Willams-Crookes: Quality Engineering (Step 4b)
QE agent: Opus. Date: 2026-10-06.
Inputs checked: _text/manifest.json, digest/Sir-William-Crookes-On-Radiant-Matter-4f2b69.01.md (the only digest), deep/Sir-William-Crookes-On-Radiant-Matter-4f2b69.md, facts.json, claims.md, sites.md, all nine notes in research/ (research/files/ is empty), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. I read the whole extracted text (_text/Sir-William-Crookes-On-Radiant-Matter-4f2b69.txt). I also rendered PDF pages 7, 21 and 22 at 200 dpi (scratch copies in /tmp/qe-crookes) and read them myself.
Page numbers: "p.N" means the === page N === marker in the extracted text (PDF page). PDF pp.1-12 are printed pp.13-24 of Popular Science Monthly vol. XVI, and PDF pp.13-22 are printed pp.157-166. PDF p.23 is unnumbered (it would be p.167).
Tags: [read] = the source document; [read: page image p.N] = checked on the rendered page; [calc] = my own computation in the project venv; [memory] = recall, with confidence given.
Verdict: MINOR. Coverage is complete. The single document was digested, then deep-read in full, with page images checked where the OCR failed. My five spot checks against the page images all agree with the deep read. The central verdicts stand: the observations are Established, and the identification of the carriers as gas molecules is Historically superseded. I found four defects that can be fixed by updating facts.json, claims.md, arbiter.md (a "QE revision" section), primer.md and example.py: 1. Every role accepts Crookes's two-beam "mutual repulsion" (Fig. 18) as beam-on-beam electrostatic repulsion. My calculation says that force is far too weak to give the jump he showed. 2. The arbiter's "Never properly tested" verdict on the hot-ring mechanism leaves thermionic emission open. A calculation rules it out at red heat. 3. A demonstrated claim was missed: at the highest vacuum the tube stops conducting, with William Morgan's 1785 experiment as the precedent. 4. There are small accuracy and process slips in the primer and the reviews.
Steps 2-4 do not need to be redone.
1. Coverage
| Item | Status |
|---|---|
| Documents in manifest | 1 (Sir-William-Crookes-On-Radiant-Matter-4f2b69, 23 PDF pages, all OCR, 1 chunk) |
| Digests present | 1 of 1 |
| Deep reads | 1, covering pp.1-23, with page images at pp.7, 21 and 22 |
| Skipped files | none. The source folder holds only this PDF; I listed it, read only. |
No coverage defects.
The digest has four errors, all caught downstream and none carried into a verdict: - the ruby line is given as 6,595 (printed 6,895); - the highest vacuum is given as 1/6,754,000 atm (printed 1/20,000,000); - the molecule count is given as "one quadrillion (10^15)" with "a trillion" left (printed 10^24 and 10^18, in the British long scale); - p.7's pressure table is garbled.
facts.json, the deep read and every review flag these and use the page-image values. This is the process working as designed.
2. Shortcuts
None. Every central claim rests on the full deep read, not on the digest alone. The deep reader read the page images for the three contested numbers. The physicist re-read the dark-space and molecule-count passages himself, and the contrarian rendered p.21 at 300 dpi.
The one real limit is outside the project folder. The Investigator downloaded nothing (research/files/ is empty, and sites.md says downloads "need the owner's go-ahead"). As a result, Crookes's technical papers (the Bakerian Lecture 1879, and "Contributions to Molecular Physics in High Vacua" 1879), Stokes's referee reports RR/8/151-152, Starke 1900 and Thomson 1903 were read by no role. Every file says so honestly. The arbiter's source-gap note and the primer's source note both list them. This is a gap in the record, not a hidden shortcut. It is carried into remediation as an Investigator task (section 7, item I1).
