Tidal Forces and the Limits of the Equivalence Principle

The Sun pulls on the Earth 178 times harder than the Moon does. The Moon raises a tide twice as big. That is not a paradox, a rounding error, or a fact about oceans. It is the reason general relativity exists. A tide is not a pull. It is a difference of pulls — and differences of pulls do not obey the inverse-square law. They obey an inverse-cube law, because a difference is a derivative, and differentiating $1/d^2$ gives $1/d^3$ .

\frac{\text{tide}_{\text{Moon}}}{\text{tide}_{\text{Sun}}} \;=\; \frac{M_{\text{M}}}{M_{\text{S}}}\left(\frac{d_{\text{S}}}{d_{\text{M}}}\right)^{3} \;=\; \mathbf{2.18} \qquad\text{while}\qquad \frac{\text{pull}_{\text{Sun}}}{\text{pull}_{\text{Moon}}} \;=\; \mathbf{178.8}

`aq-sr-equivalence-principle` proved that free fall removes the pull — exactly, at a point, with no residue. It removes the $178$ . What it cannot remove is the $2.18$ , because a coordinate change can subtract a constant from an acceleration field but cannot subtract a gradient from it. That leftover gradient is a tensor, and a tensor that is non-zero in one frame is non-zero in every frame, forever. So the boundary of special relativity is not acceleration, and it is not gravity. It is the second derivative of the potential — and the ocean has been demonstrating this twice a day since before there was anyone to watch.

Key takeaways

Signature $(+,-,-,-)$ : $\mathrm{d}s^2 = c^2\mathrm{d}t^2 - \mathrm{d}x^2$ , timelike positive.
Scope, up front. The tidal tensor $E_{ij} = \partial_i\partial_j\Phi$ , its eigenvalues $(-2GM/r^3, +GM/r^3, +GM/r^3)$ , its tracelessness, and the expiry inequality $L T^2 < \delta r^3/GM$ ("a 1 m lab is inertial for 0.81 s") are owned by `aq-sr-equivalence-principle` — recalled by name here, never re-derived. This article goes past them.
Owned — THE INVERSION. Tides go as $M/d^3$ , not $M/d^2$ . So the Sun outpulls the Moon by $\mathbf{178.8\times}$ and loses the tide contest $\mathbf{2.18}$ to $1$ . Lunar tide $1.10\times10^{-6}$ m/s²; solar tide $5.05\times10^{-7}$ m/s². The ocean answers to the nearer body, not the stronger one.
Owned — geodesic deviation, derived. Two neighbouring free-fallers, a separation vector $\xi^\mu$ , and the commutator of covariant derivatives give $\boxed{\dfrac{D^2\xi^\mu}{\mathrm{d}\tau^2} = -R^{\mu}{}_{\alpha\nu\beta}u^\alpha \xi^\nu u^\beta}$ . The covariant statement of a tide — and what makes curvature an observable rather than a definition.
Owned — spaghettification, and it runs backwards. At a horizon the tidal acceleration across a body of length $L$ is $\boxed{a_{\text{tidal}} = \dfrac{L\,c^{6}}{4G^{2}M^{2}}} \propto \mathbf{1/M^2}$ . A bigger hole has a GENTLER horizon. A $10\,M_\odot$ hole tears a person apart at $1.9\times10^{7}\,g$ ; at Sgr A* ( $4.3\times10^{6}\,M_\odot$ ) the horizon tide over a human body is $\mathbf{1.0\times10^{-4}\,g}$ — you cross it without noticing. Crossover: $\mathbf{4.3\times10^{4}\,M_\odot}$ .
Owned — a uniform field has EXACTLY ZERO tide. $\partial_i\partial_j(gz) = 0$ , identically. That is why no rocket can fake gravity, and why the Rindler chart is flat. The tide is the one thing acceleration cannot counterfeit — which is precisely why it, and not acceleration, is the border.
⚠ The sign trap. The eigenvalues of $E_{ij}$ and the relative accelerations they produce are negatives of each other ( $\ddot\xi^i = -E_{ij}\xi^j$ ). Print the tensor's eigenvalues, call them accelerations, and you have swapped stretch for squeeze. The base of `aq-sr-failure-global-inertial-frames` does exactly this.
Key idea: free fall subtracts a constant from the gravitational field. It cannot subtract a gradient. All of general relativity is the study of what is left.

The core insight

Gravity is the only force that can be switched off by falling — but only its value, and only at one point. Move an inch and the field has changed slightly, and that change is still there, and nothing you do to your frame will touch it. So the honest description of gravity is not the field but the field's derivative, and the honest question is never "how strong is gravity here?" but "how differently strong is it an inch away?" The Moon beats the Sun on that question by two while losing the first by a hundred and seventy-eight. The ocean, which cannot read, gets the right answer twice a day.

Where this lives today

Every gravitational-wave detector is a tidal instrument. A passing wave carries no force: it carries a travelling ripple in $R^{i}{}_{0j0}$ , and LIGO's mirrors are free-falling test masses whose separation it modulates. The measurement is the one this article describes — two freely-falling bodies and the drift between them that no change of frame removes — at $10^{-19}$ m over $4$ km.

