The Vacuum Propeller: The Pendulum Test That Challenges Known Physics

April 25, 2026

The phrase sounds impossible before the experiment even begins: a propeller for the vacuum. A propeller is a bargain with matter; it pushes air, water, plasma, anything with enough substance to push back. The vacuum is supposed to offer no such handle. Yet in Mikolaj Baczynski’s presentation, the impossible phrase is not shouted as a triumph. It is held at arm’s length, surrounded by caveats, equations, failed runs, magnetic biases, pendulum noise, and a repeated plea for skepticism. Somewhere inside a room full of magnets and improvised precision instruments, a tiny force-like signal appears again and again, near 200 micro-newtons. It may be a clue. It may be an artifact. It may be the kind of experimental ghost that teaches more by disappearing than by surviving. But for one brief, careful story, it is worth following the swing.

The Machine Physics Forbids

A propeller belongs to the old world of visible reaction. It has blades. It has a medium. It throws something backward so that something else moves forward. The title slide of Mikolaj Baczynski’s presentation violates that instinct immediately: “Building a Vacuum Propeller.” Behind the words is an image that feels part spacecraft, part dream, part warning label. If the device were real in the strongest sense, it would not be merely a clever motor. It would be an argument with one of the deepest rules in physics: momentum has to go somewhere.

Baczynski does not begin like someone who believes that argument is already won. He begins by reaching backward fifteen months, to an earlier presentation on the possibility of building a propeller that could work in vacuum, and then steps into the new material: pendulum tests, prior experiments, and a summary of what he has found so far. Then he says the line that makes the presentation more interesting than its title: given the extraordinary claim, one must remain strongly skeptical. That sentence is not decoration. It is the hinge on which the whole story turns.

The reported signal is small by everyday standards and enormous by interpretive ones. Baczynski says that in several experimental arrangements involving static, non-uniform magnetic field geometries, he observes a reproducible force on the order of 200 micro-newtons. The number appears in beam balance tests, torsion balance tests, and pendulum timing tests. Single-magnet configurations are described as null or noisy, while certain multi-magnet geometries with strong gradients seem to produce more consistent behavior. The phrase “seem to” matters. It is the difference between a discovery and an anomaly report.

The question, then, is not whether someone has already built a reactionless engine. That framing is too easy, too theatrical, and too likely to flatten the science into belief or disbelief. The better question is sharper: what kind of evidence would be needed before physicists took a static-magnet, vacuum-momentum anomaly seriously? Baczynski’s work is best understood as a test case in that question. The presentation is not only about whether a tiny force exists. It is about how a tiny force might fool you.

What the Vacuum Gives—and Refuses

Modern physics has already ruined the simple idea of empty space. The vacuum is not nothing in the everyday sense. It is the lowest-energy state of quantum fields, and in certain conditions its structure can matter. The Casimir effect, in which two closely spaced conducting plates experience an attraction associated with altered vacuum modes, is the classic example. Vacuum polarization, dynamical Casimir radiation, the Unruh effect, and Hawking radiation all deepen the same lesson: emptiness is not as empty as common sense says.

But “not empty” does not mean “available as propellant.” This is the crucial distinction that popular discussions often blur. The vacuum can have measurable effects without being a reservoir that a machine can push against at will. Known effects usually require boundaries, acceleration, horizons, strong fields, or time-dependent conditions. In the dynamical Casimir effect, for example, the relevant system does not involve a static magnet assembly calmly extracting momentum from space. It involves rapidly changing electromagnetic boundary conditions, implemented in a superconducting circuit.

Baczynski’s proposal walks directly toward this boundary. His thought experiment imagines a quantum fluctuation as a pair of oppositely charged particles moving in opposite directions through a non-uniform magnetic field. Because the field is non-uniform, he argues, the two members of the pair could be deflected by different amounts. At the end of the fluctuation, in this speculative picture, momentum has been exchanged between the magnetic field and the vacuum. In his slides, this hypothetical thrust is labeled QF, for “Quantum Force.”

The difficulty is that quantum field theory does not treat virtual particles as tiny classical beads with clean trajectories through a field map. Nor does current physics predict that a static magnetic gradient should produce net thrust. Baczynski says this openly. His own slide titled “Current Physics Disagrees” states the problem in strong form: vacuum energy is not a local reservoir to tap, Lorentz invariance demands isotropy, and the vacuum stress-energy tensor in a static magnetic field is symmetric, leaving momentum density at zero. That is not a minor objection. It is the wall.

