Arbiter: Pappas-PT (US 5,556,418, "Method and Apparatus for Pulsed Magnetic Induction", the PAP-IMI / PAPIMI device)

Role: Step 3. I read facts.json, claims.md, physicist.md, mathematician.md, contrarian.md, and the sites.md and research/ notes as summarised by the reviewers. Tags: [read] project documents (via the reviewers' quotations of the deep read and patent text); [web] with URL; [memory] with confidence; [calc] my own arithmetic (numpy, project venv). I made 4 web calls (budget 8).

Source gap note. There is no source gap for the patent: the single project document is complete (7 pages, OCR plus page images) [read]. There is a gap elsewhere: the inventor's 1991 IECEC paper ("Energy Creation in Electrical Sparks and Discharges: Theory and Direct Experimental Evidence", vol. 4, pp. 416-423) and the ICCF-7 paper were not obtained by anyone. They are the documents that could hold the energy-balance data behind claim 1. Every statement about the device's real operating numbers (40 kV, 60-96 J per pulse, 240 kHz, 50-150 mT) rests on manufacturer figures repeated in two 2020s clinical papers [web], not on the patent (Viti 2023 gives 50-150 mT; Keilani 2025 gives 50-100 mT; see section 9). The patent gives no L, C, voltage, current, vertical scale, gap or gas pressure [read].

Reader note (plain language). Faraday induction: a changing magnetic field B induces an electric field E around it; for a uniform B in a circle of radius r, E = (r/2) dB/dt. Schwan's relation (steady-field limit): a spherical cell of radius a in a field E has its membrane voltage shifted by up to 1.5 E a. Plasma frequency: f_p = 8.98 sqrt(n) Hz, with n the electron density in m^-3.

