RankingsPrimerArbiterPhysicistMathematicianContrarianQESites logClaimsSourcesResearch notes

Meyers-Stan (Stanley A. Meyer, "Water Fuel Cell"): Science Contrarian review (Step 2)

Reviewer role: Science Contrarian (judges by experimental evidence; strongest honest case for the claims; same rigour on the inventor and the critics). Date 2026-10-09. Tags: [read] project document or saved page; [read: p.N] page; [calc] my arithmetic (constants: Faraday 96,485 C/mol; molar gas volume 24.5 L at 25 C; water HHV 285.8 kJ/mol, LHV 241.8 kJ/mol; 1 oil barrel about 6.1 GJ); [web] URL; [memory] with confidence.

Source gap (repeat from claims.md)

  • Every Meyer document in the folder (patents, newsletters, the 222-page "Birth of New Technology" brief) is Meyer's own characterisation. He is the patent holder and sold dealerships and "profit-sharing certificates". The folder holds no measurement report, no lab report, no independent document. [read: claims.md, facts.json]
  • The brief was deep-read only to p.56 in Step 1. I read Memo 424 (pp.117-124 of the PDF) and the Memo 425/426 passages that carry numbers myself. [read]
  • Not read by anyone: the 2 h 38 min New Zealand 1989 video (transcribed in the QE round; whisper transcript and deep read now exist) [corrected 2026-10-09 after QE: was "not transcribed"], the 40 photographs, Puharich's 344-page book beyond selected pages. The Einsiedeln 1989 talk (57 min) was deep-read from a whisper transcript. NIST/OERI evaluation, the UK Queen Mary report, the Fayette County decision transcript, the 1999 estate valuation and the USPTO file histories are not obtained.
  • The strongest quantitative evidence in the Investigator's finds comes from the transcribed 1995 binder at stanslegacy.com, an advocacy archive (machine transcription of scans). Where my conclusion rests on those transcriptions I say so.

Method note and one standing caution

I did the energy arithmetic myself because no document in the folder does it. All of it uses accepted laws and one question: how much gas, how much energy does it hold, how much went in. The gas energy content of water splitting is fixed: about 286 kJ per mole of water (HHV), which is 11.4 mol and 3.27 MJ for each 420 L of mixed H2+O2 [calc]. Faraday's law fixes the gas per unit of net electric charge: 1 A for 1 h releases 0.456 L H2 plus 0.228 L O2 = 0.685 L of mixed gas (25 C) [calc]. These two numbers are the yardsticks for everything below.

Claim-by-claim

Claim 1. Water as the dielectric of a resonant, pulsed high-voltage capacitor releases H2 and O2 without electrolysis and at minimal current

a. Is the mathematics internally consistent? Partly. The series-resonance formulas (f = 1/(2 pi sqrt(LC)), V_C = Q x V_in at resonance) are textbook and correctly quoted [read: Birth pp.8-9]. Three problems. (i) Eq. 9 adds the dielectric constant (dimensionless, 78.54) to ohmic impedances: a units error [read: Birth p.10 page image, per Step 1]. (ii) The "dielectric" picture needs the water to behave as an insulator at the drive frequency. For a conductor the relaxation time is tau = eps0 x eps_r / sigma. Tap water (sigma about 0.05 S/m, [memory] medium) gives tau about 14 ns, so any drive below about 10 MHz sees tap water as a resistor, not a capacitor. Only very pure water (sigma about 5.5e-6 S/m [memory] high) has a crossover near 1.3 kHz [calc]. Meyer says the process works "in all forms of natural water" and in distilled water [read: Einsiedeln L.188-189], while Hindley says it works best with pure water [read: Electronics World p.10]. The two statements cannot both be a pure-dielectric effect. (iii) 78.54 is quoted at 25 C on p.7 and at 20 C on p.52; the value at 20 C is about 80 [memory, high]. The brief gives no L, C, turns, gap, frequency or gas rate, so the circuit is not checkable [read: Step 1 inventory]. The Electronics World description is more concrete than Meyer's: a charge-pump with a diode, a rising "staircase DC potential" until "the water breaks down and a momentary high current flows", with a sensing circuit that removes drive for a few cycles [read: Electronics World p.10-11]. That is a repetitive breakdown (spark or streamer) regime, not a "no current" regime. Meyer's own circuit (blocking diode, charging choke, cell, gated bursts) is consistent with that reading.

