Reactionless Propulsion

Reactionless propulsion means thrust with no reaction on anything: momentum appears in the vehicle and nowhere does an equal and opposite momentum appear. The word does most of the damage on this page, because it is routinely used for a second and entirely different claim. A rocket throws mass backwards. A photon rocket throws light backwards. A solar sail takes momentum from photons that arrive. An electrodynamic tether takes momentum from a planet's magnetic field through the Lorentz force. All four conserve momentum, all four are real, and three of the four carry no propellant at all.

So the framing under test — a drive that needs no propellant is achievable — is true as written and false as usually meant, and separating those two readings is the whole brief. Electrodynamic Propulsion Concepts owns the real propellantless hardware and its numbers; this page owns the momentum-conservation frame and the claims that fail it.

The reason the page can be dense rather than dismissive is that the subject has a complete measured record. One specific claim with published figures, a NASA laboratory result with its own error budget, a peer-reviewed journal paper that went through five referees, and then a purpose-built instrument that resolved the question to below the photon limit. That full cycle — claim, funding, peer review, independent instrument, artefact identification — is the content here, and it is the best-documented example of a question being settled by measurement anywhere on this map.

Established The formal position is a theorem, not a preference. Momentum conservation is the conserved quantity Noether's theorem attaches to the translational symmetry of space. A reactionless drive is therefore not a hard engineering problem; it is a claim that space is not homogeneous. Frontier And almost every proposed reactionless drive knows this, which is why almost every one of them names a hidden reservoir instead: the quantum vacuum for the Eagleworks lineage, distant matter through Mach's principle for Woodward, Unruh-radiation horizons for McCulloch's quantised inertia, an unspecified electrostatic asymmetry for Buhler. Each reservoir is named precisely so the claim is not formally a violation. Both statements belong in the record: what the brief waits on is a violation of momentum conservation, and the proponents' own preferred framing is that they are not claiming one.

Handwave The EmDrive is the model case, and the claimant's own published numbers are the place to start. Roger Shawyer of SPR Ltd proposed that a tapered closed microwave cavity develops net axial thrust. In Acta Astronautica 116 (2015), under his own name and his own company, he specifies cavity Q values of 8 × 10⁷ to 2 × 10⁸, a lift-engine specific force of 667 N/kW at a vertical acceleration of 0.39 m/s², an orbital engine at 185 N/kW and an interstellar-probe engine at 304 N/kW, and projects a 10-tonne single-stage-to-orbit spaceplane putting 2 t into LEO in 27 minutes and a 9-tonne probe reaching 0.67c over a ten-year burn. His 2016 patent application reports 0.3 N at 1 kW at Q = 50,000, a mean specific thrust of 326 ± 67 mN/kW, and projects 30 N/kW at high Q with liquid-nitrogen cooling. Established The arithmetic that frames everything else: 667 N/kW is 2 × 10⁸ times the photon limit, and even the patent's room-temperature figure is about 10⁵ times it.

Frontier The NASA result was a real measurement, honestly reported, and it is worth being precise about what it said. Harold White, Paul March, James Lawrence, Jerry Vera, Andre Sylvester, David Brady and Paul Bailey at NASA Johnson's Eagleworks laboratory reported 1.2 ± 0.1 mN/kW across all configurations at 40 W, 60 W and 80 W, in vacuum below 8 × 10⁻⁶ Torr, with forward-thrust measurements of 30 to 128 micronewtons and reverse of 30 to 83. Established The authors named their own dominant error sources — thermal baseline drift at ±4.5 micronewtons and seismic noise at ±2.2 — stated that thermal effects were the main remaining uncertainty, and proposed a Cavendish balance to remove it. Established The paper went through five referees rather than the usual two at the AIAA journal. Peer review did not fail here. It worked, and passed a paper that reported a measurement and did not claim to have explained it.

Established White himself was cautious in print, and the caution is more informative than the headlines were. He said the team “may have generated an anomalous force”, that “there are no shortcuts in scientific investigation”, and that on some error sources they had “maybe put a little bit of a pencil mark through” them rather than having them “black-Sharpie-crossed-out”. Established And the number was already collapsing. 1.2 mN/kW is about 360 times the photon limit and about 500 times smaller than Shawyer's own patent figure. Handwave The same laboratory's 2011 briefing had reported historical Quantum Vacuum Plasma Thruster results of 1,000 to 4,000 micronewtons, a specific force of 0.1 N/kW, an equivalent specific impulse of about 10¹² seconds, and a projection of 50 t to Jovian orbit in 35 days on 2 MW. Established None of that survived: the lab's own figure fell by roughly two orders of magnitude in five years as its own measurement improved.

