Kruskal–Szekeres Coordinates

The Schwarzschild metric describes the geometry outside any spherical mass, but in its familiar $(t,r,\theta,\phi)$ form it breaks down at the radius $r=2M$ : the coefficient $g_{tt}=-(1-2M/r)$ vanishes there and $g_{rr}=(1-2M/r)^{-1}$ blows up. For four decades this looked like the edge of spacetime — the "Schwarzschild singularity" — and it stalled the physical interpretation of black holes. The resolution is that the curvature is perfectly finite at $r=2M$ : the Kretschmann invariant $R_{\alpha\beta\gamma\delta}R^{\alpha\beta\gamma\delta}=48M^2/r^6$ is a plain number there, blowing up only at $r=0$ . The trouble is not the geometry but the chart. Kruskal–Szekeres coordinates are the cure: a single, maximally extended coordinate system in which the metric is manifestly regular across $r=2M$ and which continues the geometry as far as it analytically goes. The construction is a chain of three moves. First the tortoise coordinate $r_\ast$ straightens the radial light rays; then null coordinates $u=t-r_\ast$ , $v=t+r_\ast$ put those rays at fixed values; finally an exponential reparametrization $U=-e^{-u/4M}$ , $V=e^{v/4M}$ spreads the infinite pile-up of coordinate time at the horizon onto a smooth, finite null line. The payoff is a diagram on which radial light cones sit at exactly $45^\circ$ everywhere, so causal structure can be read by eye. The full manifold splits into four regions — our exterior, the black-hole interior, a causally disconnected second exterior, and a white-hole interior — stitched together across a single smooth vacuum spacetime. This is the eternal Schwarzschild solution, the stage on which surface gravity, the bifurcate Killing horizon, and ultimately Hawking radiation are all written.

Key takeaways

Kruskal–Szekeres coordinates $(T,X)$ are the maximal analytic extension of the Schwarzschild geometry: one chart covering the exterior, the black-hole interior, a second asymptotic region, and a white-hole interior, all in a single smooth manifold.
The coordinate breakdown at $r=2M$ is removed — the metric coefficient never vanishes and never diverges there — so the horizon is exposed as a regular null surface, not a physical edge; the only true singularity is at $r=0$ .
Radial light rays run on the lines $T=\pm X+\text{const}$ at $45^\circ$ , so light cones, horizons and singularities can be read straight off the diagram: the horizon $r=2M$ is $T=\pm X$ , and $r=0$ is the spacelike hyperbola $T^2-X^2=1$ .
The single conformal factor $F(r)=(32M^3/r)\,e^{-r/2M}$ is smooth and strictly positive for all $0<r<\infty$ ; at the horizon $F(2M)=16M^2/e\neq0$ , which is the whole proof that $r=2M$ is a coordinate artefact.
The Schwarzschild time $t$ acts on the diagram as a Lorentz boost with rapidity $t/4M$ ; the horizon is the fixed-point set of that boost — a bifurcate Killing horizon with surface gravity $\kappa=1/4M$ , the classical seed of the Hawking temperature $T_H=1/8\pi M$ .
Key idea: reparametrize the Schwarzschild null coordinates exponentially, $U=-e^{-u/4M}$ and $V=e^{v/4M}$ , so the offending factor $(1-2M/r)$ in the metric is cancelled exactly by $e^{r_\ast/2M}$ , leaving a smooth nonzero conformal factor everywhere but $r=0$ and mapping the horizon to the regular null lines $U=0$ or $V=0$ .

The core insight

Reparametrize the Schwarzschild null coordinates exponentially, $U=-e^{-u/4M}$ and $V=e^{v/4M}$ , so that the factor $(1-2M/r)$ multiplying the metric is exactly cancelled by $e^{r_\ast/2M}$ ; the conformal factor becomes a smooth nonzero function of $r$ everywhere except $r=0$ , and the horizon $r=2M$ maps to the regular null lines $U=0$ or $V=0$ .

Where this lives today

The Kruskal picture is the working backbone of black-hole physics, not a museum piece. Compactifying $(T,X)$ produces the Carter–Penrose diagram of the eternal black hole, the standard tool for global causal structure and the setting for Penrose's 1965 singularity theorem. The bifurcate Killing horizon and its surface gravity $\kappa=1/4M$ are the classical inputs to Hawking's 1974 calculation and to black-hole thermodynamics, where the same $\kappa$ appears in the first law $\mathrm{d}M=\tfrac{\kappa}{8\pi}\mathrm{d}A$ . The two-exterior structure of regions I and III realizes the thermofield-double state, and the ER=EPR proposal of Maldacena and Susskind (2013) reads the Einstein–Rosen bridge as the geometry of entanglement between two black holes. Observationally, the horizon whose regularity Kruskal coordinates established is now imaged directly: the Event Horizon Telescope resolved the shadow of M87\ in 2019 and Sgr A\ in 2022, both consistent with a Kerr-like horizon at the predicted scale.

