Bearden: Physicist review (Step 2)

Role: PhD physicist, mainstream view. Sources read directly: the 1993 paper bundle (text layer, pdftotext; foreword, collector section, relaxation-time section, footnotes 24-25, closing summary) and US 6,362,718 (experimental section, text layer). Tags: [read], [web] URL, [memory] confidence.

Source gap: none for the text. The patent's Figs. 6-8 data were not legible in the text layer, so the measured COP values behind Fig. 8 were not seen; only the prose and the extrapolated 100 V example were read [read]. The 1994 updates were only sampled by Step 1.

Corrections to Step 1

1. claims.md item 3 and facts.json say a wire taps "200-300 V/m". The foreword says "250 V/m in the earth/ionosphere potential" [read]. Minor. (Real fair-weather field is roughly 100-150 V/m at ground level [memory, medium].)
2. facts.json (patent entry) says COP 3.44 is "extrapolated, not measured": correct, the patent says "The data was extrapolated to describe operation at an input voltage of 100 volts" [read]. Added detail Step 1 missed: 48 W / 14 W = 3.43, not 3.44 (rounding), and 3.44 is per output coil; with two output coils the patent's own numbers give about 6.9 total. Also the claim "the text reports COP significantly greater than 1" is right, but the numeric values behind Fig. 8 were never read by anyone.
3. facts.json says the 1994 "patent application" is "probably not US 6,362,718". Confirmed plausible: the 1994 text describes electrical collector processes, while the 2002 patent contains no reference to vacuum, scalar potential, Whittaker or free energy (grep of the patent text, zero hits) [read].
4. The main paper is stated as a bundle of four documents; also the Floyd Sweet "vacuum triode" is mentioned only in the Additional Comments, with no data [read].
5. Step 1's relaxation-time note is incomplete: the 1.5e-19 s copper figure is a real number (eps0/sigma, see Claim 3 below), but Bearden misreads what it means. Not an error in Step 1 itself, a note.

Claim 1: Energy = potential = ordering in the virtual photon flux; a scalar potential contains an infinite harmonic series of bidirectional wave pairs (Whittaker/Ziolkowski)

a. Mathematics. Two parts. The Whittaker (1904) result is genuine: a solution of Laplace's equation can be written as a superposition of plane waves with complex direction cosines [memory, high]. This is a representation (like a Fourier series), not a statement that the potential physically contains travelling waves. Any smooth function can be decomposed in many ways; the decomposition does not create energy. The assertion "infinite energy in each of those infolded waves" [read, line ~189 of text] has no derivation. Worked check: the energy stored in a static field is U = (eps0/2) integral of E^2 dV, finite. For a 12 V battery's terminals that is a few picojoules to microjoules depending on geometry, not infinity. "Energy and potential are identical" is a units mismatch: potential is joules per coulomb, energy is joules. The text itself writes trapped energy as V_T x Coul_T, so it needs the charge to convert one to the other, which contradicts "identical". b. Experiments. Nothing in the vacuum-photon claim has a measurable prediction. Real vacuum-fluctuation effects (Casimir force, Lamb shift) are measured [memory, high] but are small and do not supply power. c. Tested? No test is described. No data in the paper [read]. d. If true: the energy of electrostatic fields would be unbounded, and the Whittaker decomposition would need physical meaning beyond the wave equation. Quantum electrodynamics would need a new "potential reservoir" with a way to be drawn on.

Verdict (split): (i) Whittaker decomposition as pure mathematics: Established, but it carries no energy consequence. (ii) "infinite energy in each wave pair / potential links to whole universe": Refuted by calculation (field energy is finite and computable). (iii) "energy = potential = virtual photon ordering" as a physical identification: Not testable as stated (and the units differ, so as written also Internally inconsistent).

