Special relativity says a moving clock runs slow, so speed alone should have made the airborne clocks fall behind.
General relativity says a clock higher up, in slightly weaker gravity, runs fast, so altitude should have pushed them ahead.
So the eastward clock moved faster overall and lost time, while the westward clock moved slower and gained it.
The westward clock ran the other way, and the tidy slogan that moving clocks run slow does not, by itself, explain the experiment meant to prove it.
Their speed slows their clocks by roughly 7 microseconds a day; their altitude, in weaker gravity, speeds them up by roughly 45.
The first machine to carry a human-made clock measurably into the future was a scheduled passenger flight. In October 1971, a physicist and an astronomer bought airline tickets for themselves and four caesium-beam atomic clocks, buckled the clocks into passenger seats, and flew twice around the world. When they compared those clocks against identical ones kept at the United States Naval Observatory, the travelling clocks disagreed with the stationary ones by fractions of a millionth of a second. The disagreement was not measurement error. It was time dilation, and it landed close to what Einstein’s relativity had predicted.
Named for the physicist Joseph Hafele and the astronomer Richard Keating, the experiment appeared in the journal Science in 1972, across two short papers: one setting out the predicted time gains, one reporting the measured ones. Relativity had been tested before, and more precisely. What Hafele and Keating did was test it with off-the-shelf clocks on ordinary flights, at a scale a person could hold in their head.
What the flying clocks actually showed
Two effects were at work, pulling in opposite directions. Special relativity says a moving clock runs slow, so speed alone should have made the airborne clocks fall behind. General relativity says a clock higher up, in slightly weaker gravity, runs fast, so altitude should have pushed them ahead. On a jet at cruising height, both effects are small and roughly comparable in size, which is what made the experiment delicate.
Results split by direction. For the eastward trip, theory predicted the flying clocks would lose about 40 nanoseconds (±23); they lost 59 (±10). Westward, the prediction was a gain of roughly 275 nanoseconds (±21), and the clocks delivered 273 (±7).
Those error bars matter: the eastward figure is a looser fit, the westward one a very tight one.
That asymmetry has a specific cause. It comes from the fact that the relevant speed is not the plane’s speed over the ground but its speed relative to the centre of a non-rotating Earth. The planet is already turning eastward at roughly 1,600 kilometres per hour at the equator. Fly east and you add to that; fly west and you subtract from it. So the eastward clock moved faster overall and lost time, while the westward clock moved slower and gained it.
Why “time travel” is the right phrase at the wrong scale
The popular framing needs handling on one point. “Time travel into the future” fits the eastward result well enough: that clock came home having ticked less than the one left behind, so from its point of view slightly less time had passed, and it arrived a sliver into the ground clock’s future. The westward clock ran the other way, and the tidy slogan that moving clocks run slow does not, by itself, explain the experiment meant to prove it.
Scale is the other correction. This is forward-only, and it is measured in billionths of a second. Nothing here lets anything travel backward, and nothing about it resembles the version in films. The effect grows only with speed, and the speeds that would make it dramatic are ones no crewed vehicle has come close to.
The same effect, running in your phone
A more persuasive demonstration is not the 1971 flight at all. It is the satellite navigation system most people carry. GPS satellites sit at about 20,200 kilometres up and move fast, so both relativistic effects apply. Their speed slows their clocks by roughly 7 microseconds a day; their altitude, in weaker gravity, speeds them up by roughly 45. The gravitational effect wins, and the net result is that each satellite clock runs about 38 microseconds a day faster than a clock on the ground.
That is not trivial in a system that fixes position by timing. As the National Institute of Standards and Technology lays out in its account of relativity tests, the correction is built into the satellites. Left uncorrected, the timing drift would translate into position errors growing by something like ten kilometres a day. The system works because the relativity was treated as a number to subtract, and the number was right.
The furthest-travelled humans
Sergei Krikalev, a Russian cosmonaut, is the name usually attached to human time travel. Over six missions and 803 days in orbit, the velocity of his flights left him something like 0.02 seconds younger than he would otherwise have been, a figure worked through by Universe Today. Unlike the GPS satellites, though, the International Space Station orbits low enough that the speed effect outweighs the altitude one, so its crews age fractionally slower than people on the ground rather than faster.
The record has since moved. Oleg Kononenko passed Gennady Padalka’s mark in February 2024 and, by the time he returned in September that year, had accumulated 1,111 days in space across five missions, the longest cumulative total on record. His head start on the rest of us is a little larger than Krikalev’s.
It is still a fraction of a second.
The finding is real, repeatable, and quietly load-bearing for infrastructure people use every day. It is also modest by design. The only way to turn a nanosecond into a noticeable jump forward is speed we cannot yet reach, which leaves the aeroplane, the satellite, and the cosmonaut as the honest limit of what “time travel” currently means.
