Arbiter: Gift-Stephan (Step 3)

Inputs read in full: claims.md, facts.json (via the reviewers' corrections), physicist.md, mathematician.md, contrarian.md. Source tags: [read] from the documents (as reported by the reviewers, who rendered page images), [web] URL, [calc] my own recomputation (Python, .venv), [memory] with confidence. Web budget: the search tool was exhausted, so checks were done by fetching Wikipedia pages; failed checks are said so.

Source gap (repeated from Step 1, as updated by the reviewers)

• The 2001 gravity paper (Physics Essays 14(4)) has a garbled text layer. Step 1 read pp. 1-5 and 8 from images. Physicist, mathematician and contrarian each rendered pp. 6, 7, 9, 10 (mathematician: 6, 7, 9 only). That part of the gap is now closed. I did not re-render any page myself; statements about those pages are the reviewers' [read: page image] reports, and where only one reviewer saw something I say so.
• The 2009 magnetism paper (PP-16-03.PDF) has damaged equations. Sections up to about 4.2 were read by two or three reviewers; the covalent-bond sections and conclusion were only skimmed. Equations were read from context.
• NOT in the folder, so everything about them is second-hand through Gift or [memory]: Shtyrkov's aberration data, Marmet/Kelly/Sato GPS claims, Wallace's 1969 radar test, Greenspan's simulations, Aspden's 1980 force law, Hatch 2004, Miller/Demjanov/Galaev/Munera papers, the DI Herculis analysis. To obtain: those primary papers (esp. Shtyrkov, Greenspan, Aspden), and clean journal PDFs of the 2001 and 2009 papers.
• Plainly: conclusions about Shtyrkov (claim 2), Greenspan (claim 5), the Aspden force law (claim 4) and DI Herculis rest on Gift's own account and [memory], not on the primary sources.

Corrections to Step 1 (consolidated, all three reviewers agree unless noted)

1. claims.md claim 4: the 2001 paper does NOT derive perihelion precession or light bending from its field equation. Eqs 7-10 give only Newton's law. Precession comes from an imported force law F = (1 + 3v^2/c^2) G M m / r^2 attributed to Aspden; bending and redshift use a heuristic photon "mass" E/c^2, and the paper says the theory "by itself cannot predict the outcome" [read: p.6-7 images, all three reviewers]. The paper says radar delay, orbit decay and DI Herculis await "completion of our investigation". Binary-pulsar decay is asserted "qualitatively", with no calculation.
2. claims.md claim 1: "as shown by GPS clocks" overstates. The 2018 paper contains no GPS data set. Its "measured" times are the CCIR/ITU synchronisation formula, built on c in the Earth-centred inertial (ECI) frame. Only Kelly's 414 ns and Marmet's data are quoted second-hand [read].
3. claims.md claim 3: the 2018 paper concludes there are "multiple preferred frames carried along by the Earth and the Sun" (ECI for rotation, solar-system barycentre for orbit), not one ether, and cites Hatch's "clock bias" reading as a rival [read]. The CMB-as-ether suggestion is only in the 2006 paper [read].
4. Gift himself writes of Shtyrkov "We await confirmation by others" [read: 2018 p.18].
5. Omitted from Step 1: Table II (2001, p.9): gravity "strength" G m^2/(hbar c) ~ 3e-42, which matches neither proton (5.9e-39) nor electron (1.75e-45). I recomputed the proton value: 5.905e-39 [calc]. Minor; the mass used is not stated.
6. Typo in 2001 eq 15: the perihelion formula uses "m" (planet) where the Sun's mass M is meant [read, physicist and mathematician].
7. 2009 Greenspan figures: Gift quotes 2.20e14 Hz (repulsive model) vs 1.38e14 Hz "measured". The NIST H2 harmonic constant omega_e = 4401.2 cm^-1 [web: https://webbook.nist.gov/cgi/cbook.cgi?ID=C1333740&Mask=1000, fetched by the mathematician; I did not re-fetch] is 1.32e14 Hz (4401.2 x 2.998e10); the fundamental is about 1.25e14 Hz [memory, medium]. So 1.38e14 is about 5% off the standard value; whether that is Greenspan's definition or an error is unknown without his paper.

---

Numbers I recomputed where reviewers differed or leaned on one number

Mathematician's light-bending claim, which no one else checked. Gift's orbit equation (14) is u'' + u = GM/h^2 + 3GM u^2/c^2. The mathematician says Gift drops the GM/h^2 term by setting "h = infinity", but for a photon h = cR is finite. I redid the integral by hand: with h = cR the right side is (GM/c^2R^2)(5/2 + (3/2)cos 2phi). The particular solution is u = cos(phi)/R + (GM/c^2R^2)(5/2 - (1/2)cos 2phi). Setting u = 0 at phi = pi/2 + eps gives eps = 3GM/(c^2R), so total deflection 6GM/(c^2R), or 2.63 arcsec for the Sun [calc]. Keeping only the 3 cos^2 term gives 4GM/(c^2R) (GR's value); keeping only the Newtonian term gives 2GM/(c^2R). This confirms the mathematician's algebra. The caveat: it depends on the mathematician's reading of p.7 that the term was dropped by an "h = infinity" step; I did not view the page. Treat as [read: mathematician only] plus [calc].

Mathematician's gauge point on the 2009 paper: rejected. The mathematician argues that because the magnaton gauge shift A_k -> A_k + d_k alpha is a pure gradient, the field strength is zero and the construction "never generates a magnetic field". That is wrong as stated. B = curl A is exactly what is invariant under a gradient shift; this is the same gauge freedom as ordinary QED, and the magnetic field is not generated by the gauge transformation but by the source term. I do not carry this item into the verdict. The surviving, defensible points are: (i) the kinetic term for A_k is added by hand and its invariance is not shown; (ii) the field equation "box A = source" is the Lorenz-gauge form of Maxwell, so it is not invariant under a time-dependent or arbitrary shift unless the residual-gauge condition holds (this holds in QED too, since the full Maxwell equations carry the gauge-fixing); for the 2001 scalar field, a shift A -> A + d(alpha(t))/dt does not leave box A = source invariant unless box of alpha-dot is zero, so invariance is not shown (physicist and mathematician agree; not shown, not proved false).