b. Consistent with established experiments? The weak form (pulsed high voltage across plates in water produces H2 and O2) is established; it is electrolysis plus possible plasma or breakdown chemistry. The strong form (bond "switched off" by field with no charge transfer, energy cost below electrolysis) conflicts with what is known: field strengths needed to polarise a water molecule appreciably are of order 1e9 V/m [memory, medium]; the fields here are 1e6-1e7 V/m [calc: 1-20 kV over 1.5 mm gives 0.67-13 MV/m]. A different, mainstream-compatible way to get gas at low average current is to charge the capacitor in bursts and discharge it (Faraday applies to net charge, and a DC meter on a pulsed supply does not measure it). Relevant mainstream work: Puteanus et al. (IJHE 2025) prove, for a standard PEM cell model, that pulsed electrolysis cannot beat steady DC at equal average current [web: https://fis.tu-dresden.de/portal/en/publications/efficiency-improvement-by-pulsed-water-electrolysis-an-unjustified-hope(4a344c62-7950-4991-bd5a-7b5ef73fd41c).html]; a 2026 IJHE paper (Ghosh and Haverkort) summarises Shimizu et al. as finding DC better in nearly all conditions, with one small unverified gain for a 7.9 V, 1.2 A, 17 kHz inductive pulse [web: same search result set, https://repository.tudelft.nl/file/File_93f6fb16-b9fc-48f3-bc54-ac1a6e909d54?preview=1]. Other positive reports exist (a 2018 IJHE alkaline-cell paper reports 20-25 percent lower energy per mole of H2 with pulsed potential [web, via search summary only, not read]). None is a sub-thermodynamic claim; the claimed gains are over a poor DC baseline. So the mainstream literature does not exclude "pulsing helps a bad cell" but excludes "beats the enthalpy of water".

c. Tested? By whom? Replicated? This is the heart of the file. What exists, all with a stake or by sympathetic parties: 1. Hindley/Laughton/Griffin, Oct 1990, Electronics World Jan 1991 (eyewitness; Griffin is a promoter, Hindley wrote in favour). Facts: DC ammeter read "no greater than a tenth of an amp" while "claimed to be milliamps by Meyer"; "DC voltage appeared to be pulsed at tens of thousands of volts" (an appearance, no instrument named); cell and pipework cold after 20 min; hydrogen flame melted a steel bar; gas "poured off" at peak voltage; gas stopped and restarted instantly when drive was switched. The editor's own comment: "the experimental technique has more holes in it than a sieve. First among these is: what is the precise total cell energy input required to deliver a unit volume of dry gas mixture?" [read: research/files/Electronics-World-1991-01.pdf, pp.10-12; text at /tmp extraction]. No gas volume, no input power. A cold cell at a few watts is expected of any small electrolyser; "cold" is not evidence of efficiency. 2. Nagypal, Vienna, 4 Jun 1991 [read: stanslegacy transcription of binder pp.66-68]. Plane plates 3 x 2 in, 1.6 mm gap, 12 V to 600 V bursts, "consumes about 20 mA", 0.5-1 cc of gas per minute, "very little needle flame", he "can't reproduce" Meyer's waveforms, "it works in another way", but "your theory works", and "relative effectivity is about 300%" against his own method needing 10-12 W for the same gas; he offers himself as distributor (stake). My arithmetic: Faraday for 20 mA gives 0.23 mL/min [calc], so his 0.5-1 cc/min is 2-4 times the Faraday amount, and his "300%" matches that ratio. That is the single numerical hint of gas beyond the charge passed in the whole file. It is weakened by (i) 20 mA is the supply figure, with no waveform-aware measurement of net cell charge; (ii) gas volume read by eye at about 1 cc, with bubbles "50 percent remaining on plates" [read]; (iii) the minimum energy to make 0.5-1 cc/min is only 0.065-0.13 W [calc], so the comparison "10-12 W for the same gas" is against a baseline about 100 times worse than any decent electrolyser; (iv) if 20 mA flows at 12 V the input is 0.24 W, an efficiency of 27-54 percent of HHV, ordinary. 3. Perkins Technology bench observation, 8 Oct 1992 [read: stanslegacy transcription]: dissociation from about 10 V, "maximum observed current less than 1 milliamp", vessel at room temperature, a lit jet melted stainless wire. If a sustained flame (say 0.5 mL/s) came from a net cell current below 1 mA, Faraday's law would be violated by a factor of 2,600 or more (0.5 mL/s needs about 2.6 A of net charge [calc]). This is the best honest case for an anomaly: a sustained flame at a milliampere reading. It is also where the weakness is: a "flame melting stainless wire" needs only a thin wire and a small flow, no flow rate or duration is given, and the milliamp reading comes from a meter on a pulsed supply with the measurement point unstated. No one measured gas volume against net charge and input power at the same time. 4. Meyer's own table, binder pp.59-60 [read: stanslegacy transcription]: tubular array, "12.5 VDC @ 40 amps = 500 watts", "1 liter cavity @ 7 lbs / min", "1 liter cavity = 1000 cc of gas volume = 1 lb gas pressure", so 7,000 cc/min or 420,000 cc/h, set against "prior art electrolysis: 10 cc gas per amp hour at 2 V". Reading these figures with accepted physics: (i) the electrolysis baseline is wrong by a factor of about 68: Faraday gives 685 cc per amp-hour, not 10 [calc]. (ii) "1 lb pressure = 1,000 cc" is not a unit conversion: 7 psi (0.476 atm) added to a 1 L closed volume is 0.476 L of gas at 1 atm per minute, so 7 L/min is overstated by about 14.7 times [calc]. (iii) With that correction the rate is 28.6 L/h, and Faraday for 40 A is 27.4 L/h [calc]: Meyer's own 500 W test is consistent with ordinary electrolysis at 40 A, not with a new process. If instead his 420 L/h is accepted, the gas holds 3.27 MJ/h against 1.8 MJ/h in (12.5 V x 40 A), apparent efficiency 182 percent [calc]; no mainstream mechanism gives that, and it is also 15 times Faraday for the measured 40 A. The 40 A was read with a clamp-on DC ammeter, which on a pulsed waveform is itself uncertain. I flag that the transcription is a machine OCR of a scan and "lbs" may have been misread; the scan is the record. 5. Replications: Lawton 2006 (Wales), six pairs of 316L tubes, "about one amp" for one tube, under 2 A for three, with no gas rate [read: vixra PDF pp.224-233, per Investigator]; builders and forum reports (Cramton, Raju, Kelly guide) give currents of amps, i.e. tens of watts, and no gas rates. These are not controlled tests. I count no independent replication of any efficiency claim. I did not find a negative controlled replication either: the experiment has not been done properly in either direction. 6. Court: Fayette County CVH19930292, Willis and Harper v. Meyer, judgment 24 Sep 1996 for the plaintiffs, "gross and egregious fraud" per the Columbus Dispatch 2007 and the Sunday Times via Wikipedia. Meyer's own account (interested party): Judge Corzine ruled that WFC "did not have an amperage inhibiting circuit, did not have a Voltage Intensifier Circuit (VIC) and therefore the WFC water splitting process was only electrolysis"; his expert witness (Mathias Johanson) was not called; the court audio was switched off during the VIC demonstration [read: Public Notice, per Investigator]. The Sunday Times (via Fusion Facts, 1996, not seen in the original) says three court experts found "nothing revolutionary about it", conventional electrolysis. The decision transcript (filed 11 Sep 1996) is not obtained; exhibits and depositions were destroyed 18 Jun 2014 [read: docket transcription]. A fraud judgment on a dealership contract is not a physics test, but if the court's experts examined the demonstration device (an inspection of the demonstration units occurred 18 Oct 1995 per the Public Notice) their finding is a test, albeit one whose method I cannot see.

