What Is Spacetime Curvature? Gravity Without a Force
Spacetime curvature explained without the rubber sheet: what curved geometry really means, why free fall feels like nothing, and how we measure it.
Table of Contents
Drop a pen. It falls, and nothing touches it. No string, no magnet, no jet of air — and yet it picks up speed at 9.81 m/s² all the way to the floor. For two and a half centuries the explanation was a force reaching invisibly across empty space. Since 1915 the explanation has been geometry. And almost every popular account of that geometry hands you a picture that is wrong in three separate ways.
The Rubber Sheet Is Lying to You
You have seen it: a taut rubber sheet, a bowling ball sitting in the middle, a marble circling the dent. It is probably the most reproduced illustration in all of physics, and as an explanation it fails completely.
It fails first because it explains gravity by assuming gravity. The bowling ball only makes a dent because something is pulling it downwards, out of the sheet. Put the whole apparatus in orbit and the sheet stays perfectly flat. The picture smuggles in the very thing it claims to account for.
It fails second because it shows you space and leaves out time. Spacetime has four dimensions, and for everyday gravity the curvature of the time direction does nearly all the work — a point we will come back to, because it is the most surprising true thing in the subject. The sheet shows none of it.
It fails third, and worst, because it invites you to imagine spacetime being bent into something: a fifth dimension, a surrounding void, somewhere for the dent to go. General relativity makes no such claim. Its curvature is intrinsic. It is a statement about geometry as measured from inside, by rulers and clocks that never leave. There is no outside.
Curvature Means the Rules of Geometry Change
So what does curvature mean, if not a dent in a surrounding space? It means the theorems you learned in school stop holding.
Imagine a flat creature living on the surface of a sphere, with no access to any third dimension. It sets off from the equator alongside a companion a thousand kilometres away, both walking due north, both keeping their paths as straight as any local measurement can confirm. Flat geometry promises that parallel lines stay the same distance apart forever. These two meet at the pole. The creature draws a triangle from the pole down to two points on the equator and finds the angles sum to more than 180°. It draws a circle, measures the circumference and the radius along the surface, and gets a ratio smaller than 2π.
Every one of those is a discovery made without leaving the surface. That is what intrinsic curvature is, and it is the only kind general relativity uses. The sphere is the honest two-dimensional analogue — not because spacetime resembles a sphere, but because its curvature is a fact about measurements taken inside it. The mathematical object that records distances and angles from the inside is the metric tensor, written g with two indices; the object that records how that geometry fails to be flat is the Riemann curvature tensor.
Free Fall Feels Like Nothing at All
Now the step that made the whole theory possible. Einstein later called it the happiest thought of his life: a person in free fall does not feel their own weight.
Seal yourself in a windowless box and cut the cable. Let go of a ball and it hangs beside you. Tip a bottle and the water drifts away in a sphere. Spin a gyroscope, time a pendulum, run any local experiment you like, and every result matches what you would get drifting far from any mass at all. Gravity in that box is not hidden and not merely cancelled. It is absent.
That is the equivalence principle, and it has a brutal consequence. If a single free-falling observer can find no trace of gravity anywhere in the laboratory, then gravity cannot be something that exists at a point. This is why treating gravity as a force eventually had to give way, and it is the hinge on which Einstein’s general theory turns.
What One Observer Cannot See, Two Can
Put two observers in free fall instead of one, released side by side, and everything changes.
Both fall towards the centre of the Earth, which means their paths are not parallel but convergent. The gap between them closes, slowly at first and then faster, and neither of them did anything to cause it. Release them one above the other instead and the gap opens, because the lower one sits in a slightly steeper part of the field and runs away downwards.
That relative acceleration between neighbouring free-falling bodies is spacetime curvature. Not an illustration of it — the operational definition of it. In Newtonian language it is the gradient of the field, the same quantity that raises the ocean tides and stretches moons: a transverse squeeze of GMd/r³ and a radial stretch of exactly twice that, 2GMd/r³, for two bodies separated by d at distance r from a mass M. In general relativity those numbers are components of the Riemann tensor. The tide is curvature you can watch with your own eyes, twice a day.
Geodesics, and Why Nothing Pulls the Moon
If no force acts on a falling body, what decides where it goes? The geometry does. A free object follows a geodesic — the straightest path the curved spacetime permits, the four-dimensional equivalent of a great circle on the sphere.
Nothing pulls the Moon. Its worldline through spacetime is as straight as the geometry around the Earth allows, and the closed ellipse we see is what that straight line looks like when you project away the time direction and watch only the space part. The same is true of every satellite and every planet, which is why an orbit is better described as permanent falling than as a balance of forces. Light does the same thing on null geodesics, and when it passes a galaxy cluster the result is gravitational lensing — images smeared into arcs by geometry alone.
Mostly It Is Time That Curves
Here is the part almost no popular account mentions. For slow-moving objects, essentially everything we call gravity comes from the curvature of time, not of space.
The reason is a matter of scale. In one second, a thrown ball moves perhaps ten metres through space — and c × 1 s, about 300 million metres, through the time direction. A distortion of the time direction therefore has three hundred million metres of leverage for every ten the spatial distortion gets, and the ratio of the two contributions goes as the square of that. Space near the Earth is flat to an extraordinary degree. Time is not: clocks higher up genuinely run faster, and a thrown ball follows the trajectory that maximises its own elapsed time. The parabola you drew in school is an optimisation of clock rates, nothing more.
