Does the definition of c as an exact value mean that any test of whether light is isotropic, dispersionless, etc. is placed beyond test, by definition? The answer is: “No. What is placed beyond test is the defined value of c, not any experimental investigations.”[1][2][3][4] For example, the special theory of relativity is based upon a number of postulates concerning properties of the speed of light.[5] In the present SI units system where the speed of light has a defined value of c = 299,792,458 m/s exactly, nonetheless these postulates can continue to be tested as experimental technique improves.
To illustrate how the properties of light may continue to be examined within the 1983 decision, consider the hypothetical observation of anisotropy in the propagation of light.[6] In the SI system of units, anisotropy would take the form of the metre having different lengths in different directions.[Note 1] Of course, a standard of length cannot be allowed to be uncertain, so the effect of this hypothetical anisotropy would be to change the definition of the metre, by adding a directional correction. At the same time, however, the explanation of this anisotropy would be attempted by improvements in theory, and one might conjecture that the successful explanation could involve an anisotropic propagation of light. In any event, the fundamental physical phenomenon of the propagation of light must be kept separate in our thinking from the numerical value of c in the SI system of units.[8] (Separation of these notions might be a good reason to use the symbol c0 for the SI conversion factor instead of c, a practice recommended by the CGPM.)
In sum, tests of the special theory of relativity, for example, are not impeded by an exact definition of c in the SI system of units. The experiments take the form of tests of the adequacy of the definition of the metre. If there are any necessary changes in this definition due to future experimental results, they will be accompanied eventually by theoretical explanations, and these theories may indeed invoke, if need be, as yet undiscovered fundamental properties of the propagation of light.
^
Adams, S (1997). Relativity: An Introduction to Space-Time Physics. CRC Press. p. 140. ISBN0748406212. One peculiar consequence of this system of definitions is that any future refinement in our ability to measure c will not change the speed of light (which is a defined number), but will change the length of the meter!
^
Tom Wilkie (Oct 27, 1983). "Time to remeasure the metre". New Scientist. Vol. 100. Reed Business Information. pp. 258 ff. ISSN0262-4079. From now on, if a physicist should somehow discover that light is travelling faster than we had thought, then the metre will be lengthened automatically to restore the defined value of the speed of light.
^
Edwin F. Taylor, John Archibald Wheeler (1992). Spacetime physics: introduction to special relativity (2nd ed.). Macmillan. p. 5. ISBN0716723271. What will be the consequences of a future, still better, measuring technique?...will that improvement in precision change the speed of light? No. Every past International Committee on Weights and Measures has operated on the principle of minimum dislocation of standards; we have to expect that the speed of light will remain at the decreed figure of 299,792,458 meters per second, just as the number of meters in a mile will remain at 1609.344. Through the fixity of this conversion factor c, any substantial improvement in the accuracy of defining the second will bring with it an identical improvement in the accuracy of defining the meter. Is 299,792,458 a fundamental constant of nature? Might as well ask if 5280 is a fundamental constant of nature.
^
A Brillet and JL Hall (1979). "Improved laser test of the isotropy of space". Phys Rev Lett. 42 (9): 549 ff. doi:10.1103/PhysRevLett.42.549. {{cite journal}}: Unknown parameter |month= ignored (help)
^
E Richard Cohen (1988). "Variability of the physical constants". In Venzo De Sabbata, V. N. Melnikov (ed.). Gravitational measurements, fundamental metrology, and constants; NATO Advanced Study Institutes series v.230. Springer. p. 93. ISBN9027727090. Although as the result of this definition, the quantity c cannot change, it is still possible to ask if the speed of light can change. To answer this question it is necessary to have a theory of the propagation of electromagnetic waves. With such a theory one might postulate...some operationally observable mechanism that could lead to detectable changes in the speed of light.
^For example, the Michelson-Morely experiment sought to compare the speed of light in various directions, and when no difference was found the Fitzgerald contraction was postulated to account for it within the aether theory by making the lengths different in different directions. This length difference was rendered unnecessary by the theory of relativity. In a reversal of history, if a difference in the speed of light actually were found today (at a much more refined level of accuracy), in contradiction to the special theory of relativity, the roundtrip time-of-flight length measurements in different directions would be different, indicating the lengths in different directions were different in SI units. In the rotating resonator approach to measuring anisotropy, “length variations of this cavity – whether accidental or cosmic – appear as variations of laser wavelength. They can be read out with extreme sensitivity as a frequency shift...”[7]
