Throw a ball forward from a moving train, and a platform observer also accounts for the train’s motion. We might expect light from a moving torch to gain speed in the same way. In vacuum, that expectation fails.
Vacuum light speed c is special because relativity gives it the same value for all inertial observers and makes it the local limit for causal influence, including new information. This is more than the record speed of a lamp: it concerns the relationship between space, time and possible signals.
The source moves, but vacuum c does not change
An inertial observer uses an unaccelerated reference frame, in which a free body moves uniformly along a straight line. Imagine two laboratories passing each other. Each measures a light pulse’s distance and time using its own rulers and clocks. For vacuum light, both obtain c, rather than c plus the other laboratory’s speed. OpenStax: special relativity’s postulates.
This works because distances, time intervals and simultaneity between distant events do not transform between frames according to everyday rules. We cannot retain ordinary velocity addition and simply declare light an exception. All measurements need a consistent rule.
c is exactly 299,792,458 m/s. Its exact numerical value in SI is tied to the metre’s definition: length is defined through light and the second. This is a convention about units; invariance of c in inertial measurements is a physical statement tested by experiments. BIPM: definition of the metre.
Branch · calculationOptional · you can skip this
What happens to two speeds of 0.8c?
In a teaching example, a laboratory moves at speed v relative to Earth, while a craft inside moves at speed u relative to the laboratory. Both motions follow the same straight line in the same direction. Its speed w in Earth’s inertial frame is:
w = (u + v)/(1 + uv/(c)^(2)).
With u = v = 0.8c, the numerator is 1.6c, but the denominator is 1.64. The result is approximately 0.976c, rather than 1.6c. These are hypothetical speeds for checking a formula, not the capability of an existing spacecraft.
Setting u = c gives w = c for any allowed ∣v∣ < c. At low speeds the denominator is almost one, and ordinary addition is a good approximation. The term uv/(c)^(2) is dimensionless, leaving speed units for w. OpenStax: relativistic velocity transformation.
The limit concerns what can influence what
An event happens at a particular place and time: a transmitter switches on, or a receiver responds. Whether the first event could cause the second depends on whether a signal has enough time to cross their separation.
Suppose a transmitter sends a new command to a receiver one light-second away. A light-second is the distance vacuum light travels in one second. The command cannot arrive sooner than one second later in this frame. A slower carrier or equipment delay adds time; a better engine cannot remove the lower bound.
In the diagram, the horizontal axis is one spatial coordinate and time increases upward. Light lines bound the future light cone, the region a signal from the start can reach. With only one spatial coordinate shown, we see its triangular cross-section. Max Planck: spacetime and causal structure.
Can a signal from event O arrive in time?
Time t increases upward; position x increases to the right.
One horizontal unit is one light-second; one vertical unit is one second. At these scales vacuum light lines lie at 45°.
O: x = 0, t = 0 — transmission.
- A · x = 0.5 light-seconds; t = 2 s
Inside: required mean speed 0.25c. A causal connection is allowed.
- B · x = 2 light-seconds; t = 2 s
On the boundary: only transmission at c arrives in time.
- C · x = 3 light-seconds; t = 2 s
Outside: 1.5c would be needed, so a new command from O cannot arrive in time.
Inside the cone, a slower-than-light connection is allowed; on its boundary, a light-speed connection is possible. Outside it, no allowed local transmission from the start arrives in time. “Could influence” does not mean “actually caused”: the transmitter might have sent nothing.
Why not simply accelerate harder?
A body with nonzero rest mass requires ever more energy as it approaches c; finite energy cannot bring it to c. This follows from the relativistic relationship between energy and speed, rather than friction in space. A calculation and its practical limits appear in the article on travelling to other stars.
“In vacuum” matters. Light propagation through glass or another medium depends on interaction with matter. An ordinary optical signal in glass travels slower than c. This does not lower the fundamental limit to the light speed of that particular glass or make it a new standard of causality.
Branch · a wave subtletyOptional · you can skip this
Phase, pulse peak and spot describe different speeds
Phase velocity describes motion of a surface of constant phase, such as a crest of an ideal periodic wave. Under some conditions it can exceed c. An endlessly repeating wave, however, does not by itself mark the arrival of a new command. Rüdiger Paschotta: phase velocity.
Group velocity usually describes the motion of a wave packet’s envelope. In a medium that strongly reshapes a pulse, the motion of its peak cannot automatically be equated with new information. Duke researchers explain “fast light” without a faster-than-light message.
Another example is a turning beam illuminating distant surface patches in succession. The illuminated location moves rapidly; one object or message does not travel from the first patch to the next. Photons reach each patch from the source with their own delay. A number “greater than c” therefore does not establish superluminal communication.
What to check in a claim of faster-than-c motion
In general relativity the limit is local: in a comparison near one place, no causal signal overtakes vacuum light. Changing separation between very distant galaxies in an expanding Universe is defined differently, so its rate can exceed c without such local overtaking. Max Planck: the local limit, Davis and Lineweaver: cosmological recession.
Test this claim: a receiver 3 light-seconds away receives a new command 2 seconds after transmission. In flat spacetime with stationary transmitter and receiver and synchronised clocks, that causal connection is impossible. Light needs 3 seconds. A preprogrammed coincidence of signals would not transmit a new command.
c is special as the shared boundary of possible influence, rather than a ban on every motion of a drawn point. When you encounter a superluminal number, establish what moves, how time and distance were measured, and whether it carried new information between events.