Relativistic Aberration of Light

Run through vertical rain and it slants toward you. Fly through starlight and the stars slide forward. This is aberration, and at the Earth's orbital speed of $29.8\ \mathrm{km\,s^{-1}}$ it is a shift of $20.5''$ — the number James Bradley measured in 1728, and whose story Ether Drift and the Absolute Rest Frame tells. That article ends with a triangle: $\tan\alpha = v/c$ . This one begins where the triangle breaks. Push $\beta = v/c$ toward $1$ and the shift is no longer a small tilt but a wholesale rearrangement of the sky: at $\beta = 0.9$ the entire forward hemisphere is squeezed into a cap of radius $25.84^\circ$ , about five per cent of the sky. Underneath it is a structure, and that is what this page is for. In the right coordinate the transformation of the sky is

\zeta \;\longmapsto\; e^{-\varphi}\,\zeta,

a pure dilation of the Riemann sphere by the boost's rapidity — a Möbius transformation. Möbius transformations are conformal and take circles to circles. So a relativistic observer's sky is distorted, wildly, but never sheared: every angle survives, and every circle stays a circle. A moving sphere therefore still photographs with a circular outline. That is the Terrell–Penrose effect, and it is a corollary of one line of complex analysis.

Key takeaways

Signature $(+,-,-,-)$ , timelike-positive. A photon is the null four-vector $k^\mu = \frac{\omega}{c}(1,\ \cos\theta,\ \sin\theta,\ 0)$ , $k^\mu k_\mu = 0$ . $\theta$ is the photon's propagation direction from $+\hat{x}$ — the telescope points the opposite way. Getting this wrong is how the sign gets lost.
Boost convention locked: $S$ = lab, $S'$ moves at $+v\hat{x}$ . Forward ( $S \to S'$ ): $u'_\parallel = (u_\parallel - v)/(1 - vu_\parallel/c^2)$ , $u'_\perp = u_\perp/[\gamma(1 - vu_\parallel/c^2)]$ ; the inverse is the reciprocity corollary ( $v \to -v$ ).
The aberration map. $\;\cos\theta' = \dfrac{\cos\theta - \beta}{1 - \beta\cos\theta}$ , $\;\sin\theta' = \dfrac{\sin\theta}{\gamma(1 - \beta\cos\theta)}$ . To first order, $\;\boxed{\theta' - \theta \approx +\beta\sin\theta}$ . (The base printed $\theta - \theta' \approx +\beta\sin\theta$ in one place and $\approx -\beta\sin\theta$ in another. Corrected below.)
The half-angle form (owned). $\;\boxed{\tan\dfrac{\theta'}{2} = \sqrt{\dfrac{1+\beta}{1-\beta}}\ \tan\dfrac{\theta}{2}}\;$ — exact at every angle and speed. The rational map becomes a multiplication.
The Möbius theorem (owned). Stereographic coordinate $\zeta = e^{i\phi}\cot(\theta/2)$ ; then $\zeta' = e^{-\varphi}\zeta$ , with $\varphi = \operatorname{artanh}\beta$ the rapidity. That is a Möbius map, so aberration is conformal and takes circles on the celestial sphere to circles. Its two fixed points are the apex and anti-apex — the $\pm c$ of Velocity Composition and the Speed-of-Light Invariant, drawn on the sky.
Half the sky lands in a cap of half-angle $\arccos\beta$ , solid angle exactly $2\pi(1-\beta)$ : $60^\circ$ at $\beta = 0.5$ , $25.84^\circ$ at $\beta = 0.9$ , $8.11^\circ$ at $\beta = 0.99$ .
Key idea: aberration is not a distortion of the sky but a dilation of it — multiplication by $e^{-\varphi}$ in the stereographic coordinate. That is why it is conformal, why circles survive, why boosts add rapidities, and why a moving sphere still looks like a sphere.

The core insight

The rational-looking map $\cos\theta' = (\cos\theta - \beta)/(1 - \beta\cos\theta)$ is a Möbius transformation in disguise. Find the coordinate in which it is a multiplication — the stereographic coordinate on the celestial sphere — and every hard property of aberration becomes a one-line property of $e^{-\varphi}$ .

Where this lives today

Every astrometric catalogue is aberration-corrected before it is a catalogue. Gaia measures positions to tens of microarcseconds; the aberration from the Earth's orbit is $20.5''$ — a million times larger. The map on this page is applied at full relativistic order to all two billion sources: at microarcsecond precision the $O(\beta^2)$ terms are not optional.

The cosmic microwave background is the cleanest aberration measurement ever made. The Solar System moves at $v = 369.82\ \mathrm{km\,s^{-1}}$ relative to the CMB rest frame, so $\beta = 1.23357\times10^{-3}$ . The Doppler part of the boosted $k^\mu$ gives the famous dipole, $\Delta T = T_0\beta = 2.7255\ \mathrm{K} \times 1.23357\times10^{-3} = \mathbf{3.3621\ mK}$ — exactly what Planck measures. The aberration part shifts all CMB structure toward the apex by up to $\beta = 254'' = 4.24'$ . Small — but Planck detected it in 2013, measuring our velocity from the geometry of the sky rather than its temperature. And a blazar jet pointed near our line of sight has its emission swept into the same forward cone: the geometric half of why they look so bright, the photometric half being Relativistic Beaming's.