Established The resolution came from an instrument built to find the effect. M. Tajmar, O. Neunzig and M. Weikert at TU Dresden built an inverted counterbalanced double pendulum with sub-nano-newton resolution and thrust noise of 10 to 30 nano-newtons, ran the cavity across 1850 to 2000 MHz at 3 to 7 W and up to 11 W on resonance, and measured 0.003 ± 0.02, 0.007 ± 0.01 and −0.009 ± 0.01 micronewtons at 1857, 1914 and 1943 MHz. Their conclusion, verbatim: “Our measurements limit any anomalous thrust to below the photon thrust limit.” At 11 W that limit is about 36.7 nano-newtons. Published open access in the CEAS Space Journal 14:31–44 (2022), received 29 January 2021, accepted 15 July 2021.

Established State the exclusion carefully, because the looser version of this sentence is a citation defect and this page used to carry it. The peer-reviewed paper says that Dresden's “Q factor was lower by an order of magnitude compared to others, which still rules out anomalous thrusts by at least two orders of magnitude”. The companion conference abstract from Space Propulsion 2020+1 is the source of the widely repeated more than three orders of magnitude. Same team, different documents, different claims. Established The defensible statement is the photon-limit one: Dresden bounded the effect below 3.34 micronewtons per kilowatt, which excludes the Eagleworks figure of 1.2 mN/kW by more than two orders of magnitude and Shawyer's published 667 N/kW by about eight. Every figure in that sentence is cited below to the document that actually carries it.

Established The artefacts Dresden named are the most instructive part of the whole cluster. Three of them. Thermal: heating of the cavity or the electronics causes thermal expansion which shifts the centre of gravity. Magnetic: cables can be twisted to reduce stray field but “no shielding is perfect”, and the residual interacts with Earth's field. And the liquid-metal feedthrough: “we found out that surface tension forces between the pin and the liquid exist that depend on the current passing through. This can lead to several micronewtons.” Established A several-micronewton systematic hiding in the power feedthrough is the same size as the entire claimed effect, and it was found only because somebody built an instrument specifically to look for it.

Established And the two controls that actually killed the signal need no theory at all. Marc Millis, reporting Dresden's 2018 interim results, records that when the EmDrive was pointed in a non-thrusting direction, thrust was still observed; and when the power was routed through an attenuator reducing it by a factor of 10,000, thrusting at the prior level was still observed. Established A signal independent of both orientation and input power is not a thruster. That pair of controls is the single most transferable methodological result in this brief, and it is what section 6 asks of every future claim.

Established Position one, the mainstream: momentum conservation is exact and every claimed reactionless drive is an artefact. Supported by every clean test in this brief and by the specific mechanism of each identified artefact — thermal expansion, magnetic coupling, feedthrough surface tension, vibration. Frontier What would overturn it is not an argument but a signal that survives orientation reversal and a power attenuator, and no such signal exists in any source this brief could obtain.

Handwave Position two: the EmDrive works and the nulls used the wrong Q, geometry or power. Held by Shawyer. It has a live foothold: Dresden used “geometry and operating conditions close to the model by White et al.” and reported a Q an order of magnitude lower than others'. Established But the objection survives only by shrinking the claim, and Dresden's bound is below the photon limit regardless of Q — a cavity that simply leaked its microwaves would beat the measured upper bound on the EmDrive.

Speculative Position three: quantised inertia. Mike McCulloch at Plymouth proposes that inertia arises from Unruh radiation damped by horizons, so an asymmetric cavity “falls down the Unruh gradient”. Established DARPA funded it — $1.3 million through the Defense Sciences Office's Nascent Light-Matter Interactions programme, reported September 2018, beginning with 18 months of theory before funding experimental teams in Germany and Spain. Handwave McCulloch's response to the Dresden nulls is the cluster's signature move: “the problem is that the addition of a symmetric metal box will cancel the thrust from quantised inertia.” The shielding that removes the artefact is said to remove the effect. That auxiliary hypothesis was introduced after the null and cannot be checked, which is exactly the property that separates it from the Mach-effect objections in Mach Effect Thrusters.

