Special relativity is what follows from taking that fact seriously. If two observers moving past each other measure the same speed for the same flash of light, they cannot also agree on how long things take or how long things are. Moving clocks run slow, moving rulers are short, and two events that happen at the same moment for one observer happen at different moments for another. None of this is an illusion or a trick of perspective: it is what careful measurements, with clocks and rulers, give.
The two postulates
In 1905 Albert Einstein built special relativity on two statements, its postulates. The first: the laws of physics are the same in every inertial frame, every frame that moves at constant velocity. No experiment done inside a windowless train car gliding smoothly along can tell you the car is moving; a pendulum swings, a magnet attracts and a light pulse travels between two mirrors exactly as they do at rest. Galileo had already noticed this for mechanics; Einstein extended it to every law of physics, light included. The second postulate: the speed of light in vacuum is $c$ for every inertial observer, whatever the motion of its source.
The second postulate is the strange one. A passenger on a fast train who points a flashlight forward measures its light moving at $c$ relative to the train. Galileo’s rule says the platform should measure $c + v$. It does not: the platform also measures $c$. Experiments with light from fast-moving sources, from binary stars to particles flying at nearly light’s speed, have tested this again and again.
If the speed of light is the same for everyone, something else must give. The first casualty is time.
The light clock and γ
A light clock is the simplest possible clock: two parallel mirrors a distance $L$ apart, and a photon bouncing between them. One tick is a round trip, up and back, so a clock at rest ticks every
$Δt_0 = 2L/c$
Now put the clock on a train moving at speed $v$, with its mirrors one above the other, and watch it from the platform. Between leaving the bottom mirror and reaching the top one, the photon is carried forward with the train, so for the platform it travels along a slanted line. The slanted path is longer than $L$, and by the second postulate the photon still moves along it at $c$. So for the platform each tick of the moving clock takes longer than $Δt_0$.
How much longer follows from a right triangle. Call the platform’s time for one tick $Δt$. In half a tick the photon goes up $L$ while the clock moves forward $vΔt/2$, and the photon’s own path, the slanted side, is $cΔt/2$. Pythagoras gives $(cΔt/2)^2 = L^2 + (vΔt/2)^2$. Solving for $Δt$ and using $Δt_0 = 2L/c$,
$Δt = Δt_0/\sqrt{1 − v^2/c^2} = γΔt_0$ $γ = 1/\sqrt{1 − β^2}$, $β = v/c$
The factor γ, the Lorentz factor, is always at least 1. At everyday speeds it is so close to 1 that no one notices: at a tenth of light’s speed it differs from 1 by about half a percent, and for an airliner by a few parts in $10^{13}$. Near $c$ it grows without limit. The common slip in computing it is to forget to square β, or to forget the square root.
Every clock on the train runs slow by this same factor, light clocks and wristwatches alike: if they did not, the passengers could compare them and tell that they were moving, which the first postulate forbids.
Proper time and time dilation
Which clock reads $Δt_0$ and which reads $Δt$? The answer is the most useful idea in this subject. A clock that travels from one event to another reads the proper time of its journey between them. During a passenger’s nap, her watch rides with her from the nap’s start to its end, so the time it shows is the proper time $Δt_0$ of that journey. The platform cannot time the nap with one clock, because the passenger is carried from one end of the platform to the other; it uses two of its own clocks, one where the nap starts and one where it ends, synchronized with each other in its frame. When the passenger moves at constant velocity, those two clocks measure the longer time,
$Δt = γΔt_0$
This is time dilation. Two cautions keep it straight. First, proper time belongs to a clock’s journey, not to a pair of events alone: a clock that goes out and comes back between the same two events reads less than one that waits (the twin paradox, below), and two events too far apart for any clock to travel between them have no proper time at all. Second, dilation is mutual, and each frame measures it its own way. The platform times the passenger’s watch with two platform clocks, at the places where the watch is at the nap’s start and end. The passenger, in turn, times one platform clock with two of the train’s clocks, synchronized in the train’s frame, as that platform clock passes them, and finds it running slow by the same factor. There is no contradiction: the two experiments use different pairs of events and different sets of synchronized clocks, and the frames disagree about whether the other’s clocks are synchronized (the section on simultaneity explains how). If the passenger simply compares her watch with the platform clocks she passes one after another, they read ahead of it: that is the platform frame’s measurement of her slow watch, not hers of theirs.
