RankingsPrimerArbiterPhysicistMathematicianContrarianQESites logClaimsSourcesResearch notes

Su-CC: Physicist review (Step 2)

Reviewer: PhD physicist. Project: Ching-Chuan Su (C.-C. Su), National Tsing Hua University, Hsinchu, Taiwan: the "local-ether" model and "Quantum Electromagnetics". Date: 2026-10-06.

Source tags: [read] from the project documents (via the deep reads, which I checked against my own calculations where stated), [calc] my own calculation, [web] with URL, [memory] with confidence. Su is the originator and an interested party: his descriptions of special relativity and of other people's experiments are his characterisations.

Source gap (repeated from claims.md). The 2005/2008 book is not in the folder. None of the experimental papers Su cites as support (Brillet and Hall, Saburi, Krisher, Staudenmann, Mandelberg and Witten, Cedarholm, Champeney, Turner and Hill, Hay) is in the folder. Where I compare Su with those experiments, the comparison rests on Su's citation or on [memory] unless tagged [web]. Web checks used (4 of about 6 budget): Brillet-Hall abstract, Herrmann 2009 abstract, Cassini 2003 result, quasar-periodicity surveys. I could not check Hafele-Keating's measured numbers, Mandelberg-Witten's precision, Staudenmann's data, or the Brillet-Hall paper body; those stay [memory].

Abbreviations: ECI = Earth-centred inertial frame (origin at Earth's centre, axes not rotating with Earth); SR/GR = special/general relativity; LLI = local Lorentz invariance (the statement that, in a freely falling lab, physics is the same in every direction and at every uniform velocity); PPN = parametrized post-Newtonian (a standard family of weak-field gravity theories, with parameter gamma = 1 for GR).

0. The mainstream view, stated fairly (needed to read everything below)

  1. Sagnac effect and GPS. Light going round a loop that rotates takes a different time in the two directions: delta-t = 4 (Omega . S)/c^2 for rotation rate Omega and loop area S (directed). In SR/GR this is computed in a non-rotating (inertial) frame where light speed is c; for Earth, the ECI frame. GPS practice uses exactly this: pseudorange gets the correction 2 (omega_E . S)/c (Su's eq. 9), with S the triangle area satellite-receiver-Earth centre [read: EPJC p.3]. [memory, high: Ashby, Living Rev. Relativity 6, 1 (2003)]. So "Earth's rotation matters, Earth's orbital motion does not" is not Su's discovery. In GR it follows from the equivalence principle: Earth orbits the Sun in free fall, so a lab near Earth sits in a local inertial frame; orbital velocity produces only tiny tidal terms. SR/GR needs no ether for this. Su gives the same numbers by declaring that light travels at c relative to a "local ether" at rest in the ECI frame.
  2. Preferred-frame ("ether") theories. A theory with a preferred frame and clocks/rods that obey Lorentz-covariant dynamics (Lorentz ether theory) is observationally identical to SR. A theory with a preferred frame and no length contraction or clock slowing makes different predictions at second order in v/c. The Robertson-Mansouri-Sexl test theory parametrises these differences. Modern limits: Brillet and Hall 1979, anisotropy of light speed Delta-c/c = (1.5 +/- 2.5) x 10^-15 [web: https://vivo.colorado.edu/display/pubid_108412]; rotating-cavity experiments of Herrmann et al. 2009, about 1 x 10^-17 [web: https://arxiv.org/pdf/1002.1284]. These bounds are on signals tied to the sky (sidereal modulation), that is, to a frame moving at hundreds of km/s. They do not by themselves exclude a model whose preferred frame moves with the Earth. That is Su's design, and it is why his model survives them, and also why it can only be tested by experiments that move relative to the ECI frame (turntables, vehicles, satellites). I say this plainly because it is the strongest honest point in Su's favour: the model is not refuted by the famous null results.
  3. Gravity and light. In GR's weak field a light ray behaves as if the vacuum had refractive index n = 1 + (1 + gamma) GM/(c^2 r); gamma = 1 gives deflection 4GM/(c^2 b) and the Shapiro delay. Measured: gamma = 1 + (2.1 +/- 2.3) x 10^-5 from Cassini [web: https://ideas.repec.org/a/nat/nature/v425y2003i6956d10.1038_nature01997.html]. Gravitational redshift is a lowering of atomic frequencies (or equivalently clock rates) by about Phi/c^2 (Pound-Rebka 1960, Gravity Probe A 1976) [memory, high].
  4. Matter waves. The Klein-Gordon equation (relativistic wave equation for a particle with rest mass m0, natural/Compton frequency omega0 = m0 c^2/hbar) is Lorentz invariant and gives de Broglie relations and E = gamma m0 c^2. Compton scattering is derived in textbooks from energy and momentum conservation for a photon-electron collision (Compton 1923) [memory, high].