---

Claim 1: Light speed is c - v eastward, c + v westward relative to the Earth's surface (GPS, range equation, time transfer, GPS-clock Michelson-Morley)

Where reviewers agree. All three: the algebra is right; the 414 ns Sagnac delay is real, standard, and equally predicted by special relativity (SR); the argument is circular because the c in ECI is assumed and c - v is the arithmetic consequence; the one-way speed depends on a synchronisation convention [web: https://en.wikipedia.org/wiki/One-way_speed_of_light, cited by physicist and contrarian; I did not re-fetch]; Gift's own cited cavity limits (dc/c < 1e-17) are first-order-blind and show no anisotropy; Hatch, whom Gift cites, reads the same data as clock bias.

Where they differ. Physicist: weak part Established, strong part "Not testable as stated" (circular but not "internally inconsistent"). Mathematician and contrarian: strong part "Internally inconsistent (circular)" plus "Not testable". The mathematician adds that Gift uses two different frames (ECI and barycentric) each with isotropic c, and that the GPS-clock Michelson-Morley drops the second-order term, so it cannot discriminate.

Which is stronger. Circularity is a flaw in inference, not a self-contradiction among the paper's own equations, definitions or units. "Isotropic c in two frames" is exactly SR's postulate, so it is not self-contradictory, though it sits oddly with "ether". Under the scale I apply consistently ("Internally inconsistent" means contradicts itself) the physicist is right: the strong form is Not testable as stated, with the circularity noted as a logical defect. The mathematician's observation that eq 17 is first order only because two legs are timed by separate clocks (a round trip by one clock has no first-order term) is correct and important. It is the physicist's "refinement".

Tested status (contrarian's distinction). Neither "tested and failed" nor "dismissed without testing": no test is possible, because ether and SR give identical readings under a different synchronisation convention [web: https://en.wikipedia.org/wiki/Test_theories_of_special_relativity, "experimentally equivalent to special relativity, independent of the chosen synchronization"]. The Sagnac measurement itself has been done and replicated for a century (Michelson-Gale 1925, ring lasers, GPS operations).

Strongest case for Gift (contrarian, endorsed). The one-way speed of light is a convention; textbooks that say "c is the same in all directions for all observers" state a convention plus two-way measurements. Gift is right there. He is wrong in saying this "detects" anything.

Single decisive test. None exists in principle for the one-way speed. What would settle the real disagreement (convention versus physics) is an experiment sensitive to a first-order dependence that is not a closed-path or synchronisation effect. None has been shown. Practical check: any one-way anisotropy of size v/c ~ 1.5e-6 would show in modern resonator experiments at 1e-17; none does.

Final verdict (split). - Measured delays (Sagnac, GPS clock-sync difference): Established (and equally predicted by SR). - "These show a physical c - v / c + v and therefore an ether": Not testable as stated; the inference is circular.

Claim 2: Earth's orbital motion is detectable from apparatus at rest on Earth (planetary radar, Shtyrkov, Roemer, stellar Doppler)

Agree. Roemer delay, stellar Doppler, aberration and radar ranging are established and reproduced by SR; Gift concedes his end results equal SR's v/c [read: 2006]; the radar derivation assumes c in the barycentric frame and returns c +/- v at the antenna (same circularity); aberration for 29.8 km/s is 20.5" [calc]; Shtyrkov is a single-author, unreplicated, second-hand result that Gift himself says awaits confirmation.

Differ. Mathematician rates the radar reading "Refuted by calculation (reinterpretation)" and "Internally inconsistent as proof". Physicist and contrarian: Not testable as stated. The mathematician adds a strong kinematic argument: for a co-moving source and observer the light-time displacement cancels the aberration (planetary aberration), and the first-order Lorentz theorem (which Gift quotes) says no annual first-order signal is expected in SR or in an ether theory; so Shtyrkov's result, if real, contradicts the first-order cancellation, and Gift's statement "aberration is independent of the motion of the source" misses that. The contrarian and physicist also note: CMB-as-ether requires Earth moving at about 370 km/s [memory, high; Gift quotes 390 +/- 30 km/s, read], yet Roemer-type effects show 30 km/s, so the 2006 paper's two statements conflict. Gift's "multiple preferred frames" (ad hoc) escapes this.

Which is stronger. A reinterpretation of data is not a calculation that rules a claim out, so "Refuted by calculation" is too strong for radar; Not testable as stated is correct. The mathematician's cancellation argument is standard kinematics [memory, high], and it is why Shtyrkov is a surprising claim for both camps. I could not web-check Shtyrkov (search budget exhausted): replication status unknown to me; not verified.

Tested status. Roemer/Doppler/aberration: tested for centuries, consistent with SR. Shtyrkov: never properly tested (one data set, Kazan 1998-99, per Gift; no independent replication I can find; not dismissed, simply not taken up).

Strongest case. First-order effects are the sensible place to look for a preferred frame, and Shtyrkov's measurement would be cheap to replicate with a geostationary beacon and a dish (contrarian).

Single decisive test. Replicate Shtyrkov: track a geostationary beacon (two sites if possible) through a year, subtract the standard orbit-determination and pointing models, and look for a 20-arcsec seasonal term. SR predicts none; an Earth-carried ether in Gift's sense predicts about 20". One year of data.

Final verdict (split). - Roemer, Doppler, aberration, radar ranging as physics: Established. - Reading them as ether detection: Not testable as stated (same predictions as SR; for the CMB version the 30 vs 370-390 km/s mismatch is a further problem). - Shtyrkov co-moving aberration: Never properly tested (unreplicated; standard kinematics predicts no effect; source not read).

