Special Topics in 1st-Year Physics

2 Faster than Light = Back in Time

This topic is meant to provide an introduction to some of the ideas of Special Relativity, and
in particular to argue that the proscription against moving faster than light is equivalent to
the forbidding of sending signals to oneself in the past.

2.1 Postulates of Relativity

The starting point for Einstein’s reformulation of Newtonian mechanics is a postulate that
Newton also uses: that there exist a class of inertial observers for whom particles move in
straight lines with constant speed in the absence of any external forces. The rest of Newton’s
laws (or Einstein’s reformulation of them, at least in special relativity) are then aimed at
the measurements made by these special observers.
The main two principles from which special relativity follows are given by the following
additional assertions.
• Principle of Relativity: This principle (also shared by Newton) states that laws
of physics as seen by any two inertial observers moving at constant velocity relative
to one another are identical. That is, the transformation from one such observer to
another is a symmetry of the equations of motion.
• Universality of the speed of light: This principle states that every inertial observer
measures precisely the same numerical value of the speed of light, c = 299, 792, 458
m/s, regardless of the speed of their motion relative to one another.
It is the second of these that separates Einstein from Newton. It is very counter-intuitive,
because our Newtonian intuition says that if observer A measures the speed of a light ray as
being cA , and if observer B moves in the same direction as the light ray with speed (relative
to A) of vBA , then observer B should measure the light ray’s speed as being cB with
cA = cB + vBA . (2.1)
In particular, the ray should appear to be motionless if observer B moves at the speed of
light relative to observer A. (That is, cB = 0 if vBA = cA .)
Furthermore, the addition law, (2.1), seems to be a direct consequence of the addition law
for vectors. That is, suppose we measure positions relative to some fixed origin of coordinates
at rest relative to some observer A. Further suppose another observer B has position rBA
relative to this origin of coordinates. The position of a particular point on the light wave-
front (or of a photon in the light beam) relative to the same origin of coordinates is similarly
denoted r`A , while the position of the same photon as measured by B is instead called r`B .
As shown in the figure, these vectors are related by the obvious addition law:
r`A = r`B + rBA . (2.2)
The Newtonian law for addition of velocities — including (2.1) — then follows by differ-
entiating with respect to t:
dr`A dr`B drBA
v`A = = + = v`B + vBA . (2.3)
dt dt dt
How can an argument as simple as this be wrong?

Page 1 of 8
Special Topics in 1st-Year Physics

Figure 1: The position of a particle, ‘`,’ as seen by two observers, A and B, and their relation
to the position of B relative to A.

2.2 Lorentz transformations

The fallacy in the above derivation of the addition law for velocities turns to be the hidden
assumption that all observers experience the same notion of time. Einstein argued that this
must change in precisely the way required to keep the speed of light the same for every
observer.
Why would he want to do that? Ultimately he did so because of experiments: people had
sought differences in the speed of light as a function of the different velocities experienced by
an observer on the Earth’s surface due to the Earth’s rotation or its orbit around the Sun. In
all cases no difference was found, and a famous experiment by Michelson and Morley using
an interferometer was particularly convincing on this point.
How would time have to change from observer to observer in order to conspire so that
all inertial observers measure precisely the same speed of light? To answer this consider
motion in the x direction, and suppose a light ray is measured by observer A to travel a
distance ∆x = x2 − x1 in a time interval ∆t = t2 − t1 . The speed measured by A is therefore
cA = ∆x/∆t.
How should this look to observer B? The coordinates (x0 , t0 ) observer B assigns to observer
A’s coordinates (x, t) must be some linear1 combination of those measured by A:

t0 = k1 t + k2 x and x0 = k3 t + k4 x , (2.4)

or  0   
t k1 k2 t
0 = , (2.5)
x k3 k4 x
for some real coefficients k1 , k2 , k3 and k4 . In Newton’s case the coefficients would be

k1 = 1 , k2 = 0 , k3 = v and k4 = 1 , (2.6)

where v is the relative speed of observers A and B.

1
The transformation must be linear in order that it preserve vector sums. That is, in our example with
one spatial dimension if x3 = x1 + x2 then we also want x03 = x01 + x02 .

Page 2 of 8
Special Topics in 1st-Year Physics

So observer B’s version of the measurement of the speed of light would be that the light
ray travels a distance ∆x0 in time ∆t0 and so gets cB = ∆x0 /∆t0 . Given (2.4) and demanding
cA = cB = c this implies

0 = (∆x0 )2 − c2 (∆t0 )2 = (k3 ∆t + k4 ∆x)2 − c2 (k1 ∆t + k2 ∆x)2

= (k42 − c2 k22 )(∆x)2 + (k32 − c2 k12 )(∆t)2 + 2(k3 k4 − c2 k1 k2 )∆x∆t . (2.7)

