Bearden: Arbiter (Step 3)

Inputs reconciled: facts.json, claims.md, physicist.md, mathematician.md, contrarian.md. Updated 2026-10-05 with sites.md, research/ and the reviewers' Investigator-findings sections. Tags: [read] = I read it in the source PDFs (text layer); [web] = URL given; [memory] = recalled, confidence stated. Where I rely on a reviewer's [read] quote that I did not re-check, I say so.

Source gap note (repeated)

Both source documents have a usable text layer (the 1993-94 paper bundle, 49 pages; US 6,362,718). Gaps that still stand: - [Superseded by the QE revision below] The patent's Figs. 6-8 were originally unread; they have since been read as page images (primer, QE, deep read). - [Superseded] The 13 May 1994 paper (pp.42-49) has now been read in full (deep/ read). The 15 Feb 1994 update (pp.28-41) has since been deep-read (see QE revision, Round 2). - No test, replication or suppression claim in this file comes from the documents themselves; they come from web sources named below. Each is identified by who wrote it and what stake they hold: Bearden himself (co-inventor of US 6,362,718, which lists five individual inventors and no assignee [read: patent front page], and spokesman for Magnetic Energy Ltd, the private company he describes as developing the MEG [web, Check 8]; so an interested party with a commercial stake), J. L. Naudin (an independent private builder with no stated commercial stake in the MEG), free-energy news and archive sites (ZPEnergy, rexresearch, amasci). Where a verdict rests on [memory] alone I say so. (Revised 2026-10-05; see "Evidence-rule revision" at the end.)

Corrections to earlier steps (settled by me)

Web checks done (budget 8, used 5)

1. Maxwell 1865 gravity: in section 82 he derived that gravitational field energy would be negative, called this impossible for a medium, and wrote that he "cannot go any further in this direction" [web: https://royalsocietypublishing.org/rsta/article/373/2039/20140473/114863/a-paper-I-hold-to-be-great-guns-a-commentary-on ; summary via search, text of the paper at http://www.bem.fi/library/1865-001.pdf not opened]. Upgrades the physicist's/mathematician's [memory, medium] to [web]: Maxwell did not have a gravity-EM unification. The 1865 paper pre-dates the 1873 Treatise, so this refutes "Maxwell got there" only as far as Maxwell's own stated attempt; I have not checked the Treatise itself.
2. Neutrality of matter: electron-proton charge difference limit about 1e-21 e (Bressi et al. 2011) [web: https://arxiv.org/abs/1102.2766]. Upgrades the [memory, high] figure. Note what it constrains: the neutrality of ordinary matter. It supports, but is not a direct test of, "electron charge is independent of ambient potential".
3. MEG tests: Naudin 1.75 [web: see table; revised 2026-10-05, see Check 6]. No neutral, calorimetric, accredited-lab result turned up. ZPEnergy (A. Craddock, "Twenty years after Tom Bearden's MEG", 1 Jan 2021, page opened 2026-10-05) reproduces Bearden's 2000 statements and records no later test or product [web: https://www.zpenergy.com/modules.php?name=News&file=article&sid=3918].
4. Sweet VTA: the account in the sources is that Sweet reported about 330 microwatts in and over 500 W out, never had the device independently tested, kept the magnet-conditioning process to himself, and (per Bearden) feared for his life [web, search summary: http://amasci.com/freenrg/sweetvta.html, http://www.rexresearch.com/sweet/1nothing.htm; pages not opened by me]. These are Sweet's own claims as relayed by Bearden, his collaborator and promoter (an interested party), and by archive sites. Evidence status: documented in Sweet's and Bearden's accounts; no independent measurement found in either direction; no second source found yet for the "fear for his life" reason. The absence of a test is analysed under Claim 4 below. (Revised 2026-10-05: removed "source is advocate-side, so ... reliable", which weighed the account by label.)
5. Still [memory] only: NdFeB (BH)max about 300 kJ/m^3 (medium) and magnet stored energy [updated 2026-10-05: now [web], range 190-450 kJ/m^3, giving 7-22 J; see "Investigator findings (2026-10-05)", web check 11]; Maxwell used quaternions in some Treatise passages (medium); air-earth current about 2 pA/m^2 (medium). No verdict below depends on these alone.

Web checks added in the 2026-10-05 evidence-rule revision (about 10 calls; slightly over budget, because several earlier [web] items rested on search summaries of pages nobody had opened): 6. Naudin MEG v2.0 page opened [web: https://www.novakcorp.com/energy/experiments/megv2.htm]. Verbatim: "The measured COP is 1.75" (11-07-00). On closing the loop: "I am trying to close the loop, but at this moment, I have not yet succeed in this way. There are many losses during the step-down phase of the output voltage and thus the device stops after a while..." On 11-16-00: "the two secundary coils have been broken due to some sparks inside the windings generated by the presence of the High Voltage ... the High Voltage output climbs very quickly at a high value ( > 3KV ) at certain frequencies". He planned a v3.0 with nanocrystalline cores. 7. Bearden e-mail, 2 Dec 2000 [web: https://billstclair.com/www.cheniere.org/correspondence/120200.htm]. Verbatim: "arrangements have been made for the university's scientific team to not only test it, but to physically and independently replicate it from scratch, under U.S. Department of Defense auspices"; names withheld because "After observing the fierce dogmatic hassle that the orthodox establishment poured on cold fusion ... you can understand why we are not releasing which university, which DoD agency, and which independent test laboratory"; "We have obtained COP = 17 to 20, higher in some experimental build-ups, and we deliberately pushed one unit to destruction of the output coil, with the unit reaching COP>100"; Naudin "obtained COP = 7.3 in his later (third) build-up after an initial success of COP = 1.3". 8. Outcome of the university test: two search summaries state that the university scientists "observed the MEG in operation and were quite interested in performing full, independent replication", but "the university's administration refused to sign a noncircumvention agreement, and consequently the process was stopped" [web, search summary only; the underlying page, apparently Bearden's own MEG material, was not identified or opened]. A cheniere.org MEG status page of 11 April 2001 describes Magnetic Energy Ltd as "a private company and in business" and "in intense negotiations with several large financial enterprises, for the capitalization necessary"; it gives no COP measurement [web: https://www.rootaction.net/~jay/science/www.cheniere.org/megstatus.htm]. 9. Fogal device: William J. Fogal holds US 5,196,809 (filed 1 March 1991) and US 5,430,413 (filed 22 March 1993, granted 4 July 1995), "High gain, low distortion, faster switching transistor". Abstract: "A transistor in which the emitter terminal is coupled to ground through an electrolytic capacitor. The electrolytic capacitor has a capacitance of from about 0.2 μf to about 100 μf ...". It claims gain "at least twice as great as the gain for a transistor without an electrolytic capacitor" and makes no claim of output power above input power or of energy from the vacuum [web: https://patents.google.com/patent/US5430413A/en, fetched summary]. A rexresearch compilation of Fogal articles and patents exists [web: https://www.rexresearch.com/FogalTransistor/FogalChargedBarrierTransistor.html, not opened]. 10. Evans et al., Found. Phys. Lett. (2001) MEG paper: Springer page would not load (login redirect). [memory, medium, not web-confirmed]: Bearden is a co-author of that paper (AIAS group). If so, it is a co-inventor's own theoretical account, not third-party support.

Claim 1: Energy = potential = ordering in the virtual-photon flux; scalar potential contains an infinite series of wave pairs linking each charge to the universe

Agreement: all three accept Whittaker 1903/04 as real mathematics with no energy content; all three find the "energy and potential are identical" statement wrong in units. Disagreement: physicist calls the "infinite energy" part Refuted by calculation (finite U = (eps0/2) integral E^2 dV); mathematician says no contradiction is derived, only asserted, and rates the identification Internally inconsistent; contrarian rates the whole thing Not testable as stated. Weighing: Each is right about a different sentence. - Units: potential is J/C, energy is J. Bearden himself says "it's used not really as energy, but rather as excess energy per coulomb" and then writes K_T = V_T x Coul_T [read]. That is the paper switching definitions mid-argument: Internally inconsistent (mathematician and contrarian both found it; the physicist too). - "Infinite energy": a worked check. A 12 V, 1 uF capacitor stores (1/2)CV^2 = 72 uJ, finite. If "infinite energy" means stored electrostatic energy, it is refuted by that number. If it means a reservoir that the vacuum refills, then no equation or measurement is offered, so it is not testable. The physicist's refutation applies only to the first reading; the contrarian's rating to the second. I keep both. - AB effect is a legitimate point in his favour: potentials matter for quantum phase (contrarian; mathematician agrees). It does not give a reservoir. Tested/never tested/dismissed: no test exists because nothing measurable is predicted; nobody dismissed it (it was never put in testable form). What would change: the definition of energy (Noether link to time symmetry) and QED's vacuum, which has a ground state with no extractable work.

Final verdict (split): - Whittaker/Ziolkowski decomposition as mathematics: Established (carries no energy). - "Energy = potential" as written: Internally inconsistent (units). - "Each potential holds infinite energy from the universe": Refuted by calculation if read as stored field energy; otherwise Not testable as stated.