Claim 3: A cosmic ether exists, light-speed invariance and SR are invalid, Michelson-Morley was the wrong test (FLL contraction), the CMB is its radiation

Agree. The framework is Lorentz ether theory (generalised Galilean transformation with Fitzgerald-Lorentz-Larmor contraction). It is mathematically equivalent to SR for all measurable outcomes [web: https://en.wikipedia.org/wiki/Test_theories_of_special_relativity: the Robertson-Mansouri-Sexl framework with exact relativistic time dilation and length contraction "is experimentally equivalent to special relativity, independent of the chosen synchronization"; Michelson-Morley + Kennedy-Thorndike + Ives-Stilwell fix the Lorentz transformation]. Gift concedes the equivalence and then calls SR "invalidated". Gift cannot have FLL contraction hiding the ether in second-order tests and first-order results (identical to SR's) refuting SR. Modern cavity limits near 1e-17 (Gift's own refs) [read as cited].

Differ. Physicist: SR-invalid is "Tested, not supported" plus "Internally inconsistent" (non sequitur). Mathematician: "Internally inconsistent" first, then "Tested, not supported". Contrarian: "Tested, not supported", and denies the "internally inconsistent" label for the framework as such (it is consistent as a theory) but agrees the inference fails.

Which is stronger. The framework is consistent; the inference from "first-order results equal SR's" to "SR is invalid" is a non sequitur. That is a logical-structure flaw within Gift's papers (he states the results equal SR's and then says SR is refuted), so it is fair to flag, but the headline verdict is the experimental one: the claim that an ether has been detected and SR falsified is not supported by any controlled test. I keep "Tested, not supported" as the headline and list the non sequitur as a secondary note, not a second verdict, to apply the scale consistently with claim 1.

Tested status (contrarian's distinction, with a web check). - Second-order and Kennedy-Thorndike/Ives-Stilwell family and modern resonator tests: tested, null, by many independent groups [memory, high; framework itself [web] above]. - Miller 1925-33: the contrarian marked as [memory, low] that Shankland's reanalysis was contested. I checked: Shankland et al. 1955 attributed Miller's signal to statistical fluctuation and local temperature conditions, and later work (Joos got about 1/50 of Miller's size) did not reproduce it [web: https://en.wikipedia.org/wiki/Dayton_Miller]. The page does not support a live dispute, though it records that Miller's results remain discussed. DeMeo's dissent (cited by Gift) not evaluated; Demjanov, Galaev, Munera, te Winkell: not in the folder, no independent replication known to me [memory, low; check failed]. So: tested and reanalysed, not "dismissed without testing". Whether the Shankland reanalysis was fair is the one soft spot; the web summary says the signal was smaller than the recording resolution, which is a statistical (not a political) argument. - The "first-order and one-way" claims: not testable (see claims 1-2).

Strongest case. Lorentz ether theory is an honest, empirically equivalent alternative; "no ether" is a methodological choice (simplicity) rather than an experimental result; textbook phrasing "Michelson-Morley disproved the ether" overreaches (contrarian; endorsed).

Single decisive test. None can separate an ether made undetectable by construction from SR. What can fail is a detectable preferred frame: any improved cavity, clock-comparison or Ives-Stilwell test (already at about 1e-17 for cavities) finding anisotropy or a Kennedy-Thorndike signal. Existing data: null.

Final verdict (split). - "An ether has been detected and SR is falsified": Tested, not supported (secondary note: the argument is a non sequitur, since the cited results equal SR's). - "A preferred frame exists but is undetectable by construction" (weak form), "MM was the wrong kind of test", "CMB is the ether's radiation" (a relabelling; the CMB dipole gives a cosmological rest frame at about 370 km/s, which SR accommodates): Not testable as stated; the weak form is also Historically superseded but instructive (Lorentz/Poincare 1900s).

Claim 4: The equivalence principle (EP) is false; GR is invalid; replace it with a Newtonian-type scalar-graviton gauge theory (Newton, redshift, precession), no black holes, dark matter or gravity waves

Agree. The radiation argument is the old charge-in-a-gravitational-field paradox; its standard resolution is frame dependence of radiation, and the EP is a local statement [web: https://en.wikipedia.org/wiki/Paradox_of_a_charge_in_a_gravitational_field: a supported charge does not radiate in its rest frame but does in the free-fall frame; Boulware 1980: the radiation goes into a region inaccessible to the supported observer; "The equivalence principle is preserved for charged particles"]. This upgrades the physicist's and mathematician's [memory, medium] to [web], and the contrarian's similar statement. The scalar theory reproduces Newton's law and first-order redshift easily; precession and bending are imported (Aspden law; photon mass); the coupling is to rest-mass density and is not Lorentz invariant (admitted by Gift); Nordstrom's pure scalar theory gives zero bending and about -7"/century instead of +43" [web: https://en.wikipedia.org/wiki/Nordström's_theory_of_gravitation, fetched by physicist and mathematician; I did not re-fetch]. Gravitational waves: Hulse-Taylor decay agrees with GR quadrupole radiation, observed/predicted 0.9983 +/- 0.0016 [web: https://en.wikipedia.org/wiki/PSR_B1913%2B16, cited by contrarian]; GW170814 favoured tensor over scalar or vector polarisation [web: https://en.wikipedia.org/wiki/GW170814, cited by contrarian]. A spin-0 graviton predicts scalar waves, the disfavoured case. MICROSCOPE: EP holds to about 1e-15 [web: https://en.wikipedia.org/wiki/MICROSCOPE_(satellite), the 2017 first results; the page does not give the 2022 final number, so the final figure stays [memory, medium] at the 1e-15 level].

