Lakhovsky George: QE (Step 4b)

Date: 2026-10-08. Reviewer: QE (Opus). Inputs checked: _text/manifest.json, all 22 digests, deep/ (6 files), facts.json, claims.md, sites.md, research/, physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. Spot checks were made against the extracted text in _text/ (line numbers are the .txt line numbers). Tags: [read: file p.N] extracted text, [calc] my own arithmetic, [memory] recalled, with confidence.

Verdict: MINOR

The central verdicts (cell circuit refuted by calculation, geranium and MWO cures never properly tested, radium/lead tested and not supported) are well argued and survive every check below. The defects are material in the source documents that never reached the reviews, the arbiter or the primer: a second experiment with numbers (silver-circuit water sterilisation), a whole claim family with outside experimenters (radio stations and pigeon homing), deaths by haemorrhage after treatment that bear on "no contra-indication", the co-treatments in de Cigna's ulcer series, and the MWO's mains current draw. None of these should overturn a central verdict, but two sub-verdicts need rewording and one new claim needs a verdict. That can be done by updating facts.json, claims.md, a "QE revision" section in arbiter.md and the primer. Steps 2-4 do not need to be redone.

1. Coverage

Every manifest document was digested in full (22 of 22 chunks). No document was skipped. facts.json has one entry per document, and every how_read field accurately states which page ranges were left at digest level. There is no coverage defect.

2. Shortcuts

• Winey application, digest only beyond pp.34-38. Acceptable. The golf-shaft, phi-tuning and DNA-phantom material is peripheral to Lakhovsky. It is an interested party's background, and claim 7 needs only the deep-read passage.
• Radiations and Waves pp.14-48, digest only. Defect. These pages hold the clinical record on which the Claim 4 verdict "cures: Never properly tested" is weighed: Rigaux's report on the penis-cancer case (pp.22-24), the Roversi letter, the whole of de Cigna's 1937 Vienna communication (pp.30-39), "Why advanced cases of cancer cannot be cured" (pp.41-42), the New York hospital and Brooklyn urologist tables (pp.45-47) and the directions for use (p.48). The arbiter calls de Cigna's series "the best-documented series in principle", but no reviewer read it beyond the digest. As a result its co-treatments and its own caveats were missed (see 3.3 and 3.4).
• The Secret of Life ch. VII-X (pp.30-59), digest only. Defect, partly mitigated. These pages hold the silver-sterilisation experiment (p.35), the copper-spiral timeline (pp.37, 57-58), Attilj's six collar patients (p.58), the soil and water tables (pp.48-56) and the clinical observations on lunar phase and soil (p.59). The mathematician read the Contributo table directly, so the soil verdict does not rest on the digest alone. The sterilisation experiment and the Attilj cases do rest on the digest alone, and neither was weighed.
• The Secret of Life ch. I-IV (pp.5-16), digest only. Defect. This is the opening third of the founding book: radiation-guided instinct, the pigeon experiments at radio stations, auto-electrification (the duck-wing electrometer reading). It was never deep-read, and claims.md does not carry it.
• Waves That Heal pp.1-24, digest only. It holds the collar reports (Araujo with 300 patients, Attilj, Marcus, Barinque), the animal reports, Dr. Francis on deafness and the urology reports. This is acceptable for the case summaries, because the 1941 book duplicates many of them. It is a defect for the animal reports, since "animals exclude suggestion" is the books' own answer to the placebo objection (p.15) and no review weighs it.

