Fedderson-Daedalus: Quality Engineering (Step 4b)

Inputs checked: _text/manifest.json, all 15 digests in digest/, facts.json, claims.md, physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. I spot-checked the OCR text in _text/. There is no deep/ folder: this project was reviewed before Step 0 and Step 1a existed, so Step 1 and the reviewers read rendered page images instead.

Page numbers: "PDF p.N" is the page of the scan, which is also the page used by the digests and _text. Book pages of the Proceedings equal PDF pages minus 36. Rich reprint pages equal PDF pages plus 132.

Verdict: MINOR. The two papers that the central claims rest on (Trowbridge and Sabine 1890, Rich 1917) and Trowbridge's 3-page Doppler paper were read in full from page images by at least one reviewer, and the numbers I checked are correct. The defects are stale bookkeeping in facts.json, material the digests now show and the review missed (none of which changes a central verdict), a few labels outside the verdict scale, and arbiter reasoning that the primer has since overtaken. Fix by updating facts.json and claims.md, adding a "QE revision" section to arbiter.md, and updating primer.md.

---

1. Coverage

No document was skipped. There are no unsupported file types.

Defect C1 (facts.json is stale). facts.json says both PDFs have "no text layer" and gives text_available: "none". Step 0 has since OCR'd 405 of 406 and 26 of 27 pages, so full OCR text exists for both. It also says Rich PDF pp.13-19 "were not read". The physicist, mathematician and arbiter later read them. how_read and text_available must be brought up to date, along with the Source gap note in claims.md, the arbiter and the primer.

Defect C2 (bundled PDF listed as one document). STEP 1 says: "If one PDF bundles several documents, list each." The Proceedings volume holds about 22 papers, the Academy's meeting records, and memorials. facts.json gives it one entry and says "Other papers not read beyond their titles". Most of these papers are peripheral botany and chemistry, where the digest is enough. Several are electrical or physical papers relevant to this project, and they need their own entries (see section 3).

2. Shortcuts

• Proceedings, 18 of 406 pages read at Step 1. For the central claims this is acceptable: the two relevant papers were read in full. It is not acceptable that the other physics papers in the volume were neither read nor listed. The digests now cover them and show relevant material (section 3).
• Trowbridge and Sabine's data pages (PDF pp.145-157) were read by one reviewer only. The physicist and the contrarian read only the conclusions (arbiter, source-gap note). Claim 2 (period deviation) is a central experimental claim, and its data rest on one reading. My spot checks of p.151 agree with that reading. A second independent deep read is still warranted, as the arbiter asked under "Open questions".
• Rich: digest-level coverage is not the problem. The arbiter re-read PDF pp.19-22 (reprint pp.151-154) to settle the Cu-Zn heading dispute. That was done correctly.
• No deep/ files exist. This is expected because the project predates Step 1a. All [read] tags refer to page images and are in reviewer files rather than deep/ files. That is acceptable.

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

Ordered by importance to this project and to the reader.