Speculative Position four: Mach-effect coupling to distant matter conserves momentum non-locally. Woodward, Fearn and the Space Studies Institute. It is the only candidate in the set that does not require a conservation violation, which is why it is tested rather than dismissed, and it is owned in full by Mach Effect Thrusters. Handwave Position five: quantum-vacuum momentum exchange. White and the Eagleworks lineage. Eric Davis's assessment, quoted in Aerospace America, is that “there's no ‘there’ there”; there is no accepted theoretical basis, and the lab's own numbers fell two orders of magnitude in five years.

Handwave Position six: an unidentified electrostatic asymmetry force. Charles Buhler — a real NASA engineer who founded the Electrostatics and Surface Physics Laboratory at Kennedy — and Exodus Propulsion Technologies claim about 10 millinewtons from a 30 to 40 gram device of Teflon, copper tape and foam in a custom vacuum chamber, with a claimed progression from 10⁻⁴ g in 2016–20 to 1.0 g of thrust by November 2023, roughly 2,000 test articles, presentation at the Alternative Propulsion Energy Conference rather than in journals, and a patent route rather than a publication route. Handwave By 2026 the claim includes thrust persisting after the power is switched off, which is a free-energy claim rather than a propulsion one. The electrostatic physics is Electrogravitics's, where the same device class is measured against a sub-nano-newton bound; this brief carries it only because it is presented as reactionless.

Handwave Position seven: inertial and oscillating-mass drives — the Dean drive and the long patent tail. This is the largest population and the smallest theoretical content. Christopher Provatidis's 2024 review counts about 278 Google Scholar hits for “inertial propulsion”, 536 including patents, and concludes that within Newtonian mechanics it is not possible to obtain continuous net thrust: over a full revolution at constant angular velocity “the net impulse is zero”, and such devices act “like a catapult”, delivering initial velocity to the centre of mass rather than sustained thrust. Established This class is closed by mechanics rather than by experiment, which makes it the only branch here that needs no instrument.

Speculative Position eight, and it is the reading this brief is willing to state: the field's real product is metrology and the drives are the funding pretext. Nobody holds it as a stated position. It is the residual once the Breakthrough Propulsion Physics hit rate and Dresden's instrument inventory are put side by side. Frontier Position nine: publication-and-attention asymmetry, not physics, explains the persistence. A claim costs a bench, a signal generator and a conference slide. A refutation cost TU Dresden a multi-year programme, several purpose-built balance architectures and a systematic-error hunt. The claim gets coverage on announcement; the null gets covered once and then the claim reappears with an auxiliary hypothesis attached.

Established There is no instrument bottleneck, and saying so is the sharpest available criticism of the field. Dresden's balances resolve below a nano-newton with thrust noise of 10 to 30 nano-newtons. The smallest claim in this brief is four orders of magnitude above that. Established What is missing is not a measurement but a result. The instrument exists, has been built more than once, and has run.

Established The real bottleneck for a claimant is the systematic-error budget, and it is harder than the thrust measurement. Dresden's own catalogue is the specification: thermal expansion shifting the centre of gravity, imperfect magnetic shielding interacting with Earth's field, and surface tension in a liquid-metal feedthrough varying with current at the several-micronewton level. Frontier Any device claiming a micronewton has to beat all three, and the third one was invisible to everybody until it was specifically hunted.

Frontier A third bottleneck is procedural rather than technical: nobody agrees in advance what is being tested. Configuration details get disputed after the null rather than fixed before the run. That is a protocol failure with two signatures on it, and it is the only bottleneck in this brief that could be removed at no cost.

Speculative And a fourth, which is about incentives. The output asymmetry — a bench versus a multi-year programme, one round of coverage for a null versus repeated coverage for a claim — means the population of claims grows faster than the population of measurements can shrink it. That is not a physics constraint and no experiment will fix it.

Established This brief does not wait on a facility, a material or a budget line. It waits on a scientific result nobody is producing, and the adjudication names it as such rather than routing a typed edge at another brief. Every capability that a working reactionless drive would need already exists in better form than the claims require.