3. Missed material (from the digests; each checked against the text)

1. Silver-circuit sterilisation of water. The Secret of Life p.35 gives this as "FURTHER PROOF OF CELLULAR OSCILLATION". It is a communication presented by d'Arsonval to the Académie des Sciences on 15 April 1929, with M. Sesari of the Pasteur Institute. In Experiment I, B. coli at 1,128,000/cm3 was exposed to silver spirals (119 cm2 and 72 cm2). After 18 h there were 73,600 and 171,200 per cm3; after 25 h there were 0, against 43,680,000 in the control. Experiment III took the treated water and re-seeded it: untreated (a) and heated (b) samples gave 12 and 13 colonies, while the Chamberland-filtered sample (c) gave 1,474, against 946 in the control [read: Secret of Life p.35, lines ~2100-2135]. This is the second experiment in the record with numbers and a control. Its own Experiment III shows the effect stays in the water after the metal is gone and is removed by filtration. That fits a dissolved or colloidal silver agent (the oligodynamic effect, mainstream since the 1890s [memory, high]) better than "oscillation". The Investigator also listed Lakhovsky's silver-sterilisation patents FR 688887, CH 147806, FR 810280 and FR 866036 (sites.md, Leads 1). facts.json lists this experiment as a key claim, but no reviewer, the arbiter or the primer mentions it. It needs a verdict.
2. Radio stations and homing pigeons (1924-26). The Secret of Life pp.5-6 reports that on 2 July 1924, pigeons released near the Paterna radio station (Valencia) circled and lost direction. Repeat trials were run under Spanish military control, with a report by M. J. Casamajor: pigeons were released every three minutes about 8 km from the station during transmission, failed to orient, and flew home shortly after transmission stopped. A second series on 7 November 1926 gave the same result, and German tests at Kreuznach from March 1926 agreed [read: Secret of Life pp.5-6, lines 258-280]. These are experiments by other people, with a stated on/off contrast, and claims.md does not carry them. The same chapters claim a 600 V charge from shaking a duck's wing in front of an electrometer (p.10), the semicircular canals as a "radiogoniometric receiver" (p.7, p.14), and the radiation-guided moth experiments that reinterpret Fabre (pp.8-9). Related modern work (to be checked on the web by the remediation): Engels et al., Nature 509:353 (2014) reported that broadband anthropogenic electromagnetic noise in the kHz-MHz range disrupted magnetic-compass orientation in European robins [memory, medium]. That is a different mechanism from the one Lakhovsky proposed, but it is the same observation class. The record should keep it as closely related work.
3. Deaths by haemorrhage after MWO treatment. Radiations and Waves pp.41-42 has two Saint-Louis throat-cancer patients (December 1931). Each had a severe haemorrhage at or after the fifth treatment and died. Officer "Mr. H." at Val-de-Grâce (Prof. Chaumet, November 1931) had a thoracic tumour estimated at 8-10 kg that necrosed and detached; he died of haemorrhage in January 1932. Rigaux reported two further throat cases that died of haemorrhage after the fourth treatment. Lakhovsky explains that the necrotising tumour released arteries it had eroded [read: Radiations and Waves pp.41-42, lines ~1748-1843]. The arbiter and primer say only "failures admitted in advanced cases". The record actually shows five deaths following treatment-associated tumour necrosis. This matters both ways: it is evidence that the device did something to tumours (necrosis after four or five sessions), and it is an adverse-event signal that contradicts the books' "no contra-indication" and "cannot cause any injurious effects" (Waves That Heal p.33). The mathematician notes the "without exception all the patients remained well" discrepancy, but not the deaths.
