Morray-TH: Quality Engineering (Step 4b)

Reviewer: QE agent (Opus). Date: 2026-10-03. Inputs checked: _text/manifest.json (105 documents, 262 chunks), all 262 digests in digest/, facts.json (42 entries), claims.md, physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. Spot checks were made against the extracted text in _text/. Context: Steps 1-4 ran before the digests existed. Step 1 sampled the large documents and had no OCR text for the 4th- and 5th-edition scans. Those scans now have full OCR text (4thed-4e182f, 4thed-a49807, 5thed-0d3c5c), and the digests show material that the reviews never saw.

Verdict: MAJOR. The physics refutations (Claim 3 energy budget, resonance, Appendix III/IV errors) are sound and will not change. The evidence base for Claims 1, 2, 5 and 6 was built from a sample of the central book. The witness record from 1929-1938, which includes multi-kilowatt loads, instrument checks, a state securities-commission inspection, a test that Fletcher says he arranged and Moray did not attend (Moray called this "100% untrue"), and a failed rebuild by Moray's own patent agent, was never read. Some verdict wording and the tested / never-tested / dismissed classifications depend on that record, so Steps 2-4 should be redone after the deep reads below.

---

1. Coverage

2. Shortcuts

3. Missed material (from the digests; not in facts, claims, reviews, arbiter or primer)

Witness record and tests (Claims 1, 2): 1. Larger observed loads, 1929-1938. Taking the lamp counts at their face value: - Yates, 13 Dec 1930: 32 x 50 W lamps plus a 575 W iron, about 2.2 kW (4th ed. p.179-180). - Lovesy, Sep 1930: 2,150 W (5th ed. p.56-58). - Welling (Utah Secretary of State) with Mexican Consul Gaxiola, 5 Oct 1930: 10 x 100 W plus a 575 W iron; the house meter was checked (4th ed. p.182-183). - Gardner, 1936: about 40 x 200 W, about 8 kW (5th ed. p.113-115). - Todd, 1937: about 40 x 100 W (5th ed. p.110-111). - Pyper, 1936-37: 30-50 lamps; the house main was disconnected and the house was lit from the box (5th ed. p.133; 4th ed. p.156). - A. B. Jensen, spring 1938, for the Utah Securities Commissioner (Mr. Gull): 30 x 150 W lamps, a 1000 W iron and a motor for over an hour, about 5.5 kW. Meters were placed on the antenna and ground leads and found them "electrically dead". Every part was removed, weighed and tested with volt and ammeters for battery action, and the commissioner bored holes in the box (4th ed. p.132, verified in text). - A. B. Jensen, July 1938, a desert test near Johnson's Pass, Tooele County, 25-30 miles out: 35 x 150 W plus a 1000 W iron (5th ed. p.111-113). - Bringhurst, July 1938: 20 x 150 W plus 15 x 100 W plus 600 W plus 575 W, about 5.7 kW (5th ed. p.108-109).

