Physicist review: Kelly-Patrick, "The Practical Guide to Free-Energy Devices"

Reviewer: PhD physicist (Step 2). Date: 2026-10-08. Inputs read: claims.md, facts.json (summaries and key claims), the whole deep-read file (deep/PJKbook-9cc3a5.md; I read the Sweet, Chapter 3 Don Smith/Mohamed, gravity, Chapter 5, Chapter 7 Moray, Chapter 8 Papp, Chapter 10 water, Chapter 11 Newman and suppression, Chapter 13 MEG, and Appendix Don Smith/Ainslie sections in full, and the rest by section headings and summary tables), sites.md "Leads for review" and all eleven research notes. I did not re-read the 5.7 MB book text; every page cite below is the deep reader's, which I checked for arithmetic and consistency, not against page images. Tags: [read] = from the project files, [web] = URL given, [memory] = my recollection with confidence, [calc] = my own arithmetic.

Source note. The book is a compilation. Kelly says it is mainly a report of what other people say and that you may treat it "as a work of fiction" (p.11) [read]. So it is primary evidence for what the inventors and their promoters claim, and almost never evidence that a device works. The "Investigator" findings (later editions V34.5 and V34.8; Kelly's pen name and death in July 2020) do not change any claim; they are kept in section "Investigator findings kept in the record" below.

Web verification budget. I tried two checks (Ainslie COP 17 independent results; Kapanadze independent tests). Both searches returned nothing usable [web: failed]. Neither verdict below depends on them. All physics constants I use are textbook values [memory, high].

---

What mainstream physics says, stated fairly

1. Energy conservation is a law about every system, open or closed. An open system can deliver more than the operator supplies if energy flows in from somewhere identifiable: a heat pump (COP about 3 to 5), a sailing boat, a solar panel. Kelly is right about this [read: pp.12-17, 422], and the textbook agrees. The question for every device is therefore "where does the inflow come from, how large is it, and was it metered?", not "is the system closed?".
2. Two limits then apply. (a) The second law (Kelvin-Planck): a cyclic device cannot turn heat from a single temperature reservoir, such as room air, into work. (b) Fluxes: an ambient source can only deliver its power density times the collecting area. Both are ordinary physics, not dogma.
3. A permanent magnet, gravity, and a static electric field are conservative: a system that returns to its starting state after one cycle has had zero net work done on it by them. Cyclic machines built from these cannot net output power from them.
4. What would change this: a measured excess in a controlled test that survives metering of every input and the removal of every artefact, plus an identified source. The project's documents contain no such test (see per-claim answers).

---

Claim 1. "Closed system": COP above 1 is allowed because devices are open to ambient / zero-point energy

(a) Mathematics. Kelly's statement is verbal; it has no equations. The weak form (open systems can have COP > 1) is internally consistent. The strong form (any device with sharp pulses is "open" to a field that supplies useful power) has no stated source, so there is nothing to check for consistency. Kelly contradicts himself once: "over-unity is an impossibility" yet COP > 1 is fine if the extra comes from the "zero-point energy field" [read: p.422].

(b) Consistency with experiment. Weak form: fully consistent (heat pumps, solar, wind). Strong form: the measured ambient fluxes are tiny. Fair-weather atmospheric electric field about 100 V/m and vertical conduction current about 2 pA/m^2 give about 2 x 10^-10 W per square metre [calc from textbook values, memory medium for the 2 pA/m^2]. Even 100 m^2 of collector gives 20 nW. Radio field near a 50 kW broadcast station, about 1 V/m, gives E^2 / 377 ohm = 2.7 mW/m^2 [calc]; a crystal set picks up microwatts to milliwatts, which is the real, measured "free" radio energy. Zero-point energy is the lowest energy state; by definition nothing can be extracted from it (Casimir force is a difference between configurations that you can pay for once, not a supply) [memory, high]. Kelly's "a battery pours energy in from the environment" and "the tricky bit... has been done" [read: pp.16-17] is asserted without any measurement. His Lee and Yang (1957 Nobel, parity violation in weak decays) citation as proof of broken symmetry is a non sequitur: parity violation says nothing about energy supply [memory, high].

(c) Tested? Not as a general claim. Each device is a separate test (below). In the whole book the deep readers found no input and output measured together by a named independent party for most COP figures [read: claims.md cross-cutting facts].

(d) If true, what would change: conservation of energy would stay, but the identification of a usable, extractable ambient reservoir at kilowatt scale would contradict measured field strengths, the second law (if it is thermal), and quantum field theory's definition of the vacuum as the ground state (if it is zero-point).

Verdict (split): - Weak form (open systems may exceed the operator's input): Established. - Strong form (pulse or resonance circuits tap an ambient or zero-point field for kilowatts): Refuted by calculation for the named reservoirs (atmospheric electricity, radio, thermal at one temperature, vacuum ground state), and Never properly tested for any as-yet-unnamed source.