We determine k1 through k4 by demanding the right-hand side be equal to (∆x)2 − c2 (∆t)2

and so
k42 − c2 k22 = 1 , k32 − c2 k12 = −c2 and k3 k4 = c2 k1 k2 . (2.8)
This is a system of 3 equations in 4 unknowns, so we expect a one-parameter family of
solutions. The solutions can be written in terms of hyperbolic trig functions:
1 1
cosh β = (eβ + e−β ) and sinh β = (eβ − e−β ) , (2.9)
2 2
which satisfy the identity
cosh2 β − sinh2 β = 1 . (2.10)
In terms of this the solutions to (2.8) are
k3
k4 = k1 = cosh β , and = c k2 = − sinh β , (2.11)
c
where β is a free parameter (that is next related to the relative speed, v, of the two observers).
To relate β to v we use the transformation just identified:
 0   
ct cosh β − sinh β ct
= . (2.12)
x0 − sinh β cosh β x
Specialize this to the trajectory followed by the origin x0 = 0 of observer B, leading to

0 = x0 = −c t sinh β + x cosh β , (2.13)

and since the relative speed of observer B as seen by A is given by v = x/t in the above
expression, the relation between β and v is
sinh β v
tanh β = = . (2.14)
cosh β c
Dividing (2.10) by cosh2 β implies
1 1
cosh2 β = 2 =
1 − tanh β 1 − v 2 /c2
v 2 /c2
and so sinh2 β = cosh2 β − 1 = , (2.15)
1 − v 2 /c2
and so (2.12) becomes
 0   
ct 1 −v/c ct 1
0 =γ with γ=p . (2.16)
x −v/c 1 x 1 − v 2 /c2

Page 3 of 8
Special Topics in 1st-Year Physics

Figure 2: The time-dilation formula relating two observers moving with relative speed v.

The result is the usual Lorentz transformation rule

t − xv/c2 x − vt
t0 = p and x0 = p . (2.17)
1 − v 2 /c2 1 − v 2 /c2

In particular, t measures time along the trajectory of observer A (situated at x = 0). The
above equations show that this is related to the time, t0 , measured by observer B by the
usual time-dilation formula:
t
t0 = p , (2.18)
1 − v 2 /c2
as plotted in the figure. Since t0 > t moving clocks run more slowly. This is not because of
some fault of the clocks. It is because time itself runs more slowly for a moving observer.

Relativistic law for the addition of velocities

We can now derive Einstein’s addition law for velocities which replaces the newtonian law
(2.1). To do so we perform two successive Lorentz transformations with different speeds, v1
and v2 (and so different tanh β1 = v1 /c and tanh β2 = v2 /c):
 0       
ct cosh β1 − sinh β1 cosh β2 − sinh β2 ct cosh β12 − sinh β12 ct
0 = = ,
x − sinh β1 cosh β1 − sinh β2 cosh β2 x − sinh β12 cosh β12 x

where

cosh β12 = cosh β1 cosh β2 + sinh β1 sinh β2 = cosh(β1 + β2 )

sinh β12 = cosh β1 sinh β2 + sinh β1 cosh β2 = sinh(β1 + β2 ) , (2.19)

which uses the addition theorem for hyperbolic trig functions that follow from their defini-
tions. Taking the ratio of these expressions and using (2.14) gives

c (tanh β2 + tanh β1 ) v1 + v2
v12 = c tanh β12 = = . (2.20)
1 + tanh β1 tanh β2 1 + v1 v2 /c2

Page 4 of 8
Special Topics in 1st-Year Physics

This states the relativistic rule for the addition of velocities (in one direction): if B moves
relative to A with speed v1 and C moves in the same direction relative to B with speed v2
then C moves relative to A with speed v12 , given in terms of v1 and v2 by (2.20). Notice
that if v2 = c or v1 = c then v12 = c, regardless of the other speed. Whenever both v1 and
v2 are smaller than c then so is v12 . Finally notice that Taylor expanding the denominator
of (2.20) in powers of v1 v2 /c2 shows that it reduces to the Newtonian expression (2.1) (as it
must) when v1 and v2 are both much smaller than c:
h v1 v2 i
v12 = (v1 + v2 ) 1 − 2 + · · · . (2.21)
c

2.3 Relativity of simultaneity and the requirement that v ≤ c

The Lorentz transformations have a very important implication: they imply that different
observers disagree on which events in space and time are simultaneous. To see this notice that
observer A above regards events with different x but the same value of t to be simultaneous,
while observer B regards events with different x0 sharing the same t0 as being simultaneous.
To illustrate this the curves x = 0 and t = 0 are plotted in the (x, t) plane (labelled Bill),
as well as the curves x0 = 0 and t0 = 0 (labelled Jim), using the above formulae. With this
choice x0 = 0 corresponds to x = vt while t0 = 0 represents the curve t = xv/c2 , as drawn in
Figure 4.

Figure 3: Lines of fixed position and fixed time for two observers.

