Su's local-ether model: a primer
Project folder: Researchers/Su-CC. It holds 14 distinct papers and two web pages by Ching-Chuan Su (C.-C. Su), professor of electrical engineering at National Tsing Hua University (NTHU), Hsinchu, Taiwan, written 2000-2006, plus an outline and a web summary of his book Quantum Electromagnetics: A Local-Ether Wave Equation (NTHU Press, 2008; ISBN registered 2005) [read: facts.json; web: isbn.tw and NTHU alumni e-news, per sites.md]. Su had a conventional career in microwave and antenna engineering before 2000 (IEEE Transactions papers in 1988, 1989 and 1993) [web: OpenAlex, per research/2026-10-06-openalex.md]. His local-ether papers appeared in refereed journals: Eur. Phys. J. C 21, 701 (2001); Europhys. Lett. 56, 170 (2001) and 60, 1 (2002); Eur. Phys. J. B 24, 231 (2001); J. Electromagn. Waves Appl. 15-19 (2001-2005); Found. Phys. Lett. 19 (2006); others are arXiv preprints [read: facts.json, deep reads]. Su is the originator of the model and so an interested party: what he says about relativity, GPS practice and other people's experiments is his characterisation.
This primer condenses the review files (facts.json, claims.md, the deep reads, the Investigator's sites.md and research notes, physicist.md, mathematician.md, contrarian.md and arbiter.md). It adds no verdicts; the verdicts are the arbiter's, including the arbiter's "QE revision" section (items Q1-Q8), which followed the quality check in qe.md and is folded in below. The worked example (section 5) was re-derived for this primer in example.py.
Source note (repeated from claims.md and arbiter.md). All papers in the folder are readable; four are scans whose equations the deep readers read from page images. Not in the folder: the full book (outline and web summary only), Su's papers on radar echo time (JEWA 14, 1525), on Sagnac (JEWA 15, 945), the railgun paper, and every experimental paper Su cites as support (Brillet and Hall 1979, Saburi et al., Krisher et al., Staudenmann et al., Mandelberg and Witten, Champeney et al., Kennard, Barnett). Where Su is compared with those experiments the comparison rests on Su's own citation, on a web check, or on [memory], and is tagged. Two verdicts rest partly on [memory]: that no dedicated satellite Michelson-Morley test has flown, and, for the Compton verdict, the Compton-Simon coincidence experiment (Bothe-Geiger and Kimble-Dagenais-Mandel were web-checked in the QE revision). Two of Su's own papers downloaded by the Investigator were read only at abstract level: physics/0208083 (atom-radiation interaction, the likely source of one listed prediction) and the Barnett preprint (f2d), so those predictions have no size in this review.
Source tags: [read] = from Su's documents (directly or via the deep reads that read the page images); [web] = Internet source with URL; [memory] = recalled from training, with confidence; [computed] = calculated in example.py for this primer; [review: file] = a calculation done by a named reviewer and not repeated here.
Abbreviations used throughout: SR = special relativity; GR = general relativity; GPS = Global Positioning System; ECI = Earth-centred inertial frame (origin at Earth's centre, axes pointing at fixed stars, not turning with the Earth); ECEF = Earth-centred Earth-fixed frame (turns with the Earth); MM = the Michelson-Morley experiment; LEO = low Earth orbit; ppm = parts per million.
1. The idea in one paragraph
Light needs nothing to travel through, according to modern physics, and its speed comes out the same for every observer moving steadily. Su, an antenna and microwave engineer, noticed something practical: every system that times radio signals around the Earth (GPS, satellite clock comparisons, giant ring interferometers) computes the light-travel time as if light moved at a fixed speed relative to a frame that is centred on the Earth and does not spin with it. The Earth's spin shows up in those calculations; the Earth's much faster motion around the Sun does not. Planetary radar, on the other hand, is computed in a frame centred on the Sun. Su took this literally. He proposed that each heavenly body carries its own "local ether", a medium for light at rest with that body's gravity: an Earth ether near the Earth, a Sun ether between the planets. He then rebuilt a large part of physics on that idea: gravity as a slowing of light (a refractive index) instead of curved spacetime, matter as waves obeying the same ether-bound wave equation (so that mass becomes a frequency), magnetism rewritten so that only relative velocities matter, and even a new explanation for quasar redshifts. Most of the numbers he obtains are exactly the numbers relativity gives, because his formulas turn out to be relativity's formulas with a different story attached. Where his model truly differs, it predicts tiny effects of the Earth's spin that relativity says are zero. The cleanest of these, a small direction dependence of light speed in a laboratory, was later measured in Berlin and found to be at least ten thousand times smaller than Su predicted.
2. What you need to know first
Reference frames and Galilean addition of velocities. A reference frame is a set of axes and clocks used to give positions and times. In Newtonian mechanics, if a train moves at velocity u relative to the ground and a ball moves at w relative to the train, the ball moves at u + w relative to the ground (the Galilean transformation). Light breaks this rule: experiment says light speed is the same in every inertial (non-accelerating) frame. Read: Taylor and Wheeler, Spacetime Physics, chapters 1-3; or Griffiths, Introduction to Electrodynamics, chapter 12 (Electrodynamics and Relativity) [memory, high for the Griffiths chapter].
Special relativity in three formulas. With gamma = 1/sqrt(1 - v^2/c^2) (v the speed of a moving object, c = 299,792,458 m/s the speed of light): (1) moving clocks tick slowly, by a factor 1/gamma; (2) moving objects are shortened along their motion by 1/gamma (Lorentz-FitzGerald contraction); (3) clocks that are synchronised in one frame are not synchronised in another ("relativity of simultaneity"). Together these make every measurement of the two-way (round-trip) speed of light give c in every direction. Read: Taylor and Wheeler; Griffiths chapter 12.
The ether, and the Michelson-Morley experiment (1887). Nineteenth-century physics held that light was a wave in a medium, the luminiferous ether. If the Earth moved through it at v, light sent along and back an arm of length L parallel to the motion should take (2L/c)/(1 - v^2/c^2), and across the motion (2L/c)/sqrt(1 - v^2/c^2). The difference is about (L/c)(v^2/c^2). Michelson and Morley turned their interferometer and saw almost no shift, far less than the Earth's 30 km/s orbital speed should give. Three responses were tried: the ether is dragged along with the Earth (Stokes); moving objects shrink by exactly the factor needed (Lorentz and FitzGerald); there is no ether and no preferred frame at all (Einstein, 1905). Section 3 explains why the first failed and why the second and third make the same predictions. Read: any modern physics text, the chapter on special relativity (e.g. Serway, Moses and Moyer, Modern Physics, chapter 1 [memory, medium for the chapter number]).
The Sagnac effect. Send light both ways round a closed loop that rotates. The beam going with the rotation has farther to go before it catches the moving starting point; the other has less. The time difference is Delta-t = 4 Omega . S / c^2, where Omega is the rotation vector (rad/s) and S the loop's area vector (m^2, normal to the loop). It is first order in velocity (it is proportional to the rim speed over c, not its square), which is why it is easy to see. It is what ring-laser and fibre-optic gyroscopes measure. Read: Post, "Sagnac effect", Rev. Mod. Phys. 39, 475 (1967) [memory, high].
Stellar aberration. A telescope moving across the line of sight to a star must be tilted forward by an angle v_perp/c, as you tilt an umbrella when you walk in vertical rain. The Earth's orbital motion gives the annual aberration, a yearly circle of radius 30 km/s / c = 1.0e-4 rad = 20.5 arcseconds, discovered by Bradley in 1725-1728. It is what killed the dragged-ether idea: if the ether near the Earth moved with the Earth, light would arrive with no sideways velocity relative to the telescope and there would be no annual aberration. Read: Born and Wolf would be too advanced; any astronomy text (e.g. Karttunen et al., Fundamental Astronomy, the chapter on spherical astronomy) [memory, medium].
Refractive index and travel time. In a medium of refractive index n, light moves at c/n. A medium whose index changes with position bends light toward higher n, and a path through it takes the integral of n ds / c (ds an element of path length). That is all the optics needed to understand "gravity as an index". Read: Hecht, Optics, the chapters on propagation of light and geometrical optics.
