Ether Drift and the Absolute Rest Frame

In December 1725 James Bradley pointed a purpose-built zenith sector at the star γ Draconis, hoping to catch it wobbling. He was hunting stellar parallax — the annual shift a nearby star should show as the Earth swings from one side of its orbit to the other, the last missing proof that the Earth really moves. He found a wobble of about twenty arcseconds. But it was the wrong wobble: it peaked three months out of step with parallax, and every star showed it, near or far. Parallax depends on where the Earth is; this depended on where it was going. What Bradley had found was aberration. Because light travels at a finite speed and the Earth moves through space at $29.8\ \mathrm{km\,s^{-1}}$ , a telescope must be tipped slightly forward — into the direction of the Earth's motion — to catch a star's light, exactly as you tilt an umbrella forward walking through vertical rain. The tilt is $\tan\alpha = v/c = 9.94\times10^{-5}$ radians, or $20.5$ arcseconds. It was the first direct observational proof that the Earth moves, and, run backwards, a measurement of the speed of light. It also handed the ether a loaded gun. If light is a wave in a medium and the Earth ploughs through it, the measured speed of light should depend on direction: $c - v$ one way, $c + v$ the other, a fractional variation $\Delta c/c \approx 10^{-4}$ . That is a first-order effect, and by optical standards enormous. Aberration proved the ether — if it existed — was not dragged along by the Earth, so the wind should have been blowing through every laboratory on the planet. Nobody could measure it, and the reason is one of the sharpest lessons in experimental physics.

Key takeaways

Stellar aberration is the annual tilting of starlight caused by the Earth's motion: $\tan\alpha = v/c$ , giving $\alpha = 9.94\times10^{-5}\ \mathrm{rad} = 20.5''$ and an annual swing of $41''$ . It is first order in $\beta = v/c$ , it is real, and it proves the Earth moves at about $30\ \mathrm{km\,s^{-1}}$ .
Bradley set out in 1725 to measure stellar parallax and found aberration instead — a different effect, ninety degrees out of phase, and the same for every star.
It fits a stationary ether — or no ether at all — but it kills the rescue in which the Earth drags the ether along, since a co-moving ether would give no tilt.
In a stationary ether the one-way speed of light should read $c \pm v$ , so $\Delta c/c \approx 10^{-4}$ . Nobody could measure it: a one-way speed needs two clocks synchronised in advance.
Key idea: the ether wind is first order in $\beta$ and huge — but only ONE-WAY. Every practical optical experiment sends light out and back to the same clock, and on a round trip the $+\beta$ and $-\beta$ errors cancel exactly, leaving a residue of order $\beta^{2} \approx 10^{-8}$ — ten thousand times smaller. That is why the ether hid, and what The Michelson–Morley Experiment had to be built to reach.

The core insight

Two consequences of the same $\beta = v/c \approx 10^{-4}$ behave completely differently. Aberration is a directional effect, first order in $\beta$ , and it is seen: starlight arrives tilted by $\alpha = v/c = 20.5''$ . The ether wind's speed anisotropy is also first order — but only one-way, and no nineteenth-century lab could time a one-way flight of light. Fold the light back to the same clock and the first-order terms cancel identically, leaving $O(\beta^{2}) \approx 10^{-8}$ . The ether was not hiding because it was subtle; it hid behind one extra factor of $\beta$ .

Where this lives today

Aberration is no historical curiosity: it is a correction applied to every precise position measured in the sky. The Gaia spacecraft pins stellar positions to tens of microarcseconds — and aberration is $20.5''$ , about a million times larger. Every Gaia astrometric solution models the observatory's velocity exquisitely and strips the aberration out first; so does very-long-baseline radio interferometry. Get it wrong and the whole celestial reference frame tilts. The same $v/c$ appears whenever something fast must aim a beam of light. A laser terminal on a satellite in low Earth orbit, moving at $7.7\ \mathrm{km\,s^{-1}}$ across the line of sight, cannot point at where its partner appears to be: it must lead the target by a point-ahead angle of $2v/c = 51\ \mathrm{\mu rad}$ , about $10.6''$ — Bradley's tilt, doubled for the round trip and built into a steering mirror. And a preferred frame did turn up in the end, just not the one the ether promised: the cosmic microwave background carries a dipole showing the Solar System moving at some $370\ \mathrm{km\,s^{-1}}$ relative to it, $\beta = 1.2\times10^{-3}$ . But that frame is picked out by matter and radiation, not by the laws — light still travels at $c$ in every direction, in every inertial frame, and there is no wind.