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

Agreement: all three reject it. All note the third-law argument fails because fields carry momentum. Disagreement: only on label. Physicist adds Refuted by calculation for the longitudinal part; mathematician adds Internally inconsistent; contrarian keeps "Tested, not supported" and "Not testable" for the rate-of-time-flow statements and notes the partly-right AB point. Weighing: In vacuum div E = 0 forces k.E = 0 for a plane wave, a two-line calculation, so a longitudinal vacuum EM wave is excluded by the same Maxwell equations it appeals to. Experiments agree: polarisers/Malus's law, Hertz (memory, high, textbook). Charge: neutrality bound 1e-21 e [web]. Mathematician's product-rule error is verified [read], and his units point (V kg/s^2 is not amperes) is simple dimensional analysis I accept. Tested/never tested/dismissed: Tested, but by the pre-existing body of experiment, not by an experiment designed around Bearden's version; he proposed none. Not dismissed without testing, since the old experiments already contradict it. Honest caveat: "tested, not supported" here means contradicted by existing data, not a Bearden-vs-mainstream controlled trial. Final verdict (split): - Transverse vs longitudinal vacuum waves: Tested, not supported, and Refuted by calculation (div E = 0). - Charge varies with ambient potential: Tested, not supported (neutrality of matter, constant e in accelerators). - "Derivative of a product" step: Internally inconsistent. - "EM wave = oscillation of the rate of flow of time", photon-antiphoton to graviton: Not testable as stated.

Claim 3: Collector scheme: charge a long-relaxation-time collector "current-free", then discharge into a load, so a battery can run a car or grid

Agreement: all three agree the strong form fails on the energy budget, and that no data was ever published. Disagreement: physicist and mathematician say Refuted by calculation. Contrarian splits it: strong form refuted, weak form (reduce loss inside the source) Never properly tested, mechanism Internally inconsistent. Recomputed (all numbers match across reviewers): C = 1 uF, V = 12 V. Q = CV = 12 uC. Battery outlay QV = 144 uJ. Capacitor stores (1/2)CV^2 = 72 uJ. Resistive loss 72 uJ for any R. Load can get at most 72 uJ; COP at most 0.5. Bearden's K_T = V_T x Coul_T equals the 144 uJ the battery gave up, not the 72 uJ in the capacitor, so his "free" quantity is the source's own expenditure. Relaxation time: eps0/sigma for copper = 8.854e-12 / 5.96e7 = 1.49e-19 s (his 1.5e-19 s is right) [recomputed]. It is the decay time of free charge inside a bulk conductor, not a delay before current flows; the delay in a circuit is RC. His own design target of 1 ms with a "good conductor" gives sigma = eps0/tau = 8.85e-9 S/m, which is an insulator [recomputed]: Internally inconsistent. Quartz about 50 days vs his "10 days": same order, not an error [web: https://arxiv.org/pdf/0801.4182 per mathematician; I recomputed 8.85e-12/1e-17 = 8.9e5 s, about 10 days, using sigma = 1e-17 S/m, so the number depends on the chosen sigma]. The weak form, stated fairly: dissipation in a charging path can be reduced by charging in many small steps (adiabatic charging; the loss falls toward zero as steps increase) [memory, high, standard result]. That is real engineering. It lowers loss; it never makes output exceed what the battery gave up, so it is not a free-energy mechanism. The "weak form" the contrarian leaves as never tested therefore has no overunity content. Bearden's own text: "the invention violates the closed circuit practice ... but does not violate the conservation of energy" [read], while fn. 25 allows violating local conservation with a hidden source [read]. Tested/never tested/dismissed: never built with published data (Bearden publishes none; 1994 text says a competent builder could do it, per contrarian [read]); not dismissed without testing, no documentation of blocking. Bearden released the scheme openly rather than patenting it: "In this paper, I am freely giving away what required me an arduous 30 years of my life to discover" and hopes it "will trigger a thousand, or even ten thousand, scientists and engineers to develop overunity energy devices" [read: p.4]. The silence that follows (no published build of the collector, by him or by others, in any source I found) is therefore not explained by secrecy or a withheld device. Competing explanations: (a) builders tried and got no gain and did not publish null results (consistent with the energy budget above; no record of such attempts found); (b) the scheme was too under-specified to build (no component values; the 1 ms "good conductor" target is an insulator); (c) builds exist but are unrecorded in the sources searched. A search of free-energy forums and archives for collector builds with measurements would decide between them. (Added 2026-10-05.) His framing that failure with good wires means bad wires is unfalsifiable (contrarian [read], not rechecked). What would change: circuit theory/Poynting for an ordinary RC circuit. Final verdict (split): - Strong form (battery never depleted; runs a car/grid): Refuted by calculation (energy budget above). - Mechanism "current-free charging via relaxation time": Internally inconsistent. - Weak form (lower-loss charging): Plausible, untested as a loss-reduction, but yields no gain; never built or measured by Bearden.

Claim 4: Closed classical EM cannot give overunity; real overunity devices need a hidden vacuum source; Maxwell's quaternion theory, truncated by Heaviside, held this and gravity-EM unification

Agreement: all three say the conditional ("closed CEM cannot be overunity") is correct (Poynting's theorem), and that "therefore CEM is flawed" is a choice, not a deduction. Disagreement: physicist rates the hidden-source part Not testable as stated; mathematician allows Plausible, untested or Never properly tested; contrarian Never properly tested for the Sweet VTA, Not testable for the general principle. Weighing: A "hidden source" with no formula, no location and no measurable signature cannot be tested, so Not testable as stated is right for the general principle. The one named device, Sweet's VTA, is a testable claim: Sweet reported about 330 microwatts in and over 500 W out, and the accounts say he withheld the device and the magnet-conditioning process from independent test [web, Check 4: Sweet's claims as relayed by Bearden and archive sites; no second source found yet for the stated reason]. That is neither "tested and failed" nor "dismissed without testing". The record then goes silent: no independent measurement, no surviving device in a lab, no published conditioning process. Competing explanations, with what bears on each: (a) the effect was a measurement artefact and withholding prevented its discovery (consistent with no test; no direct evidence); (b) trade secrecy, the inventor protecting an unpatented process (consistent with the reported secrecy of the conditioning step); (c) fear of harm, as Bearden states (one source, Bearden's account; no corroborating record found yet); (d) the process was lost with the inventor [memory, medium, not web-checked: Sweet died in the mid-1990s]. What would decide: any surviving unit or Sweet's notes, and any contemporaneous measurement by a third party (the rexresearch and amasci compilations, not opened here, are the place to look). (Revised 2026-10-05.) Maxwell: his 1865 attempt at gravity as a field theory failed on negative energy [web, see Check 1]; the equivalence of quaternion and vector forms for classical EM and Heaviside/Gibbs's reduction (20 equations to 4) are standard [memory, high, secondary web: https://medium.com/@zachbobby8/maxwells-quaternions-vs-heaviside-s-vector-calculus-699907b2843d]. Whether the 1873 Treatise itself contains a gravity-EM link: I found no source saying it does [web negative result], and Bearden gives no derivation. Final verdict (split): - Closed CEM cannot be overunity: Established. - A hidden, freely replenished vacuum source (general): Not testable as stated. - Sweet's VTA specifically: Never properly tested (no independent test; inventor withheld it). - "Maxwell's original theory contained gravity-EM unification": Tested, not supported (historical record: his own 1865 attempt failed on negative field energy [web, Check 1]; I found no source for a later unification, and Bearden gives no derivation). - Quaternion-to-vector change: Historically superseded but instructive. - (2026-10-05, Investigator) Bearden's 2004 USPTO remarks tie the theory to the MEG (magnet as vacuum-absorbing dipole; E = -dA/dt pulses). The mechanism: Not testable as stated (the induction step is correct and energetically empty). The MEG self-powered run as the test of a vacuum source: Never properly tested. Capacitor/electret across a magnet as a "free energy generator" (2002 letter): Not testable as stated; static div(E x H) = 0 gives no net power.

Claim 5: The MEG (US 6,362,718): COP above 1 (extrapolated 3.44 per coil at 100 V); can self-power after start-up

This is the claim where the reviewers disagree most. Physicist: Never properly tested (strong form expected to fail). Mathematician: strong form Refuted by calculation, weak form Never properly tested. Contrarian: Never properly tested; refuted only conditionally on a complete power measurement; most likely artefact.

Recomputation: - 100 V x 0.14 A = 14 W; 4000 V x 0.012 A = 48 W per coil; 48/14 = 3.43. - Magnet volume 40 x 25.4 x 38.1 mm = 38.7 cm^3 = 3.87e-5 m^3 [read dimensions]. At about 300 kJ/m^3 (BH)max [memory, medium] stored energy is about 11.6 J; at B^2/(2 mu0) = 573 kJ/m^3 (B = 1.2 T) about 22 J. Claimed excess 2 x 48 - 14 = 82 W would exhaust 12-22 J in 0.14-0.27 s. [Updated 2026-10-05, see "Investigator findings (2026-10-05)", web check 11: (BH)max is now [web] at 190-450 kJ/m^3 for sintered NdFeB. Recomputed: 3.871e-5 m^3 x 190e3 J/m^3 = 7.35 J; x 450e3 = 17.4 J; x 573e3 = 22.2 J. At 82 W: 7.35/82 = 0.090 s to 22.2/82 = 0.27 s. The range to use is 7-22 J and 0.09-0.27 s; the 300 kJ/m^3 figure is the middle of it (11.6 J, 0.14 s). No verdict changes.] - Stronger form of the objection: in steady state the B-H path closes each cycle, so average power out plus losses must equal average power in. A magnet that is not demagnetised cannot supply a continuing average.

