Wheeler-Ken: Quality Engineering (Step 4b)

Inputs checked: _text/manifest.json, all 10 digests, deep/magnetism1small-6bca80.md (five deep reads), facts.json, claims.md, sites.md, research/ (notes and the 2nd-edition PDF in research/files/), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. Source text checked against _text/magnetism1small-6bca80.txt. Tags: [read] project files; [web] URL; [memory] with confidence; [calc] computed here.

Verdict: MINOR. None of the central verdicts is likely to change. But four demonstration and theory sections of the book (pp.54-96, 128-148 except the RGB pages, 157-167, 177-206) were never read beyond the digests. Several reported observations in them went unweighed. One units criticism in the deep read and the mathematician's review is wrong. One verdict label is applied loosely. The arbiter's corrections have not yet reached facts.json and claims.md. All of this is fixable by updating facts, claims, the arbiter (a QE revision section) and the primer.

---

1. Coverage

• The manifest lists one document, magnetism1small.pdf (236 pp., 6 OCR pages, 10 chunks). All 10 chunks have digests (pp.1-236, no gaps). Nothing was skipped. Pass.
• facts.json has entries for the book and two embedded items (the Coomaraswamy "Kha" text and the twisted-light press clippings). That is acceptable.
• Investigator downloads in research/files/ (2nd edition, Dollard PDFs, Tufaile papers, kathodos files) are not in the manifest, which is correct: they are Investigator finds, not project documents. They were used through the Investigator's notes. The mathematician also read the Tufaile 2024 paper's first pages directly.

2. Shortcuts

There is also a stray reference in deep/ (line 653: "The earlier section of this file (pp.97-111)"). It points to a deep-read section that does not exist in the file. claims.md already notes this. Either a deep read was lost or the line is wrong; the deep-read file should say which.

3. Missed material (from the digests and source text, not in facts.json, claims.md, reviews, arbiter or primer unless stated)

