RankingsPrimerArbiterPhysicistMathematicianContrarianQESites logClaimsSourcesResearch notes

Mathematician review (Step 2): Larson-Tyson (Realm Industries)

Reviewer role: PhD mathematician. I check derivations, notation, units and hidden assumptions. Physics-of-nature judgements belong to the physicist; where the mathematics needs a physical number, I say so and tag it. Tags: [read] = from the project documents; [web] = from a URL; [memory] = recalled, with confidence.

Source note: all inventor evidence is the applicants' own characterization in two patent applications (US 2009/0092540 A1, the "bond-angle" parent, and US 2009/0090312 A1, the "engine" CIP, a continuation-in-part meaning it adds new matter to the parent). The assignee, Realm Industries, has a commercial stake. The patents contain almost no equations: one formula (f = 1/(RC)), some parameter ranges, and three exhaust-gas graphs. So the "mathematics" under review is small, and most of my work is checking those, plus building the energy and atom-balance calculations that the patents omit.

Notation used here

  • R = resistance (ohm), C = capacitance (farad), f = frequency (hertz, cycles per second). A time constant tau = RC is in seconds. The quantity 1/(RC) is in radians per second; dividing by 2*pi gives hertz.
  • Dielectric relaxation time of water, tau = rho * eps_r * eps_0: rho = resistivity (ohm-metre), eps_r about 80 = relative permittivity of water, eps_0 = 8.854e-12 F/m.
  • eV = electron-volt, kT = thermal energy (0.0259 eV at 300 K). cm^-1 = wavenumber unit of energy (1 eV = 8065.5 cm^-1) [memory, high].
  • F = 96,485 C/mol (Faraday constant), the charge per mole of electrons. LHV = lower heating value (heat from burning, water left as vapour).

Corrections to Step 1

Step 1 (facts.json, claims.md, deep reads) is largely accurate. I re-read the deep reads and the patent text, and re-did the RC arithmetic. Points to correct or sharpen: 1. RC calculation (deep read, section 6): confirmed (C = 58.8 nF per gap, 1.12 uF per 19-gap chamber; tap water 1/(RC) about 1.4e7 rad/s; pure 18 Mohm-cm water about 7.8e3 rad/s). One refinement: Step 1 quotes "14 MHz" for 1/(RC). That is 1.4e7 rad/s, which is 2.2 MHz in hertz. The patent never says which it means, so both readings are 3 to 4 orders of magnitude above 1-22 kHz for ordinary water. Conclusion unchanged. [read; recomputed in the project venv] 2. claims.md item 4 says gasoline-phase CO2 of 18-19% "exceeds the usual gasoline maximum". Correct, and stronger: the whole gasoline-phase gas set is not atom-balance consistent (section "Claim 4" below). This is a measuring-chain problem in the patent's own graphs, not a minor oddity. 3. Step 1 notes the polarity contradiction between [0019] and claim 2 as a "reader's observation". I confirm it from the verbatim claim text: claim 2 connects the anode to the negative terminal, [0019] to the positive. By standard electrochemical convention the anode is the positive electrode in an electrolytic cell, so claim 2 is mislabelled or reversed. [read: deep read section 7; p.15, p.22] 4. The investigator reports that the CIP was granted as US 7,793,621 B2, so claims.md's remark that no grant was stated is out of date for the CIP. The granted claim 1 requires metering air with a second mass flow controller and mixing vapour, air and a fuel. That is a different (air-breathing) arrangement from Example 2 (air excluded). Granted claims were not independently verified by me beyond the investigator's note. [web: https://patents.google.com/patent/US7793621B2/en, via Investigator] 5. I verified the WO 2010/039283 paragraphs the investigator cites: [0031] "skeptics may find it difficult to believe ... could result in the production of hydrogen gas (H2) and oxygen gas (O2)"; [0032] the product is "referred to as an energy carrier, which includes but is not limited to molecules with unnatural bond angles and H2 and O2 gas"; [0036] "hydrogen and oxygen gases may be generated ... at some resonating frequencies". [web: https://patents.google.com/patent/WO2010039283A2/en] (fetched through a summarising tool, so wording is as returned; the investigator read the page directly.) 6. The tolerance on figures read from page images (about plus or minus 0.3 percentage points) is stated in the CIP deep read; my use of those numbers below stays within it.