The physical picture

You are flying at $\beta$ along $+\hat{x}$ . A star sits somewhere on the sky. Where do you see it? Two angles are in play, and confusing them is the commonest error in this subject — it is the error the base article made. $\theta$ is the direction the photon travels; the star is in the opposite direction, at $\pi - \theta$ , and that is where you point the telescope. Now boost. That photon has zero $x$ -velocity in the lab. In your frame it must acquire an $x$ -velocity of $-v$ — you are running into it — while its $y$ -velocity is squeezed to $c/\gamma$ by Transverse Velocity Transformation's rule. So the photon now moves backwards and sideways: $\theta' > 90^\circ$ . And since the star is where the photon came from, the star has moved forward, toward the direction you are flying. That is the phenomenon in one sentence: stars move toward the apex of your motion. The two statements are one statement, carrying opposite signs — which is exactly why the sign gets dropped.

Figure

THE SIGN, DRAWN — because this is the error the base article made. θ is the angle of the PHOTON'S PROPAGATION DIRECTION from +x̂ (blue arrows). The star is where the photon CAME FROM: you point the telescope at π − θ (grey). LEFT, lab frame S: a photon travels straight up, θ = 90°; the star is straight down. RIGHT, your frame S′ moving at +v x̂: the photon now also moves BACKWARD, so θ′ = 90° + β > 90° — the propagation direction rotates AWAY from the apex. But the star, being where the photon came from, rotates TOWARD the apex by the same β. Both facts are one fact. θ′ − θ ≈ +β sin θ, and the star's shift toward the apex is likewise +β sin θ. The base article printed θ − θ′ ≈ +β sin θ, which is the opposite sign

The core equations

A photon is a null four-vector. In signature $(+,-,-,-)$ , in the $xy$ -plane:

k^\mu = \frac{\omega}{c}\Big(1,\ \cos\theta,\ \sin\theta,\ 0\Big), \qquad k^\mu k_\mu = \frac{\omega^2}{c^2}\big(1 - \cos^2\theta - \sin^2\theta\big) = 0 \qquad (1)

where: $\omega$ — the photon's angular frequency in $S$ ; $\theta$ — its propagation direction from $+\hat{x}$ . Physical reading: $k\cdot k = 0$ says the photon moves at $c$ (The Minkowski Metric) — and being zero, it is zero in every frame. Everything here is a shadow of that one invariant. Boost forward to $S'$ and take the ratio of the spatial components to the time component:

\boxed{\;\cos\theta' = \frac{\cos\theta - \beta}{1 - \beta\cos\theta}\;}, \qquad \boxed{\;\sin\theta' = \frac{\sin\theta}{\gamma\,(1 - \beta\cos\theta)}\;} \qquad (2)

where: $\beta = v/c$ , $\gamma = (1-\beta^2)^{-1/2}$ . Physical reading: this is Transverse Velocity Transformation's rule applied to a velocity of magnitude $c$ — the $\gamma$ in the transverse part and the shared denominator are its two signatures. To first order, $\;\theta' - \theta \approx +\beta\sin\theta$ : the propagation direction tilts backward, so the source tilts forward, toward the apex. The same content, in the form that makes it structural:

\boxed{\;\tan\frac{\theta'}{2} \;=\; \sqrt{\frac{1+\beta}{1-\beta}}\;\tan\frac{\theta}{2}\;} \qquad (3)

where: exact — no expansion, no restriction on $\beta$ or $\theta$ . Physical reading: the half-tangent of the sky-angle is simply multiplied. The rational map (2) is a multiplication in the right coordinate — and that is the seed of the whole owned result. Finally, the coordinate that makes the multiplication obvious. Stereographically project the celestial sphere from the anti-apex onto the complex plane, $\zeta = e^{i\phi}\cot(\theta/2)$ . Then, with $\varphi = \operatorname{artanh}\beta$ the rapidity,

\boxed{\;\zeta' \;=\; e^{-\varphi}\,\zeta\;} \qquad (4)

where: $\phi$ — azimuth about the boost axis, unchanged by the boost. Physical reading: aberration is a real dilation of the Riemann sphere by the exponential of the rapidity. It is a Möbius transformation, and that one fact carries conformality, circles-to-circles, the composition law and the fixed points. (The photon's frequency transforms by the time component, $\omega' = \gamma\omega(1 - \beta\cos\theta)$ — the relativistic Doppler effect, which falls out of the same $k^\mu$ but belongs to The Relativistic Doppler Effect. Stated here, not derived.)