A common mistake is to think dilation only looks like it happens, because light from a moving clock takes time to reach us. It is not that. The platform’s clocks sit where each tick happens, so no travel time enters their readings; and dilation has consequences no appearance could have, such as muons that live long enough to reach the ground. What an observer sees (light arriving at her eyes) and what her frame measures (clocks at each place) are different things; this sim labels them separately wherever both appear.
Length contraction
If moving clocks run slow, lengths must change too, or the frames would disagree about simple facts such as whether a muon reaches the ground. An object’s length in its own rest frame is its proper length $L_0$. In any frame where it moves along its length at speed $v$, it is shorter,
$L = L_0/γ$
The contraction is along the motion only. A meter stick flying past lengthwise is short; its height and width are unchanged. (If heights changed, two sticks passing each other side by side would each have to be shorter than the other, a contradiction.) As with dilation, the effect is mutual, and measured, not seen: measuring a moving object’s length means marking where its two ends are at the same moment, and the next sections show why frames disagree about “the same moment”. The usual slip is to multiply by γ, making the moving object longer; the rest length is always the longest.
Muons reach the ground
Cosmic rays striking the upper atmosphere make muons, heavy cousins of the electron that decay with a mean lifetime of $τ_0$ = 2.20 µs at rest. Even at nearly light’s speed, a muon travels only about 660 m in 2.20 µs, yet muons made many kilometers up reach detectors at sea level in large numbers. Relativity explains it twice over, once in each frame.
In Earth’s frame the trip from a height $h$ at speed $v$ takes $t = h/v$, and the muon’s clock runs slow: it ages only $t_0 = t/γ$. In the muon’s own frame, the muon is at rest and the atmosphere rushes up at $v$; the air’s thickness is contracted to $h/γ$, and the ground arrives after $h/(γv)$, the same $t_0$. The two frames tell different stories, slow clocks or a thin atmosphere, but they agree on the number of muons that arrive, as they must, since a count is a count. The fraction that survives the trip is $e^{−t_0/τ_0}$, the exponential decay law applied to the muon’s own time.
The lightning on the train
Simultaneity is the subtlest of the three effects, and the one that untangles the others. Lightning strikes both ends of a moving train, leaving scorch marks on the platform. A platform observer standing exactly midway between the marks sees both flashes at the same moment; since she is midway and light from each end traveled at $c$, she concludes the two strikes happened at the same time.
A passenger sits at the middle of the train. The train carries her toward the front strike’s light and away from the back’s, so the front flash reaches her first. In her own frame she is at rest midway between the ends, and both flashes traveled to her at $c$. So she concludes, correctly for her frame, that the front was struck first. Both observers correct for light’s travel time; it is not a matter of signal delays. Two events simultaneous in one frame are simply not simultaneous in another. For strikes at the ends of a train of rest length $L_0$, simultaneous on the platform, the train’s clocks put the front strike earlier by
$Δt′ = vL_0/c^2$
The same offset is why each frame finds the other’s clocks slow without contradiction: clocks along the train, synchronized in the train’s frame, are out of step by $vL_0/c^2$ in the platform’s frame, the clocks at the front lagging.
The twin paradox
One twin stays on Earth; the other flies to a distant star and back. Earth’s clocks say the trip took $2D/v$ for a star a distance $D$ away. The traveler’s clock is present at both the departure and the reunion, so it reads the proper time, which for a trip at constant speed is shorter by γ: she comes home younger.
The “paradox” is the objection that, from the traveler’s point of view, Earth moved away and came back, so Earth’s twin should be the younger. The situations are not symmetric. Earth’s twin stays in one inertial frame throughout; the traveler changes frames, from the outbound one to the inbound one, at the turnaround. On a spacetime diagram (the figure) Earth’s worldline is straight and the traveler’s is bent, and the bent one between the same two events has less proper time. The turnaround is treated here as instant, an idealized limit: a real turn takes time and bends the worldline over a stretch, which changes the ages a little, less the longer the cruise.
GPS clocks, and what every observer agrees on
The Global Positioning System is a working machine that depends on relativity. A receiver finds its position by timing signals from several satellites, each carrying an atomic clock; a timing error of one nanosecond is a ranging error of 30 cm. Two effects change a satellite clock’s rate compared with a clock on the ground. Its speed makes it run slow, by special relativity’s time dilation: each second it loses $1 − 1/γ$, about $v^2/2c^2$. Its height makes it run fast, because clocks run faster where gravity is weaker. That second effect belongs to general relativity, and this sim states it rather than deriving it: a clock at distance $r$ from Earth’s center gains $(GM/c^2)(1/R − 1/r)$ of each second on one at the surface, radius $R$.
