QE (Step 4b): Gift-Stephan

QE agent, Opus. Date 2026-10-03. Inputs read: _text/manifest.json, all 6 digest files, facts.json, claims.md, physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. Spot checks were made against the extracted text in _text/, plus one page image of the 2001 paper (p.7), rendered to the scratchpad. Nothing under ~/MEGA was modified.

Page convention note: the digests and manifest number pages by PDF page. The reviewers usually use the printed page number. The 2018 and 2009 PDFs both start with a ResearchGate cover, so the two differ by one. For example, the Shtyrkov "We await confirmation" sentence is on printed p.18 = PDF p.19.

Verdict: MINOR

The five central claims and their headline verdicts are well supported. The reviewers went past Step 1's sampling: they rendered the key 2001 pages and the arbiter recomputed the decisive numbers. No central verdict is likely to change. The defects are (1) four parts of two papers that rest on sampling or a single reviewer's reading, (2) some material in the digests that no later file mentions, the main item being a second anomalous experiment (Marinov), and (3) one verdict-scale inconsistency in claim 4. All of these can be fixed by updating facts, arbiter and primer.

---

1. Coverage

No document was skipped. Coverage passes.

2. Shortcuts

Digest-only coverage is fine for peripheral material. Here every document is central to at least one claim, so these shortcuts are counted as defects:

1. 2009 magnetism paper, middle and late sections (PDF pp.5-8). Step 1 sampled them by keyword search. The mathematician and arbiter both say the covalent-bond sections and the conclusion were "only skimmed". The verdict on claim 5 rests on a calculation using eq 3.13, so it is robust. But the paper's own supporting arguments in those sections were never assessed: Earnshaw's theorem, Greenspan's results for other molecules, the HCl example, the "rule of two" and Barut's resonances. Defect (minor).
2. 2009 gauge construction (PDF pp.3-5, eqs 3.1-3.8). The equations were read "in context" from a garbled text layer. Only the contrarian viewed a page image, and only one page. The arbiter rejected the mathematician's headline gauge argument and carried forward "invariance not shown". That label rests on reconstructed equations. Defect (minor).
3. 2001 gauge construction (PDF p.4, with pp.3 and 5). No reviewer re-read p.4. The mathematician says outright that his shift-symmetry and "matter-antimatter repel" points are "my reconstruction". The arbiter's verdict for claim 4b ("Plausible, untested") and its caveat on 4d (composition dependence, "conditional on the coupling as written") both depend on what eqs 4-8 actually say. Defect (minor).
4. 2006 paper equations. Page images were never rendered. The Roemer and Doppler derivations behind claims 2 and 3 were reconstructed from prose. The key point that "Gift's final results equal SR's v/c" is well supported by the prose (digest p.5-9 [read]). The contrarian's point that Jupiter's and the Sun's motion are "easily separated" in a parenthesis was not checked against the equations. Defect (minor).
5. Shtyrkov, Figure 5 (2018, PDF pp.19-20). This is the only claimed new anomalous result in the project. Its figure was never viewed. All reviewers call it "Never properly tested" and the primer lists it as "genuinely open". Nobody looked at what the plotted data actually show: the axes, any error bars, or how a velocity was extracted. Defect (minor), and the most important gap for the reader's "open questions".

Primary sources that are not in the folder (Shtyrkov, Greenspan, Aspden, Marmet, Kelly, Hatch, Miller and Shankland, DI Herculis) are declared correctly as a source gap in every file.