The physical picture

Think of the Schwarzschild time $t$ as a clock that runs infinitely slow at the horizon. An outgoing light ray climbing toward $r=2M^+$ takes an unbounded amount of coordinate time to get there, so successive crests pile up against the surface like logs jamming at a narrows — a purely bookkeeping congestion, since a falling observer sails through in finite proper time and notices nothing. The Kruskal reparametrization is exponential precisely because the jam is exponential: $t\to+\infty$ enters through $e^{-u/4M}$ , which squeezes the entire infinite pile-up onto the finite, smooth null line $U=0$ . Nothing physical happens at $r=2M$ , and the Kruskal metric makes that mathematically manifest by staying finite and nondegenerate there. The first figure is the resulting map of the whole spacetime.

Figure

The Kruskal–Szekeres plane. Radial light rays run at 45°, so the horizon r = 2M is the pair of null lines T = ±X through the bifurcation point, and the true singularity r = 0 is the spacelike hyperbola T² − X² = 1. Region I is our exterior, II the black hole, III a second exterior, IV the white hole.

Because the two-dimensional $T$ – $X$ metric is conformally flat, $\mathrm{d}s^2_{2D}=F(r)\,(-\mathrm{d}T^2+\mathrm{d}X^2)$ , the light cones keep the flat-space $45^\circ$ opening at every point, including on and inside the horizon. That is the diagram's superpower: the future of any event is the $45^\circ$ wedge above it, and one sees at a glance that a worldline which crosses into region II is trapped, since the singularity $r=0$ is a spacelike hyperbola lying across its entire future. Hitting $r=0$ inside the hole is therefore not like falling onto a place — it is as unavoidable as reaching next Tuesday. The second figure shows the flip side of the same fact: an infall that looks frozen in Schwarzschild time is smooth and finite in Kruskal coordinates.

Figure

The same radial infall in two charts. In Schwarzschild time t the worldline takes infinite t to reach r = 2M — the coordinate freezes. In Kruskal coordinates it crosses the horizon smoothly and reaches r = 0 in finite proper time Δτ = 4M/3, about 6.6 μs for one solar mass.

The Schwarzschild time-translation symmetry $t\to t+c$ becomes, on this diagram, a Lorentz boost with rapidity $c/4M$ : constant- $t$ surfaces are straight rays through the origin, and constant- $r$ surfaces are hyperbolae. The horizon lines $T=\pm X$ are exactly the fixed-point set where the boost degenerates — the same algebra that governs a uniformly accelerated (Rindler) observer in flat space. That structural identity is why the Unruh effect for accelerated observers and Hawking radiation from black holes come from one derivation, and why the number $1/4M$ that sets the boost scale is also the surface gravity that fixes the Hawking temperature.

The core equations

The starting point is the Schwarzschild line element, the unique static spherically symmetric vacuum solution. It is regular for $r>2M$ but its coefficients misbehave at $r=2M$ , which is the problem the rest of the article solves. We work in geometric units $G=c=1$ , so $M$ is a length.

\mathrm{d}s^2=-\left(1-\frac{2M}{r}\right)\mathrm{d}t^2+\left(1-\frac{2M}{r}\right)^{-1}\mathrm{d}r^2+r^2\,\mathrm{d}\Omega^2 \qquad (1)

where: $M$ — geometric mass $GM_{\rm phys}/c^2$ ; $r$ — areal radius (a 2-sphere at fixed $r$ has area $4\pi r^2$ ); $t$ — Schwarzschild time, the proper time of a static observer at infinity; $\mathrm{d}\Omega^2=\mathrm{d}\theta^2+\sin^2\theta\,\mathrm{d}\phi^2$ . Reading: the vanishing of $g_{tt}$ and the divergence of $g_{rr}$ at $r=2M$ are the coordinate pathology we must remove. To straighten the radial light rays we introduce the tortoise coordinate, defined so that its differential absorbs the troublesome $g_{rr}$ . It pushes the horizon out to $r_\ast\to-\infty$ , spreading the pile-up that the exponential map will later tame.

r_\ast=r+2M\ln\left|\frac{r}{2M}-1\right|,\qquad \frac{\mathrm{d}r_\ast}{\mathrm{d}r}=\left(1-\frac{2M}{r}\right)^{-1} \qquad (2)

where: $r_\ast$ — tortoise coordinate; $r_\ast\to-\infty$ as $r\to2M^+$ and $r_\ast\to+\infty$ as $r\to\infty$ . Reading: radial null rays now satisfy $t=\pm r_\ast+\text{const}$ , so they are straight lines in $(t,r_\ast)$ . Building the two families of radial null rays into the coordinates themselves gives the Eddington–Finkelstein null coordinates, one constant along ingoing rays and one along outgoing rays.