1. CRT demonstration, pp.157-161 [read: digest; source text p.157]. Wheeler held 2 in x 1 in and 1 in cube neodymium magnets against a large CRT showing a camcorder feed of a spiral test pattern. He reads the image twisting opposite ways at the two poles as "proving that the centrifugal magnetism on this side of the magnet is CCW" and the other pole inverse (p.158). He also reads "four corner loops" as against the "four poles" reading of others (p.161). He calls the CRT a "cathode ray dielectric projector" (p.157) while denying electrons. The standard account is that an axial field rotates an electron-beam image, and the sense of rotation reverses with field direction (magnetic-lens or Larmor image rotation; transverse fields give v x B deflection) [memory: high]. The book shows a real, reproducible effect that standard physics predicts, with the same pole-dependent sense Wheeler reports. It needs to be in the record and weighed under claims 2 and 6. No review mentions it.
2. Gauss-meter statement and "quadrapole", pp.145-146 [read: source text p.145]: "As any gauss meter will register to you, the highest pressure points on a magnet are its divergent centrifugal edge, and secondly its convergent centripetal point at the center of every polarized side." This is a qualitative instrument claim, so the primer's "No page of the book reports a measurement" needs softening to "no quantitative measurement (no number, instrument model, units or repeats)". Close to the face of a short magnet, standard magnetostatics also puts the largest |B| near the edge. In example.py at h = 0.1 R, |B| at r = R is 0.405 B_r against 0.402 B_r on the axis, and nearer the face the edge peak grows [calc from example.py output]. So this is a point where Wheeler's observation agrees with the textbook; the "quadrapole" reading is his. It should be kept, fairly.
3. Gyroscopes and gravity, pp.60, 181, 190 [read: source text]. Wheeler quotes Eric Laithwaite ("how a 50 pound spinning wheel loses weight"; Laithwaite's "utter hostility" reception, retired from Imperial College 1981) and states that gravitational polarization is "the principle behind gyroscopic weight loss" (p.190). The p.60 "dead gyroscope spins up" passage is explicitly a thought experiment ("in the genuine model of magnetic induction ... would cause"), not a reported test. Related experimental record, never mentioned: Hayasaka and Takeuchi (Phys. Rev. Lett. 1989) reported weight loss of right-spinning gyroscopes; controlled repeats in 1990 (Faller et al.; Nitschke and Wilmarth; Quinn and Picard) found none [memory: medium; must be web-checked before use]. This is exactly the "anomaly, then tests" material the contrarian should weigh.
4. Breaking magnets, pp.78-79 [read: source text p.78]: hard-drive magnets "split and self-fold along their dielectric planes", "into four pieces like a pizza", and "on several occasions, in breaking the more powerful hard drive magnets 3 to 5" electrical charges are shot out like lightning at the instant of breaking". This is a first-hand observation, single source, without numbers. Mundane candidates: brittle fracture along the magnetisation geometry; sparks from freshly exposed, pyrophoric NdFeB fragments [memory: medium]. It is not in any file.
5. Withheld experiments, pp.61, 79 [read: source text p.61]: "I have already written 20 bits of data to a magnetized subject and reproduced the results, by means I will not explain here". Also a 7/15/2014 discovery "using a testing medium as yet undisclosed" (p.79). The arbiter's 6f covers the 100+ TB claim but not these stated-but-withheld tests. They belong in the record as Wheeler's statements (verdict: Not testable as stated; methods withheld).
6. Other demonstrations in pp.54-96, 136-141, 191 [digest]: copper square waved against the "dielectric plane" meets resistance horizontally but not vertically (p.76; an eddy-current, Lenz-law demonstration); a thin magnet rests on edge parallel to a large magnet's plane where a steel disk cannot (pp.76-77); a few-gram disk magnet shakes a brick magnet on a 500-pound refrigerator (p.76); magnetic putty forms a sphere round a cube magnet (p.63); a ring of small magnets reduces the pull on iron (p.91); pyrolytic graphite thrown CW off one pole and CCW off the other, and turning opposite to the drag direction in water (pp.68, 72, 90, 140-141); copper ring and superconducting ring dropped on a magnet, frames 1-6 (p.191). None of these is in facts.json key_claims or the reviews. Most have standard explanations (eddy currents, induced moments, diamagnetism), and they are the book's experimental core. Also: pyrolytic graphite is more diamagnetic than bismuth (about -4e-4 against -1.7e-4 [memory: medium]), which bears on Wheeler's statement that bismuth is "the only room temperature extremely diamagnetic element" (p.99).
7. Units of "Psi x Phi = Q" (pp.18, 54, 64, 211) [read: digest; calc]. Wheeler's "electrification Q" is the product of dielectric flux Psi (coulomb) and magnetic flux Phi (weber). C x Wb = (A s)(V s) = J s, the units of action (Planck's constant), not of charge or energy [calc]. The mathematician checked the p.35 unit table but not this product. It is a good units exercise for the primer.
8. Dollard's oil-cube demonstration, p.18 [read: digest]: a cube of "10-C oil" charged between plates and carried to an uncharged pair, which becomes charged. It is quoted from Dollard, so it is his claim. facts.json mentions "10-C" only as a [memory: low] source note; no reviewer weighs it. It is mundane (a polarised or charged dielectric carried bodily, as in Faraday's ice-pail and electret work [memory: medium]), but it is the one dielectric experiment in the book.
9. Smaller factual statements, never checked: atomic radii "bismuth 143, titanium 176, polonium 190, lead 154, iron 156" with no units (p.97); "most lead is depleted uranium" and bismuth "does not exist natively" (p.102; the reviewers do address "native"); Alaska oil pipeline magnetised by "dielectrically charged oil" (p.75; geomagnetically induced currents in Alaskan pipelines are a real, studied effect [memory: medium]); health effects of wireless charging (p.19); Stephen Crothers named as having "completely refuted" black holes (p.202, a related alternative-science figure); SN 1987A (p.206); lightning branching "in perfect golden ratio patterns" (p.192). They are peripheral, but facts.json should list them in one line so the record is complete.
10. Ferrocell patent US8246356 (p.4, p.163) [read]: the book cites two patents. The Investigator found US20070241745A1 only. US8246356 (the granted patent?) has not been checked. It is an interested party's record (Vanderelli) and bears on what the Ferrocell maker claims the device shows.