And the tide shapes the solar system. The same $1/d^3$ law that lets the Moon outvote the Sun tears comets apart at the Roche limit, keeps Saturn's rings from coalescing, heats Io until it is the most volcanically active body in the solar system, and has locked the Moon's rotation to its orbit. Tidal dissipation pushes the Moon away at $3.8$ cm a year and lengthens our day — all of it $\partial_i\partial_j\Phi$ , doing the one thing gravity does that free fall cannot undo.

Gravity gradiometry maps what an accelerometer cannot see. An accelerometer in free fall reads zero — that is the equivalence principle. A gradiometer, measuring the difference between two accelerometers, reads $E_{ij}$ , which is not removable. That is why the GOCE satellite mapped Earth's geoid from orbit while in free fall the whole time, and why gradiometry finds ore bodies. The instrument exists because the tide is the part of gravity you cannot fall away from.

The physical picture

Drop a laboratory and everything inside floats — `aq-sr-equivalence-principle`'s subject, exactly true at a point. Now make the laboratory bigger and watch what leaks in. Release two balls side by side. Both fall toward the centre of the Earth, so their paths converge and in the falling frame they drift together. Release two balls one above the other. The lower is in a stronger field, so it falls faster, and they drift apart. Nothing you do to the box — no boost, no rotation, no coordinate change in existence — removes those drifts. They are the fingerprint of a real mass, and what an accelerating rocket cannot fake.

Why can it not fake them? Because a uniform field has no tide at all, and the calculation is one line. For $\Phi = gz$ ,

E_{ij} = \partial_i\partial_j (gz) = \partial_i (g\,\delta_{jz}) = \mathbf{0} \qquad \text{identically, every component.}

A uniform field is exactly a coordinate choice. Its tidal tensor vanishes, its Riemann tensor vanishes (`aq-sr-rindler-metric` computes both and gets zero), and a rocket burning at $9.8\ \mathrm{m/s^2}$ in deep space reproduces every gravitational effect there is except this one. The tide is the sole survivor — and that is the entire reason it, and not acceleration, marks the border.

Now the ocean. The Earth is in free fall around the Moon: the whole planet accelerates toward it at $GM_{\text{M}}/d^2$ , and by the equivalence principle that acceleration is invisible — it is the part that cancels. What does not cancel is that the near side is closer and the far side further. The near ocean is pulled harder than the Earth's centre is; the far ocean is pulled less hard, so relative to the falling Earth it is left behind — flung outward. Two bulges. Two high tides a day. The far bulge is not the Moon pushing; it is the Moon failing to pull hard enough.

Figure

THE INVERSION — the single most counter-intuitive fact in gravitation, and the ocean settles it twice a day. Top: the DIRECT PULL of the Sun on the Earth is 5.93 × 10⁻³ m/s², and of the Moon 3.32 × 10⁻⁵ m/s². The Sun wins by a factor of 178.8. It is not close. Bottom: the TIDE-RAISING acceleration — the DIFFERENCE in pull across the Earth's diameter — is 1.10 × 10⁻⁶ m/s² for the Moon and only 5.05 × 10⁻⁷ m/s² for the Sun. The MOON wins, by 2.18 to 1. The winner has changed. Nothing about the masses or the distances changed; only the question did. A pull falls off as M/d², but a DIFFERENCE of pulls is a derivative, and differentiating 1/d² gives 1/d³ — so the tide carries an extra factor of R/d and the near body is rewarded enormously for being near. The Moon is 390 times closer than the Sun and 27 million times lighter, and 390³ beats 27 million. THIS IS WHY GENERAL RELATIVITY EXISTS: free fall cancels the 178, exactly and forever, because it is a constant. It cannot cancel the 2.18, because that is a GRADIENT — and a gradient is a tensor, and a tensor that is nonzero in one frame is nonzero in every frame that will ever be built

The core equations

\boxed{\;a_{\text{tidal}} \;=\; \frac{2GM}{d^{3}}\,R\;} \qquad\text{versus}\qquad a_{\text{pull}} \;=\; \frac{GM}{d^{2}}

where: $M$ is the perturbing body's mass, $d$ its distance, $R$ the size of the body being tidally stressed. Physical reading: the two laws that produce the inversion. The pull carries $1/d^2$ ; the tide carries $1/d^3$ , because it is a difference of pulls across a span $R$ , and a difference is a derivative. The extra factor $R/d$ is what rewards proximity so brutally. The tidal tensor whose radial eigenvalue this is — $E_{ij} = \partial_i\partial_j\Phi$ , spectrum $(-2GM/r^3, +GM/r^3, +GM/r^3)$ — is owned by `aq-sr-equivalence-principle`.