d. If true, what would change? Electrochemistry would need a field-driven, charge-free bond-breaking channel with energy cost below the water-splitting free energy (237 kJ/mol Gibbs at 25 C [memory, high]); the second law caps any electrical route at 120 percent of the enthalpy-referred input, only when heat is absorbed from the surroundings [memory, high]. Claim 1 as a fixed-energy-budget statement ("less energy than electrolysis") would not by itself break thermodynamics provided the cell is above electrolytic efficiency. It would break it if the gas energy exceeded the 3.27 MJ per 420 L figure relative to total input (including all pulse-circuit losses).

Verdict (split). - Weak form (pulsed high voltage across water in a cell yields H2 and O2): Established (as electrolysis and breakdown chemistry). - Strong form (non-electrolytic, field-only splitting; gas output not tied to net charge; or energy cost below electrolysis): Never properly tested. No neutral, instrumented test with simultaneous input energy, net charge and gas volume is on record. It was not dismissed without testing: the court and (per the Perkins note) UK experts did look, but their reports are not obtained. Meyer's only numerical efficiency evidence (binder pp.59-60) contains arithmetic and unit errors and, corrected, matches ordinary electrolysis. The anomalies worth a proper test are the Nagypal and Perkins current/gas observations.

Claim 2. "Hydrogen fracturing": thermal explosive energy beyond combustion; up to 44,000-2.5 million barrels of oil per gallon of water

a. Mathematics. No derivation exists. Memo 424 "Atomic Energy Balance of Water" (the memo whose title promises the balance) contains no calculation at all [read: Birth pp.117-124]. It is a qualitative narrative of an "Energy Aperture" in each atom that, driven by pulses, admits "Universal Energy" into the nucleus, with a "one-way energy valve", "gyroscopic action of nucleus particles" and a recycling story in which sunlight re-energises the exhaust water. The only equations there are E = h nu and p = h nu / c, with Planck's constant printed as 6.547e-27 erg s (accepted 6.626e-27) [read: p.123; value from memory, high]. The cover equation E = m c^2 is not applied to any mass. The claim is also self-contradictory: Meyer says the process "complies with the Laws of Thermodynamics" (1996) while tapping zero-point energy; "does not create energy" yet releases "atomic energy"; he calls the same effect nuclear ("nuclei decay", Fig. 1-8) and chemical (gas combustion) in different places [read: Birth p.17, p.22].

