TernionAG: Arbiter (Step 3)

Inputs read: claims.md, physicist.md, mathematician.md, contrarian.md (all three read in full), facts.json (via claims.md), and the two superseded arbiters in review/_merged/. Verdict scale: the Step 2 scale in prompt.txt. Tags: [read] from a project document (via the reviews' deep reads), [calc] arithmetic done by a reviewer and cross-checked between reviewers, [web] URL, [memory] with confidence. Reader: self-taught undergraduate; symbols are defined where they first appear.

Source gap note (repeated; it limits every verdict below)

• Every Kassner/Ternion scientific claim rests on the patentee's own text. No lab notebook, test report, independent test, replication or Kassner biography is in either source folder. The Glarus register papers are corporate only and never name Kassner or a patent. This silence is itself analysed below ("The record goes dark"); it is not treated as a neutral gap. The Investigator's run (research/, sites.md) has since added a partial biography (Berlin 1931, Arbon 1932, London 1934-36, Zürich 1943, Montana-Vermala about 1951) from the Swiss gazette and patent indexes [web, see research notes].
• The patents are page scans read by OCR; equations and tables were checked against page images only in the deep reads. Not read in depth: GB483062 pp.8-12, all drawing sheets, the rest of the GB417501 claims. GB448418 is not in the folder. Not obtained: the German priority applications, the 1936 general-meeting minutes, any record of who bought the patents in 1945, the GEC Research Laboratories history, Universal Inventions Ltd and Vibranic Ltd records, the Dipole A.-G. Arbon records.
• The IWM transcripts are automatic (Whisper) and error-prone; the Jacobsen Paperclip book and the NARA guide were digest-only; the Guardian "sonic attack" printout has no body text.
• The three Morray-TH notes are earlier AI analyses (secondary), and the essay in the project folder is interpretive. They are not primary records, so their factual statements are checked against the primary documents and cited from those; their arguments are weighed as arguments.
• Conclusions about food, colloid, rubber, seed and tissue effects rest on the patents' assertions, calculation and [memory]. Where I say "mainstream expects X" without a [web] tag, that is [memory].

Web verification (arbiter budget 8; used 4 attempts)

WebSearch is out of quota for this session (200 of 200), so three searches failed (water Debye time, Grundler/Keilmann replication, the 1945 Goodrich/Firestone purchase). WebFetch worked for three pages: - https://en.wikipedia.org/wiki/Dielectric_spectroscopy: confirms only that relaxation (dipole, ionic) is seen at roughly 10^2 to 10^10 Hz and is slow compared with resonances above 10^12 Hz; gives no value for water's relaxation time and says nothing on sharp electrolyte resonances. [web, partial] - https://en.wikipedia.org/wiki/Dielectric_heating: says RF heating of dielectrics "had been proposed in the 1930s", citing a Bell Labs US patent 2,147,689 (1937) for 10-20 MHz heating. This supports the point that dielectric heating was not unique to Kassner. [web] - https://en.wikipedia.org/wiki/Heinrich_Fr%C3%B6hlich: 404, check failed. Update 2026-10-04 (Chrome, Google, 1 search, new tab, closed): a compilation snippet gives blood at 300 MHz as relative permittivity about 65 and conductivity about 1.3 S/m (ResearchGate, "Dielectric properties of fresh human blood", snippet only, paper not opened) [web: https://www.google.com/search?q=blood+serum+conductivity+S%2Fm+dielectric+properties+of+body+tissues+blood+300+MHz]. That is blood, not serum (serum, without insulating cells, conducts somewhat more), but it confirms the order of magnitude of the serum sigma used in claim 1: sigma/(omega eps0) is about 78 at 300 MHz for these values [calc]. The contrarian's check of water's Debye time (8.27 ps, loss peak near 20 GHz) is accepted as [web, reviewer]. The earlier list below is otherwise unchanged. Original list of items then unverified: water Debye tau 8.3 ps (high confidence, and the contrarian's Debye-Stokes estimate gives 8.5 ps independently [calc]); serum conductivity about 1.5 S/m (medium; now supported to the order of magnitude by the blood figure above); the non-thermal microwave literature and the Webb/Grundler/Keilmann reports (medium, replication status unknown); the Goodrich/Firestone purchase beyond the three clippings. No verdict below turns on one of these alone. Retry when quota is restored.

Where the reviewers agreed, disagreed, and how I settled it

Agreement (all three): every claim in section A splits into a mainstream part that is fine and a Kassner-specific part that has never been tested; the plate-resonator theory in GB483062 is correct but the 1932 tuning figures do not follow from it; the closed electrolysis cycle is ruled out by the second law; the suppression thesis rests on an absence of later record, and the reviewers set against it an ordinary documented alternative (insolvency). Under the evidence rules of 2026-10-03 that absence is an anomaly to analyse, not a point scored for either side; it is analysed in "The record goes dark" below.

Disagreements and my rulings:

1. Claim 1, is the sharp-resonance idea "refuted" or "never tested"? The old Ternion-Research arbiter said "refuted by calculation" for a Q of about 50 oscillator in a bulk liquid. The physicist gives the strongest reason against: in serum at 90 cm the conduction term sigma/(omega eps0) is about 81 [calc, sigma memory medium], so a Q of 20 to 50 would be extraordinary. But no reviewer can show the patent claims a dielectric-loss line; it claims an optical readout (lightening between crossed Nicol prisms, a birefringence signal) with no spectrum. A calculation refutes a stated mechanism, and the patent states none ("resonance dispersion" is asserted, "explained below" and never explained). So I rule: Never properly tested, with a strong mainstream expectation against and low plausibility. I do not carry forward "Refuted". Ruling stronger of the arguments: the physicist's and mathematician's (a Debye band has Q about 0.3, full width ratio 13.9 [calc], about 170 times broader than the claimed 2 percent band), but they establish that the two kinds of band differ, not that the second is absent.
2. Claim 4, how total is the inconsistency? The mathematician showed the agar cases (116 cm, 0.5 or 1 cm gap) can be rebuilt if the dielectric fills 68 or 92 percent of the gap, or the glass trough walls matter; the contrarian said dielectric loading cannot repair it (1 mm water in a 1 cm gap gives eps about 1.11, raising lambda only about 5 percent). Both are right for different cases. For the 10 cm / 30 cm and 6.2 cm / 18.6 cm cases the stated geometry cannot give the stated wavelength (all three plate sizes give lambda = 3.0 D, x = 1.05, below the lowest Bessel root 1.841) [calc, three reviewers]. The agar case follows only with unstated loading. Also US2089966 contradicts itself: p.5 says closer spacing lengthens the wavelength, p.7 says it raises the natural frequency [read].
3. Claim 5, "Established" or "Plausible, untested"? Split by object: the existence of 1930s decimetre tube oscillators and radial-line theory is Established [memory high]; the specific tuning tables and the "about 40 percent" are the patentee's own, with no input or output power, so Plausible, untested and Never properly tested respectively.
4. Claim 10, "Refuted" or "Not testable"? The physicist and contrarian say refuted in single-photon form; the mathematician says "bond strength" is undefined. Ruling: split. Single-photon or weak-field change of a bond energy is Refuted by calculation (photon 6.7e-6 eV at 18.6 cm, about 3,850 times below kT and 7e5 times below an O-H bond [calc]; a 1e6 V/m field shifts an atomic level by about 5e-5 eV [calc]). A coherent collective mechanism is Not testable as stated: nobody has defined it.
5. Claim 13, how strongly does temperature tune the band? The mathematician uses water's permittivity coefficient (0.36 percent per K, so 5 K shifts lambda about 0.9 percent), the contrarian the Debye time (about 3 percent per K [memory, medium]). Both agree that a kelvin-level thermostat is attainable, which makes "most experimenters do not control it" a weak excuse. The contrarian's correction that pressure appears only in a measurement claim (claim 14 of US2089966) is accepted; temperature is a real handle, pressure barely one.
6. Electret energy. Physicist 24 J/m^3 (free-space field energy at sigma = 2e-5 C/m^2); mathematician 0.9 microjoule per cm^2 (1 mm wax slab). These are the same order of magnitude (9 J/m^3 for a 1 mm slab) and measure different things. Either way it is tiny and conventional.

Claim-by-claim arbitration and final verdicts

Numbering follows claims.md. "Settles it" gives the single experiment or calculation.

1. Sharp natural frequencies of dipolar substances (electrolytes about 2 percent wide, serum about 90 cm, water 1.28 to 1.5 cm)

• Maths: no equations; Debye part consistent. Water 1.5 cm = Debye loss peak, f = 1/(2 pi tau) = 19 GHz, lambda about 1.6 cm [calc]. The patent attributes the 1.28 cm "pure water" figure to serum (p.8), an oddity.
• Experiment: friction dispersion established; sharp bands in conducting liquids not known to dielectric spectroscopy [memory, medium]; the claim is only a lamp-and-Nicols readout, and the sample "resonance" is never separated from the plate oscillator's own size.
• Tested: no one on record; not shown dismissed. The patent was noticed: abstracted in Chemical Abstracts (1935, 29:1178), cited in chemistry reference works and in The Botanical Review (1944), and summarised by the US Court of Customs and Patent Appeals in In re Arnold (1950) as teaching that "excitation is especially strong and energy absorption especially great under conditions of resonance" [web: research/2026-10-03-internet-archive.md; research/2026-10-03-caselaw-access-project.md]. All of these repeat the claim; none reports a test for or against it.
• If true: a collective ionic resonance with Q of tens in a medium with sigma/(omega eps0) near 80 would be new condensed-matter physics.
• Update 2026-10-04: Kassner's own 1934 French texts (FR 786,294 p.1, FR 788,428 p.1) call his "natural frequencies" Debye relaxation frequencies [web: research/2026-10-04-depatisnet.md]; the two analogue sharp-line tests at 26-90 GHz (Gandhi 1980, Gos 1997) are on record and did not reproduce the claims they tested; see "Investigator findings (2026-10-04)" below. The verdict below gains one sub-split and one reservation.
• Verdict (split): water's Debye band near 1.5 cm Established (now also web-checked: relaxation time 8.27 ps at 25 C, loss peak near 20 GHz [web, contrarian]); sharp "resonance dispersion" in electrolytes, serum, colloids Never properly tested (mainstream expectation against; not dismissed without testing, simply not taken up); the conflation of "friction" and "resonance" under one name "natural frequency" is a definitional weakness, Internally inconsistent only if both are read as one feature.
• Settles it: a thermostatted 0.1-3 GHz permittivity sweep of water, 1 percent NaCl, milk, juice and serum with an empty-cell control, plus a rebuilt crossed-Nicols detector sweeping 20-100 cm. A narrow line tracking the sample is a real anomaly; its absence moves the sharp-band part to "Tested, not supported". Added 2026-10-04: with a Q of 20-50 the line at 333 MHz would be 7-17 MHz wide [calc, mathematician], easily resolved by a modern network analyser, so this is a one-afternoon measurement. Sub-split added (2026-10-04): analogue claims of sharp lines in cells and nucleic acids at 26-90 GHz (Webb/Booth, Grundler/Keilmann) are Tested, not supported (one broad sweep with frozen controls, one concurrent-control repeat; abstracts read by the Investigator only); this is evidence about the class of claim, not a test of Kassner's bands, which remain Never properly tested.

2. A resonant field permanently alters "energy content"; frozen-in orientation (electret) is "richer in energy" and re-radiates on thawing

• Maths: surface charge 0.5-2.0e-9 C/cm^2 (not e-8, checked on page image); stored energy about 1 microjoule per cm^2 [calc, two reviewers]. "Energy content" is defined by exclusion (anything that is not a temperature change), so it is not a thermodynamic quantity.
• Experiment: Kerr-type orientation and electret formation are real; the 12 kV DC field does the poling and no experiment isolates the RF component. Re-radiation "at the same wavelength" has no mechanism and clashes with picosecond relaxation on thawing.
• Tested: none; no no-RF control.
• Verdict (split): orientation and electret storage at microjoule per cm^2 Established (conventional, not resonance-specific); "permanent energy content" gain and same-wavelength re-radiation Not testable as stated; a resonance-specific, controlled version Never properly tested.
• Settles it: electret versus no-RF control at equal poling field and temperature, with a radiometer on thawing.

3. Practical effects: preservation, colloids, catalyst-free curing, seeds, effects on men and animals

• Maths: the reported temperature rises (2-8 C) are consistent with the stated powers, e.g. 574 W for 120 s into about 2 kg gives at most 8.6 K [calc, contrarian; patent figures taken from earlier deep reads], so the data do not look invented. They are far below pasteurisation, so bulk heating does not explain a 28-day milk result, but hot spots in a 1 cm gap are not excluded (no field map, no probe). 250 W into 1 kg of water warms it 3.6 K per minute [calc].
• Experiment: dielectric heating cures rubber, resin and lacquer (established, thermal); the 1930s proposal of RF dielectric heating [web: Dielectric_heating page]. Non-thermal frequency-specific food preservation or growth effects are not established [memory, medium; web failed].
• Tested: by the patentee only, one run each, no sample counts, spoilage criteria, heat-matched or off-resonance controls. The patent itself says in advance that results vary with conditions. The milk 28-day, veal 2-week and apple-juice 4-month figures appear in GB417501 only; US2089966 is qualitative.
• Verdict (split): RF/dielectric curing, drying, vulcanisation Established (thermal); resonance-specific preservation, colloid change, seed growth, catalyst-free chemistry Never properly tested (neither tested-and-failed nor shown suppressed); effects "in men and animals" Not testable as stated (a "may" with no dose or endpoint).
• Settles it: the contrarian's four-arm blind test (resonant, off-resonant at equal absorbed power, heat-only matched to the measured temperature history, untreated), n of at least 20 per arm, bacterial counts, pH, UV-visible over 28 days, preceded by the spectrum in claim 1.

