Soddy: Quality Engineering (Step 4b)

QE date: 2026-10-08. Inputs read: _text/manifest.json; all 16 digests (Interpretation 01-10, Matter and Energy 01-06); both deep reads; facts.json; claims.md; sites.md and the research/ notes list; physicist.md, mathematician.md, contrarian.md; arbiter.md; primer.md; example.py (run). Spot checks were made against the extracted text in _text/ and two page images rendered from the source PDF (scratch: /tmp/soddy_qe). Tags: [read] extracted text or a page image I looked at, [calc] my arithmetic (Python, .venv), [memory] with confidence, [web] only where an earlier role gave the URL.

Verdict: MINOR

The central verdicts (C1 to C7e in arbiter.md) are sound and no deep read requested below is likely to overturn one. The defects are: one factual error in facts.json, a misreading repeated through all three reviews and the primer ("nearly a million times"), two book-internal inconsistencies that nobody found (helium from uranium and the "thirteen per cent" uranium energy figure), digest-only coverage of several chapters that hold pertinent claims (the actinium branch hypothesis, Soddy's own helium-from-uranium experiment, the second-law-and-life passage), and three cases where pertinent material was dropped between the reviews and the arbiter or primer. All of them can be fixed by updating facts.json, claims.md, the arbiter (a "QE revision" section) and the primer.

---

1. Coverage

Both manifest documents were digested in full. Nothing was skipped. The source-gap note (no economics books or research papers in the folder) appears in claims.md, the arbiter and the primer. Coverage: no defect.

2. Shortcuts

Digest-only coverage is fine for the peripheral chapters (rays, X-rays, the periodic table, kinetic theory). It is not fine for the following, each of which holds a central claim, an experimental result or a pertinent claim that no deep read covers:

1. Interpretation PDF pp.133-148 and 165-180 (helium, origin of radium). These hold Soddy's own four-year experiment on helium from uranium and thorium, and the Soddy-Hitchins radium growth experiment, which carry claims 3 and 4. The mathematician read p.178 directly and the arbiter checked p.169. The helium measurement (pp.142-145) was read by nobody, and it contains an internal inconsistency (section 4, item 5).
2. Interpretation PDF pp.216-220 (energy of uranium). Digest only. Holds "thirteen per cent more than radium", "one hundred and sixty tons of coal" per pound of oxide, and "360,000 times" coal for radium. The 360,000 figure undoes a reading that all reviewers share (section 4, item 3).
3. Interpretation PDF pp.234-256 (thorium and actinium series, Chapter XII). Digest only, and no review weighs it. It holds the actinium "branch" hypothesis and the co-discovery of protactinium (eka-tantalum) by Cranston and Soddy and by Hahn and Meitner, which is a priority statement by an interested party (section 3).
4. Matter and Energy PDF pp.101-118 (second law, life, Le Sage). Digest only. facts.json lists "living bodies may not obey the second law" as a key claim, and no later file weighs it (section 5).
5. Matter and Energy Chapter X (PDF pp.236-257). The contrarian read pp.240-250. The rest, including the negative self-bombardment test on p.247, rests on the digest.
6. Matter and Energy Chapter I (PDF pp.13-42). Claim C7c ("energy and wealth are synonymous", coal as capital) rests on the physicist's own reading of pp.38-39 and on the digest. That is acceptable for a claim judged "Not testable as stated", but it is thin for the passage that links Soddy to his later economics.

Sampling only part of a key document is a defect under STEP 4b. Items 1-3 are the significant ones. Deep reads are requested in section 9.

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

Each item is pertinent and appears in no review, in the arbiter or in the primer unless stated otherwise.

