RankingsPrimerArbiterPhysicistMathematicianContrarianQESites logClaimsSourcesResearch notes

Su-CC: Science Contrarian review (Step 2)

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

Method. I judge by experimental evidence, not consensus. I read the Step 1 files (facts.json, claims.md, all deep reads), the Investigator's sites.md and research notes, and the source passages named below: Su's EPJC 2001 deep read with its page-image checks, Su's EPL 2001 paper (f1b, downloaded by the Investigator; I read pp.1-3 myself), Su's Barnett paper (f2d; I read the introduction, the Kennard passage and the conclusion), and the relevant sections of every other deep read. Tags: [read] from a document in the folder or research/files, [web] with URL, [memory] with confidence. Su is the originator and an interested party; his accounts of special relativity, GPS practice and other authors' experiments are his characterisations. Web checks used: 5 of about 6 budgeted.

Source gap (repeated from claims.md). The 2005/2008 book is present only as an outline and web summary. None of the experimental papers Su cites as support (Brillet and Hall, Saburi et al., Krisher et al., Staudenmann et al., Mandelberg and Witten, Cedarholm et al., Champeney et al., Turner and Hill, Hay et al., Kennard, Barnett) is in the folder. Where I discuss them the content rests on Su's citation or on [memory], flagged each time. Updated: Su's EPL 2001 Michelson-Morley paper (f1b) is now in research/files/wayback-qem/ and I have read it.

Corrections to Step 1

  1. claims.md, source-gap paragraph: it lists Europhys. Lett. 56, 170 as "cited but not in the folder". The Investigator obtained it (research/files/wayback-qem/f1b.pdf) [read: p.1-5]. It repeats the EPJC argument (GPS range formula R = Rt + 2S.omega_E/c, "about 100 times" orbital to rotational speed, Michelson-Gale 0.2 km^2, heliocentric frame for interplanetary radar). Nothing in it changes a claim. The same Step 1 flag applies to it: orbital speed (30 km/s) over rotational speed at the equator (464 m/s) is 64, not "about 100" [calc, deep read A-local.. section (b)].
  2. claims.md claim 6: "for a quasar near NGC 7319 (0.7 arcsec ...)" blurs two numbers. In the paper 0.7 arcsec is Su's own estimate of the quasar's angular diameter from an HST image; the quasar lies about 8 arcsec from the galaxy nucleus [read: deep f5c section 2 item 4 and section 4]. Also the density (6e-11 kg/m^3) is an assumed free parameter, and Su says "a better agreement can be reached simply with a slight adjustment in the size or density" [read: deep f5c section 4]. So "gives z = 2.07 against the observed 2.114" is a fit, not a prediction. claims.md should say so.
  3. claims.md claim 5: the "-11.5 Hz at 2.3 GHz in interplanetary spacecraft links" is listed as testable. Su himself says the existing Krisher et al. data cannot test it because the onboard quartz oscillator ages and a constant term is fitted: "it seems not appropriate to test the predicted constant deviation from this experiment" [read: deep f6 section (d)]. It is a prediction no existing data address, not one the data have been compared with.
  4. claims.md claim 5 and claim 3 list Hafele-Keating, GPS, Ives-Stilwell, ammonia maser, rotor (second order), Fizeau, loop Sagnac as things the model "accounts for". The deep reads show each is numerically identical to special plus general relativity computed in the ECI frame (the -40 and +275 ns of Hafele-Keating are HK's own 1972 predictions that Su adopted, not an independent calculation [read: deep f6 (b)]). Nothing in that list is evidence for the model over relativity. claims.md says this once in the claim 5 paragraph; later steps should carry it into every claim it touches.
  5. Step 1 sites.md and claims.md describe the Brillet and Hall residual as "about 34 Hz peak-to-peak attributed to a spurious signal". That is Su's account of a paper nobody in the pipeline has read. [web] confirms only the paper's headline result: Brillet and Hall, Phys. Rev. Lett. 42, 549 (1979), set Delta c/c = (1.5 +/- 2.5) x 10^-15 (https://www.arxiv.org/pdf/gr-qc/0210106 lists the figure; the JILA record is https://jila.colorado.edu/bibcite/reference/10790). The 34 Hz term, its phase and the reasons it was called spurious remain [read: Su's paraphrase only]. Marsen's statement that it "was dismissed as spurious and averaged out" is likewise second-hand (see Investigator findings below).
  6. Deep read of EPJC, Su's critique of relativity section: the deep read puts the stellar-aberration admission on p.12, not p.13. I agree with that correction; the admission also recurs in the conclusion.
  7. f5b: Su writes that light deflection was "first confirmed in as early as 1917". The deep read flags this as an error (1919 eclipse expedition) [memory, high]. Minor, but later steps should not repeat 1917.