1. Rich's own photographs contradict a naive LC period, and nobody checked them. Rich gives L = 25 mH (coil of coppered iron wire), C = "an eighth to a half microfarad, depending on the duration of the charge" from glass panes whose "absorption was very bad", a film speed of about 42,000 mm/s, and a measured spark spacing of 4.1 mm [read: Rich PDF p.14]. 4.1 mm at 42,000 mm/s is 98 μs between sparks, so the period is T ≈ 0.20 ms if there is one spark per half-cycle (0.10 ms if one per cycle). The arbiter's "check" for claim 1 computes T = 2π√(0.025 × 1.25×10⁻⁷) = 0.351 ms and does not compare it with these data. Matching the observed period needs an effective C of about 0.039 μF (or 0.0097 μF), several times below the low-frequency value [QE calc]. The likely cause is the glass dielectric's absorption, which gives a lower effective capacitance at high frequency, plus skin effect in the iron-wire coil. Both are mainstream, but this is a measured case in the folder of the "capacity depends on rate of charge" effect that Trowbridge and Sabine invoked. It belongs in claim 1's discussion and in the primer.
2. Contemporary evidence on rate-dependent capacity (dielectric absorption) in the same volume. This is directly relevant to claim 2 ("dielectric viscosity"), which no review placed in context: - Safford and Hofman, "Telephonic Specific Inductive Capacity" (PDF pp.15-18): capacity "varied with rate" of charge. They also give a table of telephonic specific inductive capacities (petroleum 1.6 ... glass 4.6, water 6.8). - Nott, "comparative examination of dielectrics" for condensers (PDF pp.281-296): residual discharges continuing for hundreds of minutes, capacity varying with charging time, and "When as short a time as 0.002 sec. was tried ... the discharge exceeded the charge, but the results were too irregular" [read: OCR PDF p.290]. That last line is a classic over-unity-looking artefact of dielectric absorption and worth a sentence in the primer. Trowbridge and Sabine's own argument (PDF pp.147, 154-155) is that glass recovers from strain slowly, so slow-method capacities should not be used for fast discharges. That is the real, accepted phenomenon of dielectric absorption in solids. Their error was extending it to air. The arbiter and the primer should make this distinction explicit. It also strengthens the "instructive" part of the claim 2b verdict.
3. A. E. Dolbear's paper "On Change of Form Affecting a Magnetic Field" (PDF pp.309-315; the digest misreads the author as "A. Emerson Dodge"). Dolbear flexes magnetised iron rings and finds that a change of form induces currents. He reports sympathetic "magnetic sound" between rings and finding the compass points with a telephone and an iron rod. From this he speculates on ring-shaped magnetic atoms, waves in the ether outside a closed ring magnet that "would not necessarily give rise to electro-magnetic effects directly", and the "Vortex Ring theory of atoms, of the properties of which I have so often spoken to the Academy" [read: OCR PDF pp.314-315]. The meeting records also show Dolbear speaking on vortex rings (PDF p.344 digest; OCR line 14801). Dolbear is a recognised name in alternative-science and ether literature. This is the most "alt-science" content in the folder, and facts.json, the reviews and the primer all miss it. It needs a facts.json entry, and probably a claim or an "also in the folder" note.
4. Joule memorial (PDF pp.384-389): a documented case of resistance to a correct result. The memorial reports that members of the Royal Society called Joule's mechanical-equivalent paper "nonsense from beginning to end ... and as a body they refused to publish it in their Transactions", with W. Thomson the only friendly voice [read: OCR PDF p.387]. It gives Joule's values (838, 781.8, 772, 772.55 ft·lb). This matters for the Step 5 Curator ("a documented story of how the claim was resisted"), and it is a mainstream example the primer could use as a counterweight. It is currently absent everywhere.
5. Trowbridge and Sabine details that the review does not carry: - Spark lengths of 0.4 cm and 1.8 cm from a glass jar gave the same period (PDF p.152). This is the authors' actual evidence against the spark-resistance explanation, and it is stronger than "one sentence" suggests. A 4.5-fold change in spark length with no change in period weighs against a resistance-dominated shift. The reviewers and the primer should weigh it. - They compared a large condenser (19 cylinders, 15.1-60.4 cm diameter, 92 cm high) with the small one, and found the periodicity less marked with the large, slower condenser (PDF pp.147, 154). This is conclusion 3 [read: OCR PDF p.159], which no claim covers. - Their discussion of Boltzmann and Klemenčič (√K = n for simple gases, deviations for complex molecules; PDF pp.155-156). This is the 1890 experimental basis for their doubt about air and should be mentioned under claim 2b. - They suggest using the LC period to measure v, the ratio of units (speed of light), after Lorenz (PDF p.146). - Minor: the mirror flew apart twice, and speed was timed to 1/315 s.
6. J. J. Thomson's 1890 energy argument is quoted by Trowbridge as his motivation (PDF pp.228-229). The mathematician notes this, but the arbiter and the primer present the energy budget as their own refutation. The primer should credit that the decisive calculation was already published in 1890.
7. Funding and meeting records (PDF pp.335-344). Rumford Fund grants of $500 to Trowbridge "for metallic spectra" (the Doppler paper is headed as Rumford-funded), Livermore on the law of gravitation at molecular distances, Whiting on heat from alternating current through the human body, and Dolbear on vortex rings. These are peripheral, but the Rumford funding answers the contrarian's standing question "was funding blocked?" (it was not).
8. Rich background not carried forward. Schuster, Hemsalech, Milner and Royds found that metal vapours leave different electrodes at unequal velocities (PDF pp.7-8). Rich gives this as prior support. The gap was varied from 0.1 to 4 mm and "many different specimens ... and several different shapes" were used (PDF p.16), but the tables do not say which gap applies to each. That is an uncontrolled variable, and the arbiter's "What would settle it" section should name it.
9. Peripheral physics in the volume, needing facts.json entries only: Cross on microphone electrode excursion (PDF pp.101-110); Barus on a high-pressure screw compressor and Tait gauge hysteresis (pp.129-144); Cross and Hales on magneto-telephone transmitters (pp.270-281); Barus on the pressure coefficient of galvanic cells (pp.297-298); Huntington on the Stutsman County meteoric iron (pp.266-270).