4. Plate pair as a tunable resonator following Bessel-zero theory

• Maths: GB483062 is correct textbook radial-line theory: lambda = 2 pi R sqrt(eps)/x with x a root of J0 or J1 (or of dJ_n/dr); R = 10 cm, x = 2.405 gives 26.1 cm against the quoted 26.2; the worked 40 cm disc gives 32.7 cm at x = 3.83; energy eq. (12) gives 9.0 microjoule and 4.97 kW at delta = 0.3 [calc, mathematician and physicist agree]. Printing slips: Fig. 4 derivative zeros 3.15, 4.3, 5.2, 6.8 should be 3.054, 4.201, 5.318, 6.706; the edge condition is stated as J0 = 0 then J1 = 0. In the empty-gap theory the gap d does not enter the frequency.
• The 1932 figures (10 cm disc at 30 cm, 6.2 cm at 18.6 cm, 39 cm at 116 cm) all equal lambda = 3 D and match no root with eps = 1; GB483062's 40 cm disc at 32.7 cm is 3.5 times shorter than GB417501's 39 cm disc at 116 cm. The claims.md wording "radius 10 cm" versus the 1932 "diameter 10 cm" must not be set side by side as one disc.
• Tested: patentee says "confirmed by experiment", no data. GB483062 was a void application (administrative).
• Verdict (split): Bessel mode theory (GB417564, GB483062) Established; the 1932 tuning rule and spacing statements Internally inconsistent with the same author's equations for the stated geometry (and with each other, US2089966 p.5 vs p.7), with the agar cases rescuable only by unstated dielectric loading, the cause unknown (lead inductance, generator tuning, or a rule of thumb are reviewer guesses, not in the source); the claim that the sample, not the apparatus, is what resonates Never properly tested.
• Added 2026-10-04: GB 435,643 (divided out of GB417501) is a plate-resonator light detector whose frequency shifts with the dielectric of a layer; DE 1,128,930 (1960) uses plate resonators for rotating fields. The identity J0'(x) = -J1(x) it relies on is correct (first zero 3.832) [calc, mathematician]; whether the field actually rotates cannot be confirmed from the machine-translated description alone, and its acceleration and "Kernfusionsforschung" statements carry no energy, current or yield figure: Not testable as stated. None of this tests the 1932 lambda = 3 D figures; the verdict above stands. The detector shows that sample and apparatus are entangled in one device, which supports the point of the settle-it test below.
• Settles it: a field-solver or bench model of the 6.2 cm disc, 1 cm gap, 1 mm water layer, and a measurement of its resonance with and without the sample. A calculation that reproduces 18.6 cm only with the generator's leads included would show the "sample resonance" was the apparatus.

5. Valves, coaxial resonators, wavemeter, polarimeter, tuning tables 0.30-4.9 m, "about 40 percent" efficiency

• Maths: circuit analysis dimensionally and numerically correct (agrees to 0.15 percent). Table 1 lambda/l is 1.5-1.9 (not the 2 or 4 of a simple line), consistent with transit-time (Barkhausen-Kurz) operation [memory, medium]; Table 3 ratios fall from 12.9 to 7.9. The 40 percent is for one two-valve push-pull circuit (Table 4) only, with no input or output power and no definition.
• Verdict (split): general feasibility of 1930s decimetre generation Established [memory, high]; the tuning tables Plausible, untested (self-reported); "about 40 percent" Never properly tested (undefined, unmeasured).
• Settles it: rebuild one circuit and measure DC input against calorimetric RF output.

6. Hand-held Tesla applicator and lamps (0.5 mm upward; 0.3-1 mm discharge path; 1951 filament-and-rod lamp)

• A spark gap and a hot filament emit broadband radiation; a narrow line at 0.5 mm (600 GHz) would be the surprise, and none is claimed with data. No dose, spectrum, efficiency or detection in any of FR732805, GB383508, US1951731, US2104697, US2558568. US2104697 is assigned to Universal Inventions Ltd, London, not Vibranic.
• Verdict (split): broadband emission Plausible, untested; therapeutic or lamp-efficiency gains Never properly tested; as stated (no dose, spectrum, endpoint) Not testable as stated.
• Settles it: spectrum and power of the spark array and lamp with oscillation on versus shunted.

7. Commercial significance: 1945 Goodrich/Firestone purchase; Voss 1987 "foresaw the microwave oven"

• Three 1945 clippings (page images) say the patents "originally were granted on discoveries made by" Dufour, Le Duc and Kassner; no seller, price, patent number or frequency; the process they describe is heating (sponge rubber cured in about one-eighth of the time). That description comes from the buyers' announcement. Voss (editor, J. Microwave Power, 1987) wrote that Kassner "did some remarkable experiments", that he "'sketched' (perhaps that is the key word) our subject", and "In truth he foresaw both the microwave oven and the industrial electromagnetic energy applicators"; he then asked "Who was Ernst Kassner?" [read: deep/Ternion-AG-JMPEE-Vol22-Pg49-Voss-0d19a3.md, printed p.56]. The remark mixes a hedge ("sketched") with a plain statement ("foresaw"). RF dielectric heating was being proposed independently in the 1930s [web: Dielectric_heating page, Bell Labs US 2,147,689, 1937].
• New (Investigator run, 2026-10-03): a B. F. Goodrich Research Center paper (Miller and Wittig, Rubber World, Oct 1950) names the Kassner patents in the portfolio as US 2,089,966 and US 2,109,843; the US agent in the chain was H. H. Giodvad Grell, who in Nov 1945 sold unnamed "patent rights" for Fr. 200,000 to Vibranic Ltd, Glarus, chaired by Max Stoffel, Ternion's founder [web: research/2026-10-03-internet-archive.md; research/2026-10-03-e-periodica-shab.md].

Two descriptions of the same patents, side by side.

• Checks on the Goodrich wording: "about 150 million cycles" is lambda = 2 m. In GB417501's milk example the five bands are 455, 355, 283, 163 and 27.5 cm; the four longest carry 573 of the 573.6 W, at 66 to 184 MHz [calc from read figures, QE revision], so Goodrich's figure is fair for where most of the stated power went. It does not describe the "natural frequencies" the patent rests its case on: serum near 90 cm (333 MHz), fruit juice near 40 cm (750 MHz), water near 1.3-1.5 cm (20-23 GHz). The summary is accurate as far as it goes and silent on the part Kassner claimed as new. Against Kassner's non-thermal reading, equally stated: the 1945 announcements, Science News Letter and Electronic Industries all describe the process as uniform heating [web: research note], and GB417501's food examples report temperature rises of 2.5 to 4.5 K (claims 2-3 above). For it: Rubber World (Apr 1945), reporting the same purchase, adds that "the excellent results obtained by electronic vulcanization are due not only to rapid uniform heating, but also may be connected in some way with electronic power itself" (no data) [web: https://archive.org/details/sim_rubber-world_1945-04-01_112_1]; and a later inventor, Orlan M. Arnold, claimed a sub-thermal effect at the dispersion frequency, with Kassner's patent cited against him as prior art (In re Arnold, 1950; Arnold v. Watson, 1955, decided on "evidence") [web: research/2026-10-03-caselaw-access-project.md].
• Is the Goodrich account wrong? There is no evidence that Goodrich's staff knew of non-thermal effects and described the patents otherwise, so this review does not call it a misrepresentation. It is reported as the characterization of a party with a stake, not as the neutral fact of what the patents do. The records that would show what Goodrich knew: Goodrich's internal test data on the Kassner process (B. F. Goodrich company records; their present location is [memory, medium] the University of Akron archives, unverified); the 1945 purchase and due-diligence files of Goodrich, Firestone and Grell, including the Vibranic Kauf- und Abtretungsvertrag of 15 Nov 1945 (Landesarchiv Glarus); and the Arnold v. Watson trial exhibits and testimony (US District Court for D.C., about 1953-55).
• Added 2026-10-04: patent-office treatment of Kassner as prior art is documented: US 2,495,429 (Spencer, Raytheon, filed 1945) lists "2,094,602 Kassner" among References cited (USPTO scan, read by two reviewers); US 2,833,657 (Raytheon 1954) cites US 2,089,966; 32 later UHF-tube patents cite US 2,109,843 [web: research/2026-10-03-google-patents.md]. Whether the examiner or the applicant made each citation is not stated. These patents frame their inventions as heating, so the citations support Voss's historical point (priority of multi-frequency applicators), not the resonance mechanism. Open anomaly, recorded not concluded: Kassner is named in the January 1945 purchase release but not in the Industry Inventions, Inc. licensing announcements (Aug 1945), which name "Leduc and Dufour"; why is not documented. Also open: whether any Kassner patent was in the Fr. 200,000 Vibranic purchase from Grell (only US 2,558,568, filed 1946, is assigned to Vibranic).
• Added 2026-10-04 (QE remediation; physicist D2, mathematician 4a, contrarian lead 14; sites.md lead 14): what the Fr. 200,000 purchase contained is partly documented by matching priority dates. Vibranic's FR 926,351 (filed 6 May 1946) claims the same priority date, 22 Apr 1943, as US 2,415,025 (H. H. Giodvad Grell and Harry W. Richards, "Method and apparatus for treating matter in a high-frequency electric field", granted 28 Jan 1947, assigned "by direct and mesne assignments" to Firestone and Goodrich; its text is about uniform heating). Vibranic's CH 250,508 ("Verfahren und Vorrichtung zur Ultrakurzwellenbehandlung", quasi-optical waves under 10 m radiated into a reflecting chamber for "Erwärmen, Altern, Vulkanisieren") shares its priority date, 4 May 1945, with Grell's own CH 254,655 [web: research/2026-10-03-google-patents.md; US 2,415,025 confirmed on the USPTO scan, files/US2415025.ocr.txt]. So the "Patentrechte" Grell sold to Stoffel's Vibranic in Nov 1945 included at least the non-US rights to Grell's own heating patents, whose US rights were held by Firestone and Goodrich. No document yet shows a Kassner patent in that package; the Vibranic purchase contract (Glarus register file) would decide. This bears on claim 7 and on explanation 3 ("continuation under other names") in "The record goes dark": it documents continuity of Stoffel's patent dealings in the same field, not continuity of Kassner's patents.
• Verdict (split): the purchase was reported Established (as a press record; the 1950 paper now names US 2,089,966 and 2,109,843 in the portfolio; the physicist's proposed label "Partly documented" is not on the Step 2 scale, and its content is carried by this sentence); that the purchase validates Kassner's resonance mechanism Not testable as stated, because a purchase is not a test in either direction: the buyers' announcement describes heating, which would account for the commercial use, and one trade report hints at more, and no document in hand gives data either way; Voss's "foresaw" Historically superseded but instructive (an early multi-frequency applicator disclosure, not a microwave oven).
• Settles it: Goodrich and Firestone patent assignment records (USPTO assignment database) for the 1945 transfer, plus the purchase files, Goodrich internal test data and the Arnold trial exhibits listed above.