3. Missed material (from my own reading of the digest and the full text)
M1. The non-conducting vacuum (pp.7-8). Defect: a demonstrated claim with no entry anywhere. - Crookes says phosphorescence diminishes beyond about one millionth of an atmosphere "until the spark refuses to pass" [read: p.7]. - In the Fig. 5 potash tube, "When the induction spark is first turned on nothing is visible - the vacuum is so high that the tube is non-conducting". Warming the potash to release "a trace of aqueous vapor" makes conduction start "instantly" [read: p.8]. On cooling, the green glow fades and "the vacuum becom[es] non-conducting" [read: p.8]. - A long footnote quotes William Morgan, Phil. Trans. 1785 (vol. lxxv, p.272), "Electrical Experiments made to ascertain the Non-conducting Power of a Perfect Vacuum". In Morgan's mercury gauge, boiled free of air, "neither the smallest [ray] of light, nor the slightest charge, could ever be procured". He also saw the colour of the light change from green to blue to violet as air leaked in [read: p.7, footnote, partly garbled]. - The Investigator found Crookes's own paper "On Electrical Insulation in High Vacua" (1878; archive.org philtrans02850338) [web: research/2026-10-06-archive-org.md].
claims.md, facts.json key_claims, the three reviews, the arbiter and the primer never treat this as a claim. Morgan appears only in the deep read; "non-conducting" appears in no review file. For a learner it is a key piece of physics. At very low pressure there are too few gas molecules to supply the ions that sustain the discharge, so the breakdown voltage rises again: the low-pressure branch of the Paschen curve [memory, high]. It also bounds claim 1: the "radiant matter" regime has an upper vacuum limit set by the discharge itself.
M2. Chemical absorbents as a way to get past the pump (p.21). Crookes reached his highest vacuum by putting absorbents in the tube: phosphoric anhydride for water vapour, potash for carbonic acid, palladium for hydrogen, carbon and then potash for oxygen. He says this goes "to much higher degrees than I can get by mere pumping" [read: p.21]. This is an early form of chemical gettering [memory, medium on the term's history]. The deep read records it, but no review, the arbiter or the primer mentions it, and it bears on how the 1/20,000,000 atm figure was reached. Add it to facts.json and the primer.
M3. "Three hundred trillions a second" (pp.22-23) can be checked. The mathematician says this figure "is conditioned on an unexplained picture ... I cannot check it". In Crookes's long-scale wording, 300 trillions is 3 x 10^20. Then 1.288 x 10^24 molecules / 3 x 10^20 per second = 4,300 s, or about 72 minutes [calc]. That matches his "before you leave this room". So the figure is self-consistent. With the modern count (3.5 x 10^22) the same rate fills the bulb in about 2 minutes [calc]. This makes a worthwhile one-line addition to the primer's side checks.
M4. Long-scale number words are not explained in the primer. The lecture prints "1,000000,000000,000000,000000 (a quadrillion)" and "a trillion" [read: page image p.22]. In the British long scale these are 10^24 and 10^18. The digest misread exactly this as 10^15 and 10^12. The physicist explains it. The primer says only "digits grouped in sixes". A reader who opens the lecture will meet "quadrillion" and should be told what it meant in 1879.
M5. Low-vacuum against high-vacuum magnetic behaviour (p.17). In a low vacuum the deflected discharge "rises and pursues its path to the positive pole". In a high vacuum the deflected stream "did not recover itself, but continued its path in the altered direction" [read: p.17]. The physicist mentions this; the arbiter and the primer do not. It is a clean way to show the difference between a current-carrying glow column and a free beam. Add one sentence to primer section 4.
M6. Smaller items, for facts.json only. - The unnamed "last year's President" at Dublin (p.4) is William Spottiswoode, President of the British Association in 1878 [memory, high; not needed for any verdict]. - The diamond "shines with as much light as a candle" [read: p.6]. - The pole-recoil reasoning for the electrical radiometer (Fig. 12, p.14) is covered by the physicist and mathematician under claim 5. No action is needed there.