No Step 1 error changes a central claim.

What the Investigator found, and how I weigh it (kept whatever my verdicts)

Priority 1, Su's own work. f1b (EPL 2001) restates the Michelson-Morley argument; f2d (Barnett, 2005) adds a new testable prediction (below); f2c (Riemann force law, 2006) ties Su to the Weber-Riemann tradition of velocity-dependent electrodynamics; f2b and f3b are the 2001 conference originals of the force law and modified Maxwell equations; the book's 2002 contents list "Unsolved and Predicted Propagation Phenomena: stellar aberration, one-way-link rotor, second-order radar Doppler shift" [read: contents.htm]. That list matches my claim 1 and claim 5 reading: Su lists his own open problem and his own discriminating tests. Not obtained: the railgun recoil papers (IEEE Trans. Magn. 42, 2193, 2006), JEWA 15, 945 (2001), Electron. Lett. 36, 1812 (2000), the book.

Priority 2, support by others. Gift (2021, Applied Physics Research 13(1)) argues from GPS timing for an Earth-bound ether identified with the ECI frame and calls Su's nested ethers "a reasonable working hypothesis" [read: Gift 2021, via research note]. Gift is a proponent of preferred-frame readings of GPS; his reading of the data is his characterisation. Marsen (viXra 1906.0336, 2105.0110; CNPS talk 2021) is an advocate who proposes a Michelson-Morley experiment in low Earth orbit and solar orbit and reads the Brillet-Hall residual as a possible signal [read: Marsen 2019]. Engineering papers cite Su's EPL formulation (TWSTFT 2015-16, a Springer GNSS handbook chapter, SAR synchronisation 2024-25) [web: OpenAlex, per research note]. How they use it (as a reference for the ECI Sagnac correction, which is standard, versus endorsing the ether) is unread. I treat these as using the formula, not testing the interpretation, but I note that is my inference; the texts were not read. Google Scholar shows 46 citations of EPJC 2001 and 32 of EPL 2001 against about 20 and 10 in OpenAlex [web: per research note], so the citing list is incomplete.

Priority 3, related work. Beckmann's 1986 local-ether proposal (the same postulate, reached independently according to Marsen and the Natural Philosophy Wiki; Su's EPJC text does not cite Beckmann [read: grep, per research note]); Field (2019, IJMPA, preferred frames, Sagnac and Hafele-Keating, cites Su); Kipreos (2024, IJMPA, an alternative transformation without differential simultaneity compatible with current relativistic experiments, cites Su). These show the underlying question (whether Lorentz symmetry's "differential simultaneity" has direct experimental support) is being published on in mainstream journals. That is a point in the model's favour as a question, not as an answer.

Priority 4, historical record. Su's record ends in 2006. One advocate source (Marsen 2021) says Su "passed away in approximately 2007"; no obituary has been found; the book appeared from NTHU Press in March 2008 with an ISBN registered in 2005. No published Comment or Reply to EPJC or EPL was found in one Google query; Scholar's cited-by lists were blocked. I analyse this under "How the claims were received" below.

How the claims were received (the record of rejection)

  • Published, not suppressed. The core papers appeared in refereed journals: Eur. Phys. J. C 21, 701 (2001); Europhys. Lett. 56, 170 (2001); Eur. Phys. J. B 24, 231 (2001); J. Electromagn. Waves Applicat. 15-19 (2000-2005); Found. Phys. Lett. 2006 [read: facts.json; web: Crossref per research note]. Su held a faculty post from 1985 and had mainstream electromagnetics output before 2000 (IEEE MTT 1988, 1993; IEEE AP 1989) [web: OpenAlex per research note]. So the standard "blocked from publication" narrative does not apply here. The quasar paper is only an arXiv preprint (2006); whether it was ever published is unknown [read: deep f5c].
  • Ignored rather than refuted. The Investigator found no Comment/Reply and no direct rebuttal. With the record this thin, silence is consistent with at least three explanations: (i) the arguments were judged to reproduce relativity numerically and so to be an interpretation with nothing to test; (ii) Su's death or withdrawal ended the exchange; (iii) the journals' readers did not engage. Evidence for (i): Su's own text says most results are "identical" to the relativistic formulas. Evidence for (ii): only Marsen. Evidence on (iii): none. What would decide: the cited-by lists of EPJC and EPL (blocked), referee reports (not obtainable), and NTHU records.
  • Experiments that touch Su's discriminating predictions were not designed to test them. The 62 Hz cavity prediction has one old comparison (Brillet-Hall, per Su), described by Su as "too early to make a decisive conclusion". The neutron Bragg prediction has one comparison, which Su calls "not good quantitatively". No other prediction has any comparison at all. That is "never properly tested", not "tested and failed".
  • The reasons given for rejection of his side claims (annual aberration unsolved, ad hoc steps in the wave equation) are ones Su himself states. I found no case of data reanalysed or omitted against Su; the interesting rejection story, if any, is that of the anomalous residual in Brillet-Hall (below), which is not Su's data.