The arbiter's statements that 4 kW and above was "never observed" and that the record covers only "1925-28" are therefore wrong as statements about the documents. They remain unmetered lamp counts by named witnesses, many associated with Moray and some state officials, and they should be listed and weighed. 2. Instrumented checks existed, though they were not power metering. Yates (1929-30) opened the building's main switch, tested whether the terminals discharged like condensers or batteries, emptied the boxes and put a crystal set on the lead-in. Jensen (1938) used meters, as above. The "no instruments" framing in the arbiter (Claim 2: "critics did not insist on instruments") needs correcting. 3. A test that, by Fletcher's account, Moray did not attend. In his 1956 and 1957 letters and in Foster's 1973 letter, Fletcher says he arranged for a New York electrical concern (Bell Labs, per Foster) to build and test the apparatus, and that Moray agreed but did not come (5th ed. p.80-88). Eyring (1963, via Vickers) said he saw the lights but could not attest because the hand-held part was withheld. This is central to the contrarian's "dismissed without testing" versus "never tested" distinction. On Fletcher's account (a participant with a stake; Moray called it "100% untrue", 5th ed. pp.84-88, and no telegram survives) it would be "test offered, inventor declined", a possibility no reviewer considered. 4. A failed rebuild. Murray O. Hayes, Moray's patent agent, who said Moray had "explained all to me", testified in Reno (1932) that he and his associates tried to build the machine from the patent papers and failed (4th ed. p.169, verified in text). This is a non-neutral replication attempt with a negative result. It is not in any review. The arbiter's only "failed duplication" is the 1965 gold work. 5. Knudsen (Dr. Vern Knudsen; the book spells it "Knudson"), a physicist, suspected hidden wires in an antenna pole (5th ed. p.123-125). This is a contemporary critical hypothesis that the arbiter's open questions ("was any lead metered?") should cite. 6. REA engineers. Ben Crim (1938) reported tests "entirely satisfactory as far as performance went" (5th ed. p.118-119, 174). Frazer's 1939 "glass test" used 12 sheets of window glass in series (5th ed. p.181). 7. Later devices. A 1943 rebuild with the remaining "Swedish stone" was run in a car from Salt Lake City through Idaho, Wyoming and Colorado, with a failed Denver demonstration (5th ed. p.191-193). Polly's file records demonstrations on 22-23 June 1957 with about a dozen witnesses. Fate 1956 says the antenna and ground were later no longer used, and that the longest continuous run was 57+ hours. These conflict with the reviews' picture of no device after 1939. 8. The 1938 E. G. Jensen letter. A 60 W lamp and condensers sat on the primary (antenna) side, and the secondary lamp went dark if the primary lamp was removed (5th ed. p.127-128; Media/photos p.4-6). This is a technical detail that bears on where power entered.

Theory and technical (Claims 3, 4):

9. The 4th ed. (p.116-117) and the 1962 speech (p.3-4) state that the first stage is "excited externally" and is then self-sustaining. This is the stored-energy / priming question; the arbiter raises it only through Hayes's magnet stroking.
10. Performance figures conflict and should all go in the inconsistency list:
    • 1962 speech: "about 1 lb per kW", a 60 kW unit, 100 lb for 300 kW.
    • 4th ed.: only 10-20 tubes in 100 meet requirements, at $500 per tube with 29 tubes per unit, and a $5.38 M development budget.
    • 1962 speech: 2 tubes in 100.
    • Hayes 1931: 4,000 W, with 50 kW "potential".
11. The Nuenergy alternative-mechanism hypothesis (spark-gap stations, then radioactive dopants) and the report that surviving tubes were physically inspected (Section 2).
12. Radiant Energy "1926", p.32-34: the device described as a corona/catalyst ion reactor, and claims of splitting nuclei of "paper, lead, tantalum…" and making tritium by electrolysis (p.61).

Claim 5:

13. The 1940-41 gold assays in the 1945 pamphlet and in Fate 1956: Boaz Mine, Norris MT; Union Assay Office SLC, serial 88193. These predate the 1965 work and are not mentioned.
14. Isotopes: Bureau of Mines radiometric analysis and AEC statements on uranium-ore upgrading (Section 2). The 4th-ed. résumé (p.188) cites AEC assays "up to 85% U3O8".
15. Rex Research: the with/without-solution comparison and the outside accelerators (Section 2). The arbiter says "no blanks", but a crude control does exist. Assay-flux contamination is still the obvious artefact to test.
16. Photo Actinia clinical table and the USDA 1940 request (Section 2). The health-frauds hearing (p.85) records a lessee's report of no improvement and the lease price ($3,210).

Claim 6:

17. Second patent action: the gov-Commerce letter image (see Section 4) and a ~1932 divisional requirement.
18. 1961: NASA first agreed to a $5 M program, then a 21 Jul 1961 letter judged the probability of success too low, and Sen. Bennett's 30 Oct 1961 letter says Moray asked for $50,000 without disclosing the device location or process (5th ed. p.99-101). This is both sides of the "Bennett killed it" story.
19. AEC (C. E. Larson to Kerber, 1970 fragment): Eyring "convinced there was no demonstrable evidence".
20. Moray Products Co. litigation (Farnsworth, Schade, Hayes; Reno 1932), the 1925 offer to Sen. Smoot, and the 1940 shooting narrative naming Frazer and an "FBI man". This is history the arbiter treats only in summary. It is fine to keep it brief, but the Hayes failed-rebuild item (point 4) comes from this litigation.

