Science Contrarian review: Larson-Tyson (Realm Industries)

Reviewer: Science Contrarian (Step 2). Standard: judge by experimental evidence, not by consensus or reputation; build the strongest honest case for the claims; apply the same rigour to the inventors and to their critics.

Source tags: [read] from the project documents (I re-read the patent texts, the Ventura County Star article and page image p.12 of US 2009/0090312 myself); [web] with URL; [memory] with confidence; [calc] my own arithmetic (shown, with assumptions). Interested parties are named as such.

Source gap: none. Both patent applications and the article were read. The Investigator's items are secondary web reports, tagged [web] as the Investigator recorded them; I did not re-open them except where stated.

0. Headline

• The inventors never measured the one thing the whole project turns on: what the gas is. No composition analysis, no energy-in or energy-out figure, no flow, no control run on plain electrolysis gas. The only gas sample that could have been analysed (the "TyLar" cylinders, Aug 2011) was shot by bomb squads unsampled. So the central claims are mostly "never properly tested", not "tested and failed".
• Two claims fail on calculation alone, whatever the experiments say: the kHz "resonance bends the bond angle" mechanism (energy-scale mismatch, section 1) and the stored "bent-molecule vapor" that lasts months (collisional relaxation and condensation, section 2).
• The inventors' own engine graphs contain an oxygen-budget anomaly (section 4) that is exactly what an oxygen-bearing gas (oxyhydrogen) would produce, and they contain no control that separates the two readings. The inventors' own 2009 PCT text concedes the product may be H2 + O2 [web].
• The record of how the work was received is thin and mostly non-technical: no independent test was ever run; critics classified it (electrolysis, "Brown's gas") without testing it; the fraud prosecution expressly did not rule on the science. I set that out in section 7.

1. Claim 1: an "RC resonating frequency" (f = 1/(RC), 1-22 kHz) bends the water bond angle without breaking bonds

Source text: US 2009/0092540 [0005], [0021], [0023]; claims. [read]

a. Mathematics internally consistent? Partly no. - The patent describes a "fixed capacitor in parallel with a variable resistor" and calls 1/(RC) a "resonating frequency" [read: [0021]]. A resistor-capacitor circuit has a time constant and a corner frequency, not a resonance; resonance needs an inductor or another energy-storing element. "Resonance" is a misnomer. No 2π is given and it is not stated whether f is in Hz or rad/s. - "Dual frequency" (the RC frequency "combined" with the supply's intrinsic frequency into a "modifying frequency") is never defined: no beat, product or mixing formula is given. Combining two linear-circuit frequencies gives sum and difference components only if there is a nonlinearity; none is specified. [read: [0021], [0023]] - Polarity contradiction: [0019] connects the anode to the positive terminal; claim 2 connects the anode to the negative terminal. [read: pp.15, 22; Step 1 deep read agrees, I confirmed both passages exist in the text layer]. - What 1/(RC) does mean physically [calc]: for plates 4 ft^2 (0.372 m^2) with the 0.1765 in (4.48 mm) gap of the examples, water (relative permittivity about 80) gives C = about 59 nF per plate pair. Bulk RC = (gap/(sigma A))(eps A/gap) = eps/sigma, independent of geometry: the dielectric relaxation time of the liquid. For nearly pure water (sigma about 5.5e-6 S/m) that gives 1/(2 pi tau) of about 1.2 kHz; for tap water (about 0.05 S/m) about 11 MHz. So the "1-22 kHz" range corresponds to very low-conductivity water, which fits the patent's wording that "pure water may be used" [read: [0018]]. In a real cell at 5 V the electrode double layer (microfarads per cm^2 [memory, medium]) dominates the impedance, not this bulk term. The 1/(RC) number is a legitimate, measurable property of the cell; it is not a molecular property.