4. Accuracy spot checks (against the extracted text and page images)
| # | Fact | Where used | Check | Result |
|---|---|---|---|---|
| 1 | Ruby line "about 6,895" | facts, claims, all reviews, arbiter, primer | [read: page image p.7] "a wave-length of about 6,895" (text layer: 6,595) | Correct |
| 2 | Highest vacuum 1/20,000,000 atm | all files | [read: page image p.21] the fraction's denominator reads 20,000,000. Cross-check: 0.01 in / 3 miles = 1/19.0 million [calc] | Correct |
| 3 | 1.288252 x 10^24 molecules; 1.288252 x 10^18 at 10^-6 atm; 408,501,731 years | facts, reviews, arbiter, primer, example.py | [read: page image p.22] footnote digits as quoted. 1.28825106 x 10^24 / 10^8 / (365 x 86400) = 408.50 million years [calc; example.py prints the same] | Correct. The OCR "408,501751" is a slip and the deep read caught it. |
| 4 | Radiometer rotation starts at half a millimetre of mercury | contrarian, arbiter, primer table 3.1 | [read: p.15] "At a pressure of half a millimetre this dark space extends to the glass, and rotation commences" | Correct |
| 5 | Dark space "about an inch on each side" in Fig. 1, pressure not stated | physicist, arbiter, primer | [read: p.3] | Correct |
Accuracy defect A1 (primer). Primer section 4, "Priority" bullet, says the focusing cup, hinged cross, paddle-wheel railway and magnetic water wheel were "all built by Gimingham [read: p.12-17]". The lecture credits Gimingham with three of these: the hinged cross (Fig. 9, p.12), the railway paddle wheel (Fig. 11, p.13) and the magnetic wheel (Fig. 17, p.17). It does not credit him with any cup-cathode or focusing tube [read: pp.10, 11, 19]. Remove "the focusing cup" from that list.
Accuracy defect A2 (deep read and facts.json, trivial). Both say PDF pp.1-12 are printed pp.13-25. The page headers show 13-24 [read: p.2 header "14", p.12 header "24"]. Correct to 13-24. The deep read also still refers to "the brief's" wrong numbers; that leftover wording is harmless.
Accuracy defect A3 (primer, untagged overstatements). - Section 1 says Crookes's tubes were "far emptier than anyone before him". This is untagged. Morgan (1785) had a mercury-boiled vacuum (above), and Geissler and Sprengel pumps came earlier. Tag it [memory] with a confidence, or soften it to "among the emptiest of his time". - Section 4 says "He was wrong only about which particles". The primer itself lists three other errors (the dark space, the wheel mechanism, the "fourth state"). Qualify the sentence: "on the particle-versus-wave question, he was wrong only about which particles".
Substantive defect P1: the Fig. 18 "mutual repulsion" (claim 4). All roles accept it without checking it.
What Crookes did: two negative terminals "close together" at one end. When the second is switched on, the first ray "jumps up from its first position, d f, to d g". He concludes the streams act "merely as similarly electrified bodies" [read: pp.18-19].
What the files say: - The physicist and mathematician show that free like charges in parallel give a net repulsion, with the magnetic attraction only (v/c)^2 of the electric push. That logic is right. - The arbiter and primer then rate "mutual repulsion" Established, and primer section 3.4 and Check-yourself Q4 teach it as the explanation. - Nobody estimated whether that force is big enough to give a visible jump.
My estimate [calc], with the reviewers' assumed 10 kV: - A 1 mA beam has line charge I/v = 1.7 x 10^-11 C/m. Its field 1 cm away is E = λ/(2π ε0 r) = 30 V/m. - Over a 30 cm path this gives a sideways deflection of about 0.07 mm (0.7 mm at 10 mA), before any neutralisation by positive ions, which the physicist and mathematician both say would reduce it further. - That is far too small for a jump visible to a lecture audience on the phosphorescent screen.
The much larger effect is the electric field around the second cathode. It is a strong field region (hundreds of volts to kilovolts across the dark space) right next to where the first beam starts, and it pushes the first beam away. Goldstein reported in about 1880 that cathode rays are deflected by a nearby cathode, and that later accounts explain this as electrostatic deflection by the neighbouring cathode's field [memory, medium; needs a web check].