4. Co-treatments and caveats in de Cigna's ulcer series. De Cigna's duodenal-ulcer patients also received a prescribed lemon-juice diet with stepped dosing, had all other medication stopped (P.G.), had constipation treated by nasal reflexotherapy or Bonnier nasal cauterisation (R.A., D.F.), and were put on rest and diet (the Carmelite case) [read: Radiations and Waves pp.34-36, lines ~1405-1505]. He wrote: "while six cases are not sufficient, and the time elapsed is too short to speak of a positive cure", and on mechanism: "the oscillatory factor ... certainly has the most significant value, but for the moment, we cannot completely prove it" [read: pp.35, 38; lines 1480, 1635]. The arbiter's "Weaknesses" list for Claim 4 should name these co-treatments, and the clinician's own caveats belong beside the "several hundred patients" figure.
5. MWO power draw. "110 or 220 volts at 42 or 50 cycles ... only use 2 to 4 amperes", with a three-step power regulator and the patient 1.3-1.5 m from the transmitter [read: Radiations and Waves p.48, line 2080]. That is about 220-880 VA drawn from the mains [calc], the only electrical power figure for the MWO in the record. The arbiter's dose estimate uses a nominal 1 W radiated without this figure. The radiated fraction is unknown, so the conclusion "the far field does not heat" may still hold. The estimate should say what input power it assumes and show the range: 1 percent of 500 W radiated is 5 W, which gives about 0.4 W/m2 at 1 m [calc].
6. Attilj's collar cancer cases (Rome, Florence congress, May 1928). The record has a relapsing sarcoma of the hand that "almost completely disappeared" in two months, a recurrent breast sarcoma "no longer palpable", and recurrence in a post-mastectomy scar [read: Waves That Heal pp.21-23; Secret of Life p.58]. They come from a hospital radiologist, not from the seller, though they are relayed in the seller's publications. They are tumour-regression claims for the passive collar, the one device the arbiter rules out as an energy supply. The arbiter and primer give Attilj only as a record "not obtained". These reports should be stated in Claim 3 beside the collar verdict.
7. Testable side-claims in the clinical text. In a Paris professor's weekly series, improvement stalled for eight days at full moon. Patients living on conducting soils (Grenelle, Javel, Auteuil) responded faster than those on insulating soils (the "paradoxical conclusion") [read: Secret of Life p.59]. Thiais (3.36 per 1000, Seine water) and Orly (0.36, own wells) lie on the same limestone; Lakhovsky recounts Dr. Simeray's village (Académie de Médecine, 15 March 1927) where cancer appeared after the village changed its water supply [read: Secret of Life p.55]. These are the claim's own internal predictions, and the soil-cancer weighing ignores them (the physicist mentions Thiais once).
8. Animal reports offered against placebo. P. Fournier-Ormonde (Institut Physiologique de la Croix Blanche, Vaucresson) reported seven horses fitted with collars, with blood analyses showing more red cells and lower viscosity. The record also has a Brisbane vet's dog, January 1940, and the racing filly Ballerina [read: Waves That Heal pp.9-12]. The book argues that plant and animal results exclude suggestion [read: p.15]. No review answers that argument. A reply is easy (owner-reported, unblinded, no controls), but it should be made.
9. Smaller items, all pertinent and absent. - In 1941 Lakhovsky claimed that the 1934 Venice and 1937 Vienna congresses recognised him as the first inventor of the therapeutic short-wave oscillator, and that "more than 50,000" short-wave units were in use. Kotzareff 1931 credits him ahead of Schliephake (sites.md, Leads 2). No reviewer weighs the priority claim. - He cites Schliephacke (Giessen) and Esau (Jena), 1928-29: short waves killed mice and sterilised milk. That is the stated reason he left thermal short waves [read: Radiations and Waves p.10]. - Dr. Hulin (Salpêtrière dentist) reported re-ossification of pyorrhoea teeth, with radiographs, from calcium-chloride ionisation under the apparatus (Paris Society of Odontology, 20 July 1933) [read: p.27]. - In a pre-operative case, a leucocyte count fell from 20,000 to 5,000 in a week [read: p.29]. - Stadler's thesis and the New York Times article of 22 August 1926 (de Kruif) are an Investigator lead (sites.md, Leads 4) that no later file uses. - The 1941 book reports Lakhovsky's own Institute cases Nos. 1-4 (January-March 1940), with red-cell counts from under 3 million to 4.8 million [read: pp.6-7].