---

1. Corrections to the reviewers (where they differ or err)

1. Contrarian arithmetic error (claim 3). The contrarian wrote that E = 2.4 kV/m on a 10 micrometre cell gives "about 36 microvolts". Correct: 1.5 x 2400 x 1e-5 = 0.036 V = 36 millivolts, a factor 1000 off [calc]. The physicist (0.15 V) and mathematician (0.11 V) were right. The contrarian's conclusion that "the device's low-frequency field is far too weak to move transmembrane potential" is therefore wrong: at the device's published parameters, ordinary induction gives a membrane shift of tens to a hundred-plus millivolts for a 10 micrometre cell, the same size as the resting potential. This changes the weight of claim 5 (see below).
2. Induced-field spread. Physicist E = 1.0e4 V/m (B = 0.15 T peak, r = 9 cm); mathematician 7.5e3 (0.1 T, r = 10 cm); contrarian 2.4e3 (0.05 T, and an average rate over a quarter period instead of the peak). Recomputed [calc]: 50 mT, r = 10 cm, peak dB/dt = 2 pi f B = 7.5e4 T/s gives 3.8e3 V/m; 150 mT, r = 15 cm gives 1.7e4 V/m. Working range 4e3 to 2e4 V/m at the coil's outer region (zero at the axis; the E grows with r). All inputs are manufacturer figures; none is measured.
3. "Several 10 kA". Physicist says 4.5 kA is "the same order" as "several 10 kA". It is not: with 60 J and 240 kHz, L = 5.9 uH, Z0 = 8.8 ohm, peak current 4.5 kA [calc]. 10 kA needs about 290 J, 30 kA about 2.6 kJ (E = L I^2 / 2) [calc]. The contrarian's flag stands: the published current figures are inconsistent with 60-96 J unless the coil inductance is smaller. This is about the manufacturer's numbers, not the patent.
4. Q and "10-100x". The physicist's Q = Z0/R = 18 as an explanation of 10-100x voltage is loose: a charged capacitor discharging into a ringing circuit does not exceed its initial voltage by Q; magnification by Q applies to a driven circuit. The mathematician's route (duty cycle 1e-5 makes peak/average about 1e5 by construction; inductive kick and step-up coupling add voltage) is the sounder explanation. Either way, no gain is needed.
5. Plasma frequency. All three agree: 100 MHz needs n about 1.2e14 m^-3 [calc]; arcs at atmospheric pressure have n of order 1e22 to 1e24 m^-3, f_p of 1e11 to 1e13 Hz [web: https://www.osti.gov/servlets/purl/1595718 and search summary, typical arc 10^22-24 m^-3; formerly [memory] medium, now checked]. The mathematician's cm^-3 range (1e14-1e17 cm^-3) is the same order or a little lower; f_p is still above 90 GHz.
6. Air breakdown by fast pulses. Checked [web]: nanosecond breakdown in atmospheric air at about 75 kV/cm vs 30 kV/cm DC (https://www.osti.gov/servlets/purl/1787665 and review https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/hve.2016.0014). A factor of about 2.5, not 1-100. Supports the weak form only.
7. Leisner 2002. Checked [web] https://pubmed.ncbi.nlm.nih.gov/11913824/ : rat ulnar fresh fractures, treated vs untreated controls, "do not support the claim that PEMF generated by the PAP-IMI stimulate osteogenesis"; healing faster in untreated rats. Confirms the reviewers; note the control was untreated, not sham.
8. Electroporation threshold. Checked [web] (https://pmc.ncbi.nlm.nih.gov/articles/PMC12045266/ and search summary): a transmembrane potential of about 1 V is the widely used threshold, Schwan's relation gives threshold E inversely proportional to cell radius. [calc] required E = dV / (1.5 a): for 1 V, 6.7e4 V/m (a = 10 um) and 6.7e5 V/m (a = 1 um); for the lower 0.2 V reversible end, 1.3e4 V/m (10 um) and 1.3e5 V/m (1 um). This is critical for "pasteurisation": the physicist's "10-100x too small" and mathematician's "about 200x" assumed PEF thresholds of 1e5-5e6 V/m for any cell. For bacteria-sized cells the device's 4e3-2e4 V/m is 10-100x short of even the reversible threshold (refutation holds with caveats below); for large mammalian cells at the coil's outer region the induced field reaches the lower (reversible) end of the range, so the physicist's blanket "refuted by calculation" is too strong. Caveats: (i) thresholds are for microsecond-to-millisecond pulses, the half-cycle here is about 2 us; (ii) tissue is not a dilute suspension; (iii) the patent's mechanism ("pressure increase") is not electroporation and is not derived.
9. Claim 4 verdict split. Physicist and mathematician say "Refuted by calculation"; contrarian says "Never properly tested". Both are partly right: the stated mechanism (single radio quantum of 4e-7 eV at 100 MHz, 1e-6 eV at 240 MHz [calc], resonating with tissue atoms to drive roughly 1 eV endothermic steps) fails an energy-per-quantum and thermalisation argument; but the experimental claim (selective chemistry in tissue) was never tested at all. I give a split verdict below.
10. Claim 2 "eigenfrequency" split. Physicist: Refuted by calculation; mathematician and contrarian: Never properly tested. The patent never defines "eigenfrequency". If read as the electron plasma frequency, it fails by about 8 orders of magnitude in density [calc]. But the patent could mean some other mode (the discharge circuit); under that reading it is untested and not distinguishable from stray-LC ringing. Verdict is split accordingly.