Claim 2: No EM force fields in vacuum, only potentials and gradients; vacuum EM waves are longitudinal stress-potential waves; charge is not quantized but varies with ambient potential

a. Mathematics. Maxwell's equations in vacuum have div E = 0 and div B = 0. For a plane wave E ~ exp(i(k.r - wt)), div E = 0 forces k.E = 0, so E is perpendicular to the propagation direction: transverse. A longitudinal vacuum wave is excluded by the equations themselves. Also, "only potentials, no fields" fails: the fields are the gradients (E = -grad phi - dA/dt), so removing fields while keeping gradients is a relabelling. Gauge invariance: adding a constant to phi changes no physical result (except the Aharonov-Bohm phase for interference around loops). So "charge varies with the ambient potential" would make physics depend on an arbitrary additive constant, contradicting gauge invariance. b. Experiments. Polarisation: light passes through crossed polarisers and obeys Malus's law, which requires two transverse polarisation states [memory, high]. Hertz's 1880s radio experiments show polarised transverse waves [memory, high]. Charge: electrons and ions accelerated through potentials from volts to many millions of volts (accelerators) carry the same e. Neutral atoms are neutral to about 1e-21 e, which excludes a potential-dependent charge [web: https://arxiv.org/pdf/0711.4636, history of the 1e-21 bound; modern interferometry proposals reach about 1e-28 e]. c. Tested. Yes, repeatedly, by many groups over more than a century, not as tests of Bearden but as tests of the underlying facts. Not replicated in his favour. Bearden himself offers no experiment. d. If true: Maxwell's equations, radio and optics would be wrong at the level of everyday engineering; particle physics assumes a fixed e.

Verdict: Tested, not supported (transverse waves, constant e). The longitudinal-vacuum-wave part also Refuted by calculation (div E = 0 in vacuum). The Aharonov-Bohm appeal does not rescue it: the effect shows potentials matter for quantum phase, not that fields vanish.

Claim 3: Collector scheme: charge a long-relaxation-time collector from a source "current-free", then switch it to a load; the source is not depleted, so a battery could run a car or a grid

a. Mathematics. The key physical point is a basic energy budget. Worked example: collector capacitance C = 1 uF charged from a 12 V battery. Charge moved Q = CV = 12 uC. Energy the battery gives up = QV = 144 uJ. Energy stored in the capacitor = (1/2)CV^2 = 72 uJ. The other 72 uJ is dissipated in the charging path (independent of the resistance). Dumping the capacitor across a load returns at most 72 uJ. Net: out is at most half of what the battery lost. To get anything out you have to move charge Q through the battery, and the battery loses Q x V. Bearden's phrase "energy but no power" is false in that sense: charge moving under a potential difference is a current and costs energy at once; there is no state in which a source's potential is transferred while no charge flows from it (a current I=dq/dt exists whenever the collector's charge changes). The relaxation time misuse: eps0/sigma for copper = (8.85e-12 F/m)/(5.96e7 S/m) = 1.5e-19 s, matching his number [read, p. "Relaxation Time and Semiconductors"]. But this is the time for excess charge inside a conductor to decay to the surface, as a result of the conduction current itself; it is not a delay before current can start. For quartz, sigma about 1e-17 S/m and eps_r about 4.5 give about 4e6 s, roughly 50 days, the same order as his "10 days" [memory, medium: quartz conductivity varies by orders of magnitude]. A circuit's real delay is the RC time constant, and RC charging always loses (1/2)CV^2 regardless of R. Footnote 25 concedes "we are easily permitted to ... violate the local energy conservation law for a closed system" because the system is "open to a hidden source" [read]. That is an assumption stated, not shown. The closing summary says the invention "does not violate the conservation of energy law" [read] while the footnotes say it violates local conservation: the paper is internally inconsistent on this point. b. Experiments. Every measured battery, capacitor and switching circuit conserves energy to within measurement error. Switched-capacitor converters are standard engineering [memory, high] and are always lossy, never over unity. c. Tested? The paper gives no data. The 1994 update says a patent application was filed [read], but no test result is given. No independent replication known to me; none found in my web checks (see Claim 5). d. If true: conservation of energy, the definition of voltage as work per charge, and circuit theory would all need changing.

Verdict: Refuted by calculation (energy budget above). "Source is never depleted" also not supported by any measurement; the dismissed-without-test question does not arise since the claim is falsified on paper, though no one appears to have built the exact device and measured it either.