4. Accuracy spot checks (against _text/)

5. Process

• Source tags: present throughout, and the [read] / [web] / [memory] / [calc] tags are used honestly. The arbiter lists the [memory] items it did not check (Li-ion 250 Wh/kg, Nagaoka date, Coulomb barrier, etc.) and says no verdict rests on them alone. OK.
• Web checks: physicist 10, mathematician 8, contrarian 4, arbiter 5 of 8 budget. The contrarian used few, but its medium-confidence items (MMRTG) were rechecked by the arbiter. OK.
• Verdict scale: applied consistently, with split verdicts stated. Claim 6(i) uses "Established" for documented historical events. That is a stretch of a scale meant for physics claims, but it is explained. One issue: the tested / never-tested / dismissed analysis omits the Fletcher-arranged test and the Hayes rebuild (Section 3), so that classification is incomplete, not mis-applied.
• Primer: follows the STEP 4 order (sections 1-9 plus answers). Symbols are defined, prerequisites have textbook pointers, and the worked example has real numbers. It suits the reader profile well. Content defects come from the inputs: Section 1 frames the demonstrations as "1925-1928" and "never metered"; the evidence table says 4-50 kW was "never observed"; and the source-gap note is stale.
• example.py: runs cleanly in the venv, reproduces every number quoted in the primer, and is commented. OK. After remediation it should add one line for the 1,193,600-lamp figure and a battery check for a 5.5 kW load over 1 h (5.5 kWh, about 160 kg lead-acid at 35 Wh/kg, against a box whose parts were weighed in 1938).

6. Remediation actions

Deep reads (most important first)

1. Books-Works-by-Moray-Sea-of-Energy-Sea-of-Energy-in-Which-the-Earth-Floats-5thed-0d3c5c, pages 32-133, Opus. List every dated demonstration and sworn statement 1925-1978 with load, witnesses, distance from power lines, what instruments or checks were used, and whether the antenna and ground were used. Extract the full Fletcher, Eyring, Wilkinson, Foster and Bennett correspondence: who offered which test, and who declined. Record the Knudsen hidden-wire suspicion and the Hayes statements.
2. Books-Works-by-Moray-Sea-of-Energy-Sea-of-Energy-in-Which-the-Earth-Floats-4thed-4e182f, pages 116-197, Sonnet. This is Moray's own 1960 history. Extract the Yates, A. B. Jensen (Securities Commission), Welling, Pyper, Wood & Webber, Hayes and Pearson statements; the Reno failed-rebuild testimony (p.169); "excited externally" (p.116-117); the development-status figures; and differences from the 5th edition.
3. Books-Works-by-Moray-Istopes-in-Ionic-Recovery-of-U308-004a4b, pages 34-40, Sonnet. Who performed the Bureau of Mines and AEC analyses? Were the before and after samples the same material? Are there controls? Do the claimed upgrades require transmutation, or are they consistent with ordinary concentration or contamination?
4. zz-Misc-Rex-Research-Thomas-H-b42bf5, pages 1-22, Sonnet. The design of the 1963-1970 gold tests: the with/without-solution controls, the assayers and accelerator operators, the blanks, and the Rudolph statistics. Does this change "no blanks / no third-party assay"?
5. zz-Misc-Nuenergy-c70186, pages 9-21, Sonnet. The spark-gap-station and radioactive-dopant mechanism, the claimed inspection of Moray's surviving tubes, and the sources the author cites for each.
6. Books-Works-by-Moray-Photo-Actina-Oxygen-Therapy-actinia-a66d5e, pages 22-54, Sonnet. The clinical results table (who compiled it, controls, follow-up), the USDA 1940 request, and the device description compared with US 2,460,707.