Differ. 1. "Redshift and Newton": physicist says trivially reproduced (Established as limits); contrarian says Plausible, untested as a gauge theory; mathematician says Established but not in dispute. Arbiter: these are Newtonian and weak-field GR results that any theory reproduces; as a derivation from the gauge theory the first-order redshift uses the photon's gravitational mass E/c^2 (the equivalence of gravitational and inertial mass, which is what the paper denies; mathematician), so it is not a clean consequence of the field equation. 2. Composition dependence (physicist only): the source is m psi-bar gamma^0 psi (rest mass of constituents), so nuclear binding energy would not gravitate; iron-56 binding is about 8.8 MeV per nucleon (about 0.94% of mass, 8.79/931.5 [calc, binding energy per nucleon [memory, high]]), so test bodies would fall differently at about 1e-3 against bounds of 1e-13 to 1e-15. Only the physicist made this argument; the contrarian agrees the source is rest mass only. Caveat I add: this depends on the theory's coupling being to constituent rest masses as written; a different definition of the source (including binding energy) would remove the argument. It is therefore Refuted by calculation only conditional on the paper's coupling as written. 3. Light bending: contrarian says the paper's 1.75" comes from a borrowed relativistic term and that Newtonian photon mass gives 0.876"; mathematician shows that Gift's own terms treated consistently give 6GM/c^2R = 2.63", not 1.75" (I verified the algebra above; dependent on his p.7 reading). These are compatible: Newtonian 0.876" (one term), GR 1.75" (the 3GM u^2/c^2 term alone), 2.63" (all terms). The paper's 1.75" arises only after discarding a term. Either way, "derived the bending" fails. 4. "Internally inconsistent" for the rocket argument: physicist (framing assumes a Newtonian gravitational force to refute a theory denying one) and contrarian (treats the two frames by different rules) agree in substance; mathematician says the Larmor argument is a "known subtlety", not a refutation, and labels it Tested, not supported. Arbiter: the Larmor argument is not an experiment; it is a theoretical argument that has a published resolution [web above], so the EP claim stands as "Tested, not supported" (the EP, in its weak form, to about 1e-15) and the flaw is noted, not given a separate "Internally inconsistent" label, except for the gauge-construction point below. 5. Gauge construction: physicist (kinetic term added by hand, not shown invariant; non-covariant coupling "allowed by assumption"), mathematician (field equation not invariant under the time-local shift; antiparticles would repel because psi-dagger psi changes sign [his inference, paper pages not all seen]), contrarian (asserted, not derived). All agree the construction does not establish what it claims.

Tested status. The EP and GR have been tested extensively by independent groups and replicated (torsion balance, lunar laser ranging, MICROSCOPE, Mercury, light deflection, Hulse-Taylor, LIGO since 2015 [memory, high; the polarisation result is [web] above]). Gift's own theory has never been tested by anyone; the paper defers its own further tests. Not suppressed: published in Physics Essays after peer review; criticisms are physics, not process (contrarian). DI Herculis apsidal motion was not checked by anyone (physicist: check failed; contrarian: [memory, low] that spin-orbit misalignment resolved it); my own check: not done, budget exhausted. Treat as unverified.

Strongest case. GR has real open issues (singularities, quantisation); the EP is a postulate; the radiation question is subtle and was treated carefully in the literature; a retarded-Newtonian approach merits a calculation (contrarian, endorsed).

Single decisive test. Compute, in Gift's own scalar theory with the paper's coupling, the differential acceleration of test masses of different nuclear binding energy; or the orbit of Mercury and light deflection solved properly from the field equation without importing Aspden's law. For experiment, LIGO's polarisation results already disfavour scalar-only radiation; a future test would be additional detectors showing scalar polarisation, which would contradict the current GW170814 evidence.

Final verdict (split). - "EP is false": Tested, not supported (weak EP to about 1e-13 to 1e-15; the radiation argument is resolved by frame dependence [web]). - Scalar-graviton gauge theory giving Newton and first-order redshift: ~~Plausible, untested~~ REVISED by QE: Internally inconsistent (the paper's own kinetic term is not invariant under its own time-local shift; the invariant version gives no waves). Newton and first-order redshift are reproduced only as limits (see QE revision). - Same theory "giving" perihelion precession and light bending: Refuted by calculation as a derivation (imported force law; pure scalar gravity gives zero bending and retrograde precession [web, Nordstrom]; Gift's own terms give 2.63", not 1.75"). - Composition independence of free fall under the theory's coupling: Refuted by calculation (REVISED by QE: no longer conditional; the coupling is confirmed as written [read: page image p.5-6]; size 1e-2 to 1e-3 for a nucleon source against bounds of 1e-13 or better). - "No gravitational waves of Einstein's kind": Tested, not supported (Hulse-Taylor, LIGO tensor polarisation). - "No black holes, no dark matter": no argument or calculation offered by Gift and none assessed by the reviewers; not independently checked: unknown.

Claim 5: Magnetism is a separate interaction mediated by massless spin-1 "magnatons", explaining the Pauli exclusion principle and chemical bonding

Agree. The field equation (eq 3.12) is the magnetostatic Poisson equation, already contained in Maxwell/QED; the coupling psi-bar gamma_k psi A_k is the spatial half of the ordinary QED coupling j.A (the contrarian adds: it is a coupling to the fermion number current, not to the magnetic moment, despite the text); eq 3.13 is the classical dipole-dipole result, and a 1/r^3 energy is the dipole formula, not what exchange of a massless vector between charges would give (1/r). The strength is far too small to be the bond or the "Pauli force": 3.6e-4 eV at the Bohr radius (Coulomb 27 eV) and 5.4e-5 eV at 1 angstrom (Coulomb 14.4 eV, bond about 4.5 eV [memory, high]) [calc]; equal to Coulomb only near 2e-13 m (contrarian; I did not recompute that figure, but (hbar/2mc)^2/r^2 = 1 gives r = hbar/2mc = 1.9e-13 m, which is right). Pauli exclusion follows from antisymmetry (spin-statistics) without any force [memory, high]. Greenspan's simulations are a classical fit with an assumed attraction, not a test of a magnetic mechanism, and are second-hand through Gift. No magnaton has been seen; no direct test is proposed or on record.

