Embedding Diagrams of the Schwarzschild Geometry

An embedding diagram is a device for making curvature visible. We cannot look at a curved three-geometry directly, so we do the next best thing: take a two-dimensional slice of it, and find a surface sitting in ordinary flat three-space whose intrinsic distances match the slice exactly. The surface then carries the geometry in a form the eye can read. For the Schwarzschild vacuum the natural slice is the equatorial plane at one instant of coordinate time, and the surface that reproduces it is Flamm's paraboloid — the iconic funnel derived by Ludwig Flamm in 1916, the year after Schwarzschild wrote down his solution. What the funnel encodes is a single, coordinate-free fact: near a mass there is more room than flat space allows. Lay rulers end to end from a sphere of circumference $2\pi r_1$ to one of circumference $2\pi r_2$ and you measure more length than $r_2-r_1$ . The paraboloid makes that excess literal — its sloping wall is longer than its flat shadow. Far away the wall lies down flat (asymptotically Euclidean); near the throat at $r=2GM$ it tips toward vertical. But the picture is also a famous trap. It shows only the spatial curvature of one time slice. It says nothing about the time–time part of the metric, which is what actually makes things fall and clocks run slow. The rubber-sheet cartoon of a ball rolling into a well borrows Earth's gravity to move the ball and quietly swaps a spatial funnel for a potential well — two errors at once. Read correctly, the diagram is the cleanest statement in relativity of what "curved space" means, and of what it does not mean.

Key takeaways

An embedding diagram draws a curved 2D slice — here the equatorial plane of a constant-time slice of Schwarzschild — as a surface in flat 3-space whose intrinsic geometry it reproduces exactly.
The slice metric $\mathrm{d}\sigma^2=(1-2GM/r)^{-1}\mathrm{d}r^2+r^2\mathrm{d}\phi^2$ has proper radial distance that grows faster than $\mathrm{d}r$ ; matching it to a surface of revolution $z(r)$ gives Flamm's paraboloid $z=\sqrt{8GM(r-2GM)}$ .
The funnel is asymptotically flat far out and meets the throat $r=2GM$ with a vertical tangent; its Gaussian curvature is $K=-GM/(c^2r^3)$ , negative and saddle-like everywhere.
The picture shows spatial curvature only. Gravitational attraction and redshift live in $g_{tt}$ , which the slice discards — so the funnel does not explain why things fall.
The construction is valid only for $r\ge 2GM$ ; below the horizon a constant- $t$ slice's $g_{rr}$ changes sign and $r$ stops being a spatial coordinate.
Key idea: demand that a surface of revolution $z(r)$ in flat 3-space have the same induced line element as the equatorial Schwarzschild slice and solve for $z(r)$ ; the result, Flamm's paraboloid, is a picture of the excess proper radial distance — of spatial curvature alone.

The core insight

Take a 2D spatial slice of the geometry, demand that a surface of revolution $z(r)$ in flat 3-space have exactly the same induced line element, and solve for $z(r)$ . For Schwarzschild this yields Flamm's paraboloid, a visual encoding of the excess proper radial distance — that is, of spatial curvature alone.

Where this lives today

The embedding picture is no museum piece — it is the standard schematic wherever the spatial geometry near a compact object matters. In black-hole imaging, the Event Horizon Telescope's 2019 image of M87 and its 2022 image of Sagittarius A are shaped by null geodesics of the full near-horizon spacetime — the photon ring follows from the complete metric, often repackaged as the optical metric of the static geometry, and not from the constant-time slice, which accounts for only part of the bending. Equatorial embedding diagrams of the rotating Kerr metric are built the same way and show how the spatial geometry of a slice changes with spin; the ergosphere and the innermost stable circular orbit do not appear in them, since both are properties of the spacetime metric rather than of any one spatial slice. In wormhole physics the two-sheet extension of Flamm's paraboloid is the base geometry that Morris and Thorne modified in 1988 to define a traversable wormhole, quantifying the exotic stress-energy a throat would need to stay open — the geometry behind decades of work on faster-than-light shortcuts and the visuals in films like Interstellar. And the diagram remains a front-line teaching tool precisely because it is a productive trap: getting students to say exactly why the funnel is not a gravity well is one of the cleanest ways to teach that gravity, in the Newtonian limit, is the curvature of time.