---

Claim 2. Sharp pulses, resonance and a tiny drive tap the field (Joule Thief, Bedini, Benitez, Don Smith, Kapanadze, FLEET, Ainslie)

(a) Mathematics. - Benitez GB 121,561 (patent's own worked example [read: pp.452-458]). Worked check [calc]: capacitor 0.222 microfarad charged to 1,200 V stores E = 1/2 C V^2 = 0.5 x 0.222e-6 x 1200^2 = 0.16 J. At 1,000 sparks per second that is 160 W, which is below the 321 W the patent says arrives at the circuit, so the supply side is self-consistent. But the patent then says the oscillating circuit carries 40 A through 15 ohm, which dissipates I^2 R = 40^2 x 15 = 24 kW. The ring lasts 72 cycles at 71.9 kHz, about 1 ms, so the energy dissipated would be 24 kW x 1 ms = 24 J per spark, 150 times the 0.16 J stored. Also with L = 22.2 microhenry and f = 71.9 kHz the reactance is omega L = 2 pi x 71.9e3 x 22.2e-6 = 10 ohm, so Q = 10/15 = 0.67; a circuit with Q below 1 does not ring 72 times. The example is internally inconsistent: its own numbers cannot supply its claimed 2,400 W (COP 6). The deep reader noted the same 24 kW problem; the energy-per-spark and Q checks are mine. - Don Smith / Mohamed formula "Power = 1/2 C V^2 F^2" and "C V^2 F^2" [read: pp.274-277]: the units are J/s^2 (watts per second), not watts. The correct 1/2 C V^2 f is the power needed to charge the capacitor f times a second. For Mohamed's example (C = 10 nF, V = 30 kV, f = 100 kHz) that is 0.5 x 1e-8 x (3e4)^2 x 1e5 = 450 kW [calc], to yield the 30 kW he wants [read: p.284]. His "40 kV of reactive power" is a voltage called a power [read: p.283]. His resonance equation (a) gives twice the textbook frequency and (b) uses the textbook factor, so they conflict by a factor 4 in capacitance [read: p.279-280]. Mohamed retracts parts of his own Part 1 [read: pp.285-295] ("talking about an electric current that doesn't exist"). - The central teaching point: reactive power. In a resonant L-C circuit the energy sloshes back and forth; the current and voltage can be very large (Q times the drive) while the real power drawn (V_rms I_rms cos(phase)) is small. The circulating VA is not available as net work; you can only take out what the drive and losses allow. Large VA readings with tiny real input are the commonest source of apparent over-unity. - "Doubling the voltage quadrupled the power" [read: p.312] is P = V^2/R for any fixed resistor, true of every circuit and therefore not evidence for anything. - Ainslie 555 circuit [read: p.2412-2414]: the deep reader's checks (the 555 timing gives about 2.9 kHz at 4% duty; the 10 ohm load coil has about 14 microhenry and 44 ohm reactance at 500 kHz, so it is not a 10 ohm resistor at that frequency) show the quoted "power" cannot be computed from V x I without the phase. Tuning "for the highest battery voltage reading" [read: p.2414] reads terminal voltage, which is not charge.

(b) Experimental consistency. Joule Thief efficiency is textbook-level (a boost converter) and Kelly's own lux test found it worse than direct drive: 139 to 221 lux/W against 470 to 918 lux/W [read: p.476]. That result is consistent with mainstream physics and is honest on Kelly's part. The Ace_Propulsion "COP 1.8" is two DC-meter readings of a pulsed waveform (12.5 mA at 2.35 V in, 8 mA at 6.6 V out), where the product of averages is not the average of the product [read: p.482; calc]. Bulb brightness (Salty Citrus, Giannopoulos, Ming Cao) is not power: a filament lights visibly at a few percent of rating and high-frequency drive makes input metering hard; Kelly himself warns of this [read: pp.314, 311-312]. The Russian "2-2.5 kW" unit has a 12 V 0.8 Ah battery, which holds about 17 s at 2 kW [read: p.301; calc]. Battery charging claims (Bedini, Callanan, Halay, Chapter 6): the book's own caution is that "we have no instrument which can directly measure the flow of radiant energy into the charging battery" [read: p.544]; Sucahyo's warning that voltage can rise with no charge behind it (surface charge) is textbook electrochemistry [read: p.556]. Kelly's FLEET battery swap rests on rested terminal voltages, with no capacities or discharge test [read: p.577].

(c) Tested? Almost entirely by the inventors or their followers. Named exceptions: Kelly's lux test (negative), Mohamed's self-reported inability to measure frequency [read: p.276], Ming Cao's rejection of Smith's 4:1 wire rule [read: p.295]. No replication with metered input and output is on record. Kapanadze: two WO patents (2008) [web: Google Patents, via Investigator] and unsourced numbers (40 W start, 5 kW out); my own search for independent tests found none [web: failed to find any, 2026-10-08]. Ainslie: Kelly says COP 17 is "not yet replicated" and defers to a Panacea document I could not find [read: p.509; web: failed]. Tseung's open-system principle ("X units supplied plus Y units led-out from the surrounding", Energetic Forum 2011) is the claim 1 argument and carries the same burden [web: via Investigator].

