Bearden: Science Contrarian review

Reviewer role: Science Contrarian (evidence over consensus; same rigour applied to the researcher and to his critics). Sources read directly: the 1993 paper and appended comments in Bearden - Final secret of free energy.pdf (text layer, pdftotext), and the text layer of US6362718.pdf. The 1994 updates were read only in the passages quoted below. Tags: [read], [web] with URL, [memory] with confidence.

Source gap note (repeated from Step 1): the text layers are usable, but patent figures (Figs. 6-8) were not read as images, so the underlying measured data points are unseen. The patent's COP table is described only in prose.

Corrections to Step 1

1. Whittaker date. facts.json says "Whittaker 1904 gave a general solution of Laplace's equation as a superposition of plane waves". Bearden's own footnote 2 [read] cites two papers: the 1903 Math. Annalen paper (general solution of the Laplace/wave equations) and the 1904 Proc. London Math. Soc. paper (two scalar potential functions). The plane-wave superposition is the 1903 paper. Minor.
2. "Output power does not change with output turns." The facts entry says the patent states output voltage can be changed "without changing output power". The patent text [read] says the voltage is "readily reduced" by cutting turns from 450 to 12, giving 106.7 V and 0.5 A per coil. That is 53 W, not 48 W (4000 V x 12 mA = 48 W). The patent's own example therefore changes the power by about 11%, and the current rises by 41.7x while voltage falls by 37.5x (the turns ratio). The patent does not state that power is unchanged.
3. COP arithmetic. 48 W / 14 W = 3.43, not 3.44 as printed [read]. Trivial, but it shows the patent's number is not carefully computed. Also, "COP 3.44 per output coil" is per coil; the two coils together would imply 96 W / 14 W = 6.9 if both really delivered 48 W, which the patent does not claim. Step 1 should say the claim is ambiguous about this.
4. Scale of the main paper. facts.json says pp. 1-~25; the PDF is 49 pages and contains three papers (confirmed). The 1994 sections include a passage stating a capacitor "collector" must be "rigidized" because ordinary capacitors release as much energy as work as they stored [read, around PDF text line 2085]. This is an internal concession (see claim 3) that Step 1 did not capture.
5. Footnote 25. Verified [read]: it states the system "is not closed" and local conservation applies "for an open system with a hidden source", with the inequality that extracted work cannot exceed initial trapped energy plus energy flowing in. Step 1's paraphrase is fair.
6. Step 1 correctly noted that US 6,362,718 never mentions the vacuum or scalar potential. Confirmed by keyword search of the text layer for the physics terms; the patent describes ordinary flux switching. Worth stressing: the theory papers and the only measured data come from two different devices.

---

Claim 1: Energy = potential = ordering in the virtual photon flux; a scalar potential contains a harmonic series of wave pairs linking every charge to the universe

a. Mathematics consistent? Mixed. The Whittaker 1903/1904 decompositions are real mathematics: any solution of the wave or Laplace equation can be written as a superposition of plane (or bidirectional) waves [read: Bearden fn. 2-3, which cite the actual papers; the Ziolkowski references are real localized-wave papers in J. Math. Phys. and Phys. Rev. A]. A mathematical decomposition is not a physical statement about hidden energy flows; a static Coulomb potential written as a sum of waves is a Fourier-style rewriting, the same 1/r function, with no extra energy. Bearden's step from "can be decomposed" to "contains infinite energy exchanged with the universe" is not derived anywhere in the paper. He himself says energy and potential have "exactly the same definition" and that this is a reason that "energy and potential are identically the same" [read]. Units differ though: energy is joules, electric potential is joules per coulomb. He concedes this later by redefining "voltage" as energy per coulomb "the way it is corrupted in electrical circuit theory" [read], so the document uses the two words in both ways. This is a definitional inconsistency, not a subtle one.

b. Consistent with experiment? No experiment is cited that distinguishes his definition from the textbook one. The Aharonov-Bohm effect (1959; observed by Chambers 1960, Tonomura 1986 [memory: high]) shows potentials have physical effects in quantum mechanics; it does not show a vacuum-supplied energy reservoir. Bearden claims AB is "scalar interferometry" proven [read, fn. 2]; the AB effect concerns the vector potential A around a flux with no field, and he does not show the scalar-potential version he needs.

c. Tested? Not as a distinct claim. The definition has no measurable consequence beyond claims 2-4. Nobody tested "energy is a virtual photon ordering" because it predicts nothing different from standard electrostatics unless one adds claim 3.

d. What would change? The definition of energy (including the Noether link between time-translation symmetry and conservation), the interpretation of potential energy, and the treatment of the vacuum in QED. Standard QED has a vacuum with zero-point fluctuations but no extractable work at the ground state.

Verdict: Not testable as stated (the definitional core, since it makes no measurable prediction on its own), with the strong form (a vacuum energy flow through the potential that a device can tap) feeding into claim 3's verdict. Researcher-side rigour note: the unit confusion between energy and potential is a genuine error in the document, not a matter of taste.