The physical picture

Imagine surveying the space around a star with rigid rulers. You measure the circumference of one sphere and get $2\pi r_1$ ; you measure a larger one and get $2\pi r_2$ . Now you lay rulers radially from the inner sphere to the outer one — and you find you need more ruler than $r_2-r_1$ , the plain difference of the two circumferential radii. That extra length has nowhere to go in a flat map, so the map bulges: it becomes a funnel. The first figure is that funnel, drawn as the exact profile $z(r)$ of Flamm's paraboloid, with the throat at $r=2GM$ and a wall that stands at $45^\circ$ where $r=4GM$ .

Figure

Flamm's paraboloid: the exact profile z(r) = 2√(2GM(r − 2GM)) of the equatorial Schwarzschild slice, revolved about the vertical axis. Heights are in units of the Schwarzschild radius R_s = 2GM. The surface meets the throat at r = 2GM with a vertical tangent, stands at 45° where r = 4GM (slope dz/dr = 1), and flattens as z ≈ √(8GM r) far away.

Cut that funnel along a plane through the axis and you see the profile the figure plots. Its meaning is entirely in the excess: a finite coordinate interval $\mathrm{d}r$ maps to a longer stretch of surface, and the stretch grows without bound as the wall goes vertical at the throat. The second figure quantifies the excess with real numbers, comparing the coordinate gap $r_2-r_1$ to the proper distance $\Delta\sigma$ a surveyor actually lays down.

Figure

The excess proper radial distance made numerical. Left: the equatorial plane seen face-on, with circles at the areal radii r = 2GM and r = 4GM. Right: a surveyor laying rulers between them measures the proper distance Δσ = 4.59 GM/c², more than twice the coordinate difference r₂ − r₁ = 2 GM/c². The bars show the ratio Δσ ∕ Δr ≈ 2.30 — the funnel is a picture of exactly this stretch.

The diagram is a snapshot, frozen at one value of Schwarzschild time. It holds no clocks and no force arrows. Two points at different $r$ on the same funnel can have wildly different proper-time rates, but you cannot read that off the shape — it lives in the time–time metric component, which the slice deliberately throws away. A flat space embedded this way would give a plain flat disk; every departure from that disk is the spatial curvature of the gravitational field, and nothing else.

The core equations

Everything starts from the exterior Schwarzschild solution — the unique spherically symmetric vacuum geometry. We work in geometric units $G=c=1$ except where a number needs restoring, so the horizon sits at $r=2GM$ , written $2M$ in the compact form. The areal radius $r$ is defined by area, not by measured distance: a sphere at coordinate $r$ has area $4\pi r^2$ .

\mathrm{d}s^2=-\left(1-\frac{2GM}{r}\right)\mathrm{d}t^2+\left(1-\frac{2GM}{r}\right)^{-1}\mathrm{d}r^2+r^2\left(\mathrm{d}\theta^2+\sin^2\theta\,\mathrm{d}\phi^2\right) \qquad (1)

where: $M$ — mass of the central body; $G,\,c$ — Newton's constant and the speed of light; $r$ — areal (circumferential) radial coordinate, a sphere at $r$ has area $4\pi r^2$ ; $t$ — Schwarzschild time, read by a stationary clock at infinity. Reading: the two coefficients $g_{tt}$ and $g_{rr}$ are reciprocals up to sign; the embedding will keep only the second. To build a picture we freeze time and pick the equatorial plane. Setting $\mathrm{d}t=0$ and $\theta=\pi/2$ (so $\mathrm{d}\theta=0$ , $\sin\theta=1$ ) collapses the four-metric to a positive-definite two-metric — a genuine 2D surface geometry.