Deriving the structure

Derivation ATHE ABERRATION MAP — from one null four-vector, with the sign the base got wrong

Aberration is not a new postulate; it is the direction-part of the Lorentz transformation applied to something that moves at $c$ . So write the photon as a four-vector and boost it, rather than reasoning about triangles — triangles are where signs go to die, and the base article for this page lost one. The four-vector route is bookkeeping-proof: it gives the map at all angles and speeds, the first-order sign as a corollary rather than an input, and the Doppler factor free, as the time component.

Step 1 — the photon. From equation (1), $k^\mu = \frac{\omega}{c}(1, \cos\theta, \sin\theta, 0)$ with $k\cdot k = 0$ . Why: $\theta$ is defined once, here — the direction the photon goes, not the direction you look. Every sign below inherits this choice, and the base's contradiction traces to exactly this ambiguity.

Step 2 — boost it. The forward Lorentz transformation ( $S \to S'$ , with $S'$ moving at $+v\hat{x}$ ) acts on the components:

k'^0 = \gamma\big(k^0 - \beta k^1\big) = \frac{\gamma\omega}{c}\big(1 - \beta\cos\theta\big), \qquad k'^1 = \gamma\big(k^1 - \beta k^0\big) = \frac{\gamma\omega}{c}\big(\cos\theta - \beta\big),

k'^2 = k^2 = \frac{\omega}{c}\sin\theta, \qquad k'^3 = k^3 = 0.

Why: only the transverse component is left alone, and that asymmetry between $k^1$ and $k^2$ is the entire origin of aberration. If the boost rescaled both alike, directions would not change.

Step 3 — read off the angles. The photon in $S'$ still moves at $c$ : $k'\cdot k' = \frac{\gamma^2\omega^2}{c^2}[(1-\beta\cos\theta)^2 - (\cos\theta - \beta)^2] - \frac{\omega^2}{c^2}\sin^2\theta = 0$ ✓. So $\cos\theta' = k'^1/k'^0$ , $\sin\theta' = k'^2/k'^0$ :

\cos\theta' = \frac{\cos\theta - \beta}{1 - \beta\cos\theta}, \qquad \sin\theta' = \frac{\sin\theta}{\gamma(1 - \beta\cos\theta)}.

∎ Why: the denominator is common to both, because it is $k'^0$ — the same denominator that appears in every velocity transformation, and the one that becomes the Doppler factor. Aberration and Doppler are the angular and temporal shadows of one boosted null vector.

Step 4 — the first-order sign, which is the base's bug. Expand for $\beta \ll 1$ :

\cos\theta' = (\cos\theta - \beta)(1 + \beta\cos\theta) + O(\beta^2) = \cos\theta - \beta\big(1 - \cos^2\theta\big) + O(\beta^2) = \cos\theta - \beta\sin^2\theta + O(\beta^2).

Write $\theta' = \theta + \delta$ ; then $\cos\theta' \approx \cos\theta - \delta\sin\theta$ . Matching,

\boxed{\;\delta \;=\; \theta' - \theta \;\approx\; +\beta\sin\theta\;}

∎ Why: positive. The propagation direction swings away from the apex; equivalently the source swings toward it, by the same $\beta\sin\theta$ . The base's equation card printed $\theta - \theta' \approx +\beta\sin\theta$ and its limits list $\approx -\beta\sin\theta$ — it contradicted itself, and the card was the wrong half. At $\theta = 90^\circ$ , $\delta = \beta = 9.9365\times10^{-5}\ \mathrm{rad} = 20.4955''$ — Bradley's number, and the same in relativity as in the ether theory, because both agree at first order. Ether Drift and the Absolute Rest Frame owns that story.

Step 5 — the source angle. Let $\alpha = \pi - \theta$ be the angle of the star from the apex — the direction you point the telescope. Then $\cos\alpha = -\cos\theta$ , and (2) becomes

\boxed{\;\cos\alpha' = \frac{\cos\alpha + \beta}{1 + \beta\cos\alpha}\;}

Why: the sign of $\beta$ flips and the formula becomes the one an observer actually uses. Note $\cos\alpha' > \cos\alpha$ always: every star moves toward the apex, none away. No exceptions, no turning points — forced, as the next card shows, by the map's being a dilation. *Verified in verify_aq-sr-relativistic-aberration.py: $k\cdot k = 0 \Rightarrow k'\cdot k' = 0$ symbolically in signature $(+,-,-,-)$ ; equations (2) recovered from the boosted components; $\sin^2\theta' + \cos^2\theta' = 1$ identically; the first-order coefficient of $\theta' - \theta$ is $+\sin\theta$ (sympy series), and the base's $-\sin\theta$ is rejected; $\cos\alpha' \geq \cos\alpha$ for all $\alpha$ and all $0 < \beta < 1$ ; forward $\circ$ inverse $=$ identity.*

Go deeper · Derivation B · 3 minABERRATION IS A MOBIUS MAP OF THE SKY — hence conformal, hence circles map to circles

Derivation BABERRATION IS A MOBIUS MAP OF THE SKY — hence conformal, hence circles map to circles

Equation (2) is an unpleasant object: a rational function of a cosine, which resists composition and hides its own symmetries. The claim of this card is that it is the same map as a multiplication, written in bad coordinates — and that once the coordinate is changed, four separate facts about aberration collapse into one property of a single number. The right coordinate is the one complex analysts use for the sphere: stereographic projection. The restricted Lorentz group acts on the celestial sphere as the group of Möbius transformations, and a boost acts as a dilation.