1. Claim-by-claim review

Claim 1. A local ether, not a universal one, carries light (earthbound: ECI frame; interplanetary: heliocentric frame); MM null explained.

a. Mathematics internally consistent? Yes for the first-order and second-order propagation-range formulas (eqs. 2, 6, 7, 8, 13). The deep read reproduced the numbers (30 m GPS Sagnac, pi/3 fringe amplitude, 62 Hz, 2 rad loop) [read: deep/A-local-ether]. I confirm by hand: eq. 13 gives arm times (2L/c)(1 + v^2/c^2) and (2L/c)(1 + v^2/2c^2), the 1887 classical-ether result; the formalism is textbook ether theory, the novelty is which frame v is taken in. Small slips: orbital/rotational speed ratio is 64 not "about 100" [read, deep]; no effect on the argument. Gaps (not inconsistencies): (i) frame choice for interplanetary radar rests on the frame the published analyses used plus a 10 microsecond estimate, with no geocentric recalculation shown [read: deep]; (ii) the interface between two local ethers is undefined; Su concedes annual stellar aberration is unsolved once Earth's ether is included [read: EPJC p.12, p.15]. This is not minor: starlight enters the atmosphere moving with the Earth's orbit through an Earth-ether co-moving with the Earth. If the wave then travels at c relative to the Earth ether, telescope and medium share the 30 km/s and annual aberration (about 20.5 arcsec) should vanish [calc: 30 km/s / c = 1.0e-4 rad = 20.6 arcsec; memory, high for Stokes-drag objection]. Aberration is observed (Bradley 1725). So the model as written contradicts an established observation; Su says so himself.

b. Consistent with established experiments? Split. - GPS, intercontinental microwave links, Michelson-Gale, fibre gyros: consistent, because the formulas are the standard ECI-frame formulas (Su cites Ashby, ref. 7). These data cannot tell the local-ether model from SR/GR with an ECI frame: the first-order round-trip Sagnac term depends only on relative velocity [read: EPJC p.3]. The Japan-USA 0.3 microsecond and Japan-Korea 40 ns numbers are of the right size [calc, deep]; I have not seen the original papers. - Michelson-Morley: Su predicts a small but non-zero second-order term from Earth's rotation (about 2.4e-12 in v^2/c^2, giving about 2.5e-4 rad in the 1887 apparatus: undetectable) [calc, deep]. SR predicts exactly zero. The 1887 experiment cannot separate them. - Brillet and Hall 1979 and its successors: Su's eq. 14 predicts a lab-fixed 2-omega (twice the turntable rotation rate) term of 62 Hz peak-to-peak at 3.39 micron, 40 deg latitude [calc: v_E^2/(2c^2) cos^2 40 deg x 0.884e14 Hz = 62 Hz]. Su quotes a reported term at the right rate of about 17 x 2 Hz, "attributed to a persistent spurious signal" [read: EPJC p.9]. 34 Hz / 0.884e14 Hz = 3.8e-13 fractional. The quoted Brillet-Hall sensitivity of 2.5e-15 [web, above] concerns the sky-fixed (sidereal) signal; a lab-fixed 2-omega term is the kind of signal an apparatus on a turntable also gets from table tilt, gravity sag, magnetic and thermal effects, so it is analysed as a systematic [memory, medium]. I did not obtain the paper to see their reasons for calling it spurious. What I can say: the measured term is about half the prediction (ratio 0.55 [calc]), Su does not discuss the ratio or phase, and no one has isolated the term's cause. The modern rotating-cavity experiments I know report limits for sidereal-modulated amplitudes; whether any of them reports a lab-fixed term below 7e-13 I could not check [memory, low]. This remains open (see claim 5). - Annual stellar aberration: not reproduced (above). - Mainstream test frameworks: all satellite and Earth-orbit-velocity tests are blind to the model by construction (point 0.2).

c. Tested? By whom? Replicated? No controlled test of the distinguishing prediction exists. The only comparison is the Brillet-Hall residual (single measurement, dismissed as spurious, never re-measured as a lab-fixed term for this purpose). Advocates (Marsen 2019, 2021; Gift 2021) have proposed or argued, not measured (see Investigator section). Reception: no published Comment or Reply on the EPJC/EPL papers found in one pass (Investigator); Scholar counts 46 (EPJC) and 32 (EPL) citations, mostly proponents and engineering papers that use the ECI Sagnac correction. Silence is not evidence either way.

d. If true, what must change? (1) Lorentz invariance would be violated for any lab moving relative to its local ether: boosts of an Earth lab at 465 m/s, and of a satellite at 7.6 km/s, would show second-order effects (3e-10 for LEO [calc: (7.6e3/c)^2/2 = 3.2e-10]). (2) SR/GR's equivalence principle would be replaced by a mechanism whose boundary between ethers is undefined. (3) The GR link between gravity and local inertial frames would need a new account of where Earth's ether ends (Moon distance? Hill sphere? GPS orbit?). Su gives no rule.

Verdict (split). - ECI-frame Sagnac correction for GPS and links, and "no Earth-orbit term in earthbound propagation": Established (as standard SR/GR practice; it does not need an ether). - The ether as the explanation (versus SR/GR in the ECI frame) at first order: Not testable as stated (identical numbers). - The second-order lab anisotropy (62 Hz cavity beat): Never properly tested (one old residual, attributed to a systematic; not dismissed after a targeted test, but not tested either). - Annual stellar aberration with an Earth ether: Tested, not supported (Su concedes it is unexplained; the observation is established).