4. Accuracy spot checks (against the OCR text)

The digests contain errors that are now known. Later steps should not copy them: - "A. Emerson Dodge" should be A. Emerson Dolbear. - Lorenz "v = 800 × 10⁶ m/sec" should be 300 × 10⁶ m/s (an OCR error). - The Rich digest repeats the printed 6.2.

5. Process

• Source tags: present and generally honest throughout. [memory] items carry a confidence level. Calculations are marked [my calculation] or [calc] and reproduced in example.py.
• Web checks: incomplete, but reported honestly. The search quota ran out, and the arbiter says so and refuses to treat "not found" as "not replicated". These medium-confidence [memory] facts that verdicts lean on remain unchecked:
• air has negligible dielectric loss at MHz (claim 2b);
• ion mobility in air of 1-2 cm²/(V·s) (claim 3);
• the spark-channel voltage-current behaviour (claim 2);
• Guthe's 1901 coherer series (cited from Rich, so effectively [read]);
• whether Rich 1917 was ever cited, replicated or refuted. The "Never properly tested" verdicts for claims 4 and 5 depend on this. The Kelvin-1853 attribution stays [memory, high]; no verdict depends on it.
• Verdict scale: mostly applied correctly. Two exceptions:
• "Established by calculation" (arbiter, claim 3) is not a scale label. Use Established for the physical statement and Refuted by calculation for the opposite hypothesis, both of which the arbiter already argues.
• "Not supported by tabulated work functions; mechanism open" (claim 5) is not a scale label either. The nearest fit is "Plausible, untested" for "the order reflects ease of electron emission", with the work-function mismatch noted. Choose a label explicitly.
• Arbiter against primer: arbiter item 7 calls a 50 Ω spark resistance "plausible" as the explanation. Primer section 5C (with example.py section C) shows that a constant resistance of that size would damp the oscillation to about 10⁻⁷ within 8 periods, inconsistent with the 8-9 periods photographed. The primer also shows that the stray "capacity of wire, 200" listed on PDF p.151 accounts for a fifth to two-thirds of the 2.9% lengthening. The arbiter has neither point. The primer is right, and the arbiter must be revised to match.
• Primer: it follows all nine STEP 4 sections in order. It suits the reader profile: symbols defined, CGS units explained from scratch, worked numerical example, check-yourself questions with answers. It repeats the source gap. Defects: it repeats the stale "no text layer" statement; it overstates Rich's "anode" wording (spot check 5); and it omits the missed items 1-4 in section 3.
• example.py: runs cleanly with .venv/bin/python and reproduces every primer number I compared: 3.0453 μs, 3.132 μs mean, 2.9%, 38/50 Ω, 463 eV, 0.534 Å, 6.83, Welch t 9.8, z = 6.4/10.4/−0.6.
• Physicist inconsistency: physicist.md says the only printed slip in Rich is Fe+Bi- 2.21, but the 6.2 slip in Cu-Fe is real (spot check 1). The arbiter has already resolved this, so physicist.md needs no rewrite.