Consequences: - The observation (the first ray moves away) is real and Established. - Crookes's reading of it (beam-on-beam repulsion of charged molecules) is probably wrong as the main cause. - The charge-sign verdict does not change, because Perrin 1895 and Thomson 1897 establish the negative charge independently.
This needs a QE revision to the claim 4 verdict, after the web check and deep read D1. Suggested split: - the first ray is displaced when a second cathode is switched on: Established; - the cause is mutual repulsion of the two streams: Refuted by calculation as the dominant cause (the space-charge force is about 100 to 1,000 times too small), or at least "Plausible, untested" if the arbiter prefers to wait for the geometry from D1.
Primer section 3.4, the section 4 table row "Parallel streams repel", the section 7 "mutual repulsion of two beams" bullet and Q4 and its answer all need rewording. The (v/c)^2 point can stay as correct physics. The new point is that the force it gives is far too small to explain the demonstration.
Substantive defect P2: the hot-ring vanes (claim 5). "Never properly tested" leaves an alternative open that a calculation excludes.
The contrarian raised thermionic emission from the red-hot platinum ring as a possible second mechanism, and the arbiter adopted "mechanism Never properly tested". The primer repeats it as "a possible early sighting of thermionic emission".
Richardson's law gives J = A T^2 exp(-W/kT), with A of about 1.2 x 10^6 A m^-2 K^-2 and a platinum work function W of about 5.6-5.7 eV [memory, high]. At red heat (900-1100 K) this gives J = 2 x 10^-20 to 2 x 10^-14 A/m^2 [calc]. For a fine-wire ring of about 10^-5 m^2, the current is at most 10^-19 A. That is zero for any mechanical purpose.
The lecture also says the ring was heated with the coil "disconnected altogether" [read: p.15], so there was no accelerating voltage. Unbiased and at red heat, the ring is a heat source next to a radiometer. The vanes are mica at 45 degrees, in a tube "exhausted to a very high point", and that is the light-mill (Crookes radiometer) effect, whose thermal explanation the arbiter itself calls accepted [memory, high].
Recommended QE revision: - hot-ring vane spin: observation Established; - thermal (radiometric) mechanism: Established, as the accepted radiometer physics; - a thermionic contribution at red heat: Refuted by calculation (show the Richardson numbers).
Keep the contrarian's proposal (biased against unbiased ring) in the record as the experiment that would show this directly. But the "genuinely open" bullet in primer section 7 should say the calculation already answers it for red heat. Crookes's own reading ("a hot wire will set [radiant matter] in motion" [read: p.15]) stays in the record.
5. Retention check
I compared the Investigator's leads (sites.md "Leads for review" and the nine research notes) and the source's content with arbiter.md and primer.md. I was looking for anything left out because a reviewer judged it negatively.
| Material | Reviewer judgement | In arbiter? | In primer? |
|---|---|---|---|
| Crookes's "now passes as Electrons" (one obituary source) | inventor's own revision, untraced original | yes, described as one documented source, not dismissed | yes, same wording |
| Puluj 1883, "transcendental" characterisation | rival with a stake | yes, stake stated | yes |
| Theosophy/Blavatsky reception | outside science | yes ("keep in the record") | yes |
| Plasma credit for the "fourth state" | retrospective, different definition | yes | yes |
| Hittorf, Goldstein, Varley priority | reduces Crookes's priority | yes (Hittorf, Goldstein); Varley only via reviews | yes, all three |
| Stokes referee reports and the "clarifying experiment" | unread | yes, top open question | yes |
| Science Museum notebooks (radiation) | not suppression | yes, with a full absence-of-evidence analysis | yes |
| Gimingham (IET SC MSS 285, dates discrepancy) | none | yes | ledger yes; dates discrepancy no (minor) |
| Busch fogging quotation; Plücker attribution | unresolved | yes | yes |
| Starke 1900 / Thomson 1903 against Crookes's mechanism | evidence against the claim | yes | yes |
| Tired glass, low/high-vacuum magnetic contrast, non-conduction, Morgan 1785 | none (not judged) | tired glass yes; others no | tired glass yes; others no |
Result: no retention defect. Nothing pertinent was dropped because it was judged negatively. Material against Crookes (Starke, Thomson, the factor of 37 in the molecule count) and material for him (the obituary revision, Stokes's favourable referee reports, the modern Harvard tube) are both kept and stated with their sources. The items missing from arbiter and primer (M1, M2, M5) were never judged; they were overlooked, and they are handled as missed material. The Smithsonian radiometer blog, Dörfel 2003 and Sella 2015 (Chemistry World) appear only in research/google.md. They are peripheral history and their omission is acceptable.
