Jeans-JH: Quality Engineering (Step 4b)

QE agent: Opus. Date: 2026-10-08. Inputs checked: _text/manifest.json, all 45 files in digest/ (read in full), the three files in deep/, facts.json, claims.md, sites.md, all six research/ notes, physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. Spot checks were made against _text/*.txt and against page images rendered from the source PDFs (read only) into /tmp/jeansqe.

Page numbers: "PDF p.N" means the === page N === marker. Gas book: printed page = PDF page minus 17. Electricity book: printed page = PDF page minus 10. Atomicity: each PDF page is a two-page spread.

Tags: [read] = from the extracted text or a page image; [web] = with URL; [memory] = from model training, with confidence; [calc] = my own arithmetic.

Verdict: MINOR. The seven central claims and their split verdicts hold up. All three documents were digested in full, and every claim has a deep read behind it. Every spot check I made against the source pages confirmed the deep reads. example.py runs and reproduces the primer's numbers. The defects can all be fixed by updating facts.json, claims.md, arbiter.md (a "QE revision" section) and primer.md. Steps 2-4 do not need to be redone. The defects are:

• (D1) The arbiter's 17 corrections and the primer's one correction have not yet been applied to claims.md, facts.json or the deep reads, so known errors are still in the source of truth.
• (D2) Three digest-only stretches carry material the reviews rely on or should weigh. The most important is Jeans's own Chapter XVIII answer to the misfit of his integer rule for polyatomic gases (Gases PDF pp.343-355). The reviewers cite that misfit against him without reporting his answer.
• (D3) The primer leaves out the Rayleigh-Jeans law, and with it the 1905 Rayleigh-Jeans exchange that sits behind claim 1. It also leaves out the contemporary evidence and dissent on collision-excited light that Jeans cites for claim 2 (Schuster-Hemsalech; Paschen against Pringsheim).
• (D4) There are small accuracy and wording issues: the 4/3 problem is described loosely, Lorentz's argument is dated 1906, hydrogen escape is computed with H2 rather than atomic H, the H2 calculation in example.py is labelled as "Jeans's recipe", the term "LEP" is not defined, and the verdict-scale wording for the "no acceleration" sub-claim needs care.

---

1. Coverage

No document was skipped, and every chunk has a digest. The _textprep.log reports 3 documents and 45 chunks, which matches. Coverage passes.

Scan notes for later readers: Gases PDF pp.155-160 hold a duplicated or out-of-order sheet (printed pp.142-143 appear twice) [read: digest .06]. This does not affect any claim, but page offsets differ by one or two around those pages.

2. Shortcuts

Deep-read coverage is good for the central claims. The deep readers read every load-bearing equation from page images and recorded their corrections. The remaining digest-only stretches were mostly acceptable, with three exceptions that bear on the claims:

• S1 (defect). Gases Ch. XVIII, "Molecular Aggregation and Dissociation" (PDF pp.343-355), is digest only. It bears on claim 1. In §§402-407 Jeans answers the non-integer n+3 values himself: he puts chlorine's 6.08 and CO2's 6.98 down to molecular aggregation (Lussana's CO2 at 85 atm gives n+3 = 9.2; wet steam gives 32, dry steam 6.6; chlorine's density drifts from 2.4807 at 20 C to 2.4502 at 200 C against Stas's 2.4501). He then proposes n+3 = 3, 5 and 6 for monatomic, diatomic and complex gases, which he says "disposes of the tremendous difficulty" of molecules keeping large internal motions without loss of energy [read: Gases text lines 16925-16955, PDF ~p.351]. §411 applies the dissipation idea once more: radiative damping of forced vibrations makes the atoms of a diatomic molecule settle into circular orbits, which gives five degrees of freedom [read: digest .12, p.353-354]. claims.md says Ch. XVIII is digest only. Yet the physicist, mathematician, arbiter and primer all present the polyatomic and halogen misfit (CO2 6.98, Cl2 6.08, Br2/I2 6.8) as evidence against Jeans, and say that "Jeans hands the problem to chemists" (true of §272 only), without reporting his Ch. XVIII answer. The answer is wrong by modern lights: the low gamma of heavy halogens and CO2 comes from partly excited vibrations, not aggregation [memory, high]. But it is his stated position and has to be weighed. A deep read is requested (DR1).
• S2 (defect, minor). Gases Ch. XVI, "Propagation of Sound" (PDF pp.319-325, §§359-366), is digest only. Jeans works out how a lag in the exchange between translational and internal energy damps sound and changes its speed. For his ideal diatomic molecule the damping from the lag is "about one-seventh as great as that caused by viscosity" [read: Gases text line 15596]. This is the measurable signature of slow energy exchange. Sound absorption and dispersion are how vibrational relaxation times such as the CO2 figure of about 7 microseconds were later measured, and the Landau-Teller paper of 1936 is titled "Zur Theorie der Schalldispersion" [memory, high on the title; the primer's further reading cites it]. Nobody in the review connects Jeans's 1904 acoustic analysis to claim 2. This is a missed link, and it can be tested. A deep read is requested (DR2).
• S3 (minor). Electricity book §§496-500 (PDF pp.448-453), on Ampère's rule and the family of element forces, are digest only. The arbiter's sub-claim 4e, "Ampère element law: Not testable as stated", and its note that this "matters to other projects in this collection" both rest on §624 and on memory. In §§498a-500 Jeans himself writes out a Helmholtz-type family of element forces with a constant k (k = 1 gives Ampère's own system). He calls a force out of the plane of r and ds "almost inconceivable" for pure action at a distance, and concludes that action-at-a-distance theories should be given up in favour of stresses in the medium [read: digest .22, pp.450-453]. Because other projects depend on the Ampère force, the exact formulas should be on record (DR3).
• Acceptable digest-only stretches: electrostatics, potential theory, harmonics, networks, magnetism and induction in the electricity book (PDF pp.1-447, 454-556), which are standard textbook material, and Gases Ch. I-VI and XI-XV. One partial exception is Gases Ch. III-IV (PDF pp.48-75). facts.json key claim 1 for the gas book ("Maxwell's distribution holds with probability 1 ... by a generalised-space argument") rests on the digest alone. It is not one of the seven central claims, and no verdict uses it. I list it as an optional deep read (DR5), which would also cover the Burbury criticism of Jeans's proof (Phil. Mag. v-vii, cited at PDF p.67) that no file mentions.

3. Missed material (from my own reading of the digests)

Items that facts.json, claims.md, the reviews, the arbiter or the primer missed or under-used, most important first:

1. The Rayleigh-Jeans law is absent from the primer (0 mentions). It is Jeans's best-known contribution to the very problem of claim 1. The lecture prints its classical form, (8 pi R T/c^3) nu^2 d nu, as the h -> 0 limit of Planck's law [read: deep read Atomicity §2d]. The 1905 Rayleigh-Jeans exchange (Nature 71, 559 and 607; 72, 54, 101 and 243) is in research/files/montambaux, and Lodge (Nature 118, 81, 1926) credits Jeans with the 8 pi R factor [web: research/2026-10-08-montambaux.md]. Gases §257 (equipartition between matter and ether as the one "failure case") is the same equilibrium that, applied to the ether's own modes, gives the ultraviolet catastrophe [read: deep read Gases §8]. The arbiter mentions both. The primer should give the reader this link in a short paragraph.
2. Jeans's Ch. XVIII aggregation answer, and the §411 dissipation argument (see S1).
3. Jeans's 1904 sound-damping prediction (see S2).
4. The contemporary evidence and dissent on collision-excited light (Gases §§248-250) are in the deep read but absent from facts.json, claims.md, the reviews (except one mention by the mathematician), the arbiter and the primer. Jeans cites Schuster and Hemsalech's arc observation (violet light from faster molecules) as support that needs no "special state". He records Pringsheim's view that "no light is ever produced in gases by collisions alone", contested by Paschen and Smithells, and his own vacuum-tube estimate that electrical action "almost" predicts emission [read: deep read Gases §7]. These are a contemporary experimental claim for, and a dissent against, the claim 2 mechanism. Under SOURCE HONESTY they belong in the record next to the verdict.
5. The British Association debate at Birmingham in 1913 (Larmor and Love defending classical dynamics, recorded in Jeans's 1914 Report §18) is an Investigator lead under priority 3 [web: research/2026-10-08-archive-org.md]. Only the contrarian mentions it, and then just "Love". It shows that the classical position Jeans had held in 1904 still had defenders in 1913. One line in the arbiter's history would carry it.
6. Jeans corrected his own molecular count by 1911. The electricity book (§150, PDF p.142) uses n = 2.75 x 10^19 per cm^3 and recalculates the gas-book radii with it [read: EM text lines 7847, 7898]. The arbiter, physicist and primer list "N = 4e19, 49% high" as a slip in the 1904 book, which is correct. But Jeans had already revised the figure to within 2% of the modern 2.69e19 [memory, high on the modern value]. This changes the story from "slip" to "the value of 1904, revised by Jeans himself by 1911".
7. The 1911 book on atomic stability (§192, PDF p.178). Jeans argues from Earnshaw's theorem that an electron cannot rest stably under inverse-square forces. He rejects Larmor's model of orbiting electron rings as "a sort of 'perpetual motion' machine" whose energy cannot be replenished [read: EM text line 10043]. This is the 1911 starting point for the 1925 non-radiating time-symmetric atom (claim 6). It shows the problem the 1925 idea was meant to solve, and it belongs in the primer's section 4c.
8. The 1911 book on action at a distance (§§153-155, PDF pp.150-152). Jeans says no a priori argument rules out action at a distance, treats the potential as "a convenient mathematical fiction", and adopts the ether stress picture only as a tentative hypothesis [read: digest .07]. This qualifies the arbiter's "real ether" summary of claim 4: in 1911 Jeans presents the ether as a working hypothesis, while the contraction is "inevitable" only on the stated condition. The arbiter's verdicts already reflect the conditional wording, but the hedging is worth one sentence.
9. Small items, optional: free electrons per atom against the specific heats of metals (EM §613, PDF p.555), an equipartition puzzle of the same family as claim 1; Jeans's own Phil. Mag. proof of Maxwell's law and Burbury's criticism of it (Gases PDF p.67); Bertrand Russell's 1927 citation of the lecture [web: research/2026-10-08-reviews-1926.md].

4. Accuracy spot checks (against the extracted text and page images)

Every load-bearing number I checked matched the deep reads. I found no error in the deep reads' arithmetic, and the example.py output agrees with them (section 6).

5. Retention check

The question is whether pertinent material from the sources or from the Investigator's findings was left out of the arbiter or primer because a reviewer judged it negatively.

• R1 (defect, minor). Jeans's own answer to the polyatomic misfit is missing (S1). The misfit is used against him in physicist claim 1b, mathematician claim 1b, arbiter claim 1b and primer section 6, while his stated counter-argument (aggregation; the n+3 = 3/5/6 rule; §411) is absent. The cause looks like digest-only coverage, not deliberate suppression. The effect is the same, though: the record shows the evidence against him and not his reply. Fix: add it beside the misfit, with the modern reading (vibrational excitation) as the reason it fails.
• R2 (minor). Contemporary support and dissent on collision-excited light (missed item 4): Schuster-Hemsalech in support, Pringsheim against, Paschen and Smithells against Pringsheim. They were dropped between the deep read and the arbiter and primer. These are the only contemporary experiments the 1904 book cites for the claim 2 mechanism.
• R3 (minor). Investigator leads not carried into the arbiter or primer: the 1913 British Association debate (Larmor, Love), and Lodge 1926 (in the arbiter, not in the primer). The Montambaux files 05, 06, 08-11 and 13 (the rest of the 1905 Rayleigh-Jeans exchange, Jeans's Proc. Roy. Soc. 1905 paper, and his 1911 Solvay report) are still unread by any role. The arbiter lists them under "Not read by any role", which is honest. They are the most direct record of how Jeans defended claim 1 in 1905, and the next Investigator run (or a reviewer update) should read them.
• Kept correctly: Stoney's 1904 dissent (kept, unanswered, flagged as an informed party), the Adams interested-party note, Zhao's mutual-energy papers (as related work, not support), the Carati-Galgani minority programme (verdict "Never properly tested", not dismissed), Tolman, Loring and Adams's reviews, the absence of Jeans from absorber-theory histories (analysed with six competing explanations), and the Wien and Taylor single-source figures (described, not labelled away). In these the arbiter and primer follow SOURCE HONESTY well.