\boxed{\;\ddot{\xi}^{\,i} \;=\; -\,E_{ij}\,\xi^{j} \;=\; -\,\partial_i\partial_j\Phi\;\xi^{j}\;}

where: $\xi^i$ is the separation between two freely-falling particles. Physical reading: the sign trap, and it is worth a warning of its own. The eigenvalues of $E_{ij}$ are $(-2GM/r^3, +GM/r^3, +GM/r^3)$ ; the relative accelerations they produce are the negatives of those, $(+2GM/r^3\,\xi_r,\ -GM/r^3\,\xi_\perp,\ -GM/r^3\,\xi_\perp)$ — stretch radially, squeeze transversely. The minus sign in $\ddot\xi = -E\xi$ separates the tensor from the motion, and printing one while naming the other swaps stretch for squeeze.

\boxed{\;\frac{D^{2}\xi^{\mu}}{\mathrm{d}\tau^{2}} \;=\; -\,R^{\mu}{}_{\alpha\nu\beta}\,u^{\alpha}\,\xi^{\nu}\,u^{\beta}\;}

where: $u^\alpha$ is the four-velocity of the reference geodesic, $R^{\mu}{}_{\alpha\nu\beta}$ the Riemann tensor. Physical reading: geodesic deviation — the covariant statement of a tide, and the reason curvature is measurable rather than merely definable. Derived below. In the Newtonian limit it collapses to the line above, fixing $R^{i}{}_{0j0} = (1/c^2)\partial_i\partial_j\Phi$ (`aq-sr-equivalence-principle`'s card).

\boxed{\;a_{\text{tidal}}(r_s) \;=\; \frac{2GM}{r_s^{3}}\,L \;=\; \frac{L\,c^{6}}{4G^{2}M^{2}}\;}\qquad \propto \frac{1}{M^{2}}

where: $r_s = 2GM/c^2$ is the Schwarzschild radius and $L$ the length of the infalling body. Physical reading: spaghettification, upside-down. The horizon tide falls as $1/M^2$ , so the more monstrous the hole, the gentler its horizon. Note the $L$ : the quantity is an acceleration, and dropping it — as this article's base does mid-derivation — leaves units of $\mathrm{s^{-2}}$ .

Where the tide comes from

Derivation AGeodesic deviation — how a tide becomes a tensor

The Newtonian tidal tensor is a fine object but it is defeasible: a Newtonian could insist that $\partial_i\partial_j\Phi$ is a statement about a force field, and force fields are frame-dependent things one might hope to argue away. The goal of this card is to redo the calculation with no force field at all — nothing but two freely-falling worldlines and the geometry between them — and to show that the tide survives, now wearing a tensor. That is the step that makes curvature an observable, and `aq-sr-equivalence-principle` states the result without deriving it. Everything below follows from two facts, and no forces.

Step 1. Set up a family of geodesics. Let $x^\mu(\tau, s)$ be a smooth one-parameter family of freely-falling worldlines: $\tau$ runs along each, $s$ labels which one. Define

u^{\mu} = \frac{\partial x^{\mu}}{\partial \tau} \quad\text{(the four-velocity)}, \qquad \xi^{\mu} = \frac{\partial x^{\mu}}{\partial s} \quad\text{(the separation vector).}

Why: $\xi^\mu$ is the honest, coordinate-free version of "the gap between two neighbouring free-fallers", and $\ddot\xi$ is what a tide is. No forces have been mentioned and none will be.

Step 2. Note the one structural fact we need. Because $u$ and $\xi$ are coordinate vector fields of the same map, they commute:

[u,\xi]^{\mu} = 0 \qquad\Longleftrightarrow\qquad \nabla_u \xi^{\mu} = \nabla_\xi u^{\mu} .

Why: this is $\partial^2 x/\partial\tau\partial s = \partial^2 x/\partial s\,\partial\tau$ promoted to covariant derivatives — legitimate because the connection is torsion-free. It is the only input beyond "the worldlines are geodesics".

Step 3. Differentiate twice along the flow and swap. The relative acceleration is $\nabla_u\nabla_u\xi$ ; using Step 2 once,

\nabla_u \nabla_u \xi^{\mu} \;=\; \nabla_u \nabla_\xi u^{\mu} \;=\; \nabla_\xi \nabla_u u^{\mu} \;+\; \big[\nabla_u, \nabla_\xi\big] u^{\mu} .

Why: we have deliberately manufactured a commutator of covariant derivatives — which is, by definition, the Riemann tensor. The curvature was not inserted; it was produced.

Step 4. Kill the first term and name the second. The worldlines are geodesics, so $\nabla_u u^\mu = 0$ identically and the first term vanishes. The second is the definition of Riemann acting on the vector $u$ :

\big[\nabla_u, \nabla_\xi\big] u^{\mu} \;=\; -\,R^{\mu}{}_{\alpha\nu\beta}\,u^{\alpha}\,\xi^{\nu}\,u^{\beta} .

\boxed{\;\frac{D^{2}\xi^{\mu}}{\mathrm{d}\tau^{2}} \;=\; -\,R^{\mu}{}_{\alpha\nu\beta}\,u^{\alpha}\,\xi^{\nu}\,u^{\beta}\;}

Why: read what has happened. Two particles in free fall, both feeling nothing — each a perfectly good local inertial frame, neither with any right to accelerate. Yet they accelerate relative to each other, and the coefficient is the Riemann tensor. Curvature is not a definition; it is a measurement, and the instrument is two falling stones and a ruler.