b. Consistent with established experiments? No, on energy scale [calc]: - Photon energy at 22 kHz is h f = 9e-11 eV (at 1 kHz, 4e-12 eV). Thermal energy kT at room temperature is 0.0257 eV; the water bending vibration (about 1595 cm^-1 [memory, high]) is 0.198 eV. A 22 kHz drive is about 10^9 times smaller than a bending quantum. Real water responses to a field are the Debye relaxation near 20 GHz [memory, medium] and infrared vibrations; there is no kHz molecular resonance. - A kHz field in conducting water mainly drives ion conduction and electrode reactions, i.e. electrolysis. The patent's own plate/bubble description ("water vapors forming on the plates ... rise to the surface", ultrasound "to detach the bubbles") [read: [0035]-[0041]] is what an electrolysis cell does. - Voltage check: each gap is across the full 5 V supply (adjacent plates alternate anode/cathode [read: [0019]]), which exceeds the 1.23 V thermodynamic water-splitting voltage. Gas evolution at those electrodes is therefore expected by textbook electrochemistry, with no new mechanism required.

c. Tested? No test of the mechanism exists on record. No measurement of the molecular angle (spectroscopy, diffraction, NMR) was made; no run with the "dual frequency" off was reported. [read] No replication found by the Investigator.

d. If true: the angle of an isolated water molecule is fixed by electronic structure; changing it needs energy of the order of an eV (section 2) and would show as infrared/Raman shifts of the bending band. Mainstream molecular physics would have to accommodate kHz drive coupling selectively to a molecular coordinate, which contradicts the equipartition and relaxation picture.

Verdict, split: - The circuit has a time constant 1/(RC) set by water conductivity: Established (trivial). - That a 1-22 kHz "resonance" bends H-O-H without breaking bonds: Refuted by calculation (energy scale) and internally inconsistent (RC is not resonant; undefined "combination"; polarity conflict). This refutes the stated mechanism, not necessarily every observation the inventors made.

2. Claim 2: the product is a distinct "vapor with unnatural bond angles", storable at 1-100 psi for months

a. Math/definitions: the patent says the angle may be ">104.5 degrees", ">109 degrees", or "180 degrees" [read: [0056]] with no energy accounting.

b. Consistent with experiment? - Energy: 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, via search summary; "estimated to lie at 11 114 +/- 5 cm^-1"]. A molecule bent that far is at most that energetic; it is not stable, it is a hot, highly excited molecule. - Lifetime [calc]: in water vapour at 35 psig (about 3.4 atm) and 300 K the collision rate per molecule is about 2e10 per second (kinetic theory, 0.3 nm diameter). Vibrational/rotational relaxation of excited water in the gas phase takes on the order of nanoseconds to microseconds [memory, medium]. "Several months" [read: [0046]] is about 10^16 to 10^19 times longer. No mechanism for a barrier to relaxation is offered; the molecule would simply relax and the energy would appear as heat. - Condensation [calc]: saturated vapour pressure of water at 25 C is about 3.2 kPa (0.46 psi) (Antoine equation [memory, high]). At 35 psig (about 50 psia) water vapour stays a gas only above about 140 C (Antoine: 52 psia at 140 C). A vessel at room temperature holding water at 35 psi would condense to liquid water, not store "vapor". A permanent gas at room temperature and 35 psi is H2, O2, or a mixture; those do not condense. This is the strongest simple physical discriminator in the project, and nobody measured it (the vessel held "vapor" with no temperature, no liquid-collection and no composition stated). - The honest "weak form": the cell produced some pressurised combustible gas. That is observed (engines, torch, Cal/OSHA's record of exploding cylinders). What the gas is has not been established.

c. Tested? No. The patent states no composition. The company's own MSDS (2011, "TyLar" gas) stated "compressed hydrogen"; the invoice for the Korea shipment said "compressed water vapor"; Cal/OSHA found hydrogen and oxygen stored in the same tanks; EPA noted the cylinders were shot on 15 Sep 2011 without sampling [web: https://response.epa.gov/sites/7236/files/SSI_polrep_2.htm; Daily News 9 Feb 2012]. These are descriptions by a regulator (EPA, Cal/OSHA) and by the company's own paperwork; the 2011 product may not be the 2007-08 product. They are single published reports each; I weigh them as real evidence of how the company itself described and handled the gas, not as chemical analysis.

d. If true: a metastable molecular isomer of H2O stable for months in collisional gas would overturn gas-phase relaxation kinetics and the equilibrium vapour pressure curve of water.