---

Claim 2: No EM force fields in vacuum, only potentials; vacuum waves are longitudinal; electron charge varies with the ambient potential

a. Mathematics consistent? The paper offers no equation here, only the substitution "PE = -grad(phi)" in place of E = -grad(phi) [read]. In vacuum the free Maxwell equations give transverse waves from div E = 0; a longitudinal photon polarization is a gauge artifact. His "Newton's third law demands pairs of forces, so no vacuum force field" argument [read] confuses force on a test charge with momentum conservation; fields carry momentum (Poynting), which handles the third law. Footnote 5 says the EM wave is "a gravitational wave... oscillation of the rate of flow of time" [read], which has no equation behind it.

b. Consistent with experiment? The transverse nature of light is directly established: polarization experiments (Malus's law, crossed polarizers, Brewster angle) are routine, and radio antennas respond to the E and B vectors [memory: high, textbook]. A purely longitudinal vacuum wave cannot be polarized transversely. Charge: Millikan-type and modern tests show constant e; neutrality of matter has been tested to extremely small limits (Bressi et al., Phys. Rev. A 83, 052101, 2011, "Testing the neutrality of matter by acoustic means in a spherical resonator", https://arxiv.org/pdf/1102.2766 [web]; the check confirmed the paper and its aim, but I did not extract the numerical limit from the excerpt, so I do not quote one). Step 1's "1e-21 e" figure is [memory] and is not verified here; treat it as order-of-magnitude. A charge that shifted with ambient potential (e.g. a charged conductor at 2 kV) would break the neutrality of ordinary matter in ordinary experiments (e.g. every electron in a body at 2,000 V "scraping feet on carpet" per the paper [read] would differ in charge); that is not seen.

c. Tested? Longitudinal vs transverse: tested exhaustively, by everyone, daily, and a quick polarizer check suffices. Charge variation with potential: not tested by Bearden; the Millikan lineage gives indirect negative evidence. Mainstream neglected to address Bearden's version specifically, but it did not need to because the existing experiments already contradict it. This is "tested, not supported" by the existing body of work, though not in a Bearden-specific experiment.

d. What would change? Maxwell's equations as physical laws, special relativity's treatment of light, all of antenna engineering and optics, QED's photon polarization states, and charge quantization.

Verdict: Tested, not supported (transverse vs longitudinal waves; charge varying with potential). The "wave = oscillation of the rate of flow of time" and "photon-antiphoton coupling into gravitons" statements: Not testable as stated. A fair note for the contrarian ledger: one thing in this claim is partly right. Potentials are more fundamental than fields in quantum theory (the AB effect), and the field-in-vacuum picture is a model; Bearden has a legitimate point to be debated, but his conclusion does not follow.

---

Claim 3: A collector takes "current-free" potential energy from a battery during a relaxation time, then powers a load separately, at no cost to the source

This is the central practical, and checkable, claim of the 1993 paper.

a. Mathematics consistent? The pseudo-equations [1]-[8] [read] are accounting identities: K_T = V_T x Coul_T. They are correct as a statement of the energy stored in charge Q at voltage V, if one accepts it is QV. But QV is the energy the source delivered, not half of it; for a capacitor charged from a battery the stored energy is QV/2 (or 1/2 C V^2) and the source supplied QV. Worked numbers: a 1 microfarad collector charged to 12 V holds 72 microjoules; the battery must have delivered Q = 12 microcoulombs at 12 V = 144 microjoules (the other half is dissipated in the charging path). Bearden's scheme says Q moves to the collector without current through the source; but any charge that ends up on the collector must have flowed from somewhere. His own 1994 update admits the ideal "no current" case needs rigid components and instantaneous charging, and that ordinary capacitors "release as much energy as work... as it stored. Hence it will also produce dissipation of the source via the amount of energy stored" [read]. So his own text concedes that real parts do not deliver the free lunch. The relaxation-time number: the dielectric relaxation time is epsilon0 / sigma. For copper with sigma = 5.96e7 S/m this is 8.854e-12 / 5.96e7 = 1.5e-19 s (I computed this; it matches Bearden's 1.5e-19 s). The number is right, but its meaning is not: it is the time for free charge density inside a conductor to decay, and it has nothing to do with current "not flowing"; current flows from the first instant because the field establishes drift immediately. The paper's "electrons reach the skin" mechanism is not the textbook meaning cited (Paul and Nasar p. 100-113) [read: cited, not checked].

b. Consistent with experiment? Charging of a capacitor from a battery has been measured countless times; the source supplies QV. No accepted experiment shows energy appearing for free. The earth-ionosphere example [read]: the fair-weather field is about 100 V/m [memory: high] but the air-earth current density is about 2 picoamperes per square metre [memory: medium, textbook value]. Even a generous 100 m^2 collecting area at a 100 V potential difference gives 2e-12 x 100 x 100 = 2e-8 W (my arithmetic, using the memory figure): 20 nanowatts, nothing like a car.

c. Tested? Bearden's paper contains no experiment, and he says so by implication ("proof-of-principle ... for open release and certification" is still pending in 1994 [read]). He claims inventors who "succeed" lose the effect with better materials [read], which is an unfalsifiable framing: any failure with good wires is attributed to the wires. No neutral, controlled test of the collector scheme is on record that I could find. The 1994 update says a competent researcher can build the step-charged capacitor device "for himself" [read], so the claim is simple to test with a scope and a calorimeter; that nobody was commissioned to do so is notable, but I found no evidence that anyone was blocked. The relevant failure is on Bearden's side: he provided no data. The claim is therefore "never properly tested" in the sense of no test on record, but it is also refuted by calculation in its strong form (below).

d. What would change? Conservation of energy for the battery-load system. Standard theory: energy delivered to the load = energy removed from the battery (+ losses). One would need a replenishing reservoir.