Frontier Two neighbouring briefs supply the physics this one borrows. Inertial Manipulation owns the origin-of-inertia question as physics; this brief takes only the Noether framing from it. Mach Effect Thrusters owns the one candidate that claims momentum conservation rather than denying it, and that brief's question is genuinely open where this one's is not.

Established What this brief depends on institutionally is the continued existence of the null-testing laboratory, and that is a real dependency with a real failure mode: the groups capable of this work are few, their funding is justified by a hope they keep disappointing, and the honest justification is the metrology rather than the hope. Section 9 makes that argument rather than assuming it.

Established The experiment that matters is already specified, because it is the one that has already killed every signal. Five controls, in order of how much they cost: orientation reversal, a power attenuator of 10⁴, magnetic shielding of the feed lines with a documented residual, thermal isolation of the feedthrough with the surface-tension term measured rather than assumed, and publication of the full systematic budget by a group with no stake in the outcome. Established Every one of those five is a control that has already been applied and has already killed a signal.

Established Two of them did the actual work and are trivially cheap. Point the device the other way; attenuate the power by four orders of magnitude. In the 2018 Dresden interim run the apparent thrust survived both, which is a complete refutation requiring no model of what the device was supposed to do.

Frontier What would be genuinely new is a pre-registered protocol. Agree the frequency, drive amplitude, geometry, settling time and reaction-mass configuration before the run, with the claimant supplying tuned hardware and the independent group supplying the balance. Established That has never been done in this subject, and it is the only experimental change on offer that would convert a disputed null into an undisputed one.

Established One more experiment is worth naming because it was attempted and did not happen. IVO Ltd's “Quantum Drive”, built on McCulloch's theory and claimed to have a power-to-thrust ratio twelve times better than existing satellite drives, flew aboard Rogue Space Systems' BARRY-1 cubesat on a Falcon 9 on 11 November 2023, intending to raise the orbit by about 60 miles. Rogue lost the satellite to a power-system failure and lost communication on 9 February 2024. IVO's chief executive: “Sadly, we never even got to turn on the Drives!” Established That is a null of opportunity, not a null result, and reporting it as a failed test would be wrong.

Handwave No engineering requirements can be stated for the drive, because there is no effect to engineer, and a requirements table here would be fiction dressed as a specification. What the subject has actually produced is an instrument specification, and that one is real and demanding.

Established The instrument class is the valuable artefact of thirty years of this work. Dresden's SpaceDrive programme lists an inverted counterbalanced double pendulum, torsion balances, a superconducting levitation thrust balance, magnetic suspension balances, a centrifugal balance for mass-fluctuation detection, fibre-optic gyroscopes and attocube laser interferometry — tested against the EmDrive, Mach-effect thrusters, asymmetric infrared laser resonators, “VEM drives”, Wallace field generators and superconducting gravity impulse generators. Established Sub-nano-newton thrust metrology in vacuum, with magnetic shielding, thermally isolated feedthroughs and a documented systematic budget, is a general-purpose capability.

Frontier And it has customers that have nothing to do with the fringe. Micronewton force metrology is exactly what qualifying real electric thrusters, cold-gas systems and drag-free spacecraft requires. Speculative The honest institutional justification for funding this line is therefore the metrology and not the hope — an argument this brief makes rather than one the funders have made.

Established The genuinely adjacent work is the real propellantless technology, and it should be listed with numbers in the same brief that dismisses reactionless drives — because the reader's actual question, can a spacecraft accelerate without carrying fuel, has the answer yes. NASA's ACS3 deployed an 80 square metre sail from a microwave-sized cubesat on 7-metre composite booms at a total vehicle mass of 16 kg, launched 23 April 2024 and deployed that August. Established Electrodynamic tethers have flown 18 times since 1967, 8 of them electrodynamic. Ion wind flew a five-metre aeroplane ten times in 2018. Beamed propulsion and photon drives are real, momentum-conserving, and bounded by 3.34 micronewtons per kilowatt.

Established The boundaries within this map are drawn deliberately and each neighbour owns something this page deliberately does not. Electrodynamic Propulsion Concepts owns every real propellantless technology and its measured performance — the list above is a signpost, not a treatment. Mach Effect Thrusters owns Woodward's mechanism, the NIAC funding and the live methodological dispute; this brief mentions the Mach effect in one paragraph and links. Electrogravitics owns ion wind and the Biefeld–Brown physics, including the electrostatic-device class that Buhler's claim belongs to.

