Summary

In 2019 and 2022 the Event Horizon Telescope released images of two supermassive black-hole candidates. Both showed bright, asymmetric rings enclosing central brightness depressions of the scale expected for black holes. The ring diameters are measured to 4–7%, and for Sgr A*, whose mass is known independently from stellar orbits, the inferred shadow agrees with general relativity to within about 10%. That agreement is what makes them scientifically powerful. The ring is not merely beautiful; it tests a model for near-horizon emission and spacetime geometry.

The more interesting question, the one I want to spend time on here, is: what would a wormhole look like? For particular traversable-wormhole metrics and particular lighting, radically different. The images are thus evidence against some alternatives, but not a geometry-independent ruling-out of every wormhole.

The Images

The Event Horizon Telescope is a planet-scale interferometer: radio dishes from Hawaii to the South Pole, each with its own hydrogen-maser clock, their recordings combined afterwards. In 2017 its baselines reached 10,700 km, nearly the diameter of Earth. At a wavelength of 1.3 mm this gives a diffraction-limited resolution of about 25 microarcseconds [[1]] — enough to resolve a grapefruit on the Moon.

The result is often called the first photograph of a black hole. It is a reconstruction. Each pair of telescopes measures one Fourier component of the sky’s brightness, a visibility; in 2017 these covered angular scales between 25 and 160 μas, sparsely, so every imaging algorithm has to add assumptions such as positivity or smoothness to turn them into a picture [[1]]. The collaboration’s protection against artefacts from its own imaging assumptions was redundancy: four teams working without contact, then three pipelines built on different software, all recovered the ring [[1]].

In April 2019 the collaboration published six simultaneous papers on M87*, the supermassive black hole at the centre of the Virgo A galaxy [[1]], [[2]], from data taken in April 2017. The ring had an angular diameter of \(42 \pm 3\) μas. From its size, calibrated against a library of simulated images, the collaboration inferred a mass \(M = (6.5 \pm 0.7) \times 10^9 \, M_\odot\) at a distance of 16.8 Mpc [[1]]. The southern arc of the ring was brighter — I will return to why.

In May 2022 the same team published results on Sagittarius A*, the Milky Way’s central black hole [[3]], also from 2017 data. The ring diameter: \(51.8 \pm 2.3\) μas, corresponding to a mass of \(\sim 4 \times 10^6 \, M_\odot\) at 8.15 kpc. M87* is roughly 1600 times more massive than Sgr A* and roughly 2000 times farther away — so the two apparent ring sizes are within 25% of each other. The universe arranged the coincidence; the EHT exploited it.

The Physics of the Ring

The ring is not the black hole itself. You cannot image an event horizon: by definition, no information escapes from it. The EHT reconstructs lensed emission from hot plasma around the compact object, including light associated with the strong-lensing region. Its central brightness depression is consistent with the predicted shadow, but the observed ring is not a resolved image of a single photon orbit.

For a non-rotating (Schwarzschild) black hole, a light ray obeys, in units with \(G = c = M = 1\) and with \(u = 1/r\),

$$ \left(\frac{du}{d\varphi}\right)^2 = \frac{1}{b^2} - u^2 + 2u^3 , \tag{1} $$

where \(r\) is the areal radius, \(\varphi\) the angle around the hole and \(b\) the impact parameter, the distance by which the ray would miss the centre if it were not bent. The term \(u^2 - 2u^3\) is a barrier with its peak at \(u = 1/3\), height \(1/27\). In ordinary units the peak is the photon sphere:

$$ r_\text{ph} = \frac{3GM}{c^2} = \frac{3}{2} R_S $$

where \(R_S = 2GM/c^2\) is the Schwarzschild radius. Light orbiting here is in unstable equilibrium: a small perturbation sends it either spiralling inward or escaping to infinity. A ray from far away clears the barrier if \(1/b^2 > 1/27\), which defines the critical impact parameter:

$$ b_c = \frac{3\sqrt{3} \, GM}{c^2} \approx 5.196 \, \frac{GM}{c^2} $$

Rays with \(b < b_c\) end on the horizon, rays with \(b > b_c\) escape, and the ray with \(b = b_c\) winds onto the photon sphere (Fig. 1). Traced backwards from a distant observer, every direction within \(b_c/D\) of the centre ends on the horizon: that patch of sky is the shadow, and lensing makes it \(b_c/r_\text{ph} = \sqrt{3}\) times wider than the photon sphere.