(d) If true: a nameable ambient energy source with kilowatt-scale flux would be required; spark and resonance behaviour would otherwise be entirely classical (Maxwell, circuit theory), so no change to electrodynamics is needed for the circuits, only an unknown supply.

Verdict: Refuted by calculation for Benitez GB 121,561 (its own numbers), Smith's "C V^2 F^2" power formula, and Mohamed's example (input charging power 450 kW against 30 kW output). Never properly tested for the specific device claims (Kapanadze, Smith's 160 kW, Ainslie COP 17, FLEET, Bedini charging): no controlled, metered test on record; they were not so much dismissed as never offered for neutral measurement. Tested, not supported for the one measured Joule Thief comparison (Kelly's own lux test).

---

Claim 3. Switched or "conditioned" permanent magnets give large gains (Annis/Eberly, Flynn, Heins, Kawai, MEG, Sweet VTA)

(a) Mathematics. The permanent magnet is a state-function store: take the flux around a closed cycle and the net work done by the magnet is zero. Output power can exceed the control (drive) power only if the output windings are also loading the magnet, which then must be re-magnetised; that costs energy that the "COP" definition leaves out. Annis/Eberly define COP as "Output power / Drive Power" for the switches only [read: p.213, 216], so COP 200 to 400 compares output with the saturation-coil power alone and ignores whatever else supplies the flux changes. Their own cited numbers (rise time of 2 microseconds; Hitachi FINEMET Bs 1.21 T; NdFeB Br 1.43 T) are ordinary material data. MEG (Bearden, US 6,362,718): "two outputs of 48 W for one input of 12 W", COP 4 per output and 8 total [read: p.1146]; Kelly says no one has replicated COP > 1 on it [read: p.1145] and then says "replicated by others" two pages later [read: p.1147]; Dave Lawton found no difference when he removed the magnet [read: p.1147], which is what the energy argument predicts. Kawai: 19.55 W in (17 V x 1.15 A) and 62.16 W out [read: facts.json; arithmetic checks], one patent run at 100 rpm, by the patentee.

Sweet VTA, the extreme case: - Kelly: >500 W at 120 V 60 Hz from <0.31 mW (COP 1.6 million) [read: p.244]. The Investigator traced the numbers to one filmed 1987 demonstration: 10 V x 31.2 microamp = 0.312 mW in, 112.5 V x 4.7 A = 528.75 W out, "power factor measured to be 1", Bearden reading a wattmeter at about 500 W [web: aboveunity.com notes via Investigator; the video was not watched]. 0.312 mW at 10 V is a 320 kohm input impedance; at 4.7 A and 112.5 V the load is 24 ohm, about five 100 W lamps in parallel [calc]. - Sweet's 1987 fund-raising letter says "one milliwatt" in, "over 200 watts" (recently "over 500") out, "up to 1,000 watts" in tests, written to "approach selected potential financial backers" [read: page image p.1, rex-sweet-sparky1987.pdf]; the 1988 paper says 10 mW in, 5,000 W out (a ratio of 500,000) [read: A-817]. The in-book spread (0.31 / 1 / 10 mW in; 500 W / 1 kW / 5 kW out) is real and is not explained [read: deep, section B]. - Sweet's earliest lab notes (as paraphrased in the AboveUnity notes) give 6 to 19 W and then 100 to 110 W out, with 10.41 V x 1.84 A (about 19 W) on the excitation coil in one reading [web: hyiq.org, aboveunity.com via Investigator]. Ashley Gray's account (1994) gives 24.2 V x 4.6 A = 111 W out from 7.2 V x 143 microamp = 1.03 mW in (COP about 108,000), with coil impedance "several hundred thousand ohms" in the field against 2 to 3 ohms outside [read: pp.248-249]. Those two numbers are an inconsistency: 7.2 V / 143 microamp is 50 kohm, not several hundred thousand, and the "2 to 3 ohm" figure does not match a 250-turn coil [read: deep, A4]. - Physical constraints [calc]: output 500 W for one hour is 1.8 MJ from a "4 x 6 x 6 in" (about 2.4 litre) box. A 20 W lamp-load run lasted 10 to 12 hours in Gray's version, about 0.8 MJ [read: p.249]. A transformer or inductor with 143 microamp of primary current cannot deliver 4.6 A: the load current appears in the primary as reflected current (ampere-turns balance). Staying at 143 microamp means the energy is coming from somewhere else, or the readings are wrong. - "Negative energy from the Dirac Sea" [read: A-817]: in Dirac's picture the negative-energy states are all filled, which is what forbids falling into them (Pauli principle); the Dirac sea is a bookkeeping device for antiparticles, not an energy supply [memory, high]. - Side claims. "Weight reduced by about a pound" at 1 kW [read: p.507]: a pound of mass is 0.45 kg, whose energy is m c^2 = 0.45 x (3e8)^2 = 4e16 J; delivering 1 kW for an hour (3.6e6 J) corresponds to 4e-11 kg [calc]. The two numbers are 10 orders of magnitude apart. "Frost on the wires when shorted" and instruments that "read zero above 1 kW" [read: p.507] are untested and also characteristic of how a failing measurement behaves; they cannot be left out of the record, but they carry no weight without a recording. - Mohamed's page cross-reference "A-1209" for the Sweet text is wrong in this edition (the text is at A-817) [read: claims.md corrections].