Claim 1. A local ether, not a universal one, carries light

Split into a weak form (ECI-frame bookkeeping for earthbound propagation, no Earth-orbit term) and a strong form (a physically preferred frame: light speed is isotropic only in the ECI frame, so Earth's rotation gives measurable second-order, lab-fixed anisotropy).

a. Mathematics. The range, Sagnac, Doppler and Michelson-Morley formulas (eqs. 1-13, 18, 20-26) are internally consistent and reproduce when recomputed: pi/3 versus 2.4e-4 of that; 1.2e-12 times cos^2(latitude); 62 Hz at 40 degrees; 2 rad for 0.2 km^2 [read: deep EPJC arithmetic table; I re-derived the 62 Hz and the 1.047 rad]. The classical-ether arm times of eq. 13 are the textbook 1887 results; the only novelty is which frame v is measured in. One internal gap: the frame choice for interplanetary radar is argued from the frame the published analyses used, and Su's own 1 microsecond agreement (1e-9 fractional) is finer than his stated 1e-8 discrimination threshold between geocentric and heliocentric frames, yet no geocentric calculation is shown to fail [read: deep EPJC (c)]. Su's own "about 100 times" is 64. b. Established experiments. First-order Sagnac data (GPS range correction, 0.3 microseconds in the Japan-USA link, Michelson-Gale 1925, fibre gyros) fit, but they fit special plus general relativity in an ECI frame with the same formula (Su acknowledges Ashby, his ref. [7]); they do not pick the ether. The step "GPS works without an orbital term, therefore the ether is Earth-bound" is weaker than Su presents it: in relativity a local inertial frame is chosen for convenience and any frame gives the same observable, so GPS success does not show that the ECI frame is physically unique [memory, high; deep EPJC point 1]. The distinctive content is a constant, lab-fixed second-order term, 1.2e-12 x cos^2(latitude), in round-trip propagation. Modern rotating-cavity experiments are the relevant data: Herrmann et al., Phys. Rev. D 80, 105011 (2009), compared two orthogonal cavities on a rotating turntable and set Delta c/c at about 1e-17 over a year [web: https://arxiv.org/pdf/1002.1284]. But those analyses look for modulation at the turntable rate and at sidereal sidebands. Su's predicted signal is tied to geographic east (the velocity of the lab in the ECI frame), so it is fixed to the laboratory and appears as a constant-phase term at twice the turntable rate, which is exactly what experimenters treat and subtract as a systematic (tilt, magnetostriction, gravity) [memory, medium; the original papers are not in the folder]. Su constructs the model so that every sidereal-modulation and annual-modulation test is null: he says so for Hughes-Drever, the maser and GPS [read: deep f6 (e), f7 section 7]. I record this honestly in both directions: it is why the model survives the null tests, and it is also why those null tests give it no support. Annual aberration: Su states it is "not yet solved" once Earth's local ether is included; the interface between two local ethers in relative motion is undefined [read: deep EPJC (g), p.12]. In the classic reading (full ether dragging, Stokes) the observed 20.5 arcsec annual aberration would vanish [memory, high]. This is an observed phenomenon that the strong model, as stated, does not yet reproduce. c. Tested? By whom? The only quantitative comparison for the strong form is Brillet-Hall (1979): Su's prediction 62 Hz peak-to-peak at 3.39 micrometres, 40 degrees latitude; a term at the predicted rate was reported at about 17 x 2 Hz and attributed to a persistent spurious signal [read: Su, EPJC p.9, via deep read]. The measured term is 0.55 of the prediction, neither zero nor equal. Brillet and Hall's own paper has not been read by anyone in the pipeline; their headline limit [web] was (1.5 +/- 2.5) x 10^-15 on the modulated anisotropy. A lab-fixed term cannot be told from systematics by that experiment alone. Not replicated, never dedicated. A clean test exists and has apparently never been done: orient the whole apparatus differently on the floor (a lab-fixed systematic moves with it; a geographic-east term does not), repeat at two latitudes (Su: proportional to cos^2 latitude, zero at a pole), or put the interferometer in low Earth orbit where the relevant speed is about 7.6 km/s and the effect about 600 times larger than on the ground [calc: (7.6e3/c)^2 = 6.4e-10 against 1.2e-12 x cos^2]. This is Marsen's proposal [read: Marsen 2019]; Marsen is an advocate, but the arithmetic is Su's formula. No satellite cavity comparison of this kind has been located in the review files. I did not search for one; say unknown. d. If true. Light has a physically preferred frame at Earth (and a nested hierarchy: planet inside star inside galaxy), the relativity principle fails at second order, the Lorentz transformation is not fundamental, and annual aberration needs an interface theory. The first-order GPS and Sagnac practice would not change.