File changes

• facts.json:
• Update how_read and text_available for the 4th- and 5th-edition scans, Isotopes, Photo Actinia, the audio/video transcripts and the Media/photos OCR.
• Add the witness statements, the Securities Commission inspection, the Fletcher test offer, the Hayes failed rebuild, the Knudsen suspicion, the Bureau of Mines and AEC assays, the clinical table, the 1940-41 assays and the 1943/1957 device activity as key_claims.
• Redate the "1926" pamphlet as a post-1945 compilation and fix 119,360 W to 119,360,000 W.
• Fix the FDA 8137 summary (contested; motion to vacate denied 20 Aug 1963).
• Give the gov-Commerce entry its own summary with the "no natural source" text.
• claims.md:
• Rewrite the Source-gap note: the scans are now OCR'd and read via digests and deep reads.
• Claim 1: add the 1929-38 observed range (about 0.6-8 kW lamp counts) separately from the 50 kW specification.
• Claim 2: add the Securities Commission and Yates checks.
• Claim 6: add the Fletcher test offer, Hayes rebuild and NASA/Bennett 1961.
• Numbers: give Moray's own 1,193,600 alongside the sons' 1,693,600.
• physicist.md, mathematician.md, contrarian.md (redo Step 2):
• Re-answer a-d for Claims 1, 2, 5 and 6 with the new record.
• The contrarian must assess the Securities Commission inspection, the multi-kW desert test and the Crim report, and must also apply the same rigour to Moray's non-attendance at the Fletcher-arranged test and to Hayes's failed rebuild.
• All reviewers should correct the patent-rejection grounds.
• Battery and stored-energy calculations should be redone for kW-hour loads, not only 110 W.
• arbiter.md (redo Step 3):
• Correct the Claim 1 strong-form wording ("no document records it ever being observed").
• Correct the Claim 2 instruments statement.
• Add "test offered, inventor declined" and "non-neutral rebuild failed" to the tested / never / dismissed classification.
• Correct the Claim 5 wording on assays, blanks, clinical data and FDA.
• Correct the Claim 6 patent-ground statement.
• Add the uranium-upgrade claim.
• Add the Knudsen hidden-lead hypothesis and the "excited externally" priming question to "What would settle it".
• primer.md (redo Step 4):
• Update the Source gap.
• Revise Section 1 (demonstrations ran 1925-1938, with later activity).
• Revise the Section 6 evidence table rows for Claims 1, 2, 5 and 6.
• In Section 7, give the learner the strongest form of the witness record next to the energy budget. The lesson is that many lamp counts by named witnesses are still not a metered input/output test, and that the decisive tests were offered and not completed.
• Add the 1,193,600 figure.
• example.py: add the two checks listed in Section 5.

Recheck

Rechecker: Sonnet, 2026-10-03. Checked the current files against the defects (Sections 1-5) and remediation actions (Section 6) above. Evidence: deep/ (6 new reads), grep of each file, facts.json (parses, 105 entries), example.py run in the venv (exit clean, new Step 5 and Step 6 output).

Remediation actions: deep reads

Defects and file changes

Remains (minor, do not affect any verdict)

• Radiant Energy "1926" pp.32-40 and 41-60 are still known only from the digest (facts.json says so openly). It was not in the Section 6 deep-read list, and the content is a peripheral description with no verdict resting on it.
• Rounding: arbiter.md says 1.37x for Moray's own figure, example.py prints 1.38x (2.39e17 / 1.74e17 = 1.373 vs 1.38 as printed from unrounded constants). Cosmetic.

Final: PASS

Evidence-rule check (2026-10-03)

Scope: arbiter.md (260 lines) and primer.md (293 lines) against the three SOURCE HONESTY rules in prompt.txt (no suppression of evidence; interested parties; absence of evidence). Read in full.