Doppler shift. A source approaching at speed u has its frequency raised by about u/c (first order). Bouncing a signal off a moving target doubles the effect. In SR the exact round-trip ratio for a target receding at radial speed u is (1 - u/c)/(1 + u/c). There is also a second-order "transverse" shift from time dilation. Read: Griffiths chapter 12 (problems on Doppler shift).
Waves, phase velocity and group velocity. A wave exp(i(k . r - omega t)) has wave vector k (magnitude 2 pi/wavelength) and angular frequency omega. Its crests move at omega/k (phase velocity); a packet of such waves moves at d omega/dk (group velocity). De Broglie's relations say a particle of momentum p and energy E is a wave with p = hbar k and E = hbar omega, where hbar = h/2 pi = 1.0546e-34 J s. Read: Griffiths, Introduction to Quantum Mechanics, chapter 2 (the free particle and wave packets).
The Klein-Gordon equation (introduced from scratch). Start from the relativistic energy-momentum relation E^2 = p^2 c^2 + m^2 c^4. Replace E by i hbar d/dt and p by -i hbar grad (the same substitutions that turn E = p^2/2m into the Schrödinger equation). You get (nabla^2 - (1/c^2) d^2/dt^2) psi = (m c / hbar)^2 psi. This is the Klein-Gordon equation, the simplest relativistic wave equation (it describes particles without spin). A plane wave solves it if omega^2 = omega0^2 + c^2 k^2, with omega0 = m c^2/hbar (the "Compton angular frequency"). "Lorentz invariant" means the equation has the same form in every inertial frame when space and time are transformed by the Lorentz transformation; it does not single out any frame. Read: Griffiths, Introduction to Elementary Particles, the chapter on quantum electrodynamics opens with it [memory, medium]; any relativistic quantum mechanics text.
Gravitational potential. Newton's potential of a mass M at distance r is -GM/r (G = 6.674e-11 N m^2/kg^2). In this primer, as in Su's papers, Phi_g = GM/r is used as a positive number, so Phi_g/c^2 is a small dimensionless number measuring how strong gravity is: 7e-10 at the Earth's surface, 2e-6 at the Sun's surface. Read: any mechanics text (Kleppner and Kolenkow, chapter on gravitation).
3. The textbook view
3.1 Sagnac, GPS and the Earth-centred frame
Symbols: Delta-t, time difference; Omega, rotation vector; S, directed area; omega_E = 7.292e-5 rad/s, the Earth's rotation rate (sidereal); r, position vector from the Earth's centre; R_t, vector from transmitter to receiver.
The Sagnac formula Delta-t = 4 Omega . S/c^2 is derived in SR by working in a non-rotating (inertial) frame where light speed is c and the loop simply moves. Michelson and Gale (1925) measured the Earth's own rotation this way with a 0.2 km^2 loop: a shift of about 0.23 fringe, close to the 0.236 expected [memory, high, per deep/A-local-ether]. GPS uses the same physics. A receiver on the ground moves eastward at up to 465 m/s while a signal is in flight for about 0.07 s, so the computed range must be corrected by R_t . (omega_E x r_receiver)/c = 2 omega_E . S / c, where S is the area of the triangle satellite-receiver-Earth centre. The correction is up to about 30-40 m [read: EPJC p.3-4; review: deep/A-local-ether]. The calculation is done in the ECI frame. The standard reference is Ashby, "Relativity in the Global Positioning System", Living Rev. Relativity 6, 1 (2003) [memory, high].
Why is the Earth's orbital speed (30 km/s, 64 times the equatorial rotation speed) not in the correction? In GR the Earth and everything near it are falling freely around the Sun. A freely falling, non-rotating frame is "locally inertial": in it, physics looks the same as in empty space, except for tiny tidal terms. The ECI frame is such a frame. So relativity predicts that the orbit does not show up in earthbound light timing, and that the spin does. This is the equivalence principle at work, and it is the central point of the whole project: the observation Su starts from is real, and relativity already explains it [review: physicist.md section 0].
3.2 Ether theories, contraction, and modern cavity tests
Symbols: L, arm length (mirror to mirror); theta, angle between the arm and the velocity v of the apparatus through the supposed ether; beta = v/c.
In a classical ether (light at c in the ether frame, rigid arms, clocks unaffected by motion), the round-trip time on an arm at angle theta is exactly
t(theta) = (2L/c) sqrt(1 - beta^2 sin^2 theta) / (1 - beta^2) ≈ (2L/c) [1 + (beta^2/2)(1 + cos^2 theta)].
An arm along v takes longer than one across v by the fraction beta^2/2 [computed: example.py Part A]. If the arm is a Fabry-Perot cavity (two mirrors facing each other, which resonate when a whole number of half-wavelengths fits between them), the resonance frequency is inversely proportional to the round-trip time and so changes by the same fraction as the cavity is turned.
Lorentz's ether theory (1904) keeps the ether but adds two physical effects: arms shorten along v by sqrt(1 - beta^2), and clocks moving through the ether run slow by the same factor. With both, the round-trip time read on the lab clock is exactly 2L/c for every orientation and every speed [computed: example.py Part B]. Lorentz's theory makes the same predictions as SR for every experiment; the difference is only whether you call one frame "really at rest". So "ether" by itself is not the issue. The issue is whether the ether comes with contraction and time dilation. If it does, nothing can detect it. If it does not, cavities show a direction dependence of order beta^2.
Physicists test this with test theories (Robertson 1949; Mansouri and Sexl 1977) that add adjustable parameters for "how much contraction" and "how much time dilation" and let experiment fix them [memory, high]. Results in the review files: - Brillet and Hall (JILA, Boulder, 1979): a 3.39 micrometre helium-neon laser locked to a cavity on a rotating table; anisotropy of light speed tied to the sky, Delta-c/c = (1.5 +/- 2.5) x 10^-15 [web: https://vivo.colorado.edu/display/pubid_108412]. - Herrmann et al. (Berlin, 2009): two crossed cavities on an air-bearing turntable rotating every 45 s, data May 2007 to June 2008; sky-tied anisotropy about 1e-17, and the constant, lab-fixed amplitudes at twice the table rate fitted separately (year means about 1e-17, see section 5) [web: https://arxiv.org/pdf/1002.1284, read by the arbiter].
One point is important for fairness. These experiments look mainly for signals that change with the sidereal day and the year, because they were testing for a preferred frame moving at hundreds of km/s relative to the Earth (for example, the frame of the cosmic microwave background). A preferred frame that moves with the Earth's centre, as Su's does, gives no sidereal signal. It gives a signal fixed to the laboratory, at twice the table rotation rate, tied to geographic east. That is why Su's model was not ruled out by the famous null results, and why the lab-fixed term is the place to look [review: physicist.md section 0 item 2].
A related family of tests looks at matter instead of light. In the Hughes-Drever experiments (1960-1961) the energy levels of nuclei in a magnetic field were watched as the Earth's rotation turned the apparatus relative to the stars; any direction dependence of the nuclear levels would have shown up as a daily change. None was seen [memory, high].
3.3 Aberration and the "dragged ether" problem
Symbol: alpha, aberration angle; v_perp, telescope velocity across the line of sight.
alpha = arctan(v_perp/c) with v_perp measured relative to the frame in which the starlight travels in straight lines at c. In SR one uses the Sun-centred (more exactly, the solar-system barycentric) frame, so v_perp includes the Earth's orbit (20.5 arcsec, annual) and the Earth's spin (0.3 arcsec at the equator, daily) [memory, high]. In any theory where the light near the Earth travels at c relative to an Earth-carried medium, the annual part should vanish unless something special happens at the boundary between the Sun's medium and the Earth's. That was the downfall of Stokes's ether drag in the 19th century [memory, high].
3.4 Gravity and light as a refractive index
Symbols: n, effective refractive index; gamma_PPN, the "parametrised post-Newtonian" parameter that measures how much space is curved per unit mass (gamma_PPN = 1 in GR); b, impact parameter (closest approach of a light ray to the Sun); M_Sun = 1.989e30 kg.