Verdict, split: - Strong form (long-lived bent-water gas, distinct from steam and H2/O2): Refuted by calculation (relaxation time; condensation at storage conditions; energy ceiling 133 kJ/mol). - Weak form ("the cell makes a pressurised gas; its identity is unknown"): Never properly tested. Not "dismissed after testing": it was dismissed by classification (Raftery, EPA, Cal/OSHA) and, as to the one physical sample, destroyed unanalysed for safety reasons. The inventors' own later text concedes H2 + O2 [web: WO 2010/039283 [0031]-[0036], https://patents.google.com/patent/WO2010039283A2/en; Investigator quote].

3. Claim 3: igniting the vapor releases energy; uses less energy than electrolysis; safer; no oxygen consumed

a. Math: no energy figure is given. [read] My reader-side arithmetic [calc]: the example generator has 4 chambers x 20 plates x 4 ft^2 = 320 ft^2 of plate area; at the patent's own 10-100 W/ft^2 that is 3.2-32 kW, i.e. 640-6,400 A at 5 V. None of this is stated as measured.

b. Consistent with experiment? - Conservation of energy limits the claim. If the gas is H2 + O2, the best case is Faraday's law: at 640 A the cell makes about 0.12 L/s of mixed gas and about 0.8 kW of combustion heat for 3.2 kW input (about 25%, because 5 V is far above the 1.23-1.48 V needed) [calc]. "Less energy than electrolysis" is then false by definition (it is electrolysis). If the gas were really bent water, it could store at most about 133 kJ/mol, less than the 237 kJ/mol Gibbs energy of splitting, so a "cheaper than electrolysis" fuel is not forbidden by conservation, but it would be an energy store charged from the supply, not a source. The patent offers no input/output numbers to test either. - "No consumption of atmospheric oxygen" (abstract) is not shown for the cars (section 4). - "Safer than electrolysis hydrogen": contradicted by the record: the 2008 tank implosion, the June 2010 fatal explosion, the 2011 Sylmar explosion, and the Cal/OSHA finding that three containers failed [web: AP via Press-Telegram, https://www.presstelegram.com/2011/08/11/family-sorrow-in-fuel-quest/; Simi Valley Acorn 25 Jun 2010]. The inventors dropped this sentence from the 2009 PCT text [web: WO 2010/039283, Investigator note]. Interested parties: Cal/OSHA and EPA are regulators; the company's counsel disputed the legitimacy of the businesses in the fraud case, not the safety findings.

c. Tested? Not by anyone with a calorimeter.

d. If true: a closed-cycle water fuel with output above input would break the first/second laws. Only the weaker "storage" version is compatible with them.

Verdict, split: "net energy / less energy than electrolysis for the same fuel value": Refuted by calculation if the product is H2 + O2 (energy conservation, Faraday) and Not testable as stated otherwise (no numbers). "Safer than electrolysis hydrogen": Tested by events, not supported (accidents on record; the inventors dropped the claim).