1. Actinium as a "branch" of uranium (Interp. PDF pp.247-250, 253-254) [read: digest 08]. Soddy argues that actinium cannot be in the main uranium line, because its alpha activity in minerals is only 1/15 to 1/16 of the uranium series where 5/8 would be expected. He proposes that about one uranium atom in twelve (about 8 per cent) decays along an actinium branch through uranium Y and eka-tantalum. Modern view [memory, high]: the actinium series descends from a separate isotope, U-235 (found by Dempster in 1935), not from a branch of U-238. U-235's share of natural uranium activity is about 4.6% [memory, high], which matches his 1/15-1/16 well. This is an instructive case for the primer's isotope section: chemistry cannot tell a separate uranium isotope from a branch, because the two are chemically identical. Suggested verdict for the arbiter: Historically superseded but instructive.
2. Protactinium priority (Interp. PDF p.252) [read: digest 08]. Soddy reports the 1919 discovery of the parent of actinium "by Cranston and the writer" (eka-tantalum) and by Hahn and Meitner (proto-actinium). This is a second priority statement by Soddy, an interested party, and should sit beside the isotope priority in C5. Hahn and Meitner's paper is usually dated 1918 [memory, medium]. Not web-checked.
3. Soddy's own helium-from-uranium experiment (Interp. PDF pp.142-145) [read: extracted text p.142]. "I was engaged for four years ... the rate of production is almost exactly what is to be expected ... about one five-hundred-thousand-millionth of the mass of the uranium or thorium per annum", which is about 2 mg of helium per thousand tons of uranium per year (p.145, digest). This is Soddy's own experiment and his independent check on the uranium life (p.169). The mathematician notes "no data given" for it, and no file examines it (see 4.5).
4. Range and period: the Geiger-Nuttall relation (Interp. PDF pp.207-211) [read: digest 06]. Soddy prints the alpha velocity and range table and says that log period against log range gives three parallel straight lines, "the reason still obscure". He used the rule to estimate ionium (200,000 years, later measured as 100,000) and uranium II (about two million years; modern mean life about 354,000 years [mathematician, web]). This is the empirical law that Gamow's 1928 tunnelling theory explained [memory, high]. The primer's tunnelling section could use it as its bridge. The physicist mentions Geiger-Nuttall only in passing.
5. Pleochroic halos (Interp. PDF pp.211-215) [read: digest 06-07]. Joly's halos in mica have ring radii that match the alpha ranges of the uranium and thorium series, from fewer than about 100 alpha particles a year acting over hundreds of millions of years. They bear on C1: they are geological evidence that alpha energies, and so decay behaviour, have stayed the same over very long times. No review uses them.
6. A negative test of induced decay (Matter and Energy PDF p.247) [read: digest 06]. Bombardment by expelled helium "would be expected to speed decay in concentrated radium, but this was not found". This is a recorded negative experiment that bears on C1 ("cannot be influenced") and on C6d (whether alpha bombardment induces transmutation). Under the contrarian brief it belongs in the record.
7. Life and the second law (Matter and Energy PDF pp.104-108) [read: digest 03]. Soddy asks whether a "molecular intelligence" would "vitiate" the second law, and whether living cells escape it by turning food energy straight into work without passing through heat: "Time alone and experiment will answer it decisively." This is a key claim in facts.json and is dropped thereafter. See section 5.
8. Le Sage gravitation and "tractate/pellate" (Matter and Energy PDF pp.113-118) [read: digest 03]. Soddy gives Le Sage's "ultra-mundane corpuscles" a sympathetic hearing as a way to make potential energy kinetic, and ends with "the real causes of tractation and pellation are unknown, though not necessarily unknowable". Le Sage gravity recurs in alternative science, so a line in facts.json is warranted.
9. Atmospheric emanation (Matter and Energy PDF p.141) [read: digest 03]. The radium emanation in the atmosphere is put at the equivalent of "hundreds of tons of pure radium", and its rays help oxygen and nitrogen combine. As an order of magnitude the inventory is plausible [calc, rough: a global radon flux of about 1e21 Bq per year [memory, low] times a mean life of 5.5 days gives about 1e19 Bq, or about 300 t of radium equivalent]. Its role in fixing nitrogen is negligible [memory, medium]. Low priority.
10. Missing elements (Interp. PDF p.287) [read: digest 09]. From Moseley's count Soddy lists six missing elements and predicts that the two heavier manganese analogues "would in all probability be most useful metals, allied to the noble metals". These are technetium and rhenium (rhenium found in 1925 [memory, high]). Low priority; a check-your-understanding item.
11. Photochemical isotope separation (Interp. PDF p.290 footnote). Merton and Hartley (Nature, 25 March 1920) proposed separating chlorine isotopes with light filtered through chlorine. This is an early form of selective photochemical (later laser) isotope separation [memory, medium]. Low priority.
12. Ramsay and Cameron's copper-to-lithium claim (Investigator lead). All three reviewers weighed it, and the physicist gave it a verdict with a Coulomb-barrier calculation. It is missing from the arbiter's verdicts and from the primer (section 5).