Passes: - No suppression of evidence. Single-source accounts (the 1939 destruction, the 2 Mar 1940 shooting, the 1943 threat, Eyring's refused wattage test, Perreault's tube inspection, the Senate lessee, Betz's note) are quoted, sourced and weighed, with "no second source found yet" wording (arbiter.md:50, 89, 112, 118-119; primer.md:187, 235). Silence in newspapers is stated as not contradiction (arbiter.md:119, 130). Evidence against Moray is also stated (Hayes rebuild, 1942 Sanford judgment, Wood's "crackpot" wording, Frazer's counter-claim). - Interested parties. The Patent Office, FDA, NASA, Sen. Bennett, AEC/Eyring, Millikan, Bell Labs, Knudsen, Fletcher, Craig, Halleck, Bearden, the family and Perreault are each reported as that party's characterization with the stake named (arbiter.md:24, 61, 111, 121, 123-124; primer.md:179-182, 187). - Absence of evidence. Silences A-D (device and detector after 1939; 74-page FBI file; missing REA, Senate, Securities, Hayes, USDA and patent records; no record of the 1940 shooting) give competing explanations, for and against, and the deciding record (arbiter.md:132-156; primer.md:192-216).

Remaining violations: 1. arbiter.md:211 ("Changes after QE", Claim 4) still calls Perreault's tube inspection "the hearsay tube inspection". That is the dismissive label the rule forbids, and it contradicts the body (arbiter.md:89), which quotes it as one published first-hand source with low weight and no second source found yet. Fix: replace with "Perreault's tube inspection (one published source, low weight, no instrument or method given)". 2. Evidence against a claim omitted by an arithmetic slip (rule 1, "cuts both ways"). arbiter.md:21, 99 and 212 give 125 J per gram of gold residue and use it to withdraw the energy-budget argument ("does not refute it"). primer.md:190 correctly shows 9.75e19 atoms/g x 8 MeV x 1.602e-13 J/MeV = 1.25e8 J/g (a factor of about 1e6 more) against a reported dose of 5-40 J/g, so the energy argument stands beside the nuclear one. The arbiter (stated as authoritative, primer.md:3) still carries the wrong number and the withdrawal, so the two files disagree and the arbiter understates evidence against the gold claim. Fix: correct arbiter.md:21, 99, 212 to match primer.md:190. 3. Absence-of-evidence gap (rule 3). The Research Institute (1963-70, rented accelerators, funded work), its gold and uranium claims, and the 1968 Air Force proposal go quiet with no later papers, outside replication, assay certificates or recorded outcome (arbiter.md:92, 102-103; primer.md:184-185). The record treats this only as "Never properly tested". It is not listed among the Silences sections or analysed with competing explanations (commercial failure; trade secrecy; absorption into another body; the Institute's own non-reproducible results; suppression) and the deciding records (Institute and Hooper et al. papers, the unnamed assayer, accelerator-facility logs, the 1968 proposal's reviewing office). Fix: add a silence E to both files.

Final: OPEN (3 issues)

Evidence-rule recheck (2026-10-03)

Scope: the three remaining violations of the "Evidence-rule check" above, plus outdated wording in claims.md and in this file. Checked by grep after editing; example.py re-run (gold Step 6 prints 1.25e+08 J per gram).

Not counted as evidence-rule items, outside the files this recheck may write: contrarian.md (Claim 5 a) still states "about 125 J per gram of residue"; example.py's printed note still says arbiter.md writes "125 J per gram" (now true only of contrarian.md). Both should be corrected when those files are next revised. The arithmetic in example.py itself is correct.

Final: PASS

Investigator update check (2026-10-04)

Scope: the "Investigator findings (2026-10-04)" sections of physicist.md, mathematician.md and contrarian.md; arbiter.md (the same section, plus the updated Silences, Changes and Web log); primer.md (all lines marked 2026-10-04); sites.md leads 1-28 and research/ notes. Checked against the saved source text in research/files/ (Senate hearing pp.1344-1345 and 1370-1376, FBI log excerpts) and by recomputing the carrying arithmetic.