Verdict (split). - Weak form (ECI bookkeeping, no orbital Sagnac term): Established as practice, but not as evidence for an ether; it coincides with relativity and cannot discriminate (Not testable as stated as a distinction). - Strong form (preferred frame; lab-fixed second-order anisotropy of about 1e-12): Never properly tested. Not dismissed after testing; there is no published refutation and no published dedicated test. One old, partly ambiguous comparison. Annual aberration unresolved by Su's own admission.

Claim 2. Gravity is an optical index, not curvature

a. Mathematics. Eq. 28 (n_g = 1 + 2GM/(c^2 r)) is the weak-field effective index in isotropic coordinates with gamma = 1 [memory, high; deep f5b]. Derived quantities reproduce: deflection 4GM/(c^2 b) = 1.75 arcsec; GPS delay 42-62 ps; Earth-Venus 232.5 microseconds against Su's 230; Su's 4GM/c^3 = 19.5 microseconds is 19.7 recomputed [read: deep f5b]. The 19.5 microsecond "disagreement" with Kenyon's textbook is probably a coordinate-convention term (the "+1" in some textbook forms), not an observable [memory, medium; the deep reader asks reviewers to confirm and I did not]. Internal inconsistency arises in claim 6, below: the index and the transition-frequency formula were derived assuming Phi_g/c^2 much less than 1, and are applied at Phi_g/c^2 = 4.2. b. Experiments. Shapiro (1971), Anderson (1975), Reasenberg (Viking, 1979) agree with the coefficient (1+gamma)/2 = 1; recent: Cassini, Bertotti et al., Nature 425, 374 (2003): gamma = 1 + (2.1 +/- 2.3) x 10^-5 [web: https://physicsworld.com/a/general-relativity-passes-cassini-test/ and the IDEAS record https://ideas.repec.org/a/nat/nature/v425y2003i6956d10.1038_nature01997.html]. Su's formula equals that coefficient by construction, so it fits to the same accuracy. Pound-Rebka, Hafele-Keating and GPS clock-rate offsets also fit: Su re-reads redshift as a lowering of atomic transition energies by potential rather than a photon change [read: deep f5c section 2 item 3]. At first order that is the same prediction as relativity. c. Tested? By whom? The numbers were tested, by others, as tests of general relativity; Su's model supplies the same numbers. Nobody has tested the model against relativity because they agree to the order measured. Su names the claimed differences (heliocentric versus barycentric frame; reference instants) and gives no size for them and proposes no test [read: deep f5b section 6]. The Sun's displacement from the barycentre (about one solar radius at about 10 m/s) is an effect the paper does not estimate [memory, high; deep f5b]. d. If true. Spacetime curvature would be unnecessary at first order; light would be a wave in a gravitationally graded medium; the equivalence principle would be replaced by the identity of the natural frequency in inertial and gravitational coupling (Su's derivation, claim 3). Beyond first order (perihelion precession, strong fields, gravitational waves, frame dragging) the paper makes no claims, and gravitational waves, light-bending near non-spherical bodies, and Mercury's perihelion are not discussed in the folder [read: deep f5a section 7, absence checked by the deep reader for GR terms]. Whether the optical analogy extends is an open (and for the second-order terms, probably decidable by existing data) question.

Verdict (split). - Weak-field, first-order (deflection, Shapiro delay, GPS delay, redshift to first order): Established numerically, as the general-relativistic values; the claim that no curvature is needed is Not testable as stated at that order (no measurable difference). - Beyond first order and in strong fields: Plausible, untested in the weak-field regime; as applied to quasars at Phi_g/c^2 = 4.2 it is Internally inconsistent (see claim 6).

Claim 3. Matter waves obey the same preferred-frame equation, and mass is a frequency