4. Accuracy spot checks (against _text/)

One inaccuracy found. Arbiter, absence-of-evidence section (repeated in primer section 6): "A 1941 New York hospital trial table appears in the 1941 book with no outcome record". The table does give an outcome per patient: 19 rows with results such as "marked improvement", "slight improvement", "no pain, swelling still present". The book adds seven case histories with outcomes and treatment counts, and a separate Brooklyn urologist table of 8 cases [read: Radiations and Waves pp.45-47]. What is missing is the hospital's own report or presentation (promised for the New York Society of Physiotherapists), any follow-up and any control group. The sentence should say that.

5. Retention check

• Investigator findings. Retained well. Kotzareff's 40 W and his "only experimental study"; Nature 1940 (Ritchie) with both halves of the judgement; Schereschewsky; the CH 135072 and CH 164027 family; COLYSA and the Swiss Colysa; the ncip.by death and the 1945 institute; Blum; the Lumen notices; Farrell; the 1950 Haut-Parleur list. All appear in the arbiter, the primer or both. Not retained:
• The silver-sterilisation patent series and the experiment it bears on (missed item 1).
• Kotzareff's priority statement over Schliephake.
• The Stadler thesis and the NYT 1926 lead.
• Reid 1989, which appears in the physicist and mathematician files only. That is acceptable.
• Material weakened because it was judged negatively. No case found where the arbiter or primer dropped material because a reviewer judged it negatively. The arbiter explicitly retains the clinical record ("Retained fully"), the Perineau on-off-on case and the Nature reviewer's concession. Two condensations do read as softening, though:
• "Failures admitted in advanced cases". This compresses five post-treatment deaths into the word "failures" (missed item 3). The compression cuts both ways: it removes evidence of an effect on tumours as well as an adverse-event signal. It is a retention defect under "Keep pertinent information": the deaths are pertinent to the Claim 4 safety verdict whatever their interpretation.
• Attilj. His collar tumour regressions are reduced to a "record not obtained" (missed item 6). Single-source evidence is still evidence; it should be stated and weighed, not only named.
• Claim 6 and the eugenic framing. The primer omits Lakhovsky's own framing: the three children as "a new race of unusual strength and perfection of form"; the proposed municipal decree requiring treatment of engaged couples; his claim that conscription unfitness would fall from 35-45 percent to 5 percent [read: Radiations and Waves pp.65-67]. These are pertinent to how the reader judges claim 6. The arbiter's claim heading mentions municipal treatment; the primer should carry them too.

6. Process

• Source tags. Present and honest throughout. Where they apply, [memory] facts carry a confidence and are flagged as non-load-bearing. Interested parties are marked: colysa.ch, Winey, the Novocure press release, Brown/Borderland, the COLYSA booklet and the Hatonn layer.
• Web checks. The arbiter made 7 of its 8 budgeted calls. It corrected the physicist's wrong Kirby citation (PMC2008681, which is Hieger and Orr 1954) and recorded the failed checks (the Nature 144:246 login redirect). Good practice.
• Verdict scale. Applied consistently, with exact scale names, and split verdicts are labelled by part.
• Corrections. The arbiter's "Corrections to earlier documents" section is exemplary. Corrections 1-10 check out where I could verify them (2, 3, 4, 5).
• Primer. It follows the STEP 4 order (sections 1-9, with answers at the end). The worked example is numerical and reproduced by code. Reader profile: SAR, Q, LC, TTFields and Fisher's exact test are defined. Not defined:
• "HR 0.63" (hazard ratio);
• "Frohlich's coherence proposal" and "biophotons", each named without a sentence of explanation;
• "crown gall" and Bacterium (Agrobacterium) tumefaciens, which are not introduced.

Minor fixes. - example.py. Runs cleanly (checked: .venv/bin/python example.py). Output matches the primer's numbers.

7. Corrections to earlier documents

1. arbiter.md (absence-of-evidence section) and primer.md (section 6, last bullet). "A 1941 New York hospital trial table appears in the 1941 book with no outcome record" is wrong as worded. The table records an outcome per patient (19 rows), and seven case histories follow [read: Radiations and Waves pp.45-47]. Replace with: "the 1941 book prints a 19-patient table with per-patient results and seven case histories from an unnamed New York hospital (1 July-21 August 1941), and an 8-case table from an unnamed Brooklyn urologist; the hospital's own report, any follow-up and any control are not on record."
2. arbiter.md Claim 4 and primer section 6 row 4c. "Failures admitted in advanced cases" should name the five deaths by haemorrhage that followed tumour necrosis after 4-5 treatments (three in Lakhovsky's own account, two reported by Rigaux), with Lakhovsky's explanation [read: Radiations and Waves pp.41-42].
3. arbiter.md Claim 4 weighing. The dose estimate should state that the only power figure for the MWO is a mains draw of 2-4 A at 110 or 220 V [read: p.48]. It should give the far-field range for 1-10 percent radiated, not only 1 W.

