Rankings Primer Arbiter Physicist Mathematician Contrarian QE Claims Sources

Gift-Stephan: Science Contrarian review

Reviewer role: Science Contrarian (judges by experiment, separates "tested and failed" from "never tested" from "dismissed without testing", builds the strongest honest case, and holds the researcher to the same standard as the critics).

Source gap (repeated from Step 1, updated)

The text layers of the 2001 gravity paper and the 2009 magnetism paper are poor. I re-read the 2001 paper from page images (pp. 6, 7, 9, 10 viewed by me, plus the 2001 equations 7-21) and the 2009 paper's key page (journal p. 14, page image) [read: page image]. The 2018 and 2006 papers were read from the text layer (2006 equations lost in extraction; the argument was followable from prose).
Still missing: all cited primary sources (Shtyrkov, Marmet, Kelly, Sato, Wallace, Greenspan, Aspden, Hatch, Miller/Shankland papers). Anything said below about them is second-hand through Gift or [memory].
Web budget: my web-search allowance was already exhausted; only a few page fetches (Wikipedia) worked. Facts I could not check on the web are tagged [memory] with confidence, or "check failed".

Corrections to Step 1

2001 paper page range: facts.json says pp. 320-326; the paper runs pp. 320-328 (page 10 of the PDF is journal p. 328: endnote, references, French abstract) [read: page image p.10]. Pages 6, 7, 9, 10 are no longer unseen. Equations 7-10 (static field equation), 11-17 (perihelion, photon-mass heuristics, light bending), 18-21 (redshift) and Table II were read.
claims.md claim 1 says the c-v / c+v result is "shown by GPS clocks". The 2018 paper contains no GPS dataset. The "measured" times (its eqs 13-16, 18-20) are taken from the CCIR/ITU synchronisation formula (eq 1), which is built on light speed c in the Earth-centred inertial (ECI) frame. Only Marmet's and Kelly's observations (e.g. 414 ns) are reported, second-hand [read]. The claim should read "derived from the standard GPS synchronisation and range formulas".
claims.md claim 3 says "a cosmic ether (preferred reference frame) exists". The 2018 paper (section 4) actually concludes there are "multiple preferred frames carried along by the Earth and the Sun respectively" (ECI for rotation, solar-system barycentric for orbit), and explicitly cites Hatch's alternative "clock bias" interpretation as a rival reading [read]. The CMB-as-ether-radiation suggestion appears only in the 2006 paper [read].
claims.md claim 2: Gift himself writes of Shtyrkov "We await confirmation by others of this very interesting experiment" and "appears to" detect [read, 2018 section 3.1]. The folder gives no replication.
claims.md claim 4: the 2001 paper says plainly that light bending and redshift are not predicted by the theory alone: "the new theory by itself cannot predict the outcome" and it uses a "(heuristic)" photon-mass argument [read: page image p.6]. So "gives... the redshift and orbit precession" is stronger than the paper. Precession is taken from Aspden's modified force law, not derived (see claim 4 below).
No other errors found in the claims; the five-claim structure is fair.

Numbers I computed (Python, .venv)

Equatorial Sagnac round-trip difference 4 omega A / c^2 with omega = 7.292e-5 rad/s, R = 6.378e6 m: 414.7 ns. This matches Kelly's 414 ns, and it is the figure standard relativity predicts. [calculation]
Magnetic dipole-dipole energy for two electron spins at r = 1 angstrom: mu0 mu_B^2 / (4 pi r^3) = 5.4e-5 eV, against a Coulomb energy of 14.4 eV. Ratio (hbar / 2 m c)^2 / r^2 = 3.7e-6. The two are equal only at r about 1.9e-13 m. A hydrogen-molecule bond is about 4.5 eV. [calculation]
Newtonian light deflection with photon "mass" E/c^2 at the solar limb: 0.876 arcsec; GR: 1.752 arcsec. [calculation]
Gift's Table II gives gravitational strength G m^2/(hbar c) = 3e-42. For the proton I get 5.9e-39; for the electron 1.75e-45. Neither matches (minor; the mass used is not stated). [calculation]