Calculation: Energy balance. Let the collector have capacitance C and be charged to V from a battery of EMF V. Charge Q = CV leaves the battery. Energy from the battery: W_batt = QV = CV^2. Energy stored: CV^2/2. A load can dissipate at most CV^2/2 = 72 microjoules for C = 1 microfarad and V = 12 V per cycle. For the load to be "free", the battery must not lose the CV^2 = 144 microjoules, but Q has left it, so its charge state has changed by exactly Q; to restore it requires 144 microjoules of external work. A battery that had delivered N cycles has delivered N x Q coulombs, which is directly measurable as the ampere-hours consumed. The claim that "no coulombs leave" while the collector gains Q coulombs of "trapped" charge cannot hold: the collector's charge is the same charge.

Verdict (split): Strong form (a battery runs a car or a grid indefinitely with no depletion): Refuted by calculation (charge and energy accounting above), and by the paper's own concession about real capacitors. Weak form (a collector's switching schemes can exploit finite relaxation times to reduce dissipation inside the source): Never properly tested, but no mechanism for net gain is offered. Bearden's "relaxation time" mechanism is also Internally inconsistent because the 1.5e-19 s dielectric relaxation time is invoked both as a barrier and as irrelevant timing (he proposes a 1 ms design target and a mechanical switch [read]).

---

Claim 4: Closed classical EM cannot give overunity, so overunity devices are open to a hidden vacuum source; Maxwell's original quaternion theory contained this and gravity-EM unification

a. Mathematics consistent? The logic is valid as a conditional: a closed, energy-conserving system cannot be overunity. The step "therefore CEM must be flawed" is a non sequitur; mainstream concludes overunity is impossible. Maxwell's 1873 Treatise did use quaternions in parts (Vol. 2, ch. 9 and elsewhere) alongside Cartesian components [memory: medium]; Heaviside and Gibbs recast it into vector form in the 1880s [memory: high]. The vector and quaternion forms are mathematically equivalent for classical EM [memory: high; a standard result: the quaternion product splits into dot (scalar) and cross (vector) parts]. That the discarded scalar part "infolded the gravity-EM unification" is unsupported by any derivation in the paper [read] (conclusion section). It is also not what Maxwell wrote; Maxwell's gravity attempts (1864) concluded the energy required for a gravity field would be negative and he did not unify [memory: medium].

b. Consistent with experiment? Conservation of energy has been tested at high precision across circuits, motors, and particle collisions. No experiment on record shows a sustained over-unity source. The vacuum-triode (Sweet) claim is attributed here to phase conjugation by the atomic nucleus, with no calculation [read].

c. Tested? Sweet's vacuum triode amplifier: Sweet is reported to have claimed 330 microwatt input giving over 500 W output, and Bearden has said Sweet refused independent testing out of fear for his life (summary of secondary sources at http://www.rexresearch.com/sweet/1nothing.htm and http://amasci.com/freenrg/sweetvta.html [web, search result summary only; the specific pages were not opened]). So: no controlled independent test exists on record for the VTA; the claims are Sweet's own plus Bearden's testimony. "Dismissed without testing" does not fit, because the inventor withheld the device and the magnet-conditioning process, so this is a refusal on the proponent's side.

d. What would change? Energy conservation and the status of Maxwell/Heaviside vector calculus as complete; the vacuum would have to supply extractable work in contradiction with ground-state arguments.

Verdict (split): The conditional ("closed CEM cannot be overunity"): Established. The inference that CEM is flawed and open to a hidden source: Never properly tested for the VTA (no independent test; withheld by the inventor), and as a general principle Not testable as stated until a specific device is named. Quaternion-gravity unification: Not testable as stated (no equations given; I could find no source for Maxwell having claimed it). Historical point on quaternions: Historically superseded but instructive (a notation debate of the 1890s, the "quaternion war"; vector analysis won on convenience, not on physics).

---

Claim 5: The MEG (US 6,362,718) delivers more output than input power (COP > 1; extrapolated 3.44 per output coil at 100 V), and can self-power after start-up

This is the only claim with numbers, and the strongest candidate for a real, testable anomaly. I read the experiment passage in full [read: patent text layer, "A number of experiments have been conducted..." through the 100 V extrapolation].