You do not need GR's mathematics for this part. For weak gravity (Phi_g/c^2 << 1), GR predicts that light travels as if the vacuum had index
n = 1 + (1 + gamma_PPN) GM/(c^2 r) = 1 + 2GM/(c^2 r) for GR.
From this follow the bending of starlight grazing the Sun, 4GM/(c^2 b) = 1.75 arcsec, and the Shapiro delay of radar echoes passing near the Sun (about 200 microseconds for Earth-Venus near superior conjunction) [memory, high; Su's paper gives 232.5 microseconds on recomputation, review: physicist.md]. The two numbers do not disagree; they assume different geometry. The round-trip delay is (4GM/c^3) ln(4 r_E r_V/b^2), with r_E and r_V the distances of Earth and Venus from the Sun and b the ray's closest approach to the Sun. It grows only logarithmically as b shrinks: 233 microseconds for a ray grazing the solar surface (b = 1 solar radius, Su's case), 205 at 2 solar radii, 189 at 3 [computed: example.py Part F]. Real radar rays cannot pass right at the limb (the solar corona disturbs them), so the quoted "about 200" corresponds to a ray a little way out. The Cassini spacecraft measured gamma_PPN = 1 + (2.1 +/- 2.3) x 10^-5 [web: https://ideas.repec.org/a/nat/nature/v425y2003i6956d10.1038_nature01997.html]. Separately, clocks deeper in a gravitational potential run slow by about Delta-Phi_g/c^2: Pound and Rebka (1960) measured this over a 22.5 m tower, Gravity Probe A (1976) with a rocket [memory, high]. In GR this shift is universal: every clock, whatever its mechanism (a hyperfine transition as in caesium and hydrogen-maser clocks, an electronic transition as in optical clocks, a nuclear one), slows by the same Delta-Phi_g/c^2. This is called local position invariance, part of the equivalence principle. It is tested by comparing two different kinds of clock while the Sun's potential at the Earth changes over the year (the orbit is slightly elliptical). One such test, hydrogen masers against caesium fountains, found the two shift alike to within 2.1e-5 of the effect [web: https://www.ncbi.nlm.nih.gov/pubmed/17359010, Ashby et al., Phys. Rev. Lett. 98, 070802 (2007), as summarised by the arbiter]. Both are hyperfine clocks, so that test does not compare electronic with hyperfine transitions; optical-clock comparisons would, and their limits were not checked in this review. In GR the exact redshift from the surface of a non-rotating mass is 1 + z = 1/sqrt(1 - 2GM/(R c^2)), which becomes infinite when R shrinks to the Schwarzschild radius 2GM/c^2 [memory, high].
What the index picture does not give: the advance of Mercury's perihelion, frame dragging, gravitational waves, or what happens in strong gravity. Those need GR's field equations [memory, high].
3.5 Matter waves and clocks
In SR a particle of rest mass m is described by the Klein-Gordon (or, with spin, the Dirac) equation. From omega^2 = omega0^2 + c^2 k^2 one gets group velocity v = c^2 k/omega, hence omega = omega0 gamma, momentum p = hbar k = gamma m v and energy E = gamma m c^2: the relativistic mass and de Broglie relations. Atomic transition frequencies of a moving atom, seen from a frame in which the atom moves at v, are lowered by 1/gamma (time dilation). For clocks near the Earth this is computed in the ECI frame, which is how the Hafele-Keating flights (1971) and GPS satellite clock rates (satellite clocks gain about 38 microseconds per day net) are predicted [memory, high; review: deep/f6].
3.6 Magnetism and frames
Symbols: q, charge; E, electric field; B, magnetic field; F = q(E + v x B), the Lorentz force; u, velocity of a magnet.
The Lorentz force uses the charge's velocity v in the frame where E and B are given. A magnet moving at u produces, in the lab, an electric field E = -u x B (to first order), so a charge at rest feels qE = q(0 - u) x B: the force depends only on the charge's velocity relative to the magnet. This "magnetic limit" of Maxwell's theory is exactly Galilean covariant at low speeds (Le Bellac and Lévy-Leblond, Il Nuovo Cimento B 14, 217, 1973) [memory, medium]. Before Maxwell, Weber (1846) and Riemann proposed forces between charges that depend directly on their relative velocities; these were set aside after Maxwell and Lorentz, though they reproduce many low-speed results [memory, high]. In the textbook theory a steady current in a closed, neutral circuit produces no electric field outside the wire. Weber-type laws predict a tiny one, proportional to the square of the electrons' drift speed over c^2. One group looked for such a field: Edwards, Kenyon and Lemon (Utah State University), Phys. Rev. D 14, 922 (1976), reported a potential between a superconducting niobium-titanium coil and its shield that grew as the square of the current (fitted exponent 2.02 +/- 0.05), of the sign an electron charge would give, surviving a bifilar (non-inductive) winding and a Faraday cage, of 3 to 13 millivolts in earlier runs and tens of millivolts later. They excluded the explanations they considered and wrote: "The signal appears to be a real field effect and is as yet unexplained" [read: deep/f2a section 13, from the scanned paper; web: https://isidore.co/misc/Physics%20papers%20and%20books/Zotero/storage/C6B7H3S5/PhysRevD.14.html]. The same authors reported an observed potential difference again in Phys. Lett. A 162, 105 (1992) [web: abstract of arXiv:1009.0474, per arbiter.md Q2]. A null test by Bartlett and Maglic (1990) is recalled by the reviewers [memory, low-medium] but was not found, so this review cannot say whether a null replication exists. The result is documented by one group; no independent second measurement was found in this review.
3.7 Compton scattering
X-rays scattered by electrons come out with a longer wavelength, Delta-lambda = lambda_C (1 - cos phi), where phi is the scattering angle and lambda_C = h/(m_e c) = 2.43 pm (the electron's Compton wavelength). Compton (1923) derived it by treating the X-ray as a particle (photon) of energy hf and momentum h/lambda and applying conservation of energy and momentum [memory, high]. A wave-only description can reproduce the shift (Schrödinger, 1927) [memory, medium]. What favours photons are coincidence experiments in which each scattered X-ray is matched with one recoil electron (Bothe and Geiger 1925; Compton and Simon 1925) and, much later, photon antibunching (Kimble, Dagenais and Mandel, Phys. Rev. Lett. 39, 691, 1977) [web: https://en.wikipedia.org/wiki/Bothe%E2%80%93Geiger_coincidence_experiment; web: https://resolver.caltech.edu/CaltechAUTHORS:KIMprl77; Compton-Simon memory, medium, not web-checked; per arbiter.md claim 3]. Antibunching tests the light from atoms, not Compton scattering by free electrons, so the Compton-specific evidence is the two 1925 coincidence experiments.
3.8 Quasar redshifts
A quasar's redshift z (fractional increase of wavelength) is, in the standard view, cosmological: it measures how much the universe expanded while the light travelled. Lensing of quasars by nearer galaxies and detection of host galaxies around quasars support this [memory, high]. A minority of astronomers with mainstream careers, notably Halton Arp and Geoffrey Burbidge, reported quasars close on the sky to much nearer galaxies, sometimes apparently linked by filaments, and argued that some quasar redshift is "intrinsic". Su cites ten such cases from these authors, among them [read: deep/f5c, Su's citations, reference details not web-verified; arbiter.md Q8]: an X-ray quasar 8 arcsec from the nucleus of the galaxy NGC 7319 with a V-shaped filament (Galianni et al. 2005); a filament from the galaxy NGC 3628 (z = 0.0028) to quasars at z = 0.995 and 2.15 (Arp et al. 2002); the NEQ3 filament with three compact objects at z = 0.1935, 0.1939 and 0.2229 (Gutierrez and Lopez-Corredoira 2004); and two "abrupt boundaries", MCG 7-25-46 (z = 0.003 against 0.098, the H-alpha emission of one ending exactly where the other's begins; Lopez-Corredoira and Gutierrez 2005) and NGC 7320 (Gutierrez et al. 2002), where Su records that the original authors read the sharp edge "instead as evidence against the connection ... and for the projection effect". Arp, Burbidge, Lopez-Corredoira and Gutierrez are proponents of physical association; these are their reports, which Su adopts. Su did no data analysis of his own on these objects. Karlsson (1971, 1977) reported that quasar redshifts cluster at preferred values (0.06, 0.30, 0.60, 0.96, 1.41, 1.96), equally spaced in log(1 + z) [read: deep/f5c, from Su's citations]. Large later surveys did not find the periodicity: Hawkins, Maddox and Merrifield (2002) on 1647 quasar-galaxy pairs [web: https://arxiv.org/pdf/astro-ph/0208117]; Tang and Zhang (2005) on 2dF and SDSS samples of tens of thousands of quasars [web: https://en.wikipedia.org/wiki/Redshift_quantization]. Burbidge and Napier (2001) reported the pattern in other samples, and proponents dispute how the large samples were selected [web: same page]. The question is contested.