8. Ternion AG company history

• Mathematician re-added the register arithmetic: 100 shares at Fr. 1,000; vote caps (one fifth, 1/5 of 54 = 10.8 rounded to 10) consistent; dissolution quorum met and unanimous (16 of 16); the 31 Dec 1936 balance sheet balances at Fr. 1,196,535.05, research account Fr. 1,120,380.15 (93.6 percent of assets); written off, assets of Fr. 76,154.90 against loans of Fr. 1,096,535.05 leave a deficit of Fr. 1,020,380.15, 10.2 times share capital [calc]. That is a plain insolvency record. Kassner is never named; only US1951731, GB417501 and GB417564 tie his patents to Ternion.
• The board's own stated reason (20 Dec 1937): "owing to a complete lack of further funds the company can no longer be continued. Since part of the patents, whose value the research account represents, has already lapsed because it was impossible to pay the costs needed to maintain them, and since ... an industrial exploitation of the remaining, still existing patents is impossible in the foreseeable future", liquidation is "the only means of preventing a bankruptcy" [read: page image AG2 p.10, deep read translation]. This is the board's account to its shareholders; it is the best document on why Ternion ended, though no auditor's or creditor's record has been found to check it against.
• New 2026-10-04 (Dipol timing): three Thurgau newspapers (Thurgauer Zeitung 30 Dec 1933, Bote vom Untersee und Rhein 2 Jan 1934, Der Landbote 3 Jan 1934) name "die Aufhebung der Dipol A.G." as one of three reasons for a fall in Arbon's 1933 tax income [web: research/2026-10-03-e-newspaperarchives.md, e.g. https://www.e-newspaperarchives.ch/?a=d&d=TGZ19331230-01.2.6.1]. The Investigator, and two reviewers, read this as the Arbon laboratory ceasing about 18 months after founding, with the formal liquidation entry only in 1938. That reading is an inference from a tax-revenue remark, not a notice of closure, and "Aufhebung" could also describe something short of the operation ending; it is the best lead, not a settled fact. If it holds, then the Fr. 1,120,380.15 research account at end 1936 (financed by loans) was not spent in Arbon after 1933: Dipol's loan account on Ternion's balance sheet is only Fr. 51,155 [read: note-14 deep read, cited by the contrarian; I did not re-check the page]. Competing explanations for the rest: patent and London work 1934-36 (Kassner is "résidant en Angleterre" on FR 788,428, filed 27 Sep 1934 [web: DEPATISnet note]); a book valuation of the patents (the board's 1937 minutes say the research account "represents" patent value); other spending. No document in hand separates these. The capitalised figure is the company's valuation, not an audited expenditure; no misvaluation is alleged. The 17 Jul 1936 minutes are the record that would decide. Dipol's signing chemist, Dr. Alfred Wiegner (b. 1905), was an analytical and dye chemist by his ETH dissertation (carbides and silicides in alloys) and diploma (dyes), not a dielectrics specialist [web: research/2026-10-04-records-and-rechecks.md]; that bears on what the laboratory could have measured, and it is weak evidence only.
• Verdict: Established as history (documents; outside the physics scale; the founding, dissolution, 1938 creditor calls and 1952 deletion are now confirmed independently in the Swiss gazette SHAB [web: research/2026-10-03-e-periodica-shab.md]). The failure is financial; it is not evidence that the effects are real or false.
• Open item (contrarian), now partly answered: Dipol A.-G., Arbon, was founded the same day (18 Apr 1932) by the same man (Stoffel), "to maintain a laboratory" for electro-physical, electro-chemical and electro-technical processes, with factory premises on the Philosophenweg, a chemist with signing authority (Dr. Alfred Wiegner) and a trademark for "Elektrische Apparate"; it was wound up alongside Ternion in 1938 [web: SHAB, research note]. That Dipol was the laboratory where Kassner worked is an inference, not stated in any notice.

9. Bauch did a Ternion UHF project during a 1934 GEC secondment

• The document: the 1996 memoriam in the Journal of the Audio Engineering Society (vol. 44 no. 6) says that in 1934 his work "led to his recommendation by the Hertz Institute for a secondment to the G.E.C. Research Laboratories in the U.K. While there he worked on a project for the Swiss Ternion company investigating UHF and decimeter techniques." [read: _text/Ternion-AG-press-FWO-Bauch-Memoriam-4ccce7.txt p.1; the sentence spans two columns and is reassembled in deep/Ternion-AG-press-FWO-Bauch-Memoriam-4ccce7.md]. It is a published obituary in a professional journal, written after his death; its own source (family, company papers, Bauch himself) is not stated.
• The other accounts: the IWM interviews (Bauch, about two hours; his wife), the Swiss Sound 1983 article and the Wireless World author note of Feb 1953 all confirm the stay at the GEC laboratories at Wembley ("I left in 1934 for England ... one and a half years ... the North Wembley Laboratories" [read: IWM part 1, machine transcript]; "a short spell at the G.E.C. Research Laboratories, Wembley" [web: https://archive.org/details/sim_wireless-world_1953-02_59_2]). None of them mentions Ternion, Kassner or UHF, and none says what project he worked on at GEC. Silence is not contradiction: none of these accounts denies a Ternion project, and the interview transcripts do not show that he was asked about one. The 1953 note calls his Hertz Institute work "supersonics"; that describes the period before GEC, not his work at GEC.
• Circumstances that fit the sentence (consistency, not a second source): Kassner was "a German subject resident in London" on his British patents of 1934-36 (GB448419 filed 31 Aug 1934 from 329 High Holborn) [read: digests], so Ternion's inventor was in London during Bauch's 1934-35 stay; both came from Berlin; and the memoriam's "UHF and decimeter techniques" matches the subject of Kassner's British filings of 31 Aug 1933 (GB448418, "Electric discharge devices and circuits for the generation of short electromagnetic waves"; GB448419, ultra-short-wave generator) [web: research/2026-10-03-internet-archive.md; read: digests]. Against over-reading it: the essay's wording, Bauch "an employee of GEC" who "collaborated jointly" with Ternion, goes beyond the memoriam, which says only that he worked on a Ternion project while seconded there.
• Verdict (split; revised 2026-10-03, see "Evidence-rule revision"): the GEC secondment (about 1934 to end 1935) Established (four accounts, 1953 to 1996); the Ternion UHF project documented in one published source (the 1996 JAES memoriam); no second source found yet; the other accounts are silent on it, not contrary (history, outside the physics scale). It is evidence for the claim, of the weight of a single obituary sentence with an unstated source.
• Settles it: the GEC Research Laboratories project lists and reports for 1934-35 (Hirst Research Centre archive), any Ternion contract or payment to GEC (Ternion's 1936 general-meeting protocol and books, Glarus; Dipol A.-G. file, Staatsarchiv Thurgau), Bauch's own naturalisation file (UK National Archives HO 405, which would list his UK employment), and the memoriam author's source (AES archives).

10. Natural-frequency fields can raise or lower covalent and ionic bond strengths (essay's claim; not in any Ternion text)

• Verdict (split): single-photon or laboratory-strength-field change of a bond energy Refuted by calculation (numbers above); a coherent collective mechanism Not testable as stated.

11. Closed electrolysis-and-burning cycle or chemical heat pump yields net energy; electrolysis assisted by a resonant standing field (displacement current), or by radiated RF (essay and Morray note 12; never claimed by Ternion)

• (Reworded 2026-10-08 on the review direction of 2026-10-07: Ternion applies energy mainly by displacement current through the sample in a plate resonator (Kassner's "standing field", US 2,089,966) and also by radiated RF through a wand applicator ("radiating field"). Earlier text here said "RF-assisted electrolysis"; that wording was too narrow and is replaced below. See "Investigator findings (2026-10-08)".)
• dG = 237.1 kJ/mol, dH = 285.8, T dS = 48.7, cell voltages 1.229 V and 1.481 V [calc, memory high]; the note's COP = dH/(W_in) is bounded at dH/dG = 1.205. COP "1.2 to 1.4" has its upper end beyond the second-law limit; COP above 1 below 1.48 V is ordinary, so the note's bench diagnostic is wrong; its Landauer argument runs backwards (kT ln 2 = 1.73 kJ/mol); its 600 GHz and 5 THz bands are 25 to 200 times shorter than Kassner's water bands and its "libration" label is wrong [memory, medium]. The patents never mention electrolysis.
• Verdict (split): closed cycle or W_total below dG Refuted by calculation; COP up to 1.2 by drawing T dS from surroundings Established and not novel; electrolysis assisted by a resonant standing field (displacement current) or by radiated RF, as a rate effect, Plausible, untested with the field power counted as input (SUPERSEDED 2026-10-08: split into catalyst-heating effect Established, net gain Plausible, untested, Kassner-type standing or radiating field Never properly tested; see the 2026-10-08 section); the note's COP "ceiling" and diagnostic Internally inconsistent.
• Settles it: isothermal calorimetry of a cell with total electrical plus RF input recorded; any result below 237.1 kJ/mol total work would be the anomaly.

12. Weaponised binding energies; defence interests suppressed Ternion

• The weapon premise inherits claim 10.
• The suppression thesis rests on the record going silent: no papers, conferences or later work by Ternion's scientists, and Bauch's Ternion project absent from his later accounts. Under the Absence-of-evidence rule that silence is analysed in its own section ("The record goes dark", below), with the competing explanations. Stated here, for and against:
• Against (documented): the board's 1937 minutes give a financial cause ("complete lack of further funds", patents lapsed for unpaid fees) and the balance sheet shows a deficit 10.2 times share capital [read: AG2 pp.8-10]; the Ternion patents were openly published, abstracted in Chemical Abstracts, cited in reference works and argued over in US courts in 1950 and 1955; the 1945 buyers announced they would license the patents "on a reasonable basis" [web: research notes]; Kassner and Stoffel carried on openly under other company names (Vibranic Ltd, General Infra Co.) recorded in the Swiss gazette [web: SHAB note]; the Marconi-death documents (1982-91) never mention Ternion or Bauch [read]; no document in the folders shows blocked publication, withheld funding, a secrecy order or seized data.
• For, or not excluded: the silence itself, which is real and unusual for a company that capitalised Fr. 1.12 million of research; financial failure and outside pressure are not exclusive (funds can be withdrawn), and the 1937 minutes name no lenders, so who stopped funding, and why, is not recorded [read: AG2 pp.6-11]; the decimetre work Bauch is said to have done in 1934-35 at GEC was in the field that became radar, where secrecy is documented in general [memory, high], though no document ties it to Ternion. The open publication of the patents counts against suppression of the patents; it says less about any unpublished research results.
• Added 2026-10-04: the premise that Kassner's record goes dark is wrong for the man and right for the research. Kassner is documented in patents from about 1921 to 1960 and is named as deceased ("†, München") on the 1962 print of DE 1,128,930, applicant Wilma Hencke, filed 14 Apr 1960 [web: DEPATISnet note; the dagger convention is [memory, high]]. Other traces: a 1937 folder "Phono-Vision, Ernst E.W. Kassner" of "Fotos und vertrauliche Firmenunterlagen" (KB Appenzell Ausserrhoden), a 1938 visiting card to Hans Kayser (Swiss Literary Archives), and General Infra Co AG, Chur (1943-53). One security-context item is new: a Swiss Federal Archives file titled "General Infra Co AG, Chur", 1946-48, in the "Schwarze Liste" series, plus police dossiers on "Kassner, Ernst (20.11.1900)" (identity with E. E. W. Kassner unconfirmed). The file titles are all that is known; they do not show that General Infra was blacklisted, and a German national running a Swiss infra-red firm in wartime fits ordinary economic-warfare record-keeping. They are the first government record touching the group, and their contents are unread. I weigh them as neither for nor against suppression.
• The essay's "American General Electric owns BAE Systems" is wrong.
• Verdict (split; suppression part revised 2026-10-03): suppression thesis Never properly tested (as history: it is a testable claim, and the records that would decide it have not been searched; the documents in hand show a financial end and open later commerce, and none shows suppression); weaponisation premise Refuted by calculation for single-photon or weak-field bond change (claim 10), with a collective mechanism undefined; the BAE statement is a factual error (outside the scale) (QE revision).
• Settles it: the records listed under "The record goes dark", above all the identity of the lenders and the 1936-37 protocols (Glarus), the Dipol A.-G. file (Thurgau), GEC Research Laboratories reports 1934-35, UK and Swiss government files on Kassner (a German national in London 1934-36 and Zürich 1943), and any patent secrecy direction on Kassner's applications.

13. Effects hard to replicate because natural frequencies shift with temperature and pressure

• No replication attempt has been found in the folders or in the searches so far (Internet Archive, Europe PMC, UK National Archives, SHAB; sites.md), so no failures are on record to explain. That silence is itself an anomaly: a patented process cited in the 1935-1955 literature and bought in 1945 left no recorded test. Competing explanations: never attempted; attempted and not published (for example in-house by the 1945 buyers); trade secrecy; the patentee's instruments not reproducible from the specifications. Deciding records: replication or test data in the Goodrich or Firestone files, and the Arnold v. Watson exhibits (see "The record goes dark"). The explanation has no tuning protocol, and the patent itself warns in advance that results vary with conditions. Temperature is a genuine tuning handle; pressure barely. Update 2026-10-04: the temperature handle depends on which band is meant. Under the relaxation reading (Kassner's French texts; Arnold), tau of water falls from about 17.7 ps at 0 C to 8.3 ps at 25 C [memory, medium, not checked], about 3 percent per K [calc], so a 2 percent band would need control to about 0.7 K; but a true Debye hump (Q about 0.3) needs no temperature control at all, a further sign that the sharp-band claim and the relaxation claim are different things. Orlan Arnold (US 2,522,082, filed 1945) states that his critical frequency shifts with temperature, concentration and voltage and must be tracked; that is the tuning protocol Kassner lacks [web: research/2026-10-04-arnold.md]. With the claimed Q of 50 a few kelvin of drift is a large fraction of the band, which a thermostat meets.
• Verdict: Not testable as stated; a falsifiable form exists (measure each sample's spectrum at its temperature first), and then it is Never properly tested.

The record goes dark

The anomaly. Ternion A.-G. was founded on 18 Apr 1932 for "participation in electro-physical enterprises and laboratories" and by 31 Dec 1936 had capitalised Fr. 1,120,380.15 of research, 93.6 percent of its assets, financed by Fr. 1,096,535.05 of loans [read: AG2 pp.6-8; web: SHAB 1932]. It produced a large patent family (GB, US, FR and CH members, at least 15 filings including GB 435,043 and FR 786,294-297 found by the Investigator) [read; web: research notes]. Yet no research paper, white paper, conference talk, test report or later scientific work by Kassner or any Ternion or Dipol scientist has been found: not in the folders, not in the Internet Archive full-text search, Europe PMC, the UK National Archives catalogue or the Swiss gazette [web: sites.md, queries listed there]. Voss asked in 1987 "who was Kassner" [read: JMPEE deep read]. For a funded research company this silence is unusual and is treated here as an anomaly to explain, not as a neutral gap. Revised 2026-10-04 (runs 2-3): the silence is narrower than first stated. It is a silence of technical reports, notebooks, papers and conference talks. It is not a silence of the man or the company group: Kassner is documented from about 1921 to his death by May 1962 (patents, 1937 and 1938 archive items, General Infra 1943-53), and the Arbon laboratory (Dipol A.-G.) appears to have ceased about 18 months after founding (see claim 8). Two limits on its size: the searches so far are runs 1-3 of the Investigator (Espacenet and ScienceDirect blocked by CAPTCHA; several archive catalogues not searchable online), and the patents themselves are a public disclosure that was read and cited (Chemical Abstracts 1935, Thorpe, Weissberger, Ellis, The Botanical Review 1944, US courts 1950 and 1955).