u=t-r_\ast,\qquad v=t+r_\ast \qquad (3)

where: $u$ — retarded (outgoing) null coordinate; $v$ — advanced (ingoing) null coordinate. Reading: in $(u,v)$ the metric is $\mathrm{d}s^2=-(1-2M/r)\,\mathrm{d}u\,\mathrm{d}v+r^2\mathrm{d}\Omega^2$ — still singular at $r=2M$ , because $(1-2M/r)\to0$ there. The decisive step is the exponential reparametrization. In region I the outgoing coordinate carries a minus sign and the horizon sits at $U=0$ or $V=0$ ; the scale $4M$ is fixed uniquely by the requirement that the horizon factor cancel.

U=-e^{-u/4M},\qquad V=e^{v/4M}\quad(\text{region I}) \qquad (4)

where: $U,V$ — Kruskal null coordinates; future horizon at $U=0$ , past horizon at $V=0$ ; the constant $4M=1/\kappa$ is the inverse surface gravity. Reading: a shift $t\to t+c$ rescales $U\to Ue^{-c/4M}$ , $V\to Ve^{c/4M}$ — a boost, not a translation. Passing to a timelike and a spacelike combination puts the metric into its final, manifestly regular form. The prefactor $F(r)$ is the conformal factor, and its smoothness across $r=2M$ is the entire content of the resolution.

\mathrm{d}s^2=\frac{32M^3}{r}\,e^{-r/2M}\left(-\mathrm{d}T^2+\mathrm{d}X^2\right)+r^2\,\mathrm{d}\Omega^2,\qquad T=\tfrac12(V+U),\ X=\tfrac12(V-U) \qquad (5)

where: $F(r)=(32M^3/r)\,e^{-r/2M}$ — conformal factor, finite and positive for all $0<r<\infty$ ; $T$ — Kruskal time; $X$ — Kruskal radial coordinate. Reading: $F(2M)=16M^2/e\neq0$ , so the metric is smooth and nondegenerate at the horizon; light cones are $45^\circ$ because the bracket is conformally Minkowski. The radial coordinate $r$ is no longer a coordinate but a function of $T$ and $X$ , fixed implicitly by the relation below. It is this equation that assigns each region and locates the horizon and singularity.

\left(\frac{r}{2M}-1\right)e^{r/2M}=X^2-T^2 \qquad (6)

where: $X^2-T^2>0$ is the exterior ( $r>2M$ ); $=0$ the horizon ( $r=2M$ ); $\in(-1,0)$ the interior; $=-1$ the singularity ( $r=0$ , i.e. $T^2-X^2=1$ ). Reading: because $r=0$ maps to $T^2-X^2=1$ , a spacelike hyperbola, the singularity is a moment of time inside the hole, not a place. Finally, the Schwarzschild time is recovered as a boost angle, making explicit that surfaces of constant $t$ are rays through the origin and that the horizon is the boost's fixed set.

\frac{T}{X}=\tanh\!\left(\frac{t}{4M}\right)\ (\text{region I}),\qquad \frac{X}{T}=\tanh\!\left(\frac{t}{4M}\right)\ (\text{region II}) \qquad (7)

where: $t$ — Schwarzschild time; $t\to\pm\infty$ corresponds to $T=\pm X$ , the horizon. Reading: the whole infinite range of $t$ is compressed into the finite wedge bounded by the null lines — the geometric statement of the coordinate freeze.

Derivation ACancelling the horizon singularity

The goal is to show explicitly why the exponential reparametrization works: to start from the Schwarzschild metric written in null coordinates, where the factor $(1-2M/r)$ still vanishes at the horizon, and to track that factor through the change of variables to $U$ and $V$ until it is cancelled term by term, leaving the smooth conformal prefactor $32M^3 e^{-r/2M}/r$ . This is the calculation that converts the vague statement "the singularity is only in the coordinates" into an exact algebraic identity, and it pins down why the magic scale is $4M$ and nothing else.

Step 1. Write the Schwarzschild metric in the Eddington–Finkelstein null coordinates of equation (3): $\mathrm{d}s^2=-(1-2M/r)\,\mathrm{d}u\,\mathrm{d}v+r^2\mathrm{d}\Omega^2$ . The two-dimensional radial part is now proportional to $\mathrm{d}u\,\mathrm{d}v$ , but the coefficient still collapses at $r=2M$ . Why: null coordinates already diagonalize the light-cone structure, so the only remaining defect is the scalar prefactor, which is what we now attack.