\mathrm{d}\sigma^2=\left(1-\frac{2GM}{r}\right)^{-1}\mathrm{d}r^2+r^2\,\mathrm{d}\phi^2 \qquad (2)

where: $\mathrm{d}\sigma$ — proper distance measured within the 2D slice; $\phi$ — azimuthal angle. Reading: the angular part $r^2\mathrm{d}\phi^2$ is exactly Euclidean, so all the curvature is packed into the radial coefficient $g_{rr}=(1-2GM/r)^{-1}>1$ . Now demand a surface of revolution $z=z(r)$ in flat Euclidean 3-space, written in cylindrical coordinates $(r,\phi,z)$ , whose induced line element equals the slice. Restricting the flat metric $\mathrm{d}z^2+\mathrm{d}r^2+r^2\mathrm{d}\phi^2$ to $z=z(r)$ gives the target form.

\mathrm{d}\sigma^2=\left(1+\Big(\frac{\mathrm{d}z}{\mathrm{d}r}\Big)^2\right)\mathrm{d}r^2+r^2\,\mathrm{d}\phi^2 \qquad (3)

where: $z(r)$ — height of the embedding surface above the flat base plane; $(r,\phi,z)$ — cylindrical coordinates of flat 3-space. Reading: the angular parts already agree because both use the same areal $r$ , so only the radial coefficients need matching. Equating the radial coefficients of $(2)$ and $(3)$ is the entire content of the embedding — one algebraic condition on the slope.

1+\left(\frac{\mathrm{d}z}{\mathrm{d}r}\right)^2=\left(1-\frac{2GM}{r}\right)^{-1} \qquad (4)

where: the left side is the flat embedding's radial coefficient; the right side is the Schwarzschild slice's $g_{rr}$ . Reading: the surface must be tilted just enough that its slanted length reproduces the stretched proper distance. Solving for the slope shows immediately why the funnel steepens: the slope diverges as $r\to 2GM^+$ .

\frac{\mathrm{d}z}{\mathrm{d}r}=\left(\frac{2GM/r}{1-2GM/r}\right)^{1/2}=\left(\frac{2GM}{r-2GM}\right)^{1/2} \qquad (5)

where: $\mathrm{d}z/\mathrm{d}r$ — slope of the embedded wall; it equals $1$ (a $45^\circ$ wall) at $r=4GM$ and diverges at the throat $r=2GM$ . Reading: a vertical wall means finite $\mathrm{d}r$ costs unbounded proper distance — the geometric signature of the horizon on this slice. Integrating the slope with $z(2GM)=0$ gives the paraboloid itself.

z(r)=2\sqrt{2GM\,(r-2GM)}=\sqrt{8GM\,(r-2GM)} \qquad (6)

where: $z(r)$ — height profile; squaring gives the parabola $z^2=8GM(r-2GM)$ , hence "paraboloid" for the surface of revolution. Reading: real only for $r\ge 2GM$ , which is precisely the exterior region the slice describes. The physically invariant content is the excess proper radial distance — the number the whole picture exists to convey.

\Delta\sigma=\int_{r_1}^{r_2}\left(1-\frac{2GM}{r}\right)^{-1/2}\mathrm{d}r \;>\; r_2-r_1 \qquad (7)

where: $\Delta\sigma$ — proper distance along a radial ruler between areal radii $r_1$ and $r_2$ . Reading: the integrand exceeds $1$ everywhere outside the horizon, so the measured radial distance always beats the coordinate difference. Finally, the intrinsic curvature of the slice is captured by a single scalar, the Gaussian curvature, which is negative everywhere — the funnel is a saddle, not a bowl.

K=-\frac{GM}{c^2\,r^3}\qquad\left(K=-\frac{M}{r^{3}}\ \text{in } G=c=1\right) \qquad (8)

where: $K$ — intrinsic Gaussian curvature of the embedded surface. Reading: $K<0$ everywhere means geodesics on the funnel diverge like on a saddle; the magnitude falls off as $r^{-3}$ , matching the tidal (Riemann) scale of the full spacetime.

Derivation ADeriving Flamm's paraboloid from the equatorial slice

The goal is to turn the abstract statement "the equatorial slice is curved" into an explicit surface a reader can see, by demanding that a surface of revolution $z(r)$ in flat 3-space carry exactly the induced metric of the constant-time equatorial Schwarzschild geometry, and then solving the resulting first-order equation for the profile. We recover Flamm's 1916 result $z=\sqrt{8GM(r-2GM)}$ and, along the way, see why the construction stops at the horizon rather than continuing inward.