Step 5. Sanity-check against Newton. For a slow observer $u^\alpha \approx (c, 0,0,0)$ only $R^{i}{}_{0j0}$ survives, and $\mathrm{d}^2\xi^i/\mathrm{d}t^2 = -c^2 R^{i}{}_{0j0}\xi^j$ . Matching the Newtonian $\ddot\xi^i = -\partial_i\partial_j\Phi\,\xi^j$ fixes $R^{i}{}_{0j0} = (1/c^2)\,\partial_i\partial_j\Phi$ — `aq-sr-equivalence-principle`'s boxed card, recovered from a purely geometric argument. Why: and now the theorem bites. In flat spacetime $R^{\mu}{}_{\alpha\nu\beta} \equiv 0$ , so $D^2\xi/\mathrm{d}\tau^2 \equiv 0$ : special relativity is incapable of producing a tide, in any chart, ever. Not "it is difficult". A tensor that vanishes in one frame vanishes in all of them. Measure a tide and you have left special relativity for good. *Verified in verify_aq-sr-tidal-forces-equivalence-limits.py: sympy builds the Riemann tensor from a general metric, confirms $[\nabla_\mu,\nabla_\nu]V^\rho = R^\rho{}_{\sigma\mu\nu}V^\sigma$ from the definition, and verifies that the Rindler chart — non-trivial metric, non-zero Christoffels — has every Riemann component identically zero, so its geodesic deviation vanishes.*

Go deeper · Derivation B · 2 minThe inversion — why the Moon beats the Sun by two while losing by a hundred and seventy-eight

Derivation BThe inversion — why the Moon beats the Sun by two while losing by a hundred and seventy-eight

Every schoolchild is told the Moon causes the tides, and most of them notice at some point that the Sun is vastly more massive and wonder why it does not win. The goal of this card is to show that the question contains its own answer, and that the resolution is not a detail of oceanography but the exact reason free fall cannot remove gravity over a finite region — which is to say, the exact reason general relativity exists.

Step 1. Write the two quantities and notice they differ. The pull on the Earth's centre from a body of mass $M$ at distance $d$ is $GM/d^2$ . The tide is the difference between the pull at the near surface and at the centre, across a span $R_\oplus$ :

a_{\text{tidal}} \;=\; \frac{GM}{(d - R_\oplus)^{2}} - \frac{GM}{d^{2}} .

Why: the first is what the equivalence principle removes. The whole Earth falls toward the Moon at $GM/d^2$ , and free fall makes that term invisible — it is exactly the "uniform part" a change of frame subtracts. What survives is the difference.

Step 2. Expand for $R_\oplus \ll d$ , using $(1 - x)^{-2} \approx 1 + 2x$ :

a_{\text{tidal}} \;\approx\; \frac{GM}{d^{2}}\left(1 + \frac{2R_\oplus}{d}\right) - \frac{GM}{d^{2}} \;=\; \boxed{\;\frac{2GM}{d^{3}}\,R_\oplus\;}

Why: there is the cube, and it came from a derivative. The tide is $-\mathrm{d}/\mathrm{d}d$ of the pull, times $R_\oplus$ — and differentiating $d^{-2}$ produces $d^{-3}$ . The extra $1/d$ is not a correction. It is the whole story.

Step 3. Take the two ratios. With $M_{\text{S}}/M_{\text{M}} = 2.71\times10^{7}$ and $d_{\text{S}}/d_{\text{M}} = 389.2$ :

\frac{\text{pull}_{\text{S}}}{\text{pull}_{\text{M}}} = \frac{M_{\text{S}}}{M_{\text{M}}}\left(\frac{d_{\text{M}}}{d_{\text{S}}}\right)^{2} = \frac{2.71\times10^{7}}{1.51\times10^{5}} = \mathbf{178.8}

\frac{\text{tide}_{\text{M}}}{\text{tide}_{\text{S}}} = \frac{M_{\text{M}}}{M_{\text{S}}}\left(\frac{d_{\text{S}}}{d_{\text{M}}}\right)^{3} = \frac{5.89\times10^{7}}{2.71\times10^{7}} = \mathbf{2.18}

Why: the winner swaps, and the reason is one power of $d$ . $389^2 = 1.5\times10^5$ is not enough to overcome the Sun's $2.7\times10^7$ ; $389^3 = 5.9\times10^7$ is. The Moon buys its victory entirely with proximity, cubed.

Step 4. Put numbers on the accelerations themselves.

a_{\text{tidal}}^{\text{Moon}} = \mathbf{1.10\times10^{-6}\ m/s^{2}}, \qquad a_{\text{tidal}}^{\text{Sun}} = \mathbf{5.05\times10^{-7}\ m/s^{2}} .

Against the pulls: $\mathbf{5.93\times10^{-3}}$ (Sun) and $\mathbf{3.32\times10^{-5}}$ (Moon) m/s². Why: note the scale. These are $10^{-7}$ – $10^{-6}$ m/s² — a ten-millionth of $g$ — and they move the entire Pacific twice a day. A tide is not strong. It is inescapable, which is a different property, and the one that matters.