2. Corrected claim list (Step 1 claims.md edited)

1. (Corrected) An oscillating arc gives transient voltage and current 10-100x above the supplied potential and average current, which the patent attributes to arc resistance going from infinite to under 0.5 ohm to about -100 ohm (a "gain" defined as a ratio of resistances). [read: p.4-5] Note: the legal claims 1 and 17 contain none of this gain language, only resonance absorption [read, per contrarian].
2. (Corrected) The discharge produces sub-microsecond bursts at "characteristic eigenfrequencies" of the gas. Burst length is stated three ways: under 1e-6 s (description), 1e-4 to 1e-6 s (claim 14), under 1e-4 s (claim 16). The oscillogram readout (min 1.3 MHz, max 240.3 MHz, mean 110 MHz) is identical on FIG. 2 and FIG. 3, has no vertical scale and is not a spectrum. [read: deep read]
3. (Corrected) The patent says its sub-breakdown-time bursts exceed air's dielectric strength by 1-100x without ionising it, giving cell-scale fields of order 1e6 V/m non-invasively. The patent prints 1e6 V/m for BOTH the membrane field (70 mV over "several Angstrom") and air strength. Real values: membrane 1.4e7 (5 nm) to 2.3e8 V/m (3 A); air about 3e6 V/m. [read: page image p.4; calc]
4. As before: cell constituents resonate with plasma eigenfrequencies and selectively absorb energy to start selected endothermic reactions (nitrogen example). No frequency is given for any element. [read: p.5-7]
5. As before, but split into: (5a) restoring transmembrane potential (cancer 15 mV, unhealthy 50 mV, healthy 70 mV, no references in the patent) is therapeutic; (5b) the pulses "pasteurize" degenerate cells or microorganisms by pressure increase. Claims 5/6 also contain drafting errors ("diameter ... 6 square millimeters"; claim 6 conflicts with the 30 cm preferred coil). [read: p.4-7] 5c. (Added by QE) The patent also states that cells with low transmembrane potential "are the source of pain generation signals" [read: p.4]; it bears directly on the later pain trials. Efficacy and penetration are asserted without numbers: "effective to the same degree as microcontact", "may even be applied over ordinary clothing", depth "can reach any desired depth" [read: p.5]; Keilani 2025 repeats the manufacturer's "about 20 cm" penetration [web: https://pmc.ncbi.nlm.nih.gov/articles/PMC12534222/]. The only current in the patent is ">1000 amps" (claim 5) [read: p.6-7].
6. (Added from the Investigator, kept in the record) The inventor's later, much stronger claims: "20% excess heat on the probe alone" (New Energy News, June 1998), "C to Fe", "Na + O -> K", the cell as a "(ATP) controlled microscopic Nuclear Reactor", ">1 GW" peak and "888 pulses at about 80 J" (Physiology of the Cell, 2006), and a 2018 Greek utility model on "free energy generation". These are not in the 1996 patent. [web per sites.md and research/files]

3. Claim-by-claim arbitration

Claim 1: 10-100x transient voltage/current via "infinite then negative gain"

• Agree: all three reviewers: peak voltage and current well above supply/average is ordinary; "gain" as a resistance ratio is not an energy gain; the patent's own sentence ("energy increase of these electrons ... is achieved by a corresponding decrease in the energy of the remaining electrons or ions" [read: p.4]) is energy-conserving; no one has seen an energy-balance test.
• Disagree: the physicist and mathematician mark the strong form "Refuted by calculation"; the contrarian marks it the same as an argument but "never properly tested" as an experiment.
• Stronger argument: the split view. The patent's argument is refuted (a resistance ratio is not a power ratio, division by an infinite open-circuit resistance carries no information; duty cycle 1e-5 gives peak/average of 1e5 with no gain [calc]). But the inventor's experimental report (1989-91 papers; 1998 "20% excess heat") has not been read and no calorimetry protocol was seen. I weigh it as an unverified single-source report by the inventor (an interested party), not as a finding that the measurement was done and failed.
• Settle it: measure electrical energy in from the supply and total energy out (calorimetry, calibrated high-voltage probe) over many pulses, and compare the per-pulse output to (1/2) C V^2. Obtain and read the 1991 IECEC paper.
• Final verdict (split): weak form (spikes far above average values): Established. Strong form as argued in the patent: Internally inconsistent and Refuted by calculation as an argument. Net-energy-gain as an experimental claim (1991 paper, 1998 letter): Never properly tested by any neutral party.

Claim 2: sub-microsecond bursts at gas "eigenfrequencies"

• Agree: bursts of ringing at the extrema of a damped oscillation are ordinary for spark-gap discharges; the same waveform is described by independent clinical authors from manufacturer data (Keilani 2025: "megahertz to gigahertz range" peaks [web: https://pmc.ncbi.nlm.nih.gov/articles/PMC12534222/]); the patent's durations are inconsistent; no spectrum or gas-dependence is shown.
• Disagree: whether the eigenfrequency explanation is "refuted".
• Stronger: I accept the physicist's arithmetic (it is checked [calc, web]) but only against the one reading where "eigenfrequency" means electron plasma frequency. The patent does not define the term, so the broader statement is not testable as stated until a gas-substitution test (air vs nitrogen vs argon, same circuit) exists. Stray-LC ringing (100 nH and 25 pF rings at 100 MHz [calc]) explains the readout equally well, and a scope's probe bandwidth could distort the numbers (1.317 MHz minimum etc.).
• Settle it: same geometry, calibrated probe or antenna, spectrum analyser or fast scope; change the gas and the electrode/lead geometry; see whether the lines move with gas or with circuit.
• Final verdict (split; 2c revised per QE, section 9): bursts exist: Established (class of device; patent oscillograms plus later descriptions, but no independent instrument measurement of this device is on record). Plasma-frequency reading of "eigenfrequency": Refuted by calculation. Patent wording "gas-characteristic eigenfrequency": Not testable as stated; operational reading "the burst spectrum changes with the gas": Never properly tested. Stated burst durations: Internally inconsistent.

