The Null Observatory

Michelson–Morley · Cleveland, 1887

Self-contained / Web 3Dⓘ

01 / Physical hypothesis

In the apparatus rest frame, uniform motion adds no orientation-dependent optical delay.

02 / Turn the apparatus

Orientation θ

1× is one turn in six minutes. Run and Pause in the transport turn the stone; light probes use a separate, slowed clock.

Hypothetical in-plane drift

Drift azimuth ψ

03 / Align the optics

Illumination

Fine mirror displacement (nm)

Wavefront angle α

Equal-glass compensating plate Argand source on

Block arm A Block arm B

04 / See the mechanism

Overlay published 1887 means

Reset restores the balanced instrument and modern physics. The record is kept; Clear record removes it.

Drag to orbit, pinch to zoom, right-drag to pan. Click an optic to inspect it; double-click to focus.

Keyboard and pointer

Focus the scene by clicking it or with Tab. Then R starts or stops rotation, Space pauses probes, H returns to the apparatus, T selects the plan view, N opens the notebook and the wheel zooms. Escape releases the keyboard. Drag to orbit; right-drag or two fingers pan.

The 1887 apparatus / inspection view

Two paths. No aether signature.

A folded interferometer on a mercury-supported stone. Watch the phase, not the apparent speed of the light.

Building the observatory

Geometry · reflection · phase · scale

Telescope eyepiece

Orientation θ

Rotation fringe shift at θ

Change after a 90° turn

1887 hypothesis ΔN₉₀ ≈ 2Lv²/(λc²)

Rotation-only optical path difference

Current total phase

One-way folded length

Vacuum round trip

Calculated from the model: synthetic, not a recorded image. Fixed crosswire on a 6 mm virtual field.

Rotation signature · full scale

ModernAether (counterfactual)1887 noon (published)

Mean-centered fringes over a full turn. Published points are drift-reduced observations, not generated noise. The dashed curve is always the old-model prediction.

Selected component

Component

Only a reflecting face can be removed; the selected component is not a mirror.

Click an optic in the scene or choose it here. Documented dimensions and reconstructed details are distinguished in the notebook.

Recorded revolution · synthetic

θ	Rotation fringe	Total OPD (nm)	Arms live

Model output from the reconstructed geometry. Never mixed with the published 1887 observations.

Notebook

Primary evidence, distinct from the live model: the published 1887 aggregate means in wavelengths, drift included.

Noon meanEvening mean

Opposite-azimuth final means (fringe)

Azimuth	Noon final	Evening final	Noon drift-reduced

Raw readings (micrometer divisions)

Series

Michelson and Morley, American Journal of Science 34 (1887), 333–345 — AIP archive

Check	Residual	Acceptance

Reproducible checks on this implementation. They do not replace experimental validation of the omitted physics or archival verification of undocumented machining details.

Δ = OPL_B − OPL_A · φ(λ) = 2πΔ(λ)/λ · I = I_A + I_B + 2√(I_A I_B) cos φ · t_pair = (2ℓ/c) √(1 − β² sin²θ)/(1 − β²) · |ΔN₉₀| ≈ 2Lv²/(λc²)

A null signal is not a blank eyepiece. Both arms share one source; at each wavelength the returning fields add, and the eye averages over the optical oscillation. The telescope can show strong fringes whose position does not depend on rotation. In the ideal lossless model each branch carries a quarter of the input, and blocking or removing a reflector removes that branch, not merely its colored trace.

Every spectral component interferes with itself, but different frequencies add incoherently. White light is summed over 81 visible samples with an illustrative source envelope. The plate rays obey Snell's law for an assumed index of 1.517, and the broadband imbalance uses an illustrative Cauchy dispersion. Removing the compensator includes an ideal lateral re-alignment of the mirror banks.

Modern physics evaluates the experiment in the apparatus rest frame: uniform translation creates no preferred propagation direction, so rotating a fixed mirror bias produces no signature. The 1887 prediction is explicitly counterfactual — a stationary aether with Galilean velocity addition and no length contraction — evaluated segment by segment with a cancellation-safe expression.