Claim 4: Closed classical EM cannot give overunity, so real devices must tap a hidden vacuum source; Maxwell's original quaternion theory, truncated by Heaviside, contained gravity-EM unification

a. Mathematics. The first sentence is correct and is the mainstream position (energy conservation follows from Maxwell's equations, Poynting's theorem). The conclusion that CEM must therefore be flawed is a choice, not a deduction: no measurement is cited that requires a hidden source. Quaternion statement: Maxwell's Treatise (1873) expressed some equations in quaternion notation but he did not calculate with it, using only scalar or vector parts, and the scalar part was zero for his field vectors [web: Bork, http://kirkmcd.princeton.edu/examples/bork_76.pdf and search summaries]. Heaviside and Gibbs's vector form is mathematically equivalent for classical EM. The gravity claim: in the 1865 paper section 82 Maxwell tried to cast gravity as a field theory, found the field energy would have to be negative, and wrote that he could not go further [web: https://royalsocietypublishing.org/doi/10.1098/rsta.2014.0473 and search summary]. So the original theory did not contain a gravity-EM unification. b. Experiments. No experiment requires or hints at a hidden electrical source. Precision tests of energy conservation in circuits and in particle physics find none missing. c. Tested? The "hidden source" is not stated precisely enough to test. The quaternion-Maxwell history is checkable in the texts and does not support him. (Quaternion/higher-symmetry reformulations such as O(3) electrodynamics are published by others [web: https://link.springer.com/article/10.1023/A:1012085429802] but are not accepted and are not tests.) d. If true: gauge theory and Poynting's theorem would have to include an unmeasured energy inflow.

Verdict (split): (i) "closed CEM cannot be overunity": Established. (ii) "real devices must tap a hidden vacuum source": Not testable as stated, and unmotivated. (iii) "Maxwell's quaternion theory contained gravity-EM unification": Refuted by the historical record (Maxwell's own 1865 gravity attempt failed); the vector-versus-quaternion notation is only a difference of form: Historically superseded but instructive.

Claim 5: The motionless permanent-magnet generator (MEG, US 6,362,718) gives output power above input power (COP above 1; extrapolated 3.44 per coil at 100 V), and can self-power after a start-up battery

a. Mathematics. Arithmetic in the patent: 100 V x 0.14 A = 14 W; 4000 V x 0.012 A = 48 W; 48/14 = 3.43 (patent says 3.44) [read]. Questions on internal consistency: (1) Multiplying average voltage by average current gives average power only if the waveforms are in phase and the load is resistive. With an inductive core pulsed at 87.5 kHz this is not guaranteed: out-of-phase (reactive) volt-amps look like power. The patent says the average of "measured voltage and current levels" agrees with oscilloscope RMS power [read], but with a high-impedance 4000 V output and a sampling instrument (a scope probe) that is exactly where large errors arise [memory, medium]. (2) The turns ratio: 40 input turns to 450 output turns is 11:1; a transformer would give about 1100 V from 100 V, not 4000 V. A voltage of 4000 V would need ringing or resonance, which stores and returns energy rather than creating it. (3) The "reduce turns to 12" example gives 106.7 V (= 4000 x 12/450, consistent) and 0.5 A, but 12 mA x 450/12 = 0.45 A; power 53 W instead of 48. Small, but the patent states output power is unchanged, which is not quite so. (4) The self-actuating claim is stated as possible ("can be built in a self-actuating form") with no run reported [read]. A permanent magnet has a stored magnetic energy but is not a fuel: cycling its flux in a lossy core with no energy input cannot yield continuing output. Note a point against Bearden's framing: the patent text never mentions the vacuum, scalar potential or free energy [read], so the physics of the 1993 theory is not in the patent; the mechanism it names (flux switching between two core paths) is ordinary transformer physics. b. Experiments. If time-averaged output really exceeded input in steady state, that contradicts conservation of energy, confirmed in every well-controlled calorimetric test of electrical devices [memory, high]. The more plausible explanation is a power-measurement artefact in a high-frequency pulsed circuit. c. Tested? By whom: the inventors' own measurements in the patent; Bearden's account of larger buildups with COP 5 and 10 [web: http://jnaudin.free.fr/meg/meg.htm and the Bearden MEG paper archive.org link in search]; Jean-Louis Naudin, a private hobbyist, reported a variant with COP 1.76 [web: http://jnaudin.free.fr/meg/meg.htm, via search summary]. An arranged independent university test under Department of Defense auspices was attempted and failed to take place [web: same search; Bearden's own account, not independently confirmed]. I found no calorimetric, accredited-laboratory or peer-reviewed replication confirming overunity. Not replicated in a controlled neutral test: so the claim has been neither confirmed nor properly tested. d. If true: conservation of energy, plus a new source for the magnetic energy; if instead it is a measurement artefact (my expectation), nothing in mainstream physics changes.