Light is the clean test case, because light has no slow regime. Treat gravity as acting only through time and you recover the old Newtonian answer for starlight bending at the Sun: about 0.87 arcseconds. General relativity predicts about 1.75. The missing half comes from the curvature of space, and the 1919 eclipse measurements came down on the side of the larger number.
The Equation, and the Numbers That Settled It
All of this is held together by ten coupled equations, conventionally written in one line:
G_μν + Λg_μν = (8πG/c⁴) T_μν
The left side is pure geometry: G_μν is the Einstein tensor, built from the Riemann curvature of the metric g_μν, and Λ is the cosmological constant. The right side is matter and energy: T_μν is the stress-energy tensor, carrying energy density, momentum and pressure. Geometry on the left, contents on the right, and the constant 8πG/c⁴ fixing the exchange rate. That form, with the signature convention (−,+,+,+), is the one Misner, Thorne and Wheeler standardised; other books move Λ or flip signs, so it is always worth checking conventions before comparing two sources. You can see the terms broken out on our Einstein field equations page.
Three numbers made the case. Mercury’s perihelion advances about 43 arcseconds per century faster than the pull of the other planets can explain — a residual Urbain Le Verrier identified in 1859, and the figure Einstein reproduced in November 1915, days before he published the final field equations. Starlight grazing the solar limb is deflected by about 1.75 arcseconds, twice the Newtonian value. And every GPS satellite carries a clock that gains roughly 45.9 microseconds a day from weaker gravity and loses about 7.2 from its orbital speed, for a net gain near 38.7 microseconds a day; the oscillators are detuned before launch so that the navigation system works at all. Curvature is in your phone.
Geometry Is Not a Metaphor
The strongest argument that spacetime geometry is physical rather than a bookkeeping device arrived in 2015, when LIGO recorded two black holes merging and the length of a four-kilometre arm changed by a thousandth of a proton’s width. Gravitational waves are curvature travelling on its own, carrying energy, with no matter involved at any point. Geometry turned out to be a medium.
Push the curvature far enough and the geometry stops being a correction to Newton and starts dictating the terms: inside a black hole’s horizon, every future-directed geodesic leads inwards. No force is required for that either. It is simply where the straight lines go.
Frequently Asked Questions
What is spacetime curvature?
Spacetime curvature is the measurable failure of flat geometry inside spacetime: initially parallel paths converge or diverge, the angles of a triangle no longer sum to 180 degrees, and a circle's circumference is no longer 2 pi times its radius. It is an intrinsic property, meaning it is detectable entirely from the inside by rulers and clocks, and it does not require spacetime to be bent into any higher dimension. Operationally, curvature is the relative acceleration between two neighbouring objects in free fall. Release two balls side by side above the Earth and they drift together at a rate of GMd over r cubed; release them one above the other and they drift apart at twice that rate. Neither ball feels any force. Mass and energy set the curvature through the Einstein field equations, and freely moving objects follow the straightest paths the curved geometry allows.
Why is the rubber sheet analogy wrong?
The rubber sheet with a bowling ball in it fails for three reasons. First, it explains gravity by assuming gravity: the ball only sags because a real downward pull acts on it, so the picture quietly uses the thing it claims to explain. Put the sheet and ball in orbit and the sheet stays flat. Second, it shows only space and omits time, yet for everyday gravity it is the curvature of the time direction that does almost all the work; a thrown ball's arc comes mostly from clocks ticking at different rates at different heights. Third, it suggests spacetime is bent into a surrounding higher dimension, which general relativity never claims. Curvature in general relativity is intrinsic, defined by measurements made inside spacetime. A better two-dimensional analogue is the surface of a sphere, whose curvature a flat creature living on it could establish without ever leaving it.
How is spacetime curvature measured?
Spacetime curvature is measured through the relative motion of free-falling bodies and the bending of light, because a single free-falling observer detects nothing at all. Two test masses released side by side converge at GMd over r cubed and two released radially separate at 2GMd over r cubed, and that tidal drift is the direct observable. Historically three measurements pinned it down. Mercury's perihelion advances by about 43 arcseconds per century more than Newtonian planetary perturbations predict, a residual Le Verrier identified in 1859 and Einstein accounted for in November 1915. Starlight grazing the Sun's limb is deflected by about 1.75 arcseconds, exactly twice the value a Newtonian corpuscle calculation gives. And GPS satellite clocks gain roughly 45.9 microseconds per day from weaker gravity while losing about 7.2 from orbital speed, a net gain near 38.7 microseconds per day that the system corrects for continuously.
Does space or time curve more in Earth's gravity?
For slow-moving objects in Earth's gravity, the curvature of time dominates overwhelmingly and the curvature of space contributes almost nothing. The reason is that a slowly moving object travels an enormous distance through the time direction for every metre it moves through space, so even a minuscule distortion of the time direction accumulates into a large effect while the spatial distortion barely registers. A thrown ball's parabola is set almost entirely by the fact that clocks higher up run faster than clocks lower down, and the ball follows the path that maximises its own elapsed time. The balance shifts only as speeds approach that of light. Light itself is the clean test case: the time part of the geometry alone reproduces the old Newtonian deflection of about 0.87 arcseconds at the Sun's limb, while general relativity predicts about 1.75 arcseconds. The missing half comes from the curvature of space.