Claim 1. Light speed is c-v eastward and c+v westward relative to the rotating Earth (GPS clocks, range equation, time transfer, GPS-clock Michelson-Morley)

a. Mathematics. Each derivation is algebraically correct given its input. The inputs are the problem. Eq 1 (synchronisation) and eq 23 (range equation) both assume light moves at exactly c in the ECI frame. Having assumed that, finding c-v for a station moving at v in that frame is the Galilean velocity addition of the assumption. This is circular: the output restates the input. Equation 37 (c_R = c - v cos theta) is the same step for time transfer. The "GPS Michelson-Morley" (eqs 13-22) calls the Sagnac-corrected time "measured by GPS clocks" and then equates it with the ether prediction; the north-south null (eq 22) is the same assumption. Eq 17 is first order in v/c and is not "unaffected by length contraction" in any meaningful sense; it is a statement about clock synchronisation along one axis.

b. Experiments. The Sagnac effect (Michelson-Gale 1925, ring lasers, GPS clock corrections) is established. Standard relativity predicts the same number; the 414 ns figure reproduces. Nothing in the paper shows a result that differs from standard GPS practice, which already includes the Sagnac correction. Gift's own cited modern resonator limits (delta c / c < 1e-17, from his refs 7-10, [read]) are two-way/isotropy tests and show no anisotropy.

c. Tested? The one-way speed cannot be measured independently of the clock synchronisation convention [web: https://en.wikipedia.org/wiki/One-way_speed_of_light]. GPS time-transfer is built on the Einstein-style convention in the ECI frame; Gift reads the Sagnac term as a speed difference, which is a different convention (rotating-frame coordinate speed), not a different physical result. No independent replication of his reading exists in the folder. Hatch, whom Gift cites, reads the same facts as clock bias [read]. I cannot name an independent confirmation; that is not the same as a refutation. The papers are in Physics Essays and Applied Physics Research (a low-visibility journal, [memory] medium).

d. If true. Special relativity's second postulate would fail, but so would the rotating-frame GPS processing that works because it assumes c is isotropic in ECI. The paper would also need a preferred frame that is the ECI frame (not the Sun frame, not the CMB), carried by the Earth's orbital motion but not its spin.

Strongest honest case for Gift: the conventionality of one-way light speed is real physics, and the textbook c = constant for all observers is a convention plus two-way measurements. Sagnac's own 1913 interpretation was as an ether detection, and Ives (1940, quoted by Gift) argued a Lorentz-ether reading is equally valid. A Lorentz-type ether theory is empirically equivalent to SR [Mansouri and Sexl, which Gift cites; [memory] high]. Gift is right that "light speed is exactly c in all directions for all observers" is not experimentally testable as stated.

Where Gift is wrong: "detects" and "unambiguous" in the abstract. A convention-dependent number cannot detect anything. Also the "ether" he needs is a different frame for every motion (ECI for rotation; barycentric for orbit), which is not an ether but a recipe.

Verdict: split. - Weak form (coordinate one-way speed in the rotating frame is c -/+ v with ECI-based synchronisation): Established, and identical to the standard Sagnac effect. - Strong form (this detects an ether and refutes SR): Internally inconsistent (circular: assumes isotropic c in ECI and concludes anisotropy) and Not testable as stated (synchronisation convention). It was not dismissed without testing; the claim is not a test.

Claim 2. The Earth's orbital motion is detectable with apparatus at rest on the Earth (planetary radar, Shtyrkov aberration, Roemer, stellar Doppler)

a. Mathematics. 2006 paper: the Roemer and Doppler derivations assume a source at rest in the preferred frame (Io-Jupiter "pulses of darkness" separated by a fixed distance L0; a stationary star), and treat the effect as the observer's velocity alone. This forces the preferred frame to be the Sun-Jupiter frame, and the resulting speed is the Earth's 29.8 km/s orbital speed. The 2006 paper suggests the CMB is the ether, but the Earth moves at roughly 370 km/s relative to the CMB (Gift quotes 390 +/- 30 km/s [read]). The Roemer data show 30 km/s, not 390. So the two statements cannot both be right. Gift handles Jupiter's and the Sun's motion with a parenthetical "easily separated" [read], which is the unsupported step. The 2018 paper then adds a different frame (SSB) for the radar equations 41-55; the same circularity as claim 1 applies (eq 42 assumes c in the SSB frame and returns c +/- v for the Earth).