Step 5. Draw the conclusion. Free fall cancels the $178$ — exactly, at every instant, because it is a constant over the falling body. It cannot cancel the $2.18$ , because that is a gradient, and a coordinate change transforms a gradient, never deletes it.

\textbf{Free fall subtracts a constant. It cannot subtract a derivative.}

Why: that sentence is the boundary of special relativity, and every other statement in this chapter paraphrases it. `aq-sr-equivalence-principle` prices the failure ( $LT^2 < \delta r^3/GM$ — a metre lab is inertial for $0.81$ s); this card explains why there is a failure to price. *Verified in verify_aq-sr-tidal-forces-equivalence-limits.py: the expansion $GM/(d-R)^2 - GM/d^2 = 2GMR/d^3 + O(R^2/d^4)$ is checked symbolically; both ratios are recomputed at fifty digits from IAU/CODATA values, giving 2.176 and 178.8; and an artefact guard perturbs $M_{\text{Moon}}$ and $d_{\text{Moon}}$ and confirms the tide moves as $M$ and as $1/d^3$ .*

Go deeper · Derivation C · 2 minSpaghettification, and why a bigger black hole is a safer one

Derivation CSpaghettification, and why a bigger black hole is a safer one

The one thing everybody knows about black holes is that they tear you apart. True, and it is a tidal effect — but the scaling is the reverse of what everybody assumes, and the reversal follows directly from the $1/d^3$ law derived above. The goal here is to compute the tide at the horizon and find that it falls with the mass of the hole, so fast that a large enough black hole has a horizon you could cross without feeling anything at all. This is also where this article's base breaks, and the break is instructive.

Step 1. Locate the horizon: $r_s = 2GM/c^2$ , which is $\mathbf{29.5\ km}$ for $10\,M_\odot$ and $\mathbf{0.085\ AU}$ for Sgr A ( $4.3\times10^{6}\,M_\odot$ ). Why: the horizon radius grows linearly with $M$ — and the tide falls as $1/r^3$ . $M^1$ against $M^{-3}$ is going to be a rout.*

Step 2. Evaluate the radial tide at the horizon, across a body of length $L$ :

a_{\text{tidal}} \;=\; \frac{2GM}{r_s^{3}}\,L .

Why: the tidal tensor's radial eigenvalue $2GM/r^3$ , at $r = r_s$ , across $L$ . (The exact Schwarzschild result differs by an $O(1)$ factor that does not touch the scaling, which is what this card is about.)

Step 3. Substitute $r_s = 2GM/c^2$ and watch the mass collapse:

a_{\text{tidal}} = 2GM\,L \left(\frac{c^{2}}{2GM}\right)^{3} = \frac{2GM\,L\,c^{6}}{8G^{3}M^{3}} = \boxed{\;\frac{L\,c^{6}}{4G^{2}M^{2}}\;} \;\propto\; \frac{1}{M^{2}} .

Why: $M^1$ from the tide's numerator, $M^{-3}$ from the horizon's cube — net $M^{-2}$ . And note the $L$ : it must be there, because the left side is an acceleration. This article's base drops it mid-derivation, writing $(2GM/L)(c^2/2GM)^3 L$ , whose left-hand side evaluates to $c^6/4G^2M^2$ — units of $\mathrm{s^{-2}}$ — while the right-hand side, correctly, keeps the $L$ . The two sides of the base's own equality are not equal. The final formula is right; the step that produces it is broken.

Step 4. Put a person ( $L = 1.8$ m) at three horizons and read the numbers.

black hole	mass	horizon tide over 1.8 m	in $g$
stellar	$10\,M_\odot$	$1.85\times10^{8}$ m/s²	$\mathbf{1.9\times10^{7}\,g}$
crossover	$\mathbf{4.3\times10^{4}\,M_\odot}$	$9.8$ m/s²	$\mathbf{1\,g}$
Sgr A*	$4.3\times10^{6}\,M_\odot$	$1.0\times10^{-3}$ m/s²	$\mathbf{1.0\times10^{-4}\,g}$
M87*	$6.5\times10^{9}\,M_\odot$	$4.4\times10^{-10}$ m/s²	$4.5\times10^{-11}\,g$

Why: you would cross Sgr A's horizon and feel a ten-thousandth of a $g$ — less than standing up quickly. Nothing local would tell you it had happened. The horizon is not a wall; it is a causal boundary, and there is nothing there to feel.*

Step 5. The general lesson. At a supermassive horizon the tide — the only coordinate-invariant local gravitational effect — is negligible, so spacetime there is locally indistinguishable from flat, and the equivalence principle holds beautifully at the most dramatic surface in physics. Why: curvature scale and "strength of gravity" are different quantities, and the horizon proves it. What is extreme about a black hole is its causal structure, not its curvature. `aq-sr-rindler-horizon` owns the causal side; the interior belongs to the general-relativity track. *Verified in verify_aq-sr-tidal-forces-equivalence-limits.py: $2GM L/r_s^3$ is shown symbolically to equal $Lc^6/4G^2M^2$ ; the base's step, $(2GM/L)(c^2/2GM)^3 L$ , is shown NOT to equal it and to carry units of $\mathrm{s^{-2}}$ ; all four table rows and the $4.3\times10^{4}\,M_\odot$ crossover are computed at fifty digits; and the $1/M^2$ scaling is asserted by perturbing $M$ .*