a. Mathematics. Eq. 36 is the Klein-Gordon equation with omega0 = mc^2/hbar; the derivations of k = omega v/c^2 (group velocity), omega = omega0/sqrt(1 - v^2/c^2), m = hbar omega/c^2 and f = f0 sqrt(1 - v^2/c^2) check symbolically [read: deep f6 (a), f8a]. m0 = hbar omega0/c^2 is a calibration of the potential coefficient, not an independent result (Su says the coefficient was chosen to give the electrostatic force) [read: deep f4a section 12]. The augmentation operator U is postulated, not derived; the ordering q Phi U makes the Hamiltonian non-Hermitian and the Hermitian Ehrenfest formula is used anyway [read: deep f4a]. The Compton paper, read for what it derives: the energy and momentum balance is built into the assignment of (omega, k) = (mc^2, mv)/hbar to the second state and the phase-product step; the result is Compton's equations rewritten [read: deep f8c section 5]. Spin cannot be treated, yet the clocks tested (Cs, H maser) are spin (hyperfine) transitions, so the speed law for them is an expectation [read: deep f6]. b. Experiments. Hafele-Keating, GPS clock rates and synchronism, Ives-Stilwell (Su's numbers reproduce: 7.264e-5 and 7.238e-5, 0.36% apart), the ammonia maser null, and the Mossbauer rotor at second order are all reproduced exactly as in special relativity; the algebra checks [read: deep f6, f7]. So these are consistent with established experiments, and equally with relativity. Where Su differs: (i) a third-order rotor orientation anisotropy of about 1e-18, roughly five orders of magnitude below the Fe-57 line width [calc, deep f7 section 6a]; (ii) a v0^2/c^2 term of about 2e-12; (iii) the -11.5 Hz constant offset (Su says existing data cannot test it); (iv) the neutron Bragg angle difference of about 1 degree. Citations to rotor experiments contain no numbers, Kundig 1963 is not cited, and the deep reader recalls that Champeney et al. and Turner-Hill were first-order null tests, not tests of (v_r^2 - v_t^2) [memory, medium/low; to check]. c. Tested? The Staudenmann et al. (1980) neutron data are the only comparison with a distinctive prediction, and it is Su's own reading of two numbers from that paper. Predicted angles +20.48 and -19.11 degrees recompute; measured +19.14 and -20.37 under Su's assumed zero, so the asymmetry has the opposite sign, while with the opposite zero it matches within 0.03 and 0.11 degrees. The zero-independent quantity, the separation 140.49 degrees, gives theta_B = 19.755 degrees against 19.761 degrees for the standard Bragg law, whereas Su's model predicts 140.39 degrees; and the quoted wavelength 1.134 angstrom is what the standard law gives from the same two readings, so it may not be independent [read: deep f8a section 3.3, own recomputation by the deep reader]. This is not a controlled test either way: the zero of the crystal-rotation scale is unspecified. It leans against Su's size of effect but cannot be called a refutation. Su says himself "the agreement is not good quantitatively." The experimenters used in the data (Werner, Colella, Overhauser) did not test it. d. If true. Mass would be a de Broglie frequency of a field in a preferred frame; inertial and gravitational mass identity would be derived rather than postulated; matter-wave Bragg and interference angles would depend on lab orientation to Earth's rotation at first order in v0/vp. That last item is large: v0/vp is 0.10 for cold neutrons, a change of about 3% in sin(theta_B). In standard quantum mechanics (Galilean covariance for slow particles) the Bragg angle depends on velocity relative to the crystal and Earth's rotation enters only via the much smaller Sagnac-type term of order Omega L/vp [memory, high; deep f8a]. I note, for the Physicist, that precision neutron crystal spectrometers measure both +theta and -theta routinely and would be expected to reveal a first-order east-west asymmetry if one existed [memory, low; check]. Su's matter-wave Sagnac formula (eq. 27) is the standard 2m Omega.S/hbar and is not distinctive.

Verdict (split). - Algebra and the second-order phenomena (Hafele-Keating, GPS clock rates, Ives-Stilwell, maser, rotor): Established numerically (identical to relativity in the ECI frame); the local-ether reading is Not testable as stated at those orders. - Distinctive predictions (Bragg angle asymmetry, third-order rotor anisotropy, v0^2 term, -11.5 Hz): Never properly tested. The one data comparison, on Su's own account "not good quantitatively", is ambiguous (sign reversed under his orientation; separation favours the standard law; wavelength possibly circular). The third-order rotor effect is probably beyond reach of current Mossbauer technique [memory, low]. - "Mass is a frequency" as an origin of inertia: Not testable as stated (a calibration m0 = hbar omega0/c^2). - Compton as wave scattering: reproduces Compton by construction; its single new prediction (a weak soft X-ray line at a few nanometres, about 7 nm at 45 degrees for 70 pm X-rays [calc, deep f8c]) was never tested; no experiment is described. Never properly tested.

