Maxwell: Quality Engineering (Step 4b)

QE agent: Opus. Date: 2026-10-03. Inputs checked: _text/manifest.json, all 10 files in digest/, facts.json, claims.md, physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py, with spot checks against _text/*.txt and two rendered page images of dynamicaltheoryo00maxw.pdf (PDF pp.33 and 52, rendered to the scratchpad, not to the source folder). Source tags as in prompt.txt: [read] document text, [read: page image p.N] rendered PDF page, [web], [memory].

Context. Step 1 ran before digests existed. facts.json says "text layer sampled" for all four documents, and no deep/ folder exists. The reviewers, arbiter and primer writer made up for much of this by rendering the key equation pages of the 1865 paper themselves. This QE pass asks whether what is still missing could change a verdict.

---

1. Coverage

• Every manifest document has a facts.json entry, and every chunk has a digest. Nothing was skipped. Coverage: no defect.
• Bookkeeping errors in facts.json: the character counts disagree with the manifest. facts.json gives Faraday's Lines as "about 149,000 characters" (manifest: 226,745), Dynamical Theory as "about 121,000" (manifest: 181,112) and Hunt as "about 36,000" (manifest: 56,445). The counts came from an earlier extraction and should be corrected.
• how_read reads "text layer sampled" for all four entries. That is not one of the values Step 1 allows, and it no longer describes what was done: all four documents were digested in full, and the 1865 paper's PDF pp.29, 31-34, 47 and 52 were read as page images during Steps 2-4.

2. Shortcuts

3. Missed material (from the digests, checked against the text where noted)

1. Maxwell's own estimate of the field strength in sunlight (1865, §108, PDF p.53). From Pouillet's solar data (83.4 ft·lb per second per ft²), Maxwell computes the peak electromotive force in sunlight at the Earth as P = 60,000,000 (his units), "about 600 Daniell's cells per metre", and about 13,000 cells per metre at the Sun's surface. He also notes the field reverses "more than a thousand million million times in a second" [read: text layer p.53]. No review file mentions it. It is an ideal check-it-yourself calculation for this reader. Converting his input: 83.4 ft·lb/s/ft² = 1217 W/m², and the modern E₀ = √(2S/(cε₀)) = 958 V/m. A Daniell cell gives about 1.1 V, so 600 cells is about 660 V/m [my calculation; Daniell EMF [memory: high]]. That is about a factor of 1/√2 below the modern figure (958/√2 = 677 V/m). The gap may be an RMS-versus-peak or units convention. It should be checked on the page image before it is used.
2. Calculated versus measured self-inductance of the British Association coil (1865, PDF pp.59-60). Maxwell computes L = 430,165 "metres" (emu) from the coil's dimensions, with corrections. A least-squares fit to 16 spinning-coil experiments gives 456,748 (6.2% higher), and a rough electric-balance measurement gives 410,000 (4.7% lower). He judges the calculated value "probably much more accurate than either of the other determinations" [read: text layer p.60]. This is the paper's only theory-versus-measurement comparison apart from the speed of light, and nobody reports it. It also shows the size of the experimental scatter in the 1860s (about 5%), which supports the arbiter's reading of the 5% light-speed agreement.
3. Maxwell credits Faraday's 1846 "Thoughts on Ray Vibrations" (1865, PDF p.14). Maxwell says that Faraday's 1846 paper proposed essentially the same electromagnetic theory of light, with transverse disturbances, but lacked the data to compute the speed [read: text layer line 416]. This bears on priority and is missing from claims 4 and 7 and from the primer.
4. Faraday rotation as Maxwell's evidence for the ether (1865, PDF p.9). Maxwell cites Faraday's magneto-optic rotation, Verdet's opposite rotation in paramagnetic bodies and Thomson's argument as physical evidence for motion in the medium wherever there is a magnetic field [digest]. The physicist mentions Verdet once, but none of the claim 5 analyses deals with the evidence Maxwell actually offered for the ether. The claim 5 verdict stands, but the "reasonable position at the time" judgment would be better supported.
5. Field theory versus Weber's action at a distance. The 1865 paper opens by rejecting Weber and Neumann's velocity-dependent action at a distance (PDF p.7). The 1856 paper (pp.54-55) describes Weber's theory as an alternative that is valuable to keep alongside the field view [digest]. Weber electrodynamics is a recurring alternative-science theme. The primer does not mention it, and claims.md has no claim covering it. Add a short note to the primer (section 4 or 7) rather than a new claim.
6. Heaviside's objection to longitudinal waves (Belrose p.8). Belrose reports that Heaviside criticised Helmholtz for allowing longitudinal as well as transverse waves [digest]. Together with Maxwell's statement that both sciences are "at a loss" about normal (longitudinal) vibrations and that such vibrations would need an electric fluid with density (1865, PDF p.49), this links directly to the scalar/longitudinal-wave theme in other projects. The primer mentions longitudinal waves only in passing (section 3.2).
7. Smaller items, none verdict-bearing: - Hunt says Heaviside's operator methods "probably influenced" Dirac [read: Hunt]. - Hunt places the Heaviside layer (the ionosphere) in his work. facts.json mentions it and the physicist marked it "unchecked". I confirmed that Hunt mentions it [read: Hunt, line 59-61]. - Maxwell's 1856 paper ends with a revolving conducting shell whose heating Foucault confirmed experimentally (p.75) [digest]. - Heaviside's government pension was £120 a year, later £220 [read: Hunt]. - Belrose credits Hertz's 1890 restatement of the equations, which used traditional notation [digest].