The claims, answered a-d

Claim 1: an "RC resonating frequency" f = 1/(RC), combined with the supply frequency, bends the H-O-H angle without breaking bonds

a. Internal consistency of the mathematics: No, in four ways. - Terminology and circuit theory. The patent describes "a fixed capacitor in parallel with a variable resistor, generating an oscillating or resonating frequency" ([0021]). A parallel R-C network has no resonance. It is a first-order system: a step in voltage relaxes with time constant RC and there is a corner (cut-off) frequency 1/(2piRC), but nothing oscillates. Oscillation needs a second energy-storage element (an inductor L, giving f = 1/(2pisqrt(LC))) or active feedback. No inductor and no oscillator is described. [read: p.16, [0021]] - Units. f = 1/(RC) is in rad/s; the factor 2pi to hertz is omitted, and the patent states ranges in kHz. A 2pi mismatch (6.28x) is small next to the next issue but shows the formula is a gesture, not a derivation. - Dependence of C and R. The patent says C depends on "applied current", plate area, plate count and gap. For a fixed geometry, capacitance of the water-filled gap does not depend on current. And for parallel plates, R = rhod/A and C = epsA/d, so RC = rhoeps_reps_0, independent of area, plate count and gap. The patent's list of design levers (area, number, gap) therefore cannot tune the frequency in this formula. Worked numbers (Example 1 geometry: 4 ft^2 = 0.372 m^2, gap 0.1765 in = 4.48 mm): C per gap = 80 x 8.854e-12 x 0.372 / 0.00448 = 58.8 nF. Tap water (rho about 20-100 ohm-m [memory, medium]) gives RC = 1.4e-8 to 7e-8 s, so 1/(RC) = 1.4e7 to 7e7 rad/s (2 to 11 MHz). To land in the claimed 1-22 kHz band the water would need rho of roughly 1e4 to 1e6 ohm-m (computed), that is, ultra-pure water of about 18 Mohm-cm (rho = 1.8e5 ohm-m [memory, high]). But [0017] says the water "need not be purified". [read; recomputed] - The one physically sensible reading would be the electrode double-layer capacitance (the interface layer at each metal/water boundary, typically 10-100 uF per cm^2 [memory, medium]) in series with the bulk water resistance. That gives a very different time constant, but the patent does not say it, gives no number, and my rough estimate for this geometry with tap water lands at tens of hertz, not kHz. It is an estimate resting on recalled values, so I report it as an indication only. What this reading would describe is the impedance of an ordinary electrolysis cell. - Further internal contradictions: electrode polarity ([0019] versus claim 2, above); a wavelength range 0.1-100 cm against "0 to 5 GHz" in [0050] (0.1-100 cm corresponds to 0.3-300 GHz; overlap only 0.3-5 GHz); component numbering slips ([0040]-[0044]). [read: deep read section 3.3] - "Dual frequency": the patent says a combined "modifying frequency" results from the RC frequency and the supply's "intrinsic" frequency, and that this beats "a single frequency stimulus that is equivalent to the sum" ([0023]). Mathematically, two sinusoids at f1 and f2 are a linear superposition; a linear system sees nothing that is not already in f1 and f2. Only a nonlinear element (diode, MOSFET switch, or the electrode reaction itself) produces sum and difference frequencies, and then they are f1 plus-or-minus f2, which the patent does not compute. No formula defines the "modifying frequency". [read: p.16]

b. Consistency with established experiments: - An "applied frequency changes a molecule's bond angle" idea has a real counterpart: infrared radiation excites the bending vibration of water (the nu2 mode at about 1595 cm^-1, 48 THz [memory, high]). That is a frequency of about 4.8e13 Hz, nine orders of magnitude above 22 kHz. A 1-22 kHz electrical oscillation carries photon energy near 1e-10 eV and cannot resonantly excite a molecular vibration. [memory, high; the ratio is arithmetic] - Microwave rotational/dielectric absorption of water peaks near 20 GHz (relaxation time about 8 ps) [memory, high]. The patent's "1.1 GHz, 2.2 GHz" EM option is in the range where water absorbs by dielectric heating, but heating is not bond-angle change, and the patent states no measurement. - At 5 V across a 4.5 mm gap of conducting water, electrolysis occurs: the water decomposition voltage is 1.23 V [memory, high]. The patent's own text says bubbles form on the plates and rise ([0035]-[0036]) and the product ignites and supports combustion. See the energy section below.