A Thought Experiment Under Suspicion

The speculative model has the texture of engineering rather than theoretical physics. It asks what happens if one treats the field gradient, particle velocity, uncertainty-limited travel distance, and vacuum energy density as inputs into an estimated force. Baczynski works through equations involving the uncertainty principle and de Broglie wavelength, then tries several methods for estimating the resulting thrust. The model is not presented as complete field theory. It is a scaffold built around a question: could the geometry of a magnetic field make the vacuum anisotropic in a usable way?

That scaffold immediately begins to strain. In one quick estimate, Baczynski defines an “effective volume,” a region treated as though it has the same magnetic moment at every point. He uses a volume of 212 cubic centimeters and a magnetic field scale of about one tesla. In field-modeling approaches, he compares a delta-y method, based on uncertainty-limited displacement, with a delta-By method, based on differences in magnetic field across adjacent points in the grid. Different methods produce different answers.

The disagreement is large enough that it becomes part of the story. One table shows a force around 134 micro-newtons for the effective-volume delta-y method at a vacuum energy density of about 4.43 × 10²¹ joules per cubic meter. Other field-modeling estimates require vacuum energy densities many orders of magnitude higher, reaching roughly 10³² to 10³⁴ joules per cubic meter depending on the method. The numbers do not converge in a way that would reassure a theorist. Baczynski asks whether the effective volume is overestimated, whether the delta-By method is underestimated, whether the grid is too loose, and then asks the most important question: is any of this valid?

That question saves the section from becoming pure speculation. In a weaker presentation, the equations would be used to manufacture authority. Here they produce uncertainty. The model is useful not because it establishes that vacuum thrust exists, but because it defines what the experimenter thinks he is looking for: a force whose sign and magnitude depend on magnetic field geometry. It gives the apparatus a target. It does not give the target a passport through known physics.

Three Instruments, One Whisper

Baczynski’s experimental case has three main witnesses: the beam balance, the torsion balance, and the pendulum. Each one speaks in a different dialect. Each one also has reasons to be doubted. That is the strange strength of the presentation. It does not rely on a single heroic instrument. It builds a pattern across several imperfect ones, then spends much of its time showing how those imperfections might have misled the experimenter.

The beam balance is the loud, noisy witness. In the slide deck, Baczynski describes the beam balance results as very noisy, yet still important because they point toward the same rough force scale as the later pendulum modeling. The beam balance tables translate millimeters of height difference into micro-newtons, with one millimeter corresponding to approximately 100 micro-newtons. Some magnet configurations show signs in the expected direction; others do not. It is suggestive, but not clean.

The torsion balance is the more persuasive witness, but it has its own vulnerability. Baczynski calls the torsion balance results convincing, yet also emphasizes that the setup is sensitive to Earth’s magnetic field. In the diagrams, magnet configurations are tested for whether false positives could arise from background magnetic interactions. One configuration is declared unreliable; another is allowed only under conditions meant to guard against bias. Here again, the pattern is not “I measured a force, therefore the vacuum moved.” It is “I measured a twist, and now I must ask what else could twist this system.”

The pendulum is the most narratively powerful witness because it turns the alleged force into time. It does not simply point or tilt. It swings, crossing sensors, losing energy, returning through zones named “in” and “out.” Its evidence is a rhythm: the ratio between in-time and out-time. Baczynski uses simulations to ask how a small force would alter that rhythm. The pendulum is the instrument that makes the anomaly feel closest to a heartbeat.

Timing the Invisible

The pendulum apparatus is simple enough to picture and difficult enough to trust. A bob of about 130 grams hangs from an arm about 1.5 meters long. The initial displacement is about seven millimeters, corresponding to a swing angle near 0.3 degrees. Sensors labeled S1 and S2 divide the motion into zones. As the pendulum moves from the out zone to the in zone, then back again, the system records crossing times and computes an inT/outT ratio.

The reason a pendulum can be useful is that a small force changes time before it becomes visually obvious. If the hypothesized force acts with the motion in one part of the swing or against it in another, the timing ratio can shift. Baczynski compares baseline magnet configurations with versions in which a single magnet is flipped, producing what he labels qfIn or qfOut configurations. If QF exists, he expects the N-S qfIn configuration to have a higher inT/outT ratio than the N-S baseline, while the S-N qfOut configuration should have a lower ratio than the S-N baseline.

The scale is delicate but not absurdly tiny relative to the pendulum’s restoring forces. Using the pendulum’s mass, arm length, and starting displacement, Baczynski estimates that the average horizontal component of gravity relevant to the motion is about 3,000 micro-newtons. Against that background, a 200-micro-newton force would be small but not invisible. It would be a perturbation of several percent of the relevant horizontal force scale, large enough to appear in timing if the apparatus were sufficiently repeatable.

