The Photon Ring

The light that came back the long way

Point a flashlight just past a black hole and most of the beam sails by, bent by a degree or two. Aim a little closer and the deflection grows — five degrees, twenty, ninety. Closer still and something strange happens: the ray bends so hard it loops behind the hole and comes back out toward you. Closer yet and it loops twice, three times, winding around the dark sphere like thread on a spool before finally peeling off. There is a precise critical distance — the photon sphere — where a ray would, in principle, circle forever. Real rays never quite reach it; they spiral close, wind some number of half-turns, and escape.

Every one of those wound-up rays lands on your retina (or your radio dish) in almost exactly the same place on the sky: a thin, bright circle hugging the edge of the black hole's dark silhouette. That circle is the photon ring. It is not made of any particular object — it is an image of the entire luminous universe around the hole, wrapped and stacked and squeezed into a line by gravity. Light from the accretion disk, from background stars, from the far side of the disk you could never otherwise see, all gets folded onto the same ring because they all took the same kind of looping path. The photon ring is, quite literally, light that came back to you the long way around.

The photon sphere and the unstable orbit

The physics begins with null geodesics — the paths light follows through curved spacetime. Around a non-spinning (Schwarzschild) black hole of mass M, there is exactly one radius at which gravity bends a light ray into a closed circular orbit:

r_ph = 3 GM/c²  =  1.5 × r_s    (Schwarzschild photon sphere)

where r_s = 2 GM/c² is the Schwarzschild radius (the event horizon). The photon sphere sits at 1.5 horizon radii. A photon placed exactly there with exactly the right direction would circle the hole indefinitely — but the orbit is unstable. Nudge the photon inward and it spirals into the horizon; nudge it outward and it spirals away to infinity. Like a pencil balanced on its tip, the closer you tune to the critical condition, the longer it lingers before falling off — and the number of loops it completes grows without bound.

The quantity an observer actually controls is the impact parameter b — roughly, how far off-axis the ray would have passed had there been no gravity. For Schwarzschild there is a critical value

b_c = 3√3 · GM/c²  ≈  5.196 GM/c²

Rays with b slightly larger than b_c swing around the hole and escape; rays with b slightly smaller plunge through the horizon. Rays tuned ever closer to b_c from above wind ever more times. The image of the sky for an observer is therefore organised by how many half-orbits n a ray executes: n = 0 (direct light), n = 1 (a half-turn behind the hole), n = 2 (a full turn), and so on. The photon ring is the superposition of all the n ≥ 1 images stacked at the edge of the shadow.

Why every sub-ring is ~535 times thinner

The unstable nature of the photon orbit has a beautiful quantitative consequence. The rate at which nearby rays diverge from the critical orbit is set by a Lyapunov exponent γ. For Schwarzschild, the divergence per full orbit (2π of winding) is governed by exactly

γ = 2π    (Schwarzschild, per full orbit in winding angle)

Concretely, the range of impact parameters Δb that produces an image shrinks geometrically with the winding: each additional full loop multiplies the demagnification by

Δb  ∝  e^(−γ)  =  e^(−2π)
e^(2π)  ≈  535.49    (per full orbit; e^π ≈ 23 per half-orbit)

So the band of sky directions feeding rays that loop one extra full turn is about 535 times narrower; loop another full turn and it is 535 times narrower again (a factor of e^π ≈ 23 for each successive half-orbit sub-ring). Because brightness is conserved along a ray (specific intensity I/ν³ is invariant), each thinner ring is correspondingly dimmer in integrated flux but identical in surface brightness. The rings pile up exponentially onto a limiting curve — the critical curve — which is the n → ∞ image of the photon sphere itself. The first ring is a few microarcseconds wide; the second is tens of nanoarcseconds; the third is utterly unresolvable but mathematically still there. This exponential self-similarity is the photon ring's defining signature, derived rigorously by Johnson, Gralla, Holz and collaborators in 2019–2020.