Competing explanations, with the evidence for and against each.

1. Commercial failure. For: the board's minutes of 20 Dec 1937: "owing to a complete lack of further funds the company can no longer be continued. Since part of the patents ... has already lapsed because it was impossible to pay the costs needed to maintain them, and since ... an industrial exploitation of the remaining, still existing patents is impossible in the foreseeable future", liquidation is "the only means of preventing a bankruptcy" [read: AG2 p.10]; the deficit after write-off is 10.2 times share capital [calc]; Dipol A.-G. was wound up alongside it in 1938 [web: SHAB]. Against, or what it leaves open: failure explains why the company ended, not why five funded years (1932-37) left no technical report or paper; the lenders are unnamed, so why funding stopped is not recorded; and the technology was not abandoned (Kassner patents sold in 1945, Kassner still working in 1943, 1951 and 1960; see 3). New: Dipol's apparent closure in 1933 (claim 8) fits failure of the laboratory earlier than the holding company; and General Infra's in-kind rights were returned and its shares cancelled on 6 Jan 1944, a second documented unwinding in the group [web: SHAB 1944, research/2026-10-03-e-periodica-shab.md]. The minutes are the board's account to shareholders; no outside check on them has been found.
2. Trade secrecy (patents as the only intended publication). For: Ternion's stated purpose was participation and exploitation, not science; its sole director (Stoffel) and its Glarus domicile holder (Oertli-Heer) are both entered as "Kaufmann" (merchant) [web: SHAB]; its value was carried as "the patents" [read: AG2 p.10]; the 1937 minutes look for "industrial exploitation", not publication; GB417501 was filed by a London agent ("My foreign correspondents have now found ..."), the usual form for a communicated patent [read; memory, high on the formula]. A company selling a process has a reason not to publish beyond the patent. Against: no document states a secrecy policy; secrecy does not explain why no paper appeared after the patents were public or after liquidation. Deciding record: lab notebooks or reports (Dipol file; Wiegner's papers), any confidentiality clauses in Ternion's or Vibranic's contracts.
3. Continuation under other names. For (documented): Dipol A.-G., Arbon (same founder, same day, laboratory and factory, chemist Dr. Alfred Wiegner, liquidated 1938) [web: SHAB]; Universal Inventions Ltd, London, assignee of Kassner's US 2,104,697 (1938) [read]; General Infra Company AG, Chur (1943-53), with "Ingenieur Ernst E. W. Kassner, deutscher Staatsangehöriger, in Zürich" on the board, exploiting "a new physical principle" for infra-red and adjoining micro-ultrashort-wave emitters [web: SHAB 1943]; Vibranic Ltd, Glarus (1945-52), chaired by Max Stoffel, buying "patent rights" for Fr. 200,000 from Giodvad Grell, the US agent in the 1945 Goodrich/Firestone sale, and later assignee of Kassner's US 2,558,568 (1951) [web: SHAB 1945; read]. So part of the darkness is an artifact of searching under "Ternion": the people and patents continue to 1953 under other names. Against: these companies also left no scientific publication found so far; General Infra's subject (infra-red emitters) is not the dipolar-resonance work; which patents Grell sold to Vibranic is not stated. Added 2026-10-04: Kassner's last document, DE 1,128,930 (1960, plate resonators for rotating fields), passed to a private applicant (Wilma Hencke, München; relation unknown). Deciding records: the Vibranic purchase contract of 15 Nov 1945 and register file (Landesarchiv Glarus); General Infra file (Staatsarchiv Graubünden); Dipol file (Staatsarchiv Thurgau); BT 31 file of Universal Inventions Ltd (Kew).
4. Work done inside an industrial laboratory (Bauch at GEC). For: the 1996 memoriam says that at the G.E.C. Research Laboratories "he worked on a project for the Swiss Ternion company investigating UHF and decimeter techniques" [read]; Kassner was in London 1934-36 [read: digests]. Contract work in a company laboratory would be reported internally, owned by the client and not published, which would fit the silence. Against: one source only so far; Bauch's interviews and 1953 note are silent on it (not contrary); no GEC report has been found. Deciding records: GEC Research Laboratories reports and project lists 1934-35 (Hirst archive); Ternion's 1936 protocol and books for payments to GEC; Bauch's HO 405 file.
5. Wartime or classification secrecy (including Operation Paperclip). For: decimetre generation in 1934-39 was the technology of early radar, where secrecy is documented in general [memory, high]; Kassner was a German national in London 1934-36 and in Switzerland during the war; his whereabouts 1938-43 are not documented; Bauch spent "the latter years of the war ... on research on high-frequency pulse technique" in Germany [web: Wireless World 1953]; the JIOA 1947 list holds many "High Frequency" specialists [read: Paperclip deep read]. Against: E. E. W. Kassner, Bauch, Hoffmann, Stoffel and the other Ternion names are absent from the NARA Foreign Scientist Case Files index and the JIOA list (only Rudolf Kassner, jet engines, Wright Field) [read: deep/operation-paperclip-foreign-scientist-case-files-191836.md]; index absence is weak evidence, and a man living in Switzerland would not be expected there. Kassner's patents were published, not withheld, and were cited openly in US courts; Kassner was publicly registered in Zürich in 1943. Added 2026-10-04: Swiss Federal Archives E2001E#1968/78#9311* ("General Infra Co AG, Chur", 1946-48, series "Schwarze Liste", reading of the series [memory, medium], unverified) and police dossiers on "Kassner, Ernst (20.11.1900)" are the first government security-context records; contents unread. The Investigator found General Infra in no online Proclaimed or Statutory List text; BIOS Final Report 866 "High Frequency Heating" has no online copy. This is context for explanation 5, not support for it. Deciding records: UK patent-office secrecy directions on his 1933-36 applications; the three Swiss Federal Archives files above; Home Office/MI5 aliens files and Air Ministry or Admiralty files on Kassner 1934-36 (Kew); Swiss Federal Archives files on foreign nationals; FIAT/BIOS/CIOS reports on German high-frequency heating; the unresolved 1941 SHAB hit for "Kassner Ultrakurzwellen".
6. Suppression (deliberate prevention of the work by an outside interest). For: the silence itself; funding withdrawal is consistent with it as well as with failure, and the lenders are unnamed; the later buyer of the patents (Goodrich) described them in heating-only terms that also served its own position (claim 7). Against: no document shows blocked publication, a withdrawn grant, a threat or seized records; the patents were openly published, abstracted, cited and licensed "on a reasonable basis" [web]; the same people continued openly in new companies for 15 years; Kassner's record runs to a documented death in 1962 with no sign of removal from public filing. Not excluded: the unread Swiss files and the unnamed lenders. Deciding records: the lenders' identity and correspondence; Stoffel's books, which he was to keep "for 10 years in his archive" after 1952 [read: AG3 p.9], so possibly in a family or St. Gallen archive; the files under 5.

Weighing, on the evidence in hand. Update 2026-10-04: the new documents move the weights modestly. The silence of Kassner's own record is replaced by a silence only of research reports, which makes "the man was removed" less natural, while the Dipol timing and General Infra's 1944 unwinding add to the failure explanation for the group. Trade secrecy gains a little from the "confidential company papers" folder (an advertising file, so ordinary business confidentiality). Wartime context gains one unread file title and nothing more. Commercial failure is the best-documented explanation for the end of Ternion A.-G. itself (1937). Continuation under other names is documented and shows the record did not stop in 1937; it moved. Neither explains why no technical report or paper from 1932-37 survives. Trade secrecy and contract work at GEC would explain that, and the memoriam sentence is the one document that points to the second. Classification and suppression are neither shown nor excluded; the documents in hand weigh somewhat against them (open patents, open later companies, absence from the Paperclip indexes) but do not reach the unpublished research, which is exactly what is missing. These explanations are not exclusive. The records named above would decide between them; they are added to "What would settle it".

Table of deciding records.

Corrected claim list (replaces claims.md where different)

1. (Updated 2026-10-04: his 1934 French patents FR 786,294 and FR 788,428 describe the natural frequencies as Debye relaxation frequencies, only a few pages read; so the "2 percent" sharp type rests on GB417501 and its twins alone.) Kassner reported two kinds of dispersion in 1932 (GB417501 itself gives two values for water: 1.5 cm friction band and 1.28 cm "characteristic of water"): the broad Debye "friction" band of water near 1.5 cm (correct physics) and a sharp, 2 percent-wide "resonance dispersion" in electrolytes and serum, seen only as a birefringence signal, with no spectrum. Water's 1.28 cm is attributed in the patent to serum.
2. The "energy content" and electret re-radiation idea is defined by exclusion; the electret part is conventional and stores about a microjoule per cm^2.
3. The practical-effects examples are single runs; only the bean example mentions a control ("similar untreated seeds", no numbers), none states a storage temperature or sample count, and the milk and veal results include hourly re-exposure; only the thermal curing and drying parts are established. The numerical shelf-life figures are in GB417501 only.
4. GB483062's Bessel-mode theory is correct; the 1932 tuning rule (lambda = 3 D, gap-dependent) does not follow from it for the stated geometry.
5. Apparatus patents are plausible 1930s UHF engineering; "about 40 percent" refers to one circuit and is undefined.
6. Applicator and lamp patents give no dose, spectrum or efficiency.
7. The 1945 purchase is a press record about RF vulcanisation; the buyers described it as heating, and Goodrich's 1950 paper treated Kassner's US 2,089,966 and 2,109,843 as marginal heating prior art. That is the characterization of an interested party (buyer and licensor), set beside Kassner's own statement that his object "is not to produce a rise of temperature". A purchase neither validates nor refutes the resonance mechanism. Voss (1987) credited Kassner with foreseeing the microwave oven and industrial applicators, while calling it a "sketch". US examiners or applicants cited his patents from 1945 on (Spencer/Raytheon, Copson/Raytheon, 32 UHF-tube patents), always as heating or apparatus art.
8. Ternion AG is a documented insolvent holding company with a same-day sister laboratory, Dipol A.-G., Arbon (both liquidated 1938); Kassner is not named in Ternion's records. Ownership changed between 1932 (Stoffel 60, Hoffmann 40 shares) and 1937 (48 and 6 represented; 46 shares unaccounted for). Newspapers of Dec 1933 to Jan 1934 suggest Dipol ceased operating about 18 months after founding; the research account still stood at Fr. 1.12 million in 1936, and where it was spent is undocumented.
9. Bauch's GEC stay is documented in four accounts (1953-1996). His work there on "a project for the Swiss Ternion company investigating UHF and decimeter techniques" is documented in one published source (1996 JAES memoriam); no second source found yet; the other accounts are silent on it, not contrary (history, outside the physics scale). He sold an "Improved Method of Radiation Therapy" to Marconi for 80 pounds in 1934-35 [read: memoriam], a lead, not a link.
10. (New, others' claims, not Kassner's.) Orlan Arnold's "critical frequency" at the centre of the dispersion range with surface activation "without destructive heating" (US 2,522,082, filed 1945): refused as "too broad and indefinite", not tested on the record seen; his "selective surface activation" has no stated mechanism: Not testable as stated; the broad Debye-centre idea is Never properly tested at his frequencies.
11. (New.) Heller and Teixeira-Pinto (Nature 183:905, 1959; US 3,095,359): 27 MHz pulsed field, chromosome aberrations in garlic root tips with "No temperature increase" noted, frequency, pulse and axis dependence; no counts, controls or statistics; pearl-chain orientation by interfacial polarisation is conventional and broad in frequency. Chromosome effect: Never properly tested (no replication found; Mittler 1976 negative in Drosophila at 2 and 10 m is a different organism, title only).
12. (New.) DE 1,128,930 (1960): see claim 4; Not testable as stated for rotation, acceleration and fusion. 10-13. The essay's claims (bond strength, net-energy cycle [claim 11 wording, 2026-10-08: "electrolysis assisted by a resonant standing field (displacement current), or by radiated RF"; verdict split in three parts, see the 2026-10-08 section], suppression and weaponisation, temperature/pressure explanation) are not Ternion's; their verdicts are above. The silence of the record after 1937 is analysed separately ("The record goes dark").