Which argument is stronger: The mathematician's magnet-drain figure is a valid supporting bound but not the heart of it, and the contrarian is right that it cannot by itself decide whether the patent's measurements are wrong; it shows only that, if the output numbers are real and sustained, the magnet cannot be the source. The decisive point is conservation of energy over a closed cycle, which does not depend on the magnet figure. The claim has two different parts, and they get different verdicts: 1. The empirical report (the patentees measured output above input, 10-75 V): this is a measurement claim. The patent supplies no calibration, bandwidth, phase angle or uncertainty; the output "4000 V average" times "12 mA average" is a product of averages, which equals average power only for in-phase waveforms, and 4000 V is 3.6x what the turns ratio gives, so spikes or resonance (which store and return energy) are in my view the likely route to a large V x I [read + recomputed; interpretation is inference]. The patent says its real-power calculation agrees with oscilloscope RMS power [read], which is a check of arithmetic on the same probes, not an independent instrument. Measured by the patentees themselves; raw data not published (Figs. 6-8 since read, see QE revision). Never properly tested. 2. The physical claim (sustained net output from a closed, fuelless cycle, including the self-running form): this is exactly what energy conservation excludes, and the exclusion needs no further experiment, only that all inputs are inside the boundary, which is what "self-powered" asserts. Refuted by calculation. This does not say the patent's bench numbers are wrong; it says that if they are right, the device was not closed. Interested parties: everything known about MEG performance comes from people with a stake in it. The patent's description (flux switching, "receives static energy from the magnetic flux of the permanent magnet", COP above 1) is the five patentees' characterization, written to obtain a patent; Bearden's later statements are those of a co-inventor speaking for the developing company. They are reported below as their accounts, beside the one outside builder's (Naudin's) and the arbiter's own calculation. The USPTO grant is an agency act whose stake is procedural; a grant does not require a performance test [memory, high]. Tested/never tested/dismissed (contrarian's distinction, kept; rewritten 2026-10-05 to state each account): - The patentees: COP about 1.7-2.6 measured at 10-75 V, 3.44 per coil extrapolated at 100 V [read]; self-running "can be built" and "it is apparent" from the COP, no run reported [read]. - Magnetic Energy Ltd builds, per Bearden (2 Dec 2000): "COP = 17 to 20, higher in some experimental build-ups", and one unit "pushed ... to destruction of the output coil" reaching "COP>100" [web, Check 7]. Documented in Bearden's own correspondence; no waveforms, input/output values or second source found yet. Physicist's earlier "COP 5 and 10" are from the same family of Bearden accounts. - J. L. Naudin (independent private builder, own instruments, no neutral observer): v2.0 "measured COP is 1.75" (7 Nov 2000); output coils destroyed by sparking when output exceeded 3 kV; and, pertinent to the self-running claim, "I am trying to close the loop, but at this moment, I have not yet succeed ... the device stops after a while" [web, Check 6]. Bearden reports later Naudin builds at COP 1.3 and 7.3 [web, Check 7; Naudin's own page for these not opened]. Naudin's failed loop closure is the only recorded attempt at the self-running form that I found; he attributes the stop to step-down losses, not to absence of excess power. It is consistent with the energy-budget verdict, but one builder's unsuccessful attempt is not a decisive test. - University/DoD independent replication: Bearden wrote in Dec 2000 that it had been arranged "under U.S. Department of Defense auspices", names withheld [web, Check 7]. The only account of its outcome I found (search summaries) says the scientists observed the MEG and wanted to replicate it, but "the university's administration refused to sign a noncircumvention agreement, and consequently the process was stopped" [web, Check 8, search summary; source page not identified]. On that account the test was stopped over a contract term demanded by the developers, before any measurement; it is neither a failed test nor a refusal to look on technical grounds. - Evans et al., Found. Phys. Lett. 2001, explains the MEG with O(3) electrodynamics and takes COP > 1 as given; it is not a test [web: https://link.springer.com/article/10.1023/A:1012085429802]; Bearden is a co-author [memory, medium, Check 10], so it is not third-party support. So: NOT "tested and failed" (no neutral controlled test is on record); NOT "dismissed without testing" by the mainstream (the patent was granted; no blocked publication is documented; the one arranged independent test stopped, on the only account found, over a non-circumvention agreement). Closest: "nobody neutral has tested it". Absence of evidence (added 2026-10-05): the record goes silent where it would be expected to continue. In 2000-2001 Bearden reported COP 17-20 builds, an arranged DoD-auspices replication, capitalization negotiations and scaled units "in the not too distant future" [web, Checks 7-8; ZPEnergy 2021]. Twenty years later ZPEnergy (Jan 2021) records no test, product or later data [web, Check 3]. Competing explanations: (a) the excess was a measurement artefact (reactive power, uncounted drive supply, ringing; supported by the patent's 1500-4000 V vs turns-ratio discrepancy, the missing power factor, and Naudin's failed loop; against it: no full-waveform analysis of any MEG has been published either way); (b) commercial failure, no capital raised (supported by the 2001 "intense negotiations" with no announced result; no record of the company's end found); (c) trade secrecy, the developers withholding detail to protect IP (supported by the non-circumvention demand and withheld names; but the patent itself is public, so the basic design is not secret); (d) absorption into another firm or programme under a new name (no evidence found); (e) suppression (Bearden's framing of the establishment's "fierce dogmatic hassle" is offered as the reason for secrecy, not as evidence that anyone blocked the MEG; no record found of interference). What would decide: Magnetic Energy Ltd corporate records (registration, dissolution date), any later patent assignments or continuations naming the inventors, the identity of the university and DoD programme and any internal memo, and Naudin's v3.0 results. Important structural point: the patent never mentions the vacuum, scalar potential or Whittaker (physicist grep, zero hits; contrarian confirmed [read]). The theory (claims 1-4) and the only device with data are not linked in the documents. Success or failure of the MEG would say nothing about claims 1-4. What would change if true: conservation of energy for a fuelless device, or magnets would need an unmeasured energy store. Final verdict (split): - Reported COP > 1 at 10-75 V: Never properly tested (only the patentees' and Naudin's own measurements; no published power waveforms or power factor; measurement artefact is the arbiter's inference, not established; the one arranged independent replication was stopped before measurement on the only account found). - 3.44 at 100 V: Not testable as stated (an extrapolation, with an unexplained 4000 V vs 1,125 V; also 3.43 and per-coil ambiguity). - Sustained self-powered net output: Refuted by calculation. - (2026-10-05, Investigator) The reported COP verdict stands with new sub-findings: the patent's cross-check (scope "RMS power", THS730) is the method Tektronix support calls invalid for non-sinusoidal waveforms (one support reply relayed by a builder; applicability to the patentees' exact procedure is inference); builders report excess only with nonlinear loads (Naudin v2.1, Flickinger; single sources, uncontrolled); no calorimetry or closed-loop run is reported. The US continuation was rejected on theory (35 USC 101) and a demonstration was requested and not supplied; the EPO granted a patent whose claim 1 requires the self-powering loop. So: rejected on theory by one examiner, invited to demonstrate, not demonstrated; not "tested and failed". Updated explanations for the silence are in "Investigator findings (2026-10-05)".

Corrected claim list

1. Energy/potential/VPF/wave pairs: Whittaker math Established (no energy content); "energy = potential" Internally inconsistent (units); "infinite energy / universe link": Refuted by calculation (stored-field reading) or Not testable as stated (reservoir reading).
2. No vacuum fields; longitudinal waves; charge varies: Tested, not supported (existing experiments) and Refuted by calculation (div E = 0); product-rule step Internally inconsistent; time-flow/graviton statements Not testable as stated.
3. Collector: Refuted by calculation (strong form; COP at most 0.5 per cycle); mechanism Internally inconsistent; lower-loss charging Plausible, untested, no gain. 3b. (added by QE revision) Fogal charge-blocking-semiconductor device, 13 May 1994 paper, Figs. 1-2: "very high variable power gain" Never properly tested (no data) and Not testable as stated in its meter-based form; closed-loop self-powered "negative resistor" Refuted by calculation; dPhi/dt current decomposition Internally inconsistent (product rule, units). 3c. (added by QE revision, Round 2) 15 Feb 1994 update: alpha (load power / source-internal dissipation) > 1 Established as ordinary, not overunity; "conventional operational efficiency always below 1" Refuted by calculation; "efficiency greater than unity" via vacuum influx Not testable as stated (sustained self-running form Refuted by calculation); step-charged capacitor collector Plausible, untested as loss reduction (eta = N/(N+1) < 1), no measurement reported; "local energy conservation does not apply to open systems" Internally inconsistent with the paper's own global-conservation claim and unsupported (no measured boundary flux).
4. Closed CEM / hidden source / Maxwell: conditional Established; hidden source Not testable as stated; Sweet VTA Never properly tested; Maxwell-gravity unification Tested, not supported (historical evidence: Maxwell's own 1865 attempt failed; no source for a later unification) [relabelled by QE revision, was "Refuted by historical record", which is not on the scale]; quaternion-to-vector Historically superseded but instructive.
5. MEG: reported COP > 1 Never properly tested (measured range now read: COP about 1.7 at 10 V to about 2.6 at 75 V); 3.44 at 100 V Not testable as stated (curve extrapolation); sustained self-powered gain Refuted by calculation, and the patent itself names the magnet's flux as the fuel and concedes demagnetisation. 4/5 (added 2026-10-05, Investigator): claim 4 gains two lines: the 2004-remarks mechanism Not testable as stated, and the MEG self-powered run as the hidden-source test Never properly tested; the capacitor-on-magnet "generator" Not testable as stated. Claim 5 gains: ordinary flux-switching circuit Plausible, untested as stated (EPO: structure old); reported COP > 1 stays Never properly tested with sub-findings (Tektronix rule against the cited cross-check; excess only on nonlinear loads in builder reports; no calorimetry); US continuation rejected on theory with an unanswered demonstration request. No headline verdict changed.