(b) Consistency with experiment. Magnetic hysteresis, eddy-current loss and reluctance switching are measured, standard physics, and every such measurement finds output below input. The only reported contrary data are in the claim's own documents, all by promoters or by Gray as builder (stake: Gray offered his build for Sweet's replication; Goldes funded Sweet).

(c) Tested? Sweet: one demonstration filmed in 1987 with Bearden present (Bearden is an advocate, not a neutral party); hyiq.org has tried to replicate "for a number of years now, without any success" [read: p.247, Weyrich quote]; Kelly knows of no replication [read: p.244]. So: tested by replicators who did not succeed, but no controlled neutral test exists either way. Annis/Eberly application: "Abandoned" [web: Google Patents via Investigator]. MEG patent: lapsed for unpaid fee 26 Mar 2006 [web, same]. Flynn US 5,455,474 lapsed 3 Oct 2007 [web, same]. Reasons are not stated in the record; abandonment and lapse are facts, not proof of failure.

(d) If true: a permanent magnet cycled around a closed path would give net work, so the magnet would no longer be a conservative store; the Dirac sea would become an energy supply. That would revise quantum field theory's vacuum, not merely add a feature to it.

Verdict (split): - Switched-magnet COP above 1 as defined by the patentees (Annis/Eberly, Kawai, MEG): Never properly tested as an independent measurement (patentee tests only); the cycle argument gives a strong theoretical expectation against it. Where the COP is defined without the magnet re-magnetisation cost, Internally inconsistent (the definition omits an input). - Sweet VTA, the claimed numbers: Never properly tested. The evidence is one filmed demonstration plus fund-raising and builder accounts, with figures that disagree with each other by factors of 10 to 10^4. Instrument artefacts (VA read as W, phase, high-frequency or distorted waveform metering, a hidden supply) have not been ruled out, and the video was not examined in this project. - Sweet VTA, the mechanism ("negative energy from the Dirac sea"): Refuted by calculation (see mass-energy check; and the filled-sea picture gives no extractable energy).

---

Claim 4. Impulse, gravity and buoyancy devices produce excess mechanical energy (Tseung, Campbell, Hardy, Milkovic, Dmitriev, Skinner, Kwok)

(a) Mathematics. Gravity is conservative: the work around any closed path is zero (curl of g is zero, Stokes). Tseung's "2 m g r" excess per pulse for a flywheel [read: p.366] has no derivation; 2 r is the height change if the whole mass were lifted by one diameter, but a flywheel centred on its axis does not change its centre-of-mass height at all. A jerk adds torque impulse, and the work done is torque times angle, which the motor already counts. Kelly's "load dropped on a bridge" analogy [read: p.362] is a stress doubling (dynamic load factor 2), not a gain in energy. Buoyancy generators (Kwok): worked check [calc]. Put 1 m^3 of air at depth h = 10 m. The buoyant force is rho g V = 1000 x 9.81 x 1 = 9.81 kN, and rising 10 m returns at most 98 kJ. But getting the air into the bottom of the column needs pressure-volume work against rho g h + atmosphere, at least rho g h V = 98 kJ, so the net is zero at best, negative with friction. Skinner [read: p.405-411]: a mass circulating at fixed height gains nothing from gravity; Kelly's "spinning chair" argument [read: p.407] explains why spinning mass stores energy you put in (energy accumulates over time), not why it gives out continuous power. Newsreel arithmetic [read: p.406]: Kelly's own check that "COP 12" would give each of 3,500 homes 2.6 W; his reading of "1200 per cent" as 1200 times gives 255 W per home.

(b) Consistency with experiment. Every careful test of flywheels, pendulums and belts finds the stored kinetic energy plus losses equal the input. The Campbell "750 W in, 2,340 W out" [read: p.366] is a replication video with no meter data and a note that the flywheel "does not appear to have a heavy flywheel" [read: p.366]; the ratio on the stated figures is 3.1 [calc]. A flywheel run up to speed stores energy that can briefly drive a generator at high power (for example 0.5 I omega^2 released over seconds), and that, plus the motor's high starting current, is the classic source of a short-lived apparent excess; a test needs the total energy in minus the total out over many minutes, with the flywheel at the same speed at start and end.