Passes: - Coverage and retention (most items). Senate record (REA payroll dates, Wood, Frazer, Tugman, Slattery/Craig motor-and-fans, the assassination-story rumour), Shinkle's four index sheets, FBI 917273 and the 2013 "No Records", CIA F-2000-01342, the 1946 treatise, Tesla 685,957, Losev, Lilienfeld, Bearden, Hagmann, the Ottawa Bray fonds, the Carmody boxes, the 1934/McConnell/Ginex figure drift, replications and the unread records are each in arbiter.md and, in condensed form, in primer.md, with the reviewers' disagreements ruled on. Quotes checked verbatim against the saved text (Wood's "unmitigated ass ... stumbled onto something", Frazer's "next 2 years", Tugman's 12 years and "very clever hoax", the $25 a day, the Slattery lines on the aluminum disk and the Germans). Negative material is kept next to favourable (Tugman's hoax opinion, "negative conclusions", the energy-budget refutation, the Patent Office grounds). - Source tags and SOURCE HONESTY. [web] items carry URLs; [memory] items carry confidence and the arbiter names the ones it did not check (FBI retention practice, electrostatic fans, Harvard 1941, carnotite). Interested parties (hostile Senate staff, Slattery, Tugman, Electrical World, Halleck, Bearden, the family) are labelled with their stake. Single-source items are quoted and weighed, not labelled away. The FBI and Senate absences are analysed with competing explanations and the record that would decide. The arbiter's corrections of reviewer overreach (Wood's sentence, "two months" vs "2 years", the 1934 lamps) are explicit. - Arithmetic re-done. 742 days = $18,550; 1,540 kWh and about 44 t of cells; 6.5e7 J and 130 F; slow-fill 1,700 s and 0.36 mW; 9.75e19 atoms/g, 1.25e8 J/g, 13 eV/atom; 5.4e-9 W/m2 and 1.2e11 to 5e12 m2; A_eff 1.2e4 m2 at 1 MHz; 5.3e5 m2 at 1 V/m. All agree, except item 2 below. - Primer against arbiter. Verdict labels for Claims 1-7 (5d split, Claim 4 split, Claim 6 reweighting, shooting kept at Never properly tested with the reviewers' label noted) match the arbiter. Primer's 31 microwatt figure is correct for the 1 MHz dipole.

Defects: 1. Retention: Slattery's Tomadelli answer is missing from arbiter.md and primer.md. In the saved Senate text (pp.1372-1373, research/files/HT_1944_Senate-REA-hearing_pt4-5_pp1370-1376...) Slattery is asked whether "Count Tomadelli" (convicted in 1920 of mail fraud over a bulb said to take energy from the atmosphere) claimed anything different from Moray and answers "I think they are pretty similar", says he "began to be very suspicious about the whole thing", calls the Tomadelli article to Mr. Lake's attention on 23 Nov 1942, and, asked whether he acted to end Moray's employment, answers "I think I did". physicist.md (Section B) records the comparison; the arbiter's "Why REA stopped paying" row and Claim 6 give only Sen. King's "nut" as Slattery's reason and say "Nothing found was dropped because a reviewer judged it negatively". This is pertinent evidence unfavourable to the claim (and bears on the termination reason and date) that was dropped. Fix: add it to the arbiter's Senate-record table row and Claim 6 (i)/(ii) as Slattery's characterisation (an accused party with a stake, who never met Moray), and one clause in primer.md Claim 6 row and Silences "1944 Senate REA hearing" item; correct the arbiter's "nothing was dropped" sentence. 2. Arithmetic wording (arbiter.md Claim 3). "1.4 kW needs A_e about 5e5 m2 ... about 1 V/m [calc]. At 1 mV/m the same aperture gives about 30 microwatts." For 5e5 m2 at 1 mV/m the power is 1.3 mW [calc: (1e-3)^2/377 x 5e5]. The 30 microwatt figure (31 microwatts) is for the 1.2e4 m2 aperture of a 1 MHz dipole, as primer.md states correctly. Fix: say "the 1 MHz dipole's 1.2e4 m2 aperture". 3. FBI 2013 entry incomplete (arbiter.md Silence B and primer.md Silence B). The saved log line reads received 12/01/2011, reopened 02/08/2013, closed 02/11/2013, 0 pages, "No Records". Both files give only received and closed. The 3-day reopen-to-close interval is relevant to explanation 3 (narrower or follow-on scope), and arbiter.md says "the log has no scope field" without noting the reopen date. Fix: add the reopen date to both and to explanation 3.

