But is it true? Is the speed of light really independent of the motion of the observer?

The technical background, which is not necessary for the rest of the discussion, is this: The spectrometer uses the wave property of interference to compare the time that light takes to travel along the two paths. If it takes equal times, then the rays combine in step (technically: in phase) and the resultant beam is bright. If they are out of step by half a wavelength, then they cancel out, and the combined beam is dark (at the particular angle). Looking into the combined beam, what one sees is a series of bright and dark rings, corresponding to reinforcement or cancellation of the light rays that have travelled along the two different paths.

Schematic and animation of the Michelson-Morley experiment. A beam of light (actually continuous, not pulsed as in the animation) is split when it strikes a transparent plate: part is transmitted, part reflected. When the two divided beams return to the plate, partial reflection and partial transmission also combines them.

The spectrometer could be rotated, so one of these paths could be parallel to the direction of the Earth's motion around the sun, while the other was at right angles. Further, by doing two measurements twelve hours apart, one could add or subtract the speed of the Earth's surface due to its rotation about its axis. With two measurements 6 months apart, one could add or subtract these speeds from that due to the motion of the Earth around the sun. (The motion of the sun around the centre of the galaxy is effectively constant over our history. The motion of the lab due to the Earth's rotation is rather slower than that due to the motion around the sun.)

All that from watching the changes in bright and dark stripes of light in an interference pattern — what a cool experiment! The diagram shows schematically how this works. Qualitatively, one can see that both paths should be slowed a little by motion with respect to the aether. But the amounts do not cancel out. Go to this link for a quantitative treatment.

In retrospect, it was indeed a cool experiment, but not in the way the experimenters thought. It is now considered as one of the most famous null results in history. In all of the different times and seasons, the orientation of the spectrometer failed to add or to subtract the putative motion of the lab through the aether.

The simplest interpretation of the results is that light travelled at the same speed with respect to the lab, whether or not the arm of the spectrometer were travelling with the Earth through the aether or at right angles to it. (One could also make explanations in which the speed of light varied, but the shape of the spectrometer changed according to its orientation, in such a way that it exactly cancelled the effect of the lab's motion.)

Many further experiments have been performed to look for variations in the speed of light with respect to relative motion, usually by looking at the speed of light in different directions, as Michelson and Morley did. One can never show that the difference is zero, one can only give an upper limit to the possible ratio of speeds in the different directions. Currently, Stephan Schiller's team in Dusseldorf gives an upper limit to the possible variation of 6 parts in 10¹⁶. The ratio of the speeds is 1.000 000 000 000 000 0 plus or minus 0.000 000 000 000 6. (Reported in Nature's research highlights, 9/6/05.)

Basically, no-one has yet found a case in which the speed of light in vacuum is different.

I should point out that a number of reports cliam that the speed of light in a gas is not isotropic, but that the effect is very small compared with that expected from a stationary aether, so the data are not immediately persuasive. For example, this paper by Reg Cahill of Flinders University puts the case for retaining the Lorentz equations but abandoning Einstein's principle of special relativity.

"But it can't be so." "It just doesn't make sense."

These are common responses — and they were certainly the responses of this author when I first read about relativity. The answer to the first is simply that it is not up to us to decide in advance what is and what is not so (in spite of what Plato might have said). That is the job for observation and experiment (as Plato's student Aristotle, and even more emphatically Galileo, might have told Plato). To the second objection, we might say that it is not up to us to tell the universe what to do. The universe just is. It is up to us to make sense of it. For scientists, this means finding theories and laws whose predictions are in agreement with what we observe in the universe. Relativity is a theory that has been very thoroughly and precisely tested, and whose predictions are in spectacularly good agreement with the behaviour of the universe.

The principle of Special Relativity, including its weird consequence that the speed of light is the same for different observers, is not illogical. It is not false. It may be surprising or upsetting. Deep down, I think that most people who object to the principle of Special Relativity are saying "It may be true, but it wouldn't be true if I had designed the universe" or "I don't like the principle of relativity". To this objection, the universe is unlikely to register offence.

Notice that the objections mentioned in the previous paragraphs are in fact not objections to a theory, but objections to the results of experiments. The invariance of the speed of light for different observers is an experimental observation. That clocks run at different rates in different frames is an experimental observation. If you think that these are weird, then it's the universe that you are calling weird, not our theories about it. You can't wait for some new theory that isn't weird: any theory that did not make the same predictions as relativity (in the cases that we have measured so far) would be immediately recognised as false, because it wouldn't agree with experiments.

This argument doesn't stop the observations from being weird. But those are the observations. It seems that we have three options: (i) accept them, (ii) forget about them or (iii) go and live in a universe where they aren't true.

See also More on the Michelson-Morley experiment.

Science and other ways of looking at the world

I am a physicist, so you may think that I am biassed about physics and about the scientific method. Further, physics is an orthodoxy: only on a small number of questions at the very periphery of knowledge will you find serious disagreement among physicists. Further, the orthodoxy is institutionalised: someone who believed that apples fell up would find it difficult to obtain a employment as a professional physicist. So the physics world view reinforces itself. Can you trust it?

Let's make the question stronger: would you trust physics with your life? Many readers will have flown in an aircraft, which had been designed by engineers with a good understanding of the relevant areas of physics. It was navigated using a system designed by engineers and physicists with a good knowledge of relativity. How would you feel about travelling in an aircraft designed according to the principles of a one of the systems of thought that arrives at quantitatively different conclusions from those of physics and engineering? How many times do you trust your life to physics and to its applied science, engineering?

So why is physics, in general, an orthodoxy? Suppose that there are different ways of looking at a problem. Do they give different answers? If they do, we conduct experiments and we abandon, or at least qualify, those that give the wrong answers. If they give the same answers, then we look to see whether they are actually equivalent. For instance, Newton's and Hamilton's laws of mechanics appear to be quite different ways of looking at the world. Newton's system is more like a philosopher's 'cause and effect', whereas Hamilton's is more like 'purpose'. However, one can be derived from the other, so we see them as logically equivalent and related to each other fairly simply via calculus.

So, in most cases, we can look at competing theories and discard some of them because they fail the test of experiment. Those that remain are usually equivalent. Sometimes, theories only differ in exotic ways that are (currently) impossible to test. So, for instance, there are currently competing theories of quantum gravity, and none of them is accepted: they are often regarded as speculative.

Sometimes we retain for everyday use theories that, in exotic situations, are shown to be false. So, for instance, when speeds and energies are low, we use Galilean relativity and Newtonian mechanics, because they are considerably easier to use and because, when speeds and energies are low, they give the same answers as relativistic mechanics, to excellent approximations.

If you're interested, we show film clips of measurements of the speed of light and of (UHF) radio waves in a multimedia tutorial on The Nature of Light.