Step 2. Differentiate the definitions $U=-e^{-u/4M}$ and $V=e^{v/4M}$ to get $\mathrm{d}U=(1/4M)e^{-u/4M}\mathrm{d}u$ and $\mathrm{d}V=(1/4M)e^{v/4M}\mathrm{d}v$ , so that $\mathrm{d}U\,\mathrm{d}V=(1/16M^2)\,e^{(v-u)/4M}\,\mathrm{d}u\,\mathrm{d}v$ . Since $v-u=2r_\ast$ , this is $\mathrm{d}U\,\mathrm{d}V=(1/16M^2)\,e^{r_\ast/2M}\,\mathrm{d}u\,\mathrm{d}v$ . Why: the Jacobian of the exponential map is itself exponential in $r_\ast$ , and $r_\ast$ is exactly where the logarithm of the horizon factor lives — so the Jacobian is built to swallow that factor.

Step 3. Evaluate the Jacobian factor using the tortoise definition (2): $e^{r_\ast/2M}=e^{r/2M}\,e^{\ln(r/2M-1)}=e^{r/2M}\left(\tfrac{r}{2M}-1\right)=e^{r/2M}\,\tfrac{r}{2M}\left(1-\tfrac{2M}{r}\right)$ . The factor $(1-2M/r)$ has now appeared, on the same footing as the one in the metric. Why: this is the crux — the tortoise logarithm was chosen precisely so that exponentiating $r_\ast$ regenerates the horizon factor, ready to cancel.

Step 4. Invert Step 2 to write $\mathrm{d}u\,\mathrm{d}v=16M^2\,e^{-r_\ast/2M}\,\mathrm{d}U\,\mathrm{d}V$ and substitute into the metric: the offending factor cancels, $-(1-2M/r)\,\mathrm{d}u\,\mathrm{d}v=-(1-2M/r)\cdot16M^2\,e^{-r/2M}\,\tfrac{2M}{r}\,(1-2M/r)^{-1}\,\mathrm{d}U\,\mathrm{d}V=-(32M^3/r)\,e^{-r/2M}\,\mathrm{d}U\,\mathrm{d}V$ . Why: the $(1-2M/r)$ from the metric and the $(1-2M/r)^{-1}$ from the inverse Jacobian annihilate exactly, and only the smooth, strictly positive prefactor survives for every $r>0$ .

Step 5. Pass to $(T,X)$ via $U=X-T$ ... more precisely $\mathrm{d}U\,\mathrm{d}V=\mathrm{d}X^2-\mathrm{d}T^2$ from $T=\tfrac12(V+U),X=\tfrac12(V-U)$ , giving the final form (5). The horizon $r=2M$ now sits at the perfectly ordinary null lines $T=\pm X$ , and only $r=0$ (where $F\to\infty$ ) remains singular. Why: the surviving prefactor is analytic and nonzero across $r=2M$ , so the chart is valid there and can be continued smoothly into the interior — the maximal extension. *Verified: the cancellation reproduces $\mathrm{d}s^2=(32M^3/r)e^{-r/2M}(-\mathrm{d}T^2+\mathrm{d}X^2)+r^2\mathrm{d}\Omega^2$ with $F(2M)=16M^2/e$ , matching MTW §31.5 and Carroll §5.6.*

Go deeper · Derivation B · 1 minFinite proper time to reach the singularity

Derivation BFinite proper time to reach the singularity

The goal is to demonstrate quantitatively that the horizon is no barrier by computing the proper time a freely falling observer logs on the way in, and to contrast it with the infinite Schwarzschild coordinate time. We take the simplest radial geodesic — a particle released from rest at infinity — reduce the equation of motion to a first-order form using the conserved energy, and integrate the elementary result down through $r=2M$ all the way to $r=0$ , obtaining the celebrated finite answer $\Delta\tau=4M/3$ from the horizon to the centre.

Step 1. For a radial timelike geodesic the normalization $g_{\mu\nu}u^\mu u^\nu=-1$ combined with the conserved energy $E=(1-2M/r)\,\dot t$ gives $(\mathrm{d}r/\mathrm{d}\tau)^2=E^2-(1-2M/r)$ . This holds smoothly across $r=2M$ because $\tau$ and $r$ are good variables there. Why: proper time and the areal radius are geometric scalars, unaffected by the coordinate defect of $t$ , so the geodesic equation stays regular at the horizon.

Step 2. Choose the observer dropped from rest at infinity, which fixes $E=1$ . The radial equation collapses to $\mathrm{d}r/\mathrm{d}\tau=-\sqrt{2M/r}$ , taking the negative root for infall. Why: remarkably this is identical to the Newtonian free-fall speed $\sqrt{2M/r}$ , one of the coincidences that makes Schwarzschild infall so clean; the general-relativistic content is hidden in which clock reads $\tau$ .