Step 1. Write the slice metric. Fix $t$ and set $\theta=\pi/2$ in equation $(1)$ to get $\mathrm{d}\sigma^2=(1-2GM/r)^{-1}\mathrm{d}r^2+r^2\mathrm{d}\phi^2$ , a positive-definite 2D geometry. Why: only a spacelike 2-surface can be embedded in Euclidean 3-space, and freezing time is what makes the surviving metric positive-definite.

Step 2. Write the flat embedding metric. In cylindrical coordinates a surface $z=z(r)$ inherits $\mathrm{d}\sigma^2=\mathrm{d}z^2+\mathrm{d}r^2+r^2\mathrm{d}\phi^2=(1+z'^2)\mathrm{d}r^2+r^2\mathrm{d}\phi^2$ , with $z'=\mathrm{d}z/\mathrm{d}r$ . Why: choosing the same areal coordinate $r$ for the flat cylinder makes the $r^2\mathrm{d}\phi^2$ terms identical, so the two geometries can differ only in the radial coefficient.

Step 3. Equate the radial coefficients. Matching gives $1+z'^2=(1-2GM/r)^{-1}$ ; the angular parts already agree. Why: isometric embedding means "same line element", and once the angular terms coincide, a single scalar equation carries the whole matching.

Step 4. Solve for the slope. Rearranging, $z'^2=(1-2GM/r)^{-1}-1=2GM/(r-2GM)$ , so $z'=\sqrt{2GM/(r-2GM)}$ (positive root). Why: the right-hand side is positive only for $r>2GM$ and diverges as $r\to 2GM^+$ , which is already the news — the wall goes vertical at the horizon and has no real continuation below it.

Step 5. Integrate. Substituting $u=r-2GM$ gives $z=\sqrt{2GM}\int_0^{r-2GM}u^{-1/2}\mathrm{d}u=2\sqrt{2GM(r-2GM)}$ , with the constant fixed by $z(2GM)=0$ . Why: squaring yields $z^2=8GM(r-2GM)$ , a parabola, so the surface of revolution is a paraboloid — Flamm's 1916 result, reproducing MTW Box 23.2 and Carroll §5.6, and flattening to a plane as $M\to0$ .

Go deeper · Derivation B · 1 minWhy the funnel does not explain why things fall

Derivation BWhy the funnel does not explain why things fall

The goal is to show, from the geodesic equation, that the leading acceleration of a slowly moving body in the Schwarzschild field is set entirely by the time–time metric component $g_{tt}$ — the very component the embedding diagram discards — and not by the spatial curvature the funnel displays. This is the precise sense in which the rubber-sheet cartoon is wrong: it dresses up $g_{rr}$ as if it produced the force, when Newtonian gravity is a statement about the curvature of time.

Step 1. Start from the geodesic equation $\mathrm{d}^2x^\mu/\mathrm{d}\tau^2=-\Gamma^\mu{}_{\alpha\beta}(\mathrm{d}x^\alpha/\mathrm{d}\tau)(\mathrm{d}x^\beta/\mathrm{d}\tau)$ , with $\tau$ the proper time and $\Gamma$ the Christoffel symbols. Why: free fall is geodesic motion, so any "force of gravity" in the Newtonian limit must emerge from a Christoffel symbol, not from an external agent.

Step 2. Take the slow-motion limit. For $v\ll c$ the 4-velocity is dominated by its time component, $\mathrm{d}x^\alpha/\mathrm{d}\tau\approx(\mathrm{d}t/\mathrm{d}\tau,0,0,0)$ , so the radial equation reduces to $\mathrm{d}^2r/\mathrm{d}\tau^2\approx-\Gamma^r{}_{tt}\,(\mathrm{d}t/\mathrm{d}\tau)^2$ . Why: only the $tt$ Christoffel symbol survives, which already tells us the answer will depend on $g_{tt}$ alone.