4. Accuracy spot checks

Arithmetic in the arbiter and primer (percentage deviations, 1/√(μ₀ε₀), the Poynting-into-wire numbers) is reproduced by example.py (section 5).

5. Process

• example.py runs cleanly with .venv/bin/python (exit 0). Part A gives +3.65 %, +5.03 %, −0.60 %, +2.74 % and a Fizeau–Foucault spread of 5.66 %. Part B gives the printed versus physical sign of p (amplitude per wavelength 1.368 vs 0.731). Part C gives P_in = I²R = 0.5348 W, a symbolic difference of 0, and a drift speed of 2.34 × 10⁻⁴ m/s. All match the primer.
• Source tags. Present throughout and generally honest, and the reviewers flag what they did not check. Defects:
• (a) The Lodge replication tag (4.5).
• (b) The physicist cites "[read: far.pdf ...]", a filename that does not exist (it should be MaxwellOnFaradaysLines.pdf).
• (c) Page conventions are mixed. The physicist cites printed pages ("p. 465") and section numbers, while the others use PDF pages. The primer uses PDF pages consistently.
• (d) The contrarian's 439,450,000/√2 derivation is [memory: medium], and the arbiter flags it correctly.
• Web checks. Physicist 4, mathematician 2, contrarian 2, arbiter 3. The medium-confidence items left as [memory] (Lecher, Kennedy–Thorndike, the 439,450 figure) are correctly declared not verdict-bearing. Michelson–Morley and Aharonov–Bohm, which verdicts do lean on, were web-checked. Acceptable.
• Verdict scale. Mostly applied consistently, and split verdicts are well used. For claim 7 the arbiter goes off the scale ("documented history", "cause never established"), and the contrarian uses "Not supported as a general claim". This is defensible because claim 7 is history rather than physics, and the arbiter says so. Still, the primer evidence table should state plainly that claim 7 carries scale labels only for its physics content. It mostly does.
• facts.json and claims.md not updated. They still carry the Step 1 wording that the reviewers corrected:
• facts.json says the 1865 wave is "electric and magnetic", but Maxwell states it as "entirely magnetic disturbances".
• claims.md claim 4 says V = v without the μ = 1 condition.
• claims.md claim 5 omits Maxwell's own statement that his conclusions are "independent of this hypothesis".
• claims.md claim 7 lumps three different episodes together.
• Neither file mentions the (F)/(G) sign defect. The arbiter's "Corrected claim list" exists but has not been carried back into these files.
• Primer against STEP 4 and the reader profile. All nine sections are present in order, with answers at the end. Symbols are defined, SI versus Gaussian units are explained, gauge freedom is introduced in plain language, and quaternions are mentioned only in passing and not used. The worked example has real numbers and a script. The primer meets the standard. Its one factual error is inherited: "Replicated? Yes (Lodge [read: Hunt])".