Step 3. Separate and integrate: $\Delta\tau=\int_{r_2}^{r_1}\sqrt{r/2M}\;\mathrm{d}r=\tfrac{2}{3\sqrt{2M}}\left(r_1^{3/2}-r_2^{3/2}\right)$ . Every endpoint gives a finite result, including the crossing of $r=2M$ and the descent to $r=0$ . Why: the integrand $\sqrt{r/2M}$ is finite and integrable across the horizon — nothing in the proper-time bookkeeping knows a horizon is there.

Step 4. Set $r_1=2M$ , $r_2=0$ : $\Delta\tau=\tfrac{2}{3\sqrt{2M}}(2M)^{3/2}=\tfrac{2}{3}(2M)=\dfrac{4M}{3}$ . Compare with the Schwarzschild coordinate time, which diverges, $t(r)\to+\infty$ as $r\to2M^+$ . Why: the physical, invariant duration is finite and short, while the apparent freeze in $t$ is an artefact of a clock adapted to a static observer who cannot exist inside — exactly the discrepancy Kruskal coordinates dissolve. *Verified: $\Delta\tau_{2M\to0}=4M/3$ in geometric units, the standard textbook value (Hartle Ch. 12, MTW §25.5); for one solar mass this is $(4/3)GM_\odot/c^3\approx6.6\times10^{-6}\,$ s.*

Go deeper · Derivation C · 2 minSchwarzschild time translation as a Lorentz boost

Derivation CSchwarzschild time translation as a Lorentz boost

The goal is to expose the deepest structural feature of the Kruskal picture: that the innocuous time-translation isometry $t\to t+c$ of the Schwarzschild exterior acts on the maximal extension as a pure Lorentz boost, whose fixed-point set is the horizon and whose scale is the surface gravity. Establishing this turns the horizon from a metric curiosity into a bifurcate Killing horizon and provides the classical skeleton from which the Hawking temperature is later read off, by direct analogy with a Rindler observer in flat space.

Step 1. Apply $t\to t+c$ to the definitions (4). Because $u=t-r_\ast$ and $v=t+r_\ast$ shift by $\mp$ ... by $-c$ and $+c$ respectively... more directly $U=-e^{-(t-r_\ast)/4M}\to U\,e^{-c/4M}$ and $V=e^{(t+r_\ast)/4M}\to V\,e^{c/4M}$ . Why: a real translation of $t$ becomes a reciprocal rescaling of the null coordinates, which is the algebraic signature of a boost, not a translation — the isometry has changed character under the change of chart.

Step 2. Convert to $(T,X)$ using $T=\tfrac12(V+U)$ , $X=\tfrac12(V-U)$ . The reciprocal rescaling of $U,V$ turns into the hyperbolic rotation $T\to T\cosh(c/4M)+X\sinh(c/4M)$ , $X\to X\cosh(c/4M)+T\sinh(c/4M)$ — a boost of rapidity $c/4M$ . Why: this is precisely the transformation of Minkowski coordinates under a Lorentz boost, confirming that $\partial_t$ generates boosts about the origin of the $T$ – $X$ plane.

Step 3. Locate the fixed points. The boost leaves invariant only the lines $T=\pm X$ (mapped to themselves) and the single point $T=X=0$ ; the latter, the bifurcation 2-sphere, is where the Killing field $\xi=\partial_t$ vanishes, and on the lines $\xi$ is null. Its norm is $\xi^\mu\xi_\mu=-(1-2M/r)$ , which is zero on $r=2M$ . Why: a Killing horizon is by definition a null surface on which a Killing field is null, and here the boost structure makes that surface the two-sided (bifurcate) set $T=\pm X$ .

Step 4. Read off the surface gravity from the boost scale: rapidity $=\kappa\,c$ with $\kappa=1/4M$ , and quantizing a field in the smooth Kruskal vacuum restricted to region I gives a thermal spectrum at $T_H=\kappa/2\pi=1/8\pi M$ . Why: the periodicity in imaginary boost angle is $2\pi$ , which sets a temperature $\kappa/2\pi$ — the same Euclidean argument that gives the Unruh temperature of a Rindler observer, transplanted to the black-hole horizon. *Verified: surface gravity $\kappa=1/4M$ and $T_H=1/8\pi M$ (geometric units) match Wald §12.5; the boost matrix has rapidity $c/4M$ as in MTW §31.5.*

Limiting cases

Each region and limit of the Kruskal manifold reduces to a familiar geometry, which is the fastest way to sanity-check the construction.