b. Experiments. None in the folder. The claimed yield: 1 US gallon of water as H2 burned back to water gives 60 MJ, or 0.0098 barrel of oil [calc]. 2.5 million barrels is 1.5e16 J, about 2.5e8 times the chemical value and about 4.5 percent of the total rest-mass energy of the water (3.4e17 J) [calc]. Total conversion of water to energy would leave no matter; partial release would need nuclear reactions, and the combustion products are water, with no radiation, no helium, no neutrons reported by anyone. Meyer's own injector figures contradict the headline: he says 7.4 microlitres of water per injection runs a 50 hp VW at 65 mph [read: Birth p.155 (PDF p.155 text, Memo 425)]. At 2.5 million barrels per gallon, 7.4 microlitres would release 3e10 J per injection [calc], more than the whole VW would use in a year. The injector figure is instead at the chemical scale: 7.4 mg of water contains 99 J (LHV) to 117 J (HHV) as H2/O2 [calc]. A 4-cylinder four-stroke engine at 3,000 rpm has 100 power strokes per second, so 15-20 kW road load needs 150-200 J mechanical per stroke, 500-670 J of fuel at 30 percent efficiency [calc]. So even Meyer's own injector number is 3-7 times short of the chemical energy needed at 100 percent conversion, and about 5-7 times short on the more realistic basis. The claim cannot be saved by saying the 7.4 microlitre figure is idealised: the two figures differ by eight orders of magnitude and the document never reconciles them.

c. Tested? No test of the energy-beyond-combustion claim is on record, by Meyer or anyone. The NIST Office of Energy-Related Inventions evaluation reported in Spy (Dec 1991) as "a net consumer of energy, not an energy producer", and Heard (2000) citing NIST's George Lewett (28 Jun 1991), is the nearest to a test; I could not retrieve it and a web search found no NIST document on Meyer (searched: Stanley Meyer water fuel cell NIST Office of Energy-Related Inventions "net consumer of energy"; hits were the program's general reports, nistir4533 etc., not Meyer) [web: https://nvlpubs.nist.gov/nistpubs/Legacy/IR/nistir4533.pdf]. I therefore carry it as a secondhand report of a government evaluation, with the agency's own characterisation unverified. Hindley and Griffin also cite Griffin's 1993 note that he personally never got more than one third of electrical energy back from Cornish's cell, "about the norm for conventional electrolysis", while he accepts Meyer's [read: Griffin, binder pp.113-128]. Griffin's paper also says electrolysis needs "roughly 3 times the energy of the hydrogen released", which is wrong for practical alkaline or PEM electrolysers, typically 60-80 percent efficient [memory, high], and invokes zero-point energy via Wheeler and Prigogine. Kucherov US 5,632,870 (1997) cites Meyer in an electrolysis cell claimed to draw ambient heat; that is a heat-absorption route capped by the 120 percent limit, not an energy-beyond-combustion claim.

d. If true? Energy conservation (a new, usable energy reservoir tapped by 12 V pulses); for the nuclear form, an explanation for products and radiation; quantum field theory would need a mechanism for extracting vacuum energy that every Casimir and Lamb-shift calculation says is non-extractable ground state [memory, high].

Verdict. The headline yield (44,000 to 2.5 million barrels per gallon): Refuted by calculation (the calculation above, plus Meyer's own 7.4 microlitre figure showing the chemical scale). The "complies with thermodynamics yet taps zero-point energy" statement: Internally inconsistent and Not testable as stated. The weaker claim that burning the gas is "hotter" than ordinary hydrogen flame (beyond 5,000 F): Never properly tested; stoichiometric H2/O2 flames reach about 2,800 C (5,000 F) in any case [memory, high], so this part is not anomalous.

Claim 3. Water-fuel injection; vehicles that run on water

a. Mathematics. None beyond the injector droplet figures: 7.4 microlitres (50 hp VW) and 48.1 microlitres (325 hp diesel) per injection [read: Birth Memo 425]. Test: 325 hp is 242 kW; at, say, 25 injections per second (6 cylinders, 1,500 rpm) that is 9.7 kJ per injection; 48 mg of water holds 0.8 kJ of H2/O2 (LHV) [calc]. Same shortfall (about 12 times at 100 percent). The VW case: see claim 2. Griffin repeats Meyer's "25 miles per litre of water at 65 mph" [read: Griffin p. of binder 113-128]. One litre of water holds 13.4 MJ (LHV) or 15.9 MJ (HHV) as hydrogen; 25 miles (40 km) at 65 mph (29 m/s) lasts 1,385 s; at 8-15 kW road load (my assumption, [memory] medium) the work is 11-21 MJ [calc]. So the stated mileage needs 0.8 to 1.5 times the total chemical energy of the water at 100 percent engine efficiency, and 3-5 times that at realistic efficiency, before counting the electric power for the pulse circuit.

b. Consistent with experiments? The same figures show that with no outside energy, the car cannot run on hydrogen from its water. A car can of course run on a battery and a fuel. The experimental question is whether the vehicle's net energy came only from water.