6. Verdict

MINOR.

None of the defects found can change a central claim or a final verdict: - The load-bearing 1865 equations and numbers were checked on page images and re-checked here. - The new material (the sunlight estimate, the coil inductance comparison, Faraday's 1846 priority, Maxwell's ether evidence, the Weber contrast, Heaviside on longitudinal waves) adds content and learning value but does not change a verdict. - The 1856 derivations are unchecked, but claim 1 is a representation claim that Maxwell himself disclaims as physics.

The defects are: - Sampling of the two central papers with no deep reads. - One mis-tagged replication claim. - facts.json and claims.md never updated after review. - Missed material that belongs in the primer.

7. Remediation actions

Deep reads (most important first). Output goes to review/Maxwell/deep/{slug}.md, tagged [read: p.N], with page images where equations are needed.

1. dynamicaltheoryo00maxw-471f79, pp.53-60 (Sonnet). Covers §108 (energy and field strength of sunlight) and Part VII's British Association coil calculation. - Transcribe Maxwell's sunlight formula P = √(8πVW) and his numbers from the page image (p.53). - State the unit system of P = 60,000,000 and "600 Daniell's cells per metre", and convert to V/m. - Compare with the modern E₀ ≈ 958 V/m computed from his own 83.4 ft·lb/s/ft², and explain any factor (√2, RMS vs peak, units). - Report the coil data and the calculated versus least-squares versus electric-balance values of L, with Maxwell's stated reason for trusting the calculation.
2. MaxwellOnFaradaysLines-5da1a2, pp.35-55 (Opus). Covers Part II, the electro-tonic state. - From page images, transcribe Laws I-VI and the defining equations of the electro-tonic functions (α₀, β₀, γ₀). - Check whether Law I is circulation = flux (B = curl A) and whether Law VI gives EMF = −∂A/∂t, including the sign. - Check whether Theorem V/VI is a Helmholtz-decomposition-type result. - Say whether claim 2's attribution of the vector-potential idea to the 1856 paper holds as stated.
3. MaxwellOnFaradaysLines-5da1a2, pp.4-34 (Sonnet). Covers Part I, the fluid analogy. - Confirm the dictionary (velocity to force, pressure to potential, resistance k to inverse permittivity or permeability). - Check the refraction law tan θ₁/tan θ₂ = k′/k and the p = k/(4πr) source result on page images. - Check whether Maxwell states that electrostatics, magnetism and steady currents obey formally identical equations, which resolves the mathematician's open note 5.
4. dynamicaltheoryo00maxw-471f79, pp.7-24 (Sonnet). Covers the introduction and Parts I-II. - List the evidence Maxwell gives for the ether (Faraday rotation, Verdet, Thomson; p.9). - Give his statement on Faraday's 1846 "Thoughts on Ray Vibrations" (p.14). - Give his reasons for rejecting Weber and Neumann's action at a distance (p.7). - Summarise the flywheel/"reduced momentum" dynamical analogy for L, M, N (pp.15-20), which a primer reader could use.
5. HeavisideOliver-Hunt-12-960a6e, pp.5-7 (Sonnet). - Quote verbatim, column-separated, the passages on Preece blocking the 1887 paper, Biggs's removal, the Electrician cancellation, the Bath 1888 meeting (Lodge's wire waves versus Hertz), the 1893 Royal Society rejection, and the pension. - State exactly what Hunt says about Lodge's experiments, and whether anything in Hunt supports the claim that Lodge replicated Hertz.
6. Belrose-MUL-2014-Vol2-Mar-Apr-005-cdf8af, pp.5-8 (Sonnet). - Was Belrose's NEC-4 model compared with measured data, and whose? - What exactly does Belrose say about Heaviside criticising Helmholtz over longitudinal waves, and about Hertz's 1890 formulation?