The physical picture

Walk through vertical rain. Stand still and it falls straight onto your head, so the umbrella goes straight up. Start walking and the rain seems to slant towards you from ahead, and you tilt the umbrella forward. Nothing changed about the rain; what changed is that you now measure its velocity relative to you, and relative to you it has a backward horizontal component equal to your own speed. The tilt satisfies $\tan\alpha = (\text{your speed})/(\text{rain speed})$ . Starlight is the rain, the Earth is the walker, and $\beta \approx 10^{-4}$ — small but not zero, and it reverses every six months. Now ask the ether question. If light is a wave in a medium at rest in space, the Earth is not just a walker in the rain; it is a swimmer in a river. The whole laboratory is blown through a stationary ether at $29.8\ \mathrm{km\,s^{-1}}$ , and the light inside it should be swept downstream.

Figure

The ether wind. A lab sweeping through a stationary ether at v = 29.8 km/s should measure light's one-way speed as c − v downwind and c + v upwind — a FIRST-order split, Δc/c = 9.94 × 10⁻⁵ ≈ 10⁻⁴. Across the wind the change is only second order: √(c² − v²) = c(1 − 4.9 × 10⁻⁹), a 1.5 m/s drop.

Nature showed the first-order effect in the sky instead. Aberration is the same $v/c$ , but it appears as an angle rather than a speed — and angles you can measure against the fixed stars, with no clock at all.

Figure

Stellar aberration. Left — walk through vertical rain and it seems to come from ahead: tilt the umbrella forward. Right — the photon falls straight down at c while the telescope slides forward by v·t, so the tube must be tipped forward by α with tan α = v·t/c·t = v/c. For v = 29.8 km/s: α = 9.94 × 10⁻⁵ rad = 20.5″, annual swing 2α = 41″. Tilt hugely exaggerated — the true angle is 0.0057°.

Notice what the figure does not contain: an ether. The tilt follows from two facts only — light has a finite speed, and the telescope is moving. That is why aberration cannot decide between a stationary ether and no ether. What it can decide is whether the Earth drags the ether, and the verdict is brutal: if the ether inside the tube moved with the tube, the light would fall straight down the axis and there would be no tilt at all.

The core equations

The whole of aberration is one triangle. In the frame where the ether — or empty space — is at rest, light from a star at the ecliptic pole arrives with velocity $(0, -c)$ . The telescope moves with velocity $(v, 0)$ . Subtract velocities the Galilean way, with absolute time: in the telescope's frame the light travels at $(-v, -c)$ , and the tube must lie along it.

\tan\alpha = \frac{v}{c} \equiv \beta \qquad (1)

where: $\alpha$ — the aberration angle, the forward tilt of the telescope from the star's true direction; $v = 29.8\ \mathrm{km\,s^{-1}}$ ; $c = 2.998\times10^{8}\ \mathrm{m\,s^{-1}}$ . Physical reading: the tilt is set by the ratio of two speeds and nothing else — not the star's distance, not its brightness, not the medium. Since $\beta \approx 10^{-4}$ , $\tan\alpha \approx \alpha = 9.94\times10^{-5}$ radians. Turning that pure number into an angle an astronomer can read takes one conversion factor — worth doing explicitly, because forgetting the $3600$ is the classic slip.

\alpha[''] = \alpha[\mathrm{rad}] \times \frac{180}{\pi} \times 3600 = 9.94\times10^{-5} \times 206264.8 = 20.5'' \qquad (2)

where: $206264.8$ — arcseconds in one radian. Physical reading: $20.5''$ is the semi-amplitude. Over a year the Earth's velocity reverses, so a polar star traces a circle of angular radius $20.5''$ and the full swing is $41''$ — about one forty-sixth of the Moon's apparent diameter. Bradley measured just over $20''$ ; the modern constant of aberration is $20.4955''$ . Now the other face of the same $\beta$ . If the ether is real and stationary, light moves at $c$ in its frame, so a laboratory sweeping through it at $v$ must compose velocities. A pulse sent along $\hat{n}$ , at angle $\theta$ to the wind, has lab-frame speed $|c\hat{n} - \mathbf{v}|$ — to first order, a cosine.

