Anatomy of a Black Hole

Here is a claim worth resisting: a black hole is not a thing, it is a place. There is no dense ball of stuff at the centre that you could, in principle, fly up to and touch. The "surface" everyone draws, that smooth black sphere, is not made of anything. It is a boundary in spacetime, a line past which the geometry tilts so far that every road leads inward and none leads back out. Cross it and you would feel nothing. No jolt, no membrane, no heat. You would only learn, later and too late, that the exits had quietly stopped existing.

Almost everything we say about black holes seems to demand exotic new physics. It doesn't. The whole anatomy follows from a single number with units of length, the Schwarzschild radius, and a willingness to take Einstein's geometry literally. Get that one length right and the rest of the body is forced: where light can orbit, where matter's last stable orbit lies, how big a shadow the thing casts on the sky, even the faint temperature at which it glows. The black hole is mostly empty space wrapped around one scale.

The only length that matters

Start with the question a high-schooler can ask: how fast must you go to escape a mass $M$ from a distance $r$? Newtonian energy bookkeeping gives the escape velocity $v_{esc} = \sqrt{2GM/r}$. Now ask the reckless follow-up. At what radius does that escape velocity reach the speed of light, the cosmic speed limit past which nothing can be flung? Set $v_{esc} = c$, solve for $r$, and you get a length.

$$r_s = \frac{2GM}{c^2}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

This is a heuristic, a Newtonian sleight of hand. Light doesn't have escape velocity in the way a cannonball does, and the full relativistic derivation is subtler. The astonishing thing is that the answer is exactly right anyway. Karl Schwarzschild, solving Einstein's field equations from a trench on the Russian front in 1916, found the same radius falling out of the geometry with the same factor of 2 in front. It is the radius of the event horizon for a non-rotating, uncharged black hole.

Plug in numbers and the scale tells you how empty these objects are. Compress the Sun to its Schwarzschild radius and you get a sphere about 3 kilometres across. Do it to the Earth and the horizon is roughly 9 millimetres, the size of a marble. The mass hasn't changed. The horizon is just the radius at which that mass's geometry closes over itself.

Earth's r_s

~9 mm

compress Earth this small and it becomes a black hole

Sun's r_s

~3 km

the Sun's horizon would fit inside a small city

Horizon scaling

∝ M

double the mass, double the radius — linear, not square

Notice the linearity. The horizon radius scales straight with mass, so a supermassive black hole of a billion suns has a horizon a billion times wider than a one-solar-mass hole. That single fact, $r_s \propto M$, is why the biggest black holes are, in a sense, the gentlest. We'll get there.

The metric, and why the horizon isn't a wall

Equation (1) is a number. To understand the body around it, we need the geometry it sits in: the Schwarzschild metric, the rulebook for measuring distances and times outside the hole.

$$ds^2 = -\left(1 - \frac{r_s}{r}\right)c^2\,dt^2 + \left(1 - \frac{r_s}{r}\right)^{-1}dr^2 + r^2\,d\Omega^2$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Don't be put off by the symbols. The whole story is in that repeated factor $(1 - r_s/r)$, where $d\Omega^2 = d\theta^2 + \sin^2\theta\, d\phi^2$ is just the angular part on a sphere. Far away, $r \gg r_s$, the factor is essentially 1, the metric becomes flat, and physics looks ordinary. As $r$ shrinks toward $r_s$, the time term $(1 - r_s/r)$ heads to zero and the radial term $(1 - r_s/r)^{-1}$ blows up. A clock near the horizon ticks slow as seen from far away; a metre stick laid radially gets stretched.

At $r = r_s$ exactly, both terms misbehave: one vanishes, the other diverges. For decades this looked like the location of a catastrophe, a physical edge. It isn't. It's a failure of the coordinates, not the spacetime. Switch to coordinates that ride along with an infalling observer (Eddington-Finkelstein, Kruskal) and the horizon is perfectly smooth. The curvature there is finite. Nothing tears.

So where does the genuine catastrophe live? At $r = 0$, the singularity, where the curvature actually does diverge no matter what coordinates you choose. Inside the horizon the roles of $r$ and $t$ swap. The $(1 - r_s/r)$ factor changes sign, the radial direction becomes timelike, and falling to smaller $r$ becomes as unavoidable as falling forward in time. You cannot steer away from the singularity any more than you can steer away from next Tuesday. That is the real meaning of "no escape": not that the engines aren't powerful enough, but that "out" has stopped being a direction.