4. Where Su departs
The table puts Su's postulate or result next to the textbook statement. The right-hand column says what kind of difference it is: a different interpretation of the same numbers, or a real difference in a prediction.
| Topic | Textbook | Su | Kind of difference |
|---|---|---|---|
| What light moves relative to | Nothing; c in every inertial frame | c relative to a "local ether" at rest in the ECI frame near Earth, in the Sun-centred frame between planets [read: EPJC p.1] | Interpretation at first order |
| GPS and Sagnac | Computed in the ECI frame because it is locally inertial (GR) | Computed in the ECI frame because the ether is there [read: EPJC p.3-4] | Same formula, same number |
| Michelson-Morley null | SR: exactly zero at every order | Earth's orbit absent; Earth's spin gives v^2/c^2 of about 2.4e-12, too small for the 1887 apparatus [read: EPJC p.8] | Real difference at second order |
| Rods and clocks | Contract and slow (Lorentz) | Atoms do not contract: in Su's matter-wave equation the size of an atom does not change with speed; clocks do slow by 1/gamma with v relative to the ECI frame [review: physicist.md claim 3; read: f6 eq. 23] | Real difference: this is what makes a cavity anisotropy |
| Laboratory cavity | No direction dependence | Resonance lowest when the axis points east-west, by v^2/(2c^2) = 1.2e-12 cos^2(latitude) peak to peak [read: EPJC eq. 14] | The cleanest testable difference (section 5) |
| Annual aberration | Explained (Sun-centred frame) | "Not yet solved": needs a rule for the interface between two ethers [read: EPJC p.12, p.15] | A gap Su names himself |
| Gravity | Curved spacetime; weak field acts like n = 1 + 2GM/(c^2 r) | n_g = 1 + 2GM/(c^2 r) postulated, no curvature [read: f5b p.3] | Same number, by choice of the factor 2 |
| Gravitational redshift | Clocks run slow in a potential | Atomic frequencies lowered by f0/sqrt(1 + 2 Phi_g/c^2) [read: f5a; review: mathematician.md] | Same at first order; differs at second |
| Redshift of different kinds of clock | The same Delta-Phi_g/c^2 for every transition (local position invariance) | One formula, f0 (1 - Phi_g/c^2), derived only for hyperfine (spin-flip) transitions; the redshift "can be different in some other interaction" and its "transition-dependence" is listed as a prediction, with no formula or number [read: f6 eq. 25; outline p.5, p.8; arbiter.md Q1] | Possibly real, but unsized; Su's own wave equation gives a different factor for electronic transitions (below) |
| Nuclear-level isotropy (Hughes-Drever) | Null, by Lorentz invariance | Null, because the lab's speed in the ECI frame does not change as the Earth turns, "whatever the dependence of nuclear magnetic interaction on mass and speed" [read: deep/f6 p.8] | Same null; from symmetry, not a calculation |
| Matter | Klein-Gordon/Dirac, Lorentz invariant | Klein-Gordon form, but derivatives "referred to the local-ether frame"; m0 = hbar omega0/c^2 called the origin of mass [read: EPJC eq. 36; f4a] | Same algebra; the frame is an interpretation; m0 = hbar omega0/c^2 is a calibration |
| Magnetism | Lorentz force; magnetic limit is Galilean | Force from "augmented" potentials depending on velocities relative to the ether, the conductor's ions (the "matrix") and the charge; reduces to Lorentz at low speed; extra terms at higher speed, including F2, a tiny electric force on a charge at rest beside a neutral loop carrying steady current [read: f2a eq. 28, f4a] | Same at low speed; extra terms untested; F2 has the form, not the size, of one reported anomaly (Edwards et al. 1976, section 3.6) |
| Sagnac in a loop of glass or fibre | Coefficient depends on whether the loop or the Earth turns; first-order SR gives both (Post 1967) | "Various index-dependences" (n^2 for the loop's rotation, none for the Earth's) offered as a test [read: deep/f3a p.10; outline p.8] | Same coefficients as SR: does not discriminate |
| Compton | Photon + conservation laws | Wave scattering between matter-wave states, with the same conservation relations built in; predicts a weak extra line [read: f8c] | Same shift; one new, unsized line |
| Quasars | Cosmological redshift | Gravitational redshift of a gas cloud, z = sqrt(1 + 2 Phi_g/c^2) - 1, used at Phi_g/c^2 = 4.1 [read: f5c] | New mechanism, outside its own weak-field range |
The redshift of electronic transitions, from Su's own equations. This is the one new point the QE revision found, and it is worth following because it shows how a model can disagree with itself. In Su's f5a, gravity enters the matter-wave equation through the index n = 1 + 2 Phi_g/c^2. Expanding his equation (arbiter.md Q1, agreeing with a SymPy check by the f6 deep reader), the part of an atom's energy that is potential energy is redshifted by the factor (1 - g) and the part that is kinetic energy by (1 - 3g), where g = Phi_g/c^2. For an electron bound by the Coulomb force, the virial theorem says that in any change of level the potential energy changes by twice the total and the kinetic energy by minus the total [memory, high]. So an electronic transition shifts by 2(1 - g) - (1 - 3g) = 1 + g: a blueshift, against the (1 - g) redshift of a hyperfine clock. At the Earth's surface (g = 6.96e-10) the two kinds of clock would differ by about 2g = 1.4e-9 [review: arbiter.md Q1]. Su's quasar paper (f5c eq. 6) instead puts one common factor on all the terms, which would make every clock shift alike, and Su does not reconcile the two papers. Hence the arbiter's verdict: Su's statement that every transition shifts by (1 - Phi_g/c^2) is Internally inconsistent for electronic transitions; for hyperfine clocks (caesium, hydrogen maser, as used in GPS and Hafele-Keating) it is the GR value and Established. Su himself committed to no species dependence and no number, so the arbiter does not call it "Tested, not supported"; as Su stated it, the transition-dependence is Not testable as stated. The deciding step is a calculation: redo the electronic-transition redshift from f5a eq. 3 without dropping any factor of n, then compare with an optical-clock test.
Other predictions listed in Su's book outline (p.8). The outline lists 11 predictions; four were not reviewed before the QE revision [read: deep/outline; arbiter.md Q3-Q7]: - a downshift of starlight that depends on the star's mass: one clause, no formula; in weak gravity Su's only formula gives GR's surface redshift (for example of white dwarfs), so Not testable as stated as a difference; - in magnetic deflection (for example a cathode-ray tube), an effect of the accelerating electrodes moving relative to the deflection coils (the "yoke"): at first order SR gives the same change, and the two differ only by about u w/c^2 (u the electron speed, w the electrode speed), 3e-8 to 5e-7 for a 25 kV tube, far below what a spot position can resolve: Not testable as stated as a difference from SR; - a "stronger second-order effect" in the interaction of atoms with radiation: no equation, size or experiment given, and the likely source paper (physics/0208083) was not read: Not testable as stated, with a source gap; - the transition-dependent redshift, above.
