Thompson-JJ (J. J. Thomson): Quality Engineering (STEP 4b)

QE, 2026-10-09. Inputs read: _text/manifest.json; all 79 digest files (every chunk of all 11 documents, read in full); facts.json, claims.md, sites.md, the research/ notes listing, physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py (run). I did not re-read the eight deep/ files in full. Instead I checked the claims that depend on them against the extracted text and page images (see "Accuracy spot checks"). Page images were rendered to /tmp/qe_jj (scratch, not a deliverable). Tags: [read] = project text or page image (PDF page numbers unless "printed p." is given), [calc] = my arithmetic this session, [memory] with confidence, [web] with URL. I did no web searches.

Verdict: MINOR

The pipeline is complete: every chunk was digested and every document has a facts entry. The six central claims and their final verdicts hold. Remediation is still needed, for these reasons: - One worked example and one table interpretation rest on readings that are incomplete or unbalanced: the Barkla datum and the Kaufmann column labels. - Part of claim 1 (Thomson's own e/m history) is stated wrongly in the primer. - Part IV of the vortex essay (part of claim 6) was left "not assessed", although it contains testable predictions. - Several pertinent results in the digest-only chapters never reached facts.json, the arbiter or the primer.

None of these needs Steps 2-4 to be redone. Updating facts, arbiter and primer after the deep reads below is enough.

---

1. Coverage

• Manifest: 11 documents and 79 chunks. A script check found a digest file for every chunk; none is missing [read: manifest.json, digest/]. No document was skipped. facts.json has 11 entries, one per manifest document.
• Duplicate handling is acceptable. The second vortex scan (treatiseonmotionofvotexrings-0ee5cd) and the 1904/1907 printings of Electricity and Matter were digested in full. The 1907 deep read is an edition check.
• Unrecorded content in the 1904 scan. The digest of electricitymatter1904-07642b reports a reader's margin annotations in the project's copy (PDF pp.69-75). They record that p.72 is missing from the original scan and that a page from a different (1907) printing of the book was inserted at p.73, and one (p.75) links the stop-pulse passage to an external cause of radioactivity [read: digest 1904 part 02]. The PDF is therefore a composite with one page taken from the 1907 printing. facts.json says neither.
• The annotations are not primary evidence. The last one bears directly on Chapter VI, where Thomson discusses an external (Le Sage-type) energy source; see Missed material item 7.
• Defect, minor. facts.json should record both points.
• The textbook (Poynting and Thomson) is digest only. That is acceptable for a peripheral work, but see Missed material item 1: its thermal-effusion section bears on the untested vortex prediction.

2. Shortcuts

Digest-only coverage is fine for peripheral chapters. These shortcuts touch central claims:

1. Claim 1 (electron e/m and charge) rests on digest plus [memory]. The deep read of Corpuscular Theory Chapter I was lost to an overwrite (claims.md says so). Conduction Chapters V-VI, where Thomson tabulates his own e/m and e measurements, were never deep read. As a result the primer's statement that "Thomson's 1897 deflection measurement ... all agree [memory, high]" was never checked against the book, and the book contradicts it (spot check S2). The verdict "Established" does not change, but the historical account is wrong. Defect.
2. Claim 6, Part IV (vortex-atom gas theory and chemistry) is digest only and has no verdict. The arbiter lists "Gas of vortex atoms reproducing kinetic pressure: not assessed (digest only)". The digests show that Part IV makes definite, testable statements (Missed material item 1). Leaving part of a central claim without a verdict because of a reading shortcut is a defect.
3. Claim 4, counting Method I as printed in Conduction (1906) was not read. All reviews and the worked example use the 1907 Corpuscular Theory figure, Barkla's ".00025". Conduction p.339 quotes Barkla's datum as "about ·0002" and then writes "= ·00025" in the next line (spot check S1). Two documents carry the same central derivation, and the reviews read only one of them. Defect, minor.
4. Claim 2, Kaufmann table. The reviews and the arbiter compared Conduction (1906) with Corpuscular Theory (1907) only. Electricity and Matter Table II (1904/1909) carries the same table, and its labels agree with the 1907 book, not the 1906 one (spot check S5). The arbiter's statement that "the 1907 book's headings are interchanged" is therefore incomplete. Defect.
5. Positive Rays, earlier chapters (multiply charged atoms and negative ions), are digest only. facts.json lists one key claim from them ("Multiple charges ... increase with atomic weight") that no reviewer judged. Minor; acceptable if facts.json marks it as unreviewed.