c. Tested? Never independently tested. Evidence: (i) ABC television news and Columbus TV reports of the buggy, 1985-89 (Meyer says three hours of ABC footage; about 9 minutes survive per Lammertink [web: http://www.tuks.nl/wiki/index.php/Main/StanMeyersSecret]); (ii) Perkins note, 28 Feb 1993: "On none of the 3 occasions, October 1990, August 1992 and October 1992, when UK experts visited Ohio, was there a practical demonstration of a running internal combustion engine fuelled by water" [read: binder, Perkins note, via stanslegacy]; Perkins' Project TACO (VW engine, then the Thames ferry Brunel's diesel, UK demonstration planned for July 1993, deposit $125k) stopped in the record at that meeting: no document shows it was signed, run or abandoned (absence of evidence, see below); (iii) the 1994 newsletter says the buggy was still being retrofitted "to match the acceleration performance of gasoline-driven cars" [read: facts.json]; (iv) Charles Hughes told the Columbus Dispatch (2007) that he saw Meyer "power his tractor for 15 minutes on well water", "no fumes whatsoever" (one witness, recalled 20 years later, no measurement, but a first-hand account that deserves a place in the record); (v) the Sunday Times (1996, via Wikipedia and the stanslegacy summary) says Professor Laughton, as court expert, was refused a test of the car and was given a "lame excuse"; Meyer replies the visit was in 1990 and the injector was then "under US national security review" [read: Fusion Facts item and Public Notice, as summarised by stanslegacy]. A 1990 refusal is consistent with the Perkins note (no engine run on any visit). The "national security review" rests on two US foreign-filing licences cited on Birth p.1 and p.18 (492680, 10 Jul 1989; 490606, 15 Nov 1988/1989 — the two pages print different years; the 1989 is on p.1 of the viXra 233-page copy, 1988 on p.18 of the folder copy) [corrected 2026-10-09 after QE: was 1989 attributed to p.1 of the folder copy]. Such licences allow filing abroad and are routine [memory, high]; they are not evidence that a technology was classified. Where a stakeholder (Meyer) explains a refusal by security, I note it as his explanation and observe that the same man sold dealerships on a vehicle that three UK delegations never saw run.

d. If true? The energy budget above: a vehicle with no fuel input other than water would need an energy source that is not in physics, or a hidden storage (the 12 V battery, which an alternator recharges from the engine, nets zero or less).

Verdict (split). Strong form (road vehicle runs on water alone, no other energy): Refuted by calculation using Meyer's own figures (7.4 microlitres; 25 miles per litre), unless a source beyond water is supplied. The specific buggy: Never properly tested: no neutral, instrumented run (fuel/battery accounting) is on record and it was neither tested and failed nor clearly dismissed untested; UK experts say they were not shown a running water-fuelled engine.

Claim 4. Electrical Particle Generator ("gas battery"): closed loop of magnetised particles; voltage amplification

a. Mathematics. None (no voltages, currents, turns, sizes, velocities) [read: deep read of CA 1,213,671]. The patent claims voltage and current gain from a low-voltage primary and a higher-turn secondary; it never writes "power" or "efficiency", so no output-exceeds-input is stated; the reader supplies it. A coil with more secondary turns gives voltage gain in any transformer, with current falling in proportion. An output of both more voltage and more current than in is the over-unity reading.

b. Experiments. A gas has no permanent magnetisation to "magnetise individually": O2 is paramagnetic with a small susceptibility, and thermal motion randomises moments [memory, high]; ferromagnetic powders or vapour from a spark gap (claim 16) could carry moments and induce a voltage as they pass a coil, which is real induction but not net generation. US 4,613,304 (the "gas electrical hydrogen generator") claims this combination, with the plate cell on plain DC [read: deep read].

c. Tested? No test or measurement anywhere in the folder; newsletters call it a "gas battery" that "produces electric energy directly from the fuel cell gases without chemical interaction" [read]. Never built or measured in any source I have.

d. If true? Power gain from a closed circulation loop would need an energy source; the ordinary reading is a motionless transformer analogue.

Verdict. Not testable as stated (no quantities); the over-unity reading would be Refuted by calculation if any output power were quoted, but none is. Voltage gain alone: Established (transformer action).

Claim 5. Resonance in the electrode gap (1982 form: EP 0 103 656, CA 1,234,773)

a. Mathematics. No equation links gap to frequency [read: deep read]. The mechanism is that molecules "bombard back and forth" at the cavity's wavelength, velocity "increasing geometrically". Check: a hydrated proton in water has mobility of about 3.6e-7 m^2/(V s) [memory, medium]; at 1 kV over a 1.5 mm gap (E = 0.67 MV/m) the drift speed is 0.24 m/s and the transit time 6 ms, i.e. about 160 Hz; at 20 kV, 0.3 ms and 3 kHz [calc]. So an ion-transit "resonance" falls in the same region as the frequencies Meyer's patents use (kHz), which is a point in favour of a physical picture, though ion motion in a dissipative medium is drift, not resonance, and tap water's relaxation time (14 ns, above) is far shorter.

b. Experiments. The Nagypal cell (about 10 nF with 0.8 H chokes: resonance f = 1/(2 pi sqrt(0.8 H x 10 nF)) = 1.8 kHz [calc]) matches the 500-16,000 Hz burst frequencies he reports, so a series LC at kHz is how such a cell is built; that it "works" at those values says nothing about molecular resonance.

c. Tested? Not tested as a molecular-resonance claim. Puharich's patent claims molecular resonance for splitting (see claim 6).

d. If true? A narrow resonance of liquid-water molecules would be a rotational or librational line (microwave to far-infrared, about 10 GHz-10 THz [memory, high]), not a kHz one.