Figure 1. The shadow’s edge is set by the critical impact parameter \(b_c = 3\sqrt{3}\,GM/c^2\), not by the photon sphere at \(r = 3GM/c^2\). Rays traced backwards from a distant observer on the right, integrated from Eq. (1); the critical ray is the closed-form solution. Lengths in units of \(GM/c^2\).

The angular diameter of the shadow as seen by a distant observer, at a distance \(D \gg GM/c^2\), is therefore:

$$ \theta_\text{shadow} = \frac{2 b_c}{D} = \frac{6\sqrt{3} \, GM}{c^2 D} $$

Plugging in the EHT numbers for M87* (\(M = 6.5 \times 10^9 \, M_\odot\), \(D = 16.8\) Mpc), with \(GM_\odot/c^2 = 1477\) m, 1 Mpc \(= 3.086 \times 10^{22}\) m and 1 rad \(= 2.063 \times 10^{11}\) μas:

$$ \theta \approx \frac{6 \times 1.732 \times 6.5 \times 10^9 \times 1477 \, \text{m}}{16.8 \times 3.086 \times 10^{22} \, \text{m}} \approx 1.92 \times 10^{-10} \, \text{rad} \approx 40 \, \mu\text{as} $$

The EHT measured a \(42 \pm 3\) μas emission-ring diameter. An earlier version of this post did not say that the match is circular: the EHT obtained the mass I plugged in by dividing that diameter by a factor \(\alpha = 10.7\)–11.5, calibrated on simulated images, to get the gravitational radius \(GM/(c^2 D)\) [[1]]. My 40 μas is the ring’s diameter times \(6\sqrt{3}/\alpha \approx 0.9\); all it adds is that the ring sits slightly outside the shadow. The independent check is the mass from stellar dynamics, \(6.2_{-0.6}^{+1.1} \times 10^9 \, M_\odot\), which predicts a Schwarzschild shadow of \(37.6_{-3.5}^{+6.2}\) μas; the mass from gas dynamics predicts 21.3 μas, and the EHT result strongly favours the stellar value [[1]].

Spin and viewing angle change the shadow only a little: the capture radius varies by about 4% with orientation [[1]], and a Kerr shadow is 0–8% smaller than Schwarzschild’s [[3]].

The first image of this kind was computed by Jean-Pierre Luminet in 1979 [[4]], on the punch-card-fed IBM 7040 of the Paris-Meudon Observatory. Lacking drawing software, he made the picture by hand, in dots of Indian ink on negative paper [[5]]. His thin disk, seen 10° above its plane, is not M87*’s plasma, which we see about 17° from its jet axis [[1]], but the essentials are there: a dark centre and a thin, lopsided ring of lensed light [[1]], [[5]], forty years before the EHT published its image.

The Brightness Asymmetry

The southern arc of the M87* ring is brighter. Relativistic Doppler beaming in an orbiting, magnetised plasma is a natural contribution: emission from material moving towards the observer is boosted and blueshifted, while that from receding material is deboosted [[2]]. The image morphology is also shaped by the plasma and lensing, so it is not a direct map of a thin accretion disk.

The southern brightness is consistent with the orientation inferred from M87’s large-scale jet: if the spin is aligned with the jet, it points away from Earth and the material at the bottom of the image moves towards us [[1]], [[2]]. It does not, by itself, measure the black hole spin: the 2017 EHT image could not determine spin quantitatively without stronger assumptions about the emitting plasma [[2]].

Interstellar Got the Lensing Right

In 2014, the visual effects company Double Negative rendered the black hole Gargantua for Christopher Nolan’s Interstellar, tracing bundles of light rays through the geometry of a spinning (Kerr) black hole. Kip Thorne, an executive producer, collaborated on two companion papers with the visual effects team [[6]], [[7]].