(c) Tested? Skinner is the best documented historical case, from Investigator sources. British Pathe "Gravity Power!" (31 July 1939, film ID 1190.10) says 1200 per cent; the United Press story (4 April 1939) calls it Skinner's "estimate"; consulting engineer A. P. Michaels called it "a practical device"; George C. Estill (retired president of Florida Power and Light) examined it but his check "was not complete enough to allow him to form a conclusion" [web: britishpathe.com/asset/68269/; rexresearch.com/skinner/skinner.htm via Investigator]. That is one favourable letter and one inconclusive examination, not a controlled test; no later record. Campbell, Hardy, Milkovic, Dmitriev, Kwok: videos only. Kelly doubts Ribero's patent and the Tseung prototype that "stops after five minutes" [read: pp.363, 402-404]; a ten-foot version that "rotated steadily for ten months" is unnamed and unsourced.

(d) If true: gravity would not be conservative, or the flywheel would need an unknown torque. Newtonian mechanics and the equivalence principle would be affected.

Verdict (split): strong form (net energy from gravity or buoyancy in a cycle): Refuted by calculation. Weak form (any specific machine shows a measured, sustained excess): Never properly tested; the stronger cases (Campbell self-running, Skinner 1939) are single-source, unmetered, and were not subjected to neutral tests, though Skinner got an inconclusive engineer's check.

---

Claim 5. Water splitting and engines on water for far less than textbook energy (Boyce, Meyer, Cramton, Zigouras, UK generator)

(a) Mathematics. Faraday's law of electrolysis: 2 electrons per H2O, F = 96,485 C/mol. One litre of HHO (2 H2 + O2 mixed, 0 C, 1 atm; 22.41 L/mol of gas, with 3 mol of gas per 2 mol H2O) needs 5,740 C [read: deep, section H; calc agrees]. The minimum electrical work is the Gibbs free-energy voltage 1.229 V per cell: 5,740 x 1.229 = 7.05 kJ = 1.96 Wh per litre. At the thermoneutral voltage 1.48 V (no heat exchange) it is 2.36 Wh per litre, which Kelly gives as 2.34 [read: p.845-847; calc]. Between 1.229 and 1.48 V a cell takes heat from its surroundings, so "stays cold" and up to about 120% of the Kelly baseline are within textbook physics [memory, high]. Therefore: - Boyce DC: 998 W for 900 L/h is 3,992 J/L, equal to 0.70 V per cell, which is below the 1.229 V floor. - Boyce pulsed: 180 W for 900 L/h is 720 J/L, equal to 0.125 V per cell. - Cramton: 36 W for 6 L/min is 360 J/L, 0.063 V per cell [read: deep section H; calc agrees]. - Zigouras: 9.4 L/min of water is 522 mol/min, so about 17,500 L/min of gas, which needs 2.5 MW; the input is 140 W peak 2.7 kW, a ratio of order 10^4 against the book's own "COP 5-10" [read: deep section 21; calc]. The book's two numbers for the same device disagree by three orders of magnitude. - UK generator: 3 L/min of HHO carries 0.36 kW chemical [read: deep section H; calc], yet a 4 to 5.5 kW electrical load is claimed (factor of 11 to 15). Cramton's claimed diesel saving is 42 to 73 times the chemical energy of the HHO added [read: deep section H]. - Input figures conflict inside the book: Cramton 12 W, 3 A at 12 V (36 W) and 24 W motor plus 12 W rotor [read: p.827-832]; 6 x 6 W = 36 W called 30 W [read: p.837]; Boyce 216 / 211 / 1000 / 1200 / 1170 percent [read: p.770, 846]. - Meyer water capacitor, US 4,936,961 "Example 1" is a patent text [read: pp.1836-1841]. It states parameters but, per the deep reader, no measured input-output balance.

(b) Consistency with experiment. Faraday's law is among the most accurately verified laws in chemistry; electrolysers are manufactured and rated to it. Gas produced above one litre per 5,740 C would need non-Faradaic gas (chemical reaction with the electrolyte, which Boyce himself says in part [read: deep section H]) or an unmetered input. The "monatomic" or "charged water cluster" gas worth four times hydrogen [read: p.806, 821, 836] has no measurement. HHO boosters on cars: mainstream testing (alternator load, thermodynamic floor) finds no net gain [memory, medium; web check not done].

(c) Tested? Not independently for these devices. Meyer: Investigator found a 1996 Ohio court finding of "gross and egregious fraud" with three expert witnesses who found "conventional electrolysis" (Wikipedia, secondary), and the Franklin County coroner's finding of aneurysm in 1998 (Columbus Dispatch, partial) [web: en.wikipedia.org/wiki/Water_fuel_cell; dispatch.com 2007; court file not reached]. The plaintiffs and Meyer were opposing parties. That is evidence that experts examined the cell once and reached a negative conclusion; the court record itself has not been seen. No test report exists for Boyce or Cramton in the book.

(d) If true: Faraday's laws or the thermodynamic floor (Gibbs free energy of water) would fail. That is chemistry with a century of calorimetry behind it.

Verdict: Refuted by calculation for output beyond 120% of the textbook figure (all of 211% to 2,340%) and for the engine claims (chemical energy of the gas is orders of magnitude below the claimed output). Established for the baseline that boosters/electrolysers work at roughly 60 to 80% of ideal, with Kelly's own Faraday table. Tested, not supported for Meyer's cell in one court-appointed examination (secondary source; court file not read).