Differ. 1. Verdict label. Physicist: "Refuted by calculation" plus "Internally inconsistent" (redundant with Maxwell). Mathematician: "Internally inconsistent" (gauge construction), "Never properly tested" (the particle), "Refuted by calculation" (bonding). Contrarian: "Internally inconsistent" (coupling to number current, not moment, but identified with magnetisation current) and "Refuted by calculation" (bonding). 2. The mathematician's headline gauge argument (a pure-gradient shift gives zero field strength) is wrong, as explained above, and is not carried forward. His other points (non-invariance of the field equation, an unconstrained third polarisation once Lorentz covariance is dropped) are standard QFT concerns [memory, high] and stand as "not addressed by the paper". 3. Is the "magnaton" a new particle or a relabelled photon? The physicist and contrarian say the equations are Maxwell's, so nothing distinguishes it. The mathematician says "never properly tested" for the particle. Arbiter: if it is identical to the photon's magnetic part, then the claim is not new; if it is different (non-covariant, extra mode), then no experiment has looked for it, which is "never properly tested" in the weak sense, but the paper gives no distinct prediction, so a test cannot even be specified.

Which is stronger. The calculation-based refutation of the explanatory claims (bond, Pauli) is clear and uses the paper's own eq 3.13; all three reviewers support it. "Internally inconsistent" is better stated as "not shown self-consistent": the coupling, kinetic term and invariance are asserted or omitted, and the text describes a magnetic-moment coupling that the written Lagrangian does not contain (contrarian, one reviewer). Because no contradiction between the paper's own equations is demonstrated beyond that mismatch, I do not apply "Internally inconsistent" as a headline.

Tested status. The new particle: never properly tested (no experiment designed for it; dark-photon and fifth-force searches are not searches for this object [memory, medium]). Greenspan: not a controlled test; not replicated in the folder. Not suppressed: published in Progress in Physics (accepted in about 10 weeks [read]; the journal's review standard is [memory, medium/low] light); simply not taken up.

Strongest case. The opening puzzle (the sign of spin-spin magnetic interactions in different pictures) is a real teaching point; a "Pauli force" is an effective, not fundamental, force; Greenspan's classical simulations (bond electrons with assumed attraction fit vibration data) deserve an independent check, including the 1.38e14 versus 1.32e14 Hz discrepancy above (contrarian; mathematician).

Single decisive test. One calculation: with the paper's own eq 3.13, compute the singlet-triplet energy splitting of H2 at the equilibrium bond length (0.74 angstrom) and compare with the experimental bond energy of about 4.5 eV. The dipole term is of order 1e-4 eV, five orders short [calc]. For the experiment side: a search for a Lorentz-violating massless vector coupled to spin, which would already be constrained by existing precision limits (about 1e-17) and the g-2 agreement [memory, medium/high].

Final verdict (split). - A new, separate magnaton interaction that explains bonding and the Pauli principle: Refuted by calculation (about 4-5 orders too weak at atomic distances; its field equation duplicates Maxwell's). - The magnaton as a distinct particle: Never properly tested; no distinct prediction is offered, so it is hard to test. - Greenspan "experimental confirmation": Never properly tested (a fit, not a test of magnatons; not read). - Classical magnetism rewritten as a Poisson equation: Established, but not new.

---

Corrected claim list

1. The rotating Earth has a measured Sagnac delay (414.8 ns around the equator [calc]); Gift reads the east-west synchronisation difference as a physical one-way light speed of c - v / c + v, but this reading follows from assuming c in the ECI frame and is not a measurement. Verdict: delay Established; ether reading Not testable as stated.
2. Roemer, Doppler, aberration and planetary radar all show the Earth's orbital motion; Gift reads them as ether detection, and cites Shtyrkov (single author, unreplicated) for an aberration of about 20" with source and observer both fixed to the Earth. Verdicts: effects Established; ether reading Not testable as stated; Shtyrkov Never properly tested.
3. Gift's ether (a set of preferred frames, with FLL contraction) is observationally equivalent to SR and cannot refute it; "SR is invalid" is Tested, not supported. The CMB-as-ether suggestion is a relabelling.
4. The scalar-graviton gauge theory reproduces Newton's law and the first-order redshift only as limits, and is Internally inconsistent (QE revision: its kinetic term is not invariant under its own gauge shift, so gauge invariance and propagating waves cannot both hold) and Refuted by calculation for composition independence of free fall (binding energy does not gravitate). It does not derive perihelion precession or light bending (imported force law and photon mass; own terms give 2.63", not 1.75"). The EP argument rests on a radiation paradox with a standard resolution. Gravitational-wave denial is Tested, not supported; black-hole and dark-matter denial is unargued.
5. "Magnatons" are a relabelling of the magnetic part of the photon field, with a non-covariant postulate; their strength cannot account for bonding or Pauli exclusion; they have not been tested. (QE revision: the 'magnetic charge' is defined as a divergence of a single moment and then used as a nonzero constant, which is a genuine internal inconsistency for that definition; the gauge construction as a whole stays 'not shown self-consistent'. Round 2: the full 2009 paper contains no coupling value, energy or force estimate, no bond lengths for any Greenspan molecule, and cites Earnshaw only as one supporting sentence for the repulsive core; no verdict changes.)

What would settle it

Open questions

• Shtyrkov: does the 20-arcsec seasonal signal exist? Not read; not found replicated; web check failed (search budget exhausted).
• Greenspan: why 1.38e14 Hz as the "measured" H2 frequency when omega_e = 4401.2 cm^-1 gives 1.32e14 Hz (cause unknown without his paper)?
• Was Miller's data fairly reanalysed? Wikipedia's account supports Shankland, and Joos obtained about 1/50 of Miller's size, but DeMeo's counter-argument (cited by Gift) was not evaluated by anyone.
• Which mass is used in Table II (3e-42 matches neither proton nor electron)?
• Section 2018 and the 2006 paper's Aspden force law (2001 eq 12) and Wallace's 1969 radar test: not read; second-hand.
• DI Herculis apsidal motion: not checked; the contrarian recalls a spin-orbit misalignment resolution [memory, low].
• (Closed by QE) 2001 pp.3-10 and 2009 pp.3-5 were re-read from page images; see QE revision. The 2009 PDF pp.5-9 (covalent-bond sections) were then deep-read in QE Round 2; see Round 2. Greenspan's primary papers remain unread.
• Final MICROSCOPE number (2022): web page did not give it; the 1e-15 level is [web] for the 2017 result, the final figure [memory, medium].
• Whether the positronium hyperfine splitting (about 8.4e-4 eV, [memory, medium]; contrarian) supports the claim that magnetic spin-spin effects are of milli-eV order: check failed, not needed for the verdict.