Claim 2. Gravity is an optical index n_g = 1 + 2GM/(c^2 r), not curvature.

a. Mathematics. Consistent. I accept the deep read's checks: deflection integral 4GM/(c^2 b) (1.75 arcsec at the Sun), GPS delay (2GM/c^3) ln[(r_s + r_e + R)/(r_s + r_e - R)], radar delay (4GM/c^3) ln[...], 232.5 microseconds for Earth-Venus at conjunction [read: f5b; calc]. Su's quoted GPS range 43-60 ps is 42.2-62.2 ps on recomputation, and 4GM_Sun/c^3 is 19.7 not 19.5 microseconds: rounding-level. The "2" in n_g is postulated, not derived [read: f5b p.3]; it equals the GR/PPN weak-field value with gamma = 1 [memory, high]. So eq. 28 reproduces GR by construction. The "19.5 microsecond disagreement with GR" Su reports is a coordinate convention difference between textbooks (Weinberg-type "+1" term), not an observable [read: f5b, reader note; memory, medium].

b. Experiments. Matches the radar and deflection measurements to the extent GR does, because the coefficient is GR's: Shapiro et al. 1971, Reasenberg et al. 1979 (Viking), Anderson et al. 1975, Fomalont and Sramek 1976 are cited by Su as the data; Cassini 2003 gives gamma = 1 + (2.1 +/- 2.3)e-5 [web]. Su quotes no measured values from these papers [read]. What the model does not address (and Su does not claim to): perihelion precession, frame dragging, gravitational waves, strong fields. A scalar index with source GM/r has no field equation of its own and no account of how Phi_g is generated by moving or non-spherical sources. The index form also applies to light only. Su's weak-field assumption (Phi_g/c^2 << 1 [read: f5a p.3]) matters for claim 6.

c. Tested? The numbers have been tested many times, but as tests of GR. No test discriminates "index in a local ether" from "curved spacetime"; Su proposes only that frames and reference instants differ, size unstated, no test given [read: f5b section 6].

d. If true? Nothing in tested weak-field light propagation would change. What would have to change is the conceptual status of curvature, and a gravity theory beyond light would have to be built.

Verdict (split). Weak-field light deflection, Shapiro delay, GPS gravitational delay as numbers: Established (they are GR's; reproduced here by construction). "Gravity without curvature" as a replacement: Not testable as stated for light; as a complete theory of gravity: Plausible, untested only in the sense that nothing beyond light has been derived (it is not tested because it has not been formulated).

Claim 3. Matter waves obey the same preferred-frame equation; mass is a frequency; transition frequency f = f0 sqrt(1 - v^2/c^2).

a. Mathematics. The algebra checks: dispersion omega^2 = omega0^2 + c^2 k^2, group velocity gives k = omega v/c^2, omega = omega0/sqrt(1 - v^2/c^2), m0 = hbar omega0/c^2 (checked symbolically by the deep readers; my reading agrees) [read: f6, f8a, f4a]. Three points: 1. Eq. 36 is the Klein-Gordon equation; its form is Lorentz invariant [memory, high]. The preferred frame enters only by the stated interpretation (derivatives "referred to the local-ether frame") and by using a Galilean transformation to go to the atom's frame (f6 eq. 20). Result: bound-state energies scale by m0/m while the Bohr-type size is unchanged [calc from f6 eqs. 21-22: kinetic coefficient hbar^2/2m and potential coefficient (m0/m) q Phi both shrink by 1/gamma, so radius ~ hbar^2/(m0 e^2) is unchanged]. That means no Lorentz contraction of atoms or cavities in the model, which is what makes its cavity anisotropy prediction (claim 1) different from Lorentz ether theory. 2. m0 = hbar omega0/c^2 is a calibration: the potential coefficient in eq. 1 was chosen to give the electrostatic force [read: f4a, "calibration"]. The "origin of inertial mass" statement has no independent test in the paper. 3. The augmentation operator U is postulated (form and coefficient chosen), makes the Hamiltonian non-Hermitian, and the derivation is weak-field and low-speed. Spin cannot be treated, yet Hafele-Keating and GPS use caesium and hydrogen-maser hyperfine (spin) transitions: eq. (23) f = f0 sqrt(1 - v^2/c^2) is assumed for them [read: f6 p.5; f7 p.4: "verified only for electronic states"]. The Mossbauer rotor uses nuclear states: assumed too.