4. Accuracy spot checks

Defects found:

1. facts.json (Matter and Energy), mainstream_conflict: "ionium unnamed and unproved in 1911 ... this book names it only as an 'intermediate substance'" is wrong. The 1911 text reads "an intermediate substance, ionium, which does produce radium steadily with the lapse of time, intervenes. The average life of ionium is probably at least a hundred times longer than that of radium" [read: ME PDF p.224]. The deep read quotes this correctly. The same entry's "relevance" ("before ionium was isolated and named") and summary ("ionium named only as an intermediate") repeat the error. Correct statement: the 1911 book names ionium but gives no discoverer, no measured life (only "at least a hundred times" radium's) and no growth result.
2. The "1905" misprint is systematic. Matter and Energy prints 1905 where 1894-95 is meant in four places: "Before 1905 only one monatomic gas was known" and "the inert gases ... discovered in 1905" [read: page image PDF p.128]; "until 1905" for the overlooked X-rays [read: extracted text, PDF p.165]; "the X-rays by Röntgen in 1905" [read: page image PDF p.204]. Argon was found in 1894, terrestrial helium in 1895 and X-rays in 1895 [memory, high]. The record (claims.md, arbiter correction 13, primer section 7) treats this as one slip on p.204. It should say that the book consistently misprints the 1890s dates.
3. "Nearly a million times" was misread by all three reviewers, the deep read and the primer. Soddy's sentence compares radium with "a similar weight of matter undergoing any change known previously" [read: Interp. PDF p.164]. It is not a comparison with coal alone. The most energetic familiar chemical change per gram of reactants, hydrogen burning to water, gives 3,792 cal/g, and 2.91e9 / 3,792 = 7.7e5 [calc], which is "nearly a million". Soddy gives the coal-alone factor himself as "about 360,000 times as much energy as is evolved from the same weight of coal in burning" [read: Interp. PDF p.219]. The physicist ("generous"), the mathematician ("loose"), the contrarian ("generous by a factor of 2.8") and the primer (step 6: "His phrase 'nearly a million times' for coal alone is generous") should be corrected: Soddy's three figures (nearly a million against any chemical change, well over a million against coal plus oxygen, 360,000 against coal alone) are consistent with each other. Verdicts are unaffected.
4. "Thirteen per cent" (Interp. PDF pp.219-220) [read]. "The energy evolved from uranium would be some thirteen per cent greater than from the same weight of radium", so a pound bottle of oxide holds "at least one hundred and sixty tons of coal". On Soddy's own basis, 397 g x 2.9e9 cal/g x 1.13 / 8,000 cal/g = 163 t [calc], so the arithmetic is consistent. The 13 per cent is not consistent with the 1920 book's own chain, which puts three alpha emitters (uranium I, uranium II, ionium) ahead of radium [read: digest 05-06, Fig. 34]. Counting alphas per gram, (5 + 3)/5 x 226/238 = 1.52, about 50 per cent more; one extra alpha gives about 14 per cent [calc]. The figure looks like a carry-over from the 1909 text, but that is an inference: the 1909 edition was not checked. Modern figures: the full U-238 chain releases about 1.7 times the energy per gram of radium's quick chain, and 14 oz of uranium is about 250 t of coal [calc; Q-values memory, medium]. So "at least 160 tons" is low but correctly hedged. Not verdict-bearing; record it in C2 and C7.
5. Helium from uranium (Interp. PDF pp.142, 145, 169) [read p.142; calc]. The printed rate, "one five-hundred-thousand-millionth of the mass per annum" (British billion, about 2.1e-12 per year), is what one alpha per uranium atom gives with Soddy's 8e9-year life: 4/238/8e9 = 2.1e-12 [calc]. But the same edition gives uranium two alpha emitters, uranium I and uranium II [read: digest 06, PDF p.195], and these cannot be separated chemically, so purified uranium carries both. On the book's own premises the prediction is 4.2e-12 per year, and with modern values it is 5.2e-12 per year, about 5 mg per thousand tons [calc]. The claim on p.169 that the helium experiment independently gives the uranium life of 8,000 million years is therefore not consistent within the 1920 edition. With two alphas, Soddy's measured 2 mg per thousand tons implies a life about twice as long. Either his measured helium was low by about 2.5 times or the agreement he claims rests on the earlier one-alpha picture. This needs a deep read (request 1) before it goes into the arbiter. If it holds, add it to C3 as a second internally inconsistent statement alongside "58 g". The C3 verdict for the law itself does not change.
6. Digest error (no downstream effect): digest/interpretationof00sodd-f5dd74.07.md p.226 attributes "two parts of radium per million million" to Kelvin. The text says "He assumes" and means Joly; Kelvin is cited only for the thermal insulation of the core [read: Interp. PDF p.226]. The arbiter and the primer attribute it correctly to Joly.