Case	Result	Meaning
Large radius	$r\gg2M$ , $X^2-T^2\to\infty$	The physical light cones reopen to $45^\circ$ and regions I and III each approach asymptotically flat Minkowski space — two separate asymptotic infinities.
Horizon	$r=2M\ \Leftrightarrow\ UV=0$	The bifurcate Killing horizon: $U=0$ is the future horizon, $V=0$ the past horizon, and their crossing $U=V=0$ is the bifurcation 2-sphere where the boost Killing field vanishes.
Singularity	$r=0\ \Leftrightarrow\ T^2-X^2=1$	Two spacelike hyperbolae (future in II, past in IV); the Kretschmann scalar $48M^2/r^6$ diverges — genuine, coordinate-independent edges of spacetime.
Collapse spacetime	regions I and II only	A star that actually collapses replaces regions III and IV with collapsing matter; by Birkhoff's theorem the Kruskal extension applies only in the exterior vacuum.

Common trapThe Kruskal diagram is the picture of a real black hole

It is not. The full four-region Kruskal–Szekeres manifold is the eternal, time-symmetric vacuum solution — a black hole that was always there and always will be. A real black hole forms by gravitational collapse, and its spacetime has only regions I and II, glued along the future horizon to the interior of the collapsing matter. The white hole (region IV) and the second asymptotic universe (region III) are artefacts of analytically continuing the vacuum solution backward in time, and they are simply absent for an astrophysical hole. A companion trap: the Einstein–Rosen bridge joining I and III on a $T=0$ slice is not a traversable wormhole. Its throat is spacelike and dynamical — it forms and pinches off faster than light — so any timelike or null curve that tries to cross from I to III hits the singularity $r=0$ first. The two exteriors are causally disconnected, forever.

Worked examples

1 · Infall time and Hawking temperature for real black holes (fully worked)

A test object falls radially from rest far away into a Schwarzschild black hole. Using $\Delta\tau=4M/3$ for the horizon-to-centre proper time and $T_H=1/8\pi M$ for the Hawking temperature, evaluate both for a $1\,M_\odot$ hole, for Sgr A\ ( $4.3\times10^6\,M_\odot$ ), and for M87\ ( $6.5\times10^9\,M_\odot$ ). Restore SI units.

Step 1. Convert the geometric result to seconds. In geometric units $\Delta\tau=4M/3$ is a length; the physical time is $\Delta\tau/c$ with $M=GM_{\rm phys}/c^2$ , so $\Delta\tau_{\rm SI}=\tfrac{4}{3}\,GM_{\rm phys}/c^3$ . The base scale $GM_\odot/c^3=4.93\times10^{-6}\,$ s gives, for one solar mass, $\Delta\tau=\tfrac{4}{3}(4.93\times10^{-6})=\mathbf{6.6\times10^{-6}\,s}$ . The horizon-to-singularity fall of a solar-mass hole lasts under seven microseconds of the faller's own time.

Step 2. Scale linearly in $M$ , since $\Delta\tau\propto M$ . For Sgr A\: $6.6\times10^{-6}\times4.3\times10^6=\mathbf{28\,s}$ . For M87\: $6.6\times10^{-6}\times6.5\times10^9=4.3\times10^4\,$ s $=\mathbf{11.9\,hours}$ . A faller crossing M87\*'s horizon has half a day of proper time before the singularity — the bigger the hole, the gentler and longer the interior.

Step 3. Now the Hawking temperature. In SI, $T_H=\hbar c^3/(8\pi G M_{\rm phys}k_B)=6.17\times10^{-8}\,\mathrm{K}\times(M_\odot/M_{\rm phys})$ . For one solar mass, $T_H=\mathbf{6.2\times10^{-8}\,K}$ . For Sgr A\: divide by $4.3\times10^6$ to get $\mathbf{1.4\times10^{-14}\,K}$ . For M87\: $\mathbf{9.5\times10^{-18}\,K}$ .

Step 4. Read the physics. The interior lifetimes are macroscopic and grow with $M$ , while the Hawking temperatures are minuscule and shrink with $M$ — both because the only scale in the problem is $M$ itself, entering as $\Delta\tau\propto M$ and $T_H\propto1/M$ . A solar-mass hole: fall of 6.6 μs, temperature 6.2×10⁻⁸ K; M87\*: fall of 11.9 hours, temperature 9.5×10⁻¹⁸ K — the same surface-gravity scale $\kappa=1/4M$ underlies both.

2 · Locating an event on the Kruskal diagram

Find the Kruskal coordinates $(T,X)$ of the event at $r=4M$ , $t=0$ in region I, and verify it against the implicit relation (6).