4. Accuracy spot checks (against _text/magnetism1small-6bca80.txt)

1. p.46: "This angle is 42.492 (42.5~) degrees, which is also the frequency in Mhz/T of the proton, however this is incorrectly published as 42.5781" (text line 1677-1678). Correct as quoted in the primer and arbiter.
2. p.99 caption: "Single disk levitation of 3mm N45 cube neodymium with 1" cube N50 5 inches above ... identical setup using standard bismuth 'non-programmed' disk with same lifter magnet ... applied charge and field in casting" (lines 3537-3539). Correct. It supports the arbiter's correction that the lifter, not the bismuth, sets the height; claims.md and facts.json still carry the old wording.
3. p.103: 31 g and 62 g, "128 alpha-particle events over 5 days", "3.14 MeV", "(1.9 +/- 0.2) x 1019 years" [sic, superscript lost], "Bismuth also exhibits overunity ... breaks the rule of conservation" (lines 3640-3650). Correct.
4. p.217: "Replace E with energy of the dielectric, h Planck and f frequency remain the same" (line 6242-6243); p.219: "power increases on two levels exponentially as wavelengths decrease" (line 6323). Correct.
5. p.30: "D.V. = M1 x M2/ Distance^2" (line 1238). Correct.
6. p.133: Vanderelli "has confirmed this phenomenon has never been seen before" (line 4294); p.135: "Exposure times of the camera must also be considered" (line 4332). Correct.
7. Investigator's 2nd-edition finding (research/files/kathodos-magnetism-20140731.pdf, text layer extracted here): 110 pages, "2 nd Edition July 1 2014", PDF CreationDate 2014-06-28 (pdfinfo). There are no matches for RGB, ferrocell, Vanderelli or 0.381966; 42.49 appears 5 times, 137.5 6 times and Bloch 18 times. Confirmed, with one correction: there is one passing bismuth sentence ("Bismuth for example is so extremely diamagnetic due to its over-abundance of excess neutrons"), so "no bismuth matches" should read "no bismuth chapter or levitation claim". The PDF creation date (28 June) is ten days after the RGB LED "discovery" date (18 June 2014, p.133), which the 2nd edition does not mention. This adds to the arbiter's date puzzle without settling it.
8. archive.org items (research/pages/ia-.json): all three files have MD5 420583e2a0792e782432da1a774de702, identical to the project PDF (checked here). The original item was created 2014-06-25, but its current file's mtime is 2014-09-19* (the Kindle date), so the June item first held an earlier file, probably the 2nd edition. sites.md's "item created 2014-06-25" is true but could mislead; add the file date. A lead: archive.org item history may keep the June file.
9. Digest p.2 says "Dedication to the late D. Anderson, T. Martin, and T. Vanderelli". The source says only D. Anderson is "the late"; Martin and Vanderelli receive "consideration" (lines 105-107). No later file repeats the error.

All load-bearing numbers in the arbiter and primer that I checked are right.

5. Process

• Source tags. These are present and honest throughout. Reviewers mark text-layer reads as such and say which pages they did not read. The arbiter marks what it did not refetch (NIST) or did not check (Gran Sasso, Krafft).
• Web checks. These were done for the NIST proton value, Bi-209 (physicsworld), the sodium work function, Gardi (wiki) and the Tufaile papers. Two [memory: medium] items bear on verdicts without a web check: (a) the electrolysis bubble swirl as a Lorentz-force (MHD, magnetohydrodynamic) effect, used for the 6e verdict; (b) the shielded-proton value 42.5764 MHz/T in arbiter correction 6. The bismuth anisotropy figures are tagged "via search summary, primary table not read". Each needs one web check.
• Verdict scale. Mostly consistent. One loose label: all three reviewers and the arbiter give "42.4923 is the proton's true gyromagnetic ratio" the verdict Refuted by calculation. The scale reserves that label for claims that accepted laws rule out without experiment. Here the refutation is a measured constant (NMR at parts per 10^10), which is Tested, not supported (tested to extreme precision and replicated). The arithmetic of the 0.2 % gap is a calculation, but the ruling-out is experimental. The arbiter should either relabel it or state why it keeps "Refuted by calculation". The substance does not change.
• Correction to deep/ and mathematician.md (units on p.35). The deep read (line 772) and the mathematician (line 46) say the field names on p.35 are "swapped": "Dielectric field = Phi/(L x T) (Weber per cm-second)" and "Magnetic field = Psi/(L x T) (Coulombs per cm-second)". They are not swapped. Wb/(m s) = V s/(m s) = V/m, the unit of electric field strength, as Faraday's law gives (EMF = dPhi/dt per unit length). C/(m s) = A/m, the unit of magnetic field strength H, as Ampere's law gives [calc]. This is the Steinmetz/Dollard dual convention, and the book's own next lines (E = Phi/time in volts, I = Psi/time in amps) use it consistently. The mathematician's separate point stands: the "dielectric : magnetic" ratio has units of C/Wb = siemens, so 3.23606 needs a stated impedance. No arbiter or primer verdict uses the "swapped" point, so no verdict changes. Remove it from the record.
• Arbiter corrections not yet applied. claims.md still says the bismuth disk "levitates a magnet unusually high". facts.json still says "levitates a 3 mm neodymium cube 5 inches above it" and that pp.97-111, 133-176 and 207-219 are "digest only". The orchestrator should apply arbiter corrections 1-8.
• Primer vs STEP 4 and the reader profile. All nine sections are present in order, with answers at the end. Symbols are defined. Elliptic integrals are introduced in plain words. Acronyms (NMR, MRI) are defined; "CODATA" (Committee on Data for Science and Technology) in Further reading is not, which is a trivial fix. The worked example is excellent: it is checked three ways, including the off-axis brute-force check that caught a bug. Textbook chapter numbers (Griffiths 4, 5, 6, 8) are correct.
• example.py runs: .venv/bin/python example.py exits 0. Every number in primer section 5 (zero-ring tables, 581 mT, the Part C radii, Part D heights 0.0863 R and 0.0434 R, and the proton offsets of 128 and 256 kHz) matches the output.