b. Experiments. - Hafele-Keating, GPS clock offset (4.45e-10, 38.5 microseconds/day), Ives-Stilwell, ammonia maser, Mossbauer rotor at second order: Su's formulas coincide with SR/GR evaluated in the ECI frame and so agree with the data as SR/GR does [read: f6, f7; calc]. HK values -40 and +275 ns are HK's own predictions, not an independent calculation [read: f6 p.6]. Ives-Stilwell arithmetic reproduces (v^2/c^2 = 7.26e-5 and 7.24e-5), but this confirms SR equally, and the three wavelengths quoted may be representative: [memory, medium] Mandelberg-Witten's precision was about 5%, much coarser than the 0.36% internal agreement; unchecked. - Matter-wave Sagnac and COW gravity phases (eqs. 6-7 of f8b; eq. 27 of f8a): identical to the standard 2 m Omega . S/hbar and to the COW formula [calc, deep]. Standard physics also predicts no separate orbital term (the sidereal rotation already contains the 1/366 orbit contribution; free fall) [memory, high], so "Earth's orbit gives no Sagnac fringe" does not discriminate. - Compton effect as wave scattering (f8c). The deep read shows the derivation's steps are, when written out, Compton's energy and momentum balance with the electron's final (omega, k) = (m c^2, m v)/hbar built into the phase product [read: f8c section 5]. I agree: it re-derives Delta-lambda = lambda_C (1 - cos phi) from the same relations, not from a different principle. Mainstream objections: (i) semi-classical wave treatments of the Compton shift are old (Schrödinger 1927) and are not controversial as formalism [memory, medium]; the dispute is whether the radiation field needs quantising, and the evidence for that is the electron-photon coincidence experiments (Bothe and Geiger 1925; Compton and Simon 1925) and photon antibunching (Kimble, Dagenais, Mandel 1977) [memory, medium; not web-checked]. The paper does not address those, nor angular intensity or cross-section (Klein-Nishina) [read: f8c section 5, absence]. (ii) Su's one new prediction, a weak "direct radiation" at wavelength lambda_C/(gamma - 1), about 2 nm at 90 degrees for 70 pm X-rays [calc, deep], has no detector design, rate or intensity [read: f8c], and the paper gives no calculation that it exists.

c. Tested? The distinctive items: third-order Mossbauer orientation anisotropy (about 1e-18 [calc, deep]; beyond any rotor sensitivity I know [memory, low]); v_0^2/c^2 terms (about 2e-12, needs the lab moved relative to ECI); neutron Bragg-angle shift of about 1 degree. For the last, Su compared with Staudenmann et al. 1980 and wrote "the agreement is not good quantitatively"; in the deep read the measured asymmetry has the opposite sign under his assumed zero, and the zero offset and wavelength are not independent [read: f8a; calc]. Not replicated; no neutral test.

d. If true? The Lorentz transformation would be replaced for matter by a Galilean one plus frame-dependent energy scaling; atoms would not contract; spin and nuclear physics would need new derivations; Compton scattering would require showing the coincidence and statistics experiments follow from a classical field.

Verdict (split). - The relations omega = omega0/sqrt(1 - v^2/c^2), m0 = hbar omega0/c^2, de Broglie relations, matter-wave Sagnac and COW phases: Established (standard physics; the model reproduces it). - Preferred-frame reading of the same formulas (ECI, heliocentric) at the orders tested: Not testable as stated (numerically identical to SR/GR in the cases cited). - f = f0 sqrt(1 - v^2/c^2) extended to hyperfine and nuclear transitions: Plausible, untested as a derivation (agreement is by assumption; SR gives the same). - Compton as pure wave scattering: Never properly tested for the "direct radiation" prediction; the claim that it replaces photon quantisation is Tested, not supported by the coincidence and statistics experiments [memory, medium], which Su does not discuss. - Third-order Mossbauer anisotropy, v_0^2/c^2 terms: Never properly tested.

Claim 4. Magnetism can be Galilean.

a. Mathematics. Internally consistent: the algebra of eqs. 8-11, 21-22, 35-39 checks; units of eq. 39 are consistent as printed (the digest's "1/c^2" form was a transcription error) [read: f4a section 1]. The force law is F = q{-grad Phi - (dA/dt)_m + v_em x curl A}, with matrix-frame (m) derivative and relative velocity v_em, identical to the Lorentz law in the matrix frame at low speed for a neutralised current [read: f4a p.14]. The Galilean "invariance" is by labelling: every quantity is tied to a named frame; it is not form-invariance of the equations [read: f4a section 11]. Fizeau/Fresnel drag: derivation (20)-(25) passes to first order for a non-dispersive medium; the phase-speed formula c/n + v(1 - 1/n^2) is Fresnel's [read: f3a, f3c]. Hidden points: the matrix-frame derivative commutes with the wave operator only for uniform matrix velocity, then applied to rotating loops [read: f3a flag 3]; the Fizeau cancellation of the Earth-rotation term uses (1 - n^2) + n^2 = 1, not shown in the paper [read: f3c].