Step 3. Evaluate the Christoffel symbol. Since the metric is static, $\Gamma^r{}_{tt}=-\tfrac12 g^{rr}\partial_r g_{tt}=-\tfrac12(1-2M/r)\,\partial_r[-(1-2M/r)]=(1-2M/r)\,M/r^2$ . Why: the spatial coefficient $g_{rr}$ enters only through the raising factor $g^{rr}$ and cancels to leading order — the curvature drawn in the funnel makes no leading contribution.

Step 4. Restore units and read off Newton. To leading order $\mathrm{d}^2r/\mathrm{d}\tau^2\approx-GM/r^2$ , exactly Newton's law, sourced by $g_{tt}$ . Why: the acceleration that makes an apple fall comes from the curvature of time; the spatial curvature the embedding shows contributes only at post-Newtonian order (it bends light and shifts Mercury's perihelion, but it does not hold you to the ground).

Go deeper · Derivation C · 1 minThe intrinsic curvature of the slice is a negative saddle

Derivation CThe intrinsic curvature of the slice is a negative saddle

The goal is to compute the Gaussian curvature $K$ of the equatorial slice directly from its 2-metric and confirm the compact result $K=-GM/(c^2r^3)$ , establishing that the funnel is everywhere saddle-shaped and that its curvature carries the same $r^{-3}$ tidal scale as the full spacetime. This turns the qualitative "it looks like a funnel" into a quantitative invariant that is the same in any coordinate system.

Step 1. Read off the slice metric coefficients $A(r)=(1-2GM/r)^{-1}$ and $B(r)=r^2$ from equation $(2)$ , so $\mathrm{d}\sigma^2=A\,\mathrm{d}r^2+B\,\mathrm{d}\phi^2$ . Why: for a diagonal 2-metric depending only on $r$ , the Gaussian curvature has a closed form in $A$ and $B$ , so no full Riemann-tensor machinery is needed.

Step 2. Apply the surface-of-revolution formula $K=-\dfrac{1}{2\sqrt{AB}}\dfrac{\mathrm{d}}{\mathrm{d}r}\!\left(\dfrac{B'}{\sqrt{AB}}\right)$ with $B'=2r$ and $\sqrt{AB}=r/\sqrt{1-2GM/r}$ . Why: this is the standard intrinsic-curvature formula for a metric with only radial dependence, and it uses only first derivatives of $A$ and $B$ .

Step 3. Carry out the derivative. With $B'/\sqrt{AB}=2\sqrt{1-2GM/r}$ its $r$ -derivative is $2GM/r^2\cdot(1-2GM/r)^{-1/2}$ , and multiplying by the prefactor collapses everything to $K=-GM/(c^2r^3)$ (numerically verified against the metric). Why: the reciprocal factors cancel cleanly, leaving a single power law — negative for all $r>0$ , so the surface is a saddle everywhere and never a bowl.

Step 4. Check the limits. As $M\to0$ , $K\to0$ (flat plane); as $r\to\infty$ , $K\to0$ (asymptotic flatness); and the $r^{-3}$ falloff matches the tidal components of the Riemann tensor. Why: the intrinsic curvature must vanish where the geometry is flat and scale like the physical tidal field, both of which it does — a strong consistency check on the whole construction.

Limiting cases

Case	Result	Meaning
Far field, $r\gg 2GM$	$z(r)\approx\sqrt{8GM\,r}$ , slope $\to 0$	Space becomes Euclidean; the funnel flattens into a plane and $K\sim -GM/(c^2r^3)\to 0$ .
At the throat, $r\to 2GM^+$	$\mathrm{d}z/\mathrm{d}r=\sqrt{2GM/(r-2GM)}\to\infty$	The wall goes vertical; the embedding ends smoothly at finite proper distance but cannot continue inward.
Interior, $r<2GM$	$g_{rr}=(1-2GM/r)^{-1}<0$ on a $t$ -slice	The radial direction is no longer spacelike and $r$ becomes timelike; the exterior funnel simply does not describe this region.
Weak field / Newtonian	$2GM/(c^2r)\sim 10^{-6}$ at the Sun's surface	The true paraboloid is almost flat; textbook funnels exaggerate the depth enormously for visibility.
Maximal extension	two funnels joined at $r=2GM$	The Kruskal extension embeds as two asymptotically flat sheets meeting at the throat — the Einstein–Rosen bridge.