The resulting image showed the accretion disk wrapping both above and below the black hole: light from the upper face of the disk’s far side is bent over the top of the shadow, light from its lower face under the bottom [[6]]. The lensing was not artistic licence. Other things were. The disk was made by artists rather than computed from accretion physics, the Doppler shifts in colour and brightness were left out, and although the plot needed a nearly maximal spin, the film’s images use \(a/M = 0.6\), because the fast-spinning hole’s flattened shadow and off-centre disk seemed “too confusing for a mass audience” [[6]].

Nor was it the first rendering from first principles: Luminet’s came 35 years earlier. The team claims “the first Hollywood movie to attempt depicting a black hole as it would actually be seen by somebody nearby”, and found new results on the caustics seen by a camera near a fast-spinning hole, while their disk section holds “no new astrophysical insights” [[6]].

The same team published a companion paper on the wormhole in Interstellar [[7]]. That paper is where things get interesting.

What a Wormhole Would Actually Look Like

A traversable wormhole connects two regions of spacetime through a throat. The simplest example is the wormhole Homer Ellis published in 1973 [[8]], which Morris and Thorne also wrote down in their 1988 teaching paper; it is “sometimes called the Morris-Thorne wormhole metric” [[7]]:

$$ ds^2 = -c^2\,dt^2 + d\ell^2 + \left(\ell^2 + \rho^2\right)\left(d\theta^2 + \sin^2\theta\,d\varphi^2\right) $$

Here \(\ell\) is the proper radial distance from the throat, running from \(-\infty\) on one side to \(+\infty\) on the other, and \(\rho\) is the throat radius. The two sides are mirror images. For light, the same bookkeeping as Eq. (1) gives a barrier \(b^2/(\ell^2 + \rho^2)\) that peaks at the throat. The critical impact parameter is \(b_c = \rho\), and the throat is the photon sphere [[9]]. Rays with \(b < \rho\) cross the throat instead of ending anywhere, rays with \(b > \rho\) turn round at areal radius \(r = b\), and the ray with \(b = \rho\) winds onto the throat (Fig. 2).

Figure 2. Inside the critical impact parameter a wormhole transmits rays instead of capturing them. The rays of Fig. 1 for an Ellis wormhole whose throat radius \(\rho\) equals that figure’s \(b_c\), in units of \(\rho\); open circles mark throat crossings, dotted lines the far side, and the critical ray is \(r = \rho\coth\varphi\). Areal-radius chart as in [[9]]: nothing lies inside \(r = \rho\), so the inner rays rise to meet the throat.

The Ellis wormhole is a geometry lesson, not a candidate for M87*: its \(dt^2\) term is the same everywhere, so it has no mass and could not hold the stars and gas whose motions weigh M87* and Sgr A*. What it shows is that the centre of a wormhole’s image depends on what shines, and where:

  • Distant skies on both sides. Thomas Müller visualised the Ellis wormhole in various environments [[7]], [[10]], and the Interstellar team rendered their own this way [[7]]. Through the throat you see the far side’s sky compressed into a disk, a primary image with thinner higher-order images crowding towards its edge; around it, your own sky, distorted by lensing, with an Einstein ring just outside the edge [[7]]. The signature is a double celestial sphere. Nothing is captured, so there is no shadow in the sense a black hole has. The disk is a window.
  • Glowing gas on both sides. Ohgami and Sakai surrounded an Ellis wormhole with optically thin dust, static or flowing through the throat [[9]]. Because light can orbit at the throat, “a bright ring appears in the image, just as in Schwarzschild spacetime”; the ring “solely confirms neither a black hole nor a wormhole”. Light from the dust beyond the throat brightens the centre, to about a third of the ring’s peak against a sixth for a black hole (read off their Fig. 8). The central depression, which the EHT measures against the ring [[1]], is shallower, not reversed.
  • Glowing gas on our side only. If nothing shines beyond the throat, rays with \(b < b_c\) bring back darkness and the wormhole casts a shadow of its own. This is the assumption behind Bambi’s wormhole shadows [[11]], to which I return below.

In the first case, lit by distant skies, a wormhole looks nothing like the EHT images, and “a second universe in the centre” is a faithful rendering of that setup, not a universal observational prediction.

The Images Rule Something Out

This is the point I find underappreciated. The EHT results are usually discussed as confirming that M87* and Sgr A* are black holes consistent with GR. That framing is correct. But the images are also falsifying evidence against alternatives.