What would settle it (project level, in priority order)

1. Broadband permittivity sweep (0.1-3 GHz, thermostatted) plus a rebuilt crossed-Nicols detector: decides claim 1, and with it everything resonance-specific.
2. The four-arm blind food/colloid test, preceded by (1): decides claims 2 and 3.
3. A bench or field-solver check of the 6.2 cm / 18.6 cm plate oscillator: decides whether claim 1's sample resonances were the apparatus.
4. Isothermal electrolysis calorimetry: closes claim 11 for good. (Updated 2026-10-08: record total work per mole of H2 as DC electrical energy plus real power delivered to the resonator or wand, against a heat-matched off-band control, and separately test overpotential with and without the standing field at about 1.3-1.6 cm; see the 2026-10-08 section.)
5. (Added 2026-10-04) Inspect the Arnold v. Watson trial exhibits (NARA RG 21, RG 276): they are the only known place where an evidential test of a Kassner-type principle may have been put on record. Then order the three Swiss Federal Archives files and the Arbon tax rolls.
6. Archive work: USPTO 1945 assignments, Landesarchiv Glarus dossier, GEC/Hirst archive, Dipole A.-G. Arbon: decides claims 7, 8, 9, 12. The full list of records, and which explanation of the post-1937 silence each would decide, is in "The record goes dark" (table of deciding records); in priority order: a. Ternion's 1936 general-meeting protocol, share register (Aktienbuch) and the names of the lenders of the Fr. 1.1 million (Landesarchiv Glarus), and the 1938 creditor filings. b. The Dipol A.-G. register file, balance sheets and liquidation report (Staatsarchiv Thurgau); papers of Dr. Alfred Wiegner. c. The Vibranic Ltd Kauf- und Abtretungsvertrag of 15 Nov 1945 (Landesarchiv Glarus) and USPTO assignment records for US 2,089,966 and 2,109,843. d. Goodrich and Firestone internal test data and 1945 purchase files; the Arnold v. Watson trial exhibits (D.D.C., about 1953-55). e. GEC Research Laboratories reports and project lists, 1934-35 (Hirst archive); Bauch's HO 405 naturalisation file (Kew); the source of the 1996 memoriam (AES). f. General Infra Co. register file (Staatsarchiv Graubünden); Universal Inventions Ltd (BT 31, Kew). g. (see also item 5 above and the Swiss Federal Archives files) UK, Swiss and Allied files on Kassner as a German national (Home Office/MI5 aliens files, Swiss Federal Archives police files on foreigners, FIAT/BIOS/CIOS reports on German HF heating), and any patent secrecy direction on his 1933-36 British applications.

Open questions

• Was the 90 cm / 18.6 cm "resonance" a property of the liquid or of the plate size chosen?
• Who sold the vulcanisation patents in 1945, and were any of them Ternion's or Kassner's resonance-based ones?
• Did Dipole A.-G. Arbon, rather than Ternion, house the research, and where did the Fr. 1.12 million go?
• Is there a second record (or the memoriam author's source) of Bauch's Ternion project, and was GEC a contractor to Ternion?
• Who lent Ternion the Fr. 1.1 million, and why did funding stop in 1937?
• Did the work continue under Dipol, Universal Inventions, General Infra or Vibranic, and did any of them keep research records?
• Do narrow cell-resonance reports (Webb/Booth, Grundler/Keilmann, Froehlich theory) exist as remembered, and were they replicated? Web check failed; not used as support.
• What do GB483062 pp.8-12 and the missing GB448418 add? (Updated: GB 448,418 is an ellipsoidal-reflector applicator; abstract only read.)
• Do the unread eight of ten pages of FR 788,428 (and the rest of FR 786,294) mention the sharp 2 percent resonance dispersion, or only relaxation? This decides whether the two accounts of "natural frequency" are one family-wide claim or two.
• Did Dipol A.-G. stop work in 1933, and where was the Fr. 1.12 million spent (patents, London 1934-36, a valuation)?
• What is in the Swiss Federal Archives files on General Infra and "Ernst Kassner", and are they the same man?
• Was the garlic root-tip result of Heller ever repeated under controlled conditions (conductivity and temperature logged)? Did Grundler and Keilmann respond to Gos 1997?
• Why is Kassner named in the Jan 1945 release but not in the Industry Inventions licensing notices?
• Who is Wilma Hencke, and where are Kassner's later papers?
• (Added 2026-10-08) Where does the 1.28 cm of GB417501 come from (liquid water at a higher temperature, another source, or his own unrecorded measurement)? Do Niether 2018 and Serra 2020 count the field power in an overall efficiency, and does any field method keep a net saving? What are the controls and gas analysis in Roy/Rao/Kanzius, and has anyone independent tried it?

Changes from the earlier single-folder reviews

Compared with review/_merged/Kassner-EEW-Ternion/arbiter.md and review/_merged/Ternion-Research/arbiter.md (their verdicts were treated as inputs to check, not conclusions). - Merged and renumbered. The old Kassner list had seven claims (lamps separate, Voss separate); the old Ternion-Research list had six. The new list has 13: lamp/applicator claims folded into 6, Voss folded into 7, company history and Bauch split out as 8 and 9 (documented history, not science), the essay's claims 10-13 kept apart from Kassner's. - Claim 1 softened from "Refuted" to "Never properly tested". Ternion-Research ruled "Q about 50 oscillator in bulk liquid: Refuted by calculation". I keep the calculation as strong evidence against, but the patent states no mechanism or loss spectrum to refute, only a birefringence readout. Kassner-EEW-Ternion's split verdict stands. - Claim 4 (plate resonators) sharpened. The old Kassner arbiter ruled the 1932 numbers inconsistent mainly by a factor of 2.3-3.5 and said loading was "ruled out". New: they all follow lambda = 3 D (x = 1.05, below the lowest Bessel root), US2089966 contradicts itself (p.5 vs p.7), and the agar cases can be rescued by unstated loading. "Dielectric loading is ruled out" is therefore too strong for agar. - Claim 5 split further: engineering feasibility Established; tables Plausible, untested; "about 40 percent" Never properly tested. - Claim 8 strengthened: the register arithmetic was recomputed, including the insolvency calculation (deficit 10.2 times capital), which supports the mundane explanation for liquidation. The older "1958 letter" is 14 Aug 1952; "electro-physical" not "electro-chemical"; the Dipole A.-G. Arbon lead is new. - Claim 9: Ternion-Research left "Bauch worked on a Ternion project" at "Plausible, untested". This arbiter first lowered it to Never properly tested, and the QE revision relabelled it. Both are superseded by the Evidence-rule revision below: documented in one published source, no second source found yet; silence elsewhere is not contradiction. The GEC stay itself stays Established. - Claim 10 / 11: bond strength gets a split (refuted single-photon, undefined collective); the electrolysis note's COP ceiling, Landauer and bench diagnostic errors are now itemised (Ternion-Research only said "Refuted"). - Claim 7: new web support that RF dielectric heating was independently proposed in the 1930s; "foresaw the microwave oven" stays Historically superseded but instructive. - Corrections of fact carried forward: "10 cm disc gives 30 cm" is in US2089966 and the GB417564 provisional, not GB417501; Voss's cited passage is in US2089966; GB417501's application number is 32,555/32; the gold sol is gold; GB483062 is a void application; US2104697 is assigned to Universal Inventions Ltd. - Unchanged: water Debye band Established; thermal RF curing Established; "energy content" Not testable as stated; the closed net-energy cycle Refuted; nothing is shown tested and failed, and no document in the folders shows suppression. (The suppression verdict was revised on 2026-10-03; see below.) - Not yet improved: web verification again largely failed, so the old open items (water tau, tissue dispersions, Goodrich/Firestone, Grundler/Keilmann) remain [memory].

QE revision (2026-10-03)

Inputs: qe.md (verdict MINOR) and the new deep reads: deep/Ternion-AG-patents-Process-for-Altering-...-480e58.md (GB417501 pp.9-12, page images pp.10-11), deep/Ternion-AG-patents-Plate-circuit-...-d42516.md (GB483062 addendum, pp.8-12), deep/ternion-artifacts-original-Ternion-AG-2-134a9f.md (ownership section, with AG3 pp.6-9), deep/ocr-underextracted-files.md. No verdict on physics changes. Two wording changes outside the physics scale (claims 9 and 12) are made, and several facts are added.

Revised verdict table (changes only; all others unchanged)

New evidence by claim

Claim 1 (sharp resonances). - QE item 3.4, the "coarse sweep steps" argument of Morray note 11 (secondary): a 2 percent line at 90 cm (333 MHz) is 6.7 MHz wide. A 1601-point sweep from 0.1 to 3 GHz has 1.8 MHz steps, so about 4 points fall across it [calc]. A normal broadband sweep would therefore see such a line. The argument does not explain a missed resonance. Verdict stays Never properly tested. - GB417501 p.10 gives two values for water: 1.28 cm "characteristic of water" in the sera example, against about 1.5 cm for the friction band elsewhere. My earlier remark that the patent attributes 1.28 cm to serum rested on the US twin only. The patent is internally loose about what "water's frequency" is [read: deep, p.10 l.42-60].

Claims 2 and 3 (energy content, practical effects). - Exact food data [read: page image p.10]. Milk 2 L: 2 min at each of five bands (455/230, 283/160, 355/160, 163/23, 27.5/0.6 cm/W), 20 to 23 C, "kept fresh 28 days". Veal 2.5 kg: 3 min, 20 to 24.5 C, 2 weeks. Apple juice 2 L: 1.5 min, 22 to 24.5 C, 4 months in a covered jar. Rises are 3, 4.5 and 2.5 K, not "2-8 C". If the bands are applied one after another, incident energy is 68.8 kJ (milk), 74.3 kJ (veal) and 33.0 kJ (juice), giving upper bounds of 8.8, 8.7 and 4.2 K [calc, c from [memory] medium]. The stated rises are about a third to a half of that. Still far below pasteurisation, so my earlier reading stands, but heating is not negligible and no heat-matched control exists. - "Kept still longer" for milk and veal required 5 minutes of re-exposure every hour (3.2 W and 3 W). The long keeping times therefore include continued treatment. - No storage temperature, sample count, spoilage criterion or untreated control in any food example. Only the bean example names a control ("similar untreated seeds", twice the growth after four weeks, no numbers; 30 s at 35 C, 28 and 87 cm, 1.5 W). This corrects "no controls" in claims.md and claim 3: there is one uncontrolled-in-detail control phrase. The electret recipes have no constant-field-only electret, and the 12 kV DC field alone makes a conventional electret, so the contribution of the varying field is untested. Verdicts unchanged. - Where Kassner's chemistry wavelengths come from (new). GB417501 p.11 says the "term differences of hydrogen" correspond to 2.74, 9.25 and 27.75 cm, and gives reaction bands 2.5-3.5, 8.6-9.5 and 26.5-30.5 cm [read: page image p.11]. Hydrogen fine structure, Delta E = alpha^2 Ry / (n^3 l(l+1)) (Ry the Rydberg constant in cm^-1, alpha the fine-structure constant), gives 0.3652 cm^-1 = 2.74 cm (n=2, l=1), 0.1082 cm^-1 = 9.24 cm (n=3, l=1) and 0.03607 cm^-1 = 27.72 cm (n=3, l=2) [calc, run 2026-10-03]. The n=2 value, 0.365 cm^-1 (about 11 GHz), is confirmed at https://en.wikipedia.org/wiki/Fine_structure [web]; the n=3 values rest on my calculation. The three wavelengths match to better than 0.3 percent, so the identification is very likely, but the patent never says "fine structure". The deep read said the figures "do not match known hydrogen lines"; that was a comparison with optical and 21 cm lines, not with fine structure, and is superseded. Consequence: these "most effective frequencies" for chemistry are atomic-theory numbers, not measured reaction optima. As with lambda = 3 D, the numbers are traceable to a calculation. Nothing is shown that atoms reacting in a macroscopic mixture would emit fine-structure radiation. Claim 12 of the patent (catalyst substitute) and claim 23 (chemical radiation source) stay Never properly tested. - Claims 1-24 of GB417501 are all process or use claims; none fixes a wavelength or number [read: deep, pp.11-12].

Claim 4 (plate resonators). - GB483062 pp.8-12 contain no plate dimension, wavelength, power, efficiency or measurement; they are construction variants and claims 1-35 [read: addendum]. They keep the idea of tuning by plate spacing (p.9 Fig. 24, claim 25) and by dielectric loading and loaded edges (claims 17-18, 22-24, thermostat claim 28) but give no formula. Gap does not enter the patent's own frequency equations. So the lambda = 3 D problem is neither resolved nor contradicted. Claims 22-24 show Kassner knew loading shifts the resonance, which supports the mathematician's dielectric-loading rescue as a hypothesis only. The verdict (Bessel theory Established; 1932 tuning rule Internally inconsistent for the stated geometry) stands. Correction: the clean relation is lambda/D = pi/x, so x = 1.841, 2.405, 3.832 give 1.71, 1.31, 0.82, and lambda/D of about 3 needs x of about 1.05.

Claim 8 and ownership. - 1937 attendance confirmed on the page image: Stoffel 48 shares (10 votes), Hoffmann 6 (6 votes), total 54 shares, 16 votes, majority 9 [read: AG2 p.11]. At founding (18 Apr 1932) Stoffel had 60 and Hoffmann 40. So share ownership changed and 46 shares are unaccounted for. No proxy giver, lender or other shareholder is named; Darlehen I (Fr. 1,016,535.05) and II (Fr. 80,000) lenders are unnamed [read: AG2 pp.6-11]. - The 17 Jul 1936 meeting (general meetings 1-4 minuted together) is only referred to on p.7; its minutes are not in the folder. The 12 Aug 1952 universal assembly asserts 100 percent of capital represented but names only Stoffel and his lawyer Dr. W. Keller-Staub; liquidation balance at 31 May 1952 showed no assets or liabilities [read: AG3 pp.6, 8-9]. Stoffel was president, sole board member and liquidator throughout. Verdict unchanged (Established as history). Leads: Aktienbuch and the 1936 protocol at the Glarus registry, SHAB notices for Ternion and Dipol A.-G. Arbon, Staatsarchiv St. Gallen for Stoffel. - Provenance of the register papers: requested from the Glarus commercial registry (Silvia Aschwanden, deputy head); first posted 30 Jan 2017, apparently lost, re-sent [read: OCR of the .docx TIFFs].