This kind of figure is called a space-time diagram, with time increasing towards the top
of the page and spatial lengths extending roughly horizontally. For Bill lines of simultaneous
events lie at fixed t and so are parallel to the x axis. For Jim simultaneous events have fixed
t0 and so are parallel to the x0 axis. So there is no one unique observer-independent slicing of
space-time up into space and time. This is called ‘relativity of simultaneity’. The relativity
of simultaneity poses potential problems, since a goal of physics is to predict the future from
the past. But how can this be possible if observers cannot agree on what the future, present
and past are? A problem of principle blocking the prediction of the future from the past is
called a ‘causality’ problem.

Page 5 of 8
Special Topics in 1st-Year Physics

The reason Einstein gets away with it is the proscription in relativity against moving
faster than light. So far all we know is that all observers must agree on the value of the
speed of light, c, but (so far) nothing prevents things moving faster than c. What gives rise
to the requirement that nothing moves faster than light is the causality problem to do with
people not all agreeing on what the future and past are. Limiting the speed of travel to
luminal and sub-luminal speeds does the job because it turns out that people always agree
on the ordering of events that are close enough together that one can get from one to the
other while moving only at luminal or sub-luminal speeds.
To see how this works, picture the surface in space-time swept out by a light wave that is
emitted in all directions from a particular event, P : that is, from a particular place at a given
time. This light waves sweeps out a surface in space-time called the light-cone of P . The
future light-cone describes all points to which the light wave eventually goes. (Similarly the
past light-cone is the set of all points from which an incoming spherical light wave arriving
at P would have come.)

Figure 4: Space-time diagram of the future and past light-cone of an event P .

If we choose P as our origin of coordinates, then the light cone is defined by the set of
points, {t, r}, that satisfy −c2 t2 + r · r = 0. All such points are said to be light-like separated
(or null separated) from P . Similarly, the set of points lying within the interior of the light-
cone at P satisfy −c2 t2 + r · r < 0 and are said to be time-like separated from P . Finally,
space-like separated points (from P ) satisfy −c2 t2 + r · r > 0.
What is important is that (as we saw above) the space-time interval (also called the
invariant interval),
(∆s)2 := −c2 (∆t)2 + ∆r · ∆r , (2.22)
between two events (where ∆t = t1 − t2 and ∆r = r1 − r2 ) is Lorentz-invariant; that is its
value is the same for all inertial observers. (The special case that observers agree on which

Page 6 of 8
Special Topics in 1st-Year Physics

events form a null interval, ∆s = 0, is the statement that observers agree on the speed of
light.) Therefore (∆s)2 > 0 for any two events for which there exists an observer for whom
they are simultaneous.
Because only events that are space-like separated from P can be regarded as being simul-
taneous with P for some observers, all observers agree on the temporal ordering of all events
separated by null and time-like intervals. But this means that all observers agree on the
ordering of all events that can exchange signals travelling at most at the speed of light. The
ambiguity of the ordering of space-like separated events does not cause causality problems
only because these events can never influence one another, but this only assured if we know
that information can travel faster than light.2

(a) The first signal from Jane to Jim. (b) The second signal from Bill to Bertha.

Figure 5: Illustration how two signals sent faster than light between pairs of observers can
be arranged to send a signal from one observer to him/herself in their past. Here Jim and
Jane are at rest relative to each other but are moving with constant speed relative to Bill
and Bertha (who are also at rest relative to one another).

Another way to make the same argument is to show that if signals could be sent faster
than light, then it would also be possible for some observers to send signals to themselves
backwards in time. The breakdown in causality then manifests itself through the ‘Back to
the Future’ paradox, wherein someone send a message back in time to prevent their own
birth taking place.
To see how this happens consider two pairs of inertial observers, where the members of
each pair are at rest relative to each other, but where the two pairs move relative to one
another with speed v (as in the figure). In the figure Jim and Jane are not moving relative to
one another but are moving at constant speed relative to Bill and Bertha. In the left panel
2
All bets are off once relativity combines with quantum mechanics, however, since then the uncertainty
principle means that if you really know you pass exactly through event P , then your momentum is sufficiently
uncertain that there is some probability that your speed might be greater than light. The synthesis of
relativity and quantum mechanics is nonetheless consistent and causal, but this consistency is very delicate
(requiring, for instance, the existence of ‘antiparticles’ — sharing exactly the same mass and exactly opposite
charges — for every existing species of particles).

Page 7 of 8
Special Topics in 1st-Year Physics

of the figure Jane sends Jim a signal (from B to A) at precisely the moment that she passes
Bertha. By assumption the signal moves much more quickly than light (where infinite speed
would connect points that Jim and Jane agree are simultaneous). The signal is received by
Jim just as he passes Bill, and in the right panel Bill (who also sees the signal) signals Bertha
(from A to C), also at a speed much faster than light (where infinite speed for them would
connect points that they agree are simultaneous). As the figure shows, the reflected signal
arrives back to Bertha (and Jane) before she would have sent the original signal, since C is
in the past of B.

Page 8 of 8