Verdict. Not testable as stated (no dimensions, no frequency, no data). Never properly tested as a physical mechanism.

Claim 6. Puharich's prior claim (US 4,394,230; ELF book): resonant splitting at 80-100 percent efficiency, 114.9 percent energy return in 0.9 percent saline at 100 mW

a. Mathematics. The accounting is internally consistent: 100 mW for 30 min = 180 J; 10.8 cc H2 + 5.4 cc O2 = 16.2 cc; HHV 12.77 J/cc gives 206.9 J; ratio 1.149 [read: deep read pp.18-20]. The weakness: input is the product of rms V and rms I (25-38 mA, 4-2.6 V) for a modulated waveform with a stated reactive component, so real power and net charge are not given [read]. Check by Faraday: 10.8 cc H2 at 289 K is 4.55e-4 mol, needing 87.9 C, a mean net current of 49 mA over 1,800 s [calc]. The reported 25-38 mA rms is lower, but rms of an AC-modulated waveform is not the net DC the electrolysis needs, so Faraday is neither violated nor satisfied by the data given. The 120 percent figure he cites (dH/dG) is the real thermodynamic limit for reversible electrolysis with heat taken from the surroundings [memory, high], so 114.9 percent is not over-unity per se, but a cell at 2.6-4 V is far above the 1.23-1.48 V thermoneutral range and cannot be that close to reversible with real electrodes, unless the energy accounting (power, gas measured by chromatograph in "two laboratories", unnamed) is wrong.

b-c. Gas analysis by GC/MS "two laboratories", error plus or minus 10 percent, labs not named [read]. No replication on record. The patent cites no Meyer work; Meyer's US 4,798,661 cites it [read].

d. If true? Same as claim 1 strong form, with water splitting on a path below the Gibbs free energy.

Verdict. Never properly tested (not tested independently, no calorimetry, no net-charge measurement). The 114.9 percent arithmetic in itself: not over-unity if heat is absorbed, so Plausible, untested as a thermodynamic statement; its mechanism (tetrahedral water at 109 deg 28 min "shattering" at harmonics): Not testable as stated.

Meyer's outside-confirmation and finance claims (not numbered claims, but weighed)

  • "Labs around the world confirm": the only third-party observations found are Hindley/Griffin/Laughton 1990, Perkins 1992, Nagypal 1991 and Sussex "breadboard produced hydrogen" (from the Perkins note; the underlying report was not found). None reports input energy against gas energy; two of the four sources (Griffin, Nagypal) wanted to become distributors or partners. The statement "Nagypal: your theory works" is true to the letter but sits beside "can't reproduce your waveforms" and "it works in another way".
  • Hindley's argument that US Section 101 patents are "independent, critical, scientific and engineering confirmation" [read: EW p.11] is Hindley's characterisation. The patent text contains no measurements, and US examiners do not routinely test inventions [memory, medium; not verified by web]. A real Patent Office "demonstration" would appear in the file wrappers of US 4,936,961 and 5,149,407, which are not obtained and are the cheapest decisive record in this review (USPTO Patent Center).
  • Hans Nieper, 14 Dec 1996, "the tested overunity effect"; the 1997 Fowler fax calls it "scientifically verified by the German Association of Vacuum Field Energy" [read: Fowler memo via stanslegacy]. That is a letter, from a WFC-aligned party, not a test report.
  • The Einsiedeln talk (1989) gives one quantitative claim: "once you hit resonance we can excite it for 5 seconds and you are producing gas for 94 seconds" [read: deep read, L.195-]. Gas continuing 19 times as long as the drive ran contradicts the Electronics World witnesses ("gas generation ceased ... instantly as the voltage driving circuit was switched off"). Dissolved-gas supersaturation and bubbles nucleating on plates can explain post-drive bubbling; Meyer's version is not a measurement.