The shadow, the diameter, and what sets it

Inside the critical curve lies the black-hole shadow: the region of the image from which no light reaches the observer, because every ray aimed there terminates on the horizon. For a Schwarzschild hole the shadow is a perfect disk whose angular diameter, seen from distance D, is

θ_shadow = 2 b_c / D = 2 · 3√3 · GM/(c²D) = 2√27 · GM/(c²D)
        ≈ 10.39 · GM/(c²D)

The crucial point is that the diameter is the gravitational radius GM/c² multiplied by a pure number (≈ 10.4 for Schwarzschild). The shadow is appreciably larger than the horizon itself — about 2.6 times the horizon diameter — because gravitational lensing magnifies the silhouette. The photon ring traces the edge of this shadow. Its diameter therefore measures the mass directly, with almost no dependence on the unknown details of the glowing plasma. That theoretical cleanliness is the whole reason astronomers care.

The key numbers for real black holes

Two black holes are large enough on the sky to image at horizon scale: M87*, the giant in the galaxy Messier 87, and Sgr A*, the one at the centre of our own Milky Way.

Property	M87*	Sgr A*
Mass	≈ 6.5 × 10⁹ M☉	≈ 4.3 × 10⁶ M☉
Distance	≈ 16.8 Mpc (55 Mly)	≈ 8.15 kpc (26,700 ly)
Gravitational radius GM/c²	≈ 3.8 µas on sky	≈ 5.0 µas on sky
Predicted shadow diameter	≈ 40 µas	≈ 52 µas
EHT measured ring diameter	42 ± 3 µas (2019)	51.8 ± 2.3 µas (2022)
Schwarzschild radius (physical)	≈ 1.9 × 10¹⁰ km (≈ 128 AU)	≈ 1.3 × 10⁷ km (≈ 0.08 AU)
Dynamical (light-crossing) time	days–weeks	minutes

Sgr A* is about 1,500 times less massive than M87* but roughly 2,000 times closer, so the two subtend almost the same angular size — both near 50 µas. The catch is timescale: Sgr A*'s gas orbits in minutes, so its image flickers within a single observation, while M87*'s structure is stable for days. That is why M87* yielded the first clean image (10 April 2019) and Sgr A* took three more years (12 May 2022) of painstaking, motion-corrected reconstruction.

How you actually measure it

The photon ring is observed by very-long-baseline interferometry (VLBI). The angular resolution of a telescope of aperture B observing at wavelength λ is θ ≈ λ/B. To resolve 40 µas at λ = 1.3 mm you need B ≈ λ/θ ≈ 9,000 km — an aperture the size of the Earth. The Event Horizon Telescope synthesises exactly that by combining radio dishes from Hawaii to Spain to the South Pole, recording with atomic-clock timing and correlating the data after the fact. The result is the now-iconic orange ring.

But that bright orange ring is not the pure photon ring. It is dominated by the direct (n = 0) image of the hot accretion flow, with the unresolved n = 1 photon ring riding on top of it; together they blur into a feature roughly 10 µas thick. The clean photon ring — the narrow n ≥ 1 stack — is buried inside. Isolating it requires a few additional tricks: pushing baselines longer (so the interferometric "visibility" on long baselines is dominated by the sharp ring, which decays slowly with baseline length while the fuzzy direct emission decays fast), and exploiting the ring's universal shape. Broderick and collaborators reported in 2022 a feature in the M87* data consistent with the n = 1 ring, and the proposed Black Hole Explorer (BHEX) — a radio dish in space forming baselines longer than Earth's diameter — is designed expressly to detect and measure the photon ring's diameter to ~1% precision.

Worked example: M87* on the sky

Let us compute M87*'s shadow size from first principles. Take M = 6.5 × 10⁹ M☉ and D = 16.8 Mpc.

Step 1 — the gravitational radius in length. With GM☉/c² = 1.477 km,

GM/c² = 6.5 × 10⁹ × 1.477 km ≈ 9.6 × 10⁹ km ≈ 64 AU

Step 2 — the angular gravitational radius. Convert the distance: D = 16.8 Mpc = 16.8 × 3.086 × 10¹⁹ km ≈ 5.18 × 10²⁰ km. Then

(GM/c²)/D = 9.6 × 10⁹ / 5.18 × 10²⁰ ≈ 1.85 × 10⁻¹¹ rad
         = 1.85 × 10⁻¹¹ × (206,265 × 10⁶ µas/rad) ≈ 3.8 µas

Step 3 — the shadow diameter. Multiply by the Schwarzschild factor 2√27 ≈ 10.39:

θ_shadow = 10.39 × 3.8 µas ≈ 39.5 µas ≈ 40 µas

This matches the EHT-measured ring diameter of 42 ± 3 µas remarkably well — the small excess is exactly what you expect because the bright emission ring sits slightly outside the critical curve. Run the calculation in reverse and the measured ring diameter delivers the mass: this is how the EHT confirmed M87*'s mass at 6.5 × 10⁹ M☉, in agreement with the independent stellar-dynamics estimate and settling a decades-old factor-of-two dispute with the gas-dynamics value.

