Testing Einstein's Theory at the Milky Way's Supermassive Black Hole

The Shadow That Shouldn’t Fit

For decades, we have told ourselves a story about black holes. They are the simplest objects in the universe. No matter how chaotically a star collapses, no matter how violently galaxies merge, the black hole that remains can be described by exactly three numbers: its mass, its spin, and its electric charge. That is the Kerr metric, the solution to Einstein’s equations that has held since 1963. It is beautiful. It is elegant. And it might be wrong.

But in 2022, a team of more than 300 researchers did something that would have been impossible a generation ago. They pointed the Event Horizon Telescope at Sagittarius A*, the supermassive black hole at the center of our own Milky Way, and they took its picture. Not a blurry approximation. Not a computer simulation. An actual image of the shadow of a black hole 27,000 light years away.

The image looked exactly like what Einstein’s theory predicted.

That alone was remarkable. But what happened next was even stranger. The researchers, led by Kazunori Akiyama and colleagues from institutions around the world, did not just compare the image to Kerr predictions. They systematically tested every alternative they could think of. Black holes with extra charges. Black holes that were not actually black holes but objects with surfaces. Black holes described by metrics that bend spacetime in ways Einstein never imagined.

And in every case, the shadow said no.

The results, published in The Astrophysical Journal Letters as part of the first Sagittarius A Event Horizon Telescope results, represent the most precise test of general relativity ever performed at the center of a galaxy. The authors found that the observed image size of Sgr A is within about 10 percent of the Kerr prediction (Akiyama et al., 2022). That margin of error is not a failure. It is a triumph. It means that if there is any deviation from Einstein’s theory, it is hiding in the noise.

But the real story is not just that Einstein was right again. It is that we now know exactly where to look if he is wrong.

What the Shadow Actually Tells Us

The Simple Geometry of Nothing

Here is the strange thing about black hole shadows. They are not shadows in the way you think. A shadow on the ground is caused by an object blocking light. A black hole shadow is caused by the absence of light that never had a chance to exist.

Photons that pass too close to a black hole get trapped. They spiral inward, never to escape. Photons that pass at just the right distance skim the edge of the event horizon and get bent into orbits. And photons that pass farther away continue on their path, slightly deflected. The result is a dark region in the center of the image, surrounded by a bright ring of light. That dark region is the shadow.

The size of that shadow is not arbitrary. It is determined by the mass of the black hole and the geometry of spacetime around it. For a Kerr black hole, the shadow is almost circular, with a diameter that depends on the spin. For alternatives to Kerr, the shadow can be larger, smaller, or distorted.

This is what made Sgr A* such a perfect target. We already knew its mass with extraordinary precision from decades of tracking stars orbiting it. The authors used that prior constraint on the mass to distance ratio to predict exactly how big the shadow should be if Einstein was right. Then they compared that prediction to the actual image.

The shadow was the right size.

The Library of Doom

But the authors did not stop there. They built a library of simulations. Not just Kerr black holes, but black holes described by every alternative metric they could find. Black holes with extra dimensions. Black holes with scalar fields. Black holes that break the no hair theorem, the idea that black holes can only have three properties.

This is where the work gets genuinely clever. The authors did not just simulate what each alternative black hole would look like. They also simulated how the surrounding plasma would emit light, how the accretion disk would appear, and how the image would be blurred by interstellar scattering. They created thousands of synthetic images and asked: which ones match the data?

The answer was almost none. The authors found that the observed image size constrains deviations from Kerr to within about 10 percent for a wide class of alternative metrics (Akiyama et al., 2022). Some specific alternatives, like those with a nonzero NUT parameter (a mathematical trick that produces a black hole with a kind of gravitomagnetic monopole), were ruled out even more tightly.

The Black Hole That Might Not Be a Black Hole

What If There Is No Event Horizon?

Here is a question that keeps physicists up at night. What if black holes are not actually black holes? What if there is no event horizon, no point of no return, but instead some kind of exotic compact object with a surface?

This is not idle speculation. The information paradox, the firewall problem, and the breakdown of general relativity at singularities all suggest that something is missing from our picture. Perhaps black holes are actually gravastars. Or boson stars. Or fuzzballs. Or any of a dozen other proposals that replace the event horizon with something else.

The Event Horizon Telescope observations of Sgr A gave the authors a chance to test this idea directly. If Sgr A had a surface, it would behave differently in two crucial ways.

First, the shadow would be different. A compact object with a surface would not trap light as efficiently as a black hole. Photons that would have fallen into the event horizon might instead hit the surface and either be absorbed or reflected. This would change the size and shape of the dark region.

Second, the spectrum would be different. If the surface absorbed infalling matter and reemitted it as thermal radiation, we would see a distinct glow. If it reflected radiation, we would see echoes.

The authors tested both possibilities. For a thermal surface, they used the observed image size and the broadband spectrum of Sgr A* to conclude that such a surface can be ruled out. The object is too dark. It absorbs too much. For a fully reflective surface, they found it unlikely, though not completely impossible (Akiyama et al., 2022).