This is geometric optics plus scalar interference, not a time-domain Maxwell solve. It omits coating polarisation, detailed diffraction, lens aberrations, air turbulence, thermal elasticity, rotation-induced relativistic corrections and a surveyed local velocity projection. A near-null result constrains round-trip anisotropy; it does not by itself measure one-way light speed or prove all of special relativity.

The 1887 paper gives key dimensions and folded-path drawings, not a manufacturing drawing. The sixteen mirrors are positioned so that each one-way path is exactly 11 m. Mount geometry, lamp, telescope, cover and brickwork are reconstructions, not archival measurements.

The scale control crosses astronomical, terrestrial, laboratory and micrometer frames instead of placing an astronomical unit and a nanometer in one mesh; readouts stay in meters. The field lens has metric wavelength but normalized amplitude. Probes move at one hundred-millionth of light speed in optical-distance coordinates; they are timing markers, not photons.

Documented and reconstructed parts

Part	Model specification	Evidence status
Sandstone block	1.50 × 1.50 × 0.30 m	Documented
Annular wooden float	Ø1.50 m outside; Ø0.70 m inside; 0.25 m thick	Documented
Iron trough	15 mm walls; approximately 10 mm clearance	Documented
Reflecting faces	50 mm diameter, speculum metal	Documented
Parallel glass plates	50 × 75 × 12.5 mm	Documented
Effective one-way arm	Approximately 11 m	Documented; model set to 11.000 m
Sampling	16 marks; one rotation in six minutes	Documented
Mirror centers and mounts	Four banks, repeated diagonal folding	Reconstructed; not a recovered coordinate survey
Lamp, telescope, hood, brickwork	Period-appropriate functional structure	Details and textures are illustrative

Case Western Reserve University institutional history·Exact vacuum light speed (NIST)

.museum-body--immersive { overflow: auto !important; } .museum-main--fill, .museum-runtime { height: auto !important; min-height: 100vh; overflow: visible !important; } .museum-runtime > .exhibit-root, .museum-runtime > :not(#mw-about):not(noscript):not(script) { display: none !important; } .museum-runtime #mw-about { display: block !important; position: static; width: min(100%, 800px); margin: 24px auto; padding: 24px; box-sizing: border-box; } .museum-runtime [data-about-close], .museum-runtime #mw-about > h2 { display: none !important; }

1887 · Physics

Michelson–Morley Interferometer, 1887

Turn a mercury-floated stone interferometer and compare the stationary-aether prediction with the missing fringe shift.

In 1887 Albert Michelson and Edward Morley floated a sandstone slab on mercury in Cleveland and folded two light paths of about eleven meters across it with sixteen mirrors. If light traveled through a stationary aether, turning the apparatus a quarter turn should have moved the white-light fringes by roughly four tenths of a fringe. They saw a shift far smaller than that.

This exhibit computes the reconstructed folded geometry, Snell refraction through the splitter and compensating plate, and the optical path difference between the arms. The eyepiece, rotation signature and recorded revolution come from that model. The 1887 prediction mode is an explicit counterfactual — a stationary aether with Galilean velocity addition — and modern physics gives no orientation-dependent signal.

The published 1887 noon and evening means are shown separately as historical evidence, including the drift the authors recorded. Mirror mounts, lamp, telescope and brickwork are reconstructions, and the model omits air turbulence, thermal drift, diffraction and polarisation. A near-null result constrains round-trip anisotropy; it does not by itself prove special relativity.

Specifications

Experiment: Michelson and Morley, Cleveland, July 1887
Effective arm: About 11 m one-way, folded by 16 speculum mirrors
Expected shift: ≈ 0.4 fringe for a quarter turn at 30 km/s (aether hypothesis)
Reported shift: Less than 0.02 fringe, probably less than 0.01
Rotation: Stone floated on mercury; 16 marks, one turn in six minutes
Modeled: Folded ray geometry, glass dispersion, scalar interference, published means

All exhibits

Two paths. No aether signature.

In the current issue