3. Missed material

These items are in the digests (checked against the text where marked) but are absent from facts.json, claims.md, the reviews, the arbiter and the primer, or are misstated there.

1. Vortex Rings, Part IV: testable predictions [read: digest second scan pp.133-148; first scan pp.134-149]. - Gas properties:
   • Thomson predicts that in a monatomic vortex gas the mean molecular velocity falls as temperature rises (p.136), unlike ordinary kinetic theory.
   • He proposes thermal effusion as the deciding experiment. Ordinary theory gives p/p′ = (T/T′)^½ (Maxwell). Vortex theory gives p/p′ = (t/t′)^m with m > 1 (p.137).
   • pV comes out "a little less" than Boyle's value, which he says agrees with Regnault (p.136).
   • Chemistry:
   • Valency equals the number of linked rings, at most six, which he supports by WCl6 and Lothar Meyer (pp.144-145).
   • His valency table makes nitrogen, phosphorus, potassium and mercury univalent and carbon divalent (p.148).
   • A folded figure-of-eight ring is proposed for NO and NO2.
   • Dielectric strength of gases is linked to the ratio of paired time to free time (p.140, citing Phil. Mag. June 1883).
   • The 1909 Poynting-Thomson Text-Book states the kinetic-theory thermal-effusion law (gas streams from cold to hot; flux ∝ T^-½) [read: textbook digest p.214]. A book Thomson co-signed therefore gives the law his 1883 prediction contradicts.
   • Knudsen's thermal transpiration measurements (about 1909-10) supported the square-root law [memory, medium; to web-check]. On that record the prediction would be "Tested, not supported". The reviewers must decide that after a deep read.
2. Part I of the vortex essay. It claims a circular ring is stable for all small displacements of its core line, and that an elliptic vibration takes 0.289 of the time the ring needs to travel one circumference (pp.59-61 second scan). No one checked it. It is lower priority than Part IV.
3. Thomson's own e/m and e values, which bear on claim 1 [read: Conduction p.135 page image; digest pp.134-139, 163-175]. - The 1897 deflection method gives "e/m, viz. 7·7 × 10^6" and a hydrogen-ion mass "770 times" the cathode-ray particle. The digest misread this as 7.7 × 10^7 (correction C3). - The energy (thermocouple) method gives 1.17 × 10^7. - The 1898 cloud-chamber charge was 6.5 × 10^-10 esu, about 35% high. It was revised to 3.4 × 10^-10 in 1901-02. H. A. Wilson got 3.1 × 10^-10. - The "1/1700" figure (p.173) comes from Kaufmann, Simon and others (e/m about 1.86 × 10^7), not from Thomson's 1897 numbers. - This is a good lesson for the reader: the first measurement was about a factor of 2.3 off. None of it appears in the arbiter or the primer.
4. Universal positive carriers. Corpuscular Theory Chapter I claims that at very low pressure every gas gives the same two positive carriers, with e/m = 10^4 and 5 × 10^3, and concludes that positive electricity comes in two units: the hydrogen atom and twice it [read: digest pp.35-38]. This is a claim Thomson later dropped. Today it would be read as H+ and H2+ from hydrogen contamination [memory, high]. It is not in facts.json or claims.md.
5. Thomson's own specific-heat paradox for free electrons. In Corpuscular Theory Chapter IV, Thomson computes that the corpuscle count conductivity needs, as Rubens' reflection data imply, would require about 6 cal per °C per cc of silver. The whole metal needs about 0.6 cal. He calls this a contradiction [read: digest pp.99-100]. Chapter IV (p.90) also concludes that free-corpuscle conduction is "at most only a part" of metallic conduction, because thermoelectric effects do not change at melting. - This is the electron specific-heat problem, which was solved only by Fermi-Dirac statistics (Sommerfeld 1927) [memory, high]. - For a learner it is a fine example of a correct anomaly found by a classical model. It is absent everywhere, along with his Wiedemann-Franz derivation and the Hall-sign difficulty.
6. Thomson's pre-1913 thinking on transmutation and atomic energy, which bears on claim 5 [read: Conduction digest pp.333-334, 344; E&M digest p.162]. - From Bumstead's 1906 finding that lead heats about twice as much as zinc for equal X-ray absorption, Thomson suggests the rays release atomic energy, "a way to tap atomic energy". - He reports his own null search for radioactive products in metals exposed to X-rays for several hours, and similar null radium experiments made with Bumstead. - He suggests testing for transmutation by keeping radium in contact with lead. - So his 1913 caution about helium from salts follows earlier negative searches of his own. That belongs in the claim 5 record.
7. Le Sage gravity by X-rays [read: E&M 1909 digest pp.175-179; 1904 pp.174-179; 1907 pp.176-183]. - In Chapter VI Thomson suggests that very penetrating Röntgen rays could replace Le Sage's corpuscles. This would give inverse-square gravity, proportional to the product of the masses if absorption is proportional to density. He notes the objection that gravity would then travel at light speed. - He says radium's output is "not a valid argument" against an external energy source. He places the source inside the atom only because radioactivity is transient. - facts.json describes this as "the rejection of an external-radiation (Le Sage-type) source", which misstates it (correction C5). It is pertinent alternative-science material, and the project's copy carries a margin annotation on the related passage. - Related items: his radium lifetime estimate of "about 50,000 years" (p.171); his estimate of 10^19 erg stored per gram of hydrogen is already recorded.
8. The sun as a source of corpuscles. Conduction Chapter VIII suggests that the sun, and luminous stars generally, emit corpuscles, and links this to Arrhenius's account of the aurora [read: digest p.211]. This is an early form of the solar-wind idea. It is not recorded.
9. Ionising radiation from all elements. Conduction Chapter XIII reports ionisation in closed vessels by Campbell, Strutt and others that lead shielding never reduces below about one third, plus a diurnal variation. Thomson weighs and rejects a radium impurity [read: digest pp.423-432]. This is historical context; later understood as cosmic rays and K-40 [memory, medium]. Low priority.
10. X-ray diffraction and speed in Thomson's own 1906 text, which bears on claim 3.
    • Conduction Chapter XX reports Haga and Wind's slit broadening as diffraction evidence. Thomson rereads the implied wavelength as the pulse thickness (pp.661-662).
    • It also reports Marx's finding that the speed of Röntgen rays equals the speed of light [read: digest pp.659-662].
    • Reviewers cite Haga and Wind only from [memory]. The primary text is in the folder.
11. Atomic charges in molecules, a tension inside the corpus. Positive Rays (1913) argues that the atoms in compound molecules are not oppositely charged (CO and phosgene counts; nickel carbonyl; H2 + Cl2 under light) [read: digest pp.84-86]. Corpuscular Theory (1907) argues that like atoms in O2 are oppositely charged, and uses this to explain the paramagnetism of oxygen [read: digest p.155]. Positive Rays also claims that mercury ionises by losing "one or eight" corpuscles, and nothing in between (pp.68-70). None of this is recorded.
12. Corpuscle radius. Electricity and Matter gives a ≈ 5 × 10^-14 cm (p.66). The 1906-07 books give about 10^-13 cm. This is a small inconsistency inside the corpus that the claim 2 discussion could note.