b. Experiments. Roemer delay, stellar Doppler and aberration are established. All are predicted by SR and by Lorentz ether theory; the first-order fractional change v/c (about 1e-4 for 29.8 km/s [calculation]) is the same. Gift concedes the theory is "in close agreement with special relativity" and that Mansouri and Sexl showed first-order tests cannot distinguish them, then rejects that as "incorrect" because the "light speed variation" contradicts the postulate [read, 2006 section 3]. That is a reinterpretation of the same data, not new data. Shtyrkov (geostationary Intelsat 704, Kazan, 1998-99) is the one claim that would be a new result. By the normal analysis aberration depends on the relative velocity of observer and source; with both fixed to the Earth the expected effect is zero [memory, medium; I could not web-check]. If Shtyrkov saw a seasonal 30 km/s signal, a mundane cause (Earth-based tracking and ephemeris corrections, satellite orbit determination which is itself done in a frame that includes aberration) would need to be excluded; Gift's figure 5 is second-hand, from a Natural Philosophy Alliance proceedings paper and a minor journal [read: refs 47-48].

c. Tested? The Roemer, Doppler and aberration measurements: tested for centuries, consistent with SR. Shtyrkov: no replication in the folder; Gift says so himself; I found no source I could check (web check failed: search budget exhausted). Wallace's 1969 radar test (Gift's ref 53) is cited as supporting c +/- v; I have not read it [memory low: it is Wallace's claim to have tested light speed independence of the source; I do not know its reception].

d. If true. Aberration for co-moving source and observer would be non-zero, contradicting the basic aberration kinematics, and deep-space ranging pipelines (JPL ephemerides in the SSB frame, which Gift uses) would need a frame-dependent c that the pipeline does not use.

Strongest case: first-order effects were historically the right place to look; DeMeo-type reviews and the conventionality argument keep an open door; Shtyrkov's data would be cheap to replicate with an amateur or university dish and a geostationary beacon.

Verdict: split. - The effects themselves (Roemer, Doppler, aberration of stars): Established. - That they "detect an ether": Not testable as stated (equivalent to SR under a synchronisation change); the 30 km/s vs 390 km/s mismatch adds an internal problem for the CMB version. - Shtyrkov's co-moving-source aberration: Never properly tested (one unreplicated report; not dismissed, simply uncontested and unreplicated).

Claim 3. A cosmic ether exists, light-speed invariance and special relativity are invalid, second-order tests such as Michelson-Morley were the wrong test (FLL contraction); the CMB is the ether's thermal radiation

a. Mathematics. Internally this is a Lorentz-ether (generalised Galilean transformation, "GGT") theory with Fitzgerald-Lorentz-Larmor contraction and clock retardation. It is consistent as a theory, and it is mathematically equivalent to SR for all measurable results under a clock-synchronisation change (Mansouri-Sexl; Gift admits it, then dismisses it). The statement "SR is invalidated" depends on equating a convention-dependent coordinate speed with a physical speed. In 2006 the argument that the 'measured' contracted-rod/retarded-clock speed differs from the 'true' speed means no experiment can ever show the 'true' speed; so the preferred frame is untestable by construction. The paper cannot have both "ether detected" and "FLL contraction hides it in second-order tests". The CMB suggestion: the CMB frame (dipole, about 370 km/s) is observationally a preferred frame for cosmology but not for light propagation; [memory] high.

b. Experiments. Modern rotating-resonator and cavity tests limit anisotropy to about 1e-17 (Gift's refs 7-10, [read] as cited; Eisele, Nevsky, Schiller PRL 103 090401 (2009)). Gift's reply is that these are second-order and FLL contraction cancels them. That reply is the point of Lorentz ether theory and makes it unfalsifiable by those tests; but then also the same tests do not support the ether.

c. Tested? Michelson-Morley and successors: tested, null. Dayton Miller 1925-33: claimed small positive drift; reanalysed by Shankland et al. 1955 as temperature artefacts (Gift's refs 5-6). Gift implies Miller was "forgotten". The honest status: Miller's results were examined and explained, but I know of arguments that the Shankland reanalysis was itself contested [memory, low; not checked, check failed]. Other claimed positives (Demjanov, Galaev, Munera, te Winkell) are cited by Gift as "ignored" [read]; none appear in the folder, none replicated independently as far as I can tell [memory low]. I would call the second-order null results "tested, not supported", and the first-order and one-way claims "not testable" (claims 1-2).

d. If true. Special relativity would be an effective theory in a preferred frame, tensions with particle experiments that test Lorentz invariance to extreme precision (the limits quoted above), and the time dilation of fast muons and GPS clock corrections would need a preferred-frame explanation. Nothing else in the Standard Model would need changing if the ether were undetectable by construction.