Where light orbits, and where matter gives up

Now we can read the rest of the anatomy straight off the geometry, each feature a fixed multiple of $r_s$.

First, light. Outside the horizon, in the curved space of equation (2), there's a special radius where the bending is just strong enough to bend a light ray into a circle. A photon aimed exactly right will orbit the black hole. This is the photon sphere, and for a Schwarzschild hole it sits at

$$r_{ph} = \tfrac{3}{2}\,r_s = 1.5\,r_s.$$

These orbits are unstable, like a pencil balanced on its point. Nudge the photon inward and it spirals into the hole; nudge it outward and it escapes to infinity. Nothing can sit comfortably at $1.5\,r_s$. But the possibility of orbiting light is what carves the shadow we observe, so the photon sphere matters far more than its knife-edge instability suggests.

Now matter. A massive particle, a chunk of an accretion disc, can hold a stable circular orbit much closer than light can in flat space. But not all the way down. There's a radius below which stable circular orbits simply cease to exist: the innermost stable circular orbit, the ISCO. For a non-rotating black hole it sits at

$$r_{ph} = 1.5\,r_s, \qquad r_{ISCO} = 3\,r_s = 6\,\frac{GM}{c^2}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Inside the ISCO, there's no stable orbit to fall onto. Anything that drifts within $3\,r_s$ spirals inexorably inward, plunging across the horizon. This is not a soft edge. The ISCO is why accretion discs have an inner rim: the hot, glowing gas circling a black hole abruptly runs out of stable orbits at $3\,r_s$ and pours over the edge. The light we see from feeding black holes is, in large part, the last scream of matter from just outside the ISCO. Three radii, one ruler: horizon at $r_s$, photon sphere at $1.5\,r_s$, ISCO at $3\,r_s$.

The horizon, the photon sphere, and the last stable orbit are not three discoveries. They are three tick marks on a single ruler, and the ruler's length is fixed the moment you know the mass.

The shadow: what a telescope actually sees

In 2019 the Event Horizon Telescope released the first image of a black hole, a fuzzy orange ring around a dark centre in the galaxy M87. It is tempting to say we "photographed the event horizon". We didn't. The horizon is smaller than the dark patch in that image. What we photographed is the shadow, and its size is set by the photon sphere, not the horizon.

Here's the geometry. Trace light rays backward from your eye toward the black hole. Rays that pass far out swing by and continue, carrying an image of the sky behind. Rays aimed close enough get captured and fall in, so those directions look dark. The boundary between "swings by" and "gets captured" is sharp, and it's controlled by the unstable photon orbits at $1.5\,r_s$. Light skimming the photon sphere can wind around several times before escaping, which is why a bright photon ring traces the shadow's rim. Working through the relativistic ray-bending gives the shadow's apparent radius as $\sqrt{27}/2 \cdot r_s$, so its diameter is

$$D_{shadow} = \sqrt{27}\;r_s \approx 5.2\,r_s$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Look at how clean that is. The shadow is about 2.6 times wider than the horizon, and the entire calculation reduces to one irrational number, $\sqrt{27}$, times the Schwarzschild radius. There is no free parameter, no fudge factor. Measure the shadow's angular size on the sky, know the distance to the galaxy, and equation (4) hands you the mass directly. The EHT did exactly this for M87 and got about 6.5 billion solar masses, in agreement with completely independent measurements from the motion of nearby stars and gas. A formula carved out of Schwarzschild's 1916 geometry was confirmed by a planet-sized radio telescope a century later.

A black hole has a temperature

Everything so far is pure geometry, classical general relativity with no quantum mechanics in sight. Stephen Hawking's 1974 result broke that. Apply quantum field theory to the curved spacetime just outside the horizon and the black hole is not truly black. It glows, faintly, with a thermal spectrum, as if it had a temperature.

The mechanism is subtle and the cartoon of "virtual pairs splitting at the horizon" is more misleading than helpful, so I'll give the result and the scaling rather than a wrong story. The Hawking temperature is

$$T_H = \frac{\hbar c^3}{8\pi k_B G M}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

The mass sits in the denominator, which gives the counter-intuitive sign. Bigger black holes are colder. A solar-mass black hole has a Hawking temperature around 60 nanokelvin, far colder than the 2.7-kelvin cosmic microwave background bathing it, so in practice it absorbs vastly more than it emits and grows rather than shrinks. A supermassive hole of a million suns is colder still, around 12 picokelvin. To actually evaporate, a black hole must be tiny: the temperature climbs as it loses mass, the radiation accelerates, and the final moments are a runaway flash. Stellar-mass holes radiate faster than supermassive ones for the same reason a teaspoon cools faster than an ocean, but every astrophysical black hole today is gaining mass, not losing it. Evaporation is a story for a far emptier, colder universe than ours.