Incidental, not a defect: the textbook's advertisement pages list "Theo Kassner, Gold Seeking in South Africa" (Griffin) [read: textbook digest p.272]. This is probably unrelated to E. E. W. Kassner of the Ternion project and is noted only as a possible name lead.

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

example.py runs cleanly with the .venv interpreter, and its printed numbers match primer §5 and the answer to Q2 [read: run output]. Part D shows the e/m products only. It does not show that the 1904 book labels the columns like the 1907 book.

5. Retention

I checked that material the Investigator found (sites.md "Leads for review") was kept in the arbiter or primer when a reviewer judged it negatively or did not discuss it. - Kept: the Nature 1913 letter; the 1921 second edition; the 1922 Proc. A paper; Structure of Light and Beyond the Electron; Kragh; LaFave (with the reviewers' caveats, stated fairly); Rayleigh's "specks"; the 1913-1924 controversy on both sides (Redgrove, Nelson, the Royal Society referee); Wendt-Irion; Paneth-Peters; CUL MS Add. 7654; Morrisson; Navarro. Good. - Dropped: - C. N. Wall (Phil. Mag. 48:378, 1924), an extension of the Stokes-Thomson pulse theory by another author, listed under "Supporting information". The physicist mentions it ("not read, so no judgement"). The arbiter and primer drop it. Defect, minor: it belongs in the claim 3 record and in the primer's Further reading or open-questions list. - Nelson's identification of X3 as "tritium" [sites.md, Nelson row]. This is an alternative-science author's characterization. It points to a nuclear-production reading of the mass-3 line, and claim 5 is exactly where such a reading belongs. No review, the arbiter or the primer records it. Defect, minor. Record it as Nelson's characterization, beside the physics note that tritium was unknown until 1934 and that H3+ and HD+ are the mainstream assignments [memory, high]. - Tolman (1922, thermodynamics of the discharge-tube claims) and the Trinity College fonds THMJ/THMF are dropped from the primer. These are trivial omissions. I flag them only for completeness. - Nothing was dropped because a reviewer judged it negatively. Pertinent source material was dropped for lack of reading: Missed material items 1, 5, 6 and 7. Item 7, the X-ray Le Sage gravity speculation, is also mischaracterized in facts.json.

