Michelson and Morley's experiments (see the Introduction to this topic) showed that there was no difference in the speed of light when the earth was moving through the ether in different directions, suggesting that there was no such thing as an ether. However, the properties of the ether underpinned much of physics and, understandably, physicists were not willing to give it up easily. In the 1890s, G.F. Fitzgerald and H.A. Lorentz independently proposed that any length (including Michelson and Morley's experimental apparatus) must shrink in the direction of motion through the ether by a factor = . In fact, Fitzgerald and Lorentz saw that for the laws of physics to be preserved in all inertial reference frames, the Galilean transformations of Newtonian physics had to be replaced. However, no rationale or theory was provided for these particular transformations; Fitzgerald and Lorentz deduced their transformations from the mathematics of electromagnetism and not from any understanding of the relativistic nature of motion. It was not until 1905 that Einstein's theory showed the rationale behind the Lorentz transformations (sometimes called the Lorentz-Fitzgerald transformations).

It is possible to derive the lorentz transformations from the postulates of Special Relativity). However, the derivation is long and not particularly enlightening because there are several assumptions which are hard to justify without delving deeply into the mathematics of spacetime. The result of the derivation is:

Δx = γ(Δx' + vΔt) | |||

Δt = γ(Δt' + vΔx/c^{2}) |

where:

γâÉá |

What does all this mean? The primed variables (

Figure %: Relative motion of *F* and *F'*.

Δx' = γ(Δx - vΔt)Δt' = γ(t - vx/c^{2}) |

Also, the Lorentz transformation in the

Note that in the limit *v* < < *c* (that is, when the velocity involved is nowhere near the speed of light),
*γ* 1 and the transformations reduce to *x* = *x'* + *vt'* and *t* = *t'*. As we would expect
(from the correspondence principle), these are the familiar Galilean transformations. We will now
see how the lorentz transformations can be easily applied to show the results we have already derived.

If two events are simultaneous in *F'*, then *Δx'* = *x'* and *Δt'* = 0. Plugging into the equation
for *Δt* we find: *Δt* = , which is non-zero unless *x'* = 0 or *v* = 0.
Thus the events do not occur simultaneously in frame *F* (*Deltat* 0 implies that there is a time
difference between the events).

If two events occur at the same place in *F'* then *Δx'* = 0 and *Δt'* = *t'*. Using the second
equation, the separation in time between the events in *F* is: *Δt* = *γΔt'* (for *Δx'* = 0). Similarly if events occur in the same place in *F*, *Δx* = 0 and *Δt* = *t*. Then the second
inverse transformation tells us: *Δt'* = *γΔt* (for *Δx* = 0). Thus we again have
arrived at the seeming-contradiction we saw in Section
2. However, here it is
clear
that one equation applies when *Δx* = 0 and one when *Δx'* = 0; the nature of the Lorentz
transformations themselves assure us that these cannot both the satisfied for any two events.

In the section on length contraction we noted that any measurement of length
requires that the coordinates of the ends of the object be recorded simultaneously. To measure the length of a moving train,
for example when might place two time bombs, primed to go off simultaneously, at opposite ends of the train. The length of
the train is the distance between the explosions. Note that if the explosions were not simultaneous (say the
explosion at the rear occurred first), the train would move between the explosions and you would measure
an incorrect length (too long, in this case). Thus if we have a pole of length *l'* in frame *F'* and it is lying
along the *x'*-axis, what is the length in *F*? In *F* we make our simultaneous measurements and we
have *Δx* = *x* and *Δt* = 0. From the first Lorentz transformation we have: *Δx'* = *γΔx* (for *Δt* = 0). *Δx* is by definition the length in *F*, and since the pole is not
moving in *F'*, *Δx'* is its length in *F'*. Thus *l* = *l'*/*γ*, just as we discovered in Section 2. We could also analyze a
situation when a pole is at rest in *F*, and find
the seeming-contradictory result *l'* = *l* /*γ*. As we have seen the former equation applies only to
situations where *Δt* = 0 and the latter to those where *Δt'* = 0. Everything depends on which frame
the simultaneous measurements are made. (See Section
2.)