Two further tests Su proposes elsewhere were also classed: the Sagnac index-dependences (same as SR, Established, not discriminating) and a second-order measurement of the cosmic microwave background (CMB) dipole. For the CMB, Su treats the distant sources of the radiation as one more enclosing local ether, concedes that the dipole (about 300 km/s, v/c about 1e-3) cannot tell his model from SR, and says "a crucial test" needs the second-order term, giving no amplitude or method [read: deep/EPJC pp.11-12]. The order of magnitude, (v/c)^2 x 2.73 K, is about 3 microkelvin, and SR also predicts a second-order term from the Sun's motion that the paper does not separate [review: arbiter.md Q7]. Not testable as stated.
The exact point of departure is the row "Rods and clocks". Su keeps the classical ether's light propagation, keeps time dilation (his eq. 23, f = f0 sqrt(1 - v^2/c^2), v relative to ECI), but does not have length contraction of matter. Everything that distinguishes his model from SR at second order flows from that combination plus the choice of the ECI frame. Everything else (first-order Sagnac, the GR index, Klein-Gordon algebra, the low-speed Lorentz force) reproduces standard physics, which the reviewers checked line by line and found correct as algebra [review: all three reviews; arbiter.md section 1].
Where the inventor's starting points are sound, said plainly: - The ECI frame is the frame GPS and two-way satellite time transfer actually use, and the Earth's orbital velocity is not corrected for. Su's engineering observation is correct. Its explanation in SR/GR is the equivalence principle; Su's is an ether. GPS cannot tell them apart [arbiter.md claim 1]. - A preferred frame is not by itself ruled out. Lorentz's ether theory is consistent with every experiment. Recent papers in a mainstream journal (Field 2019; Kipreos 2024, both in Int. J. Mod. Phys. A) discuss preferred frames and absolute simultaneity and cite Su [web: OpenAlex abstracts, per sites.md]. - The refractive-index picture of weak gravity is a standard and useful way to calculate light bending and delays [memory, high]. - Su stated numerical predictions where his model differs, named the one observation he could not explain (annual aberration), and reported the one unfavourable comparison he knew of ("too early to make a decisive conclusion"; "the agreement is not good quantitatively") [read: EPJC p.9; f8a p.8]. The Contrarian rates this as the behaviour of a falsifiability-minded author [review: contrarian.md].
5. Worked example: the cavity "quadrupole", predicted and measured
Su's cleanest prediction is that a laser cavity on a turntable changes its resonance frequency as it turns, with a minimum when it points east-west. We derive it, show why SR predicts zero, put in numbers for the experiment Su used (Brillet and Hall, 1979) and the later one that tested it (Herrmann et al., 2009). Every step is reproduced in example.py; run it with
.venv/bin/python review/Su-CC/example.py
It ran cleanly and every check passes.
Step 1. The round-trip time on a moving arm (classical ether). Work in the ether frame. An arm of length L moves at velocity v; the arm makes angle theta with v. On the way out, light leaves the near mirror and must reach the far mirror, which moves v t1 during the flight time t1. The light covers the vector (L cos theta + v t1, L sin theta) at speed c:
(c t1)^2 = (L cos theta + v t1)^2 + (L sin theta)^2.
On the way back the near mirror comes to meet the light:
(c t2)^2 = (L cos theta - v t2)^2 + (L sin theta)^2.
Each is a quadratic in the time. Solving and adding the positive roots gives (SymPy does it in Part A)
t = t1 + t2 = (2L/c) sqrt(1 - beta^2 sin^2 theta)/(1 - beta^2), beta = v/c.
Expanding for small beta: t ≈ (2L/c)[1 + (beta^2/2)(1 + cos^2 theta)]. This is Su's eq. 13 [read: EPJC p.8] and the textbook 1887 result. Along v (theta = 0) the bracket is 1 + beta^2; across v (theta = 90 deg) it is 1 + beta^2/2. The fractional difference is beta^2/2.
Step 2. Why this means a frequency change. A cavity resonates when the round trip holds a whole number N of wavelengths, so f = N/t. A fractional change Delta-t/t gives Delta-f/f = -Delta-t/t. The cavity's frequency is lowest when it points along v and highest across it, with peak-to-peak fractional change beta^2/2. In Su's model v is the lab's velocity in the ECI frame, which is the Earth's rotation speed at the site, pointing east: v = omega_E R_E cos(latitude). That gives Su's eq. 14, Delta-f/f = 1.2e-12 cos^2(latitude) [read: EPJC p.8].
Step 3. The same calculation with Lorentz contraction. Now let the arm, of rest length L at angle theta0 in its own frame, be shortened along v by sqrt(1 - beta^2). Its ether-frame components are (L cos theta0 sqrt(1 - beta^2), L sin theta0). Putting this into Step 1's formula, the quantity under the square root becomes L^2 (1 - beta^2) whatever theta0 is, and the time is 2L/(c sqrt(1 - beta^2)). The lab clock runs slow by sqrt(1 - beta^2), so it reads exactly 2L/c (example.py Part B). No angle, no speed: SR (and Lorentz's ether theory) predict no effect at any order. So the whole prediction rests on matter not contracting, which is the case in Su's model [review: physicist.md claim 3].
Step 4. Numbers for Brillet and Hall (Boulder, about 40 deg N). With omega_E = 7.292e-5 rad/s and R_E = 6.378e6 m: - v at the equator = 465.1 m/s; at 40 deg, v = 465.1 x cos 40 deg = 356.3 m/s; beta = 1.19e-6. - beta^2/2 = 7.06e-13 [computed]. (Su: 1.2e-12 x cos^2 40 deg = 7.0e-13.) - The 3.39 micrometre laser frequency is c/lambda = 8.84e13 Hz. - Predicted peak-to-peak swing: 7.06e-13 x 8.84e13 Hz = 62.5 Hz [computed]. Su wrote 62 Hz [read: EPJC p.9]; the small difference is his rounder Earth radius and solar-day rotation rate. - Su reports that Brillet and Hall saw a term at the right rate, "merely about 17 x 2 Hz", which they "attributed to a persistent spurious signal" [read: EPJC p.9]. 34 Hz is 3.8e-13 fractional, 0.54 of the prediction [computed]. (Nobody in this review obtained the Brillet-Hall paper; the 34 Hz and its attribution are Su's account. A web search found only the abstract.) - Check that the series is good enough: at this beta the exact formula and beta^2/2 agree to seven digits (example.py Part C).
Step 5. The later test: Herrmann et al. 2009 (Berlin, 52.5 deg N). Two cavities at right angles are compared on one turntable; the measured quantity is the difference of their frequencies. With cavity 1 at angle a from east and cavity 2 at a + 90 deg, the difference of their shifts is -(beta^2/2)(cos^2 a - sin^2 a) = -(beta^2/2) cos 2a. That is a signal at twice the rotation rate, fixed in the laboratory, with amplitude beta^2/2 (Part D checks the identity). - v at Berlin = 465.1 x cos 52.5 deg = 283.1 m/s; beta^2/2 = 4.46e-13 [computed; the arbiter got 4.5e-13]. - Herrmann et al. fit the constant lab-fixed amplitudes at twice the table rate separately from the sky-tied ones. Their year means are C0 = -0.2 +/- 11.7 and S0 = -10.2 +/- 14.4, in units of 1e-18 [web: https://arxiv.org/pdf/1002.1284, Fig. 4 caption, as read by the arbiter]. The combined amplitude is sqrt(C0^2 + S0^2) = 1.0e-17, consistent with zero. - Prediction over measurement: 4.46e-13 / 1.0e-17 = 4.4e4 [computed]. Even if the amplitude conventions differ by the factor 2 that the review did not check, the gap is more than ten thousand. - The authors say individual lab-fixed values wander by several standard errors and "vary in magnitude and phase and thus average out" over the year [web: same, per arbiter]. Su's effect could not average out: it is tied to geographic east and is constant all year. That is what lets a year of data separate it from drifting systematics.