c(\theta) = \left|c\hat{n} - \mathbf{v}\right| \approx c - v\cos\theta \qquad (3)

where: $\theta$ — angle between the light's direction and the ether wind; $c(\theta)$ — the one-way speed measured in the laboratory. Physical reading: light running with the lab's motion measures $c - v = 2.99770\times10^{8}\ \mathrm{m\,s^{-1}}$ ; against it, $c + v = 2.99830\times10^{8}$ . The split is $\Delta c/c = \beta = 9.94\times10^{-5}$ — first order, and by optical standards a giant. Across the wind, $\sqrt{c^{2}-v^{2}} = c(1-\beta^{2}/2)$ : a drop of $4.9\times10^{-9}$ , or $1.5\ \mathrm{m\,s^{-1}}$ . Second order, and negligible.

Where these results come from

Derivation AStellar aberration and the velocity triangle

This is the article's owned structural result, and the only measurement in this band that actually worked. The goal is to derive the aberration angle from nothing but a finite light speed and a moving observer, extract the number 20.5 arcseconds, and then run the argument backwards — because if you know the Earth's orbital speed and you measure the tilt, you have measured the speed of light. On the way we see why an ether dragged along by the Earth is dead on arrival.

Step 1. Put a star at the ecliptic pole and work in the frame where the ether (or empty space) is at rest. Its light arrives with velocity $\mathbf{u} = (0, -c)$ : straight down, no horizontal component. Why: choosing the pole strips away the star's own geometry, so any tilt we find can only come from the observer's motion.

Step 2. Give the telescope the Earth's velocity $\mathbf{V} = (v, 0)$ . In the telescope's frame the light's velocity is $\mathbf{u}' = \mathbf{u} - \mathbf{V} = (-v, -c)$ — still falling at $c$ , but now drifting backward at $v$ . Why: Galilean subtraction with absolute time is the only kinematics available in 1728, and it is right to first order in $\beta$ ; relativity changes the angle only at $O(\beta^{2}) \approx 10^{-8}$ .

Step 3. Point the tube along $\mathbf{u}'$ , not $\mathbf{u}$ . In time $t$ the photon drops $ct$ while the telescope advances $vt$ , so the tube must lean forward by $\alpha$ with $\tan\alpha = vt/ct = v/c$ . Why: a photon entering the objective must land on the eyepiece, and the eyepiece slides forward while it falls. The tube aligns with the light's velocity relative to itself — exactly the umbrella.

Step 4. Put in numbers: $\beta = 9.94\times10^{-5}$ , so $\alpha \approx \beta$ radians; multiply by $206264.8$ to get $\alpha = \mathbf{20.5''}$ . Over six months $\mathbf{V}$ reverses, so the star swings through $2\alpha = 41''$ and traces an annual ellipse. Why: it is the reversal, not the tilt, that makes aberration observable — a constant offset shared by all stars would be invisible, but a $41''$ oscillation in step with the Earth's velocity cannot be missed.

Step 5. Run it backwards, then kill the dragged ether. Solving (1) for $c$ with $\alpha = 20.4955''$ and $v = 29.8\ \mathrm{km\,s^{-1}}$ gives $c = 2.999\times10^{8}\ \mathrm{m\,s^{-1}}$ — three figures, from a star and a plumb line. Now suppose the Earth carried the nearby ether with it: inside the tube the light would travel at $c$ with respect to the tube, straight down its axis, so $\mathbf{u}' = (0,-c)$ and $\tan\alpha = 0$ . Why: no tilt, no annual ellipse, no aberration. But aberration is there, at exactly the value an undragged medium predicts — the constraint Fresnel's partial-drag coefficient was built to satisfy, and which Stokes' full-drag theory of 1845 had to contort itself to evade. *As a check: the velocity triangle gives $\tan\alpha = v/c$ symbolically; the explicit rad → arcsec conversion gives $\alpha = 20.50''$ and $2\alpha = 41.0''$ ; $c = v/\tan(20.4955'') = 2.999\times10^{8}\ \mathrm{m\,s^{-1}}$ ; $1\,\mathrm{AU}/c = 499\ \mathrm{s}$ ; and the dragged-ether case returns $\alpha = 0$ identically.*

