QE (Step 4b): Pappas-PT (US 5,556,418, "Method and Apparatus for Pulsed Magnetic Induction", the PAP-IMI / PAPIMI device)

QE date: 2026-10-04. Inputs read: _text/manifest.json, the extracted text (_text/US5556418-dd083a.txt, all 7 pages), digest/US5556418-dd083a.01.md, deep/US5556418-dd083a.md, facts.json, claims.md, sites.md, the research/ notes (by way of sites.md and the reviewers' citations) and three research/files texts checked directly (Leisner 2002, Viti 2023, Keilani 2025), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md and example.py. I rendered page 4 of the patent at 300 dpi to check the exponents (scratch: /tmp/qe-pappas/).

Verdict: MINOR

None of the central verdicts needs to change. The arbiter already caught and corrected the one load-bearing reviewer error: the contrarian's "36 microvolts", which should be 36 millivolts. The primer follows the arbiter and does its arithmetic correctly. The defects are these: Step 1 files are stale and inaccurate (how_read, claims 2 and 3); several of the Investigator's pertinent findings were dropped between the reviews and the arbiter and primer (retention); one possible "record goes dark" item was never analysed; and there are a few reader-profile and tagging gaps. All of these can be fixed by editing facts.json, claims.md, arbiter.md (adding a "QE revision" section) and primer.md.

---

1. Coverage

• The project has one document. It was digested and then deep-read in full, and the deep reader checked every exponent and both oscillograms on rendered page images. Nothing was skipped. No coverage defect.
• The Investigator's research/files hold 25 saved documents. They are not in the manifest, so they are outside Step 0/1a coverage, and the reviewers used them through sites.md and the research notes. Two of them are the inventor's own long-form statements: "Pappas' Physiology of the Cell and its Atomic Energy", 2006, 67 pp., which the reviewers only grepped, and the 1998 New Energy News letter. The FDA EIR (2002, 17 pp., scanned) is the main historical record. See Remediation R7 (optional reads).

2. Shortcuts

• Defect S1 (Step 1 record inaccurate). facts.json how_read says "digest ... no deep read ordered", and claims.md says "read via digest only (no deep reads)". A full deep read exists (deep/US5556418-dd083a.md, with page images), and every reviewer used it. The PROCESS rule says how_read must describe what was actually done. Both files need correcting.
• No central claim rests on a digest alone. The deep read covers every central claim, the derivation (resistance ratio), the numbers and the 17 claims. No sampling defect.
• The digest made one small error that the deep read corrected and that did not propagate. The digest called 110.033 MHz the "reference frequency", but the scope readout labels it "average" [read: p.3, extracted text "average repetitive 110.033MHz"; deep read section 4].

3. Accuracy spot checks (against the extracted text and page image)

Arithmetic: I reran example.py (exit 0) and checked by hand the reviewers' and arbiter's numbers: C = 75 nF, L = 5.86 uH, Z0 = 8.84 ohm, I0 = 4.5 kA, B = 95 mT (N = 5), induced E 3.8e3 to 1.7e4 V/m, Schwan dV 113 mV (a = 10 um), tau_m = 150 ns, plasma density 1.24e14 m^-3 at 100 MHz, h f = 4.1e-7 eV, the contrarian's loop inductance 8.1 uH, f = 205 kHz and I = 3.9 kA, and the arbiter's 293 J for 10 kA. All are correct. The contrarian's 36 uV is wrong (it should be 36 mV), and the arbiter correctly flagged and fixed it.

4. Missed material (from the digest, the text and the Investigator's findings)

Items that facts.json, claims.md, the arbiter or the primer missed or under-used:

1. The patent's own current figure. The induction coil is designed "to provide very high rates of power, current >1000amps", and claim 5 says "at least 1000 amps" [read: p.6-7]. This is the only electrical current stated in the patent. The mathematician used it (B = 21 mT at 1 kA, 5 turns). The primer's worked example uses only manufacturer figures and never mentions the patent's own number.
2. "Cells with a low transmembrane potential are the source of pain generation signals" [read: p.4]. This is a distinct claim and bears directly on the later pain trials (Viti 2023/2025). It is missing from claims.md and the primer; the physicist treats it only in passing.
3. Unquantified efficacy and penetration statements [read: p.5]: "found to be effective to the same degree as microcontact"; "may even be applied over ordinary clothing"; penetration depth "proportional to the intensity of the applied field ... can reach any desired depth". Keilani 2025 repeats the manufacturer's claim of "a penetration of the body tissue depth of about 20 cm" [web: https://pmc.ncbi.nlm.nih.gov/articles/PMC12534222/]. Neither file records these.
4. Prior art cited by the patent: Czerlinski US 4,369,345, Gordon US 4,622,953 and Liboff et al. US 5,160,591 (ion cyclotron resonance, "regulating transmembrane ion movement utilizing selective harmonic frequencies") [read: p.1; web per sites.md]. Liboff is the nearest mainstream-adjacent version of claim 4 and has a stated frequency-field relation (f = qB/2 pi m). The physicist, mathematician and contrarian all discuss it, but it is absent from the arbiter and the primer (see Retention).
5. Nordenstrom. The patent calls his invasive electrode cancer method "quite successful" and builds its problem statement on it [read: p.4]. The 1989 Cure-Pappas-Eichler patent is described as "ideal in applying Nordenstrom's method" [web: https://patents.google.com/patent/US5068039A/en]. Nordenstrom is absent from the arbiter and the primer, and the Investigator listed it as "not searched this run".
6. Google legal-removal notice. The query "Energy Creation in Electrical Sparks" Pappas, which is the title of the missing 1991 IECEC paper, returned Google's notice that one result was removed after a legal request (Lumen) [web: research/2026-10-04-google-search.md; sites.md]. This bears on the one document the arbiter calls decisive for claim 1. The arbiter's absence-of-evidence section A does not mention it. Under SOURCE HONESTY it should be set out as an anomaly with competing explanations: copyright takedown of a pirated proceedings copy, which is common for IEEE/AIChE material; a takedown by a rights holder or the inventor's family; or something else. It should also say what would decide it: a Lumen Database lookup of the notice, and the library copy of the IECEC proceedings.
7. Blocked lead: a December 2011 PDF by Pappas on an OverUnity forum, whose snippet mentions "non conventional energy machines comprising arc and negative resistance" (login required, not opened) [web per sites.md]. This is a possible later statement of claim 1 in the inventor's own words. It is missing from the arbiter's open items.
8. Earlier single-subject PAPIMI human reports (refractory seizures, skin wound healing, fatigue syndrome; Viti 2023 refs 40-42) [web per sites.md]. These are the supporting record before 2023. They are missing from the arbiter and the primer.
9. FDA 2005 energy figure: "capable of delivering 100 watts at 2 apps (2 seconds x 50 Joules) of energy for every two pulses" [web: research/files/seattletimes-2007-11-19-FDA-letter-papimi.pdf, per physicist and sites.md]. This 50 J/pulse figure is the only non-manufacturer energy figure in the record (from an interested regulator), and it sits beside the manufacturer's 60/96 J. It is not in the arbiter or the primer.
10. Em-Probe, Inc. is named as an assignee on one WO listing, and Pappas was a lecturer or "professor" at TEI Piraeus (a 2009 Greek blog, snippet only) [web per sites.md]. These are minor historical leads, absent downstream; listing them is enough.

5. Retention check (SOURCE HONESTY: "Keep pertinent information")

I compared the Investigator's "Leads for review" (sites.md) item by item with arbiter.md and primer.md. Most of the record is retained well. The primer's "Safety record, kept in full" paragraph, the trials table (positive and negative) and the later inventor claims (20% excess heat, C to Fe, Na + O -> K, >1 GW, 2018 "free energy" utility model) are all present and fairly tagged. The arbiter's section 8 lists the patent family, its predecessors, imitators, supporters and critics.