6. Remediation actions

Deep reads (most important first)

1. Daedalus-Fedderson-109-c85345, PDF pp.145-159, Opus. Do a second independent full read of Trowbridge and Sabine "Electrical Oscillations in Air" from page images. Questions: - Transcribe all period tables for the large and small air condensers and the glass jar. - Does the 0.4 cm against 1.8 cm spark-length test exclude a resistance-driven period shift? - What exactly do Figs. 1-7 show? Is the "periodicity" a systematic pattern beyond the reading error? - What do they say about the Boltzmann and Klemenčič results, and about the magnetic-field experiment?
2. Oscillatory-spark-discharges-between-unl-f2ef0a, PDF pp.6-24, Sonnet. Verify every Rich table against page images. Questions: - What gap length applies to each table? - Is there any Bi-Zn data? - Does the observed 4.1 mm spark spacing match his stated L and C, and what does it imply for effective capacitance? - List every place he says "anode", "cathode", "negative" or "positive" for the favoured electrode.
3. Daedalus-Fedderson-109-c85345, PDF pp.309-316, Sonnet. Read A. E. Dolbear, "On Change of Form Affecting a Magnetic Field". Questions: - What apparatus, magnitudes (ampere-turns, deflections) and controls did he use? - What exactly does he claim about ether waves outside a closed ring magnet and about vortex-ring atoms? - Are the induced currents explained by ordinary magnetostriction (Villari effect) and flux change?
4. Daedalus-Fedderson-109-c85345, PDF pp.281-296, Sonnet. Read Nott on condenser dielectrics. Questions: - Extract the residual-discharge and charge-time data, and the "discharge exceeded the charge" observation with its conditions. - Summarise the dielectric-absorption evidence relevant to Trowbridge and Sabine's "dielectric viscosity".
5. Daedalus-Fedderson-109-c85345, PDF pp.384-389, Sonnet. Read the Joule memorial. Questions: - Transcribe exactly the account of the Royal Society's rejection (dates, which paper, what was said, when it was published) and the mechanical-equivalent values. - Note what the memorial says about how established scientists resist new fundamental conceptions.
6. Daedalus-Fedderson-109-c85345, PDF pp.228-230, Sonnet. Read Trowbridge's "Motion of Atoms" as page images at 300 dpi. Questions: - Read the two oscillation periods (".0000003"?) and the sensitivity fraction ("1/100 of a wave length"?). - Is a line-width measurement given?

Not deep-read (the digest is enough; record them in facts.json): Safford and Hofman, PDF pp.15-18 (capacity varies with charge rate); the meeting records, PDF pp.335-344 (Rumford grants, Dolbear on vortex rings, Livermore); the peripheral physics papers listed in section 3 item 9; and all botany, chemistry and memorial pages.

Web checks (arbiter's budget, once the quota allows)

• Citations or replications of D. L. Rich, Phys. Rev. 10, 140 (1917): search Google Scholar, the APS "cited by" listing, and Loeb's Fundamental Processes of Electrical Discharge in Gases.
• The dielectric loss of air at MHz frequencies.
• Ion mobility in air.
• A. E. Dolbear's biography and his vortex-atom writings.
• The Joule / Royal Society rejection, as told in a modern history.

File changes

• facts.json:
• Update both entries' how_read to "page images; full OCR text now available (Step 0)" and set text_available to "yes (OCR)".
• Remove "pages 13-19 not read" from the Rich entry.
• Split the Proceedings entry into one entry per paper. Full entries are needed for Trowbridge and Sabine; Trowbridge "Motion of Atoms"; Dolbear; Nott; Safford and Hofman; and the Joule memorial. Brief entries are enough for Cross; Cross and Hales; Barus (2 papers); Huntington; the meeting records; and one grouped entry for the botany and chemistry papers and the other memorials.
• Add Rich's apparatus numbers (L, C, 4.1 mm spacing) and the Cu-Fe 6.2/6.83 slip.
• claims.md:
• Update the Source gap note: the text is now available, and the remaining gap is Feddersen's own Annalen papers.
• Add, as a peripheral claim or an "also in the folder" note, Dolbear's claim that a change of magnet form produces currents and ether waves (vortex-ring atoms).
• Optionally add the Trowbridge and Sabine conclusion that periodicity depends on the condenser size.
• arbiter.md, add a "QE revision" section:
• (a) Replace the "50 Ω plausible" reasoning in item 7 with the primer's damping argument and the stray-capacity result. The claim 2 verdicts stay the same.
• (b) Add the Rich 4.1 mm period check under claim 1. The verdict stays the same, but note that effective C is rate-dependent.
• (c) Relabel claim 3 to Established, with the opposite hypothesis Refuted by calculation, crediting J. J. Thomson 1890.
• (d) Give claim 5's mechanism a scale label.
• (e) Soften "Rich's prose says anode" to "mixed usage".
• (f) Weigh the glass-jar spark-length test (deep read 1) and dielectric absorption in solids (Nott, Safford) under claim 2b.
• (g) Add an assessment of Dolbear's claims after deep read 3 and its web check.
• (h) Add the gap-length control to "What would settle it".
• primer.md:
• Update the Source gap paragraph.
• Fix the "anode" wording in section 4.
• Add a short box on dielectric absorption in solids versus air, with Nott's "discharge exceeded the charge" as an instructive artefact.
• Add the Rich period check as a second small worked example; extend example.py to match.
• Credit J. J. Thomson's 1890 energy argument in Step E.
• Add Dolbear and the Joule rejection story to section 7 or 9.
• example.py: add the Rich spacing-to-period and effective-capacitance calculation:

dt = 4.1 mm / 42,000 mm/s = 97.6 μs C_eff = T² / (4π² × 0.025 H) ≈ 0.039 μF (taking T = 2 dt)

• Digests: no rewrite needed, but in later files correct "A. Emerson Dodge" to Dolbear and "800 × 10⁶" to 300 × 10⁶.

After remediation, a Sonnet QE recheck appends a "Recheck" section to this file.

Recheck

Checked against the current facts.json (19 entries, valid JSON), claims.md, arbiter.md (with its "QE revision" section), primer.md, example.py (runs; section H reproduces 97.6 us, 195 us, 0.0386 uF) and deep/ (two files).

Defects

• C1 (stale facts.json, text_available, "pp.13-19 not read"): FIXED. Entries say OCR text is available, and the Rich entry records the full read.
• C2 (bundled PDF listed as one document): FIXED. The Proceedings volume is split into per-paper and grouped entries, including Dolbear, Nott, Safford and Hofman, the Joule memoir and the meeting records.
• Shortcut, Trowbridge and Sabine data read by one reviewer only: FIXED. deep/Daedalus-Fedderson-109-c85345.md is a second independent full read, and the arbiter notes two independent reads.
• Missed item 1 (Rich 4.1 mm spacing and effective C): FIXED. It is in the arbiter (QE revision item 7), the primer (worked example, check question and answer) and example.py section H.
• Missed item 2 (dielectric absorption, Nott, Safford): FIXED as coverage. It is in the primer box and the arbiter. The Nott and Safford passages still rest on OCR quotes and digests (see remaining issue 2).
• Missed item 3 (Dolbear): PARTLY FIXED. It has a facts.json entry, a primer note and an arbiter assessment, but the requested deep read was not done (remaining issue 2).
• Missed item 4 (Joule memorial): PARTLY FIXED. It is in the primer and arbiter, but the deep read was not done (remaining issue 2).
• Missed items 5-6 (Trowbridge and Sabine details; Thomson 1890): FIXED. The deep read corrected the QE's own misreadings (1.3 cm, not 1.8 cm; chronograph 1/500 s; no data for the large condenser). Thomson 1890 is credited in the arbiter and the primer.
• Missed items 7-9 (Rumford funding, Rich gap-length and Bi-Zn controls, peripheral papers): FIXED. The arbiter and facts.json carry them.
• Verdict scale labels (claim 3, claim 5 mechanism, "anode" wording): FIXED. They are relabelled in the arbiter table, and the primer evidence table matches.
• Arbiter against primer (50 ohm spark resistance): FIXED. Item 7 is marked SUPERSEDED and the damping argument is carried over.
• Claim 2a and 2b verdicts: changed to "Tested, not supported" in the QE revision, with reasons shown. This is consistent with the scale.
• Source gap statements (claims.md, arbiter, primer): FIXED. The text is now described as available, and the Feddersen gap is stated.
• Digest errors (Dodge, 800e6): FIXED in later files. No occurrence of "800" is left as a speed value, and "Dodge" appears only as a note of the digest's error.
• example.py: runs cleanly and includes the new Rich calculation.