Strongest case: Lorentz ether theory is an honest, empirically equivalent alternative; relativity's textbook "no ether" is a methodological choice (Occam) more than an experimental result; Gift is right to point out that some textbook statements overreach (e.g. "ether disproved by Michelson-Morley" ignores the first-order loophole).

Verdict: split. - "An undetectable-by-construction preferred frame exists" (the weak form): Not testable as stated. - "An ether has been detected and SR is falsified": Tested, not supported (second-order and resonator tests find nothing) plus the internal problems above; the "detection" rests on convention-dependent first-order effects (claims 1-2). It was not dismissed without testing: the experiments were done, but Miller's reanalysis history is worth a fresh independent read.

Claim 4. The equivalence principle is false (a charge in an accelerating rocket radiates, one hovering above a mass does not), so GR is invalid; replace it with a Newtonian-type quantum gauge theory with massless spin-0 gravitons (Newton's law, redshift, perihelion), with no black holes, dark matter or Einsteinian gravity waves

a. Mathematics. (i) Radiation argument (2001 section 2): uses the Larmor formula P = (1/4 pi eps0)(2/3) q^2 a^2 / c^3 in rocket B but declares zero radiation in rocket A. Larmor holds in inertial frames. The standard resolution is that radiation is frame-dependent: Rohrlich (1965) and Boulware (1980) show a supported charge does not radiate in its own rest frame and does in the free-fall frame, the radiation going into a region the co-accelerating observer cannot reach; the same holds for the charge in the accelerating rocket [web: https://en.wikipedia.org/wiki/Paradox_of_a_charge_in_a_gravitational_field (Rohrlich, Boulware)]. Gift cites Peierls and calls his resolution "arbitrary" [read]; this is what is called the "radiation paradox", resolved in the sense that the equivalence principle is local. The inconsistency is treating A by one rule and B by another. (ii) Gauge theory (2001 sections 4-5; eqs 7-10, read from page image p.6): the scalar equation box A = -m psi-bar gamma^0 psi, with the source called "mass density", reduces in the static case to the Poisson equation and Newton's law. That is the intended reduction of a massless scalar exchange and is correct as far as it goes. But psi-bar gamma^0 psi is the particle number density (times m), so the source is rest mass only: no coupling to kinetic or field energy. "Rigidity" and "gauge invariance gives the inverse square law" are asserted. The coupling term is admitted not Lorentz invariant and allowed by assumption (Postulate-style) [read]. (iii) Perihelion (eqs 12-15, page image p.7): the force law F = (1 + 3v^2/c^2) G M m / r^2 is taken from Aspden (1980) as a retardation result and not derived in the paper. I checked that the Binet equation (13)-(14) follows from that law and gives the GR perihelion advance 6 pi G M / (a c^2 (1 - e^2)) per orbit; that coefficient is exactly the GR one, so it is tuned or borrowed rather than a prediction of the scalar theory. (iv) Light bending (eqs 14-17): equation 14 with the 3GM(u/c)^2 term is the GR photon orbit equation, applied to a photon with Newtonian "mass" E/c^2. The Newtonian-mass calculation gives 0.876 arcsec [calculation]; the paper's value of 1.75 arcsec arises from the borrowed relativistic term. A massless scalar coupled to mass density does not couple to light at all (a photon has no rest mass), which is why the paper falls back on a heuristic. (v) Redshift (eqs 18-21): hv - hv' = G M m / R with m = h v / c^2 gives delta nu / nu = G M / (R c^2). This is the standard weak-field result and a famous heuristic. Correct to first order. (vi) Table II: the strength 3e-42 for gravity does not match the proton (5.9e-39) or electron (1.75e-45) [calculation]. Minor.