4. Claim 4: the vapor runs engines on its own (Honda GX340/GX390 without outside air; Toyota FJ-40 with gasoline shut off, Figs 11-13)

I rendered p.12 of US 2009/0090312 (Fig. 11) and read it. It agrees with the deep read: O2 falls from 21 to 1-4% on gasoline; CO is a dead-flat line at 10.00 from about sample 20 to 71; after the switch O2 returns to about 20.5-20.8; at about sample 147 it steps to about 21.0. [read: page image p.12]

a. Internal consistency of the data (my own check, extending Step 1) [calc]: Per 100 volumes of dry exhaust, burning carbon in air uses 1 volume of O2 per CO2 and 0.5 per CO. The O2 available is N2/3.76. Using the values read from the graphs:

A negative "left for water" is impossible for a hydrocarbon burned in air: burning gasoline also makes water, which needs additional oxygen (about 1 volume O2 per 2 volumes H2O, roughly 7 volumes for these carbon levels). So even the plateau in Fig. 11 leaves far too little oxygen for a real gasoline H/C ratio, and the late-gasoline readings of Figs 11-13 exceed the available air oxygen outright. Either the analyser was mis-calibrated or saturated (CO flat at 10.00 suggests a range limit; the IMR 2800P family lists an NDIR CO range of 0-10/20 vol% [web: https://mallimages.ofweek.com/Upload/guige/2018/09/29/381/%E8%A7%84%E6%A0%BC%E4%B9%A6.pdf, search summary; the exact unit used is unknown]) or extra oxygen was entering with the "vapor". The negative values cluster at or just before the switchover (samples 72-78, 36-42, 37-50), which is when the vapor valve was being opened; in Fig. 13 the text's "gasoline" range (1-36) leaves samples 37-50 unaccounted for [read: [0062]; deep read]. I cannot tell which explanation holds; the patent does not discuss it. This is the same observation Step 1 recorded (CO2 18-19% above the gasoline maximum of about 15%), made stronger by the carbon-oxygen balance.

Other internal points, confirmed against the figures: - Text says CO2/CO "undetectable" and, in Fig. 11/12 passages, levels "unchanged" when the engine is turned off. The plotted vapor-phase traces show CO2 about 0.2-0.6% and CO about 0.2-0.4% (Figs 12-13), and O2 steps up by about 0.3-1.0 points at engine-off. The text overstates the cleanliness. [read] - Fig. 13 gasoline phase is stated as samples 1-36 but the graph shows gasoline-type exhaust to about 51; CO is stated as ">5%" but reaches 10%. Both errors understate the data (drafting errors, not in the inventors' favour). - Hydrocarbons are plotted on a "%" axis to 9%, where exhaust analysers normally report ppm [memory, medium]; only Fig. 11 has an HC trace. - Fig. 12 and 13 are called "repeats" of the FJ-40 run in [0061]-[0062] but the figure list calls Fig. 12 "another automobile" [read: [0018]-[0020]].

b. Consistent with experiment? The vapor-phase exhaust (O2 about 20.0-20.8, CO2 and CO a few tenths of a percent) is what you get if (i) air is still drawn through the engine, and (ii) the fuel burns without taking much oxygen from that air. An oxygen-bearing gas does exactly that [calc]: a 2:1 H2:O2 mixture carries its own oxidiser, so the exhaust oxygen stays near ambient. A fuel that needed air oxygen would deplete it. If hydrogen alone burned in the intake air, the dry exhaust would read 20.3% O2 only at an air-fuel ratio about 24 times lean, far beyond hydrogen's ignition limit (lean limit roughly 4% H2 in air [memory, medium]). The observed 0.5-1.0 point O2 deficit is about 3% of the air's oxygen [calc]. For comparison, a 4.2 L six-cylinder at about 700 rpm needs about 4 kW just to overcome friction (assumed FMEP 1.5 bar [memory, low]) and so about 9-15 kW of fuel energy at 25-40% indicated efficiency, which would consume roughly 35-55% of the air's oxygen if the fuel took oxygen from the air [calc, assumptions stated: air-flow 7 g/s from 25% volumetric efficiency]. The graphs therefore show either the engine ran on a gas that supplied its own oxygen, or it was not delivering power from air-oxygen combustion. Both are consistent with the inventors' account that the "vapor" is an oxidiser-bearing fuel; neither distinguishes bent water from H2 + O2.