Defects (pertinent material dropped or given unevenly):

• R-a (supporting side thinner than the critical side in the primer). The primer keeps the full adverse record: the FDA letter's rat deaths, tachycardias and fatal MI; the infant with rhabdomyosarcoma; "plant and seed germinator"; the Seattle Times; the California judgment. It omits every supporter account the Investigator found: Wallach's and Pappas's statements that the treatment succeeded (Seattle Times); Dr Shri Kant Mishra's paper (FDA EIR exhibit 23); the Karampatsou memoriam (a collaborator since 1996 who reports conference presentations in Salzburg and use in Cuba, Vietnam and Swiss clinics); and the earlier single-subject reports (item 4.8). SOURCE HONESTY says the rule "cuts both ways". Each should be stated with its source and its stake, as the arbiter does for Karampatsou in section 6D.
• R-b (related work dropped). Liboff's ion cyclotron resonance (item 4.4) and Nordenstrom's electrochemical therapy (item 4.5) appear in the reviews but in neither the arbiter nor the primer. They are the two pieces of related work the patent itself relies on. Together with the Cure-Pappas-Eichler 1989 precursor (in the arbiter only by number; absent from the primer, which names only the lightning-conductor patent as the precursor), they show where the idea came from.
• R-c (imitators and regulatory items missing from the primer). The primer leaves out the imitators: Horkel DE 10 2007 030325, a near copy listing Pappas among the original assignees; Pulsartec EP 2062614; and Healing Machines WO 98/57704. It also leaves out FDA Import Alert 89-08, the 2003 FDA Warning Letter (LifeWorks; device name redacted, CIRCARE labels it PAP-IMI) and the Health Canada 2007 advisory. The arbiter keeps them in section 8. The primer need only list them briefly.
• R-d (unanalysed anomaly). The Google legal-removal notice (item 4.6) is not analysed anywhere.

No item was dropped because a reviewer judged it negatively. The omissions look like condensation, not suppression. Under the rule they are still defects, but minor ones.

6. Process

• Source tags. Present and mostly honest throughout. Three issues:
• The arbiter's web checks for arc electron density (OSTI 1595718) and nanosecond breakdown (OSTI 1787665) are marked "search summary", which means the pages were not opened. The primer then cites OSTI 1787665 as a plain [web] source for "75 kV/cm" without that caveat. The contrarian cited a different URL (PMC4110921) for the same figure, also from a search summary. The primer should say "[web: search-result summary of ...]" or the page should be opened.
• "50-150 mT" is attributed to both clinical papers, but Keilani says 50-100 mT (spot check 5).
• The primer's [memory, medium] items (Cone; Binggeli and Cameron on tumour membrane potential; tissue conductivity; the tau_m resistivities) are not load-bearing for verdicts. Acceptable as tagged.
• Web checks. The physicist made no web calls (budget about 6) despite a medium-confidence [memory] fact that its claim-2 "Refuted by calculation" verdict depends on (arc electron density). The mathematician also made none, but tagged its comparable facts as high-confidence textbook values, which is defensible. The arbiter repaired the gap with 4 web checks (arc density, ns breakdown, electroporation threshold, Leisner). The physicist's skipped checks are a process defect, now remedied downstream. No further action is needed beyond the tag fix above.
• Verdict scale. Applied consistently on the whole, and the split verdicts are well explained. Two wording points:
• Claim 2c is given a double verdict, "Never properly tested / Not testable as stated". The scale wants one per part. The arbiter's own reasoning (the term is undefined until a gas-substitution test exists, which would make it testable) points to Not testable as stated for the patent's wording and Never properly tested for the operational reading "the burst spectrum changes with the gas". The verdict should be split explicitly.
• Claim 5a, "Plausible, untested", sits awkwardly beside two small randomised sham-controlled pilots that were run and were positive. The scale defines "Plausible, untested" as "no real test yet". The arbiter's reason (pilots only, not replicated, subjective endpoints) should be stated in the verdict cell, e.g. "Plausible, untested at adequate scale (two small sham-controlled pilots positive, not replicated)". Otherwise the reader sees "untested" next to "tested: yes".
• Primer vs STEP 4 and the reader profile. All nine sections are present and in order. The worked example is excellent: one pulse from capacitor to membrane, with real numbers, and reproduced by example.py. The answers are correct (I checked Q1, Q2 and Q4 by hand). Defects:
• Acronyms not defined on first use: IECEC (Intersociety Energy Conversion Engineering Conference), ICCF (International Conference on Cold Fusion), MRI, FDA, 510(k) (FDA premarket notification), IRB (institutional review board), EIR (establishment inspection report), CE marking. (HRV and EPR are defined; PEMF is not used in the primer.)
• Section 2 cites F. F. Chen and Raizer by chapter description only, which is fine, and gives "Griffiths chapter 7", which is correct.
• example.py. Runs cleanly with the project venv (exit 0) and reproduces every number in primer section 5.
• Step 1 corrections not applied. All three reviewers flagged that claims.md claim 3 states "cell-scale fields of order 10^6 V/m" as if it were the true requirement, and that claim 2 says "sub-microsecond" although claims 14 and 16 say 10^-4 to 10^-6 s and < 10^-4 s. The arbiter wrote a corrected claim list (its section 2), but claims.md and facts.json were never updated. facts.json also lacks the Greek priority-number discrepancy (printed 920200309; Espacenet GR920100309A).