Figure

SPAGHETTIFICATION RUNS BACKWARDS — the tide at a black hole's horizon FALLS as 1/M², so the bigger the monster the gentler its edge. The curve is the tidal acceleration across a 1.8 m human body at the event horizon, plotted against the hole's mass. At a 10-solar-mass stellar hole it is 1.9 × 10⁷ g and you are pulled into a thread long before you arrive. Follow the curve right and it collapses: it crosses ONE g at about 43,000 solar masses, and by the time you reach Sagittarius A — the four-million-solar-mass hole at the centre of our own galaxy — the tide across your body at the horizon is 1.0 × 10⁻⁴ g. You would drift across the event horizon of Sgr A and feel a ten-thousandth of a gravity. Nothing would happen. Nothing local COULD happen, because the tide is the only coordinate-invariant gravitational effect there is, and at that horizon it is negligible — so spacetime there is locally indistinguishable from empty flat space, and the equivalence principle holds beautifully at the most dramatic surface in physics. The horizon is not a wall. It is a one-way causal boundary, and there is nothing there to feel. Two powers of M fight: the horizon radius grows as M, the tide falls as 1/r³, and M⁻³ × M wins

Limiting cases

Case	Result	Meaning
uniform field $\Phi = gz$	$E_{ij} = 0$ identically	why acceleration is not the boundary. A rocket has no tide, so it cannot fake gravity
$L \to 0$ , $T \to 0$	tide $\to 0$	equivalence becomes exact. `aq-sr-equivalence-principle` prices it: $L T^2 < \delta r^3/GM$
$d \to \infty$ at fixed $M/d^2$	tide $\to 0$	a distant, strong source is a uniform field. This is why the Sun loses
$M \to \infty$ at the horizon	$a_{\text{tidal}} \to 0$ as $1/M^2$	a supermassive horizon is tidally gentle. Cross it feeling nothing
flat spacetime	$R^{\mu}{}_{\alpha\nu\beta} = 0 \Rightarrow \ddot\xi = 0$	a theorem, not a limitation. SR cannot produce a tide in any chart

Common trapThe sign trap — the tidal tensor and the motion it causes are negatives of each other

$\ddot{\xi}^{\,i} = -E_{ij}\xi^{j}$ . That minus sign is not decoration. The eigenvalues of $E_{ij}$ are

\left(-\frac{2GM}{r^{3}},\; +\frac{GM}{r^{3}},\; +\frac{GM}{r^{3}}\right) \qquad\text{(the TENSOR — `aq-sr-equivalence-principle` owns this)}

and the relative accelerations they produce are the negatives:

\left(+\frac{2GM}{r^{3}}\xi_r,\; -\frac{GM}{r^{3}}\xi_\perp,\; -\frac{GM}{r^{3}}\xi_\perp\right) \qquad\text{— STRETCH radially, SQUEEZE transversely.}

Print the first and call it the second and you have swapped a stretch for a squeeze — you have predicted that the oceans bulge at the poles. The base of `aq-sr-failure-global-inertial-frames` does exactly this: its eq[2] gives " $\Delta a_r \approx -2GM/r^3\,\Delta r$ , $\Delta a_\perp \approx +GM/r^3\,\Delta x_\perp$ " and labels them relative accelerations of two nearby free particles — both signs backwards, in flat contradiction with its own derivation card two panels later, which correctly says " $\ddot\xi_r > 0$ (stretch), $\ddot\xi_\perp < 0$ (squeeze)". This article's own base has them right. The two bases disagree, and only one can be correct. The physical check takes two seconds: the tide pulls you apart along the field line.

Worked examples

1 · How high is the tide, actually?

Step 1 — identify. The lunar tide-raising acceleration at Earth's surface: $a = 2GM_{\text{M}}R_\oplus/d^3$ .

Step 2 — apply. With $GM_{\text{M}} = 4.90\times10^{12}\ \mathrm{m^3/s^2}$ , $R_\oplus = 6.371\times10^{6}$ m, $d = 3.844\times10^{8}$ m,

a_{\text{tidal}}^{\text{Moon}} = \frac{2(4.90\times10^{12})(6.371\times10^{6})}{(3.844\times10^{8})^{3}} = \mathbf{1.10\times10^{-6}\ m/s^{2}} .

Step 3 — the Sun, for comparison. $a_{\text{tidal}}^{\text{Sun}} = \mathbf{5.05\times10^{-7}}\ \mathrm{m/s^2}$ — less than half, despite pulling $178.8$ times harder.

Step 4 — the bulge. The tidal potential lifts water until its weight balances: $h \sim a_{\text{tidal}} R_\oplus / g$ , giving $\sim 0.7$ m for the Moon and $0.3$ m for the Sun. Spring tides (aligned) add, neap tides (perpendicular) subtract — and the spring-to-neap ratio, about $(2.18+1)/(2.18-1) = 2.7$ , directly measures the number 2.18. Coastal tides are far larger because basins resonate, but the forcing is these two numbers alone.