b. Experiments. In the regime where the law is derived (neutral conductors, low speed) it gives the same forces as the Lorentz force: the magnetic limit of Maxwell theory is already Galilean-covariant (Le Bellac and Levy-Leblond 1973) [memory, medium; read: f4a reader note]. So the statement that "the Lorentz law is valid only under hidden restrictions" is true in the trivial sense that all approximations have restrictions; it does not conflict with experiment. Differences appear in: (i) the electrostatic term (1 + v_em^2/2c^2): in a crossed-field velocity selector the selected speed shifts by 4.6% at beta = 0.3 and 12.5% at 0.5 [calc, deep], while precision beta-ray and e/m work and Bertozzi's 1964 kinetic-energy measurements up to about 15 MeV agree with qV and relativistic kinematics [memory, medium; Bertozzi unchecked on the web]. Su combines the factor with the speed-dependent mass and an inertial term (m0 - q Phi/c^2): the combined prediction at 25 kV is close to the non-relativistic transverse deflection (0.997) where SR gives a 1/gamma reduction of about 5% [calc, deep]. I could not complete a full comparison for high energies; Su calls the high-speed extension "tentative" [read: f2a]. A likely conflict with accelerator data exists but is not shown here. (ii) A moving purely magnetic medium has no drag (Su) versus SR's drag 1 - 1/n^2 with n^2 = mu/mu0: genuinely different, no experiment cited. (iii) Barnett experiment: Su (2005) predicts non-zero induction at a stationary capacitor outside a rotating solenoid, or with a short solenoid [read: Investigator f2d abstract; not read in full].

c. Tested? (i) Not tested as a modified law; high-energy electron data exist but were not analysed against it. (ii) and (iii) never tested. No replication.

d. If true? Maxwell's equations would be Galilean in the matrix frame with an ether frame needed for retardation; the unification of E and B by Lorentz transformation and the success of relativistic electrodynamics at high speed (accelerators, synchrotron radiation) would need independent explanation. The Riemann/Weber-type force law Su ties to it (Found. Phys. Lett. 2006) belongs to a tradition whose velocity-dependent potentials were largely abandoned after Maxwell and Lorentz; it is related work, to be weighed on its predictions.

Verdict (split). - Reduction to the Lorentz force for neutral currents at low speed: Established (same predictions as the standard law). - "The Lorentz force is only valid under hidden restrictions": Not testable as stated (true of any approximation). - Extra electrostatic and inertia terms at high speed: Never properly tested against the modified law; likely in tension with high-energy electron data [memory, medium] (needs a full calculation: Mathematician/Arbiter). - No drag in a moving magnetic medium; Barnett prediction: Never properly tested.

Claim 5. Testable predictions that differ from relativity.

I went through each stated prediction and asked whether it truly differs from SR/GR.

  1. 62 Hz quadrupole cavity term. Differs from SR (SR: zero). See claim 1b. Never properly tested.
  2. One-way-link rotor (eqs. 30-33; claim 5 "sinusoidal turntable phase variation"). Su predicts a phase variation delta-phi = (omega/c^2) v_E . d, with d the link vector, as the table turns. The deep reader's guess was that SR gives a different answer through clock transport [read: f3a flag 5]. I worked it out and the situation is worse for Su than "differs from SR": if the oscillators are atomic clocks, Su's own eq. (23) (f = f0 sqrt(1 - v^2/c^2) with v relative to ECI) makes the two rotating oscillators tick at relative rate 2 v_E v_t/c^2 (v_t the table velocity of each end). Integrated over a rotation this gives V d (1 - cos alpha)/c^2 (V = 465 cos lat, alpha = table angle), which exactly cancels the propagation term V d cos alpha/c^2 up to a constant [calc, my derivation; signs as required by SR]. So within his own model the effect cancels for atomic-clock endpoints, as in SR. His eq. 33 appears to omit the clock-rate term his own eq. 23 implies. It can survive only for oscillators whose frequency does not scale with 1/gamma relative to ECI (he states other dependences "are not precluded" [read: f7 p.4]). Internally inconsistent as stated for atomic oscillators (my calculation; the Mathematician should confirm), otherwise Never properly tested.
  3. Transverse second-order radar Doppler shift -2v^2/c^2 ("a few Hz" at 10 GHz, LEO). I tested this numerically ([calc] straight-line target, stationary transceiver, light-time iteration): the exact SR round-trip ratio (1 - u')/(1 + u'), with u' the radial speed at the reflection event, differs from Su's eq. 34 (1 - 2u - 2v^2 + 4u^2, u taken at the emission instant) by 3e-12 for v^2/c^2 = 2e-8, which is O(v^3): they are the same formula. The "-2v^2/c^2" is produced by evaluating u at emission rather than at reflection: u' = u + w^2/c (w the transverse speed). So the stated discrepancy with "the standard formula" is a bookkeeping convention, not physics [calc]. Refuted by calculation as a prediction that differs from relativity (the arithmetic of "a few Hz" is fine; it is the same in SR).
  4. Constant -11.5 Hz deviation at 2.3 GHz in interplanetary links. Arithmetic correct (2.3e9 x (30 km/s)^2/2c^2 = 11.5 Hz [calc, deep]). Not tested: Su says the Krisher 1990/1993 data contain a fitted oscillator-bias term that absorbs a constant, so they cannot test it [read: f6 p.7]. Two further gaps: (i) the prediction depends on what happens to frequency when a signal crosses from the heliocentric ether into Earth's, i.e. the undefined interface of claim 1; (ii) whether the Sun's potential at the ground clock is applied is unspecified [read: f6 reader notes]. In SR/GR the whole offset is the standard barycentric-to-geocentric time scaling (about 1.48e-8 [memory, high]). Never properly tested (and not well defined).
  5. No drag in a purely magnetic moving medium. Differs from SR; untested. Never properly tested.
  6. About 1 degree neutron Bragg-angle asymmetry. Su compares with Staudenmann et al. 1980: "not good quantitatively" [read: f8a p.8]; the deep read finds the sign opposite under his assumed zero and the zero offset unknown [read: f8a]. A neutron interferometer sensitive to Bragg angle asymmetry at 1 degree is a plain, doable test. Tested, not supported by the one comparison (weakly: the comparison is ambiguous), not replicated.
  7. Third-order Mossbauer-rotor orientation anisotropy. About 1e-18 [calc, deep]; far below any rotor experiment I know [memory, low]. Never properly tested (beyond present reach).
  8. No Sagnac term from Earth's orbital motion. Same prediction in GR (free fall; the sidereal day already includes the orbit) [memory, high]. Not a differentiator. Established as a null result (earthbound loops have never shown an orbital term), but it supports SR/GR equally.
  9. Marsen's proposed low-Earth-orbit and solar-orbit Michelson-Morley experiments (Investigator finding, advocate). In Su's model a cavity at 7.6 km/s relative to ECI would show anisotropy of order 3e-10; SR/GR gives zero. This is a genuine discriminating experiment and is the cleanest proposal on the table. [memory, medium] I know of no flown dedicated satellite Michelson-Morley experiment (proposals such as OPTIS/STAR were not flown); not web-checked.