6. Process
| Check | Finding |
|---|---|
| Source tags present and honest | Yes, throughout. [memory] confidence is given each time. The primer uses [computed] for its own numbers and defines it. No [memory] item is presented as [read]. |
| Web checks of medium/low-confidence facts a verdict depends on | Partial. The physicist used 2 of about 6 web calls (one failed), the mathematician 2, the contrarian 2 (one failed). The arbiter used 7 of 8 and retagged the ruby R lines and the Loschmidt number to [web]. Still unchecked, though a verdict leans on them: Thomson 1897's gas-independence of q/m (supports verdict 6a; tagged [memory, medium] by the contrarian, "medium-high" in the primer table); the Edison-effect date of 1880 (supports the hot-ring discussion; see P2); and Goldstein's deflection by a neighbouring cathode (my own [memory, medium] in P1). The cathode-fall product was checked by two roles and failed both times. The arbiter correctly kept the claim 2 verdict off it. Defect PR1 (minor): the reviewers left web budget unused while medium-confidence facts stayed unchecked. |
| Verdict scale applied consistently | Mostly. The contrarian's claim 6 entry ("Established as arithmetic from its inputs; the input is off by about 37") avoids a scale verdict for the count. The arbiter resolved this to "Refuted by calculation". The arbiter's choice of "Historically superseded" over "Refuted by calculation" for the claim 3 carrier identity is argued explicitly, which is good. P1 and P2 above are the two places where the scale should change. |
| Primer follows STEP 4 order and the reader profile | Yes: sections 1-9 in order, answers at the end, symbols defined, OCR and keV defined, Griffiths chapter numbers correct (Ch. 2 electrostatics, Ch. 5 magnetostatics) [memory, high]. Small slips: "CRT Paddle Wheel" (further reading) uses CRT (cathode-ray tube) undefined; plus A1, A3 and M4 above. |
| example.py runs | Yes. .venv/bin/python example.py exits 0 and ends "All checks passed". Its numbers match primer section 5 (3.37 cm, 7.62 m, ratio 226, 5.9 mm, 408.50 My, 37.2 and 39.9, 3.39 x 10^-7 N, β^2 = 0.038). |
| Investigator rules | The 10-minute cap was respected (7.5 minutes). Queries are logged, blocked sites are logged with the reason, and unreached sites are listed. Deviation PR2: STEP I says to save useful documents to research/files/ "when downloadable". None were saved, and the agent cited a need for owner approval. This left the four documents that would settle most open points unread (section 7, I1). |
7. Remediation actions
Deep reads (only one manifest document; slug Sir-William-Crookes-On-Radiant-Matter-4f2b69)
| # | Pages | Question | Model |
|---|---|---|---|
| D1 | pp.18-19 (with page images of Fig. 18) | Describe the Fig. 18 two-cathode tube exactly. How far apart are terminals a and b, and where are they relative to each other and to the screen? Does the first ray's displacement (d f to d g) start at the cathode end or grow along the tube? What does Crookes say about the lower ray? Is any pressure, coil setting or angle given? Purpose: to decide between beam-on-beam repulsion and electrostatic deflection by the second cathode's field (QE P1). | Sonnet |
| D2 | pp.7-8 (page image of p.7 footnote and the pressure table) | Transcribe the Morgan 1785 footnote in full (the OCR is garbled) and the pressure table. Record every statement in the lecture that the highest vacua are non-conducting ("spark refuses to pass", "non-conducting") and the conditions under which conduction returns (QE M1). | Sonnet |
| D3 | pp.14-15 (page images of Figs. 12-13) | Fig. 13 hot-ring apparatus: how far is the platinum ring from the mica vanes, and does Crookes give the exhaustion, the battery or the ring temperature? In Fig. 12, which side of the vanes is mica and which way does the fly turn? Purpose: to support the radiometric reading in QE P2. | Sonnet |
| D4 | p.9 (page image) | Recover the OCR-garbled description of the moderately exhausted bulb A (Fig. 7) and the start of the low-vacuum comparison, so that the low-against-high vacuum contrast can be quoted exactly (QE M5). | Sonnet |
facts.json
- Add key_claims: "At the highest exhaustion the tube becomes non-conducting and the spark refuses to pass; admitting a trace of vapour restores conduction (pp.7-8)" and "Chemical absorbents in the tube (phosphoric anhydride, potash, palladium, carbon) carry the exhaustion beyond what pumping alone reaches (p.21)".