Notes (not counted): Electrical World (11 Nov 1944) is in arbiter.md but not primer.md, and the debt-suit list (Ashton-Jenkins, HOLC, Roy Ware, Brewster) is summarised in the arbiter and reduced to Sanford in the primer; both are acceptable condensation as the arbiter judged them context only. arbiter.md and the reviewers use slightly different multiples (85 vs 90 times) for the 1934 line; the arbiter's 85 is right for 18 kWh.

Final: OPEN (3 issues)

Recheck (investigator update, 2026-10-04)

Checked by grep against the current arbiter.md and primer.md and against the defects above.

Remaining: none counted. The notes in the update check (Electrical World and the debt-suit list condensed in the primer; 85 vs 90 multiple) were not counted as defects and stand as acceptable.

Final: PASS

Investigator update check (2026-10-08)

Scope: the "Investigator findings (2026-10-08)" sections of physicist.md, mathematician.md and contrarian.md; arbiter.md (the same section, plus the extended Silences A, B, G and the Web log); primer.md (all lines marked 2026-10-08); sites.md leads 29-43; research/ notes 2026-10-06 to 2026-10-08 (Q01, Q02, Q04, Q08, Q09, Q11, Q12). Checked against the saved pages in research/pages/.

Passes: - Coverage and retention (most items). All 15 leads (29-43) are weighed in the arbiter and carried into the primer: the Le Bon, Glasser and Preston provenance and the dating of the "1926" printing, Moseley's cell, the 1968 infrared paragraph, phosphors and Noffsinger's list, resonance and Keely/Tesla, the thorium and folded-cylinder passages, Rutherford's excited activity, the Elster-Geitel / Ramsay-Spencer / Millikan-Winchester / Campbell record, Plauson, Ives (arbiter only; no verdict bearing), the diamond passage, Foster 1990 and the Q09 figures. The dissenting readings (58,000 against 1.7e5 Ci; 1e-19 against 2e-17 stopping fraction; stored energy against short circuit) are stated side by side, not collapsed. Single-source items (Foster, Noffsinger, the sons' "withdrawn from libraries" and 1942 FBI question) are quoted, attributed, stake noted, and labelled "not found, which is not the same as untrue". No lead was dropped for being judged negatively. - Source tags and SOURCE HONESTY. [web] items carry URLs; [memory] items carry confidence; the arbiter lists what it did not re-fetch (Moseley's abstract, Borexino, Elster-Geitel via Lodge) and what stays [memory]. Interested parties are attributed (Campbell as rival worker, Perreault as advocate, Keely's company engineer, Plauson as patentee, the Moseley 150 kV and 0.01 nA marked as Wikipedia, not the paper). Absence of evidence is analysed (new Silence G with three competing explanations and deciding records; Silences A and B sharpened). The arbiter's own corrections (Moseley "outer sphere", Weber "never confirmed", 108mAg decay, "17.5 g" to mg) are stated plainly. - Spot checks against saved pages. Franklin 2010 "526 times smaller" (present); Ramsay and Spencer Table IX mercury 216 secs, tin amalgam 0.6 secs (present, Phil. Mag. 12, p.413); Athenaeum "Mr. Carse has tried and failed completely to repeat another of these experiments" (present); Le Bon "taken up anew" and "Cape diamonds never" (present in lebonmat.htm); Kragh: Le Bon "In 1903 he was even nominated for the Nobel Prize in physics" (present); Foster quotation matches the Q09 note. - Arithmetic re-done. 17.6 mg/yr and 4.8e-8 kg/day; 58,400 Ci and 1.69e5 Ci per kW; 6,420 Ci and 1.35e20 events for the 1928 run; Coulomb barrier 9.8 MeV; 5.8 million shortfall at 1.7 eV; 295 nm = 4.2 eV; 1.2e5 ion pairs per alpha; 1.5 microwatts and the 0.24 nA ceiling; 4.8 J/cm3; 20 kJ against 5 J (4,161x); CMB peak 1,063 um; 3.1e-6 against 460 W/m2; 6e-8 W on the thoria wire. All agree. example.py runs (rc 0, with the project .venv; system python lacks sympy) and Step 7 prints the same numbers. - Primer against arbiter. Verdict labels for 5a, 5e, 5f, Claims 3, 4 and 6 and the new parts match the arbiter's "Verdict changes" table; Silence G and the extended A and B match; the primer's "What would settle it" additions match arbiter section 10.