6. Process

• Source tags: present and generally honest. Reviewers mark material known only by title or abstract as such. The arbiter's [calc] work is explicit.
• Web checks: the arbiter used 5 of its 8 calls. One medium-confidence [memory] fact still carries a verdict without a web check: "neutral H3 is not a stable ground-state molecule". The neutral-H3 verdict "Tested, not supported" leans on it plus Kragh's abstract, and the arbiter admits the gap. Process defect, minor.
• The physicist's alpha-separation energies were later web-checked by the arbiter. Good.
• The Knudsen thermal-transpiration fact will need a web check once Part IV is assessed.
• Verdict scale:
• Applied consistently for most sub-claims.
• The arbiter and the primer (row 6a) use the label "internally consistent with errata" and the ring-mathematics note in claim 4. Neither is a scale verdict.
• Row 6d reads "No verdict".
• Fix: give the ring and vortex mathematics a scale label (e.g. "Established (as mathematics, with errata)"). Give Part IV its verdicts after the deep read.
• Primer against STEP 4 and the reader profile:
• All nine sections are present, in order, with answers at the end.
• Symbols are defined. emu/esu, Thomson scattering, bremsstrahlung, LENR and linear stability get refreshers.
• The worked example is clear and reproduced by example.py.
• Two accuracy fixes are needed: S1 and S2.
• Corrections sections: present in claims.md, the reviews and the arbiter. There is no sign yet that the orchestrator has applied the arbiter's corrections 1-8. claims.md still says "supported by three counting methods" and "up to 6 vortices". The QE recheck should confirm they are applied.

7. Corrections to earlier documents (for the orchestrator)

• C1 (arbiter, "Kaufmann table" paragraph and correction 2; primer §4 table, §7 bullet, answer 2; claims.md claim 2):
• Wrong: the statement that "the 1907 book's headings are interchanged", and the framing that the 1906 and 1907 books disagree.
• Correct framing: Electricity and Matter (1903 lectures; 1904/1907/1909 printings), Table II at PDF p.65 of the 1909 scan, labels the columns like the 1907 Corpuscular Theory. Only Conduction (1906) labels them the other way.
• Keep the e/m-product argument, and state its assumption (one common slow-speed e/m of about 1.95 × 10^7).
• Say the labelling is decided only by Kaufmann's original table [read: page images E&M p.65, Corpuscular p.48, Conduction p.668].
• C2 (primer §5 steps 1-7 and point 4; arbiter "Barkla count"; reviews):
• Add that Conduction (1906, PDF p.339) quotes Barkla's datum as "about ·0002" and then uses ·00025.
• Recomputed: 17.2 per air molecule with Thomson's constants, 11.2 with modern ones; forward check 2.58 × 10^-4 against a quoted 2.0 × 10^-4 [calc].
• The conclusion (about Z, not A) is unchanged. The "steering" lesson becomes stronger.
• C3 (Conduction digest part 04, p.135 and summary): "e/m about 7.7 × 10^7" should read 7.7 × 10^6, consistent with "770 times" [read: page image].
• C4 (Conduction digest part 21, p.668): the digest has the labels reversed ("calculated values read as 3.1 ...; observed as 3.09 ..."). The page prints ρ′ (observed) = 3·1 ... 1·5 [read: page image].
• C5 (facts.json, E&M 1909 summary): "the rejection of an external-radiation (Le Sage-type) source" should become: "Thomson places radium's energy inside the atom because radioactivity is transient. He states that radium's output is not a valid argument against an external source, and he proposes that penetrating Röntgen rays could replace Le Sage's corpuscles as a mechanism for inverse-square gravity" [read: digests E&M Ch. VI].
• C6 (primer §3.1): replace "Thomson's 1897 deflection measurement ... all agree" with the measured history: Thomson 1897, 7.7 × 10^6 (deflection) and 1.17 × 10^7 (energy method); Lenard 6.4 × 10^6; Kaufmann and Simon about 1.86 × 10^7; Thomson's 1898 e of 6.5 × 10^-10 esu revised to 3.4 × 10^-10 [read: Conduction pp.135-173, digest and page image p.135].
• C7 (facts.json, electricitymatter1904 entry): record the margin annotations in the project's copy and the page inserted from the 1907 printing.
• C8 (sites.md "Leads" and the arbiter's claim 3 and claim 5 records): keep C. N. Wall 1924 and Nelson's "tritium" characterization (Retention).