- In the summary, mention the Morgan 1785 footnote, the low/high-vacuum magnetic contrast (p.17) and the "300 trillions a second" figure (about 72 minutes in Crookes's numbers).
- Correct the printed page range to 13-24 and 157-166 (and p.167 unnumbered).
- Optionally note that the 1878 Dublin President was Spottiswoode [memory, high].
claims.md
- Add claim 7: "At the very highest vacua the tube stops conducting (the spark refuses to pass); some residual gas is needed for the discharge (pp.7-8; Morgan 1785 precedent)."
- In claim 4, split "two parallel streams repel" as an observation (the first ray is displaced when a second cathode is switched on) from Crookes's interpretation (mutual repulsion of charged streams).
arbiter.md: add a "QE revision" section
- Claim 4: after D1 and a web check of Goldstein's deflection of cathode rays by a neighbouring cathode, re-rate the repulsion. Observation (displacement of the first ray): Established. "Caused by mutual repulsion of the two streams": Refuted by calculation as the dominant cause, using the space-charge estimate in QE P1 (field of about 30 V/m at 1 cm for 1 mA, deflection about 0.1 mm over 30 cm), or Plausible, untested if the geometry leaves it open. The negative-charge verdict (Established) is unchanged.
- Claim 5: hot-ring mechanism. Change "Never properly tested" to: observation Established; radiometric (thermal) drive Established; thermionic contribution at red heat Refuted by calculation (Richardson's law, QE P2). Keep the biased/unbiased experiment as the direct test.
- New claim 7 (non-conduction): suggested verdict Established. Breakdown needs residual gas, which is the low-pressure branch of the Paschen curve, and Crookes's own 1878 "On Electrical Insulation in High Vacua" covers it (unread). Web-check the Paschen statement and retag it.
- Web-check, and retag or flag: Thomson 1897's same q/m across gases and cathode metals (verdict 6a); the Edison-effect date of 1880.
- Absence of evidence: no change.
primer.md
- Section 3.4, the section 4 table row "Parallel streams repel", the section 7 "What is solid" bullet and Check-yourself Q4 and its answer: keep the (v/c)^2 physics, but add the estimate that space-charge repulsion is far too weak to explain Fig. 18 and that the displacement is mainly the second cathode's field (after D1 and the web check). Rephrase Q4 so the learner computes the deflection.
- Section 4 "Priority" bullet: remove "the focusing cup" from the list of apparatus built by Gimingham (A1). Qualify "He was wrong only about which particles" (A3). Tag or soften "far emptier than anyone before him" (A3).
- Section 5 side checks: add the long-scale note ("quadrillion" = 10^24, "trillion" = 10^18 in 1879 British usage) (M4), and the "300 trillions a second, about 72 minutes" check (M3).
- Section 4 and section 6: add the non-conducting vacuum (claim 7) with the Morgan 1785 precedent (M1), the chemical absorbents (M2) and the low/high-vacuum magnetic contrast (M5).