Several exotic compact-object proposals — gravastars, boson stars and some wormhole metrics — can produce shadow-like features [[2]]. Horizon-scale images can constrain parameters of selected alternatives, including wormhole spacetimes, but their present resolution and the uncertain emission model leave degeneracies. The EHT morphology is consistent with Kerr-based models; it is not a unique image-level identification of an event horizon.

Size alone does some of the work for Sgr A*, whose mass stellar orbits pin down; for M87*, stellar and gas dynamics still differ by a factor of 1.8 [[1]]. Take a Morris–Thorne wormhole with \(g_{tt} = -e^{2\Phi}\) and \(\Phi = -r_0/r\), as in Bambi: its mass is \(M = r_0\) (with \(G = c = 1\)), its throat at \(r_0\) is also its photon sphere, and its critical impact parameter is \(e\,GM/c^2 \approx 2.72\,GM/c^2\), about half the Schwarzschild value. With no light from the far side, Bambi predicted \(29 \pm 4\) μas for such a wormhole at Sgr A*, against \(56 \pm 8\) μas for a Schwarzschild black hole [[11]]. The EHT inferred a \(48.7 \pm 7.0\) μas shadow; for this wormhole the fractional deficit \(\delta \approx -0.48\) is “immediately ruled out” [[3]], [[12]].

The constraint is not absolute. Wormhole geometries can have photon spheres — the Ellis wormhole’s is its throat [[9]] — and can mimic a black-hole shadow at current resolution. The Damour–Solodukhin wormhole differs from a Schwarzschild black hole by one parameter \(\lambda\); its shadow radius is about \(3\sqrt{3}\,GM/c^2 / \sqrt{1 + \lambda^2}\), and the Sgr A* image requires only \(\lambda \lesssim 0.4\) at one standard deviation [[13]]. With a tiny \(\lambda\), it casts the shadow of a black hole of the same mass. The clean portal picture belongs to a wormhole lit by distant skies; neither M87* nor Sgr A* has established that geometry or excluded every traversable shortcut through spacetime by morphology alone.

For more on wormhole theory — the Morris–Thorne construction, why a traversable throat needs stress-energy that violates the null energy condition, the disputed “wormhole on a quantum processor” experiment and closed timelike curves — see a later post. The physics is rich and ongoing. But a wormhole lit by distant skies, the way the classic renderings light it, would not look like what the EHT published. It would look like a portal.

The Astonishing Thing Is That It Worked

I want to end on this. The ring around M87* follows from a solution found in 1916 to a theory written down without any observation of a black hole, by people who were not sure black holes existed, using mathematics developed for entirely different purposes. Luminet computed the image in 1979 on punch cards, and it anticipated an image reconstructed in 2019 from data that a planet-scale interferometer took in 2017.

My scale estimate borrowed the mass from the ring. The stellar-dynamics mass predicts 37.6 μas, loosely (quoted range 34.1–43.8 μas); the 42 μas ring sits 12% above that, inside the range, and Sgr A* agrees to within about 10%. That is impressive, but it is not an exactness claim: the uncertainty budget includes the source model as well as the measurement.

The images are astonishing not because they surprised physicists — they confirmed what general relativity predicted. They are astonishing because general relativity is apparently the kind of theory that earns the right to be trusted at microarcsecond precision, at distances of 16.8 megaparsecs, around objects whose entire interiors are, by construction, hidden from us.

Peer review welcome. If you have a wormhole geometry whose shadow is indistinguishable from a Kerr black hole at current EHT resolution, I would genuinely like to read the paper.

Claims and calculations checked against the cited sources through 2026-10-24.

References

[1] Event Horizon Telescope Collaboration et al. (2019). First M87 Event Horizon Telescope results. I. The shadow of the supermassive black hole. The Astrophysical Journal Letters, 875(1), L1. DOI: 10.3847/2041-8213/ab0ec7

[2] Event Horizon Telescope Collaboration et al. (2019). First M87 Event Horizon Telescope results. V. Physical origin of the asymmetric ring. The Astrophysical Journal Letters, 875(1), L5. DOI: 10.3847/2041-8213/ab0f43

[3] Event Horizon Telescope Collaboration et al. (2022). First Sagittarius A* Event Horizon Telescope results. I. The shadow of the supermassive black hole in the center of the Milky Way. The Astrophysical Journal Letters, 930(2), L12. DOI: 10.3847/2041-8213/ac6674

[4] Luminet, J.-P. (1979). Image of a spherical black hole with thin accretion disk. Astronomy & Astrophysics, 75, 228–235.