Source summary

[web] pages fetched this step: https://en.wikipedia.org/wiki/Paradox_of_a_charge_in_a_gravitational_field; https://en.wikipedia.org/wiki/MICROSCOPE_(satellite) (2017 first result only); https://en.wikipedia.org/wiki/Dayton_Miller; https://en.wikipedia.org/wiki/Test_theories_of_special_relativity. Web pages cited by reviewers and not re-fetched by me: One-way_speed_of_light; Sagnac_effect; Nordström's_theory_of_gravitation; PSR_B1913+16; GW170814; NIST H2 webbook. Failed checks: Shtyrkov replication (search budget exhausted), positronium hyperfine value (page does not quote it), Mansouri-Sexl page (404; replaced by Test_theories_of_special_relativity). [calc]: Sagnac, aberration, dipole and Coulomb energies, light-bending integral, G m_p^2/(hbar c).

---

QE revision (after QE verdict MINOR, 2026-10-03)

Inputs added: qe.md and four new deep reads in deep/ (2001 pp.3-10 Opus, 2009 pp.3-5 Opus, 2006 pp.4-12 Sonnet, 2018 pp.18-20 Sonnet), all from page images. Web budget: WebSearch was exhausted, so checks were done with WebFetch on Wikipedia. Source tags as before.

Page convention. The digests and manifest use PDF page numbers; reviewers often used printed numbers. 2001: PDF = printed minus 318. 2018 and 2009: the PDFs start with a cover, so PDF = printed plus 1 (Shtyrkov "We await confirmation" is printed p.18 = PDF p.19). The QE placed the 2001 Lagrangian on PDF p.4; the deep read shows it is on PDF p.5 (eqs 4-6) and p.6 (eqs 7-11).

Changed verdicts

Claim 5 additional findings (2009 deep read, pp.3-5)

• Fermion-part invariance of eqs 3.6/3.7 is correct [calc]. The arbiter's earlier rejection of the mathematician's "pure gradient gives zero field" argument stands.
• The kinetic term is never written. A Klein-Gordon term or a gauge-invariant (dA/dt)^2 - (curl A)^2 term gives eq 3.8 only in a fixed gauge and for static sources (the divergence of the source must vanish) [calc]. The static equation 3.12 is Maxwell's magnetostatics. Three polarisations would remain unconstrained.
• Contrarian's "coupling to number current, not moment": partly confirmed. The coupling is nu times the Dirac vector current; by the Gordon decomposition this carries a Dirac spin moment nu*hbar/2m (wrong for the neutron, Gift's example, -1.913 nuclear magnetons [memory, high]) plus a convection current, so it is a charge-type minimal coupling, not the moment coupling the text describes. Eq 3.11 (J_m = curl M) is placed next to 3.10 and never linked to the Dirac current; it is not used again.
• Eq 3.13 (F = 3 mu0 mu1 mu2 / 4 pi r^4, repulsive) holds only for the side-by-side geometry; coaxial parallel dipoles attract with twice the force [calc]. The p.5 conclusion "parallel-spin electrons repel" silently picks the repulsive geometry; averaged over directions the dipole-dipole energy is zero. No nu value or energy estimate is given in the paper.
• Earnshaw's theorem (QE gap 3). Gift (2009 PDF p.6) argues that purely electrostatic systems cannot be stable, so a non-electric force must give matter its solidity (Round 2 correction: the paper's text is one sentence, that the magnetic interaction in the repulsive-core mechanism is 'supported by Earnshaw's theorem', not a full argument for atomic stability). The standard reply: Earnshaw's theorem is classical and static; quantum mechanics (the uncertainty principle plus Pauli antisymmetry) stabilises atoms and gives matter its size, with no extra force [memory, high]. Also, the magnaton strength is 4-5 orders below the Coulomb energy at atomic distances [calc, earlier table], so it could not supply the stabilising force anyway. Verdict for the Earnshaw argument: a valid classical theorem, misapplied to quantum matter. It does not change claim 5.
• (Closed in Round 2: these were deep-read from the 2009 PDF pp.5-9; see Round 2.) Previously thin: Greenspan's wider list (Li2, B2, C2, N2 vibrations and molecular diameters), HCl, the "rule of two", rho-meson and pion future work (isospin gauge rotation) from 2009 PDF pp.6-9 were not given a separate deep read; they are peripheral and assessed from the digest only. The pattern ("relax covariance, get a particle") also appears in 2001 Table II (pions via SU(2), unification asserted, "Gm^2/(hbar c) = 3e-42" which equals the proton-electron product 3.2e-42 [calc], not proton squared).

Claim 2: Shtyrkov, Fig. 5 (2018 deep read, PDF pp.18-20)

• [read: page image p.18] Fig. 5 plots "orbital ether drift velocity, km/s" (axis 0 to 50) against "time, days" (about day 180 to 830, 25 March 98 to 23 Dec 99): about 25-30 open circles, no error bars, no fit curve, no uncertainty anywhere in the text. Two reference lines: 29.4 km/s (apparently Shtyrkov's mean) and "astron. 29.765 km/s".
• Almost all points lie between about 28 and 31 km/s. It is an absolute speed derived from an aberration angle by tan(alpha) = v/c (eq 40; alpha = 20.49" -> 29.78 km/s [calc]), not a residual. Highest values fall in late December, lowest in summer, which is qualitatively the perihelion/aphelion pattern (true speed 29.3 to 30.3 km/s [memory, high]); but the real annual swing is as small as the scatter, so the figure does not independently demonstrate a seasonal change. Gift's "seasonal variations being evident" is asserted, not shown.
• Gift states no observable, no reduction method, no model subtracted (orbit determination, pointing, refraction, the 0.02-degree inclination motion). The extraction method is therefore unknown from the folder. Apparatus (as reported): Intelsat 704 at 66 E, receiver at station TAT-01B, Kazan, March 1998 to December 1999. Primary sources (Shtyrkov NPA Proc. 2005 pp.201-205; Eur. J. Natural History 10, 44, 2011) are not in the folder.
• Web check: Shtyrkov replication not checkable (WebSearch exhausted; Wikipedia has no page). Replication: none known to me [check failed].
• Verdict unchanged: Never properly tested (single author, no error bars, unreplicated, extraction method unknown; standard kinematics predicts no first-order aberration for a co-moving source [memory, medium]; Gift himself: "We await confirmation by others"). The "genuinely open" label in the primer should say: open only in the sense that nobody has tried to replicate it; the published figure itself is too thin to be evidence.