Remediation actions

1. Deep read 1, Trowbridge and Sabine PDF pp.145-159, Opus: DONE.
2. Deep read 2, Rich PDF pp.6-24, Sonnet: DONE. It resolved the gap-length, Bi-Zn and "anode" questions.
3. Deep read 3, Dolbear PDF pp.309-316, Sonnet: NOT DONE. The arbiter says so.
4. Deep read 4, Nott PDF pp.281-296, Sonnet: NOT DONE.
5. Deep read 5, Joule memorial PDF pp.384-389, Sonnet: NOT DONE.
6. Deep read 6, Trowbridge Doppler PDF pp.228-230: replaced by the arbiter's own page-image read. It confirmed 3e-7 s, 2.4e-6 s and the 6 mm gap. The sensitivity fraction is read as 1/100 at medium confidence and is flagged as unverified. Accepted as a partial fix.
7. Web checks (Rich citations and replications, air loss at MHz, ion mobility, Dolbear biography, Joule rejection): NOT DONE. The arbiter says none were run, and the related verdicts are labelled as unchecked.
8. New side finding, not a QE defect: the small condenser's stated geometry gives about 11,900 cm if 18 gaps act in parallel, against the quoted 5317.9 cm. It is unresolved but flagged.

Remaining issues

1. Web checks for the medium-confidence [memory] facts and for any citation or replication of Rich 1917 were not done. The "Never properly tested" verdicts for claims 4 and 5 rest on this.
2. Deep reads of Dolbear, Nott and the Joule memorial were not done. Statements about them rest on QE OCR quotes and digests. None of them changes a central verdict.
3. The Doppler sensitivity fraction (1/100 of a wave length) and the small-condenser geometry discrepancy remain unverified beyond one page-image read.

No central verdict is affected by the remaining issues.

Final: OPEN (3 issues)

Recheck 2

Checked against the current facts.json (valid JSON), claims.md, arbiter.md (including "QE revision" and "Round 2"), primer.md and deep/ (two files; the Trowbridge-Sabine file now also holds sections for the Doppler page, Joule memorial, Dolbear, small-condenser geometry and Nott).

Remaining issue 1: web checks (Rich 1917 citations or replications, air loss at MHz, ion mobility, spark-channel behaviour, Dolbear biography, Joule rejection and the 1843 against 1848 date)

REMAINS. The arbiter states that no web checks were run in either revision. The dependent verdicts are labelled as unchecked ("no replication of Rich found" means not searched), so the reporting is honest and no central verdict is affected. The user has paused this item for now, pending a quality review and a plan.

Remaining issue 2: deep reads of Dolbear, Nott and the Joule memorial

FIXED. Round 2 of the arbiter reports full reads of all three, and its corrections to the earlier OCR-based item 10 are carried into the table. Nott: the residual trains are minutes, not hundreds of minutes; "absorption" is the reviewers' interpretation; the paper concerns solids, not air. Dolbear: magnetic and acoustic apparatus with no unmagnetised-ring control; the ether and vortex-atom ideas are suggestions without a test. Joule: the quotations are accurate but secondhand, and the memorial's date is internally inconsistent. A read of Safford and Hofman is still undone, but this was a digest-level item in the first QE and does not affect any verdict.

Remaining issue 3: Doppler sensitivity fraction and small-condenser geometry

• Doppler fraction: REMAINS. It has now been read from the page image twice. The result is "1/100 of a wave length" at about 85% confidence, and 1/1000 is not fully excluded. The reference unit (wavelength or line width) is ambiguous in the paper. The arbiter flags this as open and no verdict depends on it, because the 3e-7 s, 2.4e-6 s and 6 mm values were confirmed.
• Small-condenser geometry: FIXED as an error. The height is 30.47 cm (not 80.47), so the "11,900 cm" side check is withdrawn. A real 15-17% gap remains between the printed geometry (about 4,540 cm) and the quoted 5317.9 cm. A 0.075 cm zinc thickness would reconcile it, but this is an inference. The arbiter weakens the claim 2a basis accordingly (Tested, not supported, on the quoted C only), and the discrepancy is stated honestly. This is a documented limit of the source, not a defect in the review.

Remaining after this recheck: web checks (1, paused by the user) and the unresolved Doppler sensitivity fraction (3). No central verdict is affected by either.

Final: OPEN (2 issues)