Verdict: Never properly tested. There is no neutral, controlled test (calorimetry or battery-energy-in versus load-energy-out over many cycles, including a self-running demonstration) on record, and the dismissal-without-testing question: the patent examiner granted it, so it was not dismissed as such, but no one has published an adverse controlled test either. Prior expectation, on conservation of energy, is that it will not survive such a test (strong form, COP > 1 sustained: likely Tested-not-supported once someone does it). Weak form (it is a working flux-switching transformer-like circuit): Plausible, untested as stated.

Summary table

Investigator findings (2026-10-05)

Scope: every "Leads for review" entry in sites.md and all six research/ notes dated 2026-10-05 (Google Patents, EPO Register, archive.org USPTO wrapper, JLN Labs, cheniere mirror, company records). research/files/ is empty (nothing was downloaded), so everything below rests on the Investigator's notes of web pages, tagged [web] with the URLs in those notes; I did not re-open the pages. None of these findings was weighed in the review above. Web budget: I made no new web calls; the Investigator's reads are the sources. Every finding is kept below whatever my judgement of it, and interested parties are labelled.

Source gap that remains: the Investigator read only part of the file wrappers (first 50,000 of 85,447 characters of the 2004 remarks; Bearden's declaration, the abandonment notice and the EPO 2007/2008/2013 papers unread). Statements about what Bearden told the USPTO are therefore from partial reads.

F1. The 2004 US file wrapper (continuation 10/081,847): Bearden's own bridge from the theory to the MEG [web: https://archive.org/details/gov.uspto.patents.application.10081847]

What it says (Investigator's note, with the party's stake: the remarks are an applicant's legal argument written to obtain a patent, and the declaration is by a co-inventor): the magnet is a "broken-symmetry dipole" absorbing virtual-photon energy from the vacuum (citing Lee and Yang 1957); its static field is a Prigogine "dissipative structure" and a Whittaker set of bidirectional flows; in the MEG, the Aharonov-Bohm (AB) effect keeps the magnetic field B inside the core and leaves a curl-free vector potential A outside; sharp input pulses perturb A and give "strong E-field energy pulses, by the well-known equation dA/dt = -E"; COP is compared to heat-pump COP (3-4, "theoretical maximum about 9.22"); a feedback example (COP 5 open loop, half returned, "COP = infinity"); Kron's negative resistor; local versus global energy conservation in general relativity; "time-energy".

This answers what my Claim 5 review recorded as missing ("the patent never mentions the vacuum, scalar potential or free energy"): the link exists, but in the file wrapper, not the patent. It bears on Claims 1, 2, 4 and 5. Weighing each piece, in plain terms:

• Faraday and the "dA/dt = -E" step (a, b). Mathematically the equation is correct: E = -grad(phi) - dA/dt [memory, high]. But it is ordinary induction. Outside a core, B is near zero yet A is not, and a changing core flux induces E = -dA/dt in a surrounding coil. That is what every transformer does. The energy delivered comes from whatever changes the flux (the driver, or the stored magnet energy), by Poynting's theorem; the equation contains no extra source term. AB (1959) shows A affects quantum phase of electrons going around a flux tube; it adds no energy channel [memory, high]. Test (b): transformers and flux-switching generators have been built since the 1950s and conserve energy (the EPO cites several, see F4).
• Broken symmetry and Lee-Yang. Lee and Yang 1957 proposed that the weak interaction violates parity [memory, high]; it is unrelated to magnets absorbing vacuum energy. Spontaneous symmetry breaking in a ferromagnet is real, but a broken-symmetry state is a lowest-energy state: nothing can be extracted from it without raising its energy [memory, high]. A heat pump (COP 3-4) is legitimate because it moves heat from a second reservoir; the claim needs an identified reservoir and an energy-in-versus-out count, and none is given. A single reservoir at one temperature cannot be drained for work (second law).
• Feedback example. "COP 5, half fed back, so 3 to the load, COP = infinity" is arithmetic that holds only if COP 5 is true at steady state; it adds no evidence for COP 5. Also, the examiner's windmill point (COP below 1 when all wind energy is counted) conflates COP with efficiency. The Investigator notes the applicants made that distinction fairly; I agree, but the distinction does not help unless a free source is identified.
• GR "non-conservation". In general relativity energy is conserved locally always; global conservation needs a symmetry of the spacetime, and lab-scale spacetime is flat to extreme accuracy [memory, high]. This offers no 1e-3 J of free energy on a bench. Effect on verdicts: none on 1, 2, 4. It sharpens Claim 4(ii) ("hidden vacuum source"): the file wrapper is where Bearden states the proposed mechanism, and it is still not a measurable prediction beyond COP > 1 itself. I keep Not testable as stated for the vacuum-source part.

F2. The examiner's demonstration request, and the correction it forces [web: archive.org wrapper; PlainSite docket (Abandoned, failure to respond)]

Record (agency's characterization, with the agency's stake in rejecting): 2004-01-23 non-final rejection of all 14 claims under 35 USC 101 (device "will not operate as an electrical energy generating system"; figures with COP "inoperable as violating at least the first and second laws of thermodynamics"), also section 112 and double patenting over 6,362,718; the examiner required the applicants to "demonstrate how the magnetic flux stored within the permanent magnet is sufficient to contribute usable electrical energy". Applicants (2004-05-24) asked that the demonstration requirement be held in abeyance; final rejection 2004-07-26 refused; abandoned 2005-05-16; no appeal found. The file does not say why they did not respond.

Correction to my Claim 5 text: I wrote that "the patent examiner granted it, so it was not dismissed as such". That is true for the parent (granted 2002) but incomplete. A second US examiner later rejected the continuation on thermodynamic grounds without testing and asked for a demonstration, and the applicants did not supply one. So the US record shows both: dismissal on theory (the examiner did not run a test) and a request for a demonstration that went unanswered. Competing explanations the Investigator listed: (a) no unit could show net output under examination, (b) a redundant continuation not worth fighting for (the examiner found double patenting), (c) trade secrecy (Bearden 2002: demonstrations only to prospective financiers). The file decides none. My view: (b) is plausible, but a demonstration of a self-running device would have been inexpensive next to the 12 years of EP fees (F4); the absence is an anomaly I cannot resolve. It is not evidence of failure by itself.

F3. Measurement evidence from builders (JLN Labs) [web: http://jnaudin.free.fr/meg/megv21.htm ; /ths720p.htm ; /megnot01.htm]

Findings, each by its source: - Naudin (private experimenter, builder of a MEG v2.1; he has an interest in the device succeeding): "measurement artifacts remain possible"; excess appears only with a "conditioned" carbon resistor or a metal-oxide varistor (MOV) load; "apparent" output (6.76 W, with 3.25 W total input of which 1.75 W is the control board) exceeded the Joule heat in the load; calorimetry and loop closure "not yet done"; he is "pessimistic" about calorimetry. - Tektronix customer support (the scope maker; no stake in the MEG), forwarded by builder Jon Flickinger and posted 2002: "the RMS measurement on a math waveform is bogus"; true power is the mean of v(t) i(t) point by point. The patent says its figures "agree with RMS power measured using a Textronic THS730" [read, patent]. - Flickinger (independent builder, single source, undated page): with linear resistive loads no excess; COP 1.75 to 5 only with MOV loads; he asks anyone with a linear-load COP > 1 to speak.