What the patent reports [read]: - Input voltage 10-75 V, oscillator 87.5 kHz, input coils 40 turns, output coils 450 turns, Metglas 2605SA1 core, NdFeB magnet. - "Real power measurements were computed at each data point using measured voltage and current levels, with the results being averaged over the period of the signal. These measurements agree with RMS power measured using a Tektronix THS730 digital oscilloscope." - Fig. 8 gives COP at each point, "substantially higher" output than input. - Extrapolation: 100 V, 140 mA, 14 W in; 48 W out at each of two coils at 12 mA average and 4000 V. The text says "the data was extrapolated to describe experimentally-measured operation at an input voltage of 100 volts" [read, in context: the patent describes this as an extrapolation of measured operation].

a. Mathematics consistent? The arithmetic is simple: 100 V x 0.14 A = 14 W; 4000 V x 0.012 A = 48 W; 48/14 = 3.43 (the patent prints 3.44). Two things stand out. (i) 4000 V on a 450-turn coil driven from 100 V on a 40-turn coil is a turns-ratio gain of 11.25, i.e. at most about 1,125 V for an ideal transformer; the 4000 V is 3.6x that, and the patent offers no resonance or flux-switching argument for it. (ii) The "average output voltage" and "average output current" are multiplied directly. For pulsed, reactive, 87.5 kHz waveforms the product of averages is not average power unless the waveforms are in phase and the averaging is of the instantaneous product. The patent says the averaging was of the period but gives no waveform for the power calculation, and in a high-voltage, high-impedance output at 87.5 kHz a 10:1 probe with a limited-bandwidth, poorly-grounded measurement can produce large phase error. A phase error of even a few degrees on a mostly reactive load makes real power small compared to V x I, and the sign or size of the error can flip a COP from below 1 to above 1. I am not asserting this happened; the patent reports no phase angles.

b. Consistent with experiment? If real, time-averaged, with all input energy counted including any separate supply that powers the switching circuit, COP > 1 contradicts every measured energy budget. The patent says the switching drive is supplied by a "separate power source" in the test configuration [read: "Power for this drive circuit is supplied through an additional, separate power source" in the prior-art discussion; and the sensing circuit connects an external source at start-up]. Whether the 14 W input includes the oscillator, flip-flop and FET drivers is not stated in the passage I read. If it omits them, the COP is not comparable. That is the single most important gap.

c. Tested? - By the inventors: reported in the patent, no raw data in the text. - Independent replication claimed by Jean-Louis Naudin (a hobbyist builder), reported as COP 1.76 in a secondary summary [web, from a search result summary at https://www.novakcorp.com/energy/experiments/megdiag.htm; I opened this page and it only links to Naudin's build page, so the 1.76 figure and his method are unverified by me]. - Claims of COP 5, 10 and over 100 on other builds, and of a planned independent university test under Department of Defense auspices that "failed" to happen, appear in advocate sources found by search (e.g. https://www.zpenergy.com/modules.php?name=News&file=article&sid=3918 [web, search summary only, not opened]). I cannot confirm any of these; they are proponent statements. - I found no neutral, controlled, published test with full waveforms and a calorimetric check. No source I could verify says the MEG was tested and failed, either. Hence the accurate classification is "never properly tested", not "tested and failed". - Related: the Springer paper "Explanation of the Motionless Electromagnetic Generator with O(3) Electrodynamics" (Evans et al., Found. Phys. Lett. 2001, https://link.springer.com/article/10.1023/A:1012085429802 [web, title seen in search results]) is a theory paper that takes COP > 1 as given; it is not an experimental test.

How was it treated? Unlike the cold fusion case, I found no evidence of data being reanalysed or publication being blocked; the patent was granted (US 6,362,718, 2002) [read]. Note that a patent grant is not a validation of performance: the USPTO does not require a working demonstration. Both sides have a weak record: the patentees did not publish a calorimetric, full-waveform dataset, and sceptics did not run a neutral test.

d. What would change? If a closed-loop, self-powered run with all supplies inside the measurement boundary were confirmed, conservation of energy would need an energy source; a permanent magnet's stored field energy is finite and is not drawn down by switching its flux path, so it cannot supply continuous power. Nothing in mainstream physics allows this.

The single decisive test: run the self-actuating configuration the patent itself describes (output coil feeding the input coils, external start-up battery disconnected) with a resistive load for hours. A device with a true COP > 1 keeps running and the load stays powered; a measurement artefact stops. In parallel, measure input power at the supply terminals using simultaneous voltage and current sampling and compute the true integrated product, and put the output into a calorimeter. Both are cheap.

Verdict (split): Measured COP > 1 as reported at 10-75 V: Never properly tested (only the patentees' own, unseen, data and unverified hobbyist reports; no neutral test on record; not dismissed by a blocking process that I can document). The 100 V / COP 3.44 figure: Not testable as stated as a measurement, because the patent itself labels it an extrapolation; and it contains an unexplained voltage gain (4000 V vs about 1,125 V by turns ratio). Self-powered continuous operation (true overunity): Refuted by calculation only if the power measurement is correct and complete, which cannot be judged from the patent text; I therefore do not assign refuted, and state the conditional: if input energy is fully counted, conservation of energy rules out a net gain. The most likely explanation is a measurement artefact (reactive power, probe phase, uncounted drive supply), but this is inference, not an established fact.

---

Summary ledger

Fair case for the proponent

Bearden is right that (i) potentials are physically fundamental in quantum mechanics (AB effect), (ii) closed CEM forbids overunity, so any real overunity must come from an uncounted source, and (iii) the MEG is a concrete, cheap-to-test device whose test has not been done in public by a neutral lab. Against him: the 1993 paper contains no data, mixes units, conflicts with polarization and charge experiments, and the one device with numbers does not use the theory.