[5] Luminet, J.-P. (2019). An illustrated history of black hole imaging: Personal recollections (1972–2002). Preprint, arXiv:1902.11196

[6] James, O., von Tunzelmann, E., Franklin, P., & Thorne, K. S. (2015). Gravitational lensing by spinning black holes in astrophysics, and in the movie Interstellar. Classical and Quantum Gravity, 32(6), 065001. DOI: 10.1088/0264-9381/32/6/065001

[7] James, O., von Tunzelmann, E., Franklin, P., & Thorne, K. S. (2015). Visualizing Interstellar’s wormhole. American Journal of Physics, 83(6), 486–499. DOI: 10.1119/1.4916949

[8] Ellis, H. G. (1973). Ether flow through a drainhole: A particle model in general relativity. Journal of Mathematical Physics, 14(1), 104–118. DOI: 10.1063/1.1666161

[9] Ohgami, T., & Sakai, N. (2015). Wormhole shadows. Physical Review D, 91(12), 124020. DOI: 10.1103/PhysRevD.91.124020

[10] Müller, T. (2004). Visual appearance of a Morris–Thorne-wormhole. American Journal of Physics, 72(8), 1045–1050. DOI: 10.1119/1.1758220

[11] Bambi, C. (2013). Can the supermassive objects at the centers of galaxies be traversable wormholes? The first test of strong gravity for mm/sub-mm very long baseline interferometry facilities. Physical Review D, 87(10), 107501. DOI: 10.1103/PhysRevD.87.107501

[12] Event Horizon Telescope Collaboration et al. (2022). First Sagittarius A* Event Horizon Telescope results. VI. Testing the black hole metric. The Astrophysical Journal Letters, 930(2), L17. DOI: 10.3847/2041-8213/ac6756

[13] Vagnozzi, S., Roy, R., Tsai, Y.-D., & Visinelli, L. (2023). Horizon-scale tests of gravity theories and fundamental physics from the Event Horizon Telescope image of Sagittarius A*. Classical and Quantum Gravity, 40(16), 165007. DOI: 10.1088/1361-6382/acd97b


Changelog

  • 2026-07-11: Clarified that EHT images reconstruct lensed plasma emission and a central brightness depression rather than directly resolving a photon sphere or measuring spin; narrowed the wormhole discussion to specified metrics and illumination assumptions.
  • 2026-10-03: Corrected the Sgr A* distance adopted by the EHT from 8.18 kpc to 8.15 kpc (the maser-parallax distance in EHT Sgr A* Paper I), the M87*/Sgr A* mass ratio from roughly 1500 to roughly 1600 ($6.5\times10^9 / 4\times10^6 = 1625$), and the claim that the ring was “predicted in 1916” to say it follows from the 1916 Schwarzschild solution (the photon-capture shadow itself was worked out in the 1950s–60s).
  • 2026-10-24: Corrected “forty years before the telescope existed”: Luminet’s 1979 image preceded the EHT’s 2019 image, from 2017 data. Corrected “photograph” to a reconstruction from sparse visibilities, also in “it matched a photograph”; Luminet’s image “startlingly like” the EHT’s to one sharing its essentials but showing a thin disk seen 10° above its plane; his “photographic paper” to Indian ink on negative paper; the resolution from about 20 to about 25 μas. Said the 40 μas estimate is circular, using the EHT’s ring-derived mass; added the stellar- and gas-dynamics predictions (37.6 and 21.3 μas) and replaced “a few percent” with 4–7% ring precision and the Sgr A* test (about 10%). Corrected Interstellar’s spin (\(a/M = 0.6\)), its disk and Doppler choices and its “first” claim, and the wormhole’s Einstein ring and far sky. Resolved “no shadow” against “can mimic a shadow” by naming metric and lighting; fixed the link text. Added Eq. (1), the spin caveat and Figs. 1–2.