Step 1 — the half-angle form. From (2), build $\tan(\theta'/2) = \sin\theta'/(1 + \cos\theta')$ . The denominator is

1 + \cos\theta' = \frac{(1 - \beta\cos\theta) + (\cos\theta - \beta)}{1 - \beta\cos\theta} = \frac{(1-\beta)(1 + \cos\theta)}{1 - \beta\cos\theta},

and the numerator is $\sin\theta/[\gamma(1-\beta\cos\theta)]$ . The denominators $1 - \beta\cos\theta$ cancel — the whole difficulty of the map cancels — leaving

\tan\frac{\theta'}{2} = \frac{\sin\theta}{\gamma(1-\beta)(1+\cos\theta)} = \frac{1}{\gamma(1-\beta)}\tan\frac{\theta}{2}.

Now $\gamma(1-\beta) = (1-\beta)/\sqrt{1-\beta^2} = \sqrt{(1-\beta)/(1+\beta)}$ , so

\boxed{\;\tan\frac{\theta'}{2} = \sqrt{\frac{1+\beta}{1-\beta}}\;\tan\frac{\theta}{2}\;}

∎ Why: the map is a multiplication. Every trace of the awkward denominator is gone — and the multiplier is familiar: $\sqrt{(1+\beta)/(1-\beta)} = e^{\varphi}$ , with $\varphi = \operatorname{artanh}\beta$ the rapidity of Rapidity and Hyperbolic Boosts.

Step 2 — put it on the Riemann sphere. Project stereographically from the anti-apex ( $\theta = \pi$ ) onto the complex plane:

\zeta \;=\; e^{i\phi}\cot\frac{\theta}{2} \;\in\; \mathbb{C}\cup\{\infty\},

with $\phi$ the azimuth about the boost axis. The boost leaves $k^2$ and $k^3$ untouched, so $\phi' = \phi$ ; and $\cot(\theta'/2) = e^{-\varphi}\cot(\theta/2)$ by Step 1. Hence

\boxed{\;\zeta' = e^{-\varphi}\,\zeta\;}

Why: a real dilation, of factor $e^{-\varphi} \in (0,1)$ . Rotations about the boost axis are $\zeta \mapsto e^{i\psi}\zeta$ ; boosts along it are $\zeta \mapsto e^{-\varphi}\zeta$ ; together, $\zeta \mapsto \lambda\zeta$ with $\lambda$ complex — and the general Lorentz element acts as the general Möbius map $\zeta \mapsto (a\zeta+b)/(c\zeta+d)$ . This is the isomorphism $\mathrm{SO}^+(1,3) \cong \mathrm{PSL}(2,\mathbb{C})$ , and the sky is where you can see it.

Step 3 — it is a Möbius transformation, so it is conformal. $\zeta \mapsto e^{-\varphi}\zeta$ is the Möbius map with matrix $\begin{pmatrix} e^{-\varphi/2} & 0 \\ 0 & e^{+\varphi/2}\end{pmatrix} \in \mathrm{SL}(2,\mathbb{C})$ , determinant $1$ . It is holomorphic with derivative $e^{-\varphi} \neq 0$ everywhere, hence conformal — it preserves angles, sense included. Stereographic projection is itself conformal, so the composite map on the sky is conformal. Why: the sky is never sheared. A small circular patch maps to a small circular patch: changed in size, never in shape. An observer at $\beta = 0.99$ sees a violently rearranged sky in which every local angle is nevertheless exactly right.

Step 4 — circles map to circles. A dilation of the plane manifestly takes circles to circles and lines to lines. (Generally: every Möbius map takes the family of circles-and-lines to itself.) And stereographic projection takes circles on the sphere to circles-and-lines in the plane, and back. Composing the three:

\text{circle on the sky} \;\xrightarrow{\ \text{stereo}\ }\; \text{circle or line} \;\xrightarrow{\ \times e^{-\varphi}\ }\; \text{circle or line} \;\xrightarrow{\ \text{stereo}^{-1}\ }\; \text{circle on the sky}.

∎ Why: this is the owned result. The outline of a sphere, seen from anywhere, is a circle on the observer's sky, and aberration maps that circle to another circle. So a relativistically moving sphere still presents a circular outline — it does not photograph as an ellipse, despite being Lorentz-contracted. That is the Terrell–Penrose effect (Penrose 1959; Terrell 1959), whose geometric core this card has just proved. The full effect — including the apparent rotation of a non-spherical body, which needs the light-travel-time delays too — belongs to its own article and is not authored here.