c. Tested? No test of the mechanism is on record. The patent reports no measured frequency, current, power, or gas composition [read: deep read section 5]. The only instrument named is an exhaust gas analyzer on engine tailpipes. No replication was found by the Investigator [web: sites.md]. The USPTO's abandonment (13 Sep 2013, failure to respond to an office action) and the EPO's withdrawal show that the examiners' objections were not answered; the objections themselves are unread. [web, via Investigator]

d. If true, what would change: the free-molecule water geometry is fixed by electronic structure (the quantum mechanics of the electrons), and is not a dial that a kHz current turns. A stable bent-to-wider water molecule at 1-100 psi for months would require the potential energy surface of H2O to have a second minimum that standard quantum chemistry and spectroscopy (extremely well tested for water) do not show, and a mechanism for a collisional-relaxation lifetime (picoseconds to nanoseconds in a gas) to become months. See the numbers under Claim 3.

Verdict: Internally inconsistent (the "RC resonance" formula, the stated tuning parameters and the circuit description contradict each other and standard circuit theory), and the mechanism itself is Never properly tested (no measurement of any kind).

Claim 2: the pressurised "vapor with unnatural bond angles" is a distinct substance from steam or electrolysis H2/O2 and can be stored for months at 1-100 psi

a. Internal consistency: Partly. The patent says "bond angle > 104.5 degrees ... may be 180 degrees" ([0056]) yet gives no definition of what is measured. Bond angle for a molecule in a gas or liquid is not a single sharp number: the zero-point bending motion alone spreads the angle by several degrees, and in liquid water the instantaneous angle distribution is broad [memory, medium]. A product described as "water with angle above 104.5" is therefore not a well-defined substance without an angle distribution. The Abstract and [0021] say covalent bonds are not broken; [0036] of the later PCT concedes that they may be (see item 5 above). The patent argues "distinguishable from electrolysis ... electrical property and magnetic property" ([0021]) with no measurement. Water vapour is diamagnetic and has a permanent dipole moment (1.85 D [memory, high]); these are properties of ordinary water vapour, so "electrical and magnetic properties" do not distinguish anything.

b. Consistency with experiments: A molecule with its bend pushed toward linearity is a highly excited vibrational state. The barrier to linearity of H2O is about 11,114 cm^-1 = 1.38 eV = 133 kJ/mol [web: https://www.sciencedirect.com/science/article/abs/pii/S0009261405011899 (Monodromy in the water molecule; value 11,114 plus-or-minus 5 cm^-1 returned by search summary)]. Computation: Boltzmann factor at 300 K, exp(-1.38/0.0259) = exp(-53) = 7e-24. So a thermal population of linear water is nil, and a molecule given that energy by a plate would shed it by collisions in a time of order nanoseconds at 1-7 atm gas density [memory, medium: vibrational relaxation of water bending is fast]. Storage for months is incompatible with that. Also, ordinary water vapour cannot reach 35 psi at room temperature without heating (vapour pressure of water at 25 C is about 0.46 psi [memory, high]), so a vessel pressurised to 35 psi at room temperature holds a non-condensable gas, which is what H2 + O2 would be and steam would not be. [memory, high]

c. Tested? No composition measurement exists. The one physical sample on record, three TyLar gas cylinders bound for South Korea in 2011, was destroyed unsampled by bomb squads on 15 Sep 2011; two of three detonated when shot. The MSDS (safety data sheet) called the material compressed hydrogen; the invoice called it compressed water vapor. These are EPA's and its consultants' characterizations in a hazard response, plus company documents. [web: https://response.epa.gov/sites/7236/files/SSI_polrep_2.htm via Investigator] The later patent family document (WO 2010/039283) has the inventors themselves widening the product to "energy carriers" that include H2 and O2 [web, verified above]. This is the inventors' own commercial hedging, not an admission that the theory is wrong, but it is the applicants' own text.

d. If true: quantum chemistry and spectroscopy of water (rotational-vibrational line lists) and the entire theory of molecular vibrational relaxation would need a metastable, collision-proof, wide-angle state. A nearby real and testable idea (vibrationally excited water, ortho/para spin isomers, clusters) exists but none matches months of storage.