Three numbers, and an area that counts the bits

Hawking's temperature implies a thermodynamics, and thermodynamics implies entropy. Jacob Bekenstein argued, before Hawking's calculation, that a black hole must carry entropy, otherwise you could throw a hot gas in and violate the second law of thermodynamics for the outside world. The entropy turned out to be strange in a way that still hasn't been fully digested. It scales not with the volume of the hole but with the area of its horizon.

$$S_{BH} = \frac{k_B c^3 A}{4\hbar G} = \frac{k_B\,A}{4\,L_P^2}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Read the right-hand form. The entropy is the horizon area $A$ measured in units of the Planck area $L_P^2 \approx 2.6 \times 10^{-70}\ \text{m}^2$, divided by four. Roughly, the horizon is tiled by Planck-sized patches and the entropy counts those tiles. Entropy counts microstates, so the number of hidden internal configurations of a black hole is set by its surface, not its interior. That is the seed of the holographic principle: the information content of a region seems to live on its boundary. A solar-mass black hole's entropy is around $10^{77} k_B$, dwarfing the entropy of the star that formed it by roughly twenty orders of magnitude. Most of the entropy budget of the universe is hidden behind horizons.

And how much else is hidden? The no-hair theorem, proved through the work of Israel, Carter, and Hawking, says that a black hole settles down to a state described by just three numbers: its mass $M$, its electric charge $Q$, and its angular momentum $J$. That's it. Two black holes with the same three numbers are identical, regardless of whether one formed from a collapsing star and the other from a collapsing encyclopaedia. Everything else, the chemistry, the structure, the history of whatever fell in, is erased from the outside description. A black hole is the most featureless macroscopic object in physics.

The gentleness of the giants

One last consequence is worth holding onto, because it overturns the most cinematic image we have. Spaghettification, the stretching of an infalling body by tidal forces, does not happen at the event horizon. Tidal forces come from the difference in gravity between your head and your feet, and that gradient scales as $M/r^3$. Evaluate it at the horizon, where $r = r_s \propto M$, and the tidal force at the horizon scales as $1/M^2$.

So the bigger the black hole, the weaker the tidal stretching at its horizon. For a stellar-mass black hole, you'd be torn apart well before reaching the horizon, the gradient between your head and feet exceeding what flesh can hold. For a supermassive black hole, the tidal force at the horizon is gentler than standing on Earth. You could cross the horizon of M87's monster entirely intact, feeling nothing, not even knowing you'd passed the point of no return. The lethal tides would come later, deep inside, as $r$ shrank toward zero. The doom is real, but it waits at the singularity, not at the door.

That's the anatomy. No surface, no wall, no fire at the edge. A one-way boundary set by a single length, a photon sphere and an ISCO and a shadow that are all just multiples of that length, a faint temperature that runs backwards, an entropy that lives on the skin, and three numbers that are the entire public identity of the object. Take the geometry literally and the black hole stops being a monster and becomes something stranger and cleaner: a place where spacetime folds over on itself, and a length scale where the folding completes. Some food for thought, next time you see that orange ring.

Reading further

• Schwarzschild, On the Gravitational Field of a Mass Point according to Einstein's Theory (1916). The original exact solution to Einstein's field equations, derived from the front lines of the First World War; the source of equation (1) and the metric of equation (2).
• Hawking, Particle creation by black holes (Commun. Math. Phys. 43, 1975). The paper that proved black holes radiate thermally and gave them a temperature, opening black hole thermodynamics; equations (5) and (6) descend from here.
• Event Horizon Telescope Collaboration, First M87 Results I: The Shadow of the Supermassive Black Hole (ApJL 875, 2019). The first direct image of a black hole's shadow, confirming the $\sqrt{27}\,r_s$ prediction of equation (4) a century after Schwarzschild.
• Thorne, Black Holes & Time Warps: Einstein's Outrageous Legacy (1994). The definitive popular synthesis of black hole geometry by a Nobel laureate; the best long-form companion to the geometric picture in this essay.