3.12.3.1 - The Michelson-Morley Experiment

Ether and absolute motion

The ether was a proposed medium filling space through which light waves travelled. In the stationary-ether model, light had speed $c$ relative to that medium. Motion relative to it would identify a preferred reference frame, unlike ordinary motion measured relative to another object.

If Earth moved through stationary ether, an Earth-fixed instrument would experience an ether wind. In the model, light sent parallel to that motion would make its outward and return journeys under different relative-speed conditions from light sent perpendicular to it. Equal-length perpendicular arms would therefore have different round-trip times.

This is a prediction of the particular ether model, not a statement that every conceivable ether theory must give the same result. Earth's orbital motion provided a reason to expect a measurable effect; Earth need not be absolutely at rest merely because one experiment found no effect.

Splitting and recombining light

An interferometer converts a small change in the relative optical paths of two beams into a shift of interference fringes. A common source supplies both beams so they can interfere coherently.

A half-silvered beam splitter at $45^\circ$ transmits some light along one arm and reflects some along a perpendicular arm. Mirrors return both beams to the splitter. Components of the returning beams overlap and travel to the same viewing telescope or detector, producing bright and dark fringes.

[DIAGRAM: asset_name: D.3.1 - The Michelson-Morley Experiment - Diagram 1; asset_slug: D.3.1 - The Michelson-Morley Experiment - Diagram 1; recommended_method: generated_png; caption: Perpendicular light paths recombine at a Michelson interferometer; description: Ideal thin beam splitter sends light along perpendicular round trips. Both returning beams meet the splitter and overlap towards the detector.]

Principle schematic: both returning beams meet the beam splitter. Paths are slightly separated for clarity; the glass compensation used in a real instrument is omitted.

A bright fringe occurs where the beams arrive in phase. A change of optical path difference by one wavelength moves the pattern by one fringe spacing. Thus a predicted change $\delta(\Delta L)$ corresponds to a fringe shift

For example, a path-difference change of $1.80\times10^{-7}\,\mathrm{m}$ with wavelength $6.00\times10^{-7}\,\mathrm{m}$ gives $N=0.300$ fringe spacings. That is detectable by an instrument resolving $0.05$ fringe spacings. Fractional fringe shifts are measurable; an entire bright fringe need not disappear.

In a real glass-plate instrument, a compensating plate makes the beams traverse equal total thicknesses of glass. This compensates the unequal optical delay from glass, particularly helping fringe visibility with a range of wavelengths. It does not make the arms physically equal or remove vibration, temperature drift and other sources of error.