Weighing (a, b, c): This is the most useful new evidence on the measurement question I had flagged as a probable artefact. Worked example showing why the instrument matters: a sinusoidal case, peak voltage 4000 V and peak current 12 mA, 90 degrees out of phase (a pure inductor or capacitor). The product Vrms x Irms = (4000/1.414)(0.012/1.414) = 24 W "apparent power" while the true mean power v(t)i(t) is exactly zero. For non-sinusoidal 87.5 kHz pulses the error is not that extreme but can be of either sign, and it is largest for a high-impedance reactive output. Nonlinear loads (MOV, "conditioned" carbon) make it worse: the current is not proportional to the voltage, so an RMS-times-RMS product has no defined meaning. That excess appears with nonlinear loads but vanishes with linear ones is what an instrument artefact predicts; real extra energy would not care what the load is (the patent's own concept would let a lamp light). Naudin's "apparent exceeds Joule heat" is the same statement: the heat is the real power measurement, the scope number is not. Still, these are informal builder reports: no calorimetry, uncontrolled, one source for Flickinger's linear-load result, and Naudin's COP numbers are the scope-math kind. Effect: changes the weight on the patent's measured COP 1.7 to 2.6 (Fig. 8): the measurement method named in the patent is one the instrument's maker says is invalid for these waveforms. Records of the full-waveform data behind Fig. 8 are still not available (my earlier note stands). Test (c): Naudin's and Flickinger's builds are partial replications of the signals and, by their reports, not of the net power; no one reports calorimetry. Never properly tested stays accurate for the strong form, but the informal evidence now points one way (see verdict).

F4. European patent EP 1 446 862 B1 and the EPO file [web: https://patents.google.com/patent/EP1446862B1/en ; https://register.epo.org/application?number=EP01988155]

• Granted 2013-10-23 after a 12-year file, no opposition. Granted claim 1 requires non-saturating switching and feeding part of the output power back to the switching circuit (self-powering loop). Interested party note: a grant is the office's act under the EPC; the examiner's 2006 objections (as read, pages 1-3 only) were novelty and inventive step, not conservation of energy. I note the Investigator has not read whether industrial applicability was raised, so I cannot say the EPO "accepted" COP > 1; it examined patentability of a structure.
• The EPO found the MEG structure itself anticipated by US 4,006,401 (deRivas 1977; same structure, driven from an alternating source), with the novelty only in driving from the device's own output (also FR 2,312,135 Galley 1975, US 4,883,977, US 2,866,943). Substantiating point for my weak form: the circuit as a flux-switching device is ordinary and old.
• Evidence-removal lead (Investigator): the European description lacks 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 ... self-actuating form", but keeps the 3.44 extrapolation and the demagnetisation paragraph. Why is in the unread file (description pages replaced 15.06.2005). Competing readings: (i) trimming of unsupported performance statements under examiner pressure, (ii) routine redrafting. I cannot decide; I flag it because a reader weighing Claim 5 should know the measured-COP evidence was not carried into the European text. The file's 2007 replies, 2008 annex and 2013 minutes are the records that would say.
• The extrapolation point: my earlier note stands (3.44 extrapolated). The granted EP text also keeps the self-powering claim, which has no reported run in any source I have seen. Effect on verdicts: none beyond Claim 5 (below). The EPO file and the "fees to year 12" show sustained spending on European protection through 2013 (F5), which is a fact about the company, not about the physics.

F5. Company, assignment and financing record [web: https://patents.google.com/patent/US6362718B1/en ; arc-sos.state.al.us ; Google directory results]

US 6,362,718: inventors' names until a 2006 nunc pro tunc assignment to Magnetic Energy, Ltd. (Alabama); 2006 security agreement to Colmac Holdings Limited (Canada); lapse and reinstatement 2006; fees to year 12 (2013); expired. No entity named "Magnetic Energy" in the Alabama register (an unrelated non-profit only); directory listings (Manta, chamber-of-commerce) place a "Magnetic Energy Ltd" in Huntsville as a "noncommercial research organization" (Investigator did not open them). Bearden, 15 Aug 2002 (the company's own spokesman; his statements): "successful little experimental devices", no finished units, $9 million a year for three years sought, demonstrations "with nothing hidden" only to prospective financial partners whose technical staff "do their own measurements", planned 2.5 kW and 10 kW units; plus a claim that a charged capacitor or electret on a magnet with E perpendicular to H is a "proven free energy generator" (see F6). Related: Found. Phys. Lett. 14(1) 2001 author list (15 AIAS authors) includes Bearden [web: jnaudin.free.fr/meg/meg.htm], upgrading my earlier [memory, medium] item to [web].