Claim 4. Magnetism can be Galilean

a. Mathematics. The augmented potentials (1 + v_es^2/2c^2) Phi and -(v_es/c^2) Phi, the force law of eq. 39 and its reduction to the Lorentz law for a neutralised matrix at low speed all check, and eq. 39 is dimensionally consistent as printed in SI (the digest's version was an OCR error) [read: deep f4a section 1]. The "Galilean invariance" is invariance by labelling: each velocity and time derivative is attached to a named physical frame. In the quasi-static limit the ether drops out and the law depends only on relative velocities [read: deep f4a section 11]. The Fresnel drag c/n + v(1 - 1/n^2) comes out of the polarisation-current derivation at first order for a non-dispersive medium (every step passes) [read: deep f3a flags, f3c]. Unstated assumptions: uniform medium velocity, then applied to rotating loops; third-law claim holds only for instantaneous interaction; dispersion correction outside the model [read: deep f3a flag 3, f2a]. b. Experiments. In the regime where it is derived (neutral matrix, low speed, quasi-static) it gives the same force as F = q(E + v x B), so existing evidence for the Lorentz law is not evidence against it. Differences the paper's equations imply: (F1) electrostatic force multiplied by 1 + v^2/2c^2, about 4.6% at beta = 0.3 in a crossed-field velocity selector; (F2) a force on a charge at rest near a neutral steady current, of order v_sm^2/c^2; (F3) an inertial mass shifted by the absolute potential, about 20% for a 100 kV electron [calc, deep f2a section 11]. A direct high-speed check exists: Bertozzi (1964), kinetic energy measured by calorimetry for electrons accelerated through known potentials from 0.5 to 15 MeV, time of flight giving a limiting speed c, in agreement with relativistic mechanics and qV [web: https://en.wikipedia.org/wiki/Tests_of_relativistic_energy_and_momentum]. Su's high-speed form of F1 is called "tentative" in the paper [read: deep f2a]. Whether Bertozzi rules out F1 at high speed needs to be worked through by the Physicist; I did not do it. For the Barnett case: Barnett (1912) found a minute fraction of the expected charge with the capacitor inside a rotating solenoid; Su agrees his law gives zero there. Kennard (1912, 1917) tried the capacitor inside or outside; Su says "negative results were also reported, except for the disturbances ... when the magnetization was reversed", and then writes of Kennard's outside-capacitor case that a nonvanishing signal "was actually observed when the magnetization was reversed" but "cannot discriminate" between induction and magnetostatic force; and that his prediction "seems not to be reported before" [read: f2d p.1, p.6-7]. I could not check Kennard's paper. Su's reading of Kennard's outside-capacitor result is therefore single-source (Su) and ambiguous; it should not be reduced to "already refuted" or "supported" until Kennard (1917) is read. c. Tested? By whom? Not by Su's law. The Barnett and Kennard experiments (1912-1917) are old and were interpreted at the time under other theory; their outside-solenoid or short-solenoid geometries are the closest tests of Su's new prediction. No replication with modern instruments is recorded. The railgun recoil (Riemann/Weber force) papers, 2005-2006, are the other Su claim that touches experiment; not obtained. Related tradition: Ampere and Weber longitudinal forces, which mainstream considers superseded and which I note as related work whose label is mainstream's characterisation. d. If true. The Lorentz force law would hold only under restrictions; a force on a charge at rest by a neutral steady current would exist; electrostatic force and energy gain in accelerators would depend on speed relative to the electrodes. Maxwell's equations as modified would keep the same wave solutions in the ether frame.

Verdict (split). - Low-speed, neutral-matrix limit: Established (reduces to the Lorentz force law). The claim that the Lorentz law is itself "not Galilean invariant" and so needs this replacement: Not testable as stated (it is a choice of what is held fixed between frames; a deep-read note says the textbook magnetic limit is Galilean covariant [memory, medium]). - Extra terms F1-F3 and the Barnett outside/short-solenoid prediction: Never properly tested; F1 at high speed is probably tested against it by accelerator energy gain (Bertozzi 1964), which I flag as needing a worked calculation before it could be called Tested, not supported. - Moving purely magnetic medium has no drag: Never properly tested (no experiment cited).

Claim 5. The model makes testable predictions that differ from relativity

a. Mathematics. Each prediction was recomputed: 62 Hz; the Fizeau/Sagnac coefficients; the one-way turntable phase 1.86e-3 degrees per GHz-metre (0.14 degrees for 10 GHz over 10 m at 40 degrees); the transverse radar Doppler shift -2v^2/c^2 (up to 12.5 Hz at 10 GHz for a LEO pass, "a few Hz" is a low-side statement); -11.5 Hz for 30 km/s at 2.3 GHz; third-order rotor term about 1e-18 [read: deep EPJC (e),(g); f6 (d); f7]. The one-way turntable test depends on how the two ends' clocks behave when moved; the paper does not address clock synchronisation [read: deep EPJC (g), f3a flag 5]. b. Established experiments. Where Su's formulas apply at first order (radial radar Doppler eq. 35) they agree with the Lorentz result; transverse second-order Doppler is where they differ. c. Tested? List, honestly: (1) 62 Hz cavity: one old comparison, ambiguous (above). (2) Turntable one-way phase, transverse radar Doppler, -11.5 Hz, third-order rotor: no data compared. (3) Neutron Bragg about 1 degree: one comparison, ambiguous. (4) No Sagnac term from Earth's orbit: no loop interferometer has reported one; but relativity also predicts none [memory, high; deep f8a, f8b], so this prediction does not discriminate. (5) Moving magnetic medium: none. (6) Barnett: Kennard 1912/1917 (single source: Su). (7) Orbital Michelson-Morley (Marsen): proposal only. Two transverse radar Doppler measurements bear: any precision satellite or spacecraft Doppler tracking data would constrain -2v^2/c^2, and modern tracking is analysed with standard Doppler formulae; I did not locate a published comparison and say unknown. None has been replicated. d. If true. Any one confirmed prediction would show a physically preferred frame and break the relativity principle at second order.