Verdict: split. Strong form (a stable storable wide-angle water molecule with 133 kJ/mol or more excess energy): Refuted by calculation (Boltzmann and relaxation-time arguments above, given accepted spectroscopy; the lifetime step uses a recalled order of magnitude, so I rate this firm on the thermal-population part and as a strong estimate on the lifetime part). Weak form (that the stored gas is something other than ordinary electrolysis gas): Never properly tested (no analysis was done; the single physical sample was destroyed unsampled, and the later patent text concedes H2 + O2 is possible).

Claim 3: igniting the vapour releases usable energy, using less energy than electrolysis, without toxic gases or atmospheric oxygen

a. Internal consistency: No quantities are given, so no equation can be checked, but a bound exists from the claim's own logic. If "reverting the angle" releases energy, the maximum available per molecule is the excitation energy that the bending state holds, at most the barrier-to-linearity figure, 133 kJ/mol (for an angle of 180 degrees; smaller for smaller angles). Burning hydrogen, by contrast, releases 241.8 kJ/mol of H2O (LHV [memory, high]). And energy conservation says relaxation can return no more than was put in. So even in the patent's best case the stored energy cannot exceed the electrical input to the stack; it can be a (very bad) store, not a source. The statement that making it needs less energy than electrolysis is unsupported: a wide-angle excited state costs at least as much electrical energy as its excitation energy per molecule, and the patent's own 5 V cell is a conventional cell.

Worked energy budget (my calculation; plate area is the patent's, current is bounded by the patent's stated power density 10-100 W/ft^2, since no measured current is given). Plate area 4 chambers x 20 plates x 4 ft^2 = 320 ft^2. At 30-50 W/ft^2 that is 9.6-16 kW electrical input. At 5 V that is 1,920-3,200 A. If all this current went into splitting water: H2 = I/(2F) = 5.18e-3 mmol/s per ampere, so 1,920 A gives 13 L/min H2 (plus half as much O2) with chemical energy 2.4 kW(LHV); 3,200 A gives 22 L/min and 4.0 kW(LHV). Cell efficiency on LHV at 5 V = 241.8 kJ / (2 x 96,485 x 5 J) = 25%. [computed in the project venv] So the example generator, run as an ordinary electrolyser, could deliver a chemical-energy flow of 2-4 kW from 10-16 kW of electricity. For reference, a 4.2 L six-cylinder engine at idle (roughly 500 L/min of air through the manifold at 800 rpm and volumetric efficiency of 0.3 [memory, medium assumptions]) needs of order several kW of fuel energy. The numbers are the same order of magnitude, which means the engine demonstrations are quantitatively compatible with ordinary electrolysis gas (HHO, an informal name for the stoichiometric H2/O2 mix) and with no excess-energy source.

b. Consistency with experiments: No energy figure in the documents (no joules, watts, kWh, torque, rpm, load, flame temperature) can be compared with anything. The torch result (cut thin titanium, melt tungsten, melting point 3,422 C [memory, high]) is qualitative; an oxy-hydrogen flame reaches about 2,800 C [memory, medium], and melting a thin tungsten rod in a localized flame is not self-evidently inconsistent with that. This is a point for the physicist, but I note I cannot find it contradicting ordinary H2/O2 combustion.

c. Tested? Only by the inventors (the torch and engine examples). No control run with plain electrolysis gas from the same cell, no run with the "dual frequency" switched off, no energy meter. No independent measurement.

d. If true: the first law of thermodynamics would be bypassed only if out exceeded in. The patent does not claim over-unity explicitly (no output/input figure); "less energy than electrolysis" is a lesser, still unquantified claim. If the claim were "more energy out than electrical energy in" it would require a violation of energy conservation.

Verdict: split. "Releases energy on ignition": Established for the trivial reading, since the gas burned (consistent with H2/O2). "Releases it because a bond angle reverts, with less input than electrolysis": Never properly tested (no energy-in or energy-out figure; no controls); and any reading exceeding the electrical input would be Refuted by calculation (first law). I could not construct any calculation supporting the "less than electrolysis" statement.