The Problem with Surfaces

The reasoning here is worth unpacking. A thermal surface would heat up as matter fell onto it. The hotter it got, the more it would glow. At some point, that glow would become visible in the infrared or optical spectrum. We do not see that glow. Either the surface is incredibly cold, which is hard to explain given the amount of matter falling in, or there is no surface at all.

A reflective surface is trickier. If the surface bounced light back, we might see a kind of echo in the image. But the authors found that the observed image is consistent with a black hole shadow, not a reflection pattern. A reflective surface is not ruled out entirely, but it requires fine tuning that makes it less plausible than the simple Kerr solution.

This is science at its most elegant. Take a beautiful theory. Predict something specific. Go look. If you see it, the theory survives. If you do not, the theory is wrong. The authors looked. They saw the shadow. The Kerr metric survived.

Why This Matters More Than You Think

The Mass Gap Problem

One of the most frustrating things about testing general relativity is that we usually test it in weak gravity. The orbits of planets. The bending of light by the Sun. Gravitational lensing by galaxies. All of these confirm Einstein, but they do so in regimes where the deviations from Newton are small.

Black holes are different. They are the strongest gravitational fields in the universe. Testing relativity near a black hole is like testing aerodynamics in a wind tunnel at Mach 3. If there is a flaw in the theory, this is where it will show up.

But until recently, we could only test relativity near stellar mass black holes, which have masses of a few to a few tens of suns. The Event Horizon Telescope changed that. It allowed us to test relativity near supermassive black holes, which have millions or billions of solar masses. And the results from M87, the first black hole imaged, were consistent with Kerr.

Now we have results from Sgr A*, which is 1,500 times less massive than M87. The authors found that the external spacetimes of both black holes are described by the Kerr metric, independent of their mass (Akiyama et al., 2022). This is not just a confirmation. It is a closure. It means that black holes across the entire mass spectrum behave the same way.

The Charge of Nothing

There is one more number in the Kerr metric: electric charge. In principle, a black hole could have an electric charge. In practice, astrophysical black holes are expected to be neutral because any charge would quickly be neutralized by the surrounding plasma.

But what if they are not? What if a black hole somehow retained a net charge? The authors tested this. They found that the observed shadow size constrains the charge of Sgr A* to be very small. A charged black hole would have a different shadow size, and the data do not support it (Akiyama et al., 2022).

This is important because charged black holes are the simplest deviation from Kerr. If you want to modify general relativity, one of the easiest ways is to add a charge term. But the data say no.

What the Research Does Not Prove

The Limits of the Shadow

Here is the honest truth. The Event Horizon Telescope can only measure the size of the shadow to about 10 percent precision. That is good enough to rule out large deviations from Kerr. It is not good enough to rule out small deviations.

There are alternative theories of gravity that predict deviations from Kerr at the 1 percent level or smaller. These theories are not tested by these observations. The authors are clear about this. They write that their bounds are parametric, meaning they constrain the parameters of specific alternative metrics, not all possible alternatives.

The Spin Problem

The shadow size depends on the spin of the black hole. A rapidly spinning black hole has a smaller shadow than a non spinning one. The authors used the prior constraints on the mass to distance ratio to break this degeneracy, but the spin of Sgr A* is not well known. If the spin is very different from what they assumed, the constraints could change.

This is not a flaw in the analysis. It is a limitation of the data. Future observations with higher resolution could measure the spin directly and tighten the constraints.

The Plasma Problem

The shadow is not a direct image of the black hole. It is an image of the plasma around the black hole, with a dark region in the middle. The size and shape of that dark region depend on the distribution and emission properties of the plasma. The authors used a library of simulations to account for this, but the simulations are only as good as the models.

If the plasma behaves in ways we do not expect, the inferred shadow size could be wrong. This is the biggest source of systematic uncertainty in the analysis.

What This Actually Means

• ▸The Kerr metric is not just a mathematical curiosity. It is the actual description of spacetime around real black holes, from stellar mass to supermassive. The Event Horizon Telescope has now confirmed this for two black holes of vastly different masses, closing a major loophole in our tests of general relativity.

• ▸If Sgr A* is an exotic compact object with a surface, that surface cannot be thermal. The data rule out any object that heats up and glows as matter falls onto it. A reflective surface is possible but requires fine tuning that makes it less plausible than a standard black hole.

• ▸The no hair theorem, which states that black holes have only mass, spin, and charge, passes its most stringent test yet. The shadow of Sgr A* is consistent with a Kerr black hole and inconsistent with alternatives that add extra parameters.

• ▸The charge of Sgr A* is constrained to be very small. If the black hole had a significant electric charge, the shadow would be the wrong size. This is consistent with theoretical expectations but is now an observational fact.

• ▸The 10 percent precision of the current measurement is not the end of the story. Future observations with the next generation Event Horizon Telescope could improve this to 1 percent or better, potentially revealing deviations that are invisible today.

• ▸The simplest explanation is still the best. After decades of theoretical work on alternatives to Kerr, after hundreds of papers on fuzzballs and gravastars and wormholes, the data say that black holes are exactly what Einstein predicted. They are simple. They are elegant. And they are real.