a-d for claim 5 as a whole. (a) Mostly consistent, with the two failures above (2, 3). (b) Where the prediction is the same as SR, the data agree; where it differs, no data. (c) Brillet-Hall residual (one run, 1979), Staudenmann (ambiguous), Krisher (cannot test). (d) A confirmed rotor, cavity or LEO signal would end exact local Lorentz invariance at the 1e-12 to 1e-10 level for moving labs and force a new reading of SR's equivalence of frames.

Verdict (split by prediction). Cavity 62 Hz, magnetic-medium drag, -11.5 Hz, third-order rotor, LEO MM: Never properly tested. Neutron Bragg 1 degree: Tested, not supported (one ambiguous comparison). Radar transverse Doppler as a difference from relativity: Refuted by calculation. Rotor eq. 33 with atomic clocks: Internally inconsistent (pending Mathematician). Orbital-Sagnac null: Established, not discriminating.

Claim 6. Quasar redshifts: gravitational redshift from the potential of the quasar's own gas cloud, with preferred values from repeated halving.

a. Mathematics. Arithmetic reproduces: R = 4.7e18 m, M = 2.7e46 kg, Phi_g/c^2 = 4.2, z = 2.07 vs observed 2.114, surface g about 0.8% of Earth's; Karlsson and halving sequences reproduced (eq. 11 values 6.08 ... 0.06) [read/calc: f5c]. But: - Domain violation. The formula z = sqrt(1 + 2 Phi_g/c^2) - 1 (eq. 9) comes from a wave equation derived "for weak gravitational potential, Phi_g/c^2 << 1" [read: f5a p.3]; Su applies it at Phi_g/c^2 = 4.2, and at z > 0.414 where, in his words, the Schwarzschild radius exceeds the object's radius (R_s/R = 8.4) [read: f5c; calc]. Eq. 9 differs from the GR form 1/sqrt(1 - 2GM/Rc^2) - 1, which agrees at first order only. An unstated extrapolation. - Free parameter. The density 6e-11 kg/m^3 is chosen; Su says a "slight adjustment in size or density" gives better agreement [read: f5c]. The 0.7 arcsec size is his own estimate from an image; the distance (300 million ly) assumes the quasar is at the galaxy's distance, i.e. the ejection hypothesis [read]. So the z = 2.07 is not a prediction. - The cloud is not uniform in potential. For a uniform sphere the central potential is 1.5 times the surface value [calc: Phi_c = 1.5 Phi_s], giving z = 2.69 at the centre against 2.07 at the surface under Su's formula; emission from throughout the cloud would be smeared over Delta-z of order 0.6, whereas quasar broad lines are about 0.01-0.02 wide in z (velocity widths of order 5000 km/s [memory, medium]). Su places the emitting atoms "near the surface" without justification [read: f5c]. - Dynamics. Mass 2.7e46 kg is about 1.4e16 solar masses, about 2.7e4 times Su's Milky Way figure [read: deep calc]; Newtonian free-fall time of the cloud is about 270 years and escape speed 2.9 c [read: deep calc; my check of t_ff = sqrt(3 pi/(32 G rho)) agrees]. Su's reply to the stability objection (Greenstein-Schmidt) is that surface gravity is weak, yet an escape speed above c is not weak gravity. Rich clusters weigh about 1e14-1e15 solar masses [memory, high]: this cloud exceeds that by one to two orders of magnitude, in a 1000 ly region, with no pressure or rotation support discussed.