Claim 3: beating air breakdown; 1e6 V/m at the cell non-invasively

• Agree: the patent's 1e6 V/m vs 1e6 V/m comparison is wrong; the problem does not arise for induction (Faraday fields are generated inside tissue, with no field in air needed); fast pulses exceed DC breakdown by a factor of a few, not 100.
• Disagree: the contrarian says the device's field is "far too weak" (based on the 36 microvolt error). Physicist and mathematician say the induced field is comfortably able to shift membrane voltage by 0.1 V.
• Stronger: physicist and mathematician. Recomputed [calc]: E about 4e3-2e4 V/m, membrane shift 0.06-0.25 V for a = 10 um at the low-frequency (steady) limit. Caveats the reviewers underplay: Schwan's formula is the steady-field limit and the membrane charging time of a few hundred ns to microseconds is comparable to the 240 kHz half-cycle (2 us), so the real shift will be somewhat smaller; tissue conductivity and boundaries alter E; MHz-GHz ringing adds little to induced membrane voltage; no field in tissue has been measured for this device by anyone.
• Settle it: calibrated B-dot (pickup coil) measurement of dB/dt at the applicator, then induced E in a saline or tissue phantom.
• Final verdict (split): premise "1e6 V/m is needed at the cell, so one must beat air": Internally inconsistent and Refuted by calculation (needed external field is about 5e3 V/m at low frequency). Short pulses briefly exceed DC breakdown: Established (factor of a few). 1-100x over air by sub-microsecond timing from this device: Never properly tested. Induction through air reaching tissue at membrane-relevant strength: Plausible, untested at the device (published parameters make it expected; no measurement).

Claim 4: resonant selective absorption, selective endothermic chemistry

• Agree: radio-frequency photons (4e-7 to 1e-6 eV) are about a million times too small for 1 eV chemistry and thermalise in picoseconds in water; atomic lines for nitrogen are optical/UV; bound atoms do not have the free-atom spectrum; the only radio resonance of nitrogen is 14N nuclear magnetic resonance at about 3.08 MHz/T (0.31 MHz at 0.1 T [calc], which needs a static field, and the spin-flip energy is smaller still); the inventor's later NMR/EPR family (US 7,151,372 B2) reframes but does not supply a mechanism; the later transmutation claims have no data in the record.
• Disagree: "Refuted" (physicist, mathematician) vs "Never properly tested" (contrarian).
• Stronger: the physicist/mathematician on mechanism, the contrarian on experiment; I keep both. The Mallove-style point is fair: no controlled experiment on gas-selective biological effect exists, so the experimental statement has not failed a test. But the patent supplies no frequency and no energy for the reaction, so as written there is nothing for a test to confirm.
• Settle it: a gas-substitution assay (same coil and circuit, discharge in air vs nitrogen vs argon, blinded) against a chemical or cell endpoint; for the transmutation claims, an isotope/elemental assay with controls and blanks.
• Final verdict (split): mechanism as stated (single-quantum radio resonance driving endothermic chemistry): Refuted by calculation. Experimental claim of element-selective biochemical effect: Never properly tested. Later biological-transmutation extensions (1998-2006, inventor's own): Never properly tested; in conflict with nuclear physics (Coulomb barrier, no product signatures) [memory] high; related literature (Kervran, Vysotskii, Moawad) is related work, not independent confirmation of this device.