Common trapThe funnel is a gravity well that a ball rolls into

It is not. The paraboloid shows only the spatial curvature of one constant-time slice — the excess proper radial distance $g_{rr}$ — and nothing about time. Newtonian attraction and gravitational redshift come from $g_{tt}$ , the time–time component the slice deliberately discards; Derivation 2 shows the leading $-GM/r^2$ acceleration is sourced by $g_{tt}$ alone, with the spatial curvature contributing nothing at that order. The rubber-sheet cartoon fails twice over: it needs a real, external, downward gravity to make the ball roll (assuming what it claims to explain), and it silently reinterprets a spatial-geometry funnel as a potential-energy well. Orbits are timelike geodesics of the full four-dimensional spacetime, not trajectories of balls on a curved surface. Use the diagram to see spatial curvature — never to explain falling.

Worked examples

1 · Excess proper distance near the Sun (weak field)

A radial ruler is laid from the Sun's surface, $r_1=R_\odot=6.96\times10^{8}\ \mathrm{m}$ , out to twice that areal radius, $r_2=2R_\odot$ . By how much does the measured proper distance exceed the coordinate difference $r_2-r_1$ ? Use $GM_\odot/c^2=1.477\ \mathrm{km}$ .

Step 1. Set up the invariant. The proper radial distance is $\Delta\sigma=\int_{r_1}^{r_2}(1-2GM/c^2r)^{-1/2}\mathrm{d}r$ . Because $2GM_\odot/(c^2R_\odot)\approx 4\times10^{-6}\ll 1$ , expand the integrand: $(1-2GM/c^2r)^{-1/2}\approx 1+GM/(c^2r)$ .

Step 2. Integrate the excess. The leading $1$ integrates to the coordinate difference $r_2-r_1$ ; the correction gives the excess $\Delta\sigma-(r_2-r_1)\approx (GM/c^2)\ln(r_2/r_1)=(GM/c^2)\ln 2$ .

Step 3. Put in numbers. $\Delta\sigma-(r_2-r_1)\approx (1.477\ \mathrm{km})(0.6931)=\mathbf{1.02\ km}$ . So a ruler spanning a coordinate gap of $r_2-r_1=6.96\times10^{5}\ \mathrm{km}$ is longer than that by only about a kilometre — a fractional stretch of order $10^{-6}$ , invisible to the eye. The Sun's spatial curvature adds ≈ 1 km of proper radial distance between R⊙ and 2R⊙; the textbook funnel is a caricature blown up by a factor of a million.

2 · The funnel of a solar-mass black hole (strong field)

For a black hole of one solar mass, $R_s=2GM/c^2=2.95\ \mathrm{km}$ , describe the embedding wall between the horizon $r=2GM$ and $r=4GM$ : its slope there, its height above the throat, and the proper distance a surveyor would measure.

Step 1. Slope at $r=4GM$ . From equation $(5)$ , $\mathrm{d}z/\mathrm{d}r=\sqrt{2GM/(r-2GM)}=\sqrt{2GM/2GM}=\mathbf{1}$ — a wall inclined at exactly $45^\circ$ . At the throat itself the slope diverges: the wall is vertical.

Step 2. Height of the wall. From equation $(6)$ , $z(4GM)=2\sqrt{2GM\cdot 2GM}=4GM=2R_s$ . In physical units the rim at $r=4GM$ stands $z=2R_s=\mathbf{5.9\ km}$ above the throat.

Step 3. Proper distance from the horizon. The exact integral of $(7)$ from $2GM$ to $r$ is $\Delta\sigma=\sqrt{r(r-2GM)}+2GM\ln\!\big[(\sqrt r+\sqrt{r-2GM})/\sqrt{2GM}\big]$ . At $r=4GM$ this gives $\Delta\sigma=\mathbf{4.59\,GM/c^2}$ , against a coordinate difference $r_2-r_1=2\,GM/c^2$ — a ratio of $\mathbf{2.30}$ .