Defects: 1. Retention: the defending reply on the Athenaeum page the arbiter checked is left out (arbiter.md 5f, primer.md row 5f). Both files give only Campbell's objection and "a Mr. Carse failed to repeat another, unnamed, experiment", and say the experiment is unnamed. The saved Athenaeum text (research/pages/athenaeum-1906-p1-djvu.txt, no. 4087, 24 Feb 1906, a reply defending Le Bon; author not identified in the extract) answers on those very points: it says the unnamed case "seems to be that of the quinine sulphate", which "Prof. Kalahne, of Heidelberg, has lately repeated, with the result that he finds the reaction due to chemical change, as originally pointed out by Dr. Le Bon, and not to heat, as contended by Mr. Campbell"; that Parodi (Institut Egyptien) repeated his chief phosphorescence experiments "with perfect success"; and that Rutherford's Radio-Activity mentions his "lumiere noire" and quinine-sulphate luminescence "without questioning their validity". The reply is also a party with a stake and the Kalahne report is second-hand, so it is to be weighed, not accepted; but it is pertinent evidence on the Carse point, sits in a file the arbiter says it checked, and only physicist.md (one passing mention of the quinine passage) keeps any of it. Fix: add the reply, with its stake and its second-hand status, beside Campbell's objection in arbiter.md 5f and primer.md row 5f; "unnamed" becomes "unnamed by Campbell; a reply identifies it as the quinine-sulphate luminescence". 2. Retention: Le Bon's Strutt-footnote hypothesis and its answer are not in the arbiter or primer. Lead 29 and the contrarian's section record that Le Bon's footnote hoped ordinary matter might emit slower alphas below the detection threshold (about 9,000 km/s, Rutherford), and the contrarian gives the mainstream answer (a slowed alpha is a helium atom after picking up two electrons; alpha decay is allowed only for heavy nuclei). arbiter.md 5f and Claim 2 of Q01 keep the 9,000 km/s result but not the hypothesis it was cited against, so a reader cannot see what Le Bon proposed or why it fails. Fix: one sentence in arbiter.md 5f and one in primer.md row 5f. (Safenev, named in lead 30, is also absent; a name only, no verdict depends on it.) 3. Reader profile: acronyms and notation used without definition in primer.md. First uses in the evidence table (row 3, row 4) and Silence A: Ge:Hg and HgCdTe (detector materials), Eot-Wash, Borexino, ZnS:Ag, and Pb-210, Po-210 and Ra-226 notation are not defined where they appear, although the primer defines Ci, Bq, Q, Coulomb barrier and the photoelectric effect in its added teaching passages (lines 43 and 45). N_eff is glossed in words but not as a symbol. Fix: one gloss each at first use (for example "Ge:Hg, germanium doped with mercury", "ZnS:Ag, zinc sulphide with a silver trace, a screen that flashes when an alpha hits it", "Pb-210, the lead isotope with a 22.2-year half-life").

Notes (not counted): the arbiter's and primer's "17.5 mg per year" against its own 1.76e-5 kg (17.6 mg) and example.py's 17.6 mg; arbiter "about 24 mg" of silver against example.py's 25 mg (molar mass used differs). Rounding only. Sonoluminescence, AVLIS and the giant dipole resonance are named in the reviewers and sites.md lead 35 but not in the arbiter; the arbiter's "sonochemistry; tuned-light isotope separation; bond-level resonance: Established" carries the substance, acceptable condensation. Ives (lead 40) is in the arbiter but not the primer; acceptable as it bears on no verdict.

Final: OPEN (3 issues)

Recheck (investigator update, 2026-10-08)

Checked by grep against the current arbiter.md and primer.md (both modified after the check above) and against the three defects.

Remaining: none counted. No verdict changed by these fixes, and the earlier notes (rounding of 17.5/17.6 mg and 24/25 mg; condensation of sonoluminescence, AVLIS, giant dipole resonance; Ives absent from the primer) stand as acceptable.

Final: PASS