Step 5 — the fixed points, and the composition law. A dilation has exactly two fixed points on the Riemann sphere: $\zeta = 0$ and $\zeta = \infty$ — that is, $\theta = \pi$ and $\theta = 0$ , the anti-apex and the apex. Everything else flows from one to the other. And two boosts compose by multiplying their factors:

e^{-\varphi_2}\cdot e^{-\varphi_1} = e^{-(\varphi_1 + \varphi_2)}.

Why: two facts from earlier in this chapter, reappearing as geometry. The fixed points are the two directions along the boost axis — the only two directions on the sky that do not move — and they are $\pm c$ , the fixed points that Velocity Composition and the Speed-of-Light Invariant proves are unique. Here they are, as two points on a sphere. And the composition rule is rapidity additivity, the group $(\mathbb{R},+)$ . The whole chapter is one theorem seen from different angles. *Verified in verify_aq-sr-relativistic-aberration.py: the half-angle form (3) proved symbolically from (2) and confirmed numerically on a grid of 40 speeds $\times$ 60 angles to $10^{-12}$ ; $\sqrt{(1+\beta)/(1-\beta)} = e^{\operatorname{artanh}\beta}$ symbolically; $\zeta' = e^{-\varphi}\zeta$ ; the $\mathrm{SL}(2,\mathbb{C})$ matrix has determinant $1$ ; circles map to circles — 12 random circles on the sky, each sampled at 400 points, boosted, and refitted, with residuals $< 10^{-12}$ ; conformality checked by transporting two intersecting sky-curves through the map and comparing the crossing angle to $10^{-10}$ ; the fixed points are exactly $\theta = 0, \pi$ ; and $\varphi$ -additivity under composition.*

Limiting cases

Case	Result	Meaning
$\beta \ll 1$	$\theta' - \theta \approx +\beta\sin\theta$	the same as the classical answer — which is why aberration alone cannot test relativity
$\theta = 0$ or $\pi$	$\theta' = \theta$	the apex and anti-apex are the fixed points; a star dead ahead stays dead ahead
$\theta = \pi/2$	$\cos\theta' = -\beta$	a transverse photon acquires a backward component of exactly $-\beta$
$\alpha = \pi/2$ (star abeam)	$\cos\alpha' = \beta$	half the sky lands in a cap of half-angle $\arccos\beta$ , solid angle $2\pi(1-\beta)$
$\beta \to 1$	$\arccos\beta \to 1/\gamma$	the forward cap closes down; the intensity consequence is relativistic beaming, owned elsewhere
$c \to \infty$	$\theta' \to \theta$	no aberration: the effect is $O(v/c)$ and vanishes classically
two boosts	$\varphi_1 + \varphi_2$	dilation factors multiply, rapidities add

Common trapThe sign, and the two angles — this is where the base article broke

The base article contradicted itself: its equation card gave $\Delta\theta = \theta - \theta' \approx +\beta\sin\theta$ , its limits list gave $\Delta\theta \approx -\beta\sin\theta$ . The card is the wrong half. The fix is not a sign but a definition, stated once and held:

$\theta$ = the photon's propagation direction from $+\hat{x}$ . Then $\;\theta' - \theta \approx +\beta\sin\theta$ : it tilts away from the apex.
$\alpha = \pi - \theta$ = the source direction, where you point the telescope. Then $\;\alpha' - \alpha \approx -\beta\sin\alpha$ : it tilts toward the apex.

Both are correct — the same statement about the same photon. An article that switches silently between them prints a sign error, and one did. Always say which angle you mean.

Common trapWhat relativistic aberration is NOT

It is not parallax. Parallax depends on where the observer is, and on the star's distance; aberration depends on where the observer is going, and is identical for every star. They are $90^\circ$ out of phase over a year — which is how Bradley knew he had found the wrong thing.
It is not the Doppler effect, though it is its twin: both come from boosting the same $k^\mu$ . Aberration is the spatial part, Doppler ( $\omega' = \gamma\omega(1-\beta\cos\theta)$ ) the time part, and The Relativistic Doppler Effect owns the second.
It is not relativistic beaming. Aberration says where the light goes; beaming says how bright it is.
It is not the Thomas–Wigner rotation. That turns frame axes, needs two non-collinear boosts, and needs no light at all.
It does not require an ether, and never did. The first-order law is the same either way — precisely the trouble it caused for two centuries.

Worked examples

1 · Bradley's $20.5''$ , and why relativity did not change it

The Earth orbits at $29.79\ \mathrm{km\,s^{-1}}$ . Compute the aberration of a star at the ecliptic pole, exactly — then ask by how much the relativistic answer differs from the classical one.