6. Process

• Source tags. Present and honest in all files. [memory] items are given confidence levels. The arbiter says which web checks failed (hydrogen n=3 lifetimes) and keeps those items tagged [memory]. Pass.
• Web checks. Physicist 6, mathematician 5, contrarian about 7, arbiter 6 of 8. The medium- and low-confidence facts that verdicts depend on (Bothe-Geiger and Compton-Simon dates, the Bucherer dispute, the exosphere temperature, Earth's hydrogen loss, the electron radius limit, Kennedy-Thorndike, Eucken 1912, CO2 relaxation) were checked. Left as [memory]: Grangier-Roger-Aspect (the physicist's tag; the contrarian checked it on the web), Ritz combination, Dirac 1938, Ives-Stilwell and Landau-Teller (all high confidence). Pass.
• Verdict scale. Applied consistently, and split verdicts are clearly labelled. One point to tidy: claim 5e, "a radiation field produces no acceleration", gets "Refuted by calculation" with Thomson scattering added in brackets. Jeans states it as a consequence of quantum dynamics, so a classical equation of motion alone does not refute it. What refutes it is observation (Thomson scattering, the low-frequency limit of Compton scattering) and quantum electrodynamics, which also gives the quiver [memory, high]. The arbiter should either name the quantum-mechanical calculation or use "Tested, not supported". The mathematician (Tested, not supported) and the contrarian (Refuted, by observation) already lean that way. This does not change the substance.
• The primer follows STEP 4. Sections 1-9 are present in order, with answers at the end, and example.py exists. Reader profile: most symbols are defined. Undefined: "LEP" (primer line 74; define it as the Large Electron-Positron collider at CERN), "exobase" (Check yourself Q3; define it as the height above which collisions are rare and molecules on escape orbits are lost), and "eV" (electron-volt, used in Part 3; one clause is enough). Pass, with these small fixes.
• example.py runs. .venv/bin/python example.py exits 0. Parts 1-5 reproduce the primer: the RK4 run against the formula agrees within 0.02%, e^-16 = 1.13e-7, e^-800 = 10^-347.4, 1201 e^-1200 = 10^-518.07, 13 e^-12 = 8e-5, 1.32 eV and e^-56 = 4.99e-25, the H2 table, the freeze temperatures 112, 55, 30 and 18 K, x = 30.4 and 15.2 for H2, and half-lives of 0.45 yr, 7.6e5 yr and 1.26e10 yr. Pass. For a wording issue in Part 4, see section 7, item 4.
• Corrections pipeline. The arbiter lists 17 corrections and the primer 1. None has been applied to claims.md, facts.json or the deep reads yet. For example, claims.md still says "accepted quantum theory by 1913-14" and "Moon ... and Mercury lose hydrogen, helium and most other gases". facts.json still gives only the e^-800 case, "agrees exactly" for Bucherer, and the "10^-9 s" figure. The Atomicity deep read §2c still says "RT ... per mole ... used loosely" (line 49). This is defect D1. The orchestrator should apply them.