Claim 5: transmembrane potential restoration is therapeutic; "pasteurisation"

Evidence on record, both directions, kept in full: - Leisner 2002, rat ulna fracture: no benefit, slower healing, untreated controls [web: https://pubmed.ncbi.nlm.nih.gov/11913824/]. One species, one model, no sham. - Viti 2023: single-blind randomised sham-controlled, n = 32 (17/15), chronic musculoskeletal pain; heart-rate-variability indices rose after PAPIMI, not after sham; no declared conflict [web: https://pmc.ncbi.nlm.nih.gov/articles/PMC10001461/, per reviewers]. Viti 2025: single arm, n = 30, pain fell by a median 2 points, no control. Keilani 2025: randomised, single-blind, sham-controlled, n = 20, post-COVID fatigue, improved walk and fatigue scales; no declared conflict. All use manufacturer's device and manufacturer figures. - FDA: 510(k) of 1995 not cleared for lack of clinical data; 2002 inspection and 2005 letter list adverse events (rat deaths, tremor, tachycardias, one fatal myocardial infarction during treatment) and a significant-risk Class III designation, import alert 89-08 (regulator's account; it has a regulatory stake). The sponsor-run 37-site study was uncontrolled and IRB-terminated, so it yielded no usable efficacy data. Seattle Times 19 Nov 2007 quotes Pappas ("can cure cancer and AIDS", "absolutely safe") and links deaths to the device (newspaper's characterisation; inventor is an interested party). California 2005 judgment against a distributor for marketing claims (reported by Quackwatch, which has an anti-quackery stake). No causal investigation of the deaths was found. - Cancer/AIDS: no controlled data of any kind.

• Agree: no one measured membrane potential; no controlled cancer trial exists; small sham-controlled pilots show short-term physiological/symptomatic effects; fracture healing not supported.
• Disagree: (i) mechanism: the contrarian says the field cannot move membrane potential (error, see correction 1); the physicist and mathematician say it can transiently. (ii) Pasteurisation: physicist "Refuted by calculation", mathematician/contrarian "Never properly tested".
• Stronger: (i) The physicist/mathematician. A transient membrane-voltage shift of order 0.05-0.25 V is plausible, which is the same mechanism class as magnetic nerve stimulation (E of 10-100 V/m excites neurons at far lower dB/dt [memory] high) and other PEMF devices. So the observed pain/autonomic/fatigue effects have a mundane candidate mechanism (induced currents, nerve/muscle stimulation, also possibly vibration/sensation and imperfect sham blinding at 50-150 mT) without any need for the patent's resonance story. The patent's specific "restore 15 -> 70 mV" is not supported: cells actively restore their own resting potential within ms-s, and a lasting shift is not demonstrated by any calculation. (ii) Pasteurisation: see correction 8. For bacteria-sized cells the induced field is 10-100x short; for large mammalian cells it is near the lower end of reversible electroporation at the coil's outer region. Since the patent's "pressure" mechanism is undefined and no kill experiment exists, "Never properly tested" is the defensible verdict; I do not endorse a flat refutation.
• Settle it: (a) large, independent, sham-controlled multi-centre trial with objective endpoints and verified blinding; (b) measure transmembrane potential or a voltage-dye response in cultured cells in a phantom setup; (c) microbial kill assay with temperature control and a calibrated dB/dt; (d) independent safety review of the cardiac events FDA cited (requires the case records).
• Final verdict (split): short-term pain/HRV/fatigue effects in small sham-controlled pilots: Plausible, untested at adequate scale (two small sham-controlled pilots positive, one single-arm; not replicated) (no independent replication of the same endpoint, small n, subjective outcomes; mechanism likely ordinary induction). Fracture healing: Tested, not supported (one controlled animal study). Restoring transmembrane potential as the therapeutic mechanism, and cancer/AIDS cures: Never properly tested (no tests of any kind; regulators acted on missing clinical data and marketing/safety grounds, so this is not "tested and failed" and not "dismissed after a test"). Pasteurisation by pressure: Never properly tested; induced-field estimate is short of electroporation thresholds for microbes and borderline for large cells.