6. Retention check

The question here is whether pertinent material from the sources or the Investigator was left out of the arbiter or primer because a reviewer judged it negatively. I found no case of that. Negative-verdict material is kept with care: Gardi's test, the Ufopolitics thread, Tinmanpower, Kris Lang's photograph (described, not labelled), VIDEO 73, bismuth anisotropy, the Tufaile horocycles, the 2nd-edition chronology and the 4th-edition absence analysis are all in both files.

Omissions that are not judgement-driven but should still be restored (Investigator leads in sites.md that reached the reviewers and not the arbiter or primer): - George Adams, "Physical and Ethereal Spaces", hosted on Wheeler's site in 2015: the likely origin of the term "counterspace" (projective geometry). The physicist and mathematician cite it; the arbiter and primer do not. A learner needs it, because "counterspace" is Wheeler's central term. - "The Ether and Its Vortices" (attributed to Krafft [memory: medium]), hosted by Wheeler in 2014: a vortex-atom forerunner. The arbiter mentions it only as unverified; the primer omits it. - Darkstar Publications' 2005 golden-ratio work on Plato, cited in a University of South Carolina thesis. It bears on where the phi numbers came from (claim 3), a point the primer stresses ("numbers built before data"). - Wheeler's background (Gallowglass Books 2025 interview: no physics training stated) and emediapress (proceeds offered to Dollard's EPD Laboratories; the book released free). This is biographical and stake context. The primer's "independent writer" is thin. - Substack 2025: "The plane of inertia is the dielectric 45 degrees from magnetic flux and 90 degrees from gyromagnetic precession". This is a later restatement of claims 2-3 that differs from the book. All three reviewers have it; the arbiter and primer do not (the arbiter does keep the parallel 2025 "attraction = gravity" shift). - Magnetic Egg video (probably the p.234 "What is this?" explanation): in the arbiter but not in the primer's list of unviewed records. - Vanderelli's "Electromagnetic lensing using a Ferrofluid thinfilm" report (research/files): no review file mentions it. It is the maker's own technical account, an interested party's characterisation, and should be listed next to the Tufaile work.

7. Remediation

Deep reads (all on slug magnetism1small-6bca80; most important first)