---

Claim 6. Aerial, passive and fuel-less devices deliver power (Moray, Tesla/Plauson aerials, Papp, Coler, pyramids, Joe Cell, Newman, Gray)

(a) Mathematics. - Moray. The record is of witnessed lamp and appliance ratings; no meter reading of output voltage or current appears [read: deep, C3 conclusion]. Hayes states in writing "I have no instrument in my laboratory that is able to measure the amperage or the voltage at this frequency" [read: p.604]. Hayes's 675 W is 100 W + 575 W by rating. The "635 W = 0.878 hp" has an arithmetic slip (635/746 = 0.851; 655/746 = 0.878) [read: p.603; calc]. Collected power from an aerial [calc, using E^2/377]: with 1 V/m (the field a few km from a 50 kW AM station) the power density is 2.7 mW/m^2, so even 100 m^2 of effective aperture gives about 0.27 W; 675 W would need 2.5 x 10^5 m^2 or a field of about 50 V/m (a few hundred metres from the antenna). Moray's claimed distance from lines was 26 miles and over 100 miles from radio stations (Hayes) [read: p.604], which makes it worse. The germanium diode / "valve": Kelly says the equipment "is said to have contained a germanium diode" [read: p.596], but the Marshall passage says Moray used a diode only to illustrate how his detector worked [read: p.604]; the correction already noted in claims.md. Moray's own statement that the valves "are not rectifiers... [they] stop the flow of energy... returning to the outer circuit" [read: p.610] matches the memory note that this is a one-way coupler, not a rectifier, but is a description, not a design anyone has built. - Papp: "a constant 300 horsepower output at 4,000 rpm" for 35 minutes with only the engine alternator as electrical source [read: p.642]. 300 hp = 224 kW; over 35 minutes (2,100 s) that is 470 MJ, which equals about 13 litres of petrol at 35 MJ/L [calc; memory high for heating value]. A 90 hp Volvo alternator provides on the order of 1 kW. The sealed noble-gas charge has no chemical energy; plasma recombination returns at most the ionisation energy that the electrical drive put in. Kelly's "Atomic Number below 19" for a mixture with krypton (36) and xenon (54) is his wording slip [read: p.642]. In the 27 October 1968 demonstration the cylinder detonation from a few cc of gas [read: p.642] is consistent either with a large electrical discharge, an unidentified chemical charge, or a deliberate explosive (Feynman's view, by his own account [web: hoaxes.org/comments/papparticle2.html]); the Investigator notes a Medium snippet that "Authorities found no explosives" (unopened) [web: snippet only]. - Newman. Worked example [calc]: Newman's own coil comparison: 3.35 million ft of 5-gauge wire, about 90,000 turns, inductance 25,700 H, 95 mA. Stored energy is 1/2 L I^2 = 0.5 x 25,700 x 0.095^2 = 116 J [read: pp.944-947; matches]. The energy is what the battery put in while the current rose (at 100 V the time constant is L I / V = 24 s). The collapsing field returns that same energy at switch-off; it is not additional energy. Counting the inductive "back spike" as output without subtracting what the battery put in is a double count. That is what the Investigator's finding suggests Hastings did: his output includes mechanical work plus heat plus "anomalous back spikes in the megaHertz range"; the National Bureau of Standards, using resistors at the coil terminals, made 77 measurements and found efficiency 27 to 77 per cent [web: law.justia.com Newman v. Quigg 681 F. Supp. 16 (D.D.C. 1988) via Investigator]. Both sides were parties to the suit (Newman, the Patent Office). Kelly's "800% efficiency / COP 8.0" cites no report [read: p.947]. - Don Smith claims (160 kW from 12 V, 1 A; 28.8 kVA suitcase) have no stated measurement method [read: deep, Appendix section 7]. "Sound is electromagnetic", "farad is one volt for one second", "T = RC is a frequency", "60 Hz counted as 120 cycles" are errors Kelly reproduces [read: deep Appendix section 8]. - Gray/Cal-Tech: 27 W in, 10 hp out, "reputed 100 times" vs "275 times" with no report cited, and the party presenting the device had a stake (Kelly says so) [read: deep, section M5]. - Joe Cell: operator-dependent ("persona factor"), "the only numbered case" is a truck with the fuel line connected (shandy mode) [read: deep, section 4f; claims.md]. Pyramids, Brock's rabbits, Drbal razor blades: no energy claim that can be tested as stated except the razor effect (dehydration of the edge is the mainstream account [memory, medium]).

(b) Consistency with experiment. Radio energy harvesting is real and measured at the microwatt to milliwatt level. No measurement of radiant energy at kW level exists outside the claims. Eyring's objection to Moray ("might be induction") is a standard mainstream objection, and Eyring stated the test that would answer it: operate three or four miles from all power lines [read: p.602].