For a low orbit the speed effect wins and the clock runs slow overall; for a high one, gravity wins. The two cancel for a circular orbit of radius $3R/2$, about 3,200 km up. GPS satellites orbit far higher, so their clocks gain tens of microseconds a day, enough to move positions by kilometers in a day if no one corrected them. The clocks are set before launch to run slow by just the right rate. The usual mistakes are the sign (thinking the satellite’s clock must run slow because it moves fast) and leaving one effect out.
That completes the picture. Times between events, lengths of objects and whether distant events are simultaneous all depend on the frame; the speed of light and the laws of physics do not.
Part 2 · Going further
Math you’ll need: algebra with square roots, hyperbolic functions (tanh and its inverse), and one integral, $τ = \int dt/γ$; spacetime diagrams are built from scratch.
The Lorentz transformation
Maxwell’s equations predict electromagnetic waves moving at $c = 1/\sqrt{μ_0ε_0}$ and name no frame. Under Galileo’s transformation between frames, $x′ = x − vt$ and $t′ = t$, a wave moving at $c$ in one frame moves at $c − v$ in the other, so the equations could hold in one frame only. The Michelson–Morley experiment found no sign of such a frame, and Einstein chose instead to keep the equations in every frame and change the transformation. Requiring that it be linear, that it reduce to Galileo’s at low speed, that light pulses in either direction, $x = ±ct$, move at $c$ in S′ too, and that S′ see S move at $−v$ by the same rule fixes it:
$t′ = γ(t − vx/c^2)$ $x′ = γ(x − vt)$
In matrix form, with $ct$ as the time coordinate, $\begin{pmatrix} ct′ \\ x′ \end{pmatrix} = \begin{pmatrix} γ & −γβ \\ −γβ & γ \end{pmatrix}\begin{pmatrix} ct \\ x \end{pmatrix}$; the inverse has $+γβ$. The term $vx/c^2$ is the relativity of simultaneity: events at the same $t$ but different $x$ have different $t′$. Time dilation and contraction both follow, and so does Part 1’s offset $vL_0/c^2$. The usual slips are the sign of $vx/c^2$ and a forgotten γ.
Spacetime diagrams
A spacetime diagram plots events with $ct$ up and $x$ across, at the same scale, so light moves along lines at 45°. A body at rest has a vertical worldline; one moving at speed $v$ has a straight worldline leaning from the time axis by $β = v/c$, always steeper than 45°. Light’s worldlines through an event form its light cone.
A moving frame S′ can be drawn on S’s diagram. Its time axis, $x′ = 0$, is its own origin’s worldline, $x = βct$. Its space axis, the events it calls simultaneous with the origin, is $t′ = 0$, which by the transformation is the line $ct = βx$: tilted up from the $x$ axis by as much as the $ct′$ axis is tilted from the $ct$ axis. The two primed axes close up symmetrically about the light line, which both frames share. Lines parallel to the $x′$ axis are S′’s lines of simultaneity, and they are why Part 1’s lightning strikes, simultaneous along a horizontal line in S, are not simultaneous in S′.
Reading a tilted axis needs care, because the page is not Euclidean for spacetime. The one-unit tick on the $ct′$ axis lies where the axis crosses the calibration hyperbola $(ct)^2 − x^2 = 1$, and every frame’s first tick lies on that same curve; the space ticks lie on $x^2 − (ct)^2 = 1$. On the page a moving frame’s ticks look farther apart, by $\sqrt{(1 + β^2)/(1 − β^2)}$, but each is one unit of that frame’s own time or length.
The interval and proper time along worldlines
Frames disagree about times and distances, but one combination of them every frame agrees on, the interval:
$(cΔτ)^2 = (cΔt)^2 − Δx^2$
Its sign sorts every pair of events. Timelike pairs ($(cΔt)^2 > Δx^2$) can be joined by a clock moving slower than light, and $Δτ$ is the proper time that clock reads; every frame agrees on their order, and one can cause the other. Lightlike pairs are joined by light. Spacelike pairs cannot be joined by any signal or any clock, so they have no proper time; for them $\sqrt{Δx^2 − (cΔt)^2}$ is the proper distance, the separation in the frame where they are simultaneous, and their order depends on the frame. The mistake to avoid is Euclid’s habit of adding the squares: spacetime subtracts them.