Step 1 — identify. A star at the ecliptic pole is abeam: the photon arrives perpendicular to the Earth's velocity, $\theta = 90^\circ$ . With $\kappa = 20.49552''$ , the IAU constant of aberration at J2000,

\beta = \frac{\kappa}{206264.806''\ \mathrm{rad}^{-1}} = 9.93651\times10^{-5}, \qquad v = \beta c = 29.789\ \mathrm{km\,s^{-1}}.

Step 2 — apply the map. Exactly, from (2) with $\cos\theta = 0$ :

\cos\theta' = -\beta \qquad\Longrightarrow\qquad \theta' = \arccos(-\beta) = 90^\circ + 0.0056932^\circ.

Step 3 — the boundary number. The shift is $\theta' - 90^\circ = \mathbf{20.49552''}$ , to every digit given. The first-order formula $\delta = \beta\sin\theta = \beta$ gives $\beta \times 206264.806 = \mathbf{20.49552''}$ — the same number. The two first differ in the ninth significant figure: the correction is $\beta^3/6 = 3.4\times10^{-8}$ arcseconds.

Step 4 — cash it out. Bradley measured about $20.2''$ in 1728, low by $1.4\%$ , with a homemade zenith sector. And here is the sting: the relativistic prediction is identical to the classical one, at any precision reachable for the next two centuries. The theories part company only at $O(\beta^2)$ — which is why the ether survived as long as it did, and the subject of the portrait card below. ( $\gamma$ Draconis sits at ecliptic latitude $74.9^\circ$ , so its annual aberration ellipse is nearly a circle — semi-axes $20.50'' \times 19.79''$ . Bradley chose well.)

2 · A starship at $\beta = 0.9$ — where does the sky go?

You accelerate to $0.9c$ . What happens to the stars?

Step 1 — identify. $\gamma = 2.294$ , $\varphi = \operatorname{artanh}(0.9) = 1.4722$ , and the dilation factor is

e^{-\varphi} = \sqrt{\frac{1-\beta}{1+\beta}} = \sqrt{\frac{0.1}{1.9}} = \mathbf{0.229416}.

Every star's stereographic coordinate is multiplied by $0.229416$ . That is the entire answer; the rest is reading it out.

Step 2 — the star that was abeam. A star at $\alpha = 90^\circ$ , square to your left, moves to

\tan\frac{\alpha'}{2} = \sqrt{\frac{1-\beta}{1+\beta}}\tan 45^\circ = 0.229416 \qquad\Longrightarrow\qquad \alpha' = \mathbf{25.842^\circ},

equivalently $\cos\alpha' = \beta = 0.9$ exactly. The star that was at right angles is now $26^\circ$ from dead ahead.

Step 3 — the boundary number. That star was on the boundary of the forward hemisphere. So the entire forward half of the sky — $2\pi$ steradians, half of everything — is now inside a cap of half-angle $25.842^\circ$ , of solid angle

\Omega' = 2\pi\big(1 - \cos\alpha'\big) = 2\pi(1-\beta) = 0.6283\ \mathrm{sr} = \mathbf{5.00\%\ of\ the\ sky} = 2063\ \text{square degrees},

a compression of exactly $1/(1-\beta) = \mathbf{10\times}$ . Meanwhile the rear is stretched: a star $1^\circ$ off the anti-apex swings out to $4.36^\circ$ off — the same map running the other way, as it must, because a dilation shrinks toward one fixed point exactly as it expands away from the other.

Step 4 — cash it out. Push harder and the sky collapses forward. At $\beta = 0.99$ the forward hemisphere fits inside $8.11^\circ$ ( $0.5\%$ of the sky); at the LHC's $\gamma = 7460$ , inside $\arccos\beta = 1.34\times10^{-4}\ \mathrm{rad} = \mathbf{27.6''}$ — half the universe compressed into an angle the size of Bradley's $20.5''$ . One formula spans both. And through all of it every circle on the sky is still a circle: the sky is dilated, never sheared.

Figure

THE PUNCHLINE — half the sky, squeezed forward. The SAME star field (grey dots, isotropic) as seen at β = 0, 0.5, 0.9, 0.99, plotted on the observer's sky with the direction of motion (the APEX) at the centre. The orange ring is the image of the CELESTIAL EQUATOR of the motion — the set of stars that were at 90°, square abeam. Its radius is arccos β: 90° at rest, 60.00° at β = 0.5, 25.84° at β = 0.9, 8.11° at β = 0.99. EVERYTHING INSIDE THAT RING WAS THE FORWARD HALF OF THE SKY — 2π steradians, half of all there is to see — compressed into a solid angle of exactly 2π(1−β), i.e. 25%, 5%, and 0.5% of the sky. The apex at the centre and the anti-apex at the rim are the two FIXED POINTS of the map, and they are the ±c of the composition law. Note that the ring stays a perfect CIRCLE at every speed: the map is conformal