Step 6. Why the ether has to be local, and what a satellite would see (Part E). - If the ether were fixed to the Sun, v would be the 29.8 km/s orbital speed and beta^2/2 = 4.9e-9, about 7000 times the 40 deg value, and it would swing with the time of day. Brillet and Hall's 2.5e-15 limit on sky-tied signals rules that out by a factor of over a million. This is why Su needs the ether to move with the Earth's centre. - In the 1887 Michelson-Morley apparatus (11 m light path, 590 nm light), the orbital speed would give a phase swing of about 1.2 rad (Su quotes pi/3 for his rounder inputs); the Earth's spin alone gives about 3e-4 rad, invisible [computed; read: EPJC p.8]. - In low Earth orbit (7.6 km/s relative to the ECI frame) Su's model gives beta^2/2 = 3.2e-10; SR gives zero. This is the satellite Michelson-Morley experiment proposed by Su's advocate J. Marsen (viXra 1906.0336, 2019) [read: research/files/vixra]. It is the same physics as the ground cavity term, so with the ground term now bounded 4e4 times below Su's prediction it adds little for Su's model as written [arbiter.md claim 5].
What the example shows. (1) Su's prediction is honest arithmetic from his postulates: classical-ether propagation, no contraction, ECI frame. (2) The single 1979 comparison he had was a residual about half his prediction, which he called inconclusive. (3) A rotating-cavity experiment designed for another purpose measured the same kind of lab-fixed term over a year and found it at most about 1e-17, which is 4e4 times below the prediction. The arbiter's verdict on the lab-fixed cavity anisotropy is therefore Tested, not supported, with two caveats the arbiter keeps: the Berlin experiment was not designed to test Su (an incidental test), and the geometric equivalence of Su's single-cavity prediction and the two-cavity difference was argued, not derived from Su's equations (Step 5 does it with Su's eq. 13, which is the natural reading) [arbiter.md D1]. A side consequence: a lab-fixed term of 3.8e-13 in 1979 and under 1e-17 in 2008 means the 1979 residual was most probably a systematic, an inference the arbiter labels as such.
6. Evidence table
Verdicts are the arbiter's, on the scale defined in prompt.txt. "Established" for a part means the numbers match experiment and accepted theory, which for much of this project means they are SR's or GR's numbers.
| Claim | Tested? | Result | Replicated? | Verdict |
|---|---|---|---|---|
| 1a. ECI-frame Sagnac correction; no Earth-orbit term in earthbound timing | Yes: GPS, Michelson-Gale 1925, Japan-USA link (0.3 microsecond effect), gyros [read: EPJC; memory, high] | Agrees, as SR/GR in the ECI frame also does | Yes, daily | Established (not evidence for an ether) |
| 1b. The ether as the explanation at first order | Cannot be: identical numbers | n/a | n/a | Not testable as stated |
| 1c. Lab-fixed cavity anisotropy, 5e-13 to 7e-13 | Brillet-Hall 1979 (residual 3.8e-13, called spurious, per Su); Herrmann et al. 2009 (year mean about 1e-17) | Not seen at 4e4 times below prediction | The 2009 bound is one year-long experiment; incidental test | Tested, not supported |
| 1d. Annual aberration with an Earth-carried ether | Aberration observed since 1725 | Su concedes his model does not yet reproduce it | Yes | Tested, not supported |
| 2a. Light bending, Shapiro delay, GPS gravitational delay (42-62 ps) from n_g | Yes, as tests of GR (Cassini gamma_PPN = 1 to 2e-5) | Agree, by construction | Yes | Established |
| 2b. Gravity without curvature | No measurable difference stated; index covers light only | n/a | n/a | Not testable as stated; second-order redshift and frame effects Never properly tested |
| 3a. de Broglie and Lorentz-factor relations; Hafele-Keating, GPS clock rates, Ives-Stilwell, maser, rotor at second order; matter-wave Sagnac and gravity phases | Yes | Agree, identical to SR in the ECI frame | Yes | Established |
| 3b. Preferred-frame reading of those formulas | Identical numbers | n/a | n/a | Not testable as stated |
| 3c. m0 = hbar omega0/c^2 as origin of inertia; inertial = gravitational mass derived | A calibration of a coefficient; circular | n/a | n/a | Not testable as stated |
| 3d. f = f0 sqrt(1 - v^2/c^2) for hyperfine and nuclear clocks | Assumed by Su, not derived; SR gives the same | n/a | n/a | Plausible, untested (as a derivation) |
| 3e. Third-order rotor anisotropy (about 1e-18), v0^2/c^2 terms, neutron Bragg-angle asymmetry (about 1 deg) | Bragg: one comparison by Su with Staudenmann et al. 1980, "not good quantitatively"; sign depends on an unknown zero; others beyond reach | Ambiguous; leans against Su's size of effect | No | Never properly tested |
| 3f. Compton as wave scattering | Shift reproduced by construction; "direct radiation" line unsized | Coincidence (Bothe-Geiger 1925 [web]; Compton-Simon 1925 [memory, medium]) and antibunching (Kimble et al. 1977 [web]) favour photons | n/a | New line Never properly tested; wave-only replacement of photons Tested, not supported |
| 3g. Hughes-Drever nuclear-level isotropy | Yes (1960-1961) [memory, high] | Null, as Su's model also gives, by symmetry not calculation | Yes | Established, not discriminating (QE revision) |
| 3h. Gravitational redshift (1 - Phi_g/c^2) for every transition | Hyperfine clocks: GPS, Hafele-Keating; H maser vs Cs alike to 2.1e-5 (Ashby et al. 2007 [web]). Electronic vs hyperfine: optical-clock limits not checked | Su's own f5a equation gives (1 + g) for electronic transitions against (1 - g) for hyperfine, about 1.4e-9 apart at Earth's surface [arbiter calc, Q1]; Su commits to no species dependence | n/a | Hyperfine: Established. Electronic: Internally inconsistent. Transition-dependence as Su states it: Not testable as stated (QE revision Q1) |
| 4a. Galilean force law reduces to Lorentz at low speed; Fresnel drag at first order | Yes, as for the standard law | Agrees | Yes | Established |
| 4b. "Lorentz law holds only under hidden restrictions" | True of any approximation | n/a | n/a | Not testable as stated |
| 4c. Extra terms F1 (electrostatic force x (1 + v^2/2c^2), 4.6% at 25 kV), F3 (inertia shifted by potential, 20% for a 100 kV electron); Barnett outside or short solenoid; no drag in a moving magnetic medium | Not against Su's law. Bertozzi 1964 (electrons to 15 MeV agree with relativity) may be in tension with a literal high-speed F1; not worked through | Unknown | No | Never properly tested |
| 4d. F2: electric force near a neutral loop with steady current, quadratic in current | Not as a test of Su. Edwards, Kenyon and Lemon 1976 (superconducting coil) reported an unexplained potential quadratic in current, electron sign, mV to tens of mV; again in 1992 (same group) | Su's F2 for their coil: about 35 microvolts at 10 A, 89 at 16 A [arbiter calc, Q2]: same form and sign, 24 to 320 times smaller unless an unmeasured current-crowding factor is that large | One group; a 1990 null test is recalled but not found | Never properly tested (kept after weighing Edwards et al.; QE revision Q2) |
| 5a. One-way turntable link with atomic clocks | Not done | Su's own clock formula (eq. 23) cancels the effect his eq. 33 predicts [arbiter calc, D2] | n/a | Internally inconsistent; with non-atomic oscillators Never properly tested |
| 5b. Transverse radar Doppler -2v^2/c^2 (up to 12.5 Hz at 10 GHz, LEO) as a difference from SR | Standard Doppler tracking daily | Same as SR once both are evaluated at the same instant (agreement to 2e-19) [arbiter calc, D3] | n/a | Refuted by calculation (as a difference) |
| 5c. -11.5 Hz offset at 2.3 GHz in interplanetary links | Su says existing data cannot test it (fitted oscillator bias) | n/a | n/a | Never properly tested (ill defined without an interface rule) |
| 5d. No Sagnac term from Earth's orbit | Never seen | Also GR's prediction | Yes | Established, not discriminating |
| 5e. Sagnac index-dependences (n^2 for loop rotation, none for Earth rotation) | Harzer 1914, Dufour and Prunier 1942, as cited by Su via Post 1967; fibre gyroscopes | Same coefficients as first-order SR (Post 1967) | n/a | Established, not discriminating (QE revision Q6) |
| 5f. Star-mass-dependent stellar downshift; yoke/electrode effect in magnetic deflection; stronger second-order atom-radiation effect; second-order CMB dipole test | No formula or number (downshift, atom-radiation, CMB); yoke effect differs from SR only by about 1e-7 | n/a | n/a | Not testable as stated (QE revision Q3, Q4, Q5, Q7) |