The simulation uses an Euler-style, energy-based time-evolution method. It models a single swing, then uses the final potential energy of one swing as the input to the next, keeping the number of swings below twenty because beyond that the simulated period stops corresponding well to experiment. The sensor hardware and software are described as having two-microsecond minimum accuracy, while the simulation time resolution is set at ten microseconds based on the experimental spread. These are the numbers of a system trying to hear a whisper without mistaking the room for the voice.

The Planet in the Apparatus

No tabletop magnetic experiment escapes the planet. Earth’s magnetic field is not dramatic by the standards of permanent magnets, but it is everywhere, and it has direction. In Baczynski’s test location, the slide deck gives a field strength of about 53.6 microtesla and a dip angle of about 69 degrees. The pendulum’s maximum swing angle is only about 0.3 degrees. That means the apparatus is not merely swinging through empty geometry. It is swinging through a tilted planetary field.

Baczynski divides the Earth-field influence into three forces or torques. The first is a vertical torque associated with the dip angle, tending to align the magnet in the vertical direction. The second is a compass-like directional torque, tending to align magnets north-south; he argues this can be minimized by aligning the pendulum swing north-south and is partly absorbed by the structure. The third is a force pulling an aligned magnet toward Earth’s magnetic pole, which he treats as essentially zero for practical purposes because Earth’s field can be assumed uniform across the apparatus.

The vertical dip-angle effect is the one he treats as significant. It can bias the pendulum timing depending on the magnet arrangement. In the diagrams, the direction of the hypothetical quantum force is compared against the direction of the force arising from the single magnet’s interaction with Earth’s field. Some configurations create negative bias; the opposite configurations can create positive bias, where the Earth-field-induced force points in the same direction as the hypothesized QF. This is a dangerous overlap. A signal that has the same sign as a systematic is never innocent until proven otherwise.

This is where the story becomes a precision-measurement story rather than a vacuum-propulsion story. A 200-micro-newton anomaly sounds large enough to chase, but permanent magnets have moments, alignments, hysteresis, and residual asymmetries. A small uncanceled coupling to Earth’s field can imitate purpose. The experiment is not asking whether a magnet can feel Earth. Of course it can. It is asking whether, after every known magnetic influence is subtracted, randomized, reversed, and controlled, something else remains.

The Fifteen Percent Problem

Baczynski’s pendulum results are filtered through strict pass criteria. Both N-S and S-N base-magnet configurations must agree with corresponding simulations. The difference in the inT/outT ratio must exceed the standard deviation or interquartile range by more than five times. The test is not merely asking for a shift. It is asking for the shift to appear in the predicted direction, under paired configuration changes, with enough separation from the measured spread to look deliberate rather than accidental.

In the slide deck, the conclusion from passing tests is that the modeled QF is on the order of 200 micro-newtons. Baczynski notes that this value agrees with the rough 200-micro-newton scale indicated by beam balance tests. He also gives a probability estimate: the chance of such a correlation being random is roughly 0.1 to 0.8 percent. That number is provocative, but it also raises questions not fully answered in the presentation. A probability like that depends on the statistical model, the number of attempted trials, the selection rule, and whether the criteria were fixed before the data were examined.

The transcript adds the detail that changes the emotional temperature of the claim: roughly 15 percent of the pendulum tests passed the criteria. To get results that passed, Baczynski says he had to run the tests many, many times. He walks through an example that does not pass and calls parts of it bogus or untrustworthy. In a polished announcement, those failed runs might have vanished. Here they stay in the room. They make the reported signal less clean, but the presenter more credible.

The conclusion he reaches is not triumphant. He says that as much as he likes the pendulum experiments, and as valuable as he thinks they are, he does not have the resources required to build a pendulum that satisfies the test criteria. He plans to focus next on torsion balance tests and asks for collaboration at every level. This is not the ending of a discovery talk. It is the ending of an honest anomaly talk: here is the signal, here are the reasons to doubt it, here is why I cannot finish this alone.

No Free Lunch

The most important slides in the deck may be the ones titled “No Free Lunch.” They are where the apparatus answers back. The first problem is assembly placement. Baczynski models what happens if the magnet assembly is shifted three millimeters inward or outward. The apparent force changes are enormous compared with the claimed signal: roughly 1,350 to 2,500 micro-newtons depending on the assembly. Linearized against the 200-micro-newton target, this implies that placement repeatability between tests must be better than or equal to about 30 micrometers.

Thirty micrometers is not a casual tolerance. It is smaller than the width of many human hairs. It is not impossible in precision engineering, but it is unforgiving in a manually reconfigured pendulum experiment involving magnets, holders, gaps, blockers, and repeated flips. If a configuration change meant to reverse a magnetic pattern also changes the mass placement by an invisible amount, the pendulum will faithfully measure the wrong thing. It will not know whether the shift came from vacuum momentum or from the assembly sitting slightly differently in its V-shaped base.