Established Two further seams are worth stating explicitly. History of Electrogravitics owns the 1950s funding history in full, and nothing on this page reaches back before 1990 except the inertial-propulsion class, which is mechanics rather than history. Frontier Gravity Modification owns the superconductor and frame-dragging claims and shares this brief's instrument literature without sharing its argument; the Tajmar group appears in both because it tested both.

Established The most useful institutional record here is NASA's Breakthrough Propulsion Physics project, and its most useful output is its selection rule rather than its results. NASA/TM-2004-213406 records it: “reliability of assertions is more important than the implications”, with criteria scores multiplied rather than added, so a single zero on credibility kills a proposal regardless of how large its payoff would be. Established That is a discriminator, stated by a sympathiser, that any funder of this class of claim could apply tomorrow.

Established The outturn is equally public. NASA/TM-2004-213082 records that of 16 incremental research tasks, about a third were found not viable, a quarter had clear opportunities for sequels, and the rest remain unresolved. Frontier A hit rate that is honestly published is worth more to a reader than the hit rate itself, and it is the number against which every subsequent programme in this cluster should be read.

Established Marc Millis is the most useful voice in the subject precisely because he is a sympathiser who insists on method. His 2018 Estes Park workshop report, with about 30 participants, gives the field's own criteria for whom to believe: whether experimenters systematically check for false positives, whether theoreticians engage conflicting viewpoints, and whether they focus on make-or-break issues — while he “will ignore ‘believers’ or pedantic pundits”. Established The workshop's most consequential contribution that year was software engineer Jamie Ciomperlik's numerical simulations showing that with the right parameters a false positive thrust could result from vibrational effects — a mechanism for the artefact rather than an assertion of one.

Frontier The funding pattern in this subject is that money buys claims reliably and measurements rarely. DARPA's $1.3 million bought a theory programme. Private and enthusiast funding buys conference presentations. The measurement that settled the EmDrive came from a university chair with a multi-year instrument programme, and it is the only kind of spending in the record that produced a number anyone can check. Speculative The correct institutional posture is therefore not refusal but instrumentation: when a claim is tested at metrology standard it resolves quickly and publicly.

Established And one institutional exhibit runs the other way, in the claimants' favour. Fearn and Woodward built their own symmetry null in 2012 and published it. Tajmar told the proponents' own workshop in 2016, about his own data, “I cannot then say for sure that what I have seen is real or some noise”. Frontier Those are both the behaviour the field says it wants, performed by people on opposite sides of the question, and a brief that only counted failures would miss them.

Established The principal hazard is financial and it is not hypothetical. Propellantless propulsion attracts investment solicitations, and the vocabulary — quantum vacuum, Unruh radiation, Mach's principle — is borrowed from real physics, which makes claims hard for a non-specialist to sort. A device with a patent, a NASA-veteran founder and a flight manifest looks exactly like a technology company until the measurement is asked for.

Frontier A second consideration cuts the other way and deserves saying plainly. The nulls in this brief exist because somebody was funded to look. Dresden's instruments were built to find these effects. A posture that treats the whole subject as disreputable produces no nulls, and then the claims persist unrefuted — which is a worse epistemic outcome than the current one.

Established Third, there is a specific and avoidable harm in mis-citation, and this page carried one. Quoting a conference abstract's number while citing the peer-reviewed paper is a small error that propagates: it overstates a bound, and it hands the claimant a legitimate objection about rigour. Established The correction — state the photon-limit bound, cite each figure to the document that carries it — costs nothing and is more damaging to the claim, not less.

Speculative Fourth, and only in the counterfactual: a genuine reactionless drive would be a strategic capability before it was a transport technology, and the governance conversation would begin from a standing capability rather than from a research programme. There is no basis for expecting that and it is not a reason to constrain anything now.

Handwave If a reactionless drive were real the consequences would be total — launch costs, transit times, station-keeping and the entire mass budget of spaceflight would be rewritten at once. That is precisely why the claim persists through three decades of nulls, and why stating the payoff clearly is part of explaining the persistence rather than a concession to it.

Established The civilizational payoff already banked is not the one anyone wanted, and it is real. Sub-nano-newton force metrology in vacuum, with a documented artefact catalogue, exists because people spent thirty years chasing something that was not there. That capability qualifies real thrusters and drag-free spacecraft. Frontier A field can fail at its stated goal and still leave a useful instrument behind, and this is the cleanest example of that on this map.