3. Missed material (from the digests; absent from facts, claims, reviews, arbiter and primer)

1. Marinov's "absolute motion" measurement. The 2006 paper says Marinov measured the Earth's absolute velocity relative to the ether with a modified Sagnac (coupled-mirror) experiment and got 303 ± 20 km/s. Gift compares this with the CMB dipole value of 390 ± 30 km/s [read: 2006 PDF p.12, text line ~540; ref 44, Marinov 1980, Gen. Rel. Grav.]. It appears in no later file. It is an experimental anomaly claim of exactly the kind the contrarian is asked to weigh. It is also relevant to the "30 vs 370 km/s" argument the reviewers use against the CMB-ether idea, because Gift cites Marinov as matching the CMB scale. Its replication status needs a web check.
2. Weisskopf quotation (2006 p.12): the CMB "provides an absolute frame of reference". Gift uses it as authority for the preferred-frame claim. The arbiter's point that the CMB is a cosmological rest frame, which SR accepts, answers it, but the quotation is not mentioned.
3. Earnshaw's theorem argument (2009 PDF p.6 [read: text line 326]). Gift argues that, because purely electrostatic systems cannot be stable, a non-electric (magnetic) force is needed for the solidity of matter. This is a substantive physics argument and nobody answers it. The standard reply is that Earnshaw's theorem is classical and static, and that quantum mechanics (the uncertainty principle plus Pauli antisymmetry) gives stability [memory, high]. This belongs in the arbiter's claim 5 section and makes a good primer teaching point.
4. Greenspan's wider results (2009 PDF p.6): the attractive-electron assumption was also said to give correct vibrational frequencies and molecular diameters for Li₂, B₂, C₂ and N₂ (digest; isotopes as printed). The reviewers discuss H₂ only.
5. Chemistry examples (2009 PDF p.7-8): the HCl bond, the "rule of two", Gillespie on the repulsive core, Moore and Luder on the limits of the Schrödinger equation. Peripheral, but facts.json should at least list them.
6. Strong-force extensions. The 2009 paper (PDF p.9) says an isospin gauge rotation U(x) may give massive rho-mesons, and U(t) gives pions unified with gravitons. The 2001 Table II also claims a pion and strong-force unification; only the physicist and mathematician mention it. This is a recurring "relax covariance, get a particle" pattern and is worth one line in facts and primer.
7. Marmet and Couture on light deflection (2001 p.7 [read: page image, this QE]). Right after quoting 1.75" "confirmed to within 1%", Gift cites Marmet and Couture as arguing that deflection measurements have large systematic errors. Nobody records this hedge.
8. Historical positives Gift lists (2006 p.10-11; 2018 p.5): Sagnac 1913, Michelson-Gale 1925, Bilger 1995 ring laser, and Allan et al. (around-the-world GPS Sagnac). Gift also claims the 2006 paper is the first to claim ether-drift detection from orbital motion. The arbiter covers Sagnac generally, but Allan et al. and the priority claim are not recorded.
9. Other 2018 citations not followed: Wang ("light speed depends on the observer's motion relative to ECI"), Su's local-ether model, and Gift's earlier GPS papers (2010-2014, refs 32, 34, 49). These were Gift's own prior publication of the same results. The primer's further-reading list gives only the four folder papers.
10. Digest error, for the record: the 2018 digest says "In a 1950 Nature article, Stachel". The text has Stachel, "1905 and all that", Nature 433, 215, 2005 [read: ref 41]. No later file repeats the error.

4. Accuracy spot checks

All six spot checks passed. Minor point: the mathematician's correction 6 ("text ends at about p.23 plus references", questioning 27 pages) is not a real error. The PDF has 27 pages: cover, 22 text pages, references.

5. Process

• Source tags. Present and honest throughout. Calculations are tagged [calc]. Second-hand material (Shtyrkov, Greenspan, Aspden) is labelled as such every time.
• Web checks. The search budget was exhausted in Step 2. Reviewers and arbiter used WebFetch on Wikipedia and NIST, and every failed check is stated ("check failed"). This meets the rule's fallback ("or say the check failed"). The medium/low-confidence items that verdicts depend on and that are still unchecked: whether Shtyrkov was replicated, Greenspan's standing, DI Herculis, the review standard of Progress in Physics, and now Marinov. These should get web checks in remediation.
• Verdict scale. The arbiter is mostly careful and consistent. It rejects "Internally inconsistent" for circularity and "Refuted by calculation" for reinterpretation, and gives reasons. One inconsistency: claim 4b gives the scalar-graviton theory "Plausible, untested", which the scale defines as "consistent with known physics". In the same section the arbiter says the theory breaks Lorentz invariance by assumption, does not show its gauge invariance, and (4d) predicts composition-dependent free fall excluded by about 10 orders of magnitude, conditional on the coupling as written. Those cannot both stand. Either the 4d condition is confirmed (deep read 2) and 4b changes to "Refuted by calculation" or "Not testable as stated (limits only)", or 4b is reworded to say only that the Newtonian limit is reproduced. The primer evidence table repeats this.
• Arbiter quality. Good. It rejected the mathematician's wrong "pure gradient gives zero field" argument with correct reasoning. It reconciled the different energy figures (3.6e-4 eV at a₀ against 5.4e-5 eV at 1 Å), and it recomputed the 6GM/c²R light-bending integral itself.
• Primer. It follows the nine STEP 4 sections in order and has answers at the end. It is written to the reader profile: gauge invariance, spin-statistics, test theories and the equivalence principle are introduced in plain language, and symbols are defined. The worked example uses real numbers. Step 8 corrects the arbiter's v/c scale to v²/c² for two-way tests, which is right. Small gaps: §9 and the opening list do not give the 2006 paper's real title and journal ("The Relative Motion of the Earth and the Ether Detected", J. Sci. Explor. 20(2):201-214, 2006 [read]). The opening says "from the file name and Step 1".
• example.py. Runs cleanly with .venv/bin/python. All primer §5 numbers are reproduced: v = 356.29 m/s, 3.964 ns, c ∓ v to the digit, the SymPy SR check prints 0, round trip 9.42e-15 s, Sagnac 414.8 ns, side checks 0.876, 1.751 and 2.627 arcsec, and 3.62e-4 and 5.37e-5 eV.