Figure

THE OWNED RESULT — aberration is a DILATION of the Riemann sphere. Stereographically project the celestial sphere from the anti-apex onto the plane: a star at sky-angle θ, azimuth φ, becomes the complex number ζ = e^{iφ} cot(θ/2). LEFT: the celestial sphere with the apex (θ = 0) at the top and the anti-apex (θ = π) at the projection point. RIGHT: the plane. A boost of rapidity φ_r = artanh β acts as ζ ↦ e^{−φ_r} ζ — a pure real DILATION, shrinking everything toward the origin (which is the ANTI-apex... and expanding away from ∞, which is the APEX). Those two points are the ONLY fixed points, and they are the ±c of the velocity-composition law. Because a dilation is a Möbius transformation, it is CONFORMAL and takes CIRCLES TO CIRCLES: the blue circle maps to the smaller blue circle, still a circle. That is why a Lorentz-contracted sphere still photographs with a circular outline — the Terrell-Penrose effect

Physics of the portrait

Origins · 2 minStellar aberration and how this was discovered

Stellar aberration

1728–1871 · Bradley's zenith sector · Airy's water telescope · the first-order effect that could not decide anything

This card has no physicist on it, because the discovery belongs to a phenomenon — and its owner, James Bradley, already has his own page. Ether Drift and the Absolute Rest Frame tells his story: the zenith sector at Wanstead, the star $\gamma$ Draconis chosen because it passed almost through his zenith, the hunt for parallax from December 1725, and the wrong wobble — about $20''$ , ninety degrees out of phase with parallax, identical for every star. He announced it to the Royal Society in January 1729, published as a letter to Edmond Halley in Philosophical Transactions 35, 637 (1729), with a value within $1.5\%$ of today's $20.49552''$ . What belongs here is what happened next — a lesson about the limits of a first-order measurement. Aberration proves the Earth moves. It does not prove there is no ether, because to first order in $\beta$ the relativistic map and the classical velocity triangle give the same answer — and so did Fresnel's partially dragged ether, and Stokes's fully dragged one. So in 1871 George Biddell Airy did the experiment that should have settled it: he filled a telescope with water, slowing the light inside. If the aberration were produced inside the tube, the angle should change. It did not. A phenomenon first order in $v/c$ is one almost any theory can accommodate, and for 180 years aberration accommodated all of them. The question was decided only when the argument moved to second order, to Michelson and Morley. Aberration is where the trail starts — and the reason it is so long. The relativistic law on this page is Einstein's, from §7 of Zur Elektrodynamik bewegter Körper (Annalen der Physik 17, 891, 1905), where the Doppler shift and the aberration angle drop out of the same transformed wave four-vector, exactly as here.

History

1728

— James Bradley, hunting stellar parallax with a zenith sector at Wanstead, instead finds an annual displacement of about 20 arcseconds in gamma Draconis, the same for every star and ninety degrees out of phase with parallax, and identifies it as an effect of the finite speed of light combined with the Earth's motion.

1729

— Bradley's discovery is published as a letter to Edmond Halley in Philosophical Transactions 35, 637; his constant of aberration is within about 1.5 per cent of the modern 20.49552 arcseconds.

1810

— Francois Arago looks for a change in the refraction of starlight with the Earth's motion and finds none: the first experimental trouble for a simple ether.

1871

— George Biddell Airy fills a telescope with water, slowing the light inside it, and finds the aberration angle unchanged. First-order optics cannot separate the competing ether theories.

1905

— Albert Einstein derives the aberration law from the Lorentz transformation of the light wave four-vector, in section 7 of Zur Elektrodynamik bewegter Koerper, Annalen der Physik 17, 891, alongside the Doppler formula.

1959

— Roger Penrose (Mathematical Proceedings of the Cambridge Philosophical Society 55, 137) and James Terrell (Physical Review 116, 1041) independently show that a rapidly moving sphere still photographs with a circular outline, and that a moving object looks rotated rather than contracted.

2013

— The Planck collaboration measures the aberration and modulation of the cosmic microwave background caused by the Solar System's motion, recovering a velocity of 384 km/s with an uncertainty of 78 km/s statistical and 115 km/s systematic.

Going further

Texts. Rindler, Introduction to Special Relativity, 2nd ed., §§4, 12 · Jackson, Classical Electrodynamics, 3rd ed., §11.3 · Penrose & Rindler, Spinors and Space-Time, Vol. 1, §1.2 — the celestial sphere as the Riemann sphere, the source for the owned result here. Primary: Einstein, Ann. Phys. 17, 891 (1905), §7; Penrose, Math. Proc. Camb. Phil. Soc. 55, 137 (1959); Terrell, Phys. Rev. 116, 1041 (1959).