Established Meanwhile the actual answer to the reader's question is already flying. Propellantless propulsion works: sails, tethers, beamed momentum, ion wind. The transformation people expect from a reactionless drive is being delivered, more slowly and less dramatically, by hardware that conserves momentum. Frontier Improving that hardware is where the payoff is, and Electrodynamic Propulsion Concepts has the numbers.

Speculative And the subject has a durable function as an epistemic exhibit. A claim entered a peer-reviewed journal through five referees, attracted a national laboratory, a defence research agency and a satellite launch, and was closed by one university group with a better balance and two cheap controls. That is worth preserving as a case study whatever happens to the physics.

There is no horizon on which a reactionless drive is expected, so these horizons track measurement, method and the population of claims rather than engineering milestones:

• 10 yr: Established Expect no positive result surviving the five controls in section 6; the bound is already below the photon limit and nothing on the horizon improves the instrument by an order of magnitude, because the instrument is not the obstacle. Frontier Expect the claim population to keep growing, since a claim costs a bench and a refutation costs a programme. Speculative The one genuinely uncertain item is whether the Exodus / Buhler claim reaches a journal or an independent balance; ten years without either would itself be the answer.
• 25 yr: Speculative The interesting developments will be institutional rather than physical: whether any funder adopts the Breakthrough Propulsion Physics multiplied-criteria rule explicitly, and whether pre-registration reaches this subject as it has reached others. Frontier A plausible trajectory is that the drives stop being funded as propulsion and the instrument line survives on thruster qualification and drag-free metrology, which is roughly where it already is. Handwave Anyone forecasting a working device at this horizon is forecasting a violation of momentum conservation, and no candidate exists.
• 50 yr: Handwave Contingent entirely on a result nobody is producing. A reactionless drive requires that space not be homogeneous at some accessible scale, and no experiment in or out of this subject has ever hinted at that. Speculative The residual worth stating is narrower than the popular version: what is untested is not “whether momentum is conserved” but whether some reservoir exists that nobody has named. Every named reservoir so far has been tested.
• 100 / 250+ yr: Handwave Beyond useful forecasting, and the honest statement is about the shape of the record rather than a date. Every specific mechanism proposed since 1990 has been measured and excluded at a stated sensitivity, several of them by groups that wanted them to be true. Speculative A subject whose entire named hypothesis space has been tested within thirty years has no base rate at this horizon at all — but it also has no open door that an instrument could walk through.

• Depends on No brief on this map produces what this one waits on, so no depends_on edge is recorded and the honest answer is the typed requirement below. Within the map the closest relations are borrowings rather than dependencies: Inertial Manipulation supplies the origin-of-inertia framing, Mach Effect Thrusters owns the one candidate that claims momentum conservation rather than denying it, and Electrodynamic Propulsion Concepts owns the real propellantless hardware that answers the reader's actual question.
• Requires (not on this map) A reactionless drive requires that momentum not be conserved — equivalently, by Noether's theorem, that space not be homogeneous — or else that some momentum reservoir exist which nobody has yet named, since every reservoir that has been named has been tested. Nothing about this is an engineering gap. TU Dresden's inverted counterbalanced double pendulum resolves below a nano-newton with thrust noise of 10 to 30 nano-newtons, and it bounded the EmDrive below the photon limit of 3.34 micronewtons per kilowatt — below the thrust the same cavity would get for free by radiating its power away as light. That bound excludes the Eagleworks measurement of 1.2 millinewtons per kilowatt by more than two orders of magnitude, and Shawyer's published 667 newtons per kilowatt by about eight. The instrument is four orders of magnitude better than the smallest claim it is asked to test; what is missing is a result, and no facility, budget or supply chain would supply one.
• Enables Nothing measurable. If a reactionless effect existed the enabling reach would be everything in Category I at once, but no typed enabling edge is claimed, because an edge from a capability with no demonstrated effect to a capability that would depend on it records a wish rather than a dependency. What this subject has enabled is instrumentation: sub-nano-newton thrust metrology in vacuum, with a documented artefact catalogue, now available to qualify real thrusters.
• Adjacent Precision force metrology, vacuum engineering, magnetic shielding and vibration isolation supply every number in section 2. Within this map: Electrogravitics for the electrostatic-device class and the ion-wind identification, Gravity Modification for the superconductor and frame-dragging claims tested on the same instruments, History of Electrogravitics for the funding history this brief does not restate, and Solar Sail Systems and Beam-Powered Propulsion for the propellantless technologies that do conserve momentum.