b. Experiments. The weak equivalence principle holds to about 1e-15 (MICROSCOPE satellite) [web: https://en.wikipedia.org/wiki/MICROSCOPE_(satellite), which gives 1e-15 for the 2017 result; the page does not give the 2022 final figure, which I recall as about 1e-15 [memory medium]]. Gift's argument targets the Einstein (local) equivalence principle, a broader statement; but none of the experiments shows a violation. Gravitational waves: the pulsar PSR B1913+16 orbital decay matches GR's quadrupole radiation, observed/predicted = 0.9983 +/- 0.0016 (Weisberg and Huang 2016) [web: https://en.wikipedia.org/wiki/PSR_B1913%2B16]. Direct detection by LIGO since 2015 (GW150914 [memory high]) and the three-detector event GW170814 found strong evidence for pure tensor polarisation over pure scalar or vector polarisation [web: https://en.wikipedia.org/wiki/GW170814]. A spin-0 graviton theory predicts scalar waves, which is the disfavoured option. Black holes: LIGO mergers and Event Horizon Telescope imaging [memory high]. The paper's section 5.4 treats the pulsar decay "qualitatively" [read]; no calculation of the 0.2% match is given. Dark matter is outside this theory; it is simply asserted gone.

c. Tested? The scalar-graviton theory in this form was published in 2001 in Physics Essays. As far as the folder shows, nobody tested it; the tests it faces were done on GR's predictions, and the data (LIGO polarisation, pulsar decay at 0.2%, MICROSCOPE) go against it. Nordstrom-type scalar gravity theories were shown to fail the light-bending and perihelion tests long ago [memory high]. The radiation argument is not an experiment; no experiment has found the two frames distinguishable. How was it rejected? No sign of suppression: the 2001 paper was peer-reviewed in a journal; the criticisms are physics, not process. The "binary star DI Herculis" claim (section 3) rests on the 1980s apsidal-motion discrepancy; I did not check it and the case is understood to be resolved by the stars' spin-orbit misalignment [memory low].

d. If true. The equivalence principle, GR's tensor waves, black holes, the Hulse-Taylor interpretation, LIGO's polarisation results, and the interpretation of cosmological data would all need to change, along with the fact that photons gravitate.

Strongest case: GR does have unresolved issues (singularities, quantisation) and the equivalence principle is a postulate; the radiation question was treated carefully in the literature, and Gift is right that "the frames are equivalent" is subtle. A Newtonian-with-retardation approach would be worth a calculation by the sceptic.

Verdict: split. - Radiation argument that the equivalence principle fails: Internally inconsistent (different treatment of the two cases; standard treatments resolve it as frame-dependent radiation). Not an experimental result. - Scalar spin-0 graviton theory reproducing Newton and the redshift: Plausible, untested as a gauge theory; Newton and the weak-field redshift are trivially reproduced. - Same theory giving perihelion and light bending: Refuted by calculation as a derivation (precession taken from Aspden's law, light bending uses a borrowed GR term; Newtonian photon mass gives 0.876 vs 1.752 arcsec). - "No gravitational waves of Einstein's kind, no black holes": Tested, not supported (Hulse-Taylor, LIGO tensor polarisation, EHT). Never dismissed without testing; it was overtaken by data after 2001.

Claim 5. Magnetism is a fundamental interaction separate from electromagnetism, mediated by new massless spin-1 "magnatons", explaining Pauli exclusion and chemical bonding