4. Summary table

5. What would settle it (the single most decisive items)

1. Obtain the 1991 IECEC paper (vol. 4, pp. 416-423) and the ICCF-7 paper: they contain whatever energy-balance data exist.
2. Calorimetric input/output energy test of the discharge circuit (decides claims 1 and 6).
3. B-dot measurement of dB/dt and phantom induced-E map; swept-gas, swept-geometry spectrum with a calibrated probe (decides claims 2, 3).
4. Gas-substitution blinded bioassay (decides claim 4).
5. Independent, adequately sized, sham-controlled trial, with a measure of whether sham blinding holds (decides claim 5 effects).

6. Absence-of-evidence analysis (where the record goes dark)

A. The inventor's energy-creation data (1989-1991, 1998). Silent where it should continue: the patent cites papers at Einsiedeln (1989), the Tesla Symposium (1990) and IECEC (1991), but no one in the record cites them as replicated, and the 1998 "20% excess heat" is a letter with no method. Explanations: (1) the measurements were mundane stored-energy/duty-cycle effects misread as gain (fits the patent's own sentence on energy redistribution and the 1e5 peak/average ratio; evidence: the arithmetic; against: the papers were not read); (2) the experiments were crude and were never repeated because the field (free energy / New Energy circles) has little institutional funding (evidence: IECEC is a peer-reviewed engineering venue, so it was at least accepted; against: nothing shows follow-up even by sympathetic groups); (3) the data were not pursued commercially because the medical device became the product (evidence: inventor's later focus on PAPIMI; Physiology-of-the-Cell text reasserts the claims without data); (4) suppression (no evidence found; no documented blocking of publication or funding). What would decide: the 1991 paper itself, the inventor's lab notes via his family/company, and any citing papers.

B. The US trial data. The FDA says the sponsor-run 37-site study was uncontrolled, mishandled and IRB-terminated. Explanations: trial failure due to poor conduct (supported by FDA's account, an interested regulator); possible overreach by the regulator (no evidence in the record; the inventor's side is documented only through press quotes and the distributor-misconduct items); import-alert and marketing violations by distributors (documented in FDA letters and a California judgment). The data from that study are not in the record; the FDA EIR 2002 and the study files would decide.

C. Post-2007 record in the US vs Europe. The US maintenance fees stop after 2007 (patent expired 2013); the product continued in Europe under CE marking with university trials from 2023. The record therefore did not go dark; it moved jurisdiction. The absence of any large independent trial after 30 years is an anomaly: competing explanations are low commercial incentive for a trial of a spark-gap device with an expired patent, difficulty obtaining ethics approval after the FDA record, and a belief in mainstream circles that the mechanism is absurd, so no one invests (hearsay; no source). A registered trial would decide.

D. Inventor death and succession. Death reported 11 May 2026 from one source (Facebook posts, then a memoriam 11-12 Jun 2026 by a collaborator since 1996, an interested party who claims use in Cuba, Vietnam and Swiss clinics without citing studies). The date is documented in one source; no second source was found. The continuation of the company by family is reported; further company registers (Greek GEMI, Swiss and German registers) could confirm.

7. Open questions

• What does the 1991 IECEC paper actually measure, and does it include an input-output energy account?
• What is the induced field in a phantom for the real device, and what are the true current, voltage, L and C (the manufacturer's "several 10 kA" does not fit 60-96 J)?
• Is the MHz-GHz ringing a property of the gas or of the circuit and probes?
• Are the small positive PAPIMI pilots explained by ordinary PEMF/magnetic stimulation, by blinding failure, or by placebo, and do they replicate in a larger independent trial?
• What happened in the 2002-2005 adverse events (rat deaths, the fatal myocardial infarction), and were they causally investigated?
• Are the later transmutation and "20% excess heat" claims backed by any data in the inventor's archive?
• The inventor's earlier peer-reviewed physics work on the Ampere force (Nuovo Cimento B 1983; Phys. Lett. A 1985; J. Appl. Phys. 1986): not assessed by any reviewer. It is not evidence for the medical claims but belongs in the record as his formal scientific background.