7. Remediation (one round)

Deep reads (manifest documents): none needed. The only manifest document has already been deep-read in full with page images.

R1. facts.json (US5556418.pdf entry): - how_read -> "digest + deep read (pp.1-7, with page images for pp.2-7: exponents, FIGS. 1-3, claims 14/16)". - sources -> add "[read: page image pp.2-7] (deep read)". - summary/key_claims: state that the patent prints 10^6 V/m for both the membrane field and air (an internal inconsistency: the real values are about 1.4e7 to 2.3e8 V/m and 3e6 V/m); give burst durations as stated three ways (< 1e-6 s in the description; 1e-4 to 1e-6 s in claim 14; < 1e-4 s in claim 16); add the coil current ">1000 A"; add the "pain generation signals" claim; add the Greek priority number discrepancy (printed 920200309; Espacenet GR920100309A, published 31 Mar 1994 [web per sites.md]); and record the cited prior art in full (Czerlinski, Gordon, Liboff et al.).

R2. claims.md: - Replace the header line with "digest + full deep read with page images". - Claim 2: replace "sub-microsecond" with the three stated durations. - Claim 3: say "the patent states 10^6 V/m (for both membrane and air)", not a bare "fields of order 10^6 V/m". - Split claim 5 into 5a (potential restoration / pain) and 5b (pasteurisation), as the arbiter did. - Add the arbiter's claim 6 (the inventor's later excess-heat / transmutation / GW claims).

R3. arbiter.md: append a "QE revision" section (no verdict changes needed), containing: - Claim 2c: an explicit split: patent wording Not testable as stated; operational reading "burst spectrum changes with gas" Never properly tested. - Claim 5a: verdict wording "Plausible, untested at adequate scale (two small sham-controlled pilots positive, one single-arm; not replicated)". - Related work kept in the record: Liboff US 5,160,591 (ion cyclotron resonance, f = qB/2 pi m, with a contested replication record per the physicist's [memory] medium; web-verify if it is used for anything more than context); Nordenstrom's electrochemical cancer therapy (as the patent's own model); Cure-Pappas-Eichler US 5,068,039 / 5,239,442 named with co-inventors. - Absence of evidence, item A addendum: the Google legal-removal (Lumen) notice on the query for the 1991 paper's title, with competing explanations (routine copyright takedown of a proceedings copy; a request by a rights holder or the family; other) and the deciding record (the Lumen Database notice; the IECEC 1991 proceedings, vol. 4, pp. 416-423, via a library or IEEE Xplore, if indexed). - Open items: the blocked 2011 OverUnity PDF by Pappas on "arc and negative resistance"; the earlier single-subject PAPIMI reports (Viti 2023 refs 40-42); the FDA 2005 "50 Joules" per-pulse figure beside the manufacturer's 60/96 J; Keilani's 50-100 mT (not 50-150) and its "about 20 cm" penetration claim (manufacturer's characterisation).

