Length
Contraction

Shorter, but only one way

An object moving past you at speed v is measured shorter along its direction of motion, by the same Lorentz factor γ that slows its clocks:

L = L₀ / γ = L₀ · √(1 − v²/c²)

L₀ is the proper length, measured by someone travelling with the object. Height and width across the motion don’t change at all. At 0.8c the 100-metre ship measures exactly 60 metres, still 24 metres tall. Its round portholes become ovals and its ring flattens, because everything along the direction of motion shrinks, including the people aboard, their rulers and the space between their atoms. Nobody on board notices anything.

Proxima Centauri is 4.25 light-years away in Earth’s frame. Trip times ignore speeding up and slowing down.

Climb aboard: the universe shrinks

Tap Ride along. Now the ship is at rest and its full 100 metres long. To the crew it’s the buoys, the ruler and the whole route to Proxima Centauri that rush backwards, and all of it is contracted. At 0.99c the 4.25-light-year gap to Proxima is only 0.60 light-years in the ship’s frame. Covering that at 0.99c takes 0.61 years, exactly the time the ship’s clock records according to Earth’s calculation using time dilation.

That’s the point: length contraction is reciprocal. Each observer measures the other’s lengths shortened along the line of motion, just as each measures the other’s clocks running slow. Neither is “really” shorter. Length is a measurement, and it depends on the frame you make it in.

Why lengths must shrink

Measuring the length of something moving means marking where its nose and tail are at the same moment. But observers in relative motion disagree about which distant events happen at the same moment (the relativity of simultaneity you met in the Twin Paradox). Different “same moments” give different lengths.

You can get the exact amount from time dilation alone. Stand beside a buoy and time the ship as it passes. Your single clock reads Δt = L / v. The crew see the buoy slide past from nose to tail, timed by two synchronised clocks on board, so they get Δt′ = L₀ / v. Your clock is the one present at both events, so from the crew’s point of view it runs slow: Δt = Δt′ / γ. Put those together:

L / v = (L₀ / v) / γ  ⇒  L = L₀ / γ

Time dilation, length contraction and the relativity of simultaneity aren’t three separate effects. They’re three views of one fact: the speed of light is the same for everyone.

Both views are right

At 0.866c (γ = 2) the barn’s frame says the pole is 8 m long and fits inside the 10 m barn with room to spare. Both doors shut at the same instant with the whole pole inside.

The pole’s frame says the barn is only 5 m long and the pole is 16 m. It can never be inside all at once. How can both be true? The two door flashes happen 10 m apart, and events that are simultaneous in one frame aren’t simultaneous in another. In the pole’s frame the back door shuts first, while the pole’s front end hasn’t reached it yet, and reopens before it arrives. The front door shuts about 57.8 nanoseconds later, once the pole’s rear end is already inside. Neither door ever touches the pole, in either frame.

Δt′ = γ · v · L_barn / c² ≈ 57.8 ns

Whether a door hits the pole is a physical event, so every frame must agree on it, and they do. Slow the pole below about 0.781c and it no longer fits even in the barn’s frame. Then the doors strike it, and you’ll see that in both views. What frames disagree on is “at the same time”, and with it, length.

Try it: watch the timeline under the barn. In the barn’s view the two flashes sit on top of each other. Switch to the pole’s view and they split apart.

Measuring isn’t seeing

Everything above is about what’s measured: where the nose and tail are at one instant in your frame. A camera records something else. It catches light that arrives at the same moment, and light from the far end of the ship set off earlier than light from the near end.

Press What a camera sees to see this for a camera 60 metres from the flight path. Each slice of the ship is drawn where it was when its light left. An approaching ship looks stretched, far longer than 100 m, and a receding one looks squashed. The dashed outline shows where the ship actually is at that instant.

In 1959 James Terrell and Roger Penrose showed that for a 3D object the full picture is even stranger. Light from the ship’s rear face, normally hidden, also reaches the camera, and the combination looks like the object has rotated rather than shrunk. A fast sphere still looks round. In 2025 a team in Vienna finally photographed the effect, using picosecond laser pulses and an ultrafast camera to mimic a speed of light under 2 metres per second. (Our camera view draws the ship’s side exactly but leaves out that rear face and the Doppler colour shifts.)

Real evidence

• Muons, from the muon’s side. Cosmic-ray muons are typically created about 15 km up and live just 2.2 microseconds at rest. From Earth, a muon at 0.9995c takes about 50 μs to reach the ground, and time dilation stretches its lifetime to about 69 μs, so many survive. From the muon’s point of view, its clock is normal, but the atmosphere rushing up at it is contracted by γ ≈ 32 to about 474 metres, which the ground crosses in about 1.6 μs. Two frames, two stories, one result: the muon reaches the ground.
• Pancake nuclei. At Brookhaven’s RHIC, gold nuclei collide at 100 GeV per nucleon, so γ ≈ 107. At CERN’s LHC, lead nuclei reach about 2.7 TeV per nucleon, so γ is nearly 2,900. A nucleus about 14 femtometres across is flattened in the lab frame into a disc roughly 0.13 fm thick at RHIC and about 0.005 fm at the LHC. Physicists model these collisions as two thin pancakes passing through each other, and that picture is built into how the measured particle sprays are understood.
• The same factor everywhere. Length contraction uses exactly the γ confirmed by time-dilation experiments, from storage-ring muons to atomic clocks. The theory doesn’t allow one without the other, and particle accelerators are designed around both every day.