The second problem is the initial swing. Baczynski compares starting displacements of seven millimeters against nine and five millimeters. The smaller five-millimeter displacement is especially problematic because mechanical artifacts and noise become too influential. The slide estimates that the initial swing must be controlled to about 250 micrometers. That is easier than 30 micrometers, but still a demanding requirement for an apparatus whose signal depends on comparing timing ratios across repeated configurations.

The third problem is subtler: oxygen paramagnetism. Oxygen molecules are weakly attracted by magnetic fields, and Baczynski models the resulting forces as opposite to the hypothesized quantum force. In the transcript, he estimates these forces at only about one to four micro-newtons, much smaller than the placement errors. Still, their presence is a reminder that the pendulum does not swing in a Platonic vacuum. It swings in air, among magnetic gradients, in a gravitational field, on Earth, with molecules that have their own faint preferences.

The Experiment That Could Kill the Claim

If the anomaly is to become more than a provocative pattern, the next experiment must be designed to destroy it. That is not cynicism. It is the scientific courtesy owed to any extraordinary claim. A better test would begin with calibration: known forces of 50, 100, and 200 micro-newtons applied to the pendulum or torsion balance to prove that the apparatus can recover forces of the relevant size without distortion. It would then report all runs, not just passing ones, with the randomization order, rejection criteria, environmental conditions, and confidence intervals visible.

The magnetic environment would need to be controlled as a primary variable rather than treated as background. A Helmholtz-coil system could cancel, reverse, or rotate the local magnetic field while leaving the internal magnet assembly unchanged. If the observed force tracks Earth-field orientation or field strength, that would point strongly toward magnetic torque artifacts. If it persists unchanged through carefully controlled field reversal, the anomaly becomes harder to dismiss.

The reconfiguration problem needs an engineering answer. A magnet assembly that must be touched, flipped, reseated, or nudged between trials invites micrometer-scale changes that are too large for the claim. A stronger apparatus would change magnetic configuration remotely or in situ, without altering center of mass, seating geometry, start position, or mechanical contact. It would use dummy masses with matching shape and inertia, mapped magnetic fields, blind labels, and automatic data logging. The apparatus should make it hard for the experimenter to know which configuration is supposed to succeed.

The ideal version might be a vacuum-compatible torsion balance or pendulum, tested in air, nitrogen, argon, and reduced pressure to separate air effects from magnetic and mechanical ones. It would include nonmagnetic controls, single-magnet controls, Earth-field reversal, full 90-degree and 180-degree apparatus rotations, and independent replication by a lab not invested in the outcome. The goal would not be to prove Baczynski right. The goal would be to give ordinary physics every possible chance to explain the signal. Only after that failure would the word “vacuum” deserve to grow louder.

Between Artifact and Discovery

There is a temptation, in stories like this, to make the vacuum the hero. It has the right mystery. It has a century of strange credentials. Casimir plates, black holes, accelerated observers, strong-field polarization, superconducting circuits producing photons from rapidly changing boundary conditions—these are not fantasies. They are reminders that common sense is a poor guide at the scale of quantum fields. But the same history also teaches restraint. The vacuum does surprising things under specific conditions, not whatever an analogy asks it to do.

The recent QCD spin-correlation work that Baczynski references belongs in that careful category. It explores correlations connected with quark confinement and the transition from partons to hadrons. It is relevant as an example of how modern experiments probe the structure of the vacuum and the reality of quantum correlations. It is not evidence that a tabletop static magnetic gradient can exchange net momentum with empty space. To treat it as such would be to turn a bridge into a shortcut.

The better speculative bridge is more modest. Suppose a geometry-dependent force-like anomaly really is present in these magnetic assemblies. Before naming it, one would have to subtract Earth-field torque, magnetic hysteresis, oxygen effects, sensor bias, pendulum nonlinearities, release variation, assembly placement, thermal drift, vibration, and human selection. If the anomaly survives all of that, then theory has a problem worth solving. If it does not, the experiment still has value. It will have mapped a region where intuition about magnets and micronewton mechanics goes wrong.

That is why Baczynski’s presentation is strongest when it is least triumphant. The title promises a vacuum propeller, but the body of the work reveals something more fragile and more scientifically interesting: a builder pressing an extraordinary idea against the abrasive surface of measurement. The pendulum swings. The magnets pull. Earth leans through the room. Oxygen gathers faintly in field gradients. The data sometimes align and often do not. Before anyone can build a propeller for the vacuum, the vacuum must first survive the error budget.