8. Remediation

Deep reads (most important first)

1. Radiations and Waves, pp.30-48 (Opus). Tabulate every clinical case in de Cigna's Vienna 1937 communication (pp.30-39), the "advanced cases" chapter (pp.41-42), and the New York hospital and Brooklyn tables (pp.45-47). For each case give: diagnosis and how it was confirmed (histology, radiograph), co-treatments (diet, lemon juice, nasal cauterisation, medication stopped, silver nitrate), number of sessions, follow-up, outcome and any adverse event. Quote de Cigna's own caveats. Record the operating data on p.48 (voltage, current, distance, session length).
2. The Secret of Life, pp.35-37 and 57-59 (Sonnet). Cover the silver-spiral sterilisation experiment: all tables, controls, Experiment III and its interpretation, the d'Arsonval/Sesari authorship and date. Cover the copper-spiral timeline (1924-28) and Attilj's six patients. Cover the lunar-phase and conducting-soil response observations. Keep what was measured separate from Lakhovsky's inference.
3. The Secret of Life, pp.5-16 (Sonnet). Cover the radio-station pigeon experiments (Paterna 1924-26, Casamajor's report, Kreuznach 1926): who ran them, distances, timing, the on/off contrast, and whether Lakhovsky gives a source. Also cover the duck-wing electrometer measurement (setup, 600 V), the Bombyx cotton-wool experiment and the semicircular-canal argument.
4. Radiations and Waves, pp.14-29 (Sonnet). Cover Lakhovsky's own case list, Rigaux's penis-cancer report (dates, radium and X-ray history, silver nitrate, diabetes), the Roversi letter, Dr. H.'s prostatitis figures, Hulin's dental radiographs and the pre-operative leucocyte case. Give the confirmation and co-treatment for each.
5. Waves That Heal, pp.9-24 (Sonnet). Cover the animal reports (Fournier-Ormonde's horses with blood analyses, the Brisbane dog, Ballerina), Attilj's case dates and outcomes, the Araujo series (300 patients, technique, contraindications), Dr. Francis on deafness and the anti-placebo argument on p.15.
6. Radiations and Waves, pp.1-13 (Sonnet). Cover the preface claims (Venice 1934 and Vienna 1937 priority recognition, "more than 50,000" oscillators, the Postma letter of 3 Sept 1933, Presner's own case) and Lakhovsky's Institute cases Nos. 1-4 (1939-40) with blood counts. Record what is claimed, by whom, and what sources are given.

Web checks for the remediation (arbiter or primer author)

• Engels et al. 2014 (Nature 509:353), anthropogenic electromagnetic noise and robin orientation: confirm the citation and the frequency band.
• The silver sterilisation note, C. R. Acad. Sci. 1929 (d'Arsonval, 15 April). Check whether it is on Gallica.

What each later file must change

• facts.json.
• The Secret of Life entry: add key claims for the pigeon and radio-station experiments and for the duck-wing measurement.
• Radiations and Waves entry: add the five post-treatment haemorrhage deaths, de Cigna's co-treatments and caveats, the 2-4 A mains draw, and the priority claim (Venice and Vienna).
• Update how_read after the deep reads.
• claims.md.
• Add a claim: "Animals (birds, insects) orient by emitting and detecting radiation; radio transmitters disorient homing pigeons (Paterna 1924-26, Kreuznach 1926)".
• Under claim 3, add the silver-circuit sterilisation experiment as offered "proof of cellular oscillation".
• Under claim 4, note the deaths after treatment.
• arbiter.md, "QE revision" section.
• Give a verdict for the new instinct and pigeon claim. Probably split: radio interference with homing is an observation class later science has studied; "animals navigate by their own emitted radiation" is Not testable as stated, or Tested, not supported where echolocation or olfaction now explain the cited cases. Keep both halves with sources.
• Give a verdict for the sterilisation experiment as evidence of oscillation. Probably: the effect is plausible and consistent with known silver chemistry; as proof of oscillation it is not supported, and its own Experiment III points to a dissolved agent. State that the measurement itself is retained.
• Re-word Claim 4's "No contra-indication" weighing to include the five deaths and their two readings.
• Add the de Cigna co-treatments to the weaknesses, and his caveats beside the "several hundred" figure.
• Add the MWO mains draw to the dose estimate.
• Under Claim 3, state the Attilj collar regressions.
• Apply corrections 1-3 above.
• Note the priority claim and Kotzareff's corroboration in the historical record.
• primer.md.
• Section 6: add rows for the pigeon and radio claim and for the sterilisation experiment.
• Rewrite row 4c and the "where the record goes dark" bullet (corrections 1-2).
• In "material kept on the record", add the deaths after treatment, the de Cigna co-treatments, Attilj's regressions, the animal reports and the eugenic framing of claim 6.
• Define hazard ratio, crown gall (Agrobacterium tumefaciens), biophotons and Fröhlich coherence in one sentence each.
• Section 7, "Genuinely open": add "Do radio transmitters disturb homing birds, and by what mechanism?" with the modern lead.
• Recheck (Sonnet). Append a "Recheck" section here listing which of the items above were fixed.