Sources

• [read] source folder: Bearden/Bearden - Final secret of free energy.pdf and US6362718.pdf.
• [web] https://arxiv.org/pdf/1102.2766 (Bressi et al., neutrality of matter; limit value not extracted).
• [web, summaries only] http://www.rexresearch.com/sweet/1nothing.htm, http://amasci.com/freenrg/sweetvta.html (Sweet VTA claims, no independent test); https://www.novakcorp.com/energy/experiments/megdiag.htm (page opened; Naudin figures not on it); https://www.zpenergy.com/modules.php?name=News&file=article&sid=3918; https://link.springer.com/article/10.1023/A:1012085429802.
• [memory] air-earth current density about 2 pA/m^2 (medium); Maxwell used quaternions in the Treatise and Heaviside recast it (medium/high); AB effect observed by Chambers 1960 and Tonomura 1986 (high). Not web-checked, and the verdicts above do not depend on them alone: the energy-accounting refutation uses only the arithmetic shown.

---

Investigator findings (2026-10-05)

Scope. Weighed here: every entry under "Leads for review" in sites.md and the six research notes dated 2026-10-05 (research/2026-10-05-*.md: Google Patents, EPO Register, archive.org USPTO wrapper, JLN Labs, cheniere mirror, company records). research/files/ is empty (nothing was downloaded), so I worked from the Investigator's notes and quotations, not the primary pages; every [web] item below is the Investigator's reading at the URL it gives, and I did not re-open the pages myself. Items not read by the Investigator (Bearden's 2004 declaration, the last 35,000 characters of the 2004 remarks, EPO annex pp. 4-5, the 2008 annex, the 2007 and 2009 replies, the 2013 minutes) are unknown to me too. The loads I weigh do not rest on [memory] facts except where tagged.

Standing rule applied: the review earlier concluded "I found no neutral, controlled test" and "no evidence of blocking or reanalysis". Each finding below is tested against those two statements.

F1. USPTO continuation 10/081,847: rejection on thermodynamics, demonstration requested, application abandoned (priority 4; bears on claim 5, and on question c "how was it rejected")

[web: https://archive.org/details/gov.uspto.patents.application.10081847, per research/2026-10-05-archive-org-uspto.md; PlainSite docket status "Abandoned -- Failure to Respond to an Office Action"] - 2004-01-23: examiner Shawn Riley rejected all 14 claims under 35 USC 101 ("the device will not operate as an electrical energy generating system"), and under 112 and double patenting. He required the applicants "to demonstrate how the magnetic flux stored within the permanent magnet is sufficient to contribute usable electrical energy ... and how the device then generates more power than what is available in the permanent magnet", and requested a model or demonstration. - 2004-05-24: applicants (attorney remarks plus a Bearden declaration) asked that the demonstration requirement "be held in abeyance". 2004-07-26: final rejection; the request was not held in abeyance; the examiner singled out FIG. 8 and the THS730 sentence. 2005-05-16: abandoned; no appeal in the item list. - Weighing. This changes my "how was it rejected" answer in one respect and not in another. (i) It is the first documented case of a state body rejecting the MEG on theory alone: the examiner did not test anything, and his stated ground is that COP > 1 "violates the first and second laws". By itself that is "dismissed without testing". (ii) But the same examiner asked for a demonstration, which is the neutral test this review has been saying was missing. The applicants deferred it, and let the continuation lapse. So the missing neutral test is not explained by a blocking process: a test was invited and not supplied. Why not is not in the file. Competing explanations as the Investigator sets them out, which I keep unranked: no unit could show net output under examination; abandoning a redundant continuation (double patenting over claims 1-29 of the parent) cost little; trade secrecy (see F5). None is excluded by the record. (iii) The examiner's windmill argument (COP must be below 1 because a windmill's output is below wind power) conflates COP with efficiency, as the Investigator notes; the applicants' distinction was fair. COP > 1 as such is not a violation if an unmetered source is feeding the device, as with a heat pump. The real objection is the other half of the examiner's demand: the patent itself names the magnet's stored flux as the only source and says use demagnetises it [read: patent; web: EP text, F4], and that source is finite. Illustrative order of magnitude (my arithmetic; assumes a 5 cm^3 magnet, which is a guess, and a maximum energy product of about 300 kJ/m^3 for NdFeB [memory: medium, not web-checked]): 300e3 J/m^3 x 5e-6 m^3 = 1.5 J. The patent's extrapolated surplus is 48 W - 14 W = 34 W, which would drain 1.5 J in about 0.04 s. A magnet is a finite store, not a flow, and the verdict does not depend on these two assumed numbers: any plausible magnet gives seconds at most. - Verdict effect on claim 5: none to the verdict, which stays Never properly tested. The classification of "how it was rejected" is now: rejected on theory by the USPTO, demonstration requested, not given.