Checked and correct: the primer's Q-values (Ra-226 4.871, Rn-222 5.590, Po-218 6.115, Pb-214 1.018, Bi-214 3.270, Po-214 7.833 MeV) [memory, high]; the branching figures Soddy prints for thorium C (35/65) and radium C (0.03/99.97) are close to modern Bi-212 (36/64) and Bi-214 (0.02/99.98) [memory, medium] (digest 08, not used in the record); the primer's arithmetic (all of example.py).

5. Retention

Pertinent material that was weighed negatively and then dropped:

1. Life and the second law (Matter and Energy Ch. IV). facts.json carries it as a key claim, and its own mainstream_conflict judges it "not supported ... living bodies obey the second law [memory, high]". claims.md omits it, and so do all three reviews, the arbiter and the primer. A claim was removed after a negative label, without a review and without a source beyond [memory]. Soddy offers it as an open question to be settled by experiment, and it ties to Maxwell's demon and to Szilard's 1929 analysis [memory, high]. Remediation: add it to claims.md as C8 (or under C7), have the arbiter give a verdict in the QE revision, and give it a row in the primer's departures table and evidence table. A likely verdict is "Tested, not supported" for the strong form (organisms are open systems that export entropy; muscle efficiency is measured and stays within thermodynamic limits [memory, high]). The "molecular intelligence" half is instructive in its own right: Szilard and later Landauer and Bennett showed that the demon's information processing costs at least the entropy it saves [memory, high].
2. Ramsay and Cameron (1907), copper to lithium by radium emanation, tested by Curie and Gleditsch and by Soddy (negative). An Investigator lead. The physicist gave it a verdict (Tested, not supported; mechanism Refuted by calculation, 12.5 MeV Coulomb barrier) and the mathematician and contrarian kept it. The arbiter mentions only the barrier number in its calculation list, and the primer leaves it out. It is pertinent to C6d: it is the documented test-and-failure of a transmutation claim made by a Nobel laureate before 1919. Restore it to the arbiter (a C6 sub-item with the physicist's verdict) and give it one sentence in the primer (section 3.5 or the C6 row).
3. Radium nationalisation and the mesothorium patent dispute. The arbiter keeps these as absence-of-evidence item 8C. The primer omits the Knoefler dispute and Soddy's 1913 call to nationalise radium supply [web, NYT snippets via the Investigator]. Minor: add one line under "what is genuinely open", with Soddy marked as an interested party (patent holder and campaigner).
4. Fleck and van den Broek. The 1920 book credits Alexander Fleck's three years of chemical work and van den Broek's suggestion about nuclear charge [read: Interp. PDF pp.276, 279]. The primer's priority paragraph names Hevesy, Russell, Fajans, Soddy and Todd but omits both. Minor.