Claim 4: the vapour alone runs engines (Honda GX340/GX390 with outside air excluded; Toyota FJ-40 with gasoline shut off, exhaust O2 near 20%, CO/CO2/HC near zero, three runs)

a. Internal consistency: Mixed. Examples are internally consistent in narrative but the data fail two checks. - Atom balance on the gasoline phase. Take the dry exhaust composition as percentages. Nitrogen is the tracer: air is 3.76 mol N2 per mol O2 (79/21 [memory, high]). The oxygen atoms available in the air that produced the exhaust must equal oxygen atoms in O2, CO, CO2 and in the water that is removed before measurement. For Fig. 11 (values read by eye from page images: O2 about 1.5, CO 10.0, CO2 about 11.5, HC about 2, N2 by difference about 75): oxygen in air = 39.9 atoms per 100 mol dry exhaust; oxygen found in O2 + CO + CO2 = 36.0; that leaves 3.9 to form water. Gasoline (H/C about 1.9 [memory, medium]) burning the 23.5 carbon atoms found would form about 22 oxygen atoms' worth of water. The shortfall is roughly 18 atoms out of 40, i.e. the numbers are not consistent with gasoline burning in air unless a very large amount of H2 is in the dry gas (about 35 mol%, which was not measured and is not plausible). For Fig. 12 (O2 1.2, CO 6, CO2 15.5): 41.1 available, 39.4 in the three species, leaving 1.7 where about 20 are needed. For Fig. 13 (e.g. O2 1, CO 10, CO2 15, HC 2): the three species need 42 oxygen atoms while the air supplies 38, a deficit before any water forms at all. [computed from plot values with the stated plus-or-minus 0.3 tolerance; the shortfall is far larger than that tolerance] - Also, at stoichiometric combustion the maximum dry CO2 for gasoline is 15.3% (computed from CH1.9 + 1.475(O2+3.76 N2); [memory, medium] for the H/C), so CO2 of 18-19% (Figs. 12 and 13 peaks) is impossible for dry sampling. The CO trace flat at exactly 10.00% for about 50 samples is consistent with an analyzer at the top of its range (the IMR 2800 range is [memory, low]; not checked), which makes the CO values lower bounds and the instrument saturated. - Meaning: the instrument's gasoline-phase readings do not close an atom balance. Either the analyser is mis-scaled/saturated/cross-sensitive or the readings are mis-transcribed. This weakens, but does not by itself overturn, the vapour-phase "near 20% O2, near zero CO2" readings, which are what clean air looks like and so are insensitive to calibration at the high end. - The vapour phase: O2 is 19.8-20.6%, which is 0.3 to 1 point below ambient 20.9%, and CO2 holds a residue of 0.2-0.6% which vanishes at engine-off. (Step 1 recorded this.) A hydrogen-fueled engine drawing air would add water vapour to the exhaust, diluting O2 by roughly the water mole fraction (a percent or so if the analyser is not drying the gas); this simple dilution is a candidate explanation for the sub-ambient O2. The CO2 residue could be a small gasoline remainder (float-bowl fuel and the line downstream of the shutoff), also a candidate. Both are my suggestions, not data. [inference] - The 11-minute vapour run on the FJ-40: the engine ran on residual gasoline for an unknown time; the patent never says whether the air path was blocked (it was not: Fig. 14 shows air entering the carburetor [read: CIP deep read]). Zero CO2 and CO are what a hydrogen-fueled engine shows, and what H2 in air gives. - Example 2: "only the water vapor with unnatural bond angles and not outside air" ran the Honda engines. This is exactly what a stoichiometric H2 + 1/2 O2 mixture (which contains its own oxidiser) does. It is not a distinguishing observation. [memory, high]

b. Consistency with experiments: Consistent with hydrogen/oxygen combustion in engines (hydrogen engines are well established [memory, high]; HHO added to intake is a long-standing hobby topic). Not consistent with the claim that the engine consumes "no oxygen" in the Toyota runs, since air was available; the exhaust O2 of about 20% shows dilute combustion, which is unremarkable.

c. Tested? Only by the inventors, in three runs reported in the patent, with no named witness, place, load, rpm, flow, or vapour quantity. Independent testing: none found. The authorities and an outside chemist (Dan Raftery, Purdue; quoted to a reporter) characterized the process as electrolysis [web: VC Star 21 Jun 2010 via Investigator].

d. If true: nothing in mainstream physics needs to change for an engine running on H2/O2 electrolysis gas. Only the "unnatural bond angle fuel" interpretation would.