What the graphs cannot show: that the engine was running (no rpm, load, power, vapor flow or pressure decay; only text says it was running); that the vapor was the only fuel (the carburetor float bowl and its draining time are not addressed; the air path stayed open); that the air was excluded (the FJ-40 text does not say so; Fig. 14 shows air entering). The Honda "no outside air" runs have no data at all. [read]

Strongest honest case for the claim: four things are on the record in the inventors' favour. (1) Oxyhydrogen engine operation is real, so an engine can run on a water-derived gas; the claim "something derived from water ran an engine" is credible. (2) The vapor-phase traces are reproducible across three runs (O2 about 20-20.8, CO2 about 0.2-0.6). (3) The inventors reported against their own interest: they published CO2 and CO levels that did not read "zero" and a gasoline-phase graph they could have trimmed. (4) The authorities' own accounts (Cal/OSHA, EPA) describe a working gas production operation that could fill cylinders for sale.

Against, equally: no control (same cell with plain DC, bottled H2/O2); no energy balance; undefined plate current; an oxygen budget violated in the gasoline phase; "undetectable" is false for the plotted traces.

c. Tested? By the applicants only, no independent witness, date or place named. No replication found (Investigator: none found). No neutral test exists.

d. If true: (only the strong form) an engine running on a fuel that releases energy from water with no oxygen use would need a new energy source or a violation of conservation.

Verdict: Never properly tested. The inventors' own tests are uncontrolled and do not separate the claim from oxyhydrogen. The data are anomalous in the oxygen budget. The claim was not dismissed after testing; the patent family was allowed to lapse (below).

5. Claim 5: torch cut thin titanium and nickel-plated steel and melted tungsten; 60:40 to 70:30 vapor:diesel ratio combusted on compression

a/b. Oxyhydrogen torches cut and braze thin sheet metal [memory, medium]. The adiabatic flame temperature of stoichiometric oxyhydrogen is about 3,075 K at equilibrium [web: https://etasr.com/index.php/ETASR/article/view/6132, search summary]; tungsten melts at 3,695 K (3,422 C) [web: https://en.wikipedia.org/wiki/Tungsten, search summary]. A flame cannot exceed its adiabatic temperature, so "melted a tungsten welding rod" [read: [0058]] cannot mean bulk melting in a plain oxyhydrogen flame. A tungsten rod in an oxidising flame forms tungsten trioxide, which melts at about 1,473 C [memory, medium; page "Tungsten trioxide", https://en.wikipedia.org/wiki/Tungsten_trioxide listed in the search results, value not independently read], so surface "melting" or burn-back is plausible. The patent gives no flame temperature, rod size, time or photograph. If the claim is true as bulk melting, a flame hotter than oxyhydrogen would be needed, which is itself a striking claim and not argued. The diesel ratio has "no data" [read: [0079]]; the 2009 PCT text frames it as "test experiments" [web: Investigator].

c. Tested? Once, by the applicants; no data. d. For the cutting: nothing. For a hotter-than-oxyhydrogen flame: a new combustion channel.

Verdict, split: thin-sheet cutting: Plausible, untested (consistent with an oxyhydrogen torch); tungsten melting as bulk melting: Not supported as stated (exceeds the oxyhydrogen adiabatic limit); never properly tested; diesel ratio: Not testable as stated (no data).

6. Claim 6: the 2010 explosion record

Re-read: the VC Star article (17 June 2010) says a man was killed at Realm Industries, an alternative fuel company, "about 18 months after a similar incident"; it says the company's work involved "extracting hydrogen from water"; it lists Stehl and Larson as the patent inventors; it does not name the dead man [read]. The Investigator found that the victim was co-inventor Tyson Larson, 28 [web: Simi Valley Acorn, 25 Jun 2010, https://www.simivalleyacorn.com/articles/still-no-official-cause-of-explosion-on-easy-street/]. Cal/OSHA found three pressurised containers failed (no accident citation for lack of conclusive evidence; $5,685 for other citations, appealed) [web: AP]. Cause: "accidental but undetermined". Later: 9 Aug 2011 Sylmar explosion, $540,890 Cal/OSHA penalties including "oxygen and hydrogen stored together" [web: Daily News 9 Feb 2012].