(c) Tested? Moray: many witnesses over 40 years (Hayes PhD, Eyring and Fletcher of BYU, Gardner affidavit, Todd 1971 recollection [web: rexresearch.com/moray2/morayrer.htm via Investigator]); no instrumented measurement; the device is lost or destroyed (shooting, smashing, the dwindling "Swedish stone" [read: p.597, 605]). The Investigator's leads that matter: the family says "none of the original patent applications... are any longer available at the US Patent Office. Although their file jackets are there, the contents... are gone" [read: p.605; web: rexresearch]. The patent examiner's refusal grounds are quoted in the book: "no means was provided for causing the cathode to emit an appreciable number of electrons" and "no natural source of electric wave energy is known to the Examiner" [read: p.605]; the examiner's reasoning was an objection to the stated mechanism, and the examiner did not test the device. Moray's only grant found is US 2,460,707 (1949), electrotherapeutic apparatus [web: Google Patents via Investigator]; application 550,611 of 13 July 1931 is named in several copies [web: Google snippets]. Papp: Investigator found Ansley's 1989 San Jose Mercury News account ("Too many sensible people helped him build it and run it. Too many engineers say they couldn't find a hamster inside") with a Mattel engineer killed and "eight others badly wounded", against Feynman's "three men injured" [web: hoaxes.org/comments/papparticle.html]; Papp v. Feynman/Caltech was settled out of court with "Mr. Papp was paid something" (Feynman's account) [web: papparticle2.html]. The 1983 "independent certification test" has no details in the book [read: p.643]. Newman: tested by NBS (77 measurements, 27 to 77 per cent efficiency), contrary to Hastings's accounting (above). Ainslie: see claim 2. Coler: Kelly puts it in "Doubtful Devices" [read: p.1148-1150].

(d) If true: a new ambient source (for Moray, Tesla/Plauson), or a new energy source in noble gas (for Papp), or non-conservation in a coil (Newman). Moray's patent examiner asked the right question in 1931.

Verdict (split by device): - Moray: Never properly tested (the device was demonstrated to many witnesses but never instrumented; the original is lost; the examiner's objection was not a test). The aerial energy budget makes the claimed power implausible without an unnamed source. Note the file-jacket anomaly, which the Investigator should keep chasing. - Papp, 300 hp: Never properly tested by a neutral party; mechanism: Refuted by calculation (noble gas has no energy to give beyond the electrical input, 470 MJ in 35 minutes). - Newman: Tested, not supported (NBS 27 to 77 per cent efficiency, though the Patent Office was an interested party); Hastings's COP 8 contains a double count. - Don Smith and Gray/Cal-Tech: Never properly tested; formulas Internally inconsistent (units). - Joe Cell, pyramids: Not testable as stated (operator "persona" factor; no input-output quantity defined). - Coler, Plauson, Tesla aerial plate, Rayburn TREC: Never properly tested; digest-level coverage only for several (Chapter 14, p.1901-2238 patents).

---

Claim 7. Suppression of free-energy inventors by vested interests

This is not a physics claim, so my verdict applies to the physics part only: whether the suppression argument is needed to explain why no device is on the market. The physical objections above (Benitez's own numbers, the Faraday floor, conservation in cycles) make the absence of commercial devices explicable without suppression. That does not make suppression false in any individual case.

(a) Mathematics: n/a. The essay "The Suppression of Knowledge" [read: pp.993-1005] is polemic with few attributed documents (Archibald graph, Woodhead remark); the global claims ("more than 40,000 free-energy patents confiscated", "I am reliably informed"; "10,000 Americans built free-energy devices") are uncited [read: deep section E]. US patent secrecy orders under the Invention Secrecy Act exist [memory, medium: not web-checked here], but I have no source linking them to free-energy devices.

(b)-(c) What the record shows, with the Investigator's finds kept: - Documented, single-source or party-sourced: Moray's original application contents missing from the file jackets (family statement) [web: rexresearch]; Papp's partnership and investors' suits (Ansley 1989) [web: hoaxes.org]; Newman v. Quigg, where the Patent Office refused his patent for inoperability and the court ordered an NBS test [web: Justia]; Meyer's 1996 fraud finding and 1998 death by aneurysm (coroner, hypertension), with his brother's report that Meyer said "They poisoned me" [web: Dispatch 2007; Wikipedia]; Moray's shooting and smashed equipment, relayed by Kelly without date or assailant [read: p.597]; Sweet's death in July 1995 "after two unknown visitors", from "Maurice", an unnamed associate [read: A-806 to A-808]. - Records not obtained that would decide: USPTO file for 550,611; Moray's notebooks; police or hospital records of the shooting; Franklin County coroner's 1998 report; Ohio court judgment (1996); Papp suits (Roser v. Papp; Papp v. Feynman/Caltech); coroner's record of the Mattel engineer. - Absence-of-evidence analysis, in brief. Where a record goes dark (Moray's files; Sweet's notes; Kelly's later years), the competing explanations are (1) deliberate suppression, (2) loss through poverty and death (Kelly's own record stops in 2020 because he died on 11 July 2020 [web: FESIG memorial video; OverUnity Research snippet]), (3) commercial failure, (4) the device did not work as claimed. For Moray the record can favour (1) or (4); the missing file contents are the strongest anomaly. For Kelly, (2) fully explains it.