As a physicist I treat this as an absence-of-evidence anomaly, not neutral. Explanations: commercial failure (consistent with a lender-only financing trace, a lapsed fee, a small unregistered entity), trade secrecy (Bearden's stated policy), or a working device that could not be financed. Evidence for each is thin; the one thing that would discriminate is the financiers' own measurement reports (not found) and the Colmac security document and any release (not obtained). The physics does not change either way; it does mean that the claim that investors verified the MEG rests on no document I have seen. Silence is not contradiction, and I do not count it as such.

F6. The Poynting "free generator" remark (Bearden, 2002) [web: billstclair mirror of cheniere.org/correspondence/081502.htm]

Bearden's statement is that E perpendicular to H gives a Poynting flux S = E x H / mu0 and so a charged capacitor across a magnet is a free-energy generator. The mainstream view: a static E x H exists in such a configuration (hidden momentum/Feynman's disk-type examples are textbook, Griffiths Ch. 8) [memory, high]; it is a circulating flow of field energy with zero net divergence. The energy in the fields is fixed; S being nonzero does not mean energy is leaving anywhere (the relevant quantity is div S plus the rate of field-energy change, and both are zero in a static case). It is a known and legitimate example of Poynting flow without power delivery. Not testable as a generator: no device is given. Bears on Claim 4 (hidden source): it is the mainstream term, not a new source.

F7. Quantitative check I added: the stored-magnet-energy budget [memory, medium for the energy density; no web check made, so flagged]

The patent names the magnet's flux as the energy source and says demagnetisation follows. An NdFeB magnet has a maximum energy product of about 300 to 400 kJ/m^3 [memory, medium]; for a 10 cm^3 magnet that is at most about 3 to 4 J as an upper bound on usable energy. At the patent's 22 W output (with about 12 W of input), a net surplus of about 10 W would drain 3 J in about 0.3 s. A self-running generator at COP > 1 would therefore need the magnet's energy to be recycled, not consumed; the patent gives no figure for demagnetisation rate. In an ordinary flux-switching device the magnet flux just flips between paths reversibly and no magnet energy is consumed, which is why the circuit can work as a transformer-like device with COP below 1. This is consistent with the weak form and sets a falsifiable test: measure magnet flux before and after hours of operation. No source reports it.

Verdict changes

• Claim 5 (MEG). Previous: Never properly tested (strong form expected to fail). Now split. (i) Working flux-switching, transformer-like circuit (weak form): Plausible, untested as stated, strengthened by the EPO finding that the structure is old (F4). (ii) Patent's measured COP 1.7 to 2.6 as evidence: not supported as a power measurement, because the stated method (scope-math RMS power) is declared invalid for non-sinusoidal waveforms by the instrument maker and the excess tracks nonlinear loads (F3). (iii) Strong form (sustained COP > 1 with a linear load, and self-powered running): I move this from Never properly tested to Never properly tested for the self-running part (no run on record in any source) and Tested, not supported at an informal level for linear loads (one independent builder's report, uncontrolled, no calorimetry). I do not claim a neutral controlled refutation exists; a calorimetric or battery-run test remains the deciding experiment and still has not been done. I also correct the dismissal question: US continuation rejected on 35 USC 101 with an unanswered demonstration request (F2), so "dismissed on theory, with a demonstration request left unanswered" now describes the US record.
• Claims 1, 2, 3, 4: no verdict change. The file wrapper (F1) supplies Bearden's vacuum-to-MEG mechanism but adds no measurement or calculation that would alter the earlier verdicts; the AB, Lee-Yang and GR arguments are answered above. The Poynting remark (F6) and the stored-energy budget (F7) were folded into Claim 4/5 without changing verdicts.

Records that would decide what remains open: the full Fig. 8 waveform data; calorimetry or a battery-run self-powered test; the financiers' measurement reports; EPO 2007/2008/2013 papers; the Colmac security document and any release; USPTO file of 09/656,313.