1. pp.54-96, Opus, with page images for figure pages. List every observation, demonstration and number Wheeler reports in the core magnet chapter: copper square p.76, edge-resting magnet vs steel disk pp.76-77, refrigerator p.76, putty sphere p.63, 20-bit data writing p.61, broken-magnet self-folding and the 3-5 inch discharges pp.78-79, the 7/15/2014 undisclosed test p.79, flipped ring pp.80-81, pyrolytic graphite pp.68, 72, 90, magnet ring p.91, "higher gauss = shallower field" pp.66-68. For each, give the apparatus, any measurement, repeats and controls, and quote any formula (Psi x Phi = Q p.64; "one half the product of dielectric and magnetic density" p.76). Does anything here make a prediction that differs from standard magnetostatics, or change the claim 1/2/6 verdicts?
2. pp.157-176, Sonnet, with page images. CRT section: what was on the screen, magnet orientation and distance, which way the image rotates at each pole, and what "four loops" are. Does the reported pole-dependent twist match standard electron-beam image rotation in an axial field? Ferrocell gallery: for each photo, whose it is (Wheeler, Snyder, Vanderelli), and any scale, cell gap, lighting geometry or exposure stated.
3. pp.128-148, Sonnet, with page images. RGB LED photos pp.133-136 (LED layout, exposure per colour, any polariser, Kris Lang's photo and caption); resting-position photos pp.136-142 (is any angle measured, what "phi^-2 contact" refers to); the gauss-meter statement p.145 and "quadrapole" figures pp.146-147. Does the book anywhere give a reading?
4. pp.97-111, Sonnet, with page images. The p.99 photo pair: are the disk-to-cube gap, the lifter position and the disk size visible or stated? Are the two set-ups really identical? The convex and concave castings pp.100, 105 (are the claimed 137.5077 and 85 degree angles measured?); units of the p.97 atomic radii.
5. pp.177-206, Sonnet. Gravity section (Laithwaite quote p.181, "gyroscopic weight loss" p.190: is any test of Wheeler's own described?); copper and superconducting ring drop frames p.191 (what was filmed, what ring); formula table p.183; atom model pp.193-200 (stated radii and their sources, "magneto-isotopes", beta decay); galactic jets pp.201-207. List each testable statement with its page.
6. pp.207-219, Sonnet, with page images (the Step 1b order that was never run). Render the equations on pp.210 and 212 and any formula figures. Do they state any prediction for refraction, dispersion or the photoelectric effect that differs from the standard result?

facts.json

• Apply arbiter corrections 3-5. Update how_read (pp.97-111, 133-142, 168-176, 207-219 read from the text layer in Step 2; list the new deep reads once done). Bismuth wording: the lifter sets the height and the bismuth stabilises it.
• Add key_claims: the CRT pole-dependent twist (pp.157-161); the gauss-meter edge/centre statement and "quadrapole" (pp.145-146); gyroscopic weight loss via "gravitational polarization" (p.190); the broken-magnet discharges (p.78); the withheld 20-bit and 7/15/2014 tests (pp.61, 79); Psi x Phi = Q (p.64).
• Add a one-line list of the peripheral factual statements in section 3, item 9, and US8246356 (p.4).
• Correct the 2nd-edition note: one passing bismuth sentence exists; PDF created 2014-06-28.

claims.md

• Coverage note: update to the reads actually done; drop "digest only" for the text-layer ranges.
• Claim 6: replace "levitates a magnet unusually high" with the lifter wording, and split into 6a-6f as the arbiter did, adding the CRT and gyroscope items.
• p.46: the "published" value Wheeler writes is 42.5781 (arbiter correction 6).

arbiter.md: add a "QE revision" section

• 6g (new), CRT demonstration: the observation (pole-dependent image twist) is real and expected from standard electron optics. As support for a vortex: Never properly tested, with an ordinary account available. Note the irony that the device works by the electron beam Wheeler denies, as the reviewers' own fact, not a label.
• 6h (new), gyroscopic weight loss: web-check Hayasaka and Takeuchi (1989) and the 1990 null replications. Record the anomaly and its tests in full (contrarian standard). Wheeler's own version is Not testable as stated (no test described).
• Broken-magnet discharges and withheld tests: Never properly tested for the discharges (first-hand, single source; mundane candidates stated); Not testable as stated for the withheld 20-bit and 7/15/2014 tests.
• Gauss-meter statement (p.145): say plainly that the qualitative observation agrees with standard magnetostatics, which is a point in Wheeler's favour as an observer, and that the "quadrapole" reading adds nothing measurable.
• Proton verdict label: relabel "Refuted by calculation" to "Tested, not supported", or justify keeping it.
• Corrections: withdraw the "swapped field names" point (deep/ line 772, mathematician line 46). Add the Psi x Phi = J s units point.
• Web checks: electrolysis bubble swirl mechanism (6e) and the shielded-proton value (correction 6).
• Related work and history: add George Adams's counterspace, the "Ether and Its Vortices" scan, the Darkstar 2005 golden-ratio work, the Substack "45 degrees" restatement, Vanderelli's lensing report, the Gallowglass background and the emediapress EPD note. Add the date evidence: 2nd-edition PDF created 2014-06-28 without the RGB LED (dated 18 June); archive.org item created 2014-06-25, current file 2014-09-19, so the item history may hold the June file.