Verdict. Never properly tested (for items 1-3, 5-7). Item 4 is Not testable as a discriminator. The claim as a whole is credible as a research programme: Su states falsifiable predictions with numbers, which is a strength, and has been honest about the one unfavourable comparison.

Claim 6. The same gravity-as-potential idea explains discordant quasar redshifts

a. Mathematics. The arithmetic reproduces: z = 2.058 with D = 1000 ly and 2.108 with the unrounded 1018 ly; Phi_g/c^2 = 4.2; surface g about 0.8% of Earth's; Su's fragmentation formula (11) and the 14 predicted values; the Karlsson values (10) [read: deep f5c sections 4, 6, own checks]. Internal problems: - Weak-field versus strong-field. The potential coupling n_g = 1 + 2 Phi_g/c^2 and the frequency law f = f0 / sqrt(1 + 2 Phi_g/c^2) were obtained for Phi_g/c^2 much less than 1 (explicit in the weak-potential assumption of f5a, Su's own statement [read: deep f5a section 8]). The NGC 7319 case uses Phi_g/c^2 = 4.2, and the Schwarzschild radius is 8.4 times the object's radius. In general relativity the surface of such an object is inside its own horizon; in Su's model a finite z is allowed because the formula never diverges [read: deep f5c section 3]. That is a choice of extrapolation, not a derivation. - Stability. The Newtonian free-fall time of a uniform cloud at 6e-11 kg/m^3 is about 270 years [calc, deep f5c section 5] and the Newtonian escape speed about 2.9 c. Su replies that gravity is weak at the surface (under 1% of Earth's), but gives no pressure, rotation or timescale; the mass is 2.7e46 kg, about 2.7e4 times his own Milky Way figure of 1e42 kg [read: deep f5c section 5]. I find nothing in the paper that resolves this. - Parameter freedom. Density is an assumed value; "a slight adjustment in size or density" brings z to 2.114. The calculation shows that the formula can fit, not that it predicts [read: deep f5c section 4]. - Karlsson series. Su's halving model gives ratio 1.26, with z0 = 1.956 taken from Karlsson; it matches Karlsson peaks near n = -1 to 2 and replaces 0.60 with 0.71 and 0.49, and gives nine values between 1.96 and 0.06 where Karlsson has five. His 0.22, 0.14, 0.09 are not supported by any citation [read: deep f5c section 6]. b. Experiments. Arp, Burbidge and others reported low-redshift galaxies with high-redshift quasars aligned and bridged by filaments; Karlsson 1971/1977 and Arp et al. 1990, Burbidge and Napier 2001 reported periodicity in log(1+z), all in refereed journals (ApJ, A&A, AJ) [read: deep f5c section 2, from Su's citations]. This is genuine evidence offered by astronomers with mainstream careers and I record it. Against: Hawkins, Maddox and Merrifield tested a periodicity in 1647 QSO-galaxy pairs from the 2dF surveys and found no evidence for a periodicity at the predicted frequency "or at any other frequency" [web: https://arxiv.org/pdf/astro-ph/0208117]; further SDSS-based analyses (the same search result also lists astro-ph/0506366 and arXiv:0807.2641, titles "Critical Examinations of QSO Redshift Periodicities..." and "Evidence against non-cosmological redshifts of QSOs in SDSS data"), which I did not read. Su cites NEQ3, MCG 7-25-46 and NGC 7320 himself, and for NGC 7320 records that the original authors saw the redshift break as possibly evidence of projection, not association [read: deep f5c section 2 item 12]. c. Tested? Su's formula has not been tested apart from the observation of the cited redshifts. The underlying periodicity was tested by others with larger samples (above) and not supported there; the association claim (ejection) has been contested [web: same search results; I did not read them]. The sample used by Karlsson, Arp and Burbidge was selected by proximity to bright galaxies, which was Hawkins et al.'s point of attack [memory, medium]. Hence for the periodicity: tested, with large modern surveys not supporting it. For Su's gravitational mechanism specifically: never tested, and rests on an assumed density. d. If true. Gravity-induced redshift would dominate for some quasars; "cosmological" redshift would not universally give distance; the Hubble relation would not hold for those objects; the existence of cloud-like massive objects with R much less than R_s that are stable would be required.

