The Relativistic Submarine Paradox, Explained

Picture a submarine neutrally buoyant in the ocean — density perfectly matched to the surrounding water, floating at a fixed depth without effort. Now accelerate it to a significant fraction of the speed of light.

According to special relativity, the moving submarine undergoes length contraction: it shrinks along its direction of travel, its volume decreases, and its effective density rises. More dense than the water around it, it should sink. That's the view from the ocean floor.

Flip to the submarine's reference frame, and the logic runs in exactly the opposite direction. From where the crew sits, they're stationary — it's the ocean that's rushing past. The water contracts, its density increases, and the buoyancy force strengthens. The submarine should rise.

Both conclusions follow from the same physical principle applied in different frames. And since a submarine cannot simultaneously sink and rise, something in the reasoning has to be wrong. This is the relativistic submarine paradox, first formally stated by physicist James Supplee in a 1989 paper in the American Journal of Physics — a paper that included a proposed resolution, though one built on assumptions that were, as Supplee himself acknowledged, not obviously justified.

What "Solving" This Actually Means

Before reaching for a resolution, it's worth being precise about what the target is. The paradox isn't resolved simply by picking a winner. The goal is to show that whatever the submarine does in one frame, it does in every frame — not just qualitatively, but quantitatively.

The standard test: place a rock on the ocean floor at a known horizontal and vertical distance from the submarine. Set the initial conditions so the sinking submarine hits that rock in the rest frame of the ocean. The paradox is only resolved when the calculation performed in the submarine's own frame also predicts a collision with the same rock — and not just a collision, but one with the correct geometry. Because the horizontal distance to the rock is Lorentz-contracted in the submarine frame, the required downward acceleration turns out to be γ² times larger than in the ocean frame. That's the precise target any solution has to hit.

Getting that γ² factor to emerge from first principles is where the difficulty lives.

The Gravomagnetism Fix — and Where I Had to Slow Down

Working through the force accounting in each frame, the ocean frame is relatively clean. Gravity pulls the submarine down with force γmg (relativistic mass). Buoyancy pushes up, but the contracted submarine has less volume, so buoyancy loses. Net force is downward; the submarine sinks. Fine.

The submarine frame is where it gets uncomfortable. From the submarine's perspective, the ocean floor is a dense, moving sheet of mass rushing past. Its gravitational acceleration scales up — the contracted floor packs more mass per unit area, so g gets amplified by γ. That alone sends buoyancy scaling as γ². Which means the upward force massively outpaces gravity, and the submarine rises. That's the wrong sign, by a lot.

The fix involves gravomagnetism — and I'll be honest that this is where I had to stop and rebuild the argument from scratch before it clicked. The analogy used to introduce it is clean enough: magnetism, in a relativistic reframing, isn't a separate force so much as a geometrical consequence of charge density shifting between reference frames. An electron moving parallel to a current-carrying wire experiences attraction not because of some independent magnetic law, but because in its own frame the proton density in the wire is Lorentz-contracted while the electron density is spread out, creating a net charge imbalance. Magnetism is, in this telling, just electrostatics wearing a different frame.

Gravity has the same structure. A mass current — the ocean floor rushing past — creates a gravomagnetic field that modifies how moving water responds to gravity. Where I got stuck was seeing why the gravomagnetic correction to buoyancy cancels the γ² enhancement rather than compounding it. The answer is that the gravomagnetic force acts in opposition to the gravoelectric (standard gravitational) component, and when you work through the vector arithmetic, the γ² term in the buoyancy expression picks up a factor of 1/γ² from the gravomagnetic contribution. The gamma dependencies cancel. Buoyancy in the submarine frame ends up velocity-independent — which is exactly what you need to make progress.

At that point, the gravitational force on the submarine still carries a γ² enhancement from the rest frame of the submarine (the floor is more massive and moving, so both Lorentz contraction and kinetic energy contribute). The asymmetry is resolved. The math, when carried through carefully, produces the required factor.

The Assumption That Does a Lot of Heavy Lifting

Here's where it gets less tidy. The γ² gravitational enhancement only works if you're willing to treat relativistic mass as gravitational mass — i.e., if you accept that a body's kinetic energy contributes to the gravitational field it both produces and responds to. That's not a trivial assumption.

Physicist George T. Matsas, who revisited this problem after a student caught him unable to answer it on the spot — "It was embarrassing. I thought it was a scandal that there was no answer to this paradox. So, I decided to waste time on it" — provided a full general relativistic solution that agreed with Supplee's earlier result. Two different approaches, same answer. That convergence feels like confirmation.

But a subsequent analysis of the paradox raises a pointed objection. Supplee's argument, the critique goes, assumed Newton's gravitational law holds between moving bodies and combined it with Einstein's E=mc² to get speed-dependent mass — two frameworks that aren't obviously compatible. As quoted in the analysis: "the relation between mass and energy expressed by Einstein's formula above only holds when the momentum of the body is null. In fact, the correct expression is [the full relativistic energy-momentum relation]." The coincidence between Supplee's result and the full GR result may be attributable to the specific geometry chosen — an infinite flat plane — which suppresses the spatial curvature terms that full GR would otherwise contribute.

I find this critique genuinely unresolved, not obviously wrong. The counterargument — that the two expressions are equivalent when relativistic mass is handled consistently — has some force. But the light-deflection precedent is hard to dismiss. Einstein's 1911 calculation, treating a photon as a particle with relativistic mass in Newtonian gravity, gave a deflection value that was later shown to be exactly half the correct answer once full GR was applied. The agreement between frameworks in the submarine case may be telling us something real, or it may be an artifact of a flat-space approximation that happens to cancel the terms that would otherwise differ. I don't think the current literature settles this cleanly.

What Happens When You Take the Geometry Seriously

The flat-plane approximation is where the really interesting extensions begin. Researchers studying the paradox on a spherical ocean — where you have to contend with centrifugal forces and the full Schwarzschild geometry — found that the submarine's behavior depends on its position relative to the photon ring at 1.5 times the Schwarzschild radius. Below it: the submarine sinks. At it: neutral buoyancy. Above it: the submarine rises relative to the surface. That's a result that falls entirely outside the reach of the flat-space calculation, and it connects the submarine paradox to the behavior of matter near a black hole in ways that feel less like analogy and more like structural identity.

When a body approaches a black hole's event horizon, Hawking radiation creates an outward force while gravity pulls inward. A moving body contracts, reducing its cross-section to the radiation — the same mechanism by which the relativistic submarine loses buoyancy. The submarine paradox isn't just a thought experiment about underwater physics anymore; it's a toy model for one of the genuinely open problems in theoretical physics.

Whether the flat-space resolution is rigorous or merely coincidental, then, matters quite a bit — because what you're really asking is whether special relativity coupled to Newtonian-style gravity can approximate GR in this regime, or whether the agreement was always a fluke of chosen coordinates.

That question is still open. Matsas got the right number. The derivation path to that number may not be as solid as the number itself.

Amelia Nwofor is the Science Desk Editor at Buzzrag.