Go deeper · Derivation B · 2 minThe ether wind, and the extra factor of β that hid it

Derivation BThe ether wind, and the extra factor of β that hid it

Aberration says the Earth moves through the ether without dragging it. So the wind must be blowing through every laboratory on Earth at $29.8\ \mathrm{km\,s^{-1}}$ , and the one-way speed of light must depend on direction at first order in $\beta$ — one part in ten thousand, not a small number for an optician. Yet nobody detected it for half a century. The goal here is to show exactly why: the first-order effect is real but unmeasurable, and everything you can measure with a single clock is second order.

Step 1. Compose the velocities. In the ether frame a pulse along $\hat{n}$ travels at $c\hat{n}$ and the lab moves at $\mathbf{v}$ , so the lab measures $|c\hat{n} - \mathbf{v}| = c\sqrt{1 - 2\beta\cos\theta + \beta^{2}} \approx c - v\cos\theta$ . Why: dropping $O(\beta^{2})$ leaves a clean cosine — the classic wind signature, maximal along the motion and vanishing across it.

Step 2. Read off the size: downwind $c - v = 2.99770\times10^{8}\ \mathrm{m\,s^{-1}}$ , upwind $c + v = 2.99830\times10^{8}$ , so $\Delta c/c = \beta = 9.94\times10^{-5}$ . Why: enormous, for optics. Measure it and you read the Earth's absolute velocity straight off the dial — and physics has an absolute rest frame.

Step 3. Try to measure it, and hit the wall. Equation (3) is a one-way speed: you must record when the pulse left A and when it arrived at B. Over an $11\ \mathrm{m}$ path the there-versus-back gap is $\Delta t = 2Lv/(c^{2}-v^{2}) = 7.3\ \mathrm{ps}$ — some $3700$ wavelengths of sodium light, a colossal signal. Extracting it needs clocks at A and B synchronised beforehand to a picosecond. Why: no such clocks existed. Nor can you synchronise them with a light signal, because the one-way speed of light is precisely the unknown — the argument is circular. Not a technological accident but a structural difficulty, one that relativity confronts head-on in Einstein's Two Postulates.

Step 4. So do what every real optical experiment does: send the light out and back to the same clock. Along the wind,

t_{\parallel} = \frac{L}{c-v} + \frac{L}{c+v} = \frac{2L/c}{1 - \beta^{2}} = \frac{2L}{c}\left(1 + \beta^{2} + O(\beta^{4})\right).

The $+\beta$ and $-\beta$ terms have cancelled identically. Why: you lose time fighting the wind and gain it running with the wind, and to first order the two exactly compensate. What survives is that you spend longer in the slow leg than the fast one — a second-order residue, and second order only.

Step 5. Count the cost. The survivor is $\beta^{2} \approx 10^{-8}$ , smaller than the one-way effect by $1/\beta \approx 10^{4}$ . Why: that single factor of $\beta$ is the whole story of the ether's concealment. One part in $10^{4}$ was routine; one part in $10^{8}$ demanded an instrument nobody had built — an interferometer with metres-long folded arms, floated on mercury, read to a hundredth of a fringe. Building it, and finding nothing, is The Michelson–Morley Experiment. *As a check: $c(\theta) = c - v\cos\theta$ to first order (confirmed by series expansion); $c \pm v$ to six figures; $\sqrt{c^{2}-v^{2}} = c(1 - 4.94\times10^{-9})$ ; the $11\ \mathrm{m}$ one-way gap is $7.29\ \mathrm{ps} = 3713$ sodium wavelengths; and $t_\parallel = (2L/c)(1 + \beta^{2} + O(\beta^{4}))$ , with no linear term.*

Limiting cases

Case	Result	Meaning
$c \to \infty$	$\alpha \to 0$	with infinitely fast light there is no aberration — the $20.5''$ exists only because $c$ is finite, which is why Bradley's angle measures $c$
Fully dragged ether (Stokes)	$\alpha = 0$	refuted — a co-moving ether sends light straight down the axis, and the observed $20.5''$ kills it
First order in $\beta$	$\alpha = \beta = 20.5''$ ; one-way $\Delta c/c = 10^{-4}$	the angle IS observed; the speed split is not, for want of pre-synchronised clocks
Second order in $\beta$	two-way anisotropy $\sim \beta^{2} = 10^{-8}$	all a single-clock experiment can reach — the target of The Michelson–Morley Experiment
Relativistic replacement	$c$ is the same in every direction, in every frame	the wind vanishes but the tilt survives: aberration transforms a direction, not a speed, and to first order the angle is unchanged