Step 1. Compute the tortoise coordinate at $r=4M$ : $r_\ast=4M+2M\ln|4M/2M-1|=4M+2M\ln1=4M$ , since $\ln1=0$ . So $r_\ast=4M$ here.

Step 2. At $t=0$ the null coordinates are $u=t-r_\ast=-4M$ and $v=t+r_\ast=+4M$ . Then $U=-e^{-u/4M}=-e^{1}=-e$ and $V=e^{v/4M}=e^{1}=e$ .

Step 3. Convert: $T=\tfrac12(V+U)=\tfrac12(e-e)=\mathbf{0}$ and $X=\tfrac12(V-U)=\tfrac12(e+e)=\mathbf{e}\approx2.718$ . The event sits on the $T=0$ line, as it must for $t=0$ (the moment-of-time-symmetry slice).

Step 4. Check relation (6): $X^2-T^2=e^2-0=e^2\approx7.389$ , while $(r/2M-1)e^{r/2M}=(2-1)e^{2}=e^2$ . They agree exactly. The event $r=4M$ , $t=0$ maps to $(T,X)=(0,\,e)$ , and $X^2-T^2=e^2$ confirms $r=4M$ via the implicit relation.

Physics of the portrait

Kruskal–Szekeres coordinates are named for two people who never collaborated and published the same idea, in the same year, in journals a world apart — the mark of an idea whose time had fully come.

Origins · 1 minMartin Kruskal and how this was discovered

Martin Kruskal

1925–2006, American mathematician and physicist · maximal analytic extension of the Schwarzschild geometry

Martin David Kruskal was one of the most versatile applied mathematicians of the twentieth century, remembered above all for co-discovering the soliton, but general relativists know his name from a single short paper. By the late 1950s the "Schwarzschild singularity" at $r=2M$ had confused physicists for forty years; Eddington (1924) and Lemaître (1932) had written horizon-regular coordinates without grasping their significance. Kruskal saw how to do it once and for all, constructing in 1960 the exponential null coordinates that render the metric manifestly regular across the horizon and reveal the maximal, four-region extension. Characteristically, the paper was written up and submitted to Physical Review by John Wheeler, who had grasped its importance from Kruskal's private explanation. In the same year, entirely independently, the Hungarian-Australian mathematician George Szekeres published the identical construction in a Hungarian mathematics journal, where it went largely unnoticed by physicists; the joint name records both. The coordinates became a cornerstone of black-hole theory — the setting for Penrose diagrams, the singularity theorems, the bifurcate Killing horizon, and, a decade and a half later, Hawking's derivation of black-hole radiance.

History

1916

— Schwarzschild derives the vacuum solution; the surface r = 2M appears singular

1924

— Eddington writes coordinates regular at r = 2M without recognizing what they show

1935

— Einstein and Rosen find the bridge joining two sheets of the vacuum solution

1958

— Finkelstein reintroduces horizon-penetrating null coordinates, identifying a one-way membrane

1960

— Kruskal and Szekeres independently publish the maximal analytic extension

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §31.5–31.6 · Wald, General Relativity (1984), §6.4 and §12.5 · Carroll, Spacetime and Geometry (2004), §5.6–5.7 · Hartle, Gravity (2003), Ch. 12 · Schutz, A First Course in General Relativity (2nd ed., 2009), §11.3 · Kruskal, Phys. Rev. 119, 1743 (1960); Szekeres, Publ. Math. Debrecen 7, 285 (1960).

Provenance. The equations follow MTW §31.5–31.6, Wald §6.4 and §12.5, and Carroll §5.6–5.7. A verifier should recompute: (1) the tortoise integral $r_\ast=r+2M\ln|r/2M-1|$ ; (2) the cancellation giving $\mathrm{d}s^2=(32M^3/r)e^{-r/2M}(-\mathrm{d}T^2+\mathrm{d}X^2)+r^2\mathrm{d}\Omega^2$ with $F(2M)=16M^2/e$ ; (3) the Kretschmann scalar $48M^2/r^6$ , finite ( $=3/4M^4$ ) at $r=2M$ ; (4) the radial-infall proper time $\Delta\tau=4M/3$ from $2M$ to $0$ , i.e. $6.6\,\mu$ s for $1\,M_\odot$ ; (5) surface gravity $\kappa=1/4M$ and $T_H=1/8\pi M$ . All figures are standard textbook values in geometric units $G=c=1$ .

Prerequisites. The Schwarzschild Metric · Schwarzschild Radius and the Geometric Mass · Geodesics in Schwarzschild Spacetime · Killing Vectors and Conserved Quantities in Schwarzschild · Birkhoff's Theorem. Leads to. Penrose (Carter–Penrose) Diagrams · Hawking Radiation · Eddington–Finkelstein Coordinates · Einstein–Rosen Bridge and Wormholes.