Verdict (split). - Periodicity of quasar redshifts (the premise): Tested, not supported (large 2dF survey analysis [web]); the discordant-association claims behind it remain debated and I do not settle them. - Su's mechanism (gravitational redshift from an assumed dense cloud, with halving series): Internally inconsistent in the strong-potential form (weak-field formula used at Phi_g/c^2 = 4.2; no stability argument), and Never properly tested otherwise. This is the weakest claim in the project.

Where Su is wrong or careless (same rigour applied to the inventor)

  • "About 100 times" is 64 (appears in EPJC and EPL).
  • "Agreement" with Hafele-Keating is shown with HK's own predicted numbers, and the measured -59 +/- 10 ns eastward is not quoted [read: deep f6 (b)].
  • Rotor and Ives-Stilwell comparisons assert agreement with cited papers without a single number from those papers; the Ives-Stilwell wavelengths may be representative rather than raw plate readings [memory, medium; deep f7 flag].
  • Matched data are chosen from the model's own frame: the ECI frame for earthbound and heliocentric for interplanetary, each justified by agreement with the formulas analysts happened to use.
  • The model is built so every modulation test passes (above); the remaining predictions are far below present sensitivity (1e-18) or have no data (clock synchronisation in the one-way test is unaddressed).
  • 1917 for light deflection (1919); a 19.5 microsecond "disagreement" with Kenyon that is probably a convention.
  • Quasar paper: the weak-field extrapolation and the unanswered free-fall time.
  • Reference irregularities in the quasar paper (duplicate Gutierrez and Lopez-Corredoira 2004 entry, a malformed arXiv number, a possible volume/page transposition) [read: deep f5c; memory, medium to low].

The strongest honest case for Su

  1. He states his model's departures from relativity as numbers (62 Hz, 0.14 degrees, 12 Hz, 11.5 Hz, 1 degree), names the unsolved problem (annual aberration) and the unfavourable data (34 versus 62 Hz; "not good quantitatively") himself. That is the behaviour of a falsifiability-minded author.
  2. The GPS/ECI argument is a real observation: the only frame in which the first-order Sagnac correction is applied in practice is the ECI frame, and the Earth's orbital 30 km/s is not corrected. That relativity gives the same observable does not by itself show that Su's reading is wrong; it shows that GPS cannot adjudicate. The decisive items are second-order and have not been measured by anyone for this purpose.
  3. A single published residual at twice the turntable rate exists in the best historical cavity experiment, was dismissed as spurious (by the experimenters, per Su; the original not read), and Su's model predicts a term of that kind. A dedicated north-south/east-west and re-orientation test, or a space-based interferometer, would settle it cheaply relative to its significance; no one appears to have done it.
  4. Independent authors in mainstream journals (Field 2019, Kipreos 2024) are examining whether differential simultaneity and the Sagnac frame question have experimental support; Gift (2021) reads GPS the same way.
  5. The model reproduces a great deal of standard physics with a classical, local mechanism and no curvature, and the algebra is correct throughout the deep reads.

What would settle it (for the Arbiter)

  • Reorienting rotating-cavity apparatus relative to geographic east and at two latitudes, looking for a term fixed to the Earth rather than the apparatus; or a cavity comparison in low Earth orbit.
  • Obtaining and reading Brillet and Hall (1979) for the phase and size of the 2-omega term, Kennard (1917) for the outside-capacitor result, Staudenmann et al. (1980) for the zero of the crystal scale and the wavelength, Kundig (1963) for the rotor, and the Krisher papers.
  • A worked comparison of F1 (electrostatic force factor 1 + v^2/2c^2) with Bertozzi (1964) and accelerator energy gain.
  • The cited-by lists of EPJC and EPL; the full 2008 book; the NTHU alumni e-news of 9 May 2008; NTHU records (needs Su's Chinese name).

Source log

[read] deep/*.md (all 13), facts.json, claims.md, sites.md, research/ notes, research/files/wayback-qem/f1b.pdf pp.1-3, research/files/wayback-qem/f2d.pdf (introduction, Kennard passage, conclusion). [web] Brillet-Hall record: https://jila.colorado.edu/bibcite/reference/10790 ; Herrmann et al. 2009: https://arxiv.org/pdf/1002.1284 ; Hawkins, Maddox, Merrifield: https://arxiv.org/pdf/astro-ph/0208117 ; Bertozzi 1964 (via https://en.wikipedia.org/wiki/Tests_of_relativistic_energy_and_momentum) ; Bertotti et al. Cassini 2003: https://physicsworld.com/a/general-relativity-passes-cassini-test/ . [memory] flagged in place; not relied on for any verdict except where stated as a question for the Physicist.