Provenance. Signature $(+,-,-,-)$ ; boost convention as in Transverse Velocity Transformation. $\theta$ is the photon's propagation direction, $\alpha = \pi - \theta$ the source direction. One base defect found and repaired: the base gave $\Delta\theta = \theta - \theta' \approx +\beta\sin\theta$ in an equation card and $\Delta\theta \approx -\beta\sin\theta$ in its limits list — a self-contradiction, of which the card was the wrong half. The correct relation, $\theta' - \theta \approx +\beta\sin\theta$ , is derived above from the boosted null four-vector. Every number here is recomputed in verify_aq-sr-relativistic-aberration.py, including circles-to-circles, checked by boosting twelve random sky-circles and refitting them.

Builds on. Transverse Velocity Transformation · Velocity Composition and the Speed-of-Light Invariant · Rapidity and the Additivity of Velocities · The Minkowski Metric · Ether Drift and the Absolute Rest Frame. Leads to. Relativistic Beaming · The Relativistic Doppler Effect · the Terrell–Penrose rotation.

If the stars bunch forward, does the sky get brighter ahead? Is that the same thing?

The bunching is aberration; the brightening is beaming, and they are different statements that point the same way. Aberration maps directions — the photon that was at $\theta$ is now at $\theta'$ — and says nothing about the energy it carries. Beaming is about intensity, and it is the stronger effect: the specific intensity picks up the Doppler factor to the fourth power, so the forward sky is not merely crowded but ferociously bright. This page owns the geometry; Relativistic Beaming owns the photometry. And do not quote the $1/\gamma$ "headlight cone" as an aberration result — it is a beaming result. The exact aberration statement is $\arccos\beta$ .

Why is the half-angle form so much nicer than the cosine form? Is that a coincidence?

Not a coincidence, and the reason is deep. The half-tangent $\tan(\theta/2)$ is the stereographic coordinate on the sphere. The Lorentz group acts on the celestial sphere as the group of Möbius transformations of that plane, and Möbius maps are linear-fractional: in the stereographic coordinate they are as simple as they can be. A boost — the simplest Lorentz transformation — becomes the simplest Möbius map, a dilation. The cosine form is the same map in a coordinate with no reason to be adapted to the group. The niceness of the half-angle form is the group showing through.

Does aberration prove the Earth moves — and does it disprove the ether?

The first, yes; the second, no — and the gap between those answers is two centuries of physics. Bradley's $20''$ shows unambiguously that the Earth is moving. But the effect is first order in $\beta$ , and every candidate theory reproduces the first-order answer; Airy's 1871 water-filled telescope tested exactly this and found no change. The theories part company only at $O(\beta^2)$ — which is why the ether had to be killed by Michelson–Morley and not by Bradley. Ether Drift and the Absolute Rest Frame owns that argument.

Is the "conformal" claim just aesthetics, or does it say something physical?

It is physical, and testable. Conformality means the map preserves angles: take two curves that cross on the sky — a filament of nebulosity crossing another, the two edges of a jet — and the angle at which they cross is the same for every inertial observer, however fast. Distances on the sky are frame-dependent, wildly so; areas are frame-dependent; the angle at a crossing point is not. And circles are frame-independent — the round outline of a sphere is round for everybody.

What are the fixed points of the aberration map, and why should I care?

Exactly two: the apex ( $\theta = 0$ ) and the anti-apex ( $\theta = \pi$ ). A star in either direction does not move however hard you boost; every other star flows from the anti-apex toward the apex. In the stereographic picture these are $\zeta = 0$ and $\zeta = \infty$ — the two fixed points any dilation must have. And they are not new: they are the $\pm c$ of the velocity-composition law, whose uniqueness Velocity Composition and the Speed-of-Light Invariant proves. A photon from dead ahead has velocity $-c\hat{x}$ in every frame, and $-c \oplus v = -c$ . The fixed points of an abstract algebraic map are two visible points on the sky.

Two boosts in the same direction — do the aberration angles add?

The angles do not; the rapidities do. Aberration is multiplication by $e^{-\varphi}$ in the stereographic coordinate, so two collinear boosts multiply the factors: $e^{-\varphi_2}e^{-\varphi_1} = e^{-(\varphi_1+\varphi_2)}$ . The rapidities add — Rapidity and the Additivity of Velocities' result, arriving here as an obvious property of exponentials. Angles, being $\arctan$ s of the coordinate, do not. Velocity is the wrong parameter; rapidity is the right one.

Relativistic Aberration of Light

Where this lives today

The physical picture

The core equations

Deriving the structure

Limiting cases

Worked examples

1 · Bradley's $20.5''$ , and why relativity did not change it

2 · A starship at $\beta = 0.9$ — where does the sky go?

Physics of the portrait

Going further

Before this

Where this leads

Relativistic Aberration of Light

Where this lives today

The physical picture

The core equations

Deriving the structure

Limiting cases

Worked examples

1 · Bradley's 20.5′′20.5''20.5′′, and why relativity did not change it

2 · A starship at β=0.9\beta = 0.9β=0.9 — where does the sky go?

Physics of the portrait

Going further

Before this

Where this leads

1 · Bradley's $20.5''$ , and why relativity did not change it

2 · A starship at $\beta = 0.9$ — where does the sky go?