a. Mathematics (read from page image, journal p. 14). - Eq 3.3-3.4: the global phase symmetry's Noether charge is nu times the integral of psi-bar gamma^0 psi, i.e. a constant times fermion number. Calling nu "magnetic charge" (defined in 3.1 as the divergence of the moment) is relabelling; the conserved quantity is not magnetic flux. - Eq 3.6: the coupling is nu psi-bar gamma^k psi A_k, a coupling to the fermion's spatial number-current. This is the spatial half of the ordinary vector coupling j.A of QED, not a coupling to the magnetic moment (which would need the spin term sigma^{mu nu} F_{mu nu}). The text says the interaction "involves the particle's magnetic moment" [read] while the Lagrangian does not contain the moment. - Eqs 3.8, 3.10, 3.12: the field equation is then identified with Ampere's law by setting nu psi-bar gamma^k psi = J_k and identifying this with the Amperian current of magnetisation (3.11, curl M). A Dirac number current is not a magnetisation current. The agreement with magnetostatics is because A_k is being labelled the ordinary vector potential, so "converts all of classical magnetism into a quantum theory" is just ordinary electromagnetism rewritten, and "separate from electromagnetism" is asserted, not shown. - The kinetic term added for A_k (not shown) cannot be the invariant F_{mu nu}F^{mu nu}-type term, since for a time-independent alpha(x) the shift A_k -> A_k + d_k alpha changes d_i A_k. The paper says only "add a term representing kinetic energy" [read]. I could not confirm gauge invariance of the full Lagrangian (my derivation; confidence medium). - Eq 3.13 and "potential of form 1/r^3": a massless vector exchanged between charges gives a 1/r potential. The 1/r^3 comes from the classical dipole-dipole formula (Wangsness), not from the exchange of the field in 3.6. - Greenspan's classical simulations "assume attraction" between bond electrons and fit the vibrational frequency (2.20e14 Hz repulsive vs 1.38e14 Hz measured [read]). That a fit improves when a parameter changes sign is not a test of a magnetic mechanism.

b. Experiments. Magnetic dipole-dipole between electron spins at 1 angstrom is 5.4e-5 eV against 14.4 eV Coulomb and about 4.5 eV for the H2 bond [calculation]: 5 orders of magnitude too weak, and equal to Coulomb only at 1.9e-13 m. The singlet-triplet splitting of chemical bonds is the electrostatic exchange energy from antisymmetry; magnetic spin-spin splittings are in the micro-to-milli eV range (ortho/para positronium is 8.4e-4 eV [memory medium]). Pauli exclusion is established as a consequence of antisymmetry and the spin-statistics theorem [memory high]. Gift's argument that antisymmetry identifies "no force" is a semantic point; "Pauli force" is an effective exchange effect. No magnaton has been observed [memory high]. Gift's reference to Barut's resonances in e+e- scattering is second-hand [read] and the narrow resonances in that data are not generally accepted as such [memory low].

c. Tested? Only by Greenspan's simulations (not experiments), not replicated in the folder. No neutral test was proposed; the paper itself defers "quantitative application" and renormalisability [read]. It was published in Progress in Physics, a journal with open submissions and light review [memory medium]; it was not rejected by mainstream channels or suppressed; it simply has not been taken up.

d. If true. QED's unification of electric and magnetic forces and Lorentz covariance of QFT would be wrong, and the quantum treatment of chemical bonding (which predicts bond energies to chemical accuracy) would be wrong. None of this is needed to explain magnetism.

Strongest case: the opening puzzle (the sign of the spin-spin magnetic interaction in different pictures) is a real teaching point, and Gift is correct that the "Pauli force" is an effective, not a fundamental, force, which students rarely hear. Greenspan's simulation line (classical dynamics with attractive electron interaction) is an odd result worth an independent check.

Verdict: split. - A new, separate magnaton interaction: Internally inconsistent (coupling is to number current, not to the moment; reproduces ordinary A) and Refuted by calculation as an explanation of bonding (magnetic energy is about 4e-6 of Coulomb at 1 angstrom). - Pauli exclusion as a force: Not testable as stated / replaced by standard antisymmetry (Established). - Greenspan "confirmation": Never properly tested (no replication or controlled comparison on record, to my knowledge).

Where Gift is right, where he is wrong (summary)

Right: one-way light speed is convention dependent; first-order tests are the sensible place to look for a preferred frame; textbook claims that Michelson-Morley "disproved the ether" overreach; the radiation puzzle in a gravitational field is real; the "Pauli force" is not a force. Wrong: treating convention-dependent coordinate speeds (derived from formulas that assume isotropic c) as measurements; claiming a detected ether while conceding Lorentz-ether theory predicts SR's results; one ether per experiment; borrowing GR's coefficients for the gravity theory; a magnetism field that is just the ordinary vector potential and whose strength is five orders short for chemistry; calling a data fit "experimental confirmation".