This is the project's only independent record. It documents handling of a pressurised combustible gas, not the bond-angle claim. Per the SOURCE HONESTY rule: Cal/OSHA's and the AUSA's descriptions are by parties with regulatory or prosecutorial stakes; each rests on one published source I have not cross-checked against the original reports.

Verdict: Established as a historical record (not a physics claim).

7. How the claims were received (the Investigator's findings, kept in)

I weigh each item for what it does and does not show.

• USPTO. Parent application 11/973,170 was "abandoned for failure to respond to an office action" on 3 Sep 2013 [web: Google Patents]. The office action is unread (file wrapper not retrieved). Abandonment is a procedural, funding or interest decision; it shows nothing about the examiner's technical reasoning. The CIP was granted as US 7,793,621 B2 (14 Sep 2010), after Larson's death, and lapsed 14 Sep 2018 for unpaid fees. Granted claim 1 requires metering vapor and air and mixing "vapor molecules, air, and a fuel" [web: Google Patents, partial text]. The granted claim therefore does not cover the air-free, vapor-only engine of Example 2 or the claimed independence from fuel. A US patent grant is not a check on physics.
• EPO. Search opinion 31 May 2011, first examination report 27 Jan 2012, application deemed withdrawn 5 Jul 2013 after late fees; the documents did not render, so the examiner's objections are unread [web: register.epo.org]. This is the nearest thing to an independent technical reading; it is a lead, not a result.
• Inventors' own later text. The Apr 2009 PCT renames the product "energy carriers", concedes H2 + O2 could result, softens "insufficient to break bonds" to "may be insufficient", and drops "safer than electrolysis" [web: WO 2010/039283]. This is the applicants' hedge in a commercial filing; it is also the closest thing to an admission against the strong claim.
• Outside expert. Dan Raftery (Purdue chemistry) told the VC Star the application "does not seem viable" and looked like electrolysis [web: VC Star 21 Jun 2010]. He worked from news reports and the application, not the apparatus. That is an informed opinion, not a test; no test of the apparatus by any outside scientist is on record.
• Prosecution. AUSA Dooley (interested party): "We're not alleging that this is some kind of fraudulent scientific process" [web: VC Star]. Stehl entered an Alford plea (2 Sep 2014); Rossignol was convicted (20 years); charges concern misstatements about contracts and finances [web: https://www.justice.gov/usao-ndny/pr/second-defendant-convicted-8-million-fraud-scheme-concerning-purported-alternative]. These findings bear on the commercial promoters, not on whether a bond-angle effect exists; they also do not clear the physics. The commercial track record (investors told contracts existed when they did not) is relevant to how much weight the unwitnessed, undated test claims deserve; I count it as an additional reason to want independent data, not as a refutation.
• No replication, no independent measurement, no witness was found for any claim (Investigator).
• Related work. R2D2 LLC's WO 2023/055615 claims a "structurally altered gas molecule" from water (2H:1O, bond angle 94-104 degrees, stable above 300 psig) [web: https://patents.google.com/patent/WO2023055615A1/en] and cites Realm. Stan's Legacy files the Realm patents with Stanley Meyer's water-fuel lineage [web: stanslegacy.com; advocacy archive]. Citing patents (Clentech, HMT Hydromotive) treat Realm as prior art for electrolysis-gas engines [web: Google Patents]. These show the same idea claimed by others as a patent, not as a measured result. None provides a measurement.
• Search-results removal. The Investigator reports that Google noted a DMCA notice removing 10 results on one "Sylmar/Tylar" query [web: sites.md]. Who filed it, and what was removed, is unknown. I treat it as an unexplained item for the record, not as evidence of suppression.
• Absence of evidence after 2011. Competing explanations the Investigator lists (ordinary electrolysis gas that failed commercially and legally; work continuing under another name or abroad; the death, injuries and penalties ending it) all fit the facts. Records that would decide: the USPTO and EPO file wrappers, the Korean consignee, the CD Cal court exhibits, Cal/OSHA inspection files, Strategic Sciences' Nevada filings. The sample that would have settled composition was destroyed, an unforced loss to the evidence by the safety response, which was reasonable in itself.