Recheck

Date: 2026-10-08. Checked arbiter.md ("QE revision" section, lines ~160-217), primer.md (QE revision header, sections 6-7), claims.md and facts.json against the items above. Method: grep and reading of the revised text; I did not re-read _text/ again, so the revised numbers are taken as the deep readers recorded them, except where they match my earlier spot checks.

Missed material (section 3)

1. Silver-circuit sterilisation: FIXED. arbiter.md has a full entry (d'Arsonval, 15 April 1929, Sesari; Experiments I and III; no replicate, no inert-metal control). The measurement is retained and "proof of oscillation" is judged not supported. Also in claims.md (claim 3) and the primer.
2. Radio stations and homing pigeons: FIXED. New claim 8 in claims.md, primer row 8a and "Genuinely open" item 5, an arbiter verdict split, Engels 2014 recorded as related work and web-checked.
3. Deaths by haemorrhage: FIXED. All five deaths are in facts.json, claims.md, arbiter (Claim 4 verdict reworded to "Internally inconsistent as printed") and the primer. Both readings (treatment-induced necrosis, late course of disease) are kept.
4. De Cigna co-treatments and caveats: FIXED. The "several hundred" figure is corrected (8 radiographed ulcer cases, later 18). Co-treatments and his two caveats are in the primer and arbiter.
5. MWO power draw: FIXED. Mains figure of 2-4 A at 110/220 V (220-880 VA) is in the arbiter's dose section and the primer, with the unknown radiated fraction stated.
6. Attilj's collar cases: FIXED. Stated and weighed in the arbiter (confounds named; Claim 3 verdict "Never properly tested", energy-supply verdict unchanged).
7. Testable side-claims (full-moon stall, conducting versus insulating soil in Paris, Thiais/Orly, Simeray's village): REMAINING. These are in facts.json and claims.md, but arbiter.md and the primer do not mention them (no hits for moon, Simeray, Thiais, Grenelle). The soil-cancer weighing still ignores these internal predictions.
8. Animal reports against placebo: FIXED. Arbiter paragraph and primer row 3f and bullet answer the "animals exclude suggestion" argument.
9. Smaller items: PARTLY FIXED. Priority claim and Kotzareff's corroboration, and Schliephake/Esau, are in the arbiter and primer. Stadler/NYT 1926 is in the arbiter only. Hulin's re-ossification and the leucocyte case are in facts.json and the arbiter, not the primer. Institute cases Nos. 1-4 are in facts.json only. All low weight; none changes a verdict.

Section 4 inaccuracy and section 7 corrections

• "No outcome record" for the New York table: FIXED in both files. Arbiter and primer give the 19-patient table, seven case histories and the 8-case Brooklyn table, and say what is missing (hospital report, follow-up, control). Marked as corrected in the primer header.
• Corrections 2 (deaths named) and 3 (dose uses the one power figure): FIXED.

Section 5 retention and section 6 process

• "Failures admitted" compression and Attilj softening: FIXED (see 3 and 6).
• Claim 6 eugenic framing: FIXED. In the arbiter and primer section 6 (own words, municipal decree, conscription figures).
• Primer definitions: FIXED for hazard ratio, crown gall (Agrobacterium tumefaciens), biophotons and Fröhlich coherence.
• Steps 2-4 were not redone, as intended. The central verdicts are unchanged.