All other Investigator leads (Kelvin and Armstrong, Wells, Guston and Sclove, the Bodleian gaps, the Howorth letters, secret patent 5527/16, the coin request, Kitson, Morrisson, Daly and Merricks, the Lindau lectures, the 1946 manuscript) are kept in both the arbiter and the primer, stated fairly and tagged. The arbiter's absence-of-evidence analysis (section 8) follows SOURCE HONESTY properly: it sets out four competing explanations with evidence for and against each, and names the records that would decide.

6. Process

• Source tags: present and honest throughout. [calc] is used consistently. Page references were spot-checked (section 4) and are right except the p.125/126 split noted above.
• Web checks: physicist 3 of 6, mathematician 4, contrarian 5, arbiter 5 of 8. The medium- and low-confidence facts that a verdict depends on were checked (He-3 in 1939, Blackett's forks, Margaret Todd, Re-187, geoneutrinos, the Earth's heat flow). Those left as [memory] (Rutherford-Royds 1909, Gamow 1928, Cockcroft-Walton 1932) are high-confidence and do not carry a verdict. One gap: the physicist's [memory, high] figure of 1% for Be-7 is [memory, medium] in the primer. Harmless.
• Verdict scale: applied consistently, with one inconsistency of tense. C6d is scored "Plausible, untested (1920)" although every other claim is scored on present knowledge (for example, the proton radius as "Historically superseded"). Nuclear energy release is now Established (1932-45). The arbiter should lead with the present-day verdict ("Established, by routes other than the one Soddy proposed; in 1920 Plausible, untested") so that a curator's headline does not show "Plausible, untested" for something that has been done. The contrarian already puts it this way.
• Primer against STEP 4: all nine sections in order, answers at the end, symbols defined, worked example with real numbers, no textbook chapter numbers asserted. The reader profile is respected: tunnelling, Q-value, the Coulomb barrier and fm are introduced before use. "Antineutrino" (section 3.3) is used without a definition; one phrase would fix that. Further reading names the inventor's own works.
• example.py: runs (exit 0) with .venv/bin/python example.py. Every number in primer section 5 matches its output.
• Pending corrections not yet applied: claims.md still has the pre-review wording of claims 2, 3, 5, 6 and 7 (no Hevesy, no basis for "million", claims 6 and 7 not split). facts.json does not yet carry the arbiter's corrections 7-8 (Chapter X partly read; the "58 g" and lead-to-gold slips). The arbiter's "123-147 cal/h" still stands against the primer's correction to 123-145 (primer correction 1). These are the orchestrator's job, but they are listed here so the recheck can confirm them.