Ask the tutor: Ask me to walk you through why the exponential map $U=-e^{-u/4M}$ is the unique reparametrization that smooths the horizon, or to trace an infalling worldline across all four regions of the diagram.

If Schwarzschild coordinates are so misleading, why do we still use them?

Because outside the horizon they are excellent — they are adapted to static observers and to the time-translation symmetry, so the conserved energy, the redshift, and the orbits all take their simplest form in $(t,r)$ . The chart only fails at and inside $r=2M$ , where no static observer can exist anyway. The right attitude is that different coordinates suit different questions: Schwarzschild coordinates for the exterior physics a telescope sees, Kruskal–Szekeres coordinates for the global causal structure and for anything that crosses the horizon. Neither is "more true"; they cover overlapping patches of one manifold, and Kruskal simply covers more of it.

If nothing special happens locally at the horizon, why can't you escape once inside?

The horizon is locally unremarkable but globally decisive, and these are not in tension. Locally, curvature is finite and a small freely falling laboratory measures nothing unusual as it crosses $U=0$ . Globally, the light cones have tipped so that inside region II every future-directed direction — including the one an outgoing photon takes — leads to larger $T$ and smaller $r$ , terminating on the spacelike singularity. Escape would require a spacelike (faster-than-light) worldline. So the impossibility of escape is not a local force you could fight; it is the shape of the future itself, which is exactly what the $45^\circ$ light cones of the Kruskal diagram display.

Are the white hole and the second universe physically real?

Not for a black hole that formed by collapse, which is every astrophysical one. The full four-region Kruskal manifold is the eternal vacuum solution, obtained by analytically continuing the geometry as far as it mathematically goes. For a real hole, the past horizon, the white hole (region IV), and the second exterior (region III) are replaced by the spacetime of the collapsing star and its earlier history, so they never occur. They are best regarded as a mathematical completion that reveals the symmetry of the vacuum equations, and as a useful idealization — for instance in the thermofield-double interpretation — rather than as places one could visit.

How do Kruskal coordinates differ from Eddington–Finkelstein coordinates?

Eddington–Finkelstein coordinates use one of the null coordinates ( $v$ or $u$ ) together with $r$ , and they already cross one branch of the horizon smoothly — ingoing EF coordinates cover regions I and II, so they are perfect for following infalling matter. But they are not symmetric: they regularize the future horizon while leaving the past horizon singular, and they cover only half the extension. Kruskal–Szekeres coordinates use both exponentiated null coordinates at once, so they regularize the future and past horizons simultaneously and reveal all four regions. EF is the practical half-measure; Kruskal is the full, symmetric, maximal extension.

Does a falling astronaut feel anything at the moment of crossing r = 2M?

Nothing marks the instant of crossing. What the astronaut feels is tidal force — the differential pull between head and feet — set by the curvature, i.e. by the Kretschmann scalar $\sim M/r^3$ , which is finite and smooth at $r=2M$ . For a stellar-mass hole that tidal force is already lethal well outside the horizon; for a supermassive hole like M87\* it is gentle at crossing and only grows dangerous much later, deep inside. In every case the crossing itself is a non-event locally: the astronaut cannot determine the moment of passage by any local measurement, which is the physical meaning of the horizon's regularity.

Why is r a function of T and X rather than a coordinate in its own right?

Because the two Kruskal coordinates $T$ and $X$ have already used up the two-dimensional radial-temporal plane, and $r$ is now a scalar field defined on it. Physically $r$ retains its invariant meaning — the areal radius, so that a 2-sphere at that event has area $4\pi r^2$ — but it is read off the diagram through the implicit relation $(r/2M-1)e^{r/2M}=X^2-T^2$ . Curves of constant $r$ are hyperbolae: timelike hyperbolae in the exteriors, spacelike ones in the interior. That the same $r=0$ becomes a spacelike surface inside is precisely why the singularity is a future moment rather than a central place.

What is the link between Kruskal coordinates and Hawking radiation?

The link is the boost structure. On the Kruskal diagram the Schwarzschild time-translation acts as a Lorentz boost about the bifurcation point, exactly as uniform acceleration acts in flat Minkowski space. Restricting the smooth Kruskal vacuum to region I — the only region an exterior observer can access — traces out the modes behind the horizon and leaves a thermal density matrix, by the same argument that gives an accelerated (Rindler) observer the Unruh temperature. The temperature is set by the boost scale, the surface gravity $\kappa=1/4M$ , yielding $T_H=\kappa/2\pi=1/8\pi M$ . So the classical geometry Kruskal made regular is quite literally the stage on which the quantum radiance is derived.