7. Corrections to earlier documents

Under CORRECTIONS ARE WELCOME, for the orchestrator to apply in addition to the arbiter's list:

1. physicist.md, claim 4a: "The field momentum gives 4/3 instead of 2/3" is wrong. The field momentum of a slowly moving charged shell gives the same (2/3) e^2/(a C^2) as Jeans's kinetic-energy route. The factor 4/3 is the ratio of that mass to U/C^2, where U = e^2/(2a) is the electrostatic energy [calc: (2/3)/(1/2) = 4/3; memory, high on the standard 4/3 problem]. arbiter.md ruling 2 ("appears only when the momentum route or E = mC^2 is added") and correction 11 ("energy and momentum routes differ") should say: "the momentum (and kinetic-energy) mass is 4/3 of the rest-energy mass U/C^2". The primer's section 3d already states it correctly.
2. contrarian.md, claim 5a: "uses RT for kT ... which is a unit slip" contradicts the mathematician's correction, which the arbiter accepted (correction 6). In Jeans's notation R is the gas constant per molecule. Apply the same correction there.
3. arbiter.md claim 5 and primer section 4c: "Lorentz's starlight argument (1906)" repeats the lecture's date as fact. Jeans's own 1924 Report gives the references as Lorentz, Phys. Zeitschr. 11, p.349 (1910) and BA Birmingham 1913 [web: research/2026-10-08-archive-org.md, §86]. Write it as: "Lorentz's argument (dated 1906 in the lecture; Jeans's 1924 Report cites Lorentz 1910)".
4. primer section 5, Step 4 and example.py Part 4 heading: "Jeans's recipe for H2" over-attributes. Jeans's collision-time rule was for vibrations with a fixed frequency p. Classical rotation has no fixed frequency, and in his own 1904 theory the two rotations of H2 are momentoids at the principal temperature (H2 is in his n+3 = 5 list, §§258, 270) [read: deep read Gases §§8, 10]. The stronger and simpler point is that Jeans's 1904 theory predicts gamma = 7/5 for H2 at all temperatures, while Eucken (1912) measured rotation freezing out below about 100 K. The omega-tau table is a useful "what if" extension and can stay, labelled as "applying Jeans's rate law to the J = 0 to 2 rotational step (an extension, not in Jeans)".
5. primer section 3c and arbiter claim 3b: "hydrogen (H2) at 1000 K has x = 15.2 ... hydrogen is at the edge of loss". Hydrogen in the exosphere is mostly atomic, and atomic H at 1000 K gives x of about 7.6, well inside Jeans's "certainly lost" range [calc: half the H2 value; the physicist's table gives 7.5]. State both, and say that the atomic value is the relevant one.
6. claims.md "Source note" and the primer source note should also list Gases Ch. XVI and XVIII and EM §§496-500 as digest only (or remove them from the list once DR1-DR3 are done).

8. Remediation actions

Deep reads (most important first)

Changes to facts.json and claims.md

1. Apply arbiter corrections 1-17 and primer correction 1 (D1), and the corrections in section 7 above.
2. Gas book entry: add Ch. XVIII (aggregation answer, §411), Ch. XVI (sound damping) and §§248-250 (Schuster-Hemsalech; Pringsheim against Paschen) to the summary and key_claims after DR1-DR2. Change how_read once the new deep reads exist.
3. Electricity book entry: add §§153-155 and §192 (DR4), §§498a-500 (DR3), and the n = 2.75e19 revision (DR6).
4. claims.md claim 1: add Jeans's aggregation answer and the 1905 Rayleigh exchange (the time-scale restatement). Claim 3: note that Jeans revised N by 1911.

Changes to arbiter.md (add a "QE revision" section)