F2. Bearden's own mechanism, from the 2004 remarks (priority 1; bears on claims 1, 2, 4, 5)

[web: archive.org 2004-05-24 REM text, first 50,000 of 85,447 characters read by the Investigator] The remarks connect the theory papers to the device, which the patent never does. As summarised by the Investigator: the magnet is a broken-symmetry dipole absorbing virtual-photon energy from the vacuum (Lee and Yang 1957), its static field is a Prigogine dissipative structure and a Whittaker bidirectional flow; in the MEG, the Aharonov-Bohm effect localises B in the core and leaves a curl-free A outside, and sharp input pulses perturb A so that "strong E-field energy pulses, by the well-known equation dA/dt = -E" appear in the coils; COP is distinguished from efficiency; a feedback example turns COP 5 open loop into a self-powered loop. - The correction I owe to my earlier note 6 ("the theory papers and the only measured data come from two different devices"): that is still true of the documents in the folder, but Bearden himself asserted the link in 2004, so the MEG is the intended test article of claims 1 to 4. I kept the claim 1 to 4 verdicts separate from claim 5 because the patent did not link them; the file wrapper does. - Mathematics of the link. E = -dA/dt - grad(phi) is textbook (Griffiths ch. 10) and Faraday's law in the coil is exactly that equation; so "dA/dt = -E" is correct but is just ordinary induction. It supplies no energy by itself: the energy delivered into the coil by induction is the energy the pulse source put into changing the flux (Poynting theorem). The step that would make it a new source, "the vacuum supplies extra", is asserted, not derived, in what the Investigator quotes. Also, Lee and Yang 1957 concerns parity non-conservation in weak interactions [memory: high]; I know of nothing in it about a magnet absorbing vacuum energy, and the Investigator's summary gives no derivation. I note it as a stretch of an authority, not as a proof of error, since I have not read the full 85,447 characters. - Verdicts: claims 1 and 2 unchanged (the definitional core is still Not testable as stated; the transverse-wave and charge-variation parts are still Tested, not supported). Claim 4 changes: see F6.

F3. Builders' measurement evidence: Naudin v2.1, Flickinger, Tektronix (priority 2; bears on claim 5, question c)

[web: http://jnaudin.free.fr/meg/megv21.htm; .../megnot01.htm; .../ths720p.htm; .../meg.htm, per research/2026-10-05-jlnlabs.md] What was reported, in the reporters' words: - Naudin v2.1 (status 12-06-00; ferrite magnets, Metglas C-cores, 16-20 kHz driver, 10 ohm shunt, Tektronix THS720P): the control board alone draws 1.75 W; with a 9 W lamp the DC input is 3.25 W; he credits 1.5 W of input to the lamp against "a measured OUTPUT = 6.76 Watts". "Unfortunately measurement artifacts remain possible." A "conditioned" carbon resistor or a MOV (metal oxide varistor) load "is REQUIRED for getting the output datas measured above". In most cases the apparent power "seems greater than the heat dissipated by Joule's effect in the RLoad"; calorimetry and loop closure "not yet done"; the gap "makes me pessimistic" about calorimetry. - Flickinger (independent builder; date not shown on page): COP 1.75 to 5 with MOV loads; with linear resistive loads "a waste of time ... if one expects to see any excess energy"; asks anyone with COP > 1 on linear loads to speak out. - Tektronix customer support (2002, forwarded by Flickinger and posted by Naudin): "the RMS measurement on a math waveform is bogus and has no value"; no "RMS watts" for non-sinusoidal waves; true power is the mean of v(t) i(t). The patent says its power figures "agree with RMS power measured using a Textronic THS730" [read: patent]. Weighing, kept in both directions. - For the proponents: this is the first independent hands-on record in the file that reproduces the "MEG signals" at all, and Naudin states his own doubts instead of hiding them, which is more candid than the patent. Naudin's 1.5 W versus 6.76 W is a COP of 4.5 on his framing. My earlier note that his COP 1.76 was "unverified" stands (the Investigator did not find that figure on the pages read; 1.75 W is the board's no-load draw on the v2.1 page, so check that the two numbers are not confused). - Against: (i) On Naudin's own numbers, total DC input is 3.25 W, not 1.5 W; 6.76/3.25 = 2.1 (my arithmetic), and the 1.5 W figure leaves out the 1.75 W control board. This is the "uncounted drive supply" accounting issue I raised in claim 5b, now seen in a replication: the COP depends on whether the board is inside the boundary. (ii) The excess appears only with nonlinear loads. A nonlinear load draws pulsed current at certain voltage thresholds, which is exactly where V x I averaging and phase error are worst, and my claim 5a note ("product of averages is not average power") applies. This is inference, not a finding. (iii) Apparent power greater than the Joule heat is, if power was measured correctly, a violation of energy balance within the load; the nearer explanation is that the "apparent power" overstates real power. Naudin proposes EM radiation. A sanity check (my arithmetic; assumed geometry 100 turns, 0.01 m^2 loop, 16 kHz so wavelength 3e8/16e3 = 18.75 km; small-loop radiation resistance 31,171 (N A/lambda^2)^2 ohm [memory: high], about 2.5e-13 ohm): radiated power is negligible, many orders below a watt. The 2.8 mT and 1250 V/m near fields he quotes are reactive near fields, which store energy but do not carry it away. So radiation does not rescue the gap; it points back to the measurement. (iv) The Tektronix statement damages the patent's cross-check but not necessarily its primary method. The patent says real power was computed from measured V and I and averaged over the period; that is the right method if done on sampled data. What Tektronix says is invalid is the scope-math "RMS power", which the patent cites as the agreeing second measurement. Agreement between a valid method and an invalid one is not corroboration; it also does not show the first was wrong. - Classification: these are informal builder reports, uncontrolled, with no calorimetry, loop closure or waveform record published. They are not a neutral test. For linear loads they are, however, informal evidence against excess (one builder, undated, uncorroborated here), and I state that rather than burying it. - Verdict effect on claim 5 (measured COP > 1): unchanged label, Never properly tested, with the added sub-finding "informal, uncontrolled builder reports: excess seen only with nonlinear loads, none with linear loads, no calorimetry". Confidence that the reported excess is a measurement artefact rises from "most likely inference" to "better supported inference". It is still inference; no one has shown the artefact by calorimetry or by a closed-loop run, either.