Claim 9. Verdict wording changed (table above; superseded by the Evidence-rule revision). New lead: Bauch's 1934-35 sale of "An Improved Method of Radiation Therapy" to Marconi for 80 pounds [read: memoriam], and Kassner's address as "a German subject resident in London" in 1934-36 patents [read: digests]. Neither shows a Ternion link. Search British patents 1934-36 for a Bauch application.

Claim 12. The Guardian "sonic attack" article (25 Aug 2017) was OCR'd: expert opinion on ultrasound, no data, no mention of Ternion, Kassner or RF [read: OCR]. It is not evidence for any Ternion claim. The WWII "Wirbelwind Kanone" infrasound idea it mentions is unrelated to Kassner. Verdicts as above.

Coverage items closed or noted. The two under-extracted files are now read (facts.json to be corrected). A full-text search of the Paperclip book and NARA files by QE found no Kassner (only "KASSNER RUDOLF", jet engines), Ternion or Bauch [read: QE grep]; absence from those indexes is weak evidence against a Paperclip link (a man resident in Switzerland would not be expected in them); it is weighed in "The record goes dark", explanation 5. Still not read in depth: GB417501 p.9 and p.12 beyond the text layer (clean print, low risk).

Web verification status (this revision)

• Done: fine-structure value for n=2 [web: https://en.wikipedia.org/wiki/Fine_structure]. A second page (Hydrogen_spectral_series) did not give the value.
• Not done, still [memory]: water Debye time 8.3 ps (high; independently estimated 8.5 ps [calc]); serum conductivity about 1.5 S/m (medium; underlies the Q argument in claim 1); Grundler/Keilmann and Webb non-thermal-effect reports and replication status (medium); electret charge ranges (low); the 1945 Goodrich/Firestone purchase beyond the clippings. No verdict depends on one of these alone.

Remaining open (project level)

Verdicts and the "what would settle it" list are unchanged, with one addition: the 1936 protocol and share register (Glarus) decide who financed the Fr. 1.12 million.

Evidence-rule revision (2026-10-03)

Reason: prompt.txt now carries three SOURCE HONESTY rules ("No suppression of evidence", "Interested parties", "Absence of evidence"), and the owner asked for three corrections (Bauch, Goodrich, the silent record). New inputs: the Investigator's notes (research/2026-10-03-*.md, sites.md) and the extracted memoriam text. One web check was attempted (Goodrich's own RF-vulcanisation patents) and failed: the search quota is exhausted. Every change made in this revision:

Verdict changes

All physics verdicts (claims 1-6, 10, 11, 13) are unchanged: the new material is historical and does not bear on them. Claim 7's verdicts are unchanged, but the reason given for "Not testable as stated" is rewritten (below).

Wording and content changes

1. Source gap note: the absence of lab records is flagged as an anomaly analysed in the new section, not a neutral gap; the Investigator's partial Kassner biography is noted. The Morray notes and the essay are described as interpretation checked against primary documents, instead of "neither counts as evidence".
2. Reviewer agreement paragraph: "the suppression thesis is an argument from absence with an ordinary documented alternative" rewritten so the absence is analysed, not scored.
3. Claim 1: added that the patent was abstracted and cited (Chemical Abstracts 1935, reference works, Botanical Review 1944, In re Arnold 1950) as a claim, with no test either way.
4. Claim 3: "the patent pre-excuses variation" replaced by a description of what the patent says.
5. Claim 7 (owner's correction 2): the 1945 announcement's "heating" and the 1950 Goodrich Research Center account are now reported as the characterizations of an interested party (buyer of the package and licensor of the rubber field), quoted side by side with Kassner's own statement that his object "is not to produce a rise of temperature" and that "energy content" means a change "that does not consist solely in a change in the temperature". Added: a check showing Goodrich's "about 150 million cycles" fits where most power went in the milk example but not the claimed natural frequencies; the Rubber World 1945 non-thermal remark; the Arnold cases. The Goodrich account is not called false (no evidence of knowledge); the records that would show what Goodrich knew are named (internal test data, 1945 purchase files including the Vibranic contract, Arnold v. Watson exhibits). The reason "(heating explains it)" is replaced by: a purchase is not a test in either direction. The point, put to the review, that Goodrich held its own electronic-vulcanisation patents is recorded as not yet documented in the files.
6. Claim 7: Voss's remark, previously labelled "hedged", is now quoted in full, including "In truth he foresaw both the microwave oven and the industrial electromagnetic energy applicators".
7. Claim 8: the board's 1937 reason for liquidation (funds exhausted, patents lapsed for unpaid fees, no industrial exploitation in view) quoted and identified as the board's own account; SHAB confirmation and the Dipol A.-G. findings added; verdict unchanged.
8. Claim 9 (owner's correction 1): the memoriam sentence quoted; the silent sources listed and the silence stated as not contradiction; circumstances that fit (Kassner in London 1934-36; decimetre-wave subject of his 1933 British filings) added as consistency, not a second source; the essay's overstatement ("employee of GEC", "collaborated jointly") kept as evidence against over-reading; deciding records expanded.
9. Claim 12: rewritten with evidence for and against suppression both stated; points to the new section.
10. Claim 13: "pre-excuses variation" replaced by a description.
11. New section "The record goes dark" (owner's correction 3): the anomaly (Fr. 1.12 million of research, a large patent family, no papers, reports, conference work or later science by Ternion's people), six competing explanations (commercial failure; trade secrecy; continuation under Dipol A.-G., Universal Inventions Ltd, General Infra Co. and Vibranic Ltd; contract work at GEC; wartime or classification secrecy, with the Paperclip index findings; suppression), evidence for and against each, a weighing, and a table of deciding records.
12. Corrected claim list: items 7, 8, 9 and 10-13 rewritten to match.
13. "What would settle it": item 5 expanded into a prioritised list of archive records (a-g) taken from the new section.
14. Open questions: added the lenders and why funding stopped, and continuation under other company names; the Bauch question reworded.
15. "Changes from the earlier single-folder reviews" and the QE revision table and text: the claim 9 and suppression entries marked as superseded by this revision; "nothing is shown suppressed" reworded to "no document in the folders shows suppression".
16. QE revision "Coverage items": "absence from those sources is not evidence either way" replaced by a weighed statement (weak evidence against a Paperclip link, analysed under explanation 5).

Investigator findings (2026-10-04)

Inputs: the "Investigator findings (2026-10-04)" sections of physicist.md, mathematician.md and contrarian.md (all three read in full), sites.md "Leads for review" (leads 1-29), and the research notes they cite. I did not re-run the Investigator's searches; [web] tags below are as the Investigator recorded them, except the one check noted under "Web verification". Findings are leads, not tests; every pertinent one is kept here whatever the reviewers think of it. Verdict scale as in Step 2. The in-place updates above (claims 1, 4, 7, 8, 12, 13, "The record goes dark", corrected list, settle-it list, open questions) each carry an "Added" or "Update 2026-10-04" mark.

Where the three reviewers agree on the new findings

• Kassner's own French patents (FR 786,294 p.1, FR 788,428 p.1) call the natural frequencies Debye relaxation frequencies. All three call this the most important inventor-side item. It supports the Established half of claim 1 and gives no support to a 2 percent-wide band. Only pp.1-2, 15, 17-20 of FR 786,294 and pp.1 and 6 of FR 788,428 (2 of about 10 pages) were read.
• The analogue sharp-line tests (Gandhi 1980 at 26.5-90 GHz; Gos 1997 at 41.68-41.71 GHz) are against that class of claim, at frequencies about 100 times Kassner's serum band, in cells and nucleic acids rather than serum. They do not test his bands.
• Arnold, Heller, Liebesny/Wertheim and Benedetti are further claimants of frequency-specific non-thermal effects. None is a replication of Kassner, and none supplies controls, counts or statistics. Their stated mechanisms (Debye step; interfacial polarisation) are broad in frequency.
• Goodrich's 1950 characterization is an interested party's; it is accurate for part of the apparatus range (150 MHz is 2.0 m; the 2.05-2.90 m table is 103-146 MHz [calc]) and silent on the natural frequencies and on the 0.4-1 m food examples.
• Patent citations (Spencer, Copson, UHF tubes) show Kassner was known as prior art, always framed as heating.
• Kassner's record does not stop in 1937; it runs to a death in Munich by May 1962. The suppression premise "the record goes dark" needs narrowing to research reports.
• No verdict label is changed by any reviewer apart from the sub-splits recorded below.

Disagreements and my rulings

1. Is the sharp band in the French texts "absent" or "not seen"? The contrarian writes that in the French texts "there is no separate narrow resonance, only relaxation", and that this weakens the "never properly tested" half. The physicist and mathematician say they cannot tell, because only 2 of about 10 pages were read. Ruling: the physicist and mathematician are right. Silence in unread pages is not absence, and the Investigator's own note says that. The French texts do use "relaxation" for the first frequency, so within the family the two kinds of band are at least distinguished; the contrarian's point that the family does not consistently assert a sharp resonance is open until the other pages are read. I do not move the sharp-band verdict.
2. How broad is a Debye band? The contrarian says water's peak "spans about six decades [web, mine]". The physicist (about 1.1 decades at half height) and mathematician (half-height frequency ratio 13.9, Q about 0.3 [calc]) agree with each other, and 13.9 is about 1.14 decades. A single-time-constant Debye loss curve cannot span six decades; six decades is, I think, the width over which the loss is visibly non-zero, but I have not seen the source. I use the half-height figure for the comparison with a 2 percent band (about 170 times narrower than the Debye band's full width ratio), and mark the six-decade figure as unreconciled. It does not change the argument; any reading makes a Debye band hundreds of times broader than 2 percent.
3. Serum "conduction crossover" coincidence (mathematician). At sigma 1.5 S/m and eps_r 80 the conduction-to-displacement crossover f_c = sigma/(2 pi eps0 eps_r) is 0.337 GHz, wavelength 89 cm, which matches Kassner's serum "about 90 cm" [calc]. With the web-checked blood values (65, 1.3 S/m at 300 MHz) the crossover is 0.36 GHz, 83 cm [calc]. The loss tangent is about 1.1-1.2. Conduction loss falls monotonically with frequency and has no peak, so the crossover is a smooth change of regime and cannot be a Q 20-50 line. I accept it as an observation worth a test, and as a candidate for what a broad "serum band" at that wavelength could be, not as a finding. Nobody has shown Kassner meant it, and the patent says nothing like it.
4. Claim 7 label. The physicist proposes "Partly documented". That is not on the Step 2 scale. The substance (portfolio named by Goodrich in 1950; Raytheon citations; seller, price and terms unknown) is now in claim 7, the purchase stays Established as a press and trade record, and the question of whether the purchase validates the mechanism stays Not testable as stated.
5. Claim 12 label. All three reviewers keep "Not testable as stated" for suppression; I keep Never properly tested (as history), as ruled on 2026-10-03: the thesis is testable, the deciding records are named, and they are unread. The new documents (a documented death in 1962, continued patent filing, openly litigated patents) make suppression less natural, and the unread Swiss files keep it open.
6. How far the Dipol closure of 1933 should be pressed. The physicist says it weakens "five years of experimental programme", the contrarian that the work was "real but short" and that an accounting question is sharpened, and the mathematician that spending after 1933 must have been elsewhere or a valuation. They agree on the arithmetic. They also all rest it on one remark in three reprints of a tax report. I carry it as the best reading, stated with its limit (claim 8 above). It does not alter the Established verdict on the register arithmetic.
7. Heating in Heller's experiment. The physicist finds "no temperature increase" credible only for low-conductivity water (0.005-0.08 K at 1e-4 S/m; 0.5-8 K at 0.01 S/m [calc]); the mathematician gets 0.003-0.09 K at 1e-4 S/m and 1.3-46 K at 5e-2 S/m, with the same duty cycle. These agree (field and conductivity assumptions differ). The 300-1,200 V/cm field is the Navy survey's for unicellular experiments, not stated for the root tips, and the letter gives neither conductivity nor field. Ruling: the printed information cannot exclude a thermal explanation, and does not establish one; the claim is Never properly tested.