How the claims were rejected: soundness, omissions, suppression

  • The 1996 judgment: a civil suit by two dealers (Forrest Harper and Richard Willis) about fraudulent statements in selling dealerships; $40,515.72 collected [read: docket transcription]. The reported finding "only electrolysis" is Meyer's paraphrase, the "nothing revolutionary" is the Sunday Times paraphrase; neither is the decision. What is known of procedure: trial 1-2 Feb and 8-10 May 1996; Meyer's counsel withdrew (Jul-Aug 1996) and he filed pro se; his new-trial motion with an affidavit by Dr Russell Fowler (a WFC officer) was overruled; exhibits destroyed 2014. Meyer alleges the judge switched off the court recorder (single-source, interested; recorded here as his allegation; no second source found yet) and that his expert was not called. The court record's silence about experts is a gap, not a contradiction. What would decide: the Final Judgment Entry (Min 116/166-168) and the 11 Sep 1996 decision transcript. Both should survive in the court journal.
  • Laughton: a favourable observer in 1990; in 1992 the Perkins file lists "13.4.92 Letter Bramble to AG covering Laughton Report on SM's patents" (Annex J1) [read: Perkins visit, binder p.105], and the Perkins note of 28 Feb 1993 cites a DTI-sponsored report by Queen Mary and Westfield College (Laughton's college), "early 1992", whose experts "were unable to reconcile Meyer's WFC descriptions ... with currently established scientific doctrine and literature. However they left open the possibility that WFC might prove to be an altogether new development" [read]. It is reasonable to read these as the same report (my inference: same college, same date range, same subject; not confirmed). The Sunday Times then names Laughton as the 1996 expert witness. If so, a man who observed a demonstration in 1990, wrote a report on the patents in 1992 and acted as an expert in 1996 changed his assessment, and the intermediate reports (DTI, Laughton) are not published: that is the most important anomaly for the history, and it can be settled by the DTI report and the court's decision.
  • Not found, but would be expected: any journal paper by Meyer; any measurement paper by Perkins, Sussex or Dublin Institute of Technology; the Phase 1 TACO result. Competing explanations for the silence: commercial failure (Perkins' reservations: "no redress" clause, $125k deposit, UK experts viewing WFC as violating the first and second laws), the proof of concept failing, Meyer's secrecy and security assertions, hostile suppression (Meyer's own account of threats and a takeover attempt, undocumented), or loss of the files. Evidence for each: failure (nothing was demonstrated in the next three years while the buggy was said to be "pre-production"); secrecy (Meyer's confidentiality agreement was still "under discussion"); suppression (Meyer's anecdotes only: buy-out offers of $250 million, $800 million, "a billion"; Stephen Meyer's recollection of "They poisoned me"; coroner ruled cerebral aneurysm, toxicology only lidocaine and phenytoin; the Grove City police file has no taped interviews of the two Belgian investors [read: Columbus Dispatch 2007 notes]). I find no evidence in the folder or the Investigator's finds that a research result was blocked by anyone; the only blocked item is Meyer's own refusal to let experts run the car in 1990-92. The death itself is documented as natural by the coroner; what remains unexplained is why no tape exists of the Belgians' police interviews, an absence worth the police file and coroner's report.
  • After death: patents US 4,936,961, 5,149,407, 5,293,857 assigned by the estate to Marilyn Meyer 22 Mar 2001; fees paid until 2002, then lapsed (2004, 2006); no licensee or buyer found; the 1999 estate valuation ("Water Fuel Cell: Valuation of Sole Proprietorship as of March 21, 1998") is in the probate court file, not online. A technology with $250 million offers on the table, per Meyer's family, went unclaimed after the inventor died: two readings (the offers did not exist or were not serious; the technology could not be reproduced by the buyers or the estate). The valuation would help.

Case for the claims, built honestly

  1. The Perkins and Nagypal numbers (flame at under 1 mA; 2-4 times Faraday) are real observations by non-Meyer engineers who had incentives to report success but also gave their numbers; they have never been repeated under measurement of net charge. A proper test of that one point is cheap.
  2. Several British professionals (an admiral, a dean of engineering, a chemist, a company managing director and an engineering manager) travelled to Ohio, saw gas production on tap water at low meter currents, and declined to dismiss it; Laughton's own DTI-era report reportedly "left open the possibility" of a new development.
  3. Cold operation and low meter current are consistent with a small power input; the real question (input energy per litre of gas) was never answered by anyone, so the claims were neither confirmed nor excluded.
  4. The strongest mainstream-compatible variant: pulsed capacitor charging might beat a poor DC cell, up to the 120 percent reversible limit; this is a real if modest effect worth testing.

Case against, including the inventor

  1. No document in the folder or the Investigator's finds contains a simultaneous measurement of input energy and gas output for a Meyer cell. After 20 years (1975-98) the only numbers (binder pp.59-60) include a baseline error of 68 times and a unit conversion error of about 14.7 times; corrected, they match electrolysis.
  2. Meyer's headline energy figure is incompatible with his own injector figure by eight orders of magnitude, and the 7.4 microlitre figure itself is 3-7 times short of the chemical energy needed.
  3. The brief's mathematics has a units error (Eq. 9), a typeset exponent error (Eq. 14), a wrong Planck constant, and Eq. 17 with ampere-squared as "watts".
  4. His explanation for no car test is security; the Perkins file records that no engine run was ever shown in three UK visits. He sold dealerships and certificates in the same period (the documented fraud finding); his newsletters carry changing patent counts (about 40; 42; 25; 26) and costs.
  5. Witness reports in the file come from parties who were preparing to buy in (Perkins, Griffin), distribute (Nagypal) or whose friend-of-the-family accounts date from 2007.
  6. Hindley's "independent confirmation by patents" is unsupported by the patents' content.