Step 5 — the boundary number. $1.10\times10^{-6}\ \mathrm{m/s^2}$ is one ten-millionth of $g$ , and it moves the Pacific. A tide is not strong; it is irremovable — a different property, and the only one that matters here.

2 · Would you survive crossing Sgr A*?

Step 1 — identify. Tidal stretch head-to-toe at the horizon: $a = L c^6 / 4G^2M^2$ with $L = 1.8$ m, $M = 4.3\times10^{6}\,M_\odot$ .

Step 2 — apply.

a_{\text{tidal}} = \frac{(1.8)(7.265\times10^{50})}{4(6.674\times10^{-11})^{2}(8.55\times10^{36})^{2}} = \mathbf{1.0\times10^{-3}\ m/s^{2}} = \mathbf{1.0\times10^{-4}\,g} .

Step 3 — compare. A tenth of a milli- $g$ across your body; standing up from a chair subjects you to more. You would cross the event horizon of the Milky Way's central black hole and notice nothing whatever.

Step 4 — contrast. At a $10\,M_\odot$ hole the same calculation gives $\mathbf{1.9\times10^{7}\,g}$ — you are drawn into a thread far outside the horizon and never reach it. The difference is the mass, squared, in the denominator.

Step 5 — the boundary number. Set $a_{\text{tidal}} = g$ and solve: $M = \mathbf{4.3\times10^{4}\,M_\odot}$ . Above forty thousand solar masses a horizon is a place you can walk through — the sharpest statement of the chapter's thesis: what is extreme about a black hole is its causal structure, not its curvature, and only curvature is what SR cannot handle.

Origins · 2 minTidal forces and how this was discovered

Tidal forces

1687–1919 · Newton explains the sea · Laplace writes the equations · Einstein realises the tide is the only real thing

The explanation. Tides were the first thing gravity ever explained that nobody had asked it to. Newton, in Book III of the Principia, showed that the Moon's differential attraction across the Earth produces two bulges — near side pulled toward, far side left behind — with a second, weaker pair following the Sun. The sea, twice a day, obeying an inverse-square law invented for planets. Laplace, a century later, made it a real theory of a real ocean on a rotating globe. The reinterpretation. For two hundred years the tide was an application. Then Einstein noticed it is the only part of gravity that is real. The field's value can be transformed away — stand in a falling lift and gravity is gone, exactly. The difference in the field across the lift cannot be transformed away by anything, and that difference is the tide. Gravity's observable content is not $\mathbf{g}$ ; it is $\partial_i\partial_j\Phi$ , whose relativistic name is the Riemann tensor. The quantity that shows up in the sea is the quantity that curves spacetime: general relativity's field equations are, at bottom, a statement about tides. What survives. Every gravitational-wave detector is a tide meter. A passing wave is a travelling ripple of $R^{i}{}_{0j0}$ , and LIGO measures it exactly as Newton would have understood — two free masses drifting apart and back. The instrument that confirmed the last prediction of general relativity runs on the physics of the first thing gravity ever explained.

History

1687

— Newton, in the Principia, explains the twice-daily tide as the Moon's differential attraction across the Earth, and identifies the second, weaker solar tide. It is gravity's first unrequested success.

1776

— Laplace publishes the dynamical tidal equations, replacing the static bulge with a forced wave on a rotating globe; the equations still carry his name.

1907

— Einstein's equivalence principle: a freely falling frame is locally indistinguishable from an inertial one — which quietly makes the tide the only irremovable part of gravity.

1915

— General relativity identifies the tidal tensor with the Riemann curvature, and geodesic deviation makes curvature an observable rather than a definition.

2015

— LIGO detects a gravitational wave — a travelling tide — by watching two free masses drift apart and back. The detector is, in the end, Newton's tide meter.

Going further

Texts. Misner, Thorne & Wheeler, Gravitation, ch. 1 and §11.2 · Wald, General Relativity, §3.3 · Hartle, Gravity, ch. 21 · Rindler, Relativity, 2nd ed., §§1.16, 9.3 · Newton, Principia, Book III, Props. 24–37 · Cartwright, Tides: A Scientific History (1999).

Provenance. Signature $(+,-,-,-)$ . Every number is recomputed in verify_aq-sr-tidal-forces-equivalence-limits.py, which reads this file, regexes the printed values out, asserts them at fifty digits, and carries artefact guards on $G$ , $M$ , $d$ and $c$ . Base defect repaired: the spaghettification card's middle step, $(2GM/L)(c^2/2GM)^3L$ , carries a stray $1/L$ and does not equal its own right-hand side; the correct chain is $2GM\cdot L/r_s^3 = L c^6/4G^2M^2$ . Cross-base contradiction logged: `aq-sr-failure-global-inertial-frames`'s base prints the tidal-tensor eigenvalues while calling them relative accelerations — both signs backwards, contradicting this article's base, which has them right.