Verdict: split. "An engine ran on the stored gas with gasoline shut off": Plausible, untested (consistent with known physics if the gas is electrolysis gas; no neutral observation on record). "Because of unnatural bond angles / without oxygen or with energy gain": Never properly tested, with the reported gasoline-phase data internally unreliable (atom-balance failure above).

Claim 5: torch cuts thin titanium and nickel-plated steel and melts tungsten; 60:40-70:30 vapour-to-diesel still combusts on compression

a. Internal consistency: Qualitative only. No flame temperature, flow, or tip pressure beyond "5 psi, #00 tip". The diesel statement is "in test experiments" in the PCT version and carries no data. b. Consistency: An oxy-hydrogen torch is a standard device and cuts thin metals; a hydrogen/oxygen mixture compression-ignites at modest compression ratios (the gas mixture has an autoignition temperature near 570 C [memory, medium]), so combustion in a diesel engine is not surprising. Not discriminating. c. Tested: by the inventors, qualitatively. Nothing independent. d. If true: nothing changes under the H2/O2 reading.

Verdict: Plausible, untested (as a statement about what the gas does; it is consistent with ordinary oxy-hydrogen), and Not testable as stated for any bond-angle attribution (no measurable difference from electrolysis gas is specified).

Claim 6: 17 June 2010 explosion killed one person; earlier Dec 2008 incident; authorities describe the work as extracting hydrogen from water

a. Not a mathematical claim. The consistency check I can make: an electrolysis-gas stoichiometric mixture stored at pressure is detonable (H2/O2 flammable range 4-94% [memory, medium]); two of three cylinders detonated when shot [web: EPA POLREP #2 via Investigator]. The authorities' stated mechanism (a pressurised container used "to break hydrogen and oxygen apart", per VC Star) and Cal/OSHA's later finding that H2 and O2 were stored in the same tanks [web: Daily News via Investigator] are mutually consistent and consistent with the patent's electrolysis-like cell. b. Consistent with the known hazard of premixed H2/O2. c. Documented by multiple independent records: Simi Valley Acorn (25 Jun 2010), VC Star (21 Jun 2010), AP/Press-Telegram, EPA POLREP #2, Cal/OSHA via Daily News ($5,685 citations for 2010, appealed; $540,890 penalties in 2012). The victim was co-inventor Tyson Larson, 28. Cause was "accidental but undetermined" on the fire department's ruling then. Cal/OSHA found three containers failed. These are agency and press characterizations; the inventors' side (Stehl's attorney in AP; the company's appeal of the citations) is also on record. [web via Investigator; not re-fetched by me] d. Nothing in physics changes.

Verdict: Established as an event (multiple independent sources); the cause and the identity of the gas are Never properly tested (no sample analysis).

Investigator findings kept in the record (with my mathematical view; no removal on a negative judgement)

  • WO 2010/039283: inventors hedge that the product could be H2 + O2; drop the "safer than electrolysis" sentence; diesel ratio now "test experiments" (verified in part above). Bears on claims 1-3, 5. This is the applicants' own text, made in a commercial filing; it is not an admission the bond-angle theory is wrong but it shows they could not exclude ordinary electrolysis gas.
  • US 7,793,621 B2 granted 14 Sep 2010, lapsed 2018 for fees; granted claim 1 requires air metering and a fuel. A patent grant certifies novelty and utility under US examination standards, not that the explanation is correct; the grant's claims hold no bond-angle measurement. The parent was abandoned 2013 and EP 2209550 withdrawn 2013; examiner reasoning unread (USPTO file wrapper and EPO documents not yet obtained).
  • Family of 15 applications; Chilean divisional in Nov 2011 shows prosecution continued after Larson's death.
  • R2D2 LLC WO 2023/055615 (and H2Plus filings): independent claim of a "structurally altered gas molecule" from water, H-O-H angle 94-104 degrees, stable above 300 psig, made with crossed electric and magnetic fields; the examiner cited Realm's application against it. Note the opposite angle sense from Realm's (below versus above 104.5). Mathematically the same objections apply (no distribution definition, 133 kJ/mol relaxation argument is for the wide-angle side; for angles slightly below 104.5 the energy cost is smaller but the same relaxation argument holds). I could not review R2D2's data here. [web via Investigator]
  • Prosecution: the federal case (NDNY; Rossignol convicted 2014, 20 years; Stehl Alford plea 2 Sep 2014) concerned commercial misrepresentation. The AUSA said the prosecution was "not alleging that this is some kind of fraudulent scientific process" and that the indictment "makes no judgments about the validity of the science" [web: VC Star 21 Jun 2010 via Investigator]. This is the prosecutor's statement about the scope of the charges, not a validation of the science. Timothy A. Larson (investor, not accused) "believed in the science" (prosecutor's characterization). The record of Stehl's pre-2005 process (BGX, Rainbow Technologies) has no independent test in the sources found.
  • Absence of evidence after 2011 (companies dissolved, patents lapsed, principals injured or prosecuted): competing explanations are (a) ordinary electrolysis gas that failed commercially and legally; (b) work moved abroad (the South Korea export) or under another name; (c) the death, injuries and penalties ended it regardless of merit. From my side, (a) fits every number I could compute: the energy budget, the cell, the stoichiometric oxidiser and the clean exhaust. No evidence for (b) beyond the shipment; it is not excluded. What would decide it: the Korean consignee's analysis, the CD Cal exhibits, Cal/OSHA inspection files, USPTO file wrapper.
  • Also noted by the investigator: Stan's Legacy (Meyer water-fuel archive) indexes the Realm patents as prior art for later water-fuel patents. Related hobbyist/commercial "water as fuel" work (Brown's gas/HHO, Meyer) shares the electrolysis-gas reading; the labelling as "Brown's gas" is EPA's characterization.