1. Claim 1: record Jeans's Ch. XVIII aggregation answer beside the polyatomic and halogen misfit, and why it fails (vibrational excitation explains low gamma in heavy and polyatomic molecules; aggregation is negligible in dilute gases at room temperature [memory, high]). Verdicts unchanged.
2. Claim 2: add the contemporary evidence and dissent (Schuster-Hemsalech; Pringsheim against Paschen and Smithells) and, after DR2, Jeans's sound-damping prediction as the testable consequence of his rate law and its later confirmation route (acoustic relaxation). Expect no change to verdicts. If DR2 shows a definite prediction that was later borne out, the arbiter may add "Established" for the acoustic-lag consequence as classical dynamics.
3. Claim 3: N = 4e19 is the 1904 value, revised by Jeans to 2.75e19 by 1911. Atomic H is the relevant exospheric species (section 7, item 5).
4. Claim 4: fix the 4/3 wording (section 7, item 1). Note that Jeans calls the ether stress picture a tentative hypothesis (§155). Refine 4e from DR3.
5. Claim 5: date Lorentz's argument as in section 7, item 3, and tidy the verdict wording for 5e (section 6).
6. Claim 6: add the 1911 background (§192, Larmor rings as "perpetual motion") from DR4.
7. History: one line on the 1913 Birmingham debate (Larmor, Love).

Changes to primer.md

1. Add a short paragraph (section 3a or 4a) on the Rayleigh-Jeans law: the 1905 exchange with Rayleigh, Jeans's correction of the factor (8 pi, not 64 pi; Lodge 1926), the formula (8 pi k T/c^3) nu^2 d nu, and the ultraviolet catastrophe as the "5 : infinity" final state Jeans conceded in Nature 71, 607. Link it to the lecture's h -> 0 limit.
2. Section 4a: add Jeans's aggregation answer to the polyatomic misfit and why it fails, plus the Schuster-Hemsalech and Pringsheim-Paschen evidence, in one or two sentences each.
3. Section 5, Step 4: relabel the H2 table as an extension of Jeans's rate law (section 7, item 4). Add the direct point that his 1904 theory gives gamma = 7/5 for H2 at all temperatures, against Eucken's measurement. Make the matching comment change in example.py Part 4 (code unchanged).
4. Section 3c: give atomic-H escape numbers beside H2. Define "exobase" and "LEP", and "eV" on first use.
5. Section 4c: add the 1911 background to the non-radiating atom (§192), and date the Lorentz reference correctly.
6. Further reading: Landau and Teller's title is about sound dispersion. After DR2, say that Jeans's 1904 Ch. XVI is the classical precursor.

Investigator (next run, no new deep-read slot needed)

Read the saved but unread files: Montambaux 05, 06 and 08-11 (the 1905 exchange and Jeans's Proc. Roy. Soc. 1905 paper), 07 (Phil. Mag. 1905, in full), 13 (Solvay 1911), the Jeans 1903 papers (Phil. Mag. 6, 279, the source the Milan school cites), and Hudson 1989 in full. Also open Milne 1947 and Nature 117, 46, as the arbiter lists.

---

Summary for the orchestrator

MINOR. No central verdict is at risk. Apply the arbiter's 17 corrections, which are still pending, and the 6 corrections above. Run DR1-DR4 (DR5-DR6 are optional). Then add a QE revision section to arbiter.md covering Jeans's own Ch. XVIII answer, the sound-damping link, and the background material from 1911. Revise the primer to include the Rayleigh-Jeans law, a correctly attributed H2 example and the small definitional fixes.

---

Recheck

Date: 2026-10-09. Checked by grep and reading of the current claims.md, facts.json, arbiter.md, primer.md, physicist.md, contrarian.md, example.py and the three deep reads. example.py was re-run and exits 0.

Remaining issue (1): fix physicist.md line 94 to say that the field-momentum mass equals the kinetic-energy mass (2/3) e^2/(a C^2), and that 4/3 is its ratio to U/C^2. Also mark arbiter correction 11 as superseded by correction 19. No verdict is affected.

Final: OPEN (1 issues)

Orchestrator fix (2026-10-09)

• physicist.md:94 (claim 4a): old "The field momentum gives 4/3 instead of 2/3"; new: field-momentum mass equals kinetic-energy mass (2/3) e^2/(a C^2), 4/3 is its ratio to U/C^2, with correction note.
• physicist.md:14 (item 4a, same error found as a stale copy): old "masses that differ by a factor 4/3"; new "the same electromagnetic mass ... 4/3 is the ratio to U/C^2", with correction note.
• arbiter.md:215 (correction 11): appended "(superseded by correction 19: ...)".

Final: PASS