Common trapAberration is not parallax — it goes with where the Earth is going, not where it is

This is the confusion Bradley fought through, and it is still the standard trap. Parallax is a geometric effect of the Earth's position: a nearby star shifts because you view it from opposite ends of a baseline $2\,\mathrm{AU}$ across. It differs for every star (bigger for nearer ones), and even for the closest it is under $1''$ . Aberration is a kinematic effect of the Earth's velocity: $\tan\alpha = v/c$ gives $20.5''$ for every star regardless of distance, and it is maximal when the Earth's velocity — not its displacement — lies across the line of sight. Velocity leads displacement by a quarter cycle, so the two are ninety degrees out of phase, and that mismatch is what told Bradley his $20''$ wobble could not be parallax. If you ever expect a nearer star to show more aberration, stop: aberration does not care how far away the star is.

Worked examples

1 · Bradley's angle, and the speed of light from a star

The Earth orbits at $v = 29.8\ \mathrm{km\,s^{-1}}$ and light travels at $c = 2.998\times10^{8}\ \mathrm{m\,s^{-1}}$ . Find the aberration angle in arcseconds, then run the calculation backwards to recover $c$ .

Step 1. Form the ratio: $\beta = 2.98\times10^{4}/2.998\times10^{8} = \mathbf{9.94\times10^{-5}}$ . Because $\beta \ll 1$ , $\tan\alpha \approx \alpha = 9.94\times10^{-5}$ radians — good to three parts in $10^{9}$ , so the small-angle step costs nothing.

Step 2. Convert radians to arcseconds. One radian is $(180/\pi)\times3600 = 206264.8''$ , so

\alpha = 9.94\times10^{-5} \times 206264.8 = \mathbf{20.5''}.

The annual swing is $2\alpha = \mathbf{41''}$ , about one forty-sixth of the Moon's apparent diameter ( $0.52^\circ = 1872''$ ) — invisible to the eye, but a huge angle for a zenith sector.

Step 3. Reverse it: $c = v/\tan\alpha$ . With the modern aberration constant $20.4955'' = 9.936\times10^{-5}\ \mathrm{rad}$ ,

c = \frac{2.98\times10^{4}}{9.936\times10^{-5}} = \mathbf{2.999\times10^{8}\ m\,s^{-1}},

within $0.035\%$ of the accepted value. A star, a telescope and the Earth's orbit deliver $c$ to three figures.

Step 4. Note what Bradley actually did. Not knowing $v$ well, he ran the equation the other way: from his angle and the Earth–Sun geometry he concluded sunlight takes about $8$ min $12$ s to reach us. The modern value is $1\,\mathrm{AU}/c = 499\ \mathrm{s} = 8$ min $19$ s — Bradley low by $\mathbf{1.4\%}$ , in 1728.

2 · How big is the ether wind, and why could nobody see it?

A laboratory is swept through a stationary ether at $v = 29.8\ \mathrm{km\,s^{-1}}$ . Find the predicted one-way anisotropy, express it as a timing difference over $L = 11\ \mathrm{m}$ , then show what a round trip does to it.

Step 1. One-way speeds: downwind $c - v = \mathbf{2.99770\times10^{8}\ m\,s^{-1}}$ , upwind $c + v = \mathbf{2.99830\times10^{8}\ m\,s^{-1}}$ , so $\Delta c/c = \mathbf{9.94\times10^{-5}} \approx 10^{-4}$ — one part in ten thousand.

Step 2. Turn it into a laboratory number. Over $L = 11\ \mathrm{m}$ the downwind–upwind flight-time difference is

\Delta t = \frac{L}{c-v} - \frac{L}{c+v} = \frac{2Lv}{c^{2}-v^{2}} = \mathbf{7.3\ ps},

which in wavelengths of sodium light ( $\lambda = 589\ \mathrm{nm}$ ) is $c\,\Delta t/\lambda = \mathbf{3700\ waves}$ . Not a subtle signal — thousands of fringes.