What would settle it

• Claim 3 (already settled on paper): charge a capacitor from a battery with an ammeter in series and a scope across the capacitor; integrate V x I over the cycle. The battery's integrated V x I will be twice the energy delivered to the load (stored (1/2)CV^2). A coulomb counter on the battery gives the same by Q.
• Claim 2: crossed-polariser test of any radio or optical source (Malus's law); a longitudinal wave would pass. Already done many times.
• Claim 5: (a) log full waveforms (simultaneous v(t) and i(t), bandwidth well above 87.5 kHz, 10x harmonics) at both the DC supply terminals (including oscillator and flip-flop supply) and each output coil, compute the integral of v(t)i(t) over many cycles, and put the output load in a calorimeter; (b) the patent's own self-actuating configuration: external supply disconnected after start-up, resistive load, run for hours. True COP > 1 keeps running; an artefact stops. A neutral lab, with raw data published and the unit built to the patent's dimensions, would settle it. Cost: a few thousand dollars of instruments.
• Claim 4 (Sweet VTA): would need Sweet's device and the magnet-conditioning process; not available.
• Claim 5, added 2026-10-05 (Investigator): (c) true power by Cauchy-Schwarz-safe method: record v(t) and i(t) at both supply and load, integrate the point-by-point product, and put a linear (resistive) load and a nonlinear (MOV) load in turn into a calorimeter; an artefact shows excess only in the scope-math number, not in the heat. (d) Self-powering condition: with open-loop COP c and fed-back fraction f, running needs f c >= 1; the supply-removed run is the one test independent of instrument method (it runs or it stops), and it is exactly the granted EP claim 1. (e) Magnet flux before and after hours of running, if the magnet is the claimed fuel.
• Claim 4 (MEG as test article): the same supply-removed run, for a vacuum source; and any investor-demonstration measurement report.

Open questions

• [Closed by QE revision; see below] What do Figs. 6-8 actually show? Read as page images.
• (Added 2026-10-05, Investigator) Open: why the COP plot and "all COP > 1" sentences are absent from the European text (EPO 2007/2008 papers, 2013 minutes, unread); how the applicants answered the EPO on industrial applicability; what the rest of the 2004 remarks and Bearden's declaration contain; why the applicants did not answer the 2004 demonstration request; the Colmac security agreement and any release; whether investor representatives ever measured a MEG, and where such reports are; Naudin's later builds (COP 1.3, 7.3; v3.0/v4.0) and whether he ever did calorimetry or closed the loop; which flux-switching prior-art patents (Starr to Flynn) claim COP > 1; the primary source for the university non-circumvention account.
• Does the 14 W input include the oscillator/flip-flop/driver supply?
• What produces 4000 V from a 40:450 turns ratio at 100 V (transient spikes, resonance)? If resonance or ringing, the power factor is the question.
• Is there any neutral lab report on a MEG built to the patent? None found in the sources checked.
• Does the Treatise itself (1873) contain any gravity-EM suggestion? Not checked here beyond the 1865 paper.
• Are there legitimate low-loss, step-charged collector schemes of practical value (adiabatic charging)? Real but unrelated to overunity.
• AB effect: potentials are physical in QM; does the scalar-potential version Bearden needs ("distance independent AB with Whittaker beams") predict anything measurable? He states it, untested [read per contrarian].

QE revision (after qe.md, verdict MINOR)

Inputs added: deep/Bearden-Final-secret-of-free-energy-1993-d9aa85.md (13 May 1994 paper, pp.42-49, Opus, text plus page images) and deep/US6362718-383b35.md (patent experiment, claims, figure sheets, Sonnet). The QE's second deep read (15 Feb 1994 update, pp.28-41) was completed afterwards and is reported in Round 2 below, including the verified alpha definition.