primer.md

• Opening and section 6: change "No page of the book reports a measurement" to "no page reports a quantitative measurement". Mention the p.145 gauss-meter statement and that it agrees with the textbook near the face.
• Section 4 or "Kept in the record": add the CRT demonstration with the standard image-rotation account; add gyroscopic weight loss and its experimental history once web-checked; add the broken-magnet discharges and the withheld tests as Wheeler's statements.
• Section 1 or 9: name George Adams's projective-geometry "counterspace" as the source of the term, and the Krafft vortex text. Add a sentence of background (no physics training stated, Gallowglass 2025) and the Darkstar 2005 golden-ratio work.
• Optional check-yourself item: units of Psi x Phi (C x Wb = J s).
• Add the Substack 2025 "45 degrees" restatement next to the "attraction = gravity" shift. Add the Magnetic Egg video to the unviewed records. Define CODATA.
• Evidence table: new rows 6g and 6h after the arbiter's QE revision. Apply the proton label change if the arbiter makes it.

sites.md (orchestrator, or the next Investigator run)

• Add the file-date detail for the archive.org item. Add US8246356 and the archive.org item history to recheck. Add "Hayasaka Takeuchi gyroscope weight" as a related-work query.

Corrections to earlier documents

1. deep/magnetism1small-6bca80.md, pp.1-42 section, line 772, and mathematician.md line 46: "the field names are assigned opposite" / "swapped" is wrong. Phi/(L x T) in Wb/(m s) is V/m (electric field strength), and Psi/(L x T) in C/(m s) is A/m (magnetic field strength), consistent with E = Phi/time and I = Psi/time on the same page [calc]. The siemens point about the ratio stands.
2. primer.md opening and arbiter.md "Coverage limits": "No page of the book reports a measurement" overstates. p.145 states a qualitative gauss-meter observation (no numbers) [read: source text line 4544].
3. sites.md, Leads 1 (2nd edition): "its text layer has no bismuth ... matches". There is one passing sentence on bismuth diamagnetism (no chapter, no levitation claim). Also add the PDF CreationDate 2014-06-28 [read: research/files/kathodos-magnetism-20140731.pdf, pdfinfo and text layer].
4. sites.md, Internet Archive row: the item was created 2014-06-25, but its current file dates from 2014-09-19 (mtime 1411085259) [read: research/pages/ia-magnetism1small.json].
5. digest/magnetism1small-6bca80.01.md p.2: only D. Anderson is "the late"; Martin and Vanderelli are thanked, not memorialised [read: source text lines 105-107].
6. Verdict label for "42.4923 is the true proton value" in physicist.md, mathematician.md, contrarian.md and arbiter.md: by the STEP 2 scale this is a measured conflict (Tested, not supported), not Refuted by calculation. This is for the arbiter to decide in its QE revision.

Recheck

Checked on 2026-10-08 against the current files. [read] project files, by grep and spot reads.

Fixed

• Missed material (section 3). facts.json now has key_claims for the CRT twist, the gauss-meter and quadrapole statement, gyroscopic weight loss, the broken-magnet discharges, the withheld 20-bit and 7/15/2014 tests, and Psi x Phi in joule-seconds. It also carries the peripheral items and US8246356 (also in claims.md). Fixed.
• Deep reads (section 7). pp.54-96, 97-111, 128-148, 157-176, 177-206 and 207-219 have been deep-read. This is recorded in facts.json how_read, claims.md and the arbiter. The never-run Step 1b read of pp.207-219 is done. Fixed.
• facts.json. Bismuth wording now says the lifter sets the height. Fixed. The 2nd-edition note now says "one passing sentence" and gives the 2014-06-28 creation date in primer.md. Fixed.
• claims.md. Coverage note updated, claim 6 split into 6a-6i with the lifter wording, and the p.46 value is given as 42.5781. Fixed.
• arbiter.md. A "QE revision" section exists. It has 6g (CRT) and 6h (gyroscope, Hayasaka/Takeuchi), the broken-magnet and withheld-test verdicts, and the gauss-meter point. The proton verdict is relabelled Tested, not supported. The "swapped field names" point is withdrawn (item 2), with the Psi x Phi = J s units point added. The file-date evidence is added. Fixed.
• primer.md. "Quantitative measurement" wording is applied. The CRT, gyroscope, Adams counterspace, Krafft text, Darkstar 2005, Gallowglass background, emediapress, Substack "45 degrees" restatement, Magnetic Egg and Vanderelli lensing report are all present. CODATA is defined. Rows 6g and 6h are added. Fixed.
• Retention. The earlier retention omissions (section 6) are restored in the arbiter and primer. Fixed.