The interval between two timelike events gives the proper time of just one journey between them, the straight one. A clock’s own reading depends on its whole path: along any worldline, proper time adds up piece by piece: a clock moving at speed $v$ for a time $dt$ ages $dt/γ$, so $τ = \int dt/γ$, which for straight pieces is a sum of $Δt/γ$. Between two timelike events, the straight worldline, a clock moving inertially from one to the other, has the most proper time; any bent worldline between them has less. That reversed triangle inequality is the twin paradox in one line.
Velocity addition and rapidity
A ship moving at $v$ fires a probe forward at $u′$ relative to itself. Galileo would add the two; the Lorentz transformation gives
$u = (u′ + v)/(1 + u′v/c^2)$ $u′ = (u − v)/(1 − uv/c^2)$
the second form being the speed of one body in another’s frame. Both stay below $c$ whenever $u′$ and $v$ do. The cleanest way to see why is the rapidity $φ = \operatorname{artanh} β$: for boosts along one line, rapidities add, $β = \tanh(φ_1 + φ_2 + …)$, and since tanh never reaches 1, no number of boosts reaches $c$. At low speed $φ ≈ β$, which is why Galileo’s sum works there.
The Doppler shift, and the twin paradox three ways
A source approaching at speed $v$, sending signals at frequency $f_0$ by its own clock, is received at
$f/f_0 = \sqrt{(1 + β)/(1 − β)}$
and receding, at the inverse. Two effects combine: each signal has a shorter trip than the last when the source approaches, and the source’s clock runs slow by γ. Sound needs two formulas, for a moving source and a moving listener, because air is a medium; light has none, so only relative motion matters.
The twin paradox can now be told three ways. By worldlines: the traveler’s bent worldline has less proper time. By frames: on each leg each twin finds the other’s clock slow, but at the turnaround the traveler’s line of simultaneity swings, and Earth’s “now” for her jumps ahead by years. By signals: each twin sends a light signal on each of her birthdays. The traveler receives Earth’s at the receding rate for her whole outbound leg and at the approaching rate for her whole return; Earth receives hers slowly until it sees her turn, which happens only when the light from the turnaround arrives, long after the turn itself. Counting signals, each twin finds the other’s age correctly, $N = Δτ\sqrt{(1 ∓ β)/(1 ± β)}$ per leg; the asymmetry is when each sees the switch.
Muon survival, counted in both frames
The fraction of muons that survives a trip of length $h$ at speed $v$ is $e^{−t/(γτ_0)}$ in Earth’s frame, where the lifetime is dilated to $γτ_0$, and $e^{−t_0/τ_0}$ in the muon’s, where the trip is short. The exponents are equal, $t/(γτ_0) = h/(γvτ_0)$, so the count is frame independent, as any count of events must be. The product $γvτ_0$ is the mean decay length in Earth’s frame. A muon’s energy gives its γ directly, $γ = E/mc^2$, which is how experiments know it.
In the classic version of the experiment (Frisch and Smith, 1963), a detector on a mountain counted muons of a selected speed, and the same detector then counted them at sea level; far more arrived than a non-relativistic lifetime allows, in close agreement with time dilation. Real muons also lose energy in the air, roughly 0.25 GeV per kilometer, which such an experiment must account for and this sim’s cards set aside.
Four-momentum
Energy and momentum combine into a four-vector, $(E/c, p)$, which transforms between frames exactly as $(ct, x)$ does. Its invariant length is the rest energy:
$E^2 = (pc)^2 + (mc^2)^2$ $E = γmc^2$
so that $pc = mc^2\sqrt{γ^2 − 1}$ and $β = pc/E$. At rest, $E = mc^2$; at high γ, $pc$ approaches $E$; for light, $m = 0$ and $E = pc$ exactly. The slip to avoid is taking the kinetic energy, $E − mc^2$, for $pc$. This is the sim’s one step into relativistic dynamics.
The GPS budget
For a satellite clock in a circular orbit of radius $r$, with speed $v = \sqrt{GM/r}$, the two terms per second are $−(1 − 1/γ)$ and $(GM/c^2)(1/R − 1/r)$; times 86,400 they give the offset per day. At orbital speeds $β$ is about $10^{−5}$, and computing $1 − 1/γ$ as written loses most of its digits to rounding; the sim uses the exact rearrangement $1 − 1/γ = β^2/(1 + \sqrt{1 − β^2})$ (and $γ − 1 = β^2/(s(1 + s))$, $s = \sqrt{1 − β^2}$), which is $β^2/2$ to first order and exact to rounding at any speed. The gravitational term is the weak-field result of general relativity, stated, not derived. The two cancel where $GM/(2r) = GM(1/R − 1/r)$, that is, at $r = 3R/2$.