Step 3. Fold the path back to one clock — which you must, since a one-way time needs two clocks synchronised in advance to better than a picosecond, and no such pair existed in 1880. Along the wind,

t_{\parallel} = \frac{L}{c-v} + \frac{L}{c+v} = \frac{2L}{c}\,\frac{1}{1-\beta^{2}} \approx \frac{2L}{c}\left(1 + \beta^{2}\right),

so the fractional excess over $2L/c$ is not $\beta$ but $\beta^{2} = \mathbf{9.9\times10^{-9}} \approx 10^{-8}$ .

Step 4. Count what that cost: the measurable effect is smaller than the unmeasurable one by $1/\beta \approx \mathbf{10^{4}}$ . The ether wind was never subtle — everything a laboratory could observe was suppressed by one extra power of $\beta$ . Reaching $10^{-8}$ forced a new instrument into existence: The Michelson–Morley Experiment.

Physics of the portrait

The man who proved the Earth moves was looking for something else.

Origins · 2 minJames Bradley and how this was discovered

James Bradley

1692–1762 · Oxford, Wanstead and the Royal Observatory, Greenwich

Bradley was an Oxford-trained astronomer and, from 1742, the third Astronomer Royal — Halley's successor at Greenwich. In December 1725, working with the wealthy amateur Samuel Molyneux at Kew, he mounted a zenith sector: a long telescope fixed almost vertically, so γ Draconis would pass nearly overhead and refraction would be negligible. The target was stellar parallax, which would settle Copernicanism once and for all. Within months the star was moving — by about $20''$ , but at the wrong times of year and in the wrong sense for parallax. Bradley reportedly saw the answer while sailing on the Thames, watching the boat's wind-vane swing round whenever the boat changed tack even though the wind had not shifted: the apparent direction of a flow depends on your own motion. Starlight is the wind; the Earth is the boat. He announced it in a letter to Halley, published in the Philosophical Transactions in 1729 — the first direct proof that the Earth moves, and a measurement of the speed of light into the bargain. He then spent twenty years on a second, smaller wobble, the $18.6$ -year nutation of the Earth's axis, refusing to publish until he had watched a complete cycle of the Moon's nodes; he announced it in 1748 and won the Copley Medal. Bradley never found his parallax. He found something better.

History

1725

— Bradley and Samuel Molyneux begin zenith-sector observations of γ Draconis at Kew, hunting for stellar parallax.

1728

— Bradley identifies the annual 20″ wobble as aberration: the finite speed of light combined with the Earth's orbital motion.

1729

— Bradley's account is published in the Philosophical Transactions, in a letter to Edmond Halley: the first direct proof that the Earth moves.

1818

— Fresnel's partial-drag coefficient 1 − 1/n² reconciles Arago's null result with a stationary, undragged ether.

1845

— George Gabriel Stokes proposes a fully dragged ether, and must assume the flow is irrotational to recover the aberration law.

1871

— George Biddell Airy fills a telescope with water and finds the aberration angle unchanged — as Fresnel's coefficient requires.

1879

— Maxwell writes to D. P. Todd that a terrestrial optical test of the Earth's motion is second order in v/c and looks hopeless; the letter is read to the Royal Society and printed in Nature in 1880.

Going further

Texts. Taylor and Wheeler, Spacetime Physics, 2nd ed., Ch. 1 · Rindler, Introduction to Special Relativity, 2nd ed., Ch. 1 · Whittaker, A History of the Theories of Aether and Electricity, Vol. I · Bradley, Philosophical Transactions 35, 637 (1729). The aberration law $\tan\alpha = v/c$ is derived from Galilean velocity composition; the relativistic correction enters only at $O(\beta^{2})$ and is negligible at $\beta = 10^{-4}$ . Key numerical results: the rad-to-arcsec conversion gives $\alpha = 20.50''$ and $2\alpha = 41.0''$ ; $c = v/\tan(20.4955'') = 2.999\times10^{8}\ \mathrm{m\,s^{-1}}$ ; $1\,\mathrm{AU}/c = 499\ \mathrm{s}$ ; the one-way speeds are $c \pm v$ ; the $7.3\ \mathrm{ps}$ / $3700$ -wavelength gap over $11\ \mathrm{m}$ ; and the round-trip expansion is $(2L/c)(1 + \beta^{2} + O(\beta^{4}))$ , with no linear term. Anchor values used throughout: $c = 2.998\times10^{8}\ \mathrm{m\,s^{-1}}$ , $v = 29.8\ \mathrm{km\,s^{-1}}$ , $\lambda_{\mathrm{Na}} = 589\ \mathrm{nm}$ , $L = 11\ \mathrm{m}$ , $1\ \mathrm{rad} = 206264.8''$ ; aberration constant $20.4955''$ ; Moon $0.52^\circ$ .