8. Remediation actions

Deep reads (most important first; slugs from manifest.json):

1. treatiseonmotionofvotexrings-0ee5cd, pp.133-148 (Part IV), Opus. Set out and check the vortex-gas results: - pV against Boyle, and the deviation sign compared with Regnault; - the effusion law p/p′ = (t/t′)^m with m > 1, and the claim that mean velocity falls with temperature; - the paired/free-time theory of combination; - the valency rules (at most six links, the valency table, the figure-of-eight nitrogen). Read equations from page images. Which predictions are testable, and what is known of their tests?
2. ELECTRICITY-AND-MATTER-1909-828b3e, pp.61-66 (with Conduction...-23223e pp.664-668 and The-corpuscular...-ab57d0 pp.47-48), Opus. - Reconstruct equation (1) and compute ρ at the five speeds. - Establish which column each of the three books calls observed. - Test whether either column is reproducible from Thomson's model or from Kaufmann's e/m table with a stated (e/m)₀.
3. Conduction-of-electricity-through-gases-23223e, pp.130-175, Sonnet. Tabulate every e/m and e value Thomson reports, with method, observer and date: his 1897 and 1899 values, Lenard, Kaufmann, Simon, Wiechert, his 1898-1902 cloud values, H. A. Wilson. Where does the 1/1700 come from?
4. Conduction-of-electricity-through-gases-23223e, pp.333-344, Sonnet. Record: - the Barkla datum as printed (0.0002 against 0.00025) and the full counting argument; - Thomson's interpretation of Bumstead's atomic-energy result; - his reported null searches for radioactive products after X-ray and radium exposure.
5. The-corpuscular-theory-of-matter-ab57d0, pp.16-42 (Chapter I; the lost deep read), Sonnet. What values of e/m and e are given? What exactly is claimed about the two universal positive carriers (10^4 and 5 × 10^3), and on what evidence?
6. The-corpuscular-theory-of-matter-ab57d0, pp.64-117 (Chapters IV-V), Sonnet. Check the specific-heat contradiction (6 cal against 0.6 cal), the Wiedemann-Franz ratio derivation and its comparison with Jaeger-Diesselhorst, the Hall-effect sign argument and the "at most only a part" conclusion.
7. ELECTRICITY-AND-MATTER-1909-828b3e, pp.167-179 (Chapter VI end), Sonnet. Quote Thomson's external-source and Le Sage-by-X-rays argument and the 50,000-year radium lifetime arithmetic, and say whether the argument is internally consistent. Note the margin annotation in the project's copy of the 1904 scan (electricitymatter1904-07642b pp.69-75).
8. Conduction-of-electricity-through-gases-23223e, pp.657-662, Sonnet. What does Thomson say in 1906 about the Haga-Wind diffraction evidence and the Marx X-ray velocity, and how does he reconcile them with the pulse theory?

Web checks (by the remediation step): - Barkla 1904 (Phil. Mag. (6) 7, 543): the scattered fraction for air. - Kaufmann 1901-1903: the observed mass-ratio table. - Neutral H3: the ground-state stability fact behind the neutral-H3 verdict. - Knudsen (about 1909-10) thermal transpiration, the ½-power law (for item 1).