Step 4. Read the physics. Even from the horizon out to just $r=4GM$ , a surveyor measures over twice the radial distance a flat map predicts; for one solar mass that is $\Delta\sigma\approx 6.8\ \mathrm{km}$ of proper distance across a $2.95\ \mathrm{km}$ coordinate gap. The strong-field funnel is real geometry, not artistic exaggeration: proper radial distance runs 2.3× the coordinate difference right outside a stellar-mass horizon.

Physics of the portrait

Flamm's paraboloid is the oldest picture in relativity — it predates the word "black hole" by four decades and the word "wormhole" by four decades more. Its author saw, in 1916, both the funnel and the hint of a second sheet on the far side of the throat, without any of the language we now use to describe them.

Origins · 1 minLudwig Flamm and how this was discovered

Ludwig Flamm

1885–1964 · Austrian physicist who first embedded the Schwarzschild geometry

Ludwig Flamm was an Austrian theoretical physicist in Vienna and, by marriage to Boltzmann's daughter, a member of the founding family of statistical mechanics. In 1916, only months after Karl Schwarzschild derived the exterior vacuum solution and Einstein completed general relativity, Flamm published "Beiträge zur Einsteinschen Gravitationstheorie" in the Physikalische Zeitschrift, where he constructed the surface of revolution — now his paraboloid — whose intrinsic geometry reproduces the equatorial Schwarzschild slice. Crucially, Flamm noticed that Schwarzschild's radial equation admits two branches meeting at the throat, an early glimpse of what Einstein and Rosen would formalize in 1935 as a bridge between two sheets and what later physicists called the Schwarzschild wormhole. He spent his career at the Technische Hochschule in Vienna, working on relativity, radioactivity, and technical physics. His single 1916 insight gave general relativity its most reproduced — and most misused — image.

History

1915

— Einstein presents the field equations of general relativity

1916

— Schwarzschild derives the exterior spherical vacuum solution

1916

— Flamm embeds the equatorial slice as a paraboloid of revolution

1935

— Einstein and Rosen join two exterior sheets into a bridge

1962

— Fuller and Wheeler show the bridge pinches off before any signal can cross

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §23.8 and Box 23.2 (embedding diagrams; Flamm's paraboloid) · Wald, General Relativity (1984), Ch. 6 · Carroll, Spacetime and Geometry (2004), §5.6 and §5.7 · Hartle, Gravity (2003), Ch. 7 and §9.3 · Schutz, A First Course in General Relativity (2nd ed., 2009), Ch. 10 · Flamm, L., Physikalische Zeitschrift 17, 448 (1916). The key results are self-consistent throughout: restricting equation $(1)$ to $\mathrm{d}t=0$ , $\theta=\pi/2$ yields the induced 2-metric $(2)$ ; imposing $1+z'^2=(1-2GM/r)^{-1}$ and integrating with $z(2GM)=0$ gives $z=\sqrt{8GM(r-2GM)}$ ; the weak-field geodesic acceleration $-GM/r^2$ follows from $\Gamma^r{}_{tt}$ , i.e. from $g_{tt}$ ; the Gaussian curvature $K=-GM/(c^2r^3)$ follows from the slice 2-metric; and the worked numbers — Sun excess $\approx 1.02$ km between $R_\odot$ and $2R_\odot$ , proper distance $4.59\,GM/c^2$ between $r=2GM$ and $4GM$ (ratio $2.30$ ), throat at $r=2GM$ , sphere area $4\pi r^2$ , and $R_{s,\odot}=2.95$ km — are all standard values. Flamm (1916) is the historical source for the paraboloid.

Prerequisites. The Schwarzschild Metric · The Schwarzschild Radius. Leads to. Photon Orbits and the Photon Sphere · Geodesic Equations in Schwarzschild Spacetime · Gravitational Redshift in Schwarzschild Spacetime.