b. Experiments. (i) Gravitational redshift in weak fields: Pound-Rebka and GP-A, established. (ii) The ejection hypothesis (Arp, Burbidge) and Karlsson periodicity: the claimed periodicity in a small sample (290 quasars) holds at 3 sigma, but samples of 22,497 (2dF) and 46,420 (SDSS) quasars show none [web: https://en.wikipedia.org/wiki/Redshift_quantization, citing Tang and Zhang 2005, arXiv:astro-ph/0501090], and Hawkins et al. 2002 found no quantisation in 1647 galaxy-quasar pairs. Burbidge and Napier 2001 report the pattern in new samples [web, same page]. This is a live minority position with contrary large-survey evidence. (iii) Standard view of quasar redshifts: cosmological (Hubble flow) and supported by host-galaxy detections and lensing by lower-redshift galaxies [memory, high]. Gravitational lensing and absorption-line systems at lower z in front of quasars fit the cosmological reading; I have not checked the NGC 7319 case against Galianni et al. 2005.

c. Tested? The ejection hypothesis has been tested statistically on large samples with null results by the surveys above, and defended against those by its proponents; Su's mechanism has no test of its own beyond comparing a tunable calculation to one object. Su's own testable item: predicted peaks at higher z (n = -2 to -4) a few percent above Karlsson's, based on quasars "rarely found" [read: f5c]. Not replicated.

d. If true? A new, strong-field gravitational redshift law inconsistent with GR's; quasars would not be cosmological distance indicators, undoing the use of quasar redshift distributions (including the Lyman-alpha forest) for cosmology; clouds of 1e16 solar masses in 1000 ly would need explaining.

Verdict (split). Weak-field gravitational redshift: Established. Strong-field redshift formula applied at Phi_g/c^2 = 4.2: Internally inconsistent (outside its own stated weak-field domain). Su's NGC 7319 cloud as the origin of z = 2.1 and the halving sequence: Refuted by calculation as a physical mechanism under accepted dynamics (collapse in about 270 years, escape speed above c, mass 1e16 solar, potential gradient smears lines; the model it substitutes is untested) with the caveat that the "Karlsson" periodicity it targets is itself Never properly tested in an agreed sense and contradicted by large surveys. The Arp ejection hypothesis, to which the mechanism is attached: Tested, not supported (large-sample surveys), contested by its proponents.

2. Investigator's findings: weighed and kept

Kept in the record whatever my verdict (SOURCE HONESTY). - Su's EPL 56, 170 (2001) Michelson-Morley paper [read: Investigator, f1b downloaded; web: Wayback]. Su's own statement of claim 1b; I used the EPJC treatment (same argument). Not independently re-derived from f1b. - Barnett 2005, Riemann-force 2006, AP-S 2001 abstracts, arXiv physics/0208083 (own work). New testable item: Barnett induction (claim 4/5). Weigh as above. - Gift 2021 (Applied Physics Research 13(1)) [read: Investigator]. Argues GPS shows an Earth-bound ether (ECI) and calls Su's nested ethers "a reasonable working hypothesis". Gift is a preferred-frame proponent. My weighing: GPS operation in the ECI frame is established, but it is equally the standard GR/IAU choice of local inertial frame; so it does not discriminate. Kept as supporting information by another author. - Marsen 2019, 2021, CNPS talk (advocate). (i) LEO and solar-orbit Michelson-Morley proposal: the best-defined discriminating experiment (claim 5, item 9). (ii) Reading of Brillet-Hall's 2-omega residual as a possible positive signal: kept; I could not check Brillet-Hall's reasons; the residual is half the predicted size (ratio 0.55 [calc]), phase unknown. (iii) Source of the c.2007 death date and website removal: one advocate source, no second source found; competing explanations for the silence after 2006 (death, illness, retirement, withdrawal) are the Investigator's list. The 2008 NTHU Press edition (alumni e-news, 9 May 2008) would be posthumous if the date is right; the alumni notice and Su's Chinese name are the deciding records. I add no weight to the death date for physics. - Beckmann (1986/1987, Galilean Electrodynamics): same local-ether postulate reached independently, per Marsen and the CNPS wiki; Su's EPJC text does not cite Beckmann [read: grep]. Physics: Beckmann's version also has light moving at c relative to the local gravitational field; the two share the same distinguishing second-order predictions and the same difficulty with aberration [memory, low: Beckmann's treatment of aberration not checked]. - Field 2019 (IJMPA) and Kipreos 2024 (IJMPA) cite Su: mainstream-journal treatments of preferred frames and of absolute simultaneity compatible with current experiments. Consistent with my point 0.2: a preferred frame is compatible with data when it carries Lorentz-covariant clock and rod behaviour; the hard cases are those where it does not (Su's atoms do not contract). - Engineering citations (TWSTFT, Springer GNSS handbook, SAR synchronisation): use the ECI-frame Sagnac correction; how they treat the ether interpretation is unknown. The correction itself is standard. - Citation counts (Scholar 46 and 32) and no Comment or Reply found (one pass, Cited-by lists blocked by a CAPTCHA): absence of a rebuttal is not evidence of validity; as a record anomaly it is explained by the model's numbers matching standard practice, and by the low profile of the venues for the stronger claims. - CosmoQuest 2025 thread closed without discussion: noted, no weight. - Records not reached: IEEE Xplore AP-S abstracts, APS meeting abstracts, HyRead e-book, scholar.archive.org book PDF, Brillet-Hall paper, Beckmann text. The first three bear on the derivations I mark "stated, not derived".