---

6. Remediation actions

Deep reads (most important first)

1. APRExistenceoftheEther2018-3b34f8, PDF pp.18-20, Sonnet, page images. Shtyrkov section 3.1 and Figure 5: what exactly is plotted (axes, units, error bars, number of points)? How was the "orbital velocity" extracted from the aberration data? Does the figure show the full 29.8 km/s, or a residual fitted to it? Does Gift say what model was subtracted (satellite orbit determination, pointing)?
2. A-Negation-of-Einsteins-General-Theory-of-Relativ-1f31d1, PDF pp.3-6, Opus, page images. Section 4.1 Lagrangian and the time-local gauge step: write out eqs 4-8 exactly. Is the coupling m ψ̄γ⁰ψA with m the constituent fermion mass, so that the source is rest mass only? Is a kinetic term for A given, and is it invariant under A → A + (1/m)dα/dt? Does the paper address antiparticles or the sign of ψ̄γ⁰ψ? This decides arbiter verdicts 4b and 4d.
3. PP-16-03-9f62ec, PDF pp.3-5, Opus, page images. Eqs 3.1-3.13 exactly: the definition of "magnetic charge", the phase transformation, the coupling term (number current, or magnetic moment?), the kinetic term for A_k, and how eq 3.11 links the Dirac current to the magnetisation current. Confirm or correct the arbiter's "not shown self-consistent" wording and the contrarian's "coupling to number current, not moment".
4. Ether-Detected-Gift-Stephen-a93a8e, PDF pp.4-12, Sonnet, page images. The Roemer and Doppler equations (2.1-2.6 and the Doppler set), including how Jupiter's and the Sun's motion are "easily separated". Also the Marinov 303 ± 20 km/s passage and the Weisskopf CMB quotation, with Gift's reference details.
5. PP-16-03-9f62ec, PDF pp.5-9, Sonnet, text plus page images where garbled. Sections 4.1-5: the Earnshaw argument, the Greenspan list of molecules and claimed diameters, Barut's resonances, the HCl and covalent-bond reasoning, and the future-work claims (rho-mesons, pions). Does the paper give any quantitative estimate of magnaton strength at bond distances?

Web checks (arbiter, about 4 calls)

• Marinov's coupled-mirror experiment (1980 and later), its reception and any replication attempts.
• Shtyrkov's aberration papers (NPA proceedings, 2005-2010) and any replication or critique.
• Greenspan's electron-attraction H₂ papers (1990s), their reception and the H₂ vibrational value used.
• DI Herculis apsidal-motion resolution (spin-orbit misalignment, Albrecht et al. 2009) [memory, medium].

Changes to later files

• facts.json: 2006 entry: add the Marinov and Weisskopf passages and the claim of priority for detecting orbital motion. 2009 entry: change how_read once deep reads 3 and 5 are done; add Earnshaw, the wider Greenspan list, HCl and rule-of-two, and the rho-meson/pion future work. 2001 entry: add Table II pions, the Marmet and Couture hedge on light bending, and confirm that p.7 drops GM/h² via h² = ∞ [read: page image p.7]. 2018 entry: list Gift's own prior GPS papers (refs 32, 34, 49) and Wang and Su.
• claims.md: add Marinov as supporting evidence under claim 3 (an anomalous experimental claim, second-hand). Revise the source-gap note: the gap on 2001 pp.6, 7, 9, 10 is closed; p.4 is open until deep read 2 is done.
• arbiter.md: add a "QE revision" section. (a) Resolve the 4b/4d inconsistency from deep read 2. (b) Add Marinov to claim 3's tested-status list, with a tested/replicated status from the web check. (c) Answer the Earnshaw argument under claim 5. (d) Update Shtyrkov from deep read 1, keeping "Never properly tested" unless the figure shows something different. (e) Upgrade the light-bending caveat to [read: page image p.7]. (f) Note the PDF-page vs printed-page convention.
• primer.md: fix the 2006 paper's title and journal in the opening list and in §9. Revise evidence table row 4b to match the arbiter revision. Add Marinov to §7 "genuinely open", or explain it under 3a once checked. Add a short Earnshaw-versus-quantum-stability note to §4 or §7, a good teaching point for this reader. Remove "via the mathematician" from answer 4, since QE confirmed the page.
• example.py: no change needed.

---

Recheck

Recheck agent (Sonnet), 2026-10-03. Read: qe.md, the four new deep/ files (headers and key passages), the QE revision section of arbiter.md, facts.json, claims.md, primer.md. example.py was re-run with the project venv and still reproduces the primer numbers (3.62e-04 and 5.37e-05 eV side checks seen; no error). Nothing under ~/MEGA was touched.

Defects (QE sections 2 and 5)

Missed material (QE section 3)

Remediation actions

Remaining issues

1. 2009 paper pp.5-9 (Earnshaw, Greenspan list, HCl, rule of two, rho/pion) never deep-read; the Earnshaw reply is from [memory, high]. No verdict depends on it.
2. Web checks for Shtyrkov replication and Greenspan's H2 value failed (search budget). These are stated as unchecked in the arbiter and primer.

Both are low impact and already disclosed in the files. They are not defects that could change a verdict, but they were listed by the QE as defects and are not closed.

Final: OPEN (2 issues)

Recheck 2

Recheck 2 agent, 2026-10-03. Read: qe.md (including Recheck), arbiter.md (Round 2 section), the second block of deep/PP-16-03-9f62ec.md, and the files' timestamps. Only qe.md was written. The user's instruction on the remainder: pause for now; the user will review quality and return with a plan.

Final: OPEN (1 issues)