Spin, frame dragging, and the D-shaped curve

A real astrophysical black hole spins, and spin distorts the ring. For a rotating (Kerr) black hole, frame dragging — the wholesale twisting of spacetime by the hole's rotation — breaks the spherical symmetry of the photon sphere. Co-rotating (prograde) photons can hold a circular orbit closer to the hole, down to r = GM/c² for a maximally spinning hole, while counter-rotating (retrograde) photons are pushed out to r = 4 GM/c². The single photon sphere becomes a photon shell spanning a range of radii.

Projected onto the sky, the critical curve is no longer a centred circle. It shifts sideways (the approaching, frame-dragged side is brighter and displaced) and flattens on one edge into a characteristic "D" shape. The size, displacement and asymmetry of this curve encode both the spin magnitude a and the observer's inclination. The Lyapunov exponent γ — the 2π that gave us the factor 535 — also becomes a function of viewing angle and spin, so the spacing of the sub-rings carries spin information too. Measuring this is the long game: the diameter gives mass, and the shape gives spin, completing a direct readout of the Kerr metric's two parameters and a sharp test of the no-hair theorem.

Discovery, names, and key dates

• 1916 — Schwarzschild. Karl Schwarzschild solves Einstein's field equations for a point mass, implicitly fixing the photon sphere at 3 GM/c² (though he did not name it).
• 1959 — Darwin. Charles Galton Darwin (grandson of the naturalist) works out strong gravitational lensing near a Schwarzschild hole and finds the geometric series of looping images, anticipating the exponential ring structure.
• 1973 — Bardeen. James Bardeen computes the appearance of a Kerr black hole against a bright background, producing the first calculation of the asymmetric "shadow" and its D-shaped boundary.
• 1979 — Luminet. Jean-Pierre Luminet publishes the first realistic simulated image of a black hole with an accretion disk, showing the lensed top-and-bottom of the disk and the bright photon ring rim — decades before any observation.
• 2019 — EHT first image. On 10 April 2019 the Event Horizon Telescope collaboration releases the first image of M87*, a 42 µas ring, confirming the shadow's predicted size.
• 2019–2020 — the photon-ring theory. Michael Johnson, Samuel Gralla, Daniel Holz, Alexandru Lupsasca and collaborators formalise the n-indexed sub-ring decomposition, the e^(2π) demagnification, and the interferometric strategy to isolate the ring.
• 2022 — Sgr A* image. On 12 May 2022 the EHT releases the image of our own Galaxy's black hole, a 51.8 µas ring, despite its minute-scale variability.
• 2020s — BHEX and space VLBI. The proposed Black Hole Explorer mission targets a clean detection of the n = 1 photon ring from a space-based baseline.

Common misconceptions and subtleties

• The photon ring is not the accretion disk. The bright ring in the EHT images is mostly direct light from glowing gas. The true photon ring is a thin, gas-independent feature underneath it. They overlap on the sky but are physically distinct — one is emission, the other is geometry.
• The shadow is bigger than the event horizon. A frequent error is to call the dark disk "the event horizon." The shadow's diameter is ≈ 10.4 GM/c², about 2.6 times the horizon's diameter (2 r_s = 4 GM/c²). Lensing magnifies the silhouette; you are seeing the projection of the photon sphere, not the horizon itself.
• Photons do not "orbit forever" in the ring you see. The closed circular orbit at the photon sphere is unstable, so no real escaping photon sits on it. The ring is built from photons that came arbitrarily close and then peeled off after a finite number of half-loops.
• The ring is not perfectly circular for a spinning hole. Frame dragging makes the Kerr critical curve off-centre and flattened. For M87*'s low inclination the deviation from a circle is only a few percent, which is why the first image looked round — but the asymmetry is real and is the target for spin measurements.
• Higher-order rings are not just "fainter copies." Each sub-ring is a remapping of the entire sky, including parts hidden behind the hole, demagnified by ~535 and rotated. The n = 1 ring shows the back of the disk; the n = 2 ring shows the front again, and so on — an infinite nested series of universe-wide images.