What would settle it

  1. A neutral bench run of a Meyer-type cell (Lawton/Nagypal geometry) with: a true-RMS and DC-coupled current probe at the cell, a power analyser on the supply, a calibrated gas collection (volume per minute), and temperature (calorimetry), run on tap and deionised water; compare gas volume with Faraday's 0.685 L per A h of net charge and the 3.27 MJ per 420 L energy content. A result of more gas than Faraday predicts for net charge, or more gas energy than total input, would be the first real anomaly.
  2. The records: Fayette County decision transcript (11 Sep 1996) and Final Judgment Entry; USPTO file wrappers of US 4,936,961 and 5,149,407; the DTI/Queen Mary report; the Laughton 13 Apr 1992 report; NIST OERI evaluation (Lewett 28 Jun 1991); the probate valuation; the Grove City police file and coroner's report; the Perkins TACO outcome (Peterborough, Livett Launches, Brunel).

Corrections to Step 1

  1. claims.md claim 1 and the Nagypal quote: the Investigator's correction stands, and I add the Faraday comparison. Nagypal's gas rate (0.5-1 cc/min at 20 mA) is 2-4 times the Faraday value for his stated current (0.23 mL/min), which is what his "300%" appears to express. The quote "your theory works" in the newsletter omits "can't reproduce" and the distributorship offer.
  2. facts.json / claims.md treat the energy claims as "no measurement, derivation or experiment" for claim 2; it is correct, and I add that Memo 424 itself (not read in Step 1) contains no calculation, and that Meyer's p.60 "test" does exist in the 1995 binder (not in the folder), with the errors listed above. The "Requested deep reads" table in claims.md lists Memo 424; that read is now done here (qualitative only).
  3. Step 1 claim 1 numbers: "12 V x 30 = 360 V" and similar were treated as data; the Electronics World description (diode charge pump, breakdown detection, momentary high current) shows the real regime is repetitive breakdown rather than a "no current" dielectric charging. Claim 1's description should say so.
  4. The Sunday Times/Fusion Facts "guilty of fraud" wording: the archive's own commentary (and the docket) shows a civil judgment, Fayette County, not Franklin; the Investigator's correction stands. Also the Dispatch's "Franklin County 1994 settlement" refers to the Brooks case (93CV005477), a different case.
  5. Investigator lead list: the "UK DTI-sponsored report by Queen Mary and Westfield College" and "Laughton Report on SM's patents" (Perkins Annex J1, 13 Apr 1992) are probably one document; the Investigator lists them as separate leads. Not confirmed.
  6. Birth p.18 (folder copy) prints "foreign grant license #490606 issued Nov. 15, 1988"; p.1 of the viXra copy says 1989 [corrected 2026-10-09 after QE: was p.1 of the folder copy] The patents' dates should be checked: the Step 1 fact "Nov 1989" depends on the OCR of p.1.
  7. The "dielectric constant of about 5" printed in Electronics World (p.10) is the magazine's error (water is about 78-80 [memory, high]); not Meyer's.
  8. Griffin's statement that electrolysis needs "roughly 3 times the energy of the hydrogen released" is wrong for practical electrolysers (60-80 percent efficient) and should not be repeated as mainstream fact.

Web verification (budget about 6; used 2)

  • Puteanus (IJHE 2025) and the Ghosh/Haverkort summary of Shimizu: confirmed by search results [web: fis.tu-dresden.de link and TU Delft repository link above]; I did not read the PDFs myself beyond the Investigator's saved copy of Puteanus.
  • NIST OERI "net consumer of energy": check failed; no NIST page on Meyer found [web search above]. The statement stays a secondhand report (Spy 1991; Heard 2000, via the Investigator).
  • Dielectric constant 78.54 at 25 C, water HHV/LHV, Faraday constant: [memory, high], not rechecked.
  • Proton mobility 3.6e-7 m^2/(V s), tap-water conductivity, 1e9 V/m polarisation field: [memory, medium]; not web-checked; the verdicts do not depend on them (they support remarks only).

Verdict summary

  1. Claim 1: weak form Established; strong form Never properly tested (arithmetic in Meyer's own table matches ordinary electrolysis).
  2. Claim 2: Refuted by calculation (2.5 million barrels per gallon); "complies with thermodynamics while tapping ZPE" Internally inconsistent.
  3. Claim 3: strong form Refuted by calculation on Meyer's own figures; the buggy itself Never properly tested.
  4. Claim 4: Not testable as stated (voltage gain alone Established).
  5. Claim 5: Not testable as stated; never tested as a mechanism.
  6. Claim 6 (Puharich): Never properly tested; 114.9 percent is under the reversible 120 percent bound, so not over-unity by itself.