7. Corrections to earlier documents (for the orchestrator)

1. facts.json, Matter and Energy entry (summary, relevance, mainstream_conflict): the 1911 book names ionium ("an intermediate substance, ionium", ME PDF p.224). Delete "ionium unnamed", "names it only as an 'intermediate substance'" and "before ionium was isolated and named". Replace them with: ionium named, no discoverer, life "at least a hundred times" radium's, no growth result yet.
2. claims.md (corrections list), arbiter.md (correction 13), primer.md section 7 ("Small slips"): the "1905" date is misprinted four times in Matter and Energy (PDF pp.128 twice, 165, 204), where the true dates are 1894-95 (argon, helium, X-rays). Say "the book misprints the 1890s dates as 1905" instead of a single slip.
3. physicist.md (Claim 2a), mathematician.md (Claim 2a and table 2b), contrarian.md (section 2 item 7; section 5 item 6; corrections item 7), deep/interpretationof00sodd-f5dd74.md (section 3 table and the closing list), primer.md step 6: "nearly a million times" compares radium with "any change known previously", not with coal alone. Against hydrogen and oxygen it is 7.7e5 [calc]. Soddy gives the coal-alone factor as 360,000 himself [read: Interp. PDF p.219]. The phrase is not generous. Replace the sentence in primer step 6 with: "Soddy prints all three bases: nearly a million against the most energetic chemical change, well over a million against coal plus oxygen, and 360,000 against coal alone (p.219)."
4. digest/interpretationof00sodd-f5dd74.07.md p.226 ("Kelvin's assumption of two parts of radium per million million"): the assumption is Joly's.
5. arbiter.md section 4, C6d: lead with the present-day verdict (Established since 1932-45 by other routes); keep "Plausible, untested as of 1920" as the historical qualifier.
6. arbiter.md and primer.md: restore Ramsay-Cameron copper to lithium (Tested, not supported; mechanism Refuted by calculation, after the physicist) and the life-and-second-law claim from facts.json (section 5).
7. arbiter.md section 1 and section 4 C2 ("123-147"): apply primer correction 1 (123-145).
8. primer.md: the 33/100 split citation should read "Interp. PDF pp.125-126".

8. What each later file must change (MINOR remediation)

facts.json: correction 1; add as key claims the actinium-branch hypothesis (Interp. pp.247-250), the protactinium co-discovery (p.252, Soddy an interested party), Soddy's helium-from-uranium measurement (pp.142-145), the Geiger-Nuttall rule as Soddy states it (p.210), the pleochroic-halo evidence (pp.211-215), the uranium energy figures (13 per cent, 160 tons, pp.219-220), the self-bombardment negative test (ME p.247) and Le Sage (ME pp.116-118); update how_read with the new deep reads.

claims.md: apply the arbiter's corrections 1-5 (not yet applied) and QE corrections 2 and 3; add the life-and-second-law claim; add the actinium branch to C3 or C5.

arbiter.md (new section "QE revision"): verdicts for (a) life and the second law, (b) Ramsay-Cameron, (c) the actinium branch (suggested: Historically superseded but instructive; the separate U-235 parent explains the 1/15-1/16 activity), (d) the helium-from-uranium consistency, after deep read 1 (possibly a second "Internally inconsistent" sentence under C3), (e) the 13 per cent figure under C2 or C7. Restate C6d per correction 5. Apply corrections 2, 3 and 7.

primer.md: step 6 sentence (correction 3); "1905" (correction 2); define antineutrino; one sentence each on Ramsay-Cameron, the life-and-second-law claim, the actinium branch (as an isotope example in section 3.3 or 4), Geiger-Nuttall as the bridge to tunnelling in section 3.5, and Fleck and van den Broek in the priority paragraph; new rows in the evidence table for the added claims. Optional check-yourself question: from Soddy's 1/15-1/16 actinium activity, estimate the U-235 fraction of uranium.

Entries (Step 5): after the arbiter revision, update the Soddy entry's verdict rows (C6d headline) and its artifacts list (the protactinium papers, the 1913 isotope papers and Soddy's helium experiment as named records).