New evidence

1. Patent Figs. 6-8 closed as an open question [read: page image p.3-4, deep read s.4-6]. Measured points run 10-75 V; Fig. 8 COP rises from about 1.7 (10 V) to about 2.6 (75 V, about 22 W out, about 8.5 W in, approximate reading). The 100 V figures (14 W in, 48 W out, 3.44) are a polynomial-fit extrapolation beyond the data. Fig. 7 is most consistent with output per coil (inference; the axis is unlabelled). Fig. 6E/6F show output peaks of about 1500 V at 75 V in, against about 844 V from the 11.25 turns ratio, and near-sinusoidal ringing of about 23 us period, which points to resonance and reactive power. No power factor, load value, bandwidth or uncertainty is given. This sharpens, and does not change, the "never properly tested" verdict.
2. Input power: 14 W (and the measured inputs) is V x I into the two power FETs only (Figs. 6C-6D). The oscillator, flip-flop, one-shot and gate-driver supply is not stated to be included; no quiescent figure is given [read: deep read s.1]. The open question "does 14 W include the logic supply?" is answered as: not stated, probably not. At 10 V (under 1 W input) this omission would matter most.
3. The patent names the energy source (the patentees' own characterization) [read: p.11-12]: "the system receives static energy from the magnetic flux of the permanent magnet", it is "an open dissipative system ... from the magnetic flux stored within the permanent magnet", "Continued operation ... causes demagnetization of the permanent magnet", and it is compared to reactor fuel rods. The self-running form is argued ("it is apparent") from COP; no self-running test is reported. Effect on the argument: my earlier text called the magnet-drain bound "supporting". It is now direct. The patentees name the magnet as the fuel, so the test is whether the magnet can supply the claimed excess. Stored energy about 11.6 J (at an NdFeB (BH)max of about 300 kJ/m^3 [memory, medium; web check not completed]) to 22 J (B^2/2mu0 at 1.2 T) in 38.7 cm^3 [recomputed from patent dimensions]; claimed excess 82 W (2 x 48 - 14) would exhaust it in 0.14-0.27 s; even the measured 75 V excess (about 14 W) would use 11-22 J in about 1-2 s. The verdict does not rest on this figure alone: the patent itself names the magnet as the fuel [read: p.11-12], and an order-of-magnitude change in (BH)max (stored energy anywhere from about 1 J to about 220 J) would still give seconds at most, not a long life, at 82 W excess (220 J / 82 W = 2.7 s). If a web check is run later, replace the [memory, medium; web check not completed] tag with [web] and the URL. Caveats kept: the web check FAILED (search budget exhausted). [Updated 2026-10-05: the check was later run (web check 11, "Investigator findings (2026-10-05)"); the figure is now [web], 190-450 kJ/m^3, so the store is 7-22 J; it lasts 0.09-0.27 s at 82 W excess and 7.35/13.5 = 0.54 s to 22.2/13.5 = 1.6 s at the measured 75 V excess of about 13.5 W (22 - 8.5) [recomputed].] Also, the magnet energy is not a rigorously consumable store when flux is merely switched between paths, so this is a bound on the patent's own stated mechanism, not a proof about every possible mechanism. The closed-cycle argument remains the decisive one; the patent's own wording now makes it applicable without any added assumption. Claim 5 verdict unchanged, the evidence is stronger.
4. Pulse-width inconsistency [read: p.11, p.14-15]: the description gives 11.45 microseconds (1/87.5 kHz is 11.43 us, consistent with Fig. 6A/6B), but claims 10 and 26 say "approximately 11.5 milliseconds", a factor of about 1000 (about 87 Hz). Almost certainly a unit slip, but it is in the granted claims. Other drafting errors found: "output coils 26, 28" for 29, 30; "1.575 mm" for inches; claims 2-3 "said of said second input coil"; claims 9/25 without antecedent. No claim recites a COP, demagnetisation or a measured self-run.
5. The 13 May 1994 paper (titled "Practical Overunity Electrical Devices") [read: pp.42-49]: - Contains no measurements at all: no voltages, currents, powers or waveforms. Only capacitor prices ($400-2,000 per about 1 uF). - Fogal charge-blocking semiconductor (CBS): "Fogal's marvelous semiconductor blocks passage of electrons into its output terminal, but passes displacement current dØ/dt into it" [read]. The paper gives nothing on structure, ratings or source. The device does have a public record outside the paper (added 2026-10-05, Check 9): William J. Fogal's own patents US 5,196,809 (filed 1991) and US 5,430,413 (granted 1995), "High gain, low distortion, faster switching transistor", describe a transistor whose emitter is coupled to ground through a 0.2-100 uF electrolytic capacitor and claim gain "at least twice" that of the bare transistor, with no claim of output power above input or of energy from the vacuum [web]. So there are two characterizations: the inventor Fogal's (in his patents: a higher-gain transistor, ordinary amplification from a supply) and Bearden's (a dm/dt blocker enabling overunity). Bearden's is his own interpretation of another inventor's device; Fogal's patents do not support or contradict the overunity reading, they simply do not make it. Fogal's patent text was read only through a fetched summary. Fig. 1 (open loop): CBS into a step-up transformer primary. Claimed "very high variable power gain" with no number. Bearden concedes an ammeter at the CBS output reads the dPhi/dt as ordinary current and that V x I then shows energy conserved between primary and secondary, and that adding load draws more primary current (the standard reflected-load behaviour). The gain exists only in an "actual dissipative power" that, by his own p.44 admission, no instrument he knows can separate. Also: the CBS "uses some of the power to power itself" and can "fail catastrophically", which is incompatible with zero primary dissipation. - Fig. 2 (closed loop): a switch removes the DC supply and a secondary winding feeds an AC-to-DC converter back to the CBS, giving a self-powered "negative resistor", "driven by ... the vacuum". No run is reported. The loop contradicts Bearden's own p.46 isolation rule, since the converter then is the primary source. Energy budget: with the supply removed, load plus losses must be paid for indefinitely by an unnamed mechanism; the second law is simply declared inapplicable. - "Blocks electrons, passes displacement current" is the textbook description of a capacitor, so as described the CBS is indistinguishable from a series capacitor. Fogal's own patent structure (an electrolytic capacitor at the emitter) is consistent with that reading [web, Check 9]. - Absence of evidence (added 2026-10-05): Bearden presents the CBS in 1994 as making overunity "readily obtained"; after that I found no measurement of a CBS power gain, no Fig. 1 or Fig. 2 build, and no later Fogal patent claiming gain above input. Competing explanations: (a) the gain was a meter artefact of the kind Bearden himself describes (an ammeter that "reads dPhi/dt as normal current"), so careful builds found nothing to report; (b) commercial or trade secrecy on Fogal's side; (c) builds were made but are unrecorded in the sources searched (the rexresearch Fogal compilation was not opened). What would decide: the rexresearch compilation and contemporary press reports (a search summary cites an interview in "SN"), and any measured bench data on the Fogal transistor. - Dimensional checks (sympy, deep read s.2, s.10): dPhi/dt is V/s, not amperes; charge = Phi x m would be V.kg, not coulombs; (dPhi/dt)(dm/dt) is not amperes. Textbook displacement current is eps0 dPhi_E/dt with Phi_E the electric flux (V.m); Bearden's symbol appears to confuse flux with potential (inference). Product-rule error re-confirmed from the page image (d/dt[(Phi)(m)] = (dPhi/dt)(dm/dt)). - Energy formula printed as K = 1/2 (CV)^2 (page image confirms; units C^2, not J). Likely a typo, but it is the one formula offered as "proof". - Self-contradiction: p.43 says an ordinary capacitor dissipates as much as it stores; p.45 says "with ordinary capacitors ... the efficiency can approach 1.0 rather than 0.50". The latter is the standard step-charging result (loss falls as 1/N; example.py PART B), not overunity; the cited Am. J. Phys. papers are about ordinary quasi-static charging [memory, medium; web check not done]. - SQUID argument: measured magnetic fields between charging plates confirm that displacement current equals the lead conduction current, which supports Maxwell and contradicts the idea of a separate, extra current. No citation is given; the CBS device itself is offered as the "best proof", which is circular. The measurement history [memory, medium-low] was not web-checked.
6. Minor: foreword 250 V/m and p.18 "200-300 V/m" both occur.
7. The 15 Feb 1994 update (pp.28-41) is now deep-read [read: deep read, pp.31-32 and page image p.31]. Verified definition (p.31-32): "operational efficiency Alpha" is "the average power expended in the load to power it, divided by the average power expended inside the source to dissipate its dipolarity", i.e. alpha = /

   . This differs from true efficiency eta = P_load / (P_load + P_internal): alpha = eta/(1 - eta), eta = alpha/(1 + alpha) [recomputed]. Worked check: 12 V battery, 0.1 ohm internal resistance, 10 ohm load gives alpha = 100 but eta = 0.990 [recomputed]. So alpha > 1 is ordinary, and Bearden's statement that conventional operational efficiency "is always underunity" [read: p.32] is wrong for his own definition. His Fig. 2 slide (p.31) uses a third quantity, Eff = P_L/P_T = I_L/I_T, which is true efficiency. Measurement for step charging: none. The collector is an ordinary capacitor charged in "several hundred" steps [read: p.30]; support is only the Am. J. Phys. teaching papers, and p.35 admits a competent researcher can build it "to prove it experimentally". Physics: efficiency N/(N+1), 99.7% at N = 300, never above 1 [recomputed].

8. Output above total input (new, listed per instruction): the update states it as a permissibility argument, not a result [read: pp.28, 30, 32, 37]. p.28: the source "can operate at an efficiency greater than unity without violation of the laws of physics", shortfall supplied by the vacuum; p.30: "one can extract all the free EM energy one wishes, from any electrical power source"; p.32: with alpha > 1 the device "can conceivably run itself". No input or output number is reported anywhere in pp.28-41.

Verdict changes and additions

No headline verdict for claims 1-5 changed. What changed: one new item (3b), one relabel (Maxwell-gravity), stronger evidence for claim 5, and (Round 2) one further item (3c) from the 15 Feb 1994 update.

Round 2 (deep reads of the 15 Feb 1994 update, pp.28-41, and the 13 May 1994 paper, pp.42-49)

Input: deep/Bearden-Final-secret-of-free-energy-1993-d9aa85.md, now holding both reads (no web checks were run in it; its [memory] items are not load-bearing for any verdict below).

What the new read found [read unless tagged]: - Alpha definition and its gap from true efficiency: see New evidence item 7. - The update contains no measurement of any kind: no voltages, currents, powers, waveforms or test report. Its numbers are "several hundred" steps (p.30), a tentative 98% Al / 2% Fe collector alloy (p.29), 25,000 MOSFET "components" (p.35, unverified), and the "one half" loss assertion (p.31). Figs. 1a/1b (p.29) are circuit drawings without component values. - Patent application mentioned (p.28) has no number or date; nothing links it to US 6,362,718. - "Suicide circuit" argument (p.31-33): the "one half" is asserted, not derived. It matches only the matched-load special case (R = r, eta = 50%); a 12 V battery into 10 ohm with 0.1 ohm internal resistance delivers 99% [recomputed]. His Fig. 2 slide blames parallel parasitic branches, his text blames half the energy restoring the source: two different mechanisms. Also, verbatim: "this 'suicide circuit' practice is guaranteed to keep the power meter on your home or business, and to keep the meter on the gas pump ... Perhaps one may be forgiven for suspecting that, at the deepest levels of financial control, this may be the real purpose in seeing that the existing interpretation of classical EM stays 'as is.'" [read: pp.32-33]. Bearden states this as a suspicion; no document, person or event is offered in support, and I found none. (Reworded 2026-10-05: previously labelled "conspiracy-style", which dismissed it by label.) - Open-system thermodynamics (pp.33-34, notes 10-12): the second law and energy conservation apply to open systems; the Lindsay-Margenau step "lower entropy, therefore higher energy" (note 10) does not follow (same energy, lower entropy is possible). The "Maxwell's Demon" appeal (note 7) is not supported by Landauer/Bennett analyses [memory, high]. - Step charging: eta = N/(N+1), see item 7. Bearden's own claim of no electron current through the source contradicts the Ampere-Maxwell continuity he cites on p.36 [recomputed reasoning]. - Charge-quantisation claim (p.37, ref 6): stated as a consequence of his own premise; no experiment. - Heaviside quote, Occam's-razor caveat for experimenters (p.35) and the admission that his own device is unproven (p.35) are recorded.

Verdict changes: the headline verdicts for claims 1-5 do not change. The 13 May paper's items (3b) are unchanged from the earlier QE revision. One new item, 3c, is added to the Corrected claim list and the verdict table above, because the read found statements of output above total input (the vacuum "makes up the difference"). None comes with a measurement. The deep read's suggested labels are kept: permissibility claim Not testable as stated, sustained self-running form Refuted by calculation (same energy-budget argument as claims 3 and 5).

Web checks in this revision

Attempted two (NdFeB (BH)max; Fundaun et al. 1992 content). Both FAILED: the web search tool reported its per-session budget exhausted. Both facts remain [memory, medium] (NdFeB later retagged [web], 2026-10-05, web check 11 under "Investigator findings"). Neither alters a verdict: the closed-cycle energy argument does not use them.