Scope. The tidal tensor, its eigenvalues, its tracelessness and the expiry inequality are `aq-sr-equivalence-principle`'s; the flatness of the Rindler chart is `aq-sr-rindler-metric`'s; the horizon's causal structure is `aq-sr-rindler-horizon`'s; the Schwarzschild metric and the black-hole interior belong to the general-relativity track. What is used here is the Newtonian $2GM/r^3$ at $r_s$ , exact in scaling and $O(1)$ in coefficient.

Builds on. The Equivalence Principle · Proper Acceleration and the Comoving Frame · Rindler Coordinates · Gravitational Time Dilation. Leads to. The Failure of Global Inertial Frames · The Unruh Effect · From Special to General Relativity.

Why are there two tidal bulges rather than one?

Because the tidal tensor's radial eigenvalue is a stretch, and a stretch is symmetric: it pulls outward on both ends of a diameter. Work in the frame falling with the Earth's centre. The near ocean is closer to the Moon and pulled harder than the centre, so relative to the centre it moves toward the Moon. The far ocean is further and pulled less hard, so relative to the centre it lags away from the Moon. Two outward displacements, on opposite sides — and the Earth rotates under both, giving two high tides a day. The far bulge is often called "centrifugal", which is a needless and frame-dependent complication: in the freely-falling frame there is no centrifugal force, and the far bulge is the Moon failing to pull hard enough on water further away than the centre is.

If free fall removes gravity, in what sense do tides remain?

Free fall removes the field's value at a point — exactly, with no residue. What it cannot remove is the field's gradient, and the reason is algebra rather than physics: a coordinate transformation adds a constant to the acceleration field, and adding a constant to a function does not change its derivative. So $\partial_i\Phi$ can be set to zero at a point by choosing to fall, while $\partial_i\partial_j\Phi$ sits there unaffected. The relativistic version is sharper: $R^{\mu}{}_{\alpha\nu\beta}$ is a tensor, and a tensor non-zero in one frame is non-zero in every frame, because tensors transform linearly and a linear map cannot send a non-zero object to zero. Your only knob is the size of the region, and `aq-sr-equivalence-principle` prices that: $LT^2 < \delta r^3/GM$ .

Is spaghettification really gentler at a bigger black hole? That seems backwards.

It is, and it stops being backwards once you see where the two powers of $M$ come from. The tidal acceleration at radius $r$ goes as $M/r^3$ — that part does grow with mass. But the horizon moves outward as $r_s = 2GM/c^2 \propto M$ , so you are evaluating the tide at a radius that grows linearly. Together: $M/r_s^3 \propto M/M^3 = 1/M^2$ . Three powers of mass in the denominator beat one in the numerator, and the horizon tide falls as the square of the mass. At $10\,M_\odot$ it is $1.9\times10^{7}\,g$ and you are torn apart far outside the horizon; at Sgr A* it is $1.0\times10^{-4}\,g$ and you drift across without noticing. The crossover — horizon tide across a human body exactly $1\,g$ — is about $4.3\times10^{4}\,M_\odot$ .

Does the far-side tidal bulge require centrifugal force?

No, and invoking it obscures the physics. The clean frame is the one freely falling with the Earth's centre: no centrifugal force, no rotation of the Earth–Moon system to worry about, and the tide is exactly the residue of the lunar field after its value at the centre is subtracted. That residue points outward at both ends of the Earth–Moon line, because the tidal tensor's radial eigenvalue is a stretch. The centrifugal story arises only if you insist on the rotating Earth–Moon frame, where you must then add a centrifugal term — same answer, more machinery, and an inertial force that a change of frame could remove. The tide cannot be removed by a change of frame; the centrifugal force can. That alone tells you which is the real one.

Why does the geodesic-deviation derivation never mention a force?

Because there is no force, and that is the point. Both particles are in free fall: each at rest in its own local inertial frame, each with a zero accelerometer reading, neither with the slightest right to accelerate. The derivation uses two ingredients only — both worldlines are geodesics ( $\nabla_u u = 0$ ), and the separation vector commutes with the tangent ( $\nabla_u\xi = \nabla_\xi u$ , the equality of mixed partials). Everything else is the commutator of covariant derivatives, which is the Riemann tensor by definition. The relative acceleration emerges from geometry alone. That is the precise sense in which general relativity replaces the gravitational force: not by explaining it, but by showing that the only observable it ever had — the tide — is a property of the manifold.

Could a very cleverly designed accelerating frame mimic a tidal field?

No, and it is a theorem rather than a challenge. Any accelerated frame in flat spacetime is a coordinate system on Minkowski space, and Minkowski space has $R^{\mu}{}_{\alpha\nu\beta} = 0$ identically. Geodesic deviation then gives $D^2\xi^\mu/\mathrm{d}\tau^2 = 0$ for every pair of free-fallers, in every chart, however exotic. You may build a frame with position-dependent clock rates (Rindler), with a horizon (Rindler), with non-zero Christoffel symbols (any curvilinear chart) — but never with a non-zero tide, because the tide is a tensor and the tensor is zero. Conversely, measure a tide and you have measured curvature, and no chart will hide it. That is the whole boundary between the two theories, and it is why this chapter's thesis is that the border is curvature, not acceleration.