8. Items kept in the record whatever the verdict (from the Investigator)

Patent family WO 94/01176, EP 0648140 B1 (granted 2001), DE 69329908, RU 2126278 C1, US maintenance fees to 2007; predecessors US 5,068,039 and US 5,239,442, EP 0402552 A1 (lightning conductor, the spark-gap/LC design before medical use); US 7,151,372 and US 2006/0220648 (NMR/EPR reframing); imitators (DE 10 2007 030325, EP 2062614, WO 98/57704); supporters (Viti 2023/2025, Keilani 2025, Wallach); critics and regulators (Leisner 2002, FDA 2003/2005, California 2005, Health Canada 2007). The Greek priority number printed as 920200309 vs recorded 920100309 is a probable misprint (unverified).

Web-check log

• Electroporation threshold and Schwan relation: [web] https://pmc.ncbi.nlm.nih.gov/articles/PMC12045266/ (confirmed ~1 V threshold, inverse dependence on radius).
• Leisner 2002: [web] https://pubmed.ncbi.nlm.nih.gov/11913824/ (confirmed).
• Arc electron density 1e22-1e24 m^-3: [web] https://www.osti.gov/servlets/purl/1595718 (search summary).
• Nanosecond breakdown 75 vs 30 kV/cm: [web] https://www.osti.gov/servlets/purl/1787665 (search summary).
• Not re-checked (not verdict-dependent): 14N gyromagnetic ratio 3.077 MHz/T ([memory] high), TMS thresholds ([memory] high), Kervran/Vysotskii literature (related work only).

9. QE revision (2026-10-04, per qe.md; QE verdict MINOR)

Inputs: qe.md, the existing full deep read of the single manifest document. No new deep reads exist in deep/ (QE ordered none; the optional reads of "Physiology of the Cell" and the FDA EIR were not done, so nothing below rests on them). I made no new web calls. [calc] = my arithmetic.

Verdict changes. No central verdict changes. Two wordings are tightened (already applied in sections 2-4): - Claim 2c: the double verdict "Never properly tested / Not testable as stated" is split. Patent wording ("characteristic eigenfrequencies of the gas", never defined): Not testable as stated. Operational reading (the burst spectrum shifts when only the gas is changed): Never properly tested. - Claim 5a: "Plausible, untested" now reads "Plausible, untested at adequate scale (two small sham-controlled pilots positive, one single-arm; not replicated)". The pilots were run, so "untested" beside "tested: yes" misleads. Reasons: small n, subjective endpoints, manufacturer-supplied device, no independent replication. - Claims 1, 3, 4, 5b, 6 and their splits are unchanged.

Corrections of fact (QE spot checks). - Field range: Viti 2023 gives 50-150 mT; Keilani 2025 gives 50-100 mT [web, checked by QE against the saved PMC XML]. My 4e3-2e4 V/m range (section 1.2) used 50 and 150 mT, so it stands as a range over the two sources. At Keilani's 100 mT and r = 15 cm the field is about 1.1e4 V/m [calc: 2 pi x 2.4e5 x 0.1 x 0.15 / 2]. No verdict depends on the difference. - Both papers write pulse energy as "W (joules)" / "Watt (Joule)", a unit muddle in the source; the reviews and I read it as joules. - The OSTI figures for arc electron density (1595718) and ns breakdown 75 vs 30 kV/cm (1787665) came from search-result summaries; the pages were not opened. Read them as [web: search-result summary]. The contrarian's PMC4110921 for the same figure is also a summary. The order-of-magnitude conclusions hold (plasma frequency is off by about 8 orders in density). - The digest labelled 110.033 MHz "reference"; the scope readout says "average" [read: p.3]. No downstream effect. - The contrarian's 36 microvolt error is already corrected in section 1.1 (36 millivolts).

Related work kept in the record. - Liboff et al., US 5,160,591 (ion cyclotron resonance, "regulating transmembrane ion movement utilizing selective harmonic frequencies"), cited by the patent [read: p.1]. Nearest mainstream-adjacent relative of claim 4, with a stated rule f = qB/(2 pi m); it needs a static field, as NMR does. The physicist's remark on its contested replication record is [memory] medium and not web-checked here; it is context, not a verdict input. - Nordenstrom's invasive electrochemical cancer therapy, called "quite successful" in the patent, is the method the patent set out to replace non-invasively [read: p.4]. The 1989 Cure-Pappas-Eichler patent US 5,068,039 (with US 5,239,442) is described as "ideal in applying Nordenstrom's method" [web: https://patents.google.com/patent/US5068039A/en] and is the direct precursor of the ions-by-remote-action idea. Nordenstrom's own claims are not assessed here; related work, not corroboration of this device. - Czerlinski US 4,369,345 and Gordon US 4,622,953, the other prior art cited [read: p.1].

