The Spacetime Interval: Timelike, Lightlike, and Spacelike Separation

Two observers watch the same pair of events. They agree that the events happened, but they may assign them different distances and different elapsed times.

That can sound like a problem. If one observer measures one distance and another measures a different distance, how can both still agree that the events are timelike-, lightlike-, or spacelike-separated?

The answer is that relativity does not preserve space and time as separate quantities. It preserves a particular combination of them: the spacetime interval.

If the geometric picture is unfamiliar, Light Cones: What Can Affect What in Spacetime? provides the useful starting point. A light cone places events inside, on, or outside its boundary. The spacetime interval is the calculation that makes the same classification without relying on a diagram.

Space and Time Change Together

Consider two events, \(A\) and \(B\). In one inertial frame, their coordinate differences are

$$ \Delta t = t_B-t_A $$

and

$$ \Delta x = x_B-x_A. $$

Another observer moving at a steady velocity relative to the first may obtain different values for both \(\Delta t\) and \(\Delta x\). Neither value, taken alone, is shared by every inertial observer.

But the changes are not arbitrary. When an observer's motion changes the measured distance between the events, it changes the measured elapsed time in a coordinated way. The spacetime interval captures that coordination.

In one spatial dimension, using the sign convention adopted on this page, the interval is

$$ \Delta s^2=c^2\Delta t^2-\Delta x^2. $$

Here, \(c\) is the speed of light. Multiplying \(\Delta t\) by \(c\) converts the time difference into a distance, so both terms are measured in the same units before they are compared.

The square on \(\Delta s^2\) is part of the notation. The interval itself can be positive, zero, or negative. That sign is what classifies the relationship between the two events.

What the Sign Tells Us

With the convention

$$ \Delta s^2=c^2\Delta t^2-\Delta x^2, $$

the three possibilities are:

The formula is therefore an exact version of a familiar comparison. The positive term is the square of the distance light could travel during the elapsed time. The subtracted term is the square of the events’ spatial separation. The sign tells us which squared distance is larger.

In three spatial dimensions, the idea does not change. The full spatial separation simply includes all three directions:

$$ \Delta s^2=c^2\Delta t^2-\Delta x^2-\Delta y^2-\Delta z^2. $$

The same sign rules apply under the convention used here.

Why Every Inertial Observer Agrees

Changing inertial frames changes the coordinates assigned to the events. A Lorentz transformation mixes the measured space and time differences: an observer who obtains a new \(\Delta x\) also obtains a corresponding new \(\Delta t\).

What the transformation preserves is

$$ c^2\Delta t^2-\Delta x^2-\Delta y^2-\Delta z^2. $$

This is Lorentz invariance. Every inertial observer may divide the separation differently between space and time, but all calculate the same value of \(\Delta s^2\).

That is why the classification cannot change with the observer. A positive interval cannot become negative merely because someone moves into another inertial frame. Timelike remains timelike, lightlike remains lightlike, and spacelike remains spacelike.

The interval does not erase disagreement between observers. It identifies what survives that disagreement.

One Station, Three Possible Alarms

Suppose that, in one inertial frame, an instrument emits a signal and a station is 900,000 kilometers away. Use

$$ c\approx 300{,}000\ \text{km/s}. $$

In this frame, the station remains at the same position while three possible alarm events occur along its worldline. The signal's lightlike boundary reaches the station at exactly three seconds, placing an earlier alarm outside the cone and a later alarm inside it.

Figure 1. The future light cone from the emission reaches the station after three seconds. The three possible alarm events show how negative, zero, and positive intervals map onto positions outside, on, and inside the cone.

Start with the alarm at two seconds. During that time, light could travel only 600,000 kilometers. The interval is

$$ \begin{aligned} \Delta s^2 &=(300{,}000\ \text{km/s})^2(2\ \text{s})^2-(900{,}000\ \text{km})^2\ &=(600{,}000\ \text{km})^2-(900{,}000\ \text{km})^2\ &=-4.5\times10^{11}\ \text{km}^2. \end{aligned} $$

The result is negative, so the two events are spacelike-separated. This reproduces the diagram's conclusion: the alarm lies outside the emission event's future light cone, and that particular signal could not have caused it.

At three seconds, light can travel exactly 900,000 kilometers:

$$ \Delta s^2=(900{,}000\ \text{km})^2-(900{,}000\ \text{km})^2=0. $$

That separation is lightlike. At four seconds, the time term becomes larger than the distance term, so the interval is positive and the separation is timelike.

The arithmetic and the diagram perform the same test. They ask whether the available time is too short, exactly sufficient, or more than sufficient for light to cross the distance.

A Note About the Opposite Sign Convention

Some books define the interval with the opposite overall sign:

$$ \Delta s^2=\Delta x^2+\Delta y^2+\Delta z^2-c^2\Delta t^2. $$

With that convention, timelike intervals are negative and spacelike intervals are positive. Lightlike intervals are still zero.

Nothing physical has changed. Multiplying the entire expression by \(-1\) reverses the written sign but preserves the classification. This is why a statement such as "timelike means positive" is incomplete unless the author has specified the convention being used.

The Algebra and the Cone Say the Same Thing

The light cone gives the geometry. The spacetime interval gives the frame-independent test beneath it.

• A positive interval under the convention used here places the second event inside the cone.
• A zero interval places it on the lightlike boundary.
• A negative interval places it outside the cone.

Observers can disagree about the distance and elapsed time between the events because those quantities depend on the frame. They cannot disagree about whether the separation is timelike, lightlike, or spacelike because the combination \(\Delta s^2\) does not.

That is the interval's central gift: it separates what changes with the observer from the causal structure that does not. The next question is what this leaves open for spacelike-separated events—developed in Can Two Observers Disagree About Which Event Happened First?—and why attempts to connect such events faster than light create trouble for causality.