F4. European patent EP 1 446 862 B1 and the EPO file (priorities 1, 3 and 4; bears on claim 5)

[web: https://patents.google.com/patent/EP1446862B1/en; https://register.epo.org/application?number=EP01988155; per research/2026-10-05-google-patents.md and -epo-register.md] - Granted 2013-10-23 after 12 years, two "deemed withdrawn" episodes cured by further processing, no opposition. Granted claim 1 requires non-saturating switching and feeding output power back to the switching circuit (self-power); claim 2 adds the start-up battery and switchover. The EPO's 2006 annex held that US 4,006,401 (deRivas 1977) "discloses a motionless electromagnetic generator having the same structure"; the only difference was that in E1 the coil power came from the continuously applied source, "not from the permanent magnet". Other cited art: US 4,077,001, US 4,883,977, FR 2,312,135 (Galley 1975), US 2,866,943. - The granted EP description keeps the 3.44 extrapolation and the demagnetisation paragraphs but, as read by the Investigator, drops the COP-versus-voltage plot and the sentences "output power was substantially higher than input power", the THS730 agreement, and "all significantly greater than 1" [web, text of both documents compared]. Drawing sheets and pages were replaced on 2005-06-15; the reason is in the unread file. Weighing. - Strongest honest case for the proponents: a second patent office, with a different standard from the USPTO's, found the claimed self-powered loop novel and inventive over decades of flux-switching generators and did not (in the pages read) refuse on thermodynamic grounds. The applicants also paid for 12 years of prosecution. - Against and neutral: (i) a grant is not a performance test and the EPO annex pages read concern Art. 84, 54 and 56, not Art. 57 industrial applicability; whether it was raised later is in unread documents. (ii) The point matters for the physics: the EPO's own reading makes the self-power feedback the only novelty over a 1977 device. That is the one feature for which no demonstration exists. The patented novelty and the unverified claim coincide. (iii) The plot of the measured points (the only measured data in the project) disappeared from the European text while the unmeasured, extrapolated 3.44 stayed. I cannot say why: it may be routine conforming of drawings under examiner objection, or a deliberate trimming of evidence. Both are open; the file would decide. I record it and draw no conclusion about intent. - Verdict effect: none. It sharpens the settling test: the granted European claim is exactly the self-powered, non-saturating, battery-disconnected run named as the decisive test in claim 5.

F5. Bearden letter, 15 Aug 2002, and the company's state (priorities 1 and 4; bears on claim 5, absence of evidence)

[web: https://billstclair.com/www.cheniere.org/correspondence/081502.htm, mirror of cheniere.org, whose own domain showed only a privacy banner; per research/2026-10-05-cheniere-mirror.md] Bearden, as co-inventor and company spokesman (an interested party): Magnetic Energy Ltd has "a series of successful little experimental devices that prove the overunity and the concepts"; no finished power units; $9 million a year for three years sought; "We do not sell stock to the public, put on public demonstrations, etc. Instead, we give demonstrations (with nothing hidden) to potential financial partners, whose technical representatives do their own measurements"; "every previous legitimate COP>1.0 system ... has been firmly defeated right at this same 'vertical cliff'" of scale-up funding; plans for 2.5 kW and 10 kW units. Postscript: a charged capacitor or electret laid across a magnet with E perpendicular to H is a "proven 'free energy generator'" by Poynting's E x H. Weighing, same rigour for the inventor as for his critics: - "Prove the overunity" is stated with no number, no measurement and no named witness. Investor-only demonstrations whose reports have not surfaced is a secrecy explanation consistent with the record and also consistent with there having been nothing to show; the record cannot separate them. The "vertical cliff" explanation for why every COP > 1 device fails at scale is unfalsifiable as framed, the same pattern I flagged in 1994 (failure with good materials is blamed on the materials). - Not suppression: nothing in the letter, or elsewhere in the record, says a regulator, journal or funder blocked the work. He says funding was sought and in negotiation; who declined, if anyone, is unknown. - The postscript device is a different, simpler claim, and it is checkable by textbook calculation: for static E and H, div(E x H) = H . curl E - E . curl H = 0, so the Poynting flux circulates with no net divergence and no energy is delivered anywhere [memory: high; Griffiths ch. 8]. The circulating flux is a known curiosity (hidden momentum, the Feynman disk puzzle), not a power source. "Proven" has no data behind it in the file. I treat it as part of claim 4 (a specific hidden-source device), Never properly tested as a measurement, with the textbook calculation counting against any net output.