What single calculation or measurement would settle this

  1. Gas chromatography or mass spectrometry of the stored gas (H2, O2, N2, H2O fractions), which distinguishes "energy carrier" from electrolysis gas in one run. This has never been done on record.
  2. An energy meter on the cell (volts and amps) and a flow meter or gas-volume measurement during a run, giving energy in per standard litre; compare with 2F per mole H2 at 5 V (about 965 kJ per mol of H2, or about 25% LHV efficiency). If the gas volume equals the Faraday prediction for H2 + 1/2 O2, the product is electrolysis gas.
  3. Run the same cell with the "dual frequency" absent (plain DC at the same average current) and compare the output. No such control is reported.
  4. Re-measure the engine exhaust with a calibrated multi-gas analyser (verifying atom balance closure on gasoline first) with the air intake sealed and a hydrogen sensor.

Limitations and source honesty

  • I read the project deep reads in full and the patent text through them; I did not re-render the page images, so the graph readings are the deep readers' by-eye values ([read: page image] per those notes). The atom-balance numbers use those values and the tolerance noted.
  • Web checks done (budget about 6; used 2): WO 2010/039283 paragraphs (verified); water barrier to linearity 11,114 cm^-1 (verified through a search summary, source named above). Not checked and left as [memory]: IMR 2800A CO range (low confidence, not load-bearing for any verdict); the vibrational relaxation time order (medium; the Refuted-by-calculation verdict on the strong form rests mainly on the thermal-population and energy-conservation arguments, which are firm); tap-water resistivity (medium; if wrong by 100x the frequency still lands in the kHz-MHz region above or near the band only for near-distilled water, so the conclusion that unpurified water does not fit is robust).
  • All Investigator facts above are tagged [web via Investigator]; I verified only the WO paragraphs myself.
  • The inventors' side: their characterizations are reported as theirs; I found no instrument data supporting or contradicting the bond-angle interpretation, so my negative verdicts on the strong form rest on calculation and on internal inconsistency, not on a failed experiment. No replication or failed replication is claimed here.

Verdict summary

  1. RC resonance mechanism: Internally inconsistent; mechanism Never properly tested.
  2. Distinct stable "unnatural-angle" vapour: strong form Refuted by calculation; weak form (it is not just electrolysis gas) Never properly tested.
  3. Energy release / less energy than electrolysis: "burns" Established; "bond-angle reversion, less than electrolysis" Never properly tested; any excess over electrical input Refuted by calculation (first law).
  4. Engine runs on the vapour: Plausible, untested as electrolysis gas; as bond-angle fuel Never properly tested; the reported gasoline-phase exhaust data fail an atom balance.
  5. Torch and diesel: Plausible, untested; bond-angle attribution Not testable as stated.
  6. Explosion record: event Established (several independent sources); cause and gas identity Never properly tested.