Updated open questions

• Closed: what Figs. 6-8 show; whether 14 W includes the logic supply (not stated, probably excluded).
• Still open: what produces the roughly 1500-4000 V output (resonance? power factor?); any neutral replication; whether any CBS device exists and what it is (partly answered 2026-10-05: Fogal's patented transistor exists; no overunity measurement of it found).
• Added 2026-10-05 (records that would decide the absence-of-evidence questions): Magnetic Energy Ltd corporate record and fate; identity of the university and DoD programme in Bearden's Dec 2000 account and the primary source for "refused to sign a noncircumvention agreement"; Naudin's v3.0 and COP 7.3 build pages; any surviving Sweet VTA unit or notes; the rexresearch Fogal compilation.
• Decisive test for the CBS device, if one is ever built: run Fig. 2 with the DC supply disconnected into a resistive load for hours, as for the MEG.

Evidence-rule revision (2026-10-05)

Audit of every claim against the four SOURCE HONESTY rules (no suppression of evidence, interested parties, absence of evidence, keep pertinent information). No research/ or sites.md existed for this project at that time (the Investigator's run of 2026-10-05 followed; see the last section); about 10 web calls were made (Checks 6-10). Changes:

1. Source gap note: "mostly advocate or hobbyist pages" (a label) replaced by naming each web source and its stake (Bearden as co-inventor and company spokesman; Naudin as independent private builder; archive/news sites).
2. Corrections table, Naudin row: opened the MEG v2.0 page. Kept COP 1.75. Withdrew the earlier search-summary details (10 ohm resistor, THS720P scope, load heating not confirming the power), which are not on the page. Added Bearden's report of Naudin's later COP 1.3 and 7.3 builds, attributed to Bearden.
3. Web Check 3: dropped "lists only proponent claims"; cites the ZPEnergy 2021 article by author and date and states what it records.
4. Web Check 4 (Sweet VTA): removed "source is advocate-side, so ... reliable", which weighed the account by its label. The account (330 uW in, over 500 W out; withheld; fear for his life) is now quoted, attributed and described as documented in Sweet's and Bearden's accounts with no second source found yet.
5. New Web Checks 6-10: Naudin's own words on COP, coil failure above 3 kV, and his unsuccessful attempt to close the loop (pertinent to the self-running claim; earlier dropped or missed); Bearden's 2 Dec 2000 e-mail (COP 17-20, COP>100 to destruction, DoD-auspices replication arranged, names withheld); search-summary account that the university test stopped over a non-circumvention agreement; Magnetic Energy Ltd status page (Apr 2001); Fogal's own patents; Bearden as co-author of the Evans et al. paper ([memory, medium], not confirmed).
6. Claim 3: added Bearden's stated free release of the collector scheme [read: p.4] and an absence-of-evidence analysis of why no build is recorded (null results unpublished; under-specified design; unrecorded builds), with the records that would decide.
7. Claim 4 (Sweet VTA): "[web, advocate sources]" replaced by the attributed account, plus an absence-of-evidence analysis (artefact; trade secrecy; fear of harm as Bearden states; process lost with the inventor) and what would decide.
8. Claim 5: added an interested-parties paragraph. The patent's mechanism and COP statements are the patentees' characterization; Bearden's later figures are a co-inventor's; the USPTO grant is not a performance test; the Evans et al. paper is not third-party support if Bearden co-authored it.
9. Claim 5, "Tested" section rewritten: "COP 5, 17-20, 100 ... proponent statements, unverified" and "not neutral" replaced by each account quoted with source and stake (patentees; Magnetic Energy Ltd per Bearden; Naudin; the university/DoD arrangement and its reported end; Evans et al.). The earlier wording "failed to occur" is replaced by the only account of the outcome found: stopped over a non-circumvention agreement before any measurement.
10. Claim 5: added an absence-of-evidence analysis of the silence after 2001 (artefact; commercial failure; trade secrecy; absorption under a new name; suppression), with the evidence for and against each and the records that would decide (company registration and dissolution, later patent filings, the identity of the university and DoD programme, Naudin v3.0).
11. Claim 5, empirical part: "Not neutral" replaced by "measured by the patentees themselves"; the artefact explanation is marked as the arbiter's view ("in my view the likely route"). The outdated "Figs. 6-8 unread by anyone" is pointed to the QE revision.
12. QE revision item 3: "The patent concedes the energy source" reworded to "The patent names the energy source (the patentees' own characterization)".
13. QE revision item 5, Fogal CBS: added Fogal's own patents and his characterization (a higher-gain transistor with an emitter electrolytic capacitor; no claim of gain above input or of vacuum energy) beside Bearden's overunity characterization, plus an absence-of-evidence analysis of the CBS record after 1994.
14. Round 2: "unsupported conspiracy-style assertion" replaced by the verbatim "suicide circuit ... power meter on your home" passage, described as Bearden's stated suspicion with no supporting evidence offered or found.
15. Updated open questions: added the specific records that would settle the MEG, Sweet and Fogal silences.

Verdicts: none changed. Reasons: (a) the newly quoted Bearden figures (COP 17-20, >100) come with no waveforms or input/output values, and the only outside builder (Naudin) reports 1.75 with his own instruments, so "Reported COP > 1: Never properly tested" still describes the record; (b) the reported end of the DoD-auspices test (stopped before measurement) is neither a failed test nor a mainstream dismissal, so it does not move the verdict toward "Tested, not supported" or "Dismissed without testing"; (c) Naudin's unsuccessful loop closure agrees with "Sustained self-powered net output: Refuted by calculation" but that verdict rests on the closed-cycle energy budget, not on his attempt; (d) Fogal's patents describe ordinary amplification and supply no measurement of gain above input, so the CBS items stand; (e) the Sweet VTA remains "Never properly tested".

Investigator findings (2026-10-05)

Inputs: sites.md ("Leads for review"), the six research/ notes dated 2026-10-05 (Google Patents, EPO Register, archive.org USPTO wrapper, JLN Labs, cheniere mirror, company records), and the "Investigator findings (2026-10-05)" section of physicist.md, mathematician.md and contrarian.md. research/files/ is empty (nothing was downloaded), so every [web] item below is the Investigator's reading of a page, carried by the three reviewers and by me at second hand. I re-opened none of those pages. Source gap that remains, stated by all three reviewers and by me: the Investigator read only the first 50,000 of 85,447 characters of the applicants' 2004 remarks, did not read Bearden's declaration, the abandonment notice, the EPO annex pp.4-5, the 2007/2008/2009 replies or the 2013 minutes. What Bearden told the USPTO and the EPO is therefore known from partial reads.

Web checks in this pass (budget 8, used 2): (11) NdFeB energy product: sintered Nd2Fe14B grades span about 190-450 kJ/m^3 (N52 about 414 kJ/m^3), bonded magnets about 70-130 kJ/m^3 [web: https://en.wikipedia.org/wiki/Maximum_energy_product and https://www.samaterials.com/content/maximum-energy-product-in-magnetic-materials.html, search summaries; pages not opened]. This upgrades the [memory, medium] "about 300 kJ/m^3" to [web]: 38.7 cm^3 gives about 7 J (190) to 17 J (450) [recomputed: 3.871e-5 x 190e3 = 7.35 J; x 450e3 = 17.4 J]; with the 22 J upper value from B^2/2mu0 the range becomes 7-22 J (0.09-0.27 s at 82 W), which supersedes the 11.6-22 J and 0.14-0.27 s written earlier in this file (those lines now carry a pointer here). The patent does not say which magnet material was used (unknown here); a ferrite magnet (energy product far lower [memory, medium]) would give a smaller store, so the bound stands. (12) Attempted to re-open Naudin's ths720p.htm page (the Tektronix note) with the fetch tool: connection refused (the same refusal the earlier arbiter run met). The Tektronix text therefore remains the Investigator's reading in Chrome [web: http://jnaudin.free.fr/meg/ths720p.htm, not re-checked by me].

Where the reviewers agree on the new findings

• Every finding is carried; none is dropped. All three weigh the same eight items: the 2004 USPTO file wrapper and Bearden's mechanism; the examiner's demonstration request; Naudin v2.1, Flickinger and the Tektronix statement; the EPO grant and the removal of the COP plot from the European text; the financing and assignment trail; the 15 Aug 2002 letter; the Poynting remark; the old flux-switching prior art.
• "dA/dt = -E" is correct and is ordinary induction (physicist F1, mathematician F4.2, contrarian F2). The mathematician's step is the sharpest: where B = 0, curl A = 0, so locally A = grad(chi) and E is a pure gradient with zero EMF round any loop not enclosing changing flux; the coil EMF is the Faraday value from the enclosed flux. No extra energy appears. I verified the logic [my calculation, from the same identity]. The Aharonov-Bohm (AB) effect is a quantum-phase effect for charged particles and adds no classical energy channel.
• The Tektronix rule is mathematically right: true average power is the mean of v(t) i(t); Vrms x Irms is an upper bound (Cauchy-Schwarz) equal to power only for proportional waveforms or single-frequency sinusoids. Both worked examples agree: physicist 4000 V / 12 mA 90 degrees out of phase gives 24 W apparent, 0 W true; mathematician 10 sin(wt) V and 2 sin(3wt) A gives 10 W apparent, 0 W true [recomputed: 7.07 x 1.41 = 10.0].
• The EPO found the MEG structure anticipated by US 4,006,401 (1977); the only novelty is the self-powering feedback. All three note this is also the one feature with no demonstration.
• Claim 1 and 2 and 3 verdicts: no change from any reviewer.