Later files: - facts.json: apply C5 and C7. Add the Part IV predictions, Thomson's measured e/m and e history, the universal-positive-carrier claim, the specific-heat paradox, the pre-1913 atomic-energy and transmutation searches, and Le Sage-by-X-rays. Mark "multiple charges increase with atomic weight" as unreviewed, or review it. - claims.md: apply the arbiter's corrections 1-8 (not yet applied) and C1/C2. Add to claim 6 the Part IV gas and valency predictions. Add to claim 1 that Thomson's own first values were low. - arbiter.md, "QE revision" section: - Kaufmann framing (C1). - Barkla datum (C2). - A verdict for the Part IV predictions, after deep read 1 and the Knudsen check. - Claim 5 record: Thomson's earlier null searches, Wall, and Nelson's "tritium" characterization. - Claim 3 record: Haga-Wind and Marx from Thomson's own text. - Scale labels for the ring and vortex mathematics. - Web-check the neutral-H3 fact or keep its caveat. - primer.md: - Fix §3.1 (C6), the §4 Kaufmann row and §7 bullet (C1), and §5 (C2: add the 0.0002 note and the recomputed 17.2 and 11.2). - Add a §6 row for Part IV. - Add to §7 "What is genuinely open or instructive": the specific-heat paradox as an anomaly the classical model found, and the Le Sage-by-X-rays proposal as Thomson's own speculation. - Add Wall 1924 to §9. - Update answer 2 and example.py Part D: print the 1904 labelling and the (e/m)₀ assumption. - QE recheck (Sonnet): confirm C1-C8, the eight deep reads and the four web checks, and append "Recheck".

Recheck

QE recheck, 2026-10-09. Read the current claims.md, arbiter.md (QE revision), primer.md, facts.json, sites.md, example.py (run with the .venv interpreter), the new deep sections and the digests. I did no web searches and did not redo the page-image checks.

Corrections C1-C8

• C1 (Kaufmann framing): fixed. claims.md claim 2 and QE note, arbiter QE revision and primer §6/§7 say that E&M (all printings) and the Corpuscular Theory label 3.09 ... as Kaufmann's, and only Conduction labels 3.1 ... that way. The "interchanged" and "resolved" wording survives only inside quoted "do not say" instructions and a superseded note in the arbiter. example.py Part D prints the three labellings and the (e/m)0 assumption.
• C2 (Barkla 0.0002/0.00025): fixed. claims.md claim 4, primer §5 Step 8 and point 4, facts.json and example.py Part E show 17.2 and 11.2. example.py runs and agrees.
• C3 (digest p.135, 7.7 x 10^7): remaining, minor. The correction is recorded in claims.md and the arbiter. The digest file digest/Conduction...-23223e.04.md line 45 still reads "about 7.7 x 10^7".
• C4 (digest p.668, labels reversed): remaining, minor. It is recorded in the correction lists, but the digest file was not edited. Same status as C3. I count C3 and C4 as one issue.
• C5 (Le Sage statement in facts.json): fixed. The E&M 1909 entry now gives Thomson's actual position.
• C6 (primer §3.1, e/m history): fixed. The primer has the master table, 7.7 x 10^6 and the factor 2.3, and claims.md claim 1 matches.
• C7 (margin annotations in the 1904 scan): fixed. Both the 1909 and 1904 entries in facts.json record them and the inserted 1907 page.
• C8 (Wall 1924, Nelson "tritium"): fixed. Both are in sites.md, arbiter claim 5 and primer §9 (Wall) and §4 (Nelson).

Deep reads and web checks

• Reads 1-6: done. Part IV, the Kaufmann table, the e/m and e history, the Barkla count with the Bumstead and null searches, Chapter I (two positive carriers), and Chapters IV-V (specific heat, Wiedemann-Franz, Hall sign) all have sections in deep/. The specific-heat paradox is in the primer §7 and S1.
• Reads 7 and 8: done in deep/, but not carried through. deep/ELECTRICITY-AND-MATTER-1909-828b3e.md (Le Sage, 50,000 years) and deep/Conduction...-23223e.md (Haga-Wind, Marx) now have sections. claims.md claim 3 and the primer's Le Sage paragraph use them. The arbiter's QE revision (line 175) still says both deep reads were "not done", and the primer §7 item 8 still says the Haga-Wind and Marx text was "read only by digest". Remaining: issue 1 (stale statements). The arbiter's claim 3 record has no verdict update from these reads.
• Web checks: Knudsen/Reynolds done (arbiter; Part IV verdicts), neutral H3 done (arbiter master table says "now [web]"). Barkla 1904 and Kaufmann 1901-03 were not obtained. They are listed as open items in the arbiter and primer, with the conclusions stated as conditional on them. That is acceptable disclosure, not a defect.
• Part I of the vortex essay remains unchecked. It is disclosed in the primer (open list, item 6) and was lower priority in the QE.