Ask the tutor: Ask me to walk through why the same slope condition that builds Flamm's funnel also proves the funnel can't be the "gravity well" that makes an apple fall.

Does the embedding diagram show all of the Schwarzschild spacetime's curvature?

No — it shows a small slice of it. The construction freezes time and picks the equatorial plane, so the funnel captures only the intrinsic geometry of that one two-dimensional spatial slice. All the information in the time–time component $g_{tt}$ — gravitational time dilation, redshift, and the Newtonian-limit attraction — is thrown away before the picture is even drawn. Two events at very different radii on the same funnel experience very different clock rates, but nothing in the shape tells you that. The diagram is a faithful map of spatial curvature and a complete blank on temporal curvature, which is exactly why it must never be read as a "gravity well".

Why is the radial coordinate r called the areal radius rather than a distance?

Because in curved space you cannot define a radius by measuring outward from a centre and getting a single agreed number — the proper distance you measure exceeds the coordinate difference, and there is no flat interior to measure from anyway. What survives unambiguously is the area of each sphere of symmetry: it is $4\pi r^2$ , and that equation defines $r$ . So $r$ labels spheres by their size, not by how far away they are. The whole point of the embedding diagram is to display the gap between this areal radius and the measured radial distance: the funnel is precisely the picture of "more proper distance than the change in areal radius accounts for".

What is the throat, and can the paraboloid continue below it?

The throat is the circle at $r=2GM$ where the embedding wall becomes vertical — the slope $\mathrm{d}z/\mathrm{d}r=\sqrt{2GM/(r-2GM)}$ diverges there. It sits at finite proper distance from any exterior point, so the surface simply ends; it does not run off to infinity. You cannot continue the exterior paraboloid to smaller $r$ because for $r<2GM$ the coefficient $g_{rr}$ on a constant-time slice turns negative — the radial direction stops being spacelike and $r$ becomes timelike. The interior is a genuinely different geometry that this slice does not cover. In the maximally extended (Kruskal) picture, a second exterior sheet attaches at the throat, giving the two-funnel Einstein–Rosen bridge.

If the funnel isn't a gravity well, why does the rubber-sheet demo seem to work?

It works as a cheap analogy and fails as physics. A marble circling a weighted rubber sheet does trace rosette-like orbits, but only because real, external, downward gravity pulls it into the dimple and friction bleeds its energy — both imported from outside the analogy. Worse, the sheet's shape is being read as a potential-energy well, whereas Flamm's paraboloid is a map of spatial distance. The demonstration therefore assumes gravity to explain gravity and mislabels space curvature as potential. The honest version keeps the funnel as a picture of spatial curvature only and says plainly that orbits are geodesics of four-dimensional spacetime, dominated in the slow-motion limit by the curvature of time.

Can every spatial slice of every spacetime be drawn as an embedding diagram?

No. Isometrically embedding a 2-surface in flat Euclidean 3-space is possible only when the surface's intrinsic geometry is compatible with such an embedding, and many are not. Some slices have regions whose Gaussian curvature or global shape require a higher-dimensional flat space (four, five, or more dimensions), and others admit no smooth global embedding at all. Even for Schwarzschild the choice of slice matters: the equatorial constant- $t$ slice embeds beautifully as Flamm's paraboloid, but other time slicings give different — sometimes non-embeddable — surfaces. Embedding is a helpful visualization when it exists, not a universal guarantee, which is one reason relativists rely on coordinate-free curvature invariants rather than pictures.

How does this connect to wormholes?

Directly. Flamm already noticed in 1916 that Schwarzschild's radial equation has two branches meeting at the throat, so the natural maximal extension embeds as two asymptotically flat paraboloids joined mouth to mouth — the Einstein–Rosen bridge of 1935, the prototype wormhole. But the Schwarzschild bridge is not traversable: Fuller and Wheeler showed in 1962 that it pinches off before any signal can cross. Morris and Thorne asked in 1988 what geometry a traversable throat would need, modifying the embedding profile and computing the exotic, negative-energy stress-energy required to hold it open. So the same picture that visualizes a black hole's spatial curvature is also the starting canvas for every serious discussion of wormholes.