R4. primer.md: - Define the acronyms listed in section 6 at first use. - In section 4 (Where the inventor departs) or 6 (Evidence): add a short "Supporters and the inventor's side" paragraph beside the "Safety record" paragraph. It should cover Wallach's and Pappas's success claims (Seattle Times 2007; interested parties), the Mishra paper (FDA EIR exhibit 23, not obtained), the earlier single-subject reports (seizures, wound healing, fatigue; Viti 2023 refs 40-42), the Karampatsou memoriam claims (Salzburg presentations; Cuba, Vietnam, Swiss clinics; collaborator since 1996; no studies cited), and the claimed 20 cm penetration (manufacturer, via Keilani). - Section 4: add Liboff's ion cyclotron resonance as the mainstream-adjacent cousin of claim 4 (it needs a static field, like NMR), and Nordenstrom as the invasive method the patent set out to replace; name the 1989 Cure-Pappas-Eichler "ions by remote action" patent as the direct precursor. - Section 5: mention the patent's own ">1000 A" coil current (B = 21 mT at 1 kA, 5 turns, 30 cm, a factor of 4.5 below the 4.5 kA modelled), so the reader sees what the patent itself specifies. - Section 6 "Safety record" paragraph: add Import Alert 89-08, the 2003 Warning Letter (device redacted; CIRCARE's identification) and the Health Canada 2007 advisory (via Wikipedia references, not checked directly); add the imitators (Horkel DE 10 2007 030325, Pulsartec, Healing Machines) to section 9 or section 4. - Fix the tags: Keilani is 50-100 mT; OSTI 1787665 is "[web: search-result summary]" unless opened. - Evidence table rows 2c and 5a: mirror the R3 wording.

R5. Investigator (next run, not part of this remediation round): the 1991 IECEC paper (top priority; it decides claim 1's experimental form); the Lumen notice; the ICCF-7 paper; the Internet Archive copies of papimi.gr / papimi.com; Health Canada; the Greek patent office text of GR920100309A. These are already listed as "not reached (time limit)" in sites.md, except the Lumen lookup, which should be added.

R6. QE recheck (Sonnet): confirm R1-R4 and append "Recheck" to this file.

R7. Optional reads of Investigator files (not manifest documents, so not in the deep-read list), in priority order, for whoever does R3/R4: 1. research/files/Pappas-Physiology-of-the-Cell-ralf-kollinger.pdf (67 pp., the inventor's own 2006 compilation). Question: does it contain any energy-balance (input vs output) measurement, a calibrated oscillogram, or a gas-dependence of the burst spectrum (claims 1, 2, 4)? Sonnet. 2. research/files/seattletimes-2007-FDA-EIR-papimi.pdf (17 pp., scanned). Question: what does it say about the Mishra paper (exhibit 23), the study protocol, the outcomes recorded, and the adverse events? Sonnet.

8. Summary of defects

---

Recheck

Recheck date: 2026-10-04. Read the current facts.json, claims.md, arbiter.md and primer.md against the remediation list.

• S1 / R1 (facts.json): FIXED. how_read now says digest + full deep read with page images; the 10^6 V/m inconsistency, the three burst-duration statements, the >1000 A coil current, the "pain generation signals" claim, the Greek priority-number discrepancy and the cited prior art (Czerlinski, Gordon, Liboff) are recorded.
• R2 (claims.md): FIXED. The header says digest + full deep read; claims 2 and 3 are corrected; claim 5 is split into 5a and 5b; claim 6 (the inventor's later claims) is added.
• R3 (arbiter.md, section 9 "QE revision"): FIXED. Claim 2c is split (Not testable as stated / Never properly tested); the claim 5a wording is corrected; Liboff, Nordenstrom and Cure-Pappas-Eichler are kept; the Lumen notice is analysed with competing explanations and the deciding record; the OverUnity lead, the FDA 50 J figure and Keilani's 50-100 mT are added.
• R4 (primer.md): FIXED. Acronyms are defined (IECEC, ICCF, MRI, FDA, 510(k), IRB, EIR, CE); a supporters paragraph is present (Wallach/Pappas, Mishra, earlier single-subject reports, Karampatsou/Salzburg, the 20 cm claim); Liboff, Nordenstrom and the 1989 precursor are in section 4; the patent's >1000 A coil current is in section 5; Import Alert 89-08, the Warning Letter, the Health Canada advisory and the imitators (Horkel, Pulsartec, Healing Machines) are listed; the OSTI source is tagged "search-result summary"; Keilani is given as 50-100 mT.
• P1: already remedied by the arbiter. A1 (36 uV vs 36 mV): already corrected earlier and not carried into the primer.
• R5 (Investigator's next run) and R7 (optional reads of Investigator files): out of scope for this remediation round; they remain open as follow-up work and are not defects in the reviewed files.

Final: PASS