Findings by the Investigator's four priorities, and what they do

1. The inventor's own work. FR 788,428 (filed 27 Sep 1934, "Procédé de polarisation diélectrique par résonance moléculaire électrique de substances dites dipôles", Kassner "résidant en Angleterre") repeats the GB417501 examples (gold sol, agar, milk, veal "plus de 15 jours", apple juice, serum about 90 cm): the same single runs, so a third language member is not a third experiment. ES 135,727 is the Spanish member of the same energy-content family (found on DEPATISnet; not read by anyone; physicist C1, mathematician 1b; sites.md lead 20) [web: research/2026-10-04-depatisnet.md]; the Spanish patent office (OEPM) historical archive would give its text. GB 448,418 and US 2,094,602: ellipsoidal-reflector applicator (liquid in a glass sphere at the second focus); standard optics, no result figures; one more small slip in US 2,094,602 ("a diameter of 1 cm or a radius of 2 cm"; I checked the OCR line, file research/files/US2094602.ocr.txt line 337). US 2,096,459 repeats "about 40 percent" with no method. GB 435,643: plate-resonator light detector. DE 1,128,930 (1960): rotating-field plate resonators; the mathematician's check is in claim 4. Kassner's death by May 1962. Effect: claim 1 (relaxation reading; sub-split above), claims 4-5 unchanged, claim 8 and the dark-record analysis revised.

2. Support by others. - Arnold (US 2,522,082; In re Arnold 1950; Arnold v. Watson, 237 F.2d 584, D.C. Cir. 1956): a trained dielectric chemist (JACS 1936; J. Electrochem. Soc. 1938) claimed the same principle in the language of Debye dispersion; the examiner cited Kassner; the courts refused the broad claims as "too broad and indefinite" and did not rule on the physics; the 1955 trial court said it decided "on evidence" and that Arnold "exposed the error of certain assumptions ... by predecessors" [web: research/2026-10-04-arnold.md]. The trial exhibits are not seen. The physicist notes a contradiction in activation without heating at a loss peak (absorption is greatest there) unless the sample is thin or cooled, and that skin depth at 27 MHz in weakly conducting water is about 1 m, so no surface selectivity arises from skin effect. The mathematician notes that a skin depth of 2.3 cm at 333 MHz (sigma 1.5 S/m) gives no selective heating of micrometre surface layers in the Debye picture without a different surface relaxation frequency. Both are arguments about the claim, not tests of it. Arnold is not an independent replication: he cites Kassner and his own earlier papers. His trial record is the one place an evidential test may have been put on file; it is now first in my archive list. - Heller and Teixeira-Pinto 1959, US 3,095,359 (cites US 2,089,966), and the Midwest Research Institute survey for the US Navy (1960): weighed in disagreement 7 above. The survey is an interested contractor (a hazard-standards programme) reporting a claim, not a replication; it offers Maxwell-Wagner polarisation as the mechanism. The physicist's calculation that pearl-chain alignment energy is 6e4 to 1e6 times kT for 5 micrometre cells at 3e4-1.2e5 V/m (and the mathematician's U/kT of about 470 at 1 micrometre) shows orientation needs no new physics [calc, assuming a Clausius-Mossotti factor of 0.5]. A 5 microsecond pulse has bandwidth about 0.2 MHz, 0.74 percent of 27 MHz [calc]: Heller's "frequency dependence" is a measurable claim to a percent. Heller's patent text lists outcomes at named frequencies (for example sex ratio at 25 MHz, histidine reversion at 31 MHz) without n or controls. - Liebesny and Wertheim (GB 417,863, 1.8 mm to 120 m, organism-specific), Benedetti, and Kassner GB417501 in The Botanical Review 1944; Rubber World Apr 1945 ("may be connected in some way with electronic power itself"). Claims with no data in hand. Over a frequency range of 6.7e4 [calc] "choose the frequency for the organism" has no test unless the organism-frequency map is given. - Archangelsky, "Über die spezifische Wirkung des Ultrakurzwellenfeldes" (1930s; physicist B2): a further claimant of a specific (non-thermal) short-wave effect, listed with Schliephake and Compere in the reference list of Kowarschik, Kurzwellentherapie (Springer, Wien 1936); only the bibliographic record was seen, no data [web: research/2026-10-04-related-work.md; https://link.springer.com/chapter/10.1007/978-3-662-28514-5_4]. - Muth 1927 and Krasny-Ergen 1936-37 (physicist B1): pearl-chain formation of particles in HF fields was first reported by Muth (Kolloid-Z. 41, 97, 1927), with theory by Krasny-Ergen (Hochfrequenztechnik und Elektroakustik 48, 126, 1936; 49, 195, 1937), as cited by the 1960 Navy survey, Appendix E [web: research/2026-10-04-heller-pearl-chain.md]. This is the mainstream line (induced polarisation; Maxwell-Wagner) for the orientation effects Heller reported, contemporary with Kassner's patents. Neither paper was read by a reviewer; it is in sites.md's run-4 list. - Jonas, US 2,712,713 (1950), seed treatment by HF fields, and later patents citing GB417501 (Hayashibara GB 2,145,317, 1983, animal productivity; Vergara Piccaluga EP 0,848,058, 1995, cell behaviour; "Aqua Dynamics" water-treatment patents 1990-91) [web: research/2026-10-03-google-patents.md; sites.md lead 13]. These are claims by others that cite Kassner; no data seen. Jonas bears directly on claim 3 (seed treatment). No reviewer weighed Jonas or these later citers; I record them unweighed, as further claimants, not tests, and the Jonas text (claims, any germination figures and controls) is the next thing to read for claim 3. - Against, or tests of related classes: Mittler 1976 (no genetic damage in Drosophila at 2 and 10 m; title only, different organism); Gandhi 1980 and Gos 1997 (above); the 1930s German dispute (Schliephake and Compere 1933 for a specific effect; Kowarschik 1936 and others reported only as snippets saying none was proven). The contrarian's reservation applies to all: no reviewer read the full texts of Gandhi, Gos, Mittler, Kowarschik or the Grundler and Keilmann paper. Kappe, Pieber and Dallinger (Angew. Chem. 2013) show the question of non-thermal microwave effects in synthesis is still in dispute; a 10 K local temperature error near 100 C changes a rate by a factor 2.04 for E_a = 80 kJ/mol [calc: E_a/R = 80,000/8.314 = 9,622 K; 1/363 - 1/373 = 7.386e-5 /K; product 0.711; exp(0.711) = 2.04; corrected 2026-10-04 from the factor 1.3 first carried over from the mathematician, which the expression does not give; example.py step 8d and primer.md agree on 2.04], which is how apparent non-thermal effects are usually produced and does not exclude real ones.

3. Closely related work. Webb and Booth (Nature 1969) and Grundler and Keilmann (PRL 1983) are the positive side of sharp-line reports and are recorded beside Gandhi and Gos. Nothing in the record tests Kassner's 0.3-4.9 m region. Not yet searched: Debye-Falkenhagen dispersion of electrolytes (the mainstream benchmark for "electrolyte bands" in claim 1; the physicist recalls broad, not 2 percent-wide, ionic-atmosphere relaxation in the MHz range [memory, low]).

4. Historical record. Goodrich 1950 and the Industry Inventions chain (claim 7); Dipol 1933, Wiegner, the Fr. 200,000 Vibranic purchase equal to 0.18 of the book research account [calc, mathematician: 200,000 / 1,120,380.15 = 0.179] (it shows only that 1945 did not value the portfolio at book cost), and of which part is identified by matching priority dates: Vibranic's FR 926,351 (priority 22 Apr 1943, = Grell/Richards US 2,415,025, assigned to Firestone and Goodrich) and CH 250,508 (priority 4 May 1945, = Grell's CH 254,655), so the purchase included at least Grell's foreign heating rights, and whether any Kassner patent was in it is still open (claim 7 above; sites.md lead 14); the BAR file titles (claim 12 and the dark-record section); the 1941 SHAB "Kassner Ultrakurzwellen" hit closed as false; GB 435,043 is GB 435,643; US 2,190,668 is Llewellyn. Items kept in the record without weight: the 1923-28 "Ernst Kassner" adding-machine patents and the 1929 Elektro-Physikalische Gesellschaft Kassner (identity probable or unknown); SPSL file "W Kaeser - E Kassner" at the Bodleian (which Kassner unknown); NARA matches for "Ernst Kassner" and "Giodvad" in seized-German-record and State Department lists (contexts unread); Goodrich Series VII (finding aid blocked); BIOS Final Report 866; Espacenet blocked by CAPTCHA.

Verdict table (changes from this update only)

Absence-of-evidence analysis: what changed

Details are in "The record goes dark" (anomaly paragraph, explanations 1, 3, 5, weighing, table of deciding records). In short: the anomaly is now stated as missing research reports, not a missing man; commercial failure gains (Dipol 1933, General Infra's 1944 unwinding); continuation gains DE 1,128,930; trade secrecy gains very little; wartime context gains one unread file title; suppression gains nothing documented and loses some naturalness. The silence about replication of Kassner's examples (claim 13) is unchanged: no test of any of his food, colloid or seed examples has been found in any source, and Arnold's trial exhibits are now the first record to inspect.

Web verification (this update)

Budget 8, used 1 (Chrome, Google, new tab, closed; I did not touch the existing open tab). Result: blood at 300 MHz is about eps 65, sigma 1.3 S/m (snippet; blood rather than serum). Not checked and still [memory]: water tau at 0 C (17.7 ps, medium), electret charge ranges (low), the Grundler and Keilmann replication status beyond Gos 1997 (abstract via the Investigator), Debye-Falkenhagen. No verdict turns on any of them alone. The contrarian's own check of water's relaxation time (8.27 ps) is accepted.

QE remediation (2026-10-04, Investigator update check)

• Defect 1 fixed: the Arrhenius factor for a 10 K error near 100 C at E_a = 80 kJ/mol is 2.04, not 1.3 [calc: 9,622 K x 7.386e-5 /K = 0.711; e^0.711 = 2.04]; corrected under priority 2 above and in mathematician.md 3c. No verdict changes; the point (temperature error can mimic a non-thermal effect) is stronger.
• Defect 2 fixed (arbiter part): the matched priority dates FR 926,351 = US 2,415,025 (22 Apr 1943) and CH 250,508 = CH 254,655 (4 May 1945) are added to claim 7 and to priority 4 above: the Vibranic purchase included at least Grell's foreign heating rights; no document yet shows a Kassner patent in it. Claim 7 verdicts unchanged. (The primer.md paragraph is left to the primer author.)
• Defect 3 fixed: ES 135,727 (priority 1), Archangelsky, Muth 1927 with Krasny-Ergen 1936-37, and Jonas US 2,712,713 with the later GB417501 citers (priority 2) are added, each as a claim or citation by others with no data seen; noted that no reviewer weighed Jonas.
• Defect 4 (primer.md omissions): not applicable to this file; left to the primer author.

Investigator findings (2026-10-08)

Inputs: the "Investigator findings (2026-10-08)" sections of physicist.md, mathematician.md and contrarian.md (all three read in full), the review direction of 2026-10-07, and my own checks. The new material is sites.md leads 30-40 from follow-ups Q03 (2026-10-06) and Q05 (2026-10-07). I did not re-open the Investigator's papers; [web] tags are as the Investigator recorded them (mostly abstracts or snippets). Findings are leads, not tests; every pertinent item stays in the record. Verdict scale as in Step 2.

Claim 11 wording (review direction, 2026-10-07)

Claim 11 now reads: "a closed electrolysis-and-burning cycle, or electrolysis assisted by a resonant standing field (displacement current), or by radiated RF, yields net energy". Ternion's two methods in Kassner's own words (US 2,089,966 p.3, read by the Investigator): the "standing field" (conductors set into natural oscillation so that standing waves fill the space holding the substance, "a pair of concentric tubes or a plate-resonator") and the "radiating field" (reflector-focused). Corrected in the verdict table below, claim 11 section, corrected claim list. claims.md claim 11 has no "RF-assisted" wording (it says only "if bond energy can be lowered cheaply, a closed ... cycle"); I did not edit it (not my file); the orchestrator may add the new wording there. The "RF-assisted electrolysis" phrase also appears in physicist.md C6 and contrarian.md lead 32 and in the primer; the primer update follows.

Where the reviewers agree

• Thermodynamics is unchanged by anything found: dG = 237.1, dH = 285.8, T dS = 48.7 kJ/mol, dH/dG = 1.205 [calc, memory high; all three recomputed]. A closed cycle gains nothing; the heat-pump half of the Q05 hypothesis ("somewhat like air conditioning") is correct and ordinary; a saving in electrical work must be paid at least in full by real power delivered into the water (6.6 W for 24 kJ/mol at 1 mol H2/hour [calc]).
• Kassner wrote nothing on electrolysis or bond energy; his chemistry statement is a rate claim (GB417501 p.4, acetylene plus water to acetaldehyde p.5) with no yield, energy or control. The extension to water is a fair application of his general principle for dipolar substances; whether the principle is true is claim 1.
• Water at its own Debye band has Q = eps'/eps'' of about 1 (1.14 at 19.2 GHz, 0.97 at 23.4 GHz; loss tangent 0.87 and 1.03 [calc, I reran it]). A water-filled resonator at the band is heavily damped; "lightly loaded" holds only with a low water filling factor, which means a small effect. Kassner himself wrote "the absorption of energy therefore especially great in the case of resonance" (US 2,089,966 p.3), so by his text resonance is where the sample draws most power, and where a heat-matched control matters most.
• Measured, mainstream relatives exist and are not Ternion bands: Niether 2018 and Serra 2020 (high-frequency fields lowering electrolysis cost at a catalyst), Hundt 2014 (infra-red bond-selective dissociation of water on Ni(111)), the second Wien effect (static field, about 10^8 V/m [memory, medium]).
• Roy/Rao/Kanzius (13.56 MHz, about 300 W, salt-water gas burning only while the field is on) is claimants' work with no replication on record; not refuted, not a Kassner band (22 m against 1.28 cm).
• Thermally regenerative cycles are the mainstream form of "the environment pays": a second reservoir, bounded by 1 - Tc/Th = 0.20 for 373 K and 298 K [calc].