| 6a. Weak-field gravitational redshift | Pound-Rebka, Gravity Probe A | Agrees | Yes | Established |
| 6b. Su's redshift formula at Phi_g/c^2 = 4.1 for the quasar near NGC 7319 (z = 2.07 vs observed 2.114) | Fit with a chosen density; formula derived for Phi_g/c^2 << 1 | Outside its own range; cloud of 1.3e16 solar masses, escape speed 2.9c, free-fall time 272 years, centre z = 2.69 vs surface 2.07 | n/a | Internally inconsistent |
| 6c. Karlsson periodicity (premise) | Large surveys (Hawkins et al. 2002; Tang and Zhang 2005) | Not found; Burbidge and Napier 2001 report it in other samples | Contested | Tested, not supported (contested) |
| 6d. Su's halving series predicting new peaks 0.22, 0.14, 0.09 | No data cited | n/a | No | Never properly tested |
| 6e. Quasar-galaxy cases (NGC 7319, NGC 3628, NEQ3, Arp 220, M82, NGC 383) and the abrupt-redshift-boundary argument (MCG 7-25-46, NGC 7320) | Su's cases are other authors' observations, quoted at second hand; no analysis by Su | A sharp boundary is what Su's single-potential rule needs, and also what unrelated galaxies at different distances would show; the original NGC 7320 authors favour projection | n/a | Abrupt-boundary argument Not testable as stated; physical association not assessed, left open (QE revision Q8) |
The record after 2006 (absence of evidence). Su's publications stop in 2006 [read: sites.md, from Crossref, arXiv and the Wayback Machine]. His advocate J. Marsen wrote in 2021 that Su "passed away in approximately 2007" and that his website was removed; the Natural Philosophy Wiki repeats this citing Marsen. This is documented in one published advocate source; no obituary or university notice has been found yet [read: research/files/vixra/2105.0110v1-Marsen.pdf, per sites.md]. The book appeared from NTHU Press in March 2008 (ISBN registered 2005), consistent with a posthumous edition if the date is right. Yet the Wayback Machine holds captures of Su's NTHU site as late as 2017 [read: research/2026-10-06-wayback-machine.md]. No published Comment or Reply to his main papers was found in one search (Google Scholar's cited-by lists were blocked). Competing explanations set out by the arbiter: death around 2007 (Marsen's account; consistent with the clean stop and the 2008 book); illness or retirement (no evidence either way); withdrawal after limited reception (no engagement found to answer); suppression (nothing in the record supports it, and the core papers appeared in refereed journals). The deciding records are an NTHU department or alumni notice from 2007-2008, newspaper obituaries, and the book's front matter; all need Su's Chinese name, which this review has not established [arbiter.md section 5]. The lack of rebuttals is consistent with referees finding nothing to rebut where the numbers equal relativity's, with the author's activity ending, or with low engagement; silence is not evidence of validity, and it is not evidence against.
Related and supporting work found by the Investigator (kept whatever the verdicts). Petr Beckmann proposed the same local-ether postulate in 1986-1987 (Einstein Plus Two; the journal Galilean Electrodynamics), described by Marsen and the Natural Philosophy Wiki as reached independently; Su's EPJC paper does not cite him [web; read: grep]. S. J. G. Gift (2021, Applied Physics Research 13(1)) argues from GPS timing for an Earth-bound ether and calls Su's nested ethers "a reasonable working hypothesis"; Gift is a proponent of preferred-frame readings [read: research/files]. Engineering papers on satellite time transfer, a Springer GNSS handbook chapter and radar synchronisation (2015-2025) cite Su's ECI-frame Sagnac formula; whether they engage with the ether reading is unknown (texts not read) [web: OpenAlex]. Su's own later papers add a Barnett-experiment prediction (arXiv physics/0512179) and a Riemann-type force law (Found. Phys. Lett. 2006) [read: research/files/wayback-qem]. Two other citing papers were found and not weighed by the reviewers, because neither bears on a verdict: J. Wu, H.-W. Tsao and Y.-R. Huang (National Taiwan University and the Chung-Shan Institute of Science and Technology), "Reviewing Michelson Interferometer Experiment and Measuring the Speed of Starlight", J. Mod. Phys. 10, 539 (2019), who cite Su's EPL paper as recent interest in Michelson-Morley, argue for a wave-ether model with zero drag, and report that starlight arrives at a different time from a local light source, which they read as starlight not travelling at c [read: research/files/others/Wu-Tsao-2019-JMP.pdf, abstract and p.1]; and J.-E. Persson, "The misunderstood Sagnac effect", Physics Essays 34 (2021), which cites Su (text not read) [web: OpenAlex, per sites.md]. Edwards, Kenyon and Lemon (1976, 1992), section 3.6, are related experimental work: they did not test or cite Su, but their reported anomaly has the form of his F2 term.
7. What a learner should take away
What is solid. - The first-order Sagnac effect and its use in GPS and satellite time transfer, computed in the ECI frame. Su's algebra for it is correct, and his engineering observation (the spin matters, the orbit does not) is true. - The weak-field index n = 1 + 2GM/(c^2 r) as a way to compute light bending and radar delay. Su's integrals are right; they give GR's numbers because the factor 2 is GR's. - The Klein-Gordon algebra from omega^2 = omega0^2 + c^2 k^2 to the relativistic mass and de Broglie relations. - The classical-ether round-trip formula (section 5, Step 1) is a good exercise in kinematics, and Su uses it correctly.
What is wrong, and why the mistakes are instructive. - Same numbers, different story, is not evidence. Su lists many experiments his model "accounts for" (GPS, Hafele-Keating, Ives-Stilwell, Mossbauer rotors, neutron interferometry). In every case his formula is relativity's formula evaluated in the ECI frame. Agreement with data that two theories predict equally cannot choose between them. The lesson is to ask, for every "confirmation", whether the rival theory predicts the same. - The model's distinctive content is one missing ingredient. Su keeps time dilation but not length contraction. Section 5 shows that this single choice creates the cavity anisotropy, and that adding contraction removes it exactly. The Berlin data say nature does contract, to about one part in 10^4 of the effect Su predicted. - An ether that moves with the Earth meets the problem that killed Stokes's ether in the 19th century. Annual aberration needs a rule for what happens to light at the boundary between the Sun's ether and the Earth's. Su said so himself. - Check that a formula is used inside the range it was derived for. The quasar calculation uses a weak-field formula at Phi_g/c^2 = 4.1, a Newtonian potential where the escape speed would be 2.9 c, and a density chosen to fit. - Evaluate rival formulas at the same instant. The "-2v^2/c^2" radar Doppler difference vanishes when the radial speed is taken at the reflection event rather than the emission event. A good reminder that apparent second-order effects are often bookkeeping. - Check a model against its own pieces. Su's one-way turntable prediction cancels when his own clock-rate formula is applied to the two clocks on the table.
What is genuinely open. - Whether any version of a nested-ether model can reproduce annual aberration; nobody has written an interface rule. - Whether the Edwards-Kenyon-Lemon potential is real and whether it has anything to do with F2: measuring the current distribution in the coil, a repeat with an ordinary copper coil at Su's predicted 0.1 microvolt level, and reading the 1990 and 1992 papers would settle it [arbiter.md Q2]. - What Su's wave equation really predicts for the gravitational redshift of an optical (electronic) clock, done without dropping factors of n, and how it compares with optical-clock tests [arbiter.md Q1]. - Whether Su's full force law (the extra terms F1 and F3) is ruled out by high-energy electron data such as Bertozzi 1964; nobody in the review worked it through. - The neutron Bragg-angle asymmetry: one ambiguous comparison; an independent wavelength and a known zero would settle it. - Whether the Herrmann bound is exactly equivalent to Su's prediction once his equations are applied to the two-cavity geometry with a real spacer (section 5, Step 5 uses the natural reading). - The quasar association and periodicity question, which is outside what Su's mechanism can decide. - What happened to Su after 2006, and who saw the 2008 book through the press.