Changes to each later file: - facts.json: - Correct how_read for all four entries, for example "digest (all pages) + page images pp.29, 31-34, 47, 52 (Steps 2-4) + deep read pp.…". - Correct the character counts from the manifest. - Dynamical Theory summary: say the wave is stated as "entirely magnetic disturbances"; add V = v for μ = 1; add the (F)/(G) sign defect, the sunlight estimate, the coil L comparison and the Faraday 1846 credit. - Faraday's Lines: add the results from deep reads 2 and 3. - Hunt: add the pension amounts, the Dirac remark and the Heaviside layer (now confirmed). - Belrose: add the Helmholtz longitudinal-wave remark. - claims.md: Replace the claims with the arbiter's "Corrected claim list" (μ = 1 condition; ether disclaimed in §73-75; (F)/(G) signs; three separate Heaviside episodes). Optionally add one sentence to claim 4 on Faraday's 1846 priority. - arbiter.md: Add a "QE revision" section that: - (a) Corrects the Lodge tag: Hunt documents Lodge's own wire-wave experiments, not a replication of Hertz. Replication of Hertz becomes [memory] or [web] if a source is found. - (b) Records the British Association coil comparison (about 5-6% scatter between calculation and measurement) as context for claim 4's 5% agreement. - (c) Notes Maxwell's stated ether evidence (Faraday rotation) under claim 5 without changing the verdict. - (d) States the outcome of deep read 2 for claim 2. Change the verdict only if Law VI is not the rate-of-change form. No verdict changes are expected. - primer.md: - Fix the claim 4 row in the evidence table ("Replicated?" column) and any other "Lodge replicated" wording. - Add a short second worked example or an exercise: Maxwell's 1865 sunlight field strength, set against the modern 958 V/m, with the unit conversion explained. Extend example.py to match. - Add a sentence in section 4 on Faraday's 1846 priority. - Add a sentence in section 4.3 on the Faraday-rotation evidence for the ether. - Add a short note in section 4 or 7 on the field theory versus Weber's action-at-a-distance choice, and on Maxwell's and Heaviside's treatment of longitudinal waves, flagged as relevant to the scalar-wave theme elsewhere in the collection. - Use the British Association coil scatter as the 1860s error yardstick in section 7. - physicist.md: Correct "far.pdf" to MaxwellOnFaradaysLines.pdf. Optional, since reviewer files are not normally revised under MINOR. Record the correction in the arbiter's QE revision instead. - example.py: Add a Part D for the sunlight field strength if the primer adds it.

Recheck

Recheck agent: Sonnet. Date: 2026-10-03. Checked: deep/ (4 files), facts.json, claims.md, arbiter.md ("QE revision"), primer.md, example.py (runs, exit 0), physicist.md.

Defects (QE sections 1-5)

Remediation actions

Open issues: (1) deep read 1 not run, so the sunlight and coil figures are unverified on page images; (2) primer.md Lodge replication cell; (3) primer.md missing the added material listed above; (4) example.py Part D absent.

Final: OPEN (4 issues)

Recheck 2

QE Recheck 2, Sonnet, 2026-10-03. Checked: deep/ (the Dynamical Theory file now has a "QE round 2 deep read: pp.53-60" section), facts.json, claims.md, arbiter.md, primer.md, and example.py, which runs with exit 0.

The material now in primer.md is: - Faraday's 1846 priority. - Faraday-rotation evidence for the ether. - Field theory versus Weber's action at a distance. - Longitudinal waves, including Maxwell's remarks and Belrose's Heaviside–Helmholtz remark. - The coil scatter of 5-6 % as an 1860s error yardstick. - A sunlight worked example in section 5.2.

Final: PASS