Other checks

• claims.md no longer says "supported by three counting methods" unqualified or "up to 6 vortices" (claim 6 now says stable for 3 to 6, seven neutral, eight unstable). Arbiter corrections 1-8 are applied in claims.md.
• Verdict scale: row 6a now reads "Established as mathematics, with errata"; rows 6d-6h have scale verdicts.
• Coverage and the 79-chunk digest set are unchanged.

Remaining defects

1. Stale statements in arbiter.md line 175 and primer.md §7 item 8 (deep reads 7 and 8 said to be undone or digest-only, although they are done and used elsewhere).
2. Digest files not edited for the C3/C4 errors (recorded only in correction lists).

Final: OPEN (2 issues)

Orchestrator fix (2026-10-09)

Changes made (each carries an inline "[corrected 2026-10-09 ...]" note): 1. arbiter.md QE revision - The "Not done" paragraph (former line 175) is corrected. Deep reads 7 and 8 are done, and the note names their sections; the original wording is kept in quotation. The open list (item 8) no longer lists them, but the cited Haga-Wind, Wind and Marx papers stay open. - New section 10, "Deep reads 7 and 8: what they mean for claim 3". Read 8 shows that in 1906 Thomson turned Haga-Wind diffraction into a pulse thickness (about 2 x 10^-8 cm), so diffraction as such does not decide pulses against waves (Fourier equivalence). The deciding tests are the characteristic lines and the Duane-Hunt cutoff, which crystal diffraction made measurable. - Verdict unchanged: the strong form stays Tested, not supported, with its grounds refined, and the arbiter says so plainly. A pointer is added at the claim 3 verdict line. Read 7: the 50,000-year arithmetic is correct and is an upper limit. The Le Sage-by-X-rays proposal is kept in the record with Thomson's own objections and his unperformed calculation, and gets no verdict (it is not a central claim and no reviewer assessed it). 2. primer.md - §7 open item 8: "read only by digest" is corrected. - The Le Sage paragraph: "deep read (not yet done)" is corrected and the deep reader's two gaps are added. - Source note item 1: chapter VI and Conduction ch. XX are now listed as deep read. - The §4 X-rays row, §6 row 3c (test column) and the §7 "X-rays as pulses" lesson now give the refined grounds. Verdict text unchanged. - The header records the second update. 3. digest/Conduction...-23223e.04.md: p.135 now reads 7.7 x 10^6 (C3). The same error is also corrected in the same file's p.137 line and its chunk summary. 4. digest/Conduction...-23223e.21.md p.668: the labels are corrected (C4). Observed (Kaufmann) is 3.1 ... 1.5 and calculated is 3.09 ... 1.65, as the page prints, with a note that the other two books label the columns the other way.

Stale-copy search: no other "7.7 x 10^7" in review files, deep/ or facts.json. facts.json how_read already lists both deep reads.

Check (QE standard)

• Remaining defect 1 (stale statements): arbiter and primer no longer say reads 7 and 8 are undone. A grep for "not done", "read only by digest" and "not yet done" finds only the corrected notes. The new arbiter weighing quotes the deep reads accurately: I checked it against the deep sections (p.662 quote, 2 x 10^-8 cm with the exponent caveat, Marx with no number, 4.8 x 10^8 h and 5 x 10^10 cal). Fixed.
• Remaining defect 2 (digest C3/C4): both are corrected in place. C3 is consistent with 1/(1.3 x 10^-7) and with "770 times". C4 follows QE spot check S5. Fixed.
• The new [memory, high] items (Fourier/white-light equivalence, Duane-Hunt, characteristic lines) are tagged and flagged as not web-checked, in line with the arbiter's existing list. No verdict rests on a new unchecked item, because the verdict is unchanged.
• Not part of this fix, noted for completeness: arbiter correction 4 also lists Corpuscular digest corrections (p.30, p.89). The QE recheck did not list them as remaining, and I did not check them.
• Site not re-rendered (render_site.py not run, as instructed).

Orchestrator, 2026-10-09: arbiter correction 4 Corpuscular digest items applied after checking page images (PDF pp.30, 90): digest .01 p.30 e = 3.1 (was 8.1) x 10^-10 esu; digest .03 p.89 V = (2/3)(a theta/e) log(N1/N2), N1/N2 = 2.36 x 10^17 (was 2.86), Sb/Bi 1/30 volt and log 1.33 (was 1/80, 1.88).

Final: PASS