Builds on. The Luminiferous Ether · Galilean Transformations · Absolute Space and Absolute Time. Leads to. The Michelson–Morley Experiment · Lorentz–FitzGerald Contraction Hypothesis · The Constancy of the Speed of Light · Einstein's Two Postulates.

If aberration proves the Earth moves, doesn't it also prove the ether exists?

No, and this is the most important thing to take away. The only ingredients in the derivation are a finite light speed and a moving telescope — the word "ether" never enters. Aberration would appear in exactly the same amount if space were empty and light simply travelled at $c$ , which is what relativity says. What it does settle is a question within ether theory: it rules out the Earth carrying the nearby ether along, since a co-moving ether gives zero tilt. It constrains ether models; it is not evidence for them.

Why is the aberration angle the same for every star, when parallax is not?

Because aberration is not about geometry, it is about velocity. Parallax is a triangle in space: you view the star from two ends of a baseline, so the shift is the baseline over the distance, and nearer stars shift more. Aberration has no triangle in space at all — it is a triangle in velocity, made of the light's $c$ and the Earth's $v$ , and distance never appears. Every photon arriving here, from the nearest star or the most distant quasar, travels at $c$ when it arrives, so every one is tilted by the same $v/c$ . That distance-independence is how you tell the two apart in the data.

The ether wind is one part in ten thousand. Why was it so hard to see?

Because it lives in the one-way speed of light, and every practical optical measurement is a round trip. A one-way speed needs two clocks, at separated places, synchronised in advance to picoseconds — and you cannot synchronise them by sending a light pulse, because the one-way light speed is exactly what you are trying to determine. So experimenters had no choice but to send light out and back to a single clock. On that round trip the head wind and the tail wind cancel to first order, and the survivor is $\beta^{2} \approx 10^{-8}$ instead of $\beta \approx 10^{-4}$ . One extra factor of $\beta$ is the whole difference between trivial and heroic.

Doesn't the Earth's rotation, or the Sun's motion through the galaxy, also count?

Yes, and experimenters counted on it. The wind at any laboratory is the vector sum of everything the lab is doing: the Earth's rotation (up to $0.46\ \mathrm{km\,s^{-1}}$ at the equator), its orbit ( $29.8\ \mathrm{km\,s^{-1}}$ ), the Sun's motion through the galaxy. The orbital term dominates and reverses over six months, which is why experiments were repeated at different hours and seasons. Even if the Solar System were momentarily at rest in the ether, the Earth's orbit alone guarantees a wind of $29.8\ \mathrm{km\,s^{-1}}$ half a year later — no way to be unlucky twice.

Bradley was hunting parallax. How did he know his 20″ wobble wasn't it?

By its timing. Parallax follows the Earth's displacement, so a star should sit furthest from its mean position when the Earth is at the ends of the baseline perpendicular to the line of sight. Aberration follows the Earth's velocity, and velocity leads displacement by a quarter cycle. Bradley's γ Draconis reached its extremes three months from where parallax demanded — a ninety-degree phase error no instrumental care could explain. The size was wrong too: $20''$ dwarfs any real stellar parallax (the nearest star manages under $1''$ ), and the wobble appeared in every star he checked.

Does relativity get rid of aberration along with the ether?

No. Aberration survives, because it is a statement about directions, not speeds. Relativity insists every observer measures the light's speed to be $c$ , so the ether wind — an anisotropy of speed — is abolished outright. But the direction light arrives from still transforms between frames, and the angle agrees with $\tan\alpha = v/c$ to first order in $\beta$ ; corrections appear only at $O(\beta^{2})$ , far below anything Bradley could see. The observation stands, the number stands, and only the mechanism is replaced — the same fate as Fresnel's drag coefficient.