8. Check yourself
- A GPS signal takes 0.075 s to reach a receiver on the equator. The receiver moves east at 465 m/s. What is the largest range correction the Earth's rotation can cause, and what would it be if the Earth's 29.8 km/s orbital speed had to be included instead?
- Using Su's eq. 14, what peak-to-peak frequency swing would his model predict for a 1064 nm laser cavity at the equator? At the North Pole?
- In section 5, Step 3, contraction makes the cavity anisotropy disappear. Su's model has time dilation of atomic clocks but no contraction. Explain in two sentences why time dilation alone cannot remove the angle dependence.
- Su's quasar cloud has Phi_g/c^2 = 4.1. Compute z from Su's formula z = sqrt(1 + 2 Phi_g/c^2) - 1, and say what the GR formula 1 + z = 1/sqrt(1 - 2 Phi_g/c^2) gives.
- Why does the absence of an orbital Sagnac term in GPS not discriminate between Su's model and general relativity?
9. Further reading
Mainstream. - E. F. Taylor and J. A. Wheeler, Spacetime Physics, 2nd ed. (Freeman, 1992): special relativity from scratch, with the train-and-light examples this primer leans on. - D. J. Griffiths, Introduction to Electrodynamics, chapter 12 (Electrodynamics and Relativity): Lorentz transformation of fields, Doppler shift. - N. Ashby, "Relativity in the Global Positioning System", Living Reviews in Relativity 6, 1 (2003): the ECI frame, the Sagnac correction and clock rates in GPS, the mainstream account of exactly the observations Su starts from [memory, high]. - C. M. Will, "The Confrontation between General Relativity and Experiment", Living Reviews in Relativity (2014 edition): PPN parameters, light bending, Shapiro delay, gravitational redshift [memory, high]. - D. Mattingly, "Modern Tests of Lorentz Invariance", Living Reviews in Relativity 8, 5 (2005): test theories (Robertson-Mansouri-Sexl) and cavity experiments [memory, high]. - Y. Z. Zhang, Special Relativity and its Experimental Foundations (World Scientific, 1997): one-way versus two-way light speed, synchronisation conventions, Lorentz ether theory [memory, medium]. - A. Brillet and J. L. Hall, "Improved laser test of the isotropy of space", Phys. Rev. Lett. 42, 549 (1979) [web: https://vivo.colorado.edu/display/pubid_108412, abstract]. - S. Herrmann et al., "Rotating optical cavity experiment testing Lorentz invariance at the 10^-17 level", Phys. Rev. D 80, 105011 (2009) [web: https://arxiv.org/pdf/1002.1284]. - E. J. Post, "Sagnac effect", Rev. Mod. Phys. 39, 475 (1967) [memory, high]. - M. Le Bellac and J.-M. Lévy-Leblond, "Galilean electromagnetism", Il Nuovo Cimento B 14, 217 (1973) [memory, medium]. - N. Ashby et al., Phys. Rev. Lett. 98, 070802 (2007): hydrogen masers against caesium fountains as a test of local position invariance [web: https://www.ncbi.nlm.nih.gov/pubmed/17359010]. - W. F. Edwards, C. S. Kenyon and D. K. Lemon, "Continuing investigation into possible electric fields arising from steady conduction currents", Phys. Rev. D 14, 922-938 (1976): the reported potential from a superconducting coil, quadratic in current (section 3.6) [read: deep/f2a section 13; web: https://isidore.co/misc/Physics%20papers%20and%20books/Zotero/storage/C6B7H3S5/PhysRevD.14.html]. - E. Hawkins, S. J. Maddox and M. R. Merrifield, "No periodicities in 2dF Redshift Survey data" (2002) [web: https://arxiv.org/pdf/astro-ph/0208117].
Su's own key works (in the folder unless noted). - "A local-ether model of propagation of electromagnetic wave", Eur. Phys. J. C 21, 701-715 (2001). The core paper: GPS, Michelson-Morley, the 62 Hz prediction, aberration admitted unsolved. - "Reinterpretation of the Michelson-Morley experiment based on the GPS Sagnac correction", Europhys. Lett. 56, 170 (2001) (research/files/wayback-qem/f1b.pdf). - "A wave-propagation model for gravitational effects on light deflection and radar echo time", J. Electromagn. Waves Appl. 15, 259 (2001). - "A local-ether wave equation and speed-dependent mass and quantum energy", Eur. Phys. J. B 24, 231 (2001). - "A local-ether wave equation unifying gravitational and electromagnetic forces", J. Electromagn. Waves Appl. 16, 403 (2002); "A local-ether wave equation and the consequent electromagnetic force law", J. Electromagn. Waves Appl. 16, 1275 (2002). - "Reinterpretation of the effects of the Earth's rotation and gravity on the neutron-interferometry experiment", Europhys. Lett. 60, 1 (2002). - "Reinterpretation of Fizeau's experiment with moving medium in accord with the Sagnac effect due to earth's rotation", J. Electromagn. Waves Appl. 19, 559 (2005). - arXiv preprints: physics/0211019 (modified Lorentz force), physics/0208082, physics/0208084, physics/0208085, physics/0506211 (Compton), physics/0608164 (quasars), physics/0512179 (Barnett). - Quantum Electromagnetics: A Local-Ether Wave Equation, NTHU Press, 2008 (not in the folder; outline and web summary only).
Advocates and related work (stake noted). - J. Marsen, "The Local-Ether Model and Quantum Electromagnetics Theory of Prof. Ching-Chuan Su", viXra 2105.0110 (2021), written "to promote the work of the late Prof. Ching-Chuan Su"; and viXra 1906.0336 (2019), the satellite Michelson-Morley proposal. - S. J. G. Gift, "Detection of the Ether Using the Global Positioning System", Applied Physics Research 13(1), 12 (2021); a proponent of preferred-frame readings of GPS. - P. Beckmann, Einstein Plus Two (Golem Press, 1987): the same local-ether postulate [web, per sites.md; text not read]. - J. H. Field, Int. J. Mod. Phys. A 34, 1930014 (2019) and E. T. Kipreos, Int. J. Mod. Phys. A (2024): mainstream-journal discussions of preferred frames and simultaneity that cite Su [web: OpenAlex abstracts].
Answers to Check yourself
- Largest correction = (receiver speed) x (flight time) = 465 x 0.075 = 35 m, reached when the receiver moves straight along the line of sight (the actual correction is R_t . v/c, smaller for other geometries). With 29.8 km/s it would be 29,800 x 0.075 = 2,235 m, about 2.2 km. GPS works to metres without such a term, which is Su's argument, and also what GR's free-fall (equivalence principle) argument predicts.
- At the equator Delta-f/f = 1.2e-12 x cos^2 0 = 1.2e-12. The frequency is c/1064 nm = 2.82e14 Hz, so the swing is 1.2e-12 x 2.82e14 = 338 Hz peak to peak. At the North Pole cos 90 deg = 0: no rotation speed, no effect. (That cos^2(latitude) dependence is the direct test the arbiter names: compare two latitudes.)
- Time dilation slows the lab clock by the same factor whatever way the cavity points, so it rescales every round-trip time equally and cannot change their ratio. Only something that depends on orientation, such as a contraction along v, can cancel the extra beta^2/2 on the arm that points along v.
- Su: z = sqrt(1 + 8.2) - 1 = sqrt(9.2) - 1 = 3.03 - 1 = 2.03 (the arbiter got 2.04 from unrounded inputs; Su, with 4.2, got 2.07). GR: 1 - 2 x 4.1 = -7.2, so the square root is of a negative number and there is no answer: the cloud would be inside its own Schwarzschild radius (8.2 times its radius). The two formulas agree only when Phi_g/c^2 is small, and Su's was derived only for that case.
- In GR the Earth falls freely around the Sun, so a frame centred on the Earth and not rotating is locally inertial, and the orbital velocity drops out of local physics except for tidal terms. GR therefore predicts no orbital Sagnac term too. A null result that both theories predict cannot favour either.