9. Deep-read requests (most important first)

1. interpretationof00sodd-f5dd74, PDF pp.141-146, 169, 195, Opus: Soddy's helium-from-uranium and thorium experiment. Exact measured quantities, accumulation times, the stated prediction ("one five-hundred-thousand-millionth ... per annum"), and how many alpha particles per uranium atom that prediction assumes. Does the 1920 text (uranium I and II, p.195) keep the prediction consistent, and does p.169's claim that helium gives 8,000 million years hold on the book's own numbers?
2. interpretationof00sodd-f5dd74, PDF pp.216-221, Sonnet: energy of uranium against radium. Quote the "thirteen per cent", "160 tons of coal", "light London for a year", "360,000 times" and "a thousand times as much as the uranium itself" passages. On what chain and alpha count does the 13 per cent rest? Compare with the 1920 chain in Fig. 34 and the table on p.199.
3. interpretationof00sodd-f5dd74, PDF pp.246-256, Sonnet: actinium origin. Quote the activity ratio (1/15-1/16 against 5/8), the branching estimate (1 in 12, about 8 per cent), uranium Y (Antonoff 1911), the eka-tantalum/proto-actinium discovery (Cranston and Soddy; Hahn and Meitner) with the dates and wording Soddy uses, and the branched radium C and thorium C fractions.
4. matterenergy00soddiala-f13420, PDF pp.101-118, Sonnet: second law, "molecular intelligence", whether living things obey the second law, Le Sage gravitation and "tractate/pellate". Exact claims, hedges and any experiment proposed.
5. interpretationof00sodd-f5dd74, PDF pp.207-215, Sonnet: alpha velocity and range table, the range-period log-log rule ("reason still obscure"), the life estimates made from range (ionium 200,000 y, uranium II), and the pleochroic halos (sizes, ages, alpha counts per year) as evidence for constant decay over geological time.
6. matterenergy00soddiala-f13420, PDF pp.236-257, Sonnet: the whole of Chapter X (the contrarian read pp.240-250 only). Sun shrinkage (40 miles per thousand years), "mainspring of the universe", the pound of uranium against 100 tons of coal, the 1/10,000,000,000 per year release, the self-bombardment negative test (p.247), the time-standard remark, and the closing energy-and-civilisation argument.
7. interpretationof00sodd-f5dd74, PDF pp.49-57 and 106-132, Sonnet: how Soddy presents the heat measurement (133 cal/h, the experimenters if named, half a grain over sixteen years), Tait's limits on geological time, and his own statement of the rival external-energy hypothesis ("straining at a gnat and swallowing a camel"). These are the C2 passages that rest on digests now.
8. matterenergy00soddiala-f13420, PDF pp.13-42, Sonnet: Chapter I "Physical History". Coal as capital, "energy and wealth are synonymous", the gold "guinea stamp" argument and the exact wording, for C7c and for the link to the later economics.

Recheck

Recheck date: 2026-10-08. Re-read claims.md, facts.json, arbiter.md (including section 10, "QE revision"), primer.md and grep-checked physicist.md, mathematician.md, contrarian.md, deep/interpretationof00sodd-f5dd74.md and digest 07.

Remaining (2): 1. The "generous" and "loose" wording for "nearly a million times" still stands in physicist.md, mathematician.md, contrarian.md and the deep read table (QE correction 3). The correction is recorded in claims.md and the arbiter, so no verdict is affected, but the source files still carry the error. 2. digest 07 p.226 still says "Kelvin's assumption" (QE correction 4). Recorded in claims.md.

Final: OPEN (2 issues)

Orchestrator fix (2026-10-08)

• physicist.md:45: "generous" -> "is stated on Soddy's own basis" (with note).
• mathematician.md:125: "Loose; both bases printed" -> "Not loose: the basis Soddy states; both bases printed" (with note).
• contrarian.md:30: "generous by a factor of 2.8" -> "the basis Soddy states" (with note).
• deep/interpretationof00sodd-f5dd74.md:102 (table row) and :141: "generous" / "holds only loosely" -> Soddy's stated basis (with notes).
• digest/interpretationof00sodd-f5dd74.07.md:29: "Kelvin's assumption" -> "Joly's" (with note).

Final: PASS