Claim 3: Marinov and Weisskopf (2006 deep read, PDF pp.11-12)

• [read: page image p.11-12] "Marinov [44] succeeded in measuring the absolute motion of the Earth relative to the ether by using an ingenious modification of the Sagnac experiment. He obtained a value of 303 +/- 20 km/s, which is comparable to the value measured relative to the CMBR." Ref 44: Marinov, Gen. Rel. Grav. 12, 57 (1980). No apparatus details, direction or data are in the paper. Gift's CMB figure is 390 +/- 30 km/s with no citation. The two differ by 87 km/s, about 2.4 sigma on combined errors [calc: sqrt(20^2 + 30^2) = 36]. "Comparable" is loose, and a speed match without a direction says little. (The usual modern CMB figure is about 370 km/s [memory, medium]; not web-verified.)
• Web: [web: https://en.wikipedia.org/wiki/Stefan_Marinov] Marinov used coupled mirrors and coupled shutters; reported light-speed anisotropy from 1974; peer-reviewed publications 1970-1982; his claims "have not found acceptance"; no successful replication is recorded on that page. So the status is: Never properly tested in the sense of no independent replication on record; not "tested and failed" either, since I found no neutral test report. The source is Wikipedia only and it gives no detail of any attempted repeat; a reader who wants to settle this should read Marinov's 1980 paper and look for independent coupled-shutter repeats. This does not change the claim 3 headline verdict, because Marinov's value of 303 km/s is also inconsistent with Gift's own "30 km/s Roemer" and 390 km/s CMB values (three different speeds for the Earth through one ether). Add to claim 3's tested-status list.
• Weisskopf quotation [read: p.11; ref 43: Weisskopf, Am. Scientist 71, 473, 1983; quote not checked against the original]: "Nature has provided an absolute frame of reference". This is the CMB as a cosmological rest frame, which SR accepts: SR forbids detecting motion relative to a medium by local optics, but does not forbid a distinguished cosmological frame. It does not support an ether carrying light, and Weisskopf himself says the Einstein light-speed postulate stands.
• 2006 equations (page images p.6-9): the Roemer (2.1-2.12) and Doppler (2.13-2.23) derivations reduce to the textbook first-order results 2 T0 v/c and 2 f v/c; the FLL factors drop out at first order. Internal slip: (1.2) lowers a moving system's frequency while (2.3) and (2.7) shorten the measured period (second order only). "Easily separated" for Jupiter's and the Sun's motion is a parenthesis with no method; the Sun's motion through the ether is not separated at all in the equations. Gift also states Maxwell's proposal, Delta t = 2 t0 v/c (eq 3.1), is "identical" to (2.12), but there t0 is the light crossing time of Earth's orbit and v the Sun's speed, a different, far smaller quantity than the Roemer period effect. Priority claim: "for the first time" ether-drift detection for uniform motion [read: p.10]. None of this changes claims 2 or 3.

Web checks this round

• DI Herculis: [web: https://en.wikipedia.org/wiki/DI_Herculis] the apsidal-motion discrepancy was resolved in 2009 by Albrecht, Reffert and Snellen: the stars' spin axes lie roughly in the orbital plane (tilts over 70 and 80 degrees), and including the resulting oblateness the precession matches GR. This upgrades the contrarian's [memory, low] to [web] and removes the 2001 paper's "most glaring failure of GTR" (ref 28) as an objection to GR. Gift's 2001 paper predates this (2001).
• Marinov: see above (Wikipedia only).
• Failed: Shtyrkov replication, Greenspan reception and the H2 value (WebSearch budget exhausted and no Wikipedia page for either). The 1.38e14 vs 1.32e14 Hz question stays open.

Other items recorded (QE section 3)

Marmet and Couture hedge on deflection data; Allan, Weiss and Ashby 1985 (Science 228, 69) among Gift's Sagnac positives (already covered under claim 1's Sagnac replication); Gift's earlier GPS papers (refs 32, 34, 49) republish the same claims; 2001 Table I (one linear equation vs ten nonlinear) and the 2001 claim that GR "predicts dark matter" (a misattribution: dark matter is an observational inference [memory, high]); the 2001 paper asserts but does not calculate the PSR 1913+16 decay and, per the above, its gauge-invariant form has no waves. QE spot check: the 2018 digest's "1950 Nature, Stachel" is 2005 (Nature 433, 215) [read: ref 41]; no later file repeats it.

Round 2 (second QE round: deep read of 2009 PDF pp.5-9, sections 4.1-5 and references)

Input added: the second block of deep/PP-16-03-9f62ec.md (Opus, text layer for pp.5-9 plus page images p.6 and p.8). PDF p.n = journal p.(n+10). I read that deep read in full; I did not re-render any page myself, so page-level statements below are [read: deep read, page image p.6 / p.8].

Result: no verdict changes. The deep read closes the last "thin" item from the first QE revision (the covalent-bond sections, Greenspan's list, Earnshaw, the rho/pion future work). Everything it found fits the existing verdicts for claim 5 and sharpens three statements.