3. Corrections to Step 1

  1. claims.md claim 5 and the radar prediction. claims.md lists a "-2v^2/c^2 transverse second-order radar Doppler shift (a few Hz for a 10 GHz low-orbit satellite)" among predictions that "differ from relativity". My numerical check ([calc] above) shows the exact SR round-trip ratio equals Su's expression once u is evaluated at the emission instant; the "discrepancy" is the choice of evaluation time. Correct to: Su's formula is numerically equal to SR; no distinguishing prediction.
  2. claims.md claim 5, rotor. The deep read (f3a flag 5) suggests SR cancels the rotor term through clock transport while Su's model, with "no time dilation of moving clocks", predicts the full term. Su's model does have clock-rate scaling: eq. 23 of f6 and eq. 8 of f7, f = f0 sqrt(1 - v^2/c^2) with v relative to ECI. It implies the same cancellation for atomic oscillators. Correct: Su's eq. 33 omits a term his own transition-frequency formula implies.
  3. claims.md claim 1 "Sagnac needs the first-order term; relativity cannot supply it". This is Su's argument (EPJC section 4); the Step 1 text reports it faithfully, but it should not stand as physics: first-order Sagnac is the standard SR/GR result in an inertial (ECI) frame. Su himself cites Ashby. The word "Consequently" in claim 1 (explanation vs equivalence) should read: same numbers, different interpretation.
  4. claims.md claim 2. The "230 microseconds" and deflection numbers are GR's own; "reproduces without curvature" is by choice of the factor 2, not derived. Add "by construction".
  5. deep/f3a reader note on rotor says "probably differs from SR because Su has no clock time dilation". Corrected by item 2.
  6. Claim 3 wording "mass is a frequency ... referred to the local-ether frame". Add: the equation is Lorentz invariant in form (Klein-Gordon); the frame enters only through interpretation and the Galilean step; atoms do not contract in the model.
  7. Brillet-Hall. The Step 1 deep read quotes "about 1.5e-15 limit" from memory; web-confirmed as Delta-c/c = (1.5 +/- 2.5)e-15 [web]. Note that this limit applies to sidereal signals, not to a lab-fixed 2-omega term; the 34 Hz peak-to-peak term (3.8e-13) is far larger than that limit and is not constrained by it.
  8. claims.md claim 6. Add: eq. 9 is outside its own weak-field derivation; the cloud's potential is not uniform (central z = 2.69 vs surface 2.07 under Su's formula [calc]); large-survey null results on quasar periodicity exist [web].
  9. Minor numerical notes retained: GPS orbital/rotational speed ratio 64 not "about 100"; GPS gravitational delay 42.2-62.2 ps vs Su's 43-60; 4GM_Sun/c^3 = 19.7 not 19.5 microseconds; none changes a verdict.

4. What would settle the open items

  • Lab-fixed 2-omega term: a modern turntable cavity experiment, with the lab-fixed component reported and cross-checked at several latitudes (prediction scales as cos^2 latitude, 7e-13 at 40 degrees), plus reversal of tilt and gravity sag. Possibly the existing data of rotating-cavity groups already contain it as a "constant" term.
  • A cavity or clock-comparison experiment on a satellite (Marsen's proposal): prediction 3e-10 at LEO; GR zero.
  • Rotor test with oscillators not subject to 1/gamma scaling, or confirmation that atomic ones cancel.
  • Interface rule between local ethers (Su has none): until then aberration, the -11.5 Hz prediction and the boundary of "Earth's ether" are undetermined.

5. Verdict summary

  1. Local ether: ECI Sagnac and no orbital term, Established; ether as explanation, Not testable as stated; 62 Hz cavity, Never properly tested; annual aberration, Tested, not supported.
  2. Gravity as optical index: numbers Established (GR's own, by construction); replacement of curvature, Not testable as stated.
  3. Matter waves: standard relations Established; preferred-frame reading Not testable as stated; hyperfine/nuclear extension Plausible, untested; Compton wave-only reading Tested, not supported (memory, medium); third-order effects Never properly tested.
  4. Galilean magnetism: low-speed neutral reduction Established; extra terms and magnetic-medium drag Never properly tested (tension with accelerator data suspected, memory, medium).
  5. Differing predictions: mostly Never properly tested; radar Doppler Refuted by calculation (not different from SR); rotor Internally inconsistent for atomic clocks; neutron Bragg Tested, not supported; orbital-Sagnac null Established but not discriminating.
  6. Quasars: weak-field redshift Established; Phi/c^2 = 4.2 use Internally inconsistent; NGC 7319 mechanism Refuted by calculation under accepted dynamics; ejection hypothesis Tested, not supported (contested).