Where they differ, and my weighing

1. Does the new evidence move claim 5 to "Tested, not supported"? Physicist: split the claim and place the strong form (sustained COP > 1 with a linear load) at "Tested, not supported at an informal level" (one independent builder, uncontrolled). Mathematician and contrarian: stay at Never properly tested, with sub-findings. I side with the mathematician and contrarian. The scale defines "Tested, not supported" as controlled experiments that did not support the claim. What exists is: Naudin's own page (a single builder who has an interest in the MEG succeeding and who states "measurement artifacts remain possible", "calorimetric ... not yet done"), and Flickinger's undated "MEG Notes" (a single source on Naudin's page, linear loads show nothing). Nobody reports calorimetry, a true v(t) i(t) integral, or a closed-loop run, either way. I state Flickinger's linear-load result as evidence against excess, in full, and I do not reduce it to a label; but it is one uncontrolled report. The physicist agrees it is not a neutral controlled refutation; our difference is only the label, and the scale's own wording settles it.
2. How strong is the artefact reading now? Physicist: the patent's method is one the instrument maker calls invalid; the excess tracks nonlinear loads, which is what an artefact predicts. Contrarian: "better supported inference", but with a caution that the patent's primary method (real power from sampled V and I, averaged over the period) is valid if done on sampled data, and the scope-math "RMS power" is only its cross-check; agreement between a valid and an invalid method is not corroboration but does not show the first wrong. The contrarian is right and the physicist overstates slightly: the patent says its real-power calculation agrees with RMS power measured on a Textronic THS730 [read: patent, per the earlier deep read]. Two further limits, which I add: (a) the Tektronix reply addresses scope math waveforms in general, relayed through Flickinger (a builder), and whether it describes the exact THS730 procedure the patentees used is an inference (Flickinger's own comment ties it to "Bearden's and JLN's THS720P scope readings", which is his reading); (b) the patent gives no raw data to show whether the first method was sampled or averaged. So: the instrument-maker evidence removes the cross-check as support and shows how a large over-reading is arithmetically possible, but it does not show that the primary calculation was wrong. The artefact reading is now better supported inference (my own view as before), not established.
3. Naudin's numbers. The contrarian caught an accounting point that I verify [recomputed]: on v2.1 the total DC input is 3.25 W (control board 1.75 W plus 1.5 W attributed to the lamp), so 6.76 W / 3.25 W = 2.08, not the 4.5 (6.76 / 1.5) that results if the board is outside the boundary. This is the same drive-supply question as the patent's 14 W (FET supply only; logic supply not stated). Three different numbers read "1.75" in this record and must not be confused: the v2.0 page "measured COP is 1.75" (7 Nov 2000) [web, Check 6], the v2.1 control-board no-load draw of 1.75 W, and Flickinger's "COP 1.75 to 5" for MOV loads. They are different quantities from different builds; the Investigator did not find "COP 1.75" on the v2.1 pages. Bearden's reports of Naudin builds at COP 1.3 and 7.3 (2 Dec 2000) are not linked to any page found, and I do not infer that 7.3 was a nonlinear-load figure. Naudin's radiation explanation: mathematician (area 0.03 m^2, radiation resistance about 2e-16 ohm) and contrarian (100 turns, 0.01 m^2, about 2.5e-13 ohm) used different assumed geometries; I recomputed both from R = 31,171 (NA/lambda^2)^2 with lambda = 18.75 km and get 2.3e-16 and 2.5e-13 ohm [recomputed, formula from the reviewers, [memory, high]]. Either way radiated power is negligible; and 1250 V/m as a plane wave would be E^2/377 = 4.1 kW/m^2 [recomputed], so it is a reactive near field. Naudin's own radiation suggestion does not account for the gap. His measurement method for the 2.8 mT and 1250 V/m figures is not read.
4. Claim 4: does the MEG make the hidden-source idea testable? Contrarian: yes, relabel the MEG instantiation Never properly tested (was Not testable as stated). Physicist: no, still Not testable as stated (the file wrapper gives a mechanism, not a measurable prediction beyond COP > 1). Mathematician: unchanged. My weighing: both are right about different sentences. The mechanism (a magnet as broken-symmetry dipole absorbing vacuum energy; AB-localised B; E = -dA/dt pulses) offers no formula for how much power the vacuum supplies; that stays Not testable as stated, and the induction step is shown energetically empty. But Bearden himself names the MEG as the device and the vacuum as the source (2004), and the granted EP claim 1 requires the self-powering loop; that device-level prediction (it runs with the supply removed) is testable and no run is on record. So the MEG-specific hidden-source claim is Never properly tested; the general principle and the mechanism are Not testable as stated. One tension must be stated, not hidden: the 2000 patent text names the magnet's flux as the fuel and concedes demagnetisation [read], for which the energy-budget refutation (finite store, 7-22 J) applies; the 2004 remarks name the vacuum as the source, which that calculation does not address, and the examiner had asked the applicants to show "how the magnetic flux stored within the permanent magnet is sufficient". The two descriptions of the source differ, both by the patentees. Neither has a measurement behind it.
5. The Poynting remark (Bearden 2002 P.S.: a charged capacitor or electret laid across a magnet with E perpendicular to H is a "proven free energy generator"). Physicist: Not testable as a generator, no device given. Mathematician: mathematically empty. Contrarian: Never properly tested as a measurement, the textbook calculation counts against. For static fields with no free current, div(E x H) = H . curl E - E . curl H = 0 [recomputed from the vector identity], so the flux circulates and delivers nothing; this is the textbook hidden-momentum / Feynman-disk case [memory, high; Griffiths Ch. 8 per reviewers]. "Proven" has no data behind it in the file. My verdict: Not testable as stated (no load, no device, no measurement named); the static Poynting argument supplies no net power. This is a specific named configuration, so I list it in claim 4 and keep it in the record.
6. EPO text trimming. All three flag, and none infers intent: the granted EP description keeps the extrapolated "3.44" and the demagnetisation paragraphs but lacks the COP-versus-voltage plot and the sentences "output power was substantially higher than the input power", the THS730 agreement and "all ... significantly greater than 1 ... self-actuating form" [web: Google Patents text of both documents compared, per the Investigator; description pages and sheets replaced 15.06.2005, and Art. 84 clarity objections were raised in 2006]. Competing readings: (a) trimming of unsupported performance statements under examiner pressure; (b) routine redrafting or conforming of drawings; (c) a choice to carry less checkable evidence. The file (2007/2008 replies, 2013 minutes) would decide. Effect, which I judge the same as the mathematician's: the retained 3.44 now stands without the data it was extrapolated from, so it is less checkable. No verdict moves.
7. USPTO continuation: dismissed or invited to test? Physicist and contrarian both correct their earlier "no dismissal documented": the continuation 10/081,847 was rejected on 35 USC 101 (2004-01-23) on thermodynamic grounds without any test, so on that application the US record shows dismissal on theory; but the same examiner required a demonstration, the applicants asked to defer it (2004-05-24), he refused (2004-07-26), and the application was abandoned (2005-05-16, PlainSite "failure to respond"; no appeal in the item list). I adopt the contrarian's reading: this is not a blocking process that prevented a test; a test was invited and not supplied, and the file does not say why. The examiner's windmill argument conflates COP (output over operator-paid input) with efficiency (output over all input); the applicants' distinction was fair (the examiner is the agency's voice, with an agency's stake in the rejection; the applicants' remarks are legal advocacy for a patent). The core of his demand, a source large enough, is sound. The parent patent (application 09/656,313) was granted on 26 March 2002 [read: patent front page]; whether its examiner raised a utility objection or asked for a demonstration is unknown, because the parent file wrapper was not obtained (archive.org returned "Item cannot be found" [web: https://archive.org/details/gov.uspto.patents.application.09656313, per research/2026-10-05-archive-org-uspto.md]). No such demand is known; the parent file was not read. (Corrected 2026-10-05 after QE: an earlier wording said it was granted "without such a demand", which no source supports.)
8. Magnet energy bound across reviewers: physicist 3-4 J (assumes 10 cm^3, 22 W out, about 12 W in), contrarian 1.5 J (assumes 5 cm^3, 48 W - 14 W). These assume magnet volumes that are guesses. The arbiter's figure uses the patent's own 40 x 25.4 x 38.1 mm = 38.7 cm^3 [read] and the web-checked energy product range: about 7-17 J (to 22 J at B^2/2mu0), and the earlier QE item stands. The physicist's "about 12 W input" at 75 V differs from the QE reading of about 8.5 W in (approximate figure reading); I keep 8.5 W, labelled approximate, and note that no verdict depends on it. All routes give seconds at most for a claimed surplus of 10-80 W; the self-run verdict rests on the closed-cycle argument, not on this number.
9. Self-powering condition (mathematician): with open-loop COP c and a fed-back fraction f of output, sustained running needs f c >= 1. At f = 0.5 this needs c >= 2, which the patent's measured range (about 1.7 to 2.6) straddles. So the loop adds no new physics: it restates c > 1. A run with the supply removed is the decisive, instrument-independent test. I accept this and add it to "What would settle it".