Findings that confirm or sharpen claim 5 - Earnshaw's theorem, corrected. In Round 1 I described Gift as arguing that purely electrostatic systems cannot be stable, so a non-electric force must give matter its solidity. The paper is weaker. It makes one sentence at the end of section 4.1: the magnetic interaction "in the repulsive core mechanism is supported by Earnshaw's theorem [13]" (ref 13: Gottfried and Weisskopf) [read: p.6]. It is cited as support for the repulsive core only, not for atomic stability as a whole; it states no hypotheses and never mentions the uncertainty principle or antisymmetry [read: pp.5-9]. My earlier reply (a classical, static theorem; quantum mechanics stabilises atoms; the magnaton is 4-5 orders too weak to help) is my own [memory, high], not the paper's, and stands. Verdict unchanged. - Greenspan, fully listed. [15] (Physics Essays 5, 250, 1992): H1H1, H2H2 (deuterium), H1H2, H1H3 and Li7-Li7. [16] (Physics Essays 5, 554, 1992): Li7-Li7, B11-B11, C12-C12, N14-N14 [read: page image p.6]. The paper gives no bond lengths or diameters for any molecule, no force law, no attraction magnitude and no initial conditions; "correct diameters" is a bare assertion. The only numbers are the two H2 frequencies. The strong sentence ("an extremely strong indication of the correctness of the magnetic interaction model") is Gift's inference, not Greenspan's. A classical simulation has no spin, and the paper never shows that Greenspan's assumed "fully attractive" electron force is the 1/r^3 magnetic form. This supports "Never properly tested" for the Greenspan row. - 1.38e14 Hz is confirmed as printed. The page image shows 2.20e14 Hz (Coulomb-repulsion simulation) and 1.38e14 Hz (experimental); the paper never says 1.32e14. The standard fundamental is about 1.32e14 Hz [memory, high; NIST value above, [web] from the mathematician]. So the paper's benchmark is about 4.5% off the usual value; the cause (misprint or Greenspan's definition) is still unknown. It does not change the argument, since the contrast is 2.20 against about 1.3-1.4. - No quantitative estimate anywhere. pp.5-9 contain no value of nu, no coupling, no energy or force at any distance, and no ratio to Coulomb [read]. Detailed predictions are explicitly deferred (p.5; Conclusion item 2). Section 4 offers explanations, not predictions. My [calc] of 1e-4 eV against Coulomb about 14 eV (earlier table) is the estimate the paper lacks. - Luder [18] omission. Luder's book is titled "The Electron-Repulsion Theory of the Chemical Bond"; Gift cites it only for the remark that wave-mechanics results are "systematized experimental facts" and does not say its thesis opposes Greenspan's electron-attraction thesis [read: p.9 title; content [memory, medium]]. - The singlet H2 bond. The paper never names the Heitler-London / exchange-integral explanation of the singlet bond, though it cites Moore [17]; the mainstream account is the exchange energy, an electrostatic-plus-antisymmetry effect of order eV [memory, high]. The paper's claim that the singlet magnetic attraction "is enough to overcome" Coulomb repulsion at orbital overlap is asserted, not computed, and contradicts the [calc] 1e-4 eV vs about 14 eV. - 1/r^3 against 1/r. The Conclusion says the magnetic potential "is of the form 1/r^3 but become[s] dominant at high energies or short distances" [read: page image p.8]; a 1/r^3 energy is the dipole-dipole form, whereas exchange of a massless vector between sources gives 1/r. This was already noted under Claim 5 "Agree"; the Conclusion wording confirms it. - Barut and e+e- scattering. The "sharp resonances and novel asymmetries" in e+e- scattering are asserted second-hand through Barut [9] with no energies, cross sections or data [read: p.5]; Krisch [20] and Halzen-Martin [5] are named as places to look. Not checked by me; unverified, and no verdict depends on it. - Future work. The rho-meson and pion statements are hoped-for results and self-citations (Gift refs 22 and 23; ref 23 is the Hadronic Journal paper this one "supersedes"), with no masses, couplings or calculation [read: page image p.8]. "Yielded pi-mesons" is Gift's own earlier claim, not read here. Peripheral; no verdict change. - Peripheral oddities (qualitative claims presented as evidence): "Liquid oxygen will actually cling to a magnet" (ordinary paramagnetism, already explained by quantum mechanics and adding no prediction); the HCl and NaCl bonding narratives (Fig. 1 is a drawing, not data); the "rule of two" credited to magnatons "as verified by the Greenspan data" [read: p.6-8].

Verdicts changed by Round 2: none. Claims 1-4 untouched (the new read covers only the 2009 paper). For claim 5, the headline "Refuted by calculation" (bonding and Pauli), "Never properly tested" (particle; Greenspan) and "not shown self-consistent / magnetic-charge definition internally inconsistent" all stand; the new read adds that the paper itself supplies no number that could contradict the calculation.

Source gap update. The 2009 paper is now read in full (pp.3-9) from text plus page images where equations matter; the earlier gap statement that the covalent-bond sections were only skimmed is closed. Greenspan's two papers, Luder, Barut and Gillespie remain outside the folder, so everything about them is second-hand through Gift.

Verdict table (current, after Round 2: no verdict changed)

Changes the primer must make

Evidence table row 4b to Internally inconsistent and 4d unconditional; remove "via the mathematician" from the light-bending answer; add the Earnshaw versus quantum-stability note; add Marinov to the open items with the status above; correct the Shtyrkov description (figure thin, not strong open evidence); fix the 2006 paper's title and journal ("The Relative Motion of the Earth and the Ether Detected", J. Sci. Explor. 20(2):201-214, 2006 [read]); add the DI Herculis [web] resolution. Round 2 additions: say the Earnshaw point is one sentence in the paper and the quantum reply is the reviewers' own; quote the H2 benchmark as '1.38 x 10^14 Hz as printed (standard value about 1.32 x 10^14)'; list the Greenspan molecules (H2, D2, HD, HT, Li2 in [15]; Li2, B2, C2, N2 in [16]) and say no diameters are given; note Luder's book argues the opposite (electron repulsion); note the paper gives no coupling or energy estimate, so the primer's worked example (about 1e-4 eV against about 14 eV Coulomb) is the missing calculation.