Supporters' accounts, with source and stake. Wallach's and Pappas's statements that treatment succeeded (Seattle Times, 19 Nov 2007; both interested parties); Dr Shri Kant Mishra's paper as FDA EIR exhibit 23 (not obtained); earlier single-subject PAPIMI reports (refractory seizures, wound healing, fatigue syndrome; Viti 2023 refs 40-42) [web per sites.md]; the Karampatsou memoriam (collaborator since 1996, an interested party; reports presentations in Salzburg and use in Cuba, Vietnam and Swiss clinics, no studies cited). These are what supporters report; none is an independent controlled trial. They sit beside the FDA and press adverse-event record (section 3, claim 5) as reports from parties with a stake.

Imitators and regulatory items for the primer (already in section 8): Horkel DE 10 2007 030325 (near copy listing Pappas among original assignees), Pulsartec EP 2062614, Healing Machines WO 98/57704; FDA Import Alert 89-08, the 2003 Warning Letter (device name redacted; CIRCARE labels it PAP-IMI), Health Canada 2007 advisory (these last three via Wikipedia references, not checked directly).

Absence-of-evidence addendum to 6.A: the Google legal-removal notice. The query for the title of the missing 1991 IECEC paper, "Energy Creation in Electrical Sparks" Pappas, returned Google's notice that a result was removed after a legal request (Lumen) [web: research/2026-10-04-google-search.md]. This touches the one document I called decisive for claim 1. Competing explanations: (1) routine copyright takedown of a proceedings copy, common for IEEE/AIChE material; (2) a request by the inventor, his family or a company, for unknown reasons; (3) the notice concerned a different page matching the query. The notice text and removed URL are not in the record, so none can be preferred. What would decide: the Lumen Database notice, and the IECEC 1991 proceedings (vol. 4, pp. 416-423) from a library or IEEE Xplore. Added to the Investigator's next-run list. This is an anomaly to analyse, not a conclusion.

Further open items. - A December 2011 Pappas PDF on an OverUnity forum, snippet "non conventional energy machines comprising arc and negative resistance" (login required, not opened): possibly his later statement of claim 1 in his own words [web per sites.md]. - The FDA 2005 letter's "100 watts at 2 apps (2 seconds x 50 Joules) ... for every two pulses" [web: research/files/seattletimes-2007-11-19-FDA-letter-papimi.pdf, per physicist]: the only non-manufacturer energy figure (from an interested regulator), beside the manufacturer's 60/96 J. It is the same order as 60 J, so it does not resolve the "several 10 kA" mismatch of section 1.3 (10 kA still needs a coil inductance far below the 5.9 uH implied by 60 J at 240 kHz) [calc]. - The patent's own ">1000 amps" with a 5-turn, 30 cm coil gives about 21 mT (mathematician's figure), roughly 4.5 times below the 4.5 kA modelled from manufacturer figures. Induced field at r = 15 cm and 240 kHz is then about 2.4e3 V/m [calc: 2 pi x 2.4e5 x 0.021 x 0.15 / 2], at the low edge of my working range; the Schwan shift for a = 10 um is about 36 mV [calc: 1.5 x 2.4e3 x 1e-5]. Still the size of a resting potential, so claim 3's "Induction ... plausible, untested" stands. - Minor leads: Em-Probe, Inc. as assignee on one WO listing; Pappas as lecturer at TEI Piraeus (2009 Greek blog, snippet only). - Optional reads still undone: "Physiology of the Cell" (67 pp.; any input/output energy measurement, calibrated oscillogram, gas dependence?) and the FDA EIR 2002 (Mishra paper, protocol, outcomes, adverse events).

Corrected claim list: changes. Claim 5c added (pain-generation signals; unquantified efficacy and penetration; ">1000 amps"). Claims 2 and 3 wording as in section 2. Claims.md and facts.json need the matching edits (QE R1, R2), which are outside this file.