Remaining

1. Item 7: the lunar-phase, soil-conductivity and Thiais/Orly/Simeray predictions are missing from arbiter.md and primer.md. Add a short paragraph weighing them (the Contributo table the mathematician read is the relevant test) to Claim 6 or the soil section.
2. Item 9: add one sentence each to the primer for the Hulin and leucocyte cases and the Institute cases Nos. 1-4, or state that they were set aside as non-load-bearing.

Final: OPEN (2 issues)

Orchestrator fix (2026-10-08)

The arbiter, acting as orchestrator, cleared the two "Remaining" items of the Recheck. Sources read for the fix: The Secret of Life pp.51-59 (_text/), the Contributo table and Seine legend (research/files/1927-Contributo-eziologia-cancro_djvu.txt), mathematician.md Claim 5, and deep/Radiations and Waves sections 4-6. No web calls. Git and render_site.py were not run.

Correction to the Recheck: it said all four side-claims "are in facts.json and claims.md". That was partly wrong. facts.json had the full-moon and conducting-soil observation only; Thiais/Orly and Simeray were in neither file; claims.md had none of the four. Fixed at the source, as below.

Added: 1. arbiter.md, "QE revision" > new subsection "Claim 5, soil section (addition, orchestrator fix 2026-10-08): the side-predictions". The soil claim is Claim 5 in this arbiter, not Claim 6. The subsection weighs each prediction, says whether Lakhovsky's printed data test it, and gives a verdict: - Full-moon stall: Never properly tested. No data were printed, and a chance overlap with an 8-day full phase has a probability of about 0.5 [calc]. - Conducting soils respond faster: Never properly tested. The Contributo gives mortality, not treatment response. The two lists of districts are the two ends of that table, and no conductivity was measured. - Thiais/Orly and Simeray's water supply: Never properly tested. The electrical mechanism is Not testable as stated. Thiais/Orly is, in Lakhovsky's own words, a counter-example to the soil rule that he explained afterwards. Limestone is classed as insulating on pp.51 and 53 and as a conductor on p.55. - The soil-geography verdict is unchanged (Never properly tested). 2. primer.md, section 6 evidence table: new rows 5e (full-moon stall), 5f (conducting-soil response) and 5g (Thiais/Orly, Simeray). Each has its verdict. 3. primer.md, section 6, "Material kept on the record": a new bullet, "Side-predictions of the soil theory", written for the reader. It covers the Thiais/Orly lesson, the shifting rock classes and a small-numbers illustration. 4. primer.md, the same list: a new bullet, "Smaller case reports, kept but not load-bearing". It covers Hulin's re-ossification case, the 1935 leucocyte case and Institute cases Nos. 1-4, each with its weight (low) and the reason. 5. primer.md header: a note "Orchestrator fix (2026-10-08, after the QE recheck)". 6. facts.json, The Secret of Life entry: a new key_claim for Thiais/Orly and Simeray (p.55). 7. claims.md, Claim 5: an "Also offered under this claim" line listing the four side-predictions.

No central verdict changed.

Check against the two remaining items

• Item 1 (side-predictions in arbiter and primer, weighed against the Contributo table): fixed. The arbiter now has moon, Simeray, Thiais and Grenelle. Each prediction is stated, the Contributo figures are used (Grenelle 2.08, Auteuil 1.76, Javel 1.61, Gaillon 0.324, Porte-Dauphine 0.53, Champs-Elysees 0.68, Plaine Monceau about 1.07-1.08, and the Brie band 1.26-2.00), and the verdicts use the prompt.txt scale. The primer has the matching rows and a reader-level paragraph. I spot-checked the quotations against _text/ (p.55 "did not appear to be in accord with my theory"; p.59 "paradoxical conclusion"; p.53 "insulating soil (sand, limestone, etc.)") and the figures against the Contributo OCR. One limit remains: the Thiais and Orly values cannot be read in the OCR of the Contributo's Seine map, so they come from The Secret of Life alone. The arbiter says so.
• Item 2 (Hulin, leucocyte, Institute Nos. 1-4 in the primer): fixed. Each case has its own description and a stated weight, with the reasons taken from the deep read.

Final: PASS