Disagreements and my rulings

1. Label for field-assisted electrolysis at a catalyst. Physicist and contrarian: Established. Mathematician: not "Established" (abstracts only, one group, field power not stated); wrote "tested, not yet independently replicated; supported as a rate effect, net benefit unproven". Ruling: the measured overpotential reduction (100-200 mV; "equivalent to ~200 C" with the cell up 5 C [web per Investigator]) is published, peer-reviewed and has an accepted mechanism (local heating of the catalyst), so the effect is Established; I keep the mathematician's caution as a qualifier: I have read only abstracts, the two papers are different mechanisms (induction heating of magnetic nanoparticles; microwave reduction of ceria, with a patent application, an interested party), and net energy gain is Plausible, untested because the field power is not shown to be repaid (the 2026 ACS review snippet describes both a net loss and a net saving, context unseen). Stronger argument: the physicist's and contrarian's on the effect, the mathematician's on what it does not show.
2. Does it bear on the Q05 mechanism? The contrarian calls it the strongest new item for the best case; the physicist and mathematician stress that it acts on the overpotential (the 0.4-0.8 V above 1.23 V in real cells [memory, medium]), not the 237.1 kJ/mol floor, and not through water's Debye band, not through displacement current in the sample, and not through a wand. I side with the latter: it shows that an applied field can lower the voltage cost of splitting water at a catalyst; it is not a test of a Kassner standing field or radiating field.
3. Claim 10 wording. All three keep Refuted by calculation for the single-photon form and add that resonant bond-selective weakening of water is real in the infra-red at a surface (Hundt 2014). Ruling: the verdict stays; the reader gets the fuller statement (microwave photons cannot, infra-red photons can, on a surface). Kanzius-type RF claim: Never properly tested by a neutral party.
4. Two numerical errors in the notes: the reviewers disagree, so I checked. - "About 35,000 times a 1.28 cm photon" (research/2026-10-06-followup-Q03.md section D; sites.md lead 34; physicist C5 repeated it, corrected there 2026-10-08). The mathematician and contrarian say 3,400-4,700. I confirm: 0.33-0.46 eV over 9.7e-5 eV = 3,400-4,700 [calc]. The 50,000 figure is the O-H bond (4.8 eV) over the photon [calc: 49,600]. 35,000 is wrong; use about 3,400-4,700 [calc, rechecked 2026-10-08: photon hc/lambda = 1239.84 eV nm / 1.28e7 nm = 9.686e-5 eV; 0.33/9.686e-5 = 3,407; 0.46/9.686e-5 = 4,749]. (This line said 3,500-4,700 until the 2026-10-08 QE check; 3,407 rounds to 3,400.) - "0.2 V/Angstrom (2e9 V/m)" in the Q05 note. The mathematician says it should be 0.02 V/Angstrom and that Saitta's 0.35 V/Angstrom is 3.5e10 V/m, so the estimate is 17 times below it. I checked: 1 V/Angstrom = 1e10 V/m, so 2e9 V/m = 0.2 V/Angstrom and 0.35 V/Angstrom = 3.5e9 V/m [calc]. The Q05 note and the physicist (C4, 3.5e9 V/m) are right; the mathematician's correction (his section D5 and E2) is itself wrong, and the contrarian's 3.5e9 is right. The linear storage estimate (2e9 V/m for 24 kJ/mol, formula 0.5 eps0 (eps_r - 1) E^2 times 18.07 cm3/mol [calc: 1.97e9 V/m]) is thus the same order as the simulated dissociation threshold, i.e. a 10 percent saving by stored field energy needs fields near the dissociation scale, where the linear formula fails. The conclusion (the field must itself supply the energy) is unchanged. This is a correction of a correction; I say so plainly so neither wrong version is carried forward.
5. Where does 1.28 cm come from? Candidates: the water-vapour line (22.235 GHz = 1.348 cm, 5.3 percent off [calc]); liquid water warmer than 25 C (1.28 cm = 23.4 GHz = Debye tau 6.8 ps [calc]; the physicist recalls about 40-45 C, [memory, low], and I think that high, unchecked); ammonia (my own addition: Cleeton and Williams, Phys. Rev. 45:234, 1934, maximum observed at 1.25 cm [web: https://link.aps.org/doi/10.1103/PhysRev.45.234 via search abstract], 2 percent from 1.28 cm, but ammonia has no part in the patent's water). Both microwave-spectroscopy measurements (Cleeton and Williams 1934; vapour-line work from 1946, Becker and Autler, Townes and Merritt, per the contrarian [web]) are later than GB417501 (1932), so neither can be what Kassner copied; the temperature reading and "his own unrecorded measurement" stay open. Not settled; no weight on any.

Verdict table (changes from this update only)

Corrected claim list: additions

11. Closed cycle: ruled out by the state functions dG and dH. Field-assisted electrolysis is a measured overpotential effect at catalysts; Kassner's two methods (standing field by displacement current; radiating field by wand) have not been tested on electrolysis. Kassner never wrote on electrolysis.
12. Wording: "microwave photons cannot weaken a water bond; infra-red photons can, on a surface". Errors in earlier documents (kept, per CORRECTIONS ARE WELCOME): 35,000 should be about 3,400-4,700 (Q03 note section D, sites.md lead 34, physicist.md C5); the mathematician's 0.02 V/Angstrom correction is wrong, the Q05 note's 0.2 V/Angstrom stands (mathematician.md D5, E2); sites.md lead 31's vapour-line suggestion is weakened because the earliest line measurement found is later than 1932.

What would settle it: changes

• Claim 11 (decisive): isothermal calorimetry, total work per mole of H2 = DC electrical energy plus real power delivered into the resonator or wand (drive minus reflected), against a heat-matched, off-band control; hypothesis below 237.1 kJ/mol, thermodynamics at or above it. Separate allowed form: does the standing field at about 1.3-1.6 cm, or a radiating field, lower overpotential by more than the field costs?
• Read the full text of Niether 2018 (frequency, power, energy balance), Serra 2020, Roy et al. 2008 and Rao et al. 2010 (controls, gas analysis), and the 2026 ACS review's 215.6 kWh context.
• Claim 1: swept permittivity and loss of water, 0.1 M NaCl and serum, 0.2-25 GHz, thermostatted, to test whether 1.28 cm corresponds to a temperature.

Absence-of-evidence analysis: what changed

The record's silence on Ternion research reports (the "record goes dark" section) is unchanged by these leads; they are other people's work. Two new silences are noted. (1) No measurement of total work per mole below 237.1 kJ/mol was found; since the second law says it cannot go lower, the silence is expected and I weigh it as neither for nor against. (2) No independent replication of Roy/Rao/Kanzius, in favour or against, was found in a short search; silence is not contradiction, and the explanations (conventional heating or plasma effects not excluded by the abstracts; low interest after the claimants disclaimed net energy; a stated lack of access to the 2010 paper) are untested. The record that would decide: the papers' controls and gas analyses, and any independent attempt.

Related work and evidence restored (QE 2026-10-08)

These items are in all three reviews (or in two of them) and in sites.md, and were left out of this section's first draft. They are kept here as the record, whatever weight they carry; the reader weighs them.

• Puharich, US 4,394,230 (1983), water splitting by a "complex electrical wave form" (sites.md lead 35; physicist C8, mathematician C3, contrarian lead 35) [web: Google snippets via Investigator; the patent text returned nothing in Chrome and no reviewer read it]. The mathematician adds that Puharich's 1987 compilation names "22.3 GHz" as water's resonance, which is the water-vapour line (22.235 GHz) [web: snippet via Investigator]. A claim by another inventor in the same area as claim 11; unread, so no weight either way. It is the one item in the record that ties a resonance-electrolysis claimant to the same vapour line discussed under ruling 5.
• Stanley Meyer's "water fuel cell" (sites.md lead 35; all three reviews). An Ohio court in 1996 found "gross and egregious fraud" after expert testimony that the cell used conventional electrolysis [web: Wikipedia snippet via Investigator]. This is the court's characterization in an adversarial proceeding brought by investors, not a laboratory test, and it concerns one device; it says nothing about Kassner's principle. Meyer's own description of his device is not in the record (no reviewer read his patents).
• Third-party description of the Roy/Rao/Kanzius observation (mathematician B2, contrarian lead 33; sites.md row i-sis.org.uk) [web: https://www.i-sis.org.uk/canWaterBurn.php via Investigator]. Mae-Wan Ho (Institute of Science in Society, 2009; an outlet with an advocacy stance on science policy, which is its stake) wrote "The phenomenon is real, but the mechanism is entirely unknown", and that the 13.56 MHz photon carries about 1e-8 of the energy needed per molecule [checked: h x 13.56e6 Hz = 5.61e-8 eV; O-H 463 kJ/mol = 4.80 eV; ratio 1.17e-8]. An anonymous reader comment on the same page says the Raman spectra look similar "apart from a minor difference in peak intensities". It is the only third-party description of the observation found, documented in one published source; it cuts both ways (the effect is reported as real; the claimed spectral evidence is questioned; the energy per photon rules out a single-photon bond effect). No second source found yet.
• Evidence against the second-Wien model: Stovall et al., EES Catalysis 2026, which "questions the model" of field-enhanced water dissociation in bipolar membranes (sites.md lead 37; physicist B5, mathematician B3, contrarian lead 37) [web: snippet via Investigator; not read]. So the nearest measured relative of claim 2's orientation half (Onsager 1934, Bui 2021) has a published challenge to its mechanism. The measured rate effect is not in dispute in the snippets; its explanation is.
• Ambient radio as a second reservoir (physicist C9, mathematician D6, contrarian claim 11 item 5) [web: aggregator snippet via Investigator, 0.1-10 microwatt/cm2; primary survey Pinuela et al. 2013 not read]. Recomputed: 24 kJ/mol at 1 mol H2/hour = 24,000/3,600 = 6.7 W; at 1 microwatt/cm2 = 0.01 W/m2 that needs 6.6-6.7 W / 0.01 W/m2 = 660-670 m2 of perfect collector [calc] (the physicist and contrarian use 6.6 W, the mathematician 6.7 W). A plate resonator tuned at 19-23 GHz does not collect broadcast bands (roughly 0.1-6 GHz [memory, medium]) anyway. So ambient radio is not a credible source for Q05's "the environment pays"; the T dS term (48.7 kJ/mol, already counted) and a hot/cold or sunlight-type second reservoir (Carnot 0.20 at 373/298 K) remain the physical routes.

QE remediation (2026-10-08, Investigator update check)

• Defect 1 fixed (arbiter part): Puharich US 4,394,230 (with the 1987 "22.3 GHz" note) and Meyer's water fuel cell (as the Ohio court's 1996 characterization) are added above with source tags; the Meyer parenthesis in the verdict table now carries its tag. (Primer Section 4.6 is left to the primer author.)
• Defect 2 fixed (arbiter part): (a) Mae-Wan Ho's "The phenomenon is real, but the mechanism is entirely unknown", her 1e-8 photon ratio (rechecked: 1.17e-8) and the reader's "minor difference in peak intensities" comment are added, with the outlet's stake; (b) Stovall et al. 2026 questioning the second-Wien model is added as evidence against; (c) ambient radio (0.1-10 microwatt/cm2, 660-670 m2 for 6.6-6.7 W, recomputed; no collection at 19-23 GHz) is added. (Primer part left to the primer author.)
• Defect 3 (priority wording "first measured 1946", "first microwave absorption spectrum"): not applicable to arbiter.md, which already says "earliest line measurement found" and calls neither measurement the first; the wording is in primer.md and example.py, left to the primer author.
• Defect 4 fixed (reviewer files): physicist.md C5 and its verdict summary item 2 now read 3,400-4,700 with the arithmetic and a dated correction note (correction 5 added to the physicist's "Corrections to earlier documents"); mathematician.md D5 and E2 carry a dated correction note saying the 0.02 V/Angstrom correction and the "17 times below" claim are wrong (2e9 V/m = 0.20 V/Angstrom; Saitta's 0.35 V/Angstrom = 3.5e9 V/m, about 1.8 times the 1.97e9 V/m estimate [calc]), with the summary of verdict changes annotated; the mathematician's "3,500-4,700" in E1, F and the summary is aligned to 3,400-4,700. Not applicable here: sites.md leads 31 and 34 (the Investigator's file) and claims.md claim 11 (left to the orchestrator, as noted under "Claim 11 wording").
• Defect 5 fixed: ruling 4 and the corrected claim list now give 3,400-4,700 (3,407-4,749 [calc]), matching the mathematician, primer and example.py.

Web verification (this update)

Budget 8, used 1: Cleeton and Williams 1934 (above). Not re-checked by me and still as the reviewers tagged them: Niether, Serra, Hundt, Saitta (arXiv abstract rechecked by the physicist), Wien-effect papers, tau(T) of water, bulk fields of laboratory oscillators [memory, medium]. No verdict turns on one of these alone.