The sim’s model puts the ground clock at rest on a spherical Earth at its mean radius; Earth’s turning and its real, slightly flattened gravity change the net by less than a tenth of a microsecond a day, and real orbits’ small eccentricity adds a periodic term that receivers correct separately.
Reference
Common mistakes
Part 1
- Adding light’s speed to its source’s: every inertial observer measures $c$.
- Thinking dilation only looks like it happens because of light’s travel time: frames measure it with clocks where events happen, and muons show it.
- Taking one frame as the real one: the proper time belongs to the clock present at both events, and dilation is mutual.
- Multiplying a length by γ, or contracting heights: $L = L_0/γ$, along the motion only.
- Explaining simultaneity away as signal delay: both observers correct for travel time and still disagree.
- Calling the twins symmetric, or blaming the traveler’s acceleration: her worldline is the bent one.
- Getting GPS’s sign wrong, or leaving one effect out: speed slows the satellite’s clock, height speeds it up.
Part 2
- Adding the squares in the interval: spacetime subtracts them.
- Reading a tilted axis like a square one: its ticks sit on the calibration hyperbolas.
- Using Galileo’s sum near $c$, or taking the Doppler factor for γ alone.
The equation sheet
| Part 1 | |
|---|---|
| light’s speed | $c$ = 299,792,458 m/s for every inertial observer |
| the Lorentz factor | $β = v/c$; $γ = 1/\sqrt{1 − β^2}$ |
| a light clock | $Δt_0 = 2L/c$ |
| time dilation | $Δt = γΔt_0$ |
| contraction | $L = L_0/γ$ |
| a muon’s range | $d = vγτ_0$; survivors $e^{−t_0/τ_0}$ |
| simultaneity | $Δt′ = vL_0/c^2$ |
| the twins | traveler’s time $Δt_{trip}/γ$ |
| GPS | range error $= cΔt$ |
| Part 2 | |
|---|---|
| Lorentz | $t′ = γ(t − vx/c^2)$; $x′ = γ(x − vt)$ |
| the interval | $(cΔτ)^2 = (cΔt)^2 − Δx^2$ |
| along a worldline | $τ = \int dt/γ$ |
| velocities | $u′ = (u − v)/(1 − uv/c^2)$ |
| rapidity | $β = \tanh(φ_1 + φ_2 + …)$, $φ = \operatorname{artanh} β$ |
| Doppler | $f/f_0 = \sqrt{(1 + β)/(1 − β)}$ approaching |
| signals | signals seen $= Δτ\sqrt{(1 − β)/(1 + β)}$ (receding) |
| four-momentum | $E^2 = (pc)^2 + (mc^2)^2$; $E = γmc^2$ |
| GPS | $−(1 − 1/γ) + (GM/c^2)(1/R − 1/r)$ per second |
Symbols: $c$ light’s speed; $v$ a speed and $β = v/c$; $γ$ the Lorentz factor; $Δt_0$ a proper time (Part 2’s $τ$; some textbooks write it $Δτ$); $L_0$ a proper length; $τ_0$ the muon’s mean lifetime at rest, 2.20 µs; $t$, $x$ an event’s coordinates in S and $t′$, $x′$ in S′; $φ$ a rapidity; $f$, $f_0$ frequencies; $D$ a star’s distance; $E$, $p$, $m$ energy, momentum and mass; $G$, $M$, $R$ Newton’s constant and Earth’s mass and mean radius; $r$ an orbit’s radius. On spacetime diagrams $c$ = 1 ly/yr.
Sources and further reading
- OpenStax, University Physics Volume 3, the chapter “Relativity,” CC BY-NC-SA 4.0 (openstax.org): the tracks follow its topics and notation.
- E. F. Taylor and J. A. Wheeler, Spacetime Physics, for spacetime diagrams, the interval and the twin by signals.
- J. B. Hartle, Gravity: An Introduction to Einstein’s General Relativity, for the GPS budget and its gravitational term.
- D. H. Frisch and J. H. Smith, “Measurement of the Relativistic Time Dilation Using μ-Mesons,” American Journal of Physics 31, 342 (1963).
- This sim’s Field Manual (H) for the model and its limits; Frame Shift for Galilean frames; Gravitation & Satellite Orbits for the GPS orbit.
Special Relativity & Spacetime v1.1 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org