F6. Corporate and ownership trail: Magnetic Energy Ltd, Colmac Holdings (priority 4; absence-of-evidence analysis)

[web: https://patents.google.com/patent/US6362718B1/en; Alabama SOS https://arc-sos.state.al.us/CGI/CORPNAME.MBR/INPUT; Google results, per research/2026-10-05-google-patents.md and -company-records.md] - Patent filed 2000-09-06, granted 2002-03-26, in the inventors' names until a nunc pro tunc assignment (2006-06-01, effective 2006-05-30) to Magnetic Energy, Ltd., Alabama; corrective assignment 2006-06-08; security agreement 2006-06-12 (reel/frame 017759/0537) naming Colmac Holdings Limited, Canada; lapse for the 4-year fee, petition, reinstatement 2006; 8-year fee late (2010) and 12-year fee (2013) paid; expired 2020. Alabama SOS: no entity named "Magnetic Energy" (only an unrelated non-profit); directory listings (Manta, a chamber listing) show a Huntsville "Magnetic Energy Ltd" and a Bearden paper gives a P.O. box there. - Competing explanations for the silence after about 2005, with the evidence each way, kept open. (a) Commercial failure: a pledge to a lender in 2006, a late-paid fee in 2010 and no product are consistent; so is any other. (b) The device did not scale or did not show net output: consistent with the examiner's finding (F1) and builders' reports (F3), not proved by them. (c) Trade secrecy: the letter (F5) says demonstrations were private; but a patented device whose structure is fully disclosed has little left to keep secret except the measurements. (d) Absorption into another firm under another name: the Colmac pledge is the only lead; unidentified. (e) Suppression: no document found supports it; no document found excludes it. Silence in the Alabama register is not evidence of non-existence (the Investigator notes the entity may be registered in another jurisdiction or under a different exact name; Manta lists one). - The record sustains some expenditure to 2013 (fees, a 12-year European prosecution), which is against the simple story "the project vanished in 2002". It does not show any technical result in that period. Decisive records: the Colmac security-agreement text and any release (USPTO Assignment Center), Colmac in Corporations Canada, the EPO file pages not read, any investor-demonstration measurement report, the parent 09/656,313 file wrapper (not on archive.org). - Verdict effect: none. This is an unresolved anomaly, now with a documented financing trace.

F7. Related prior art and the O(3) paper (priority 3; bears on claims 4 and 5)

• Static flux-switching generators before the MEG [web: Naudin index and EPO annex]: Starr US 2,802,170 (1957), Toffolo US 3,087,108 (1963), Garron US 3,368,141 (1968), Galley FR 2,312,135 (1975), deRivas US 4,006,401 (1977), Richardson US 4,077,001 (1978), Flynn WO 00/07285 (Parallel Path Technology), and others. These are listed as related work, whatever one thinks of them. The Investigator found no performance data for any of them showing COP > 1, and I found none either; I do not read anything into that, since I did not search for it. What the list does show is that the MEG's switching structure is old, and that a hobbyist community and a patent office both place its novelty in the self-powering feedback.
• Found. Phys. Lett. 14(1), 2001 (the Evans et al. O(3) paper I cited in claim 5c): the Naudin index lists Bearden among its 15 AIAS authors [web: jnaudin.free.fr/meg/meg.htm]. I had called it a theory paper that takes COP > 1 as given; I now add that a patentee is a co-author, so it is not independent support. It remains untested theory.

Changes to my verdicts

Questions a-d for the new material: (a) the only new mathematics is the 2004 E = -dA/dt argument, internally fine and energetically empty; (b) builder results are inconsistent with energy balance if the power figures are right, which is why they point to the measurement; (c) new tests: Naudin (uncontrolled, no calorimetry, nonlinear loads), Flickinger (uncontrolled, no excess on linear loads), a requested USPTO demonstration (not given), investor demonstrations (reports not found); no replication in the controlled sense; (d) unchanged: energy conservation would need an additional source, and a permanent magnet is a finite store.

Web checks this pass: none new. The Investigator's [web] statements are carried with their URLs. My own [memory] items (Lee and Yang 1957; NdFeB energy product about 300 kJ/m^3; the small-loop radiation formula; div(E x H) = 0 for static fields) are not web-checked; no verdict in the table rests on any of them alone.

Sources added

• [web, via Investigator notes, pages not re-opened by me] https://archive.org/details/gov.uspto.patents.application.10081847 ; https://patents.google.com/patent/US6362718B1/en ; https://patents.google.com/patent/EP1446862B1/en ; https://register.epo.org/application?number=EP01988155 ; http://jnaudin.free.fr/meg/meg.htm, megv21.htm, ths720p.htm, megnot01.htm ; https://billstclair.com/www.cheniere.org/correspondence/081502.htm ; https://arc-sos.state.al.us/CGI/CORPNAME.MBR/INPUT .
• [read] research/2026-10-05-*.md (six notes) and sites.md.