What each new finding adds to the record (kept whatever its weight)

Further Investigator material kept for the record (added after QE, 2026-10-05)

These items are in research/ and sites.md and were not carried above. They are listed so the record keeps them; none changes a verdict. All [web], second hand (the Investigator's reading; I re-opened none of the pages).

1. Other arguments in Bearden's 2004 USPTO remarks (applicants' legal argument and co-inventor's declaration; first 50,000 of 85,447 characters read) [web: https://archive.org/stream/gov.uspto.patents.application.10081847/10081847-image_file_wrapper/10081847-2004-05-24-00003-REM_djvu.txt, per research/2026-10-05-archive-org-uspto.md]: - Negative energy from Dirac-sea holes, and a reference to Bedini (the remarks as summarised name Bedini; which Bedini work is not recorded). - "Time-energy": E = tc^2, energy said to be converted from time to spatial energy in sharp gradients. Units check, on the formula as summarised (the remarks themselves not re-read): t c^2 has units s x m^2/s^2 = m^2/s, not joules (J = kg m^2/s^2), so as written it is not an energy; it would need a factor in kg/s that the summary does not show [my calculation]. - The static field as a Prigogine dissipative structure (open system far from equilibrium). - Local versus global energy conservation in general relativity: Hilbert 1917, Logunov and Loskutov, Kondepudi and Prigogine. - Heat-pump analogy with figures: COP 3-4 for real heat pumps and "theoretical maximum ... about COP = 9.22", cited to Halliday and Resnick. Comment: a heat pump's COP above 1 is real because heat is drawn from a cold reservoir whose heat flow can be metered; the analogy requires the MEG's extra input to be named and metered in the same way, which is what the examiner asked for. - Already carried above: broken-symmetry dipole (Lee and Yang 1957), Whittaker flows, AB localisation and E = -dA/dt, COP 5 feedback example, Kron's negative resistor.
2. The examiner's other grounds (USPTO, 2004-01-23 non-final rejection, examiner Shawn Riley) [web: https://archive.org/stream/gov.uspto.patents.application.10081847/10081847-image_file_wrapper/10081847-2004-01-23-00002-CTNF_djvu.txt, per research note]: besides 35 USC 101, rejection under 35 USC 112 first and second paragraphs; obviousness-type double patenting over claims 1-29 of US 6,362,718; drawings objected to ("when the input power source (e.g., 38) is taken away, the remaining circuit becomes inoperable"; COP figures "inoperable as violating at least the first and second laws of thermodynamics"); title objection ("is seen as a transformer which relies on the concept of perpetual motion"); page 2 of the specification missing; Newman v. Quigg (1988) cited for the demonstration requirement. The double-patenting ground is the record behind the "redundant continuation" explanation for the 2005 abandonment (item 7 above and explanation (b) for the silence): if the continuation's claims were not patentably distinct from the parent's, dropping it cost the applicants little. It supports that explanation without excluding the others.
3. Bearden's 15 Aug 2002 letter, further content (co-inventor and company spokesman) [web: https://billstclair.com/www.cheniere.org/correspondence/081502.htm]: "every previous legitimate COP>1.0 system ... has been firmly defeated right at this same 'vertical cliff'" of scale-up funding (his characterization; no system or record is named in the letter as summarised); 10% of sales in poor countries reserved for wells, schools and clinics; planned 2.5 kW and 10 kW units; his book Energy from the Vacuum: Concepts and Principles (2002), "nearly 1,000 pages"; "Our first patent has issued, and two others are in various stages", which the Investigator matches to the US continuation (abandoned 2005) and the EP/PCT family (granted 2013).
4. Company-records leads (none opened) [web: https://www.google.com/search?q=%22Magnetic+Energy+Ltd%22+Huntsville+Alabama, per research/2026-10-05-company-records.md]: a Manta listing of Magnetic Energy Ltd, Huntsville, as "Noncommercial research organizations"; a ChamberofCommerce.com listing in Huntsville; a Facebook page "Magnetic Energy Ltd (@MagneticEngergyLtd)" describing MEG technology; a Bearden paper on pdfcoffee giving "Magnetic Energy Ltd. P.O. Box 1472 Huntsville, AL 35807" and citing Ritson, Nature 421 (2003); a ZPEnergy "Letter to JLN Laboratories" in which an AIAS writer offers to work "with your company and Magnetic Energy Ltd. to develop the MEG commercially". Bearing: the Manta classification and the P.O. Box are directory traces of the company, not registration records; the ZPEnergy letter is a lead on the AIAS group's commercial interest (relevant to the Evans et al. paper's stake). The Alabama register search found no entity named "Magnetic Energy" (row above).

Absence-of-evidence analysis, updated (claim 5)

The silence after 2001 is still an anomaly, now with a documented trace. New evidence by explanation: - (a) Measurement artefact. For: Tektronix rule and the arithmetic of apparent versus true power; excess only with nonlinear loads; apparent output above load heat; Naudin's failure to close the loop; the 4000 V vs 1,125 V turns-ratio gap and ringing; the removal of the measured plot from the European text (reason unknown). Against: no calorimetry or full-waveform analysis exists to show it; the patent's primary method may have been valid; the Investigator did not read the investor-measurement claims' substance. - (b) Commercial failure. For: a security agreement (2006) with a Canadian holding company, a late-paid 2010 fee, no product, the 2002 letter ("successful little experimental devices", funding sought), no Alabama entity found. Against the simple story that the project vanished in 2002: fees paid to 2013 and a 12-year European prosecution (two deemed-withdrawn episodes cured), which shows sustained spending. - (c) Trade secrecy. For: Bearden's stated policy of demonstrations only to prospective financiers; withheld names (2000); the demonstration request unanswered (2004). Against: the structure is published in two patent families, and what was kept private is the measurements. - (d) Absorption under another name. The Colmac pledge is the only lead; Colmac is otherwise unidentified; no evidence for or against. - (e) Suppression. Nothing found supports it; nothing excludes it. What was found is a theory-based rejection by one examiner, a granted European patent, and a demonstration invited and not given. The university non-circumvention account still rests on search summaries only. Records that would decide: the Colmac security agreement and any release (USPTO Assignment Center reel/frame 017759/0537, also 017706/0167 and 017746/0237); Colmac in Corporations Canada; the rest of the 2004 remarks and Bearden's declaration; the EPO 2006 annex pp.4-5, 2007 and 2009 replies, 2008 annex and 2013 minutes; the parent file 09/656,313; investor-demonstration measurement reports; Naudin's v3.0/v4.0 pages and the COP 1.3 and 7.3 builds.

Absence-of-evidence analysis for claims 3 and 4 (collector builds, Sweet VTA) and the Fogal device is unchanged: the Investigator searched none of them this run (rexresearch and amasci were not reached).

Verdict table after the Investigator findings

Headline verdicts for claims 1-5 do not change.

Changes to the other sections of this file (this pass)

See the edits marked "(2026-10-05, Investigator)" above and the list returned with this update.

QE remediation (Investigator update check, 2026-10-05)

• QE defect 1 (untagged "parent patent granted without such a demand"): item 7 above rewritten. Grant date 26 March 2002 [read: patent front page]; the parent file 09/656,313 was not obtained (archive.org "Item cannot be found"), so whether a demand was made is unknown.
• QE defect 2 (stale NdFeB tag and range): Web-check list item 5, the claim 5 energy-budget bullet, QE revision item 3 and the Round 1 web-check note now point to web check 11; recomputed 3.871e-5 m^3 x 190e3 / 450e3 / 573e3 J/m^3 = 7.35 / 17.4 / 22.2 J, lasting 0.090-0.27 s at 82 W and 0.54-1.6 s at about 13.5 W; web check 11 text now says the 7-22 J range supersedes 11.6-22 J.
• QE defect 3a (2004 remarks: Dirac-sea/Bedini negative energy, time-energy E = tc^2, Prigogine, Hilbert/Logunov, Halliday and Resnick COP 9.22): added under "Further Investigator material", with a units check on E = tc^2 (m^2/s, not J) and a note on the heat-pump analogy.
• QE defect 3b (examiner's other grounds: 112, double patenting over claims 1-29, perpetual-motion title objection, missing page 2, Newman v. Quigg): added, with the double-patenting ground tied to the "redundant continuation" explanation.
• QE defect 3c (15 Aug 2002 letter: "vertical cliff", 10% of sales for poor countries, 2002 book, "two others in various stages"): added.
• QE defect 3d (company-records leads: Manta, ChamberofCommerce, Facebook page, pdfcoffee P.O. Box 1472, ZPEnergy "Letter to JLN Laboratories"): added.
• Reviewer files (physicist.md, mathematician.md, contrarian.md): no edit. The QE defects name no error in them; the physicist's [memory, medium] NdFeB figure was a non-counted note, superseded here.