Dismissed without testing, or tested and failed? For the mechanism: dismissed without testing by outside experts and agencies, who classified it; refuted on first principles by calculation (sections 1-2). For the "water vapor runs engines": never tested neutrally.

8. Corrections to Step 1

1. claims.md item 4 says CO, CO2 and hydrocarbons "near zero". The plotted vapor-phase traces show CO2 about 0.2-0.6% and CO about 0.2-0.4% (Figs 12-13), roughly 5-15 times ambient CO2; hydrocarbons are plotted only in Fig. 11. The Step 1 deep read had these figures; the claims.md wording should reflect them. I confirmed against my own render of p.12.
2. claims.md item 4 says Honda engines "ran with outside air excluded". That is the applicants' text only; no data, rpm, duration or witness accompany it. It should be worded "the patent states" (the deep read did say so).
3. claims.md item 4: "gasoline-phase CO2 of 18-19% exceeds the usual gasoline maximum" is correct but understated. The oxygen-budget check in section 4 shows the late-gasoline and switchover readings exceed even the oxygen in the intake air (negative water budget), and it is the CO and CO2 together that matter.
4. claims.md item 6 says the article does not name the victim. The article (verified) does not. The Investigator identified him as co-inventor Tyson Larson [web]; claims.md should be updated, and the "similar incident" 18 months earlier is the December 2008 tank implosion.
5. claims.md / facts.json treat the folder's patent as "not stated to have been granted". The CIP (the 2009/0090312 family) was granted 14 Sep 2010 (US 7,793,621 B2); the bond-angle parent was abandoned [web: Google Patents]. The Step 1 note on this is out of date.
6. Deep-read figures: the two deep reads differ by about 0.1-0.3 points in O2 baselines and in whether the engine-off O2 step is described as "unchanged" (the 092540 deep read reports a rise from about 20.7 to 20.9; the 090312 deep read reports about 20.6-20.8 to 21.0). My render shows about 20.6-20.8 then about 21.0-21.1. Immaterial to verdicts; either reading contradicts the text's "unchanged".
7. Weak point not stated in Step 1: claim 1 is described as "RC resonating frequency". An RC circuit has no resonance; the claims list should call it the inventors' term for the RC time-constant frequency.

9. Summary table

10. What would settle the open points

1. Gas chromatography/mass spectrometry or a thermal conductivity analysis of the vapor from the example generator, plus its dew point at 35 psig and the condensate collected. A single run separates "H2 + O2" from "a new gas".
2. A run with the engine on the vapor with metered vapor flow, rpm, exhaust O2 plus lambda, and a control with bottled oxyhydrogen at matched flow. Pressure decay of the vessel over the 11-minute run gives the fuel flow directly (about 1,000-1,500 L of gas at standard conditions over 11.5 minutes if idle needs 9-15 kW [calc, assumptions above], implying a vessel of several hundred litres at 35 psig; the patent gives no vessel volume).
3. Joule meter on the cell: input energy per mole of gas, compared with 237 kJ/mol.
4. The USPTO and EPO file wrappers, for the examiners' technical objections.
5. Infrared/Raman spectrum of the vapor in the 1595 cm^-1 region.