(d) If true: the Patent Office, courts and coroners would have to be systematically wrong or complicit; that would be a historical finding, not a physics change.

Verdict: Not testable as stated for the global claim; individual episodes (Moray's missing file, Meyer's court case) are Never properly tested historically because the primary records have not been obtained.

---

Investigator findings kept in the record (and how they bear on the physics)

• Later editions: V34.5 (24 June 2019, 2,561 pages) and V34.8 (26 Feb 2020, final) exist; the preface disclaimers are unchanged; new sections include "The Buie-Morin Power System" and "Art Porter's Magnetic device" [web: tuks.nl PJKbook.pdf; archive.org/details/pkj_free_energy]. My review is of V32.9 only. Any changed figures in V34.x are unchecked.
• Identity: "Patrick J. Kelly" is a pen name for Edward Dickson Williams per the FESIG memorial (5 Aug 2020); he died 11 July 2020 [web: youtube.com/watch?v=MnrNmV-vYO4; overunityresearch topic 3932 snippet]. He ran a free-energy information site: a stake.
• Sweet: figures traced to one filmed 1987 demo (0.312 mW in, 528.75 W out), a fund-raising letter, and 1986 lab notes with much smaller outputs (6 to 19 W, then 100 to 110 W); funding by Mark Goldes (Aesop Institute), who has a stake [web: aboveunity.com; hyiq.org; rexresearch.com/sweet].
• Tseung's statement of principle ("X supplied plus Y led-out") [web: Energetic Forum 7530].
• Related work: US 11,588,421 B1 (Lyden, 2023) cites Benitez GB patents and Moray US 2,460,707; a granted patent states the applicant's characterisation, not a measurement [web: Google Patents].
• Benitez family has four GB documents (17,811/1914, 5,591/1915, 14,311/1915, 121,561/1918); Kelly lists three. My Benitez energy check applies to 121,561 [web: Google Patents].
• Skinner (Pathe and United Press), Moray witnesses, Papp (Ansley, Feynman), Newman (NBS, Hastings), Meyer (court, coroner): treated above under the claims they bear on.

Corrections to Step 1

1. claims.md claim 2 says Benitez GB 121,561 is "400 W in, 2,400 W battery power". The patent's own worked example is internally inconsistent: energy per spark 0.16 J (160 W at 1,000 sparks/s) against 24 kW of dissipation in the 40 A, 15 ohm circuit, and the circuit has Q about 0.67, so it cannot ring 72 times. This sharpens the deep reader's note (which gave the 0.16 J figure) and should be carried into the arbiter.
2. claims.md claim 5 says Boyce's DC figure is "211-216 percent of Faraday". Faraday's law itself fixes only charge per litre (5,740 C per litre of HHO), not energy; Kelly's "Faraday" baseline is the thermoneutral energy figure 2.34 Wh/L (1.468 V per cell), as the deep reader found. The summary should say "of the 2.34 Wh per litre thermoneutral figure". Under that metric the claims also fall below the Gibbs 1.229 V floor, which is a stronger result than "above the baseline".
3. claims.md claim 3: Kelly's COP for Annis/Eberly "200 to 400" is output over the switch drive power only, not over total input. The summary should say so; the standard cycle argument makes this definition the point.
4. claims.md claim 6 lists Moray "635-675 W of lamps and appliances in witnessed demonstrations". The 635 W figure conflicts with its own conversion (0.878 hp = 655 W); say "635 or 655 W". The 675 W is a rating sum (100 W + 575 W), and Hayes states he could not measure current or voltage [read: p.604].
5. facts.json and claims.md describe "Patrick J. Kelly" without noting the pen name (Investigator correction, adopted) and that V32.9 is not the final edition.
6. claims.md cross-cutting facts: "Sweet input 0.31 mW / 1 mW / 10 mW and output 500 W / 1 kW / 5 kW" are now traced: 0.31 mW and 500 W come from the 1987 video; 1 mW from the 1987 letter; 10 mW and 5 kW from Sweet's 1988 paper.
7. Deep read, Ainslie (T2): the claim that "NE555 data sheets give a maximum frequency of about 500 kHz" is [memory, medium] and unverified; not used in my verdicts.
8. Coverage caveat: the book digests (Chapters 1-2, 14, 20; Appendix pp.1901-2238, 2253-2328, 2433-2481) were not deep-read, so Heins, Flynn, Tseung frame and Asaoka claims in claim 3 rest on digest-level text, and my verdicts on them are weaker than on Sweet, Benitez or the electrolysis figures.
9. No deep read corrections were needed to the Moray "germanium diode" reading beyond the one already recorded in claims.md.

Summary of verdicts