Established “Reactionless and propellantless are the same thing.” They are not, and this is the misconception the page most wants to correct because every other error follows from it. Established Propellantless propulsion carries no fuel and still conserves momentum by pushing on something external — arriving photons, a planetary magnetic field, ambient air, a beam. It is achievable, it is achieved, and an 80 square metre sail deployed from a 16 kg spacecraft in 2024. Reactionless propulsion is the claim that no equal and opposite momentum appears anywhere, and it is the one that has been tested and bounded.

Established “NASA confirmed the EmDrive.” A NASA laboratory measured a signal of 1.2 ± 0.1 millinewtons per kilowatt and published it with a self-declared dominant thermal uncertainty of ±4.5 micronewtons and a stated need for a better balance. Established The signal then survived being pointed the wrong way and survived a 10,000-fold power attenuator, which is what an artefact does and what a thruster does not. The laboratory's own related figure from 2011 was eighty times larger and did not survive either.

Established “Peer review failed.” The AIAA journal used five referees rather than the usual two. The paper reported a measurement with an error budget and did not claim a mechanism, which is exactly what a journal should publish. Frontier What failed was the reporting layer, which converted a measurement with named systematics into a confirmation.

Established “Dresden ruled the EmDrive out by more than three orders of magnitude.” That figure circulates widely, including in an earlier version of this page, and it comes from the companion conference abstract rather than the peer-reviewed paper. Established The journal paper says “at least two orders of magnitude”, and adds explicitly that Dresden's Q factor was an order of magnitude lower than others'. The stronger and cleaner statement is that the effect was bounded below the photon limit of 3.34 micronewtons per kilowatt, which excludes Eagleworks by more than two orders and Shawyer by about eight. Attaching a conference number to a journal citation is a naming defect even when the physics conclusion is unchanged.

Established “The nulls are contested by the people who ran the experiments.” They are contested by the theories' originators, after the fact, on shielding and configuration grounds. Handwave McCulloch's version — a symmetric metal box cancels the thrust — explains the null in a way that cannot be checked. Frontier Woodward's version names a frequency, a transformer, a voltage multiplier and a settling time, and is checkable; that difference is the entire subject of Mach Effect Thrusters and it should not be flattened in either direction.

Established “The Quantum Drive was tested in orbit and failed.” It was never switched on. The cubesat carrying it lost its power system and then communications, and the operator's own statement is that the drives were never turned on. Established This is a null of opportunity, not a null result, and reporting it as evidence either way is wrong.

Handwave “Exodus has produced thrust that beats gravity.” The claim is about 10 millinewtons from a 30 to 40 gram device, presented at a conference and pursued through patents, with roughly 2,000 test articles claimed and no peer-reviewed publication in ten years. Handwave The 2026 version, in which thrust persists after the power is switched off, is a perpetual-motion claim rather than a propulsion claim, and it should be labelled as one. The comparable device class was measured to a nano-newton bound in 2024 by the group in Electrogravitics.

Handwave “Inertial drives just need better engineering.” Over a full revolution at constant angular velocity the net impulse of an oscillating-mass device is zero; it acts like a catapult, giving the centre of mass an initial velocity and no sustained thrust. Established This class is closed by Newtonian mechanics rather than by measurement, which is why the review that counts 536 patent-and-paper hits can dispose of all of them in a paragraph.

Frontier “There is no institutional value in any of this.” There is, and it is not the drives. Sub-nano-newton force metrology in vacuum, with magnetic shielding, thermally isolated feedthroughs, orientation reversal and attenuator controls, is a real capability with real customers in electric propulsion and drag-free spacecraft. Speculative The defensible sceptical reading is not that the work was worthless but that its product was the instrument rather than the drive.

Established And the framing itself. “A drive that needs no propellant is achievable” is true, and a brief that answered the sentence rather than the reading would get the wrong answer. Established What fails is the reactionless reading, and it fails at a stated number: nobody has demonstrated an anomalous force above 3.34 micronewtons per kilowatt under controls, in any source this brief could obtain.