Remaining

1. sites.md not updated. Line 65 still says the 2nd edition's text layer has "no bismuth" matches, and does not give the PDF CreationDate 2014-06-28. The Internet Archive row (line 11) and line 87 still do not give the current file date 2014-09-19. US8246356 and the item-history recheck are not added. "Hayasaka Takeuchi gyroscope weight" is not added to the related-work queries. Line 67 still calls VIDEO 73 "digest only so far".
2. digest/magnetism1small-6bca80.01.md p.2 still reads "Dedication to the late D. Anderson, T. Martin, and T. Vanderelli". Only D. Anderson is "the late".
3. deep/magnetism1small-6bca80.md line 653 still says "The earlier section of this file (pp.97-111)". The pp.97-111 section is now appended later in the file, so the pointer is still wrong.

The "swapped" wording remains at deep/ line 772 and mathematician.md line 46. The arbiter withdraws it explicitly, and no verdict relies on it, so I count it as handled. A pointer to the arbiter's withdrawal there would still help.

None of these affects a verdict.

Final: OPEN (3 issues)

Orchestrator fix (2026-10-09)

Changes made (each carries an inline "[corrected 2026-10-09 ...]" or "(added 2026-10-09 after QE)" note in the file): 1. sites.md - Leads 1, 2nd edition: "no bismuth matches" replaced by "no bismuth chapter or levitation claim; one passing bismuth sentence" (quoted). PDF CreationDate 2014-06-28 added. - Internet Archive row (table) and Leads 4 (publishing line): current file date 2014-09-19 added next to "item created 2014-06-25". - The same stale "lacks the bismuth ... material" wording in sites.md's own "Corrections to earlier documents" is also corrected. - New table rows (Recheck? yes): Google Patents US8246356 and the archive.org item history for magnetism1small. New leads: US8246356 (related work, with Vanderelli's stake) and the item history (historical record). - "Hayasaka Takeuchi gyroscope weight" added to the queries for the next run, with a related-work lead for Hayasaka-Takeuchi 1989 and Faller 1990. - VIDEO 73 lead: "digest only so far" replaced. The book's pp.97-111 are now deep-read (deep/ section "pp.97-111 (bismuth chapter)"); the video itself is still unviewed (arbiter.md line 220 agrees). 2. digest/magnetism1small-6bca80.01.md p.2: the dedication now applies "the late" to D. Anderson only; Martin and Vanderelli receive "consideration" (source text lines 105-107). 3. deep/magnetism1small-6bca80.md, pp.43-53 corrections (line 653): the pointer now names the pp.97-111 section by its heading and says it sits later in the file, after the pp.157-176 section. 4. deep/ p.35 units bullet (line 772) and mathematician.md line 46: an inline note withdraws the "opposite/swapped" reading and points to arbiter.md QE revision, Verdict changes item 2 (Wb/(m s) = V/m, C/(m s) = A/m). The siemens point is kept.

No verdict changed. Stale-copy search: facts.json, claims.md, primer.md and arbiter.md already carried the corrected dedication, bismuth and date wording. The deep pp.177-206 section (line ~1479) speaks of "swapped roles" in naming only, not of a unit error. That section agrees with the withdrawal, so it is left as is.

Check (QE standard)

• Recheck item 1 (sites.md): every sub-point is present. I verified each by grep against qe.md section 4 items 7-8 and arbiter correction 7: CreationDate 2014-06-28, mtime 2014-09-19, one passing bismuth sentence, the US8246356 row and lead, the item-history row and lead, the Hayasaka query, and the VIDEO 73 status. Fixed.
• Recheck item 2 (digest dedication): the line matches the source text. Fixed.
• Recheck item 3 (deep pointer): the named heading exists and does come after the pp.157-176 section. Fixed.
• Swapped-wording pointer: present in both files. Fixed.
• The new rows use "not reached (added 2026-10-09 after QE)" with no queries listed. No site is recorded as negative without queries. SOURCE HONESTY holds: Vanderelli's stake is marked.
• Site not re-rendered (render_site.py not run, as instructed).

Final: PASS