- Section 6 row 5d and section 7 "genuinely open": replace "possible early sighting of thermionic emission" with the Richardson estimate. Keep the biased/unbiased test as the direct check.
- Further reading: write "CRT (cathode-ray tube)".
example.py
- Add Part 5:
- (a) the field of a 1 mA and a 10 mA beam at 0.5-1 cm, and the deflection of a parallel 10 keV beam over 30 cm (P1);
- (b) Richardson current density for platinum at 900-1300 K (P2);
- (c) 1.288 x 10^24 / 3 x 10^20 s^-1 in minutes (M3).
- Have the primer cite the new part.
Investigator (next run; not part of this QE round)
- I1: with the owner's approval, download and save to research/files/:
- Stokes RR/8/151 and RR/8/152 (Royal Society, 2 pages each);
- the Bakerian Lecture 1879 (archive.org philtrans06253890);
- "Contributions to Molecular Physics in High Vacua" 1879 (philtrans07220971), which should give Fig. 18's numbers, the "laws of magnetic rotation" and dark-space/pressure pairs;
- "On Electrical Insulation in High Vacua" 1878 (philtrans02850338), for claim 7;
- Starke 1900 (Ann. Phys. 3, 101-107). Then run a review update (Steps 2-4b) for claims 2, 4, 5 and 7.
- I2: search for Goldstein's "deflection of cathode rays by a neighbouring cathode" (German: "Ablenkung der Kathodenstrahlen", about 1880) for P1.
Recheck
Date: 2026-10-06. I re-read qe.md and checked the current arbiter.md, claims.md, facts.json, primer.md, example.py and the deep read with grep. I also ran ../../.venv/bin/python example.py.
| Defect | Status | Evidence |
|---|---|---|
| P1 Fig. 18 mutual repulsion | Fixed | arbiter.md splits claim 4 into the observation (Established) and the cause (Plausible, untested). The "QE revision" section and primer 5a-ii/5a-iii carry the space-charge estimate (0.07-0.7 mm at 10 kV). The arbiter's "Plausible, untested" is the option the QE allowed. |
| P2 Hot-ring thermionic emission | Fixed | arbiter.md row 5: thermal drive Established, thermionic contribution at red heat Refuted by calculation. primer section 7 now says calculation answers it. example.py 5b prints Richardson values (at most a few 1e-14 A at red heat). |
| M1 Non-conducting vacuum (claim 7) | Fixed | claims.md claim 7, facts.json key_claims, arbiter.md row 7, primer table rows and 7a. |
| M2 Chemical absorbents | Fixed | facts.json and primer line 106. |
| M3 "300 trillions a second" check | Fixed | facts.json, primer line 161, example.py 5c (71.6 min). |
| M4 Long-scale number words | Fixed | primer line 159. |
| M5 Low/high-vacuum magnetic contrast | Fixed | primer line 103. |
| M6 Smaller items | Not checked, optional (facts.json only). | |
| A1 Gimingham and the focusing cup | Fixed | primer line 115 credits Figs. 9, 11 and 17 only. |
| A2 Printed page range 13-24 | Partly fixed | facts.json says 13-24. The deep read header (deep/Sir-William-Crookes-On-Radiant-Matter-4f2b69.md, line 3) still says 13-25. Trivial, with no effect on any verdict. |
| A3 Overstatements | Fixed | primer line 15 now says "among the emptiest vessels of his time". Line 116 now reads "On the particle-versus-wave question, he was wrong only about which particles" and lists the other errors. |
| PR1 Unchecked medium-confidence facts | Partly addressed | The arbiter row 7 notes a partial web check. A first-party check of Goldstein's deflection by a neighbouring cathode failed and is tagged [memory, medium]. This is disclosed and does not change a verdict. |
| CRT undefined | Fixed | primer line 250 now reads "CRT (cathode-ray tube)". |
| example.py | Pass | Exits 0, ends "All checks passed", and includes Part 5a-5c. |
Remaining: A2 (the deep read's page range, 13-25 should be 13-24).
Final: OPEN (1 issues)