Ives: Arbiter (STEP 3, redone after QE round 2 MAJOR)

Inputs weighed: qe.md (round 2), facts.json, claims.md (rewritten 17:01), physicist.md, mathematician.md, contrarian.md (all redone 17:04-17:10 after the 31 Ives papers arrived), deep/, sites.md, research/. This file replaces the earlier arbiter.md in full. Material from the earlier arbiter and its QE revision that is still true is carried forward (Pobedonostsev-Parshin, Faraj, Lalli, Preikschat, Sagnac, clock paradox, restored material, absence of evidence) so nothing pertinent is lost. Tags: [read] from a project document (deep read, _text, review file); [web] with URL; [memory] with confidence; [calc] my own arithmetic in this step (project venv, 2026-10-07).

Changes after QE

Verdicts that changed from the previous arbiter.md (the one written at 15:15 and revised for QE round 1):

1. Claim 4 (gravitation): Never properly tested (abstract only, low confidence) is replaced by a split. For a single static mass (Mercury 43 arcsec per century, light bending 1.75 arcsec, red shift): Established (they are general relativity's own numbers; Ives derived them, nothing fitted). The ether-and-forces reading and the transverse force as "a separate force": Not testable as stated. Extensions Ives never worked out (Shapiro delay, pulsars, waves, many bodies) and the large-interferometer conjecture: Never properly tested. The earlier statements "velocity terms presumably chosen to reach 43" and "one matched number" are withdrawn: Refuted by calculation.
2. Claim 1: still Established, but strengthened. Ives-Stilwell Part II (1941, shift 0.113 A against the 0.09 A doublet separation) answers the 1938 caveat that every earlier shift was below the doublet separation.
3. Claim 3: labels unchanged, evidence changed. "SR is internally contradictory" stays Refuted by calculation, now from the 1949 paper itself, not the 1953 summary. New sub-verdict: the 1945 claim that the relativity principle is "superfluous" is Not testable as stated, and in logic it is not shown (the principle re-enters through the premises).
4. Claim 7 (new): Ives's separate physical claims. 7a photon will go the way of the caloric: Tested, not supported. 7b radiation pressure on a moving absorber: formula Established, "earlier values wrong" Not testable as stated. 7c dishing disk: Never properly tested. 7d ripple-tank analogue: Established as a classical-wave analogy only. 7e, 7g: Not testable as stated. 7f large interferometer: Never properly tested.
5. Unchanged: claims 2 (2A Established; 2B Tested, not supported; 2C Not testable as stated; 2D Refuted by calculation), 5 and 6.

Source gap (repeated and updated)

The manifest holds 38 documents: the 7 original ones and 31 papers by Ives, 1924-1951, obtained legally under follow-up Q10. Depth of reading: - Deep-read from page images and text (equations checked): Ives-Stilwell 1938 and 1941 (Part II); Ives 1937 (JOSA 27:263, 27:310, 27:305); 1938 (28:296); 1939 (29:183); 1945 (Phil. Mag. 36:392); 1947 (Phys. Rev. 72:229); 1948 (JOSA 38:413); 1949 (JOSA 39:757); Pobedonostsev-Parshin (6 pp.); Faraj; Lalli 2014; Preikschat passages; the 1952 and 1953 mass-energy papers. - Read by a reviewer or QE in the extracted text, not deep-read: the 1942 radiation-pressure and ripple-tank papers, the 1939 disk passage, the 1951 photon passage, the 1947 historical note, Ruderfer's appendix (Moray 5th ed. pp.271-274). - Digest only: the 1924-26 photoelectric papers, 1933 Ives-Fry, 1937 27:177 and 27:389, the 1939 chronotopic-interval and double-Fizeau-wheel papers, 1940, 1943-44 wave packets, 1948 JOSA 38:879, 1950 and 1951 papers, the Rumford lecture beyond the photon passage. Verdicts on claim 7 items rest on these plus quoted lines and are marked. - Not obtained: JOSA 42:540 (1952, legal copy), the 1953 JOSA 43 notes, Rev. Sci. Instr. 9:283 (1938), Nature 168:246 (1951), Proc. Roy. Dublin Soc. 26:9 (1952), Hazelett and Turner (1979), Bridgman's 1959 memo and Lalli 2013 (known only through Lalli 2014 or an abstract), Jones JOSA 29:337 and Otting (known only through Ives-Stilwell II), the full Pobedonostsev papers (ZhTF 1989 issue 3; Khlopin preprint), Will, Sci. Am. 1974 (cited by Ruderfer). - Known digest errors, do not use: H-alpha for H-beta (1938); "He" for H2; the 1941 factor printed (1 - v/c^2)^(1/2); the 1942 ripple exponent; the JOSA 29:294 digest sentence "c_R = c y^(1/2)" (the deep read has c_R = c y, c_T = c y^(1/2)); the 1948 digest's garbled mass formula and orbit equation. - No verdict now rests on [memory] alone; the points that do rest partly on [memory] are marked below.

Numbers recomputed or checked by the arbiter [calc]

1. Mercury: GM_sun = 1.32712e20 m^3/s^2, a = 5.791e10 m, e = 0.2056: 6 pi GM/(c^2 a (1 - e^2)) = 42.98 arcsec per century; one sixth is 7.16 [calc]. The reviewers agree. Mathematician's caveat kept: "one sixth" assumes the Newtonian force is held fixed while the mass grows; if the force scales with the moving mass the deep read's integration gives one third. Both readings are special relativity; neither gives 43.
2. Light bending at the solar limb: 4GM/(c^2 R) with R = 6.957e8 m gives 1.7512 arcsec [calc]. claims.md writes 1.7516; the correct figure is 1.7512 (the physicist's and mathematician's value). The "half" value, 0.8756, results if radial contraction is dropped.
3. Ives-Stilwell II: lambda0 beta^2/2 for H2+ at 42,260 V gives 0.1095 A, at 43,000 V 0.1114 A, at 28,300 V 0.0733 A [calc]; reported computed values 0.1098, and 0.0717 at 28,300 V (2% below mine). The shift 0.113 A exceeds the 0.09 A doublet separation by a factor 1.24; the centre of gravity of two lines cannot leave the interval between them, so intensity transfer cannot fake a shift larger than the separation (the 0.09 A figure is inferred by the authors as "about half the separation", not printed).
4. Radiation pressure on an absorber moving toward a source (7b): solving momentum balance F = u(1 + b) + E_a b and energy balance E_a = u(1 + b) + F b (c = 1, b = v/c) with SymPy gives F = u(1 + b)/(1 - b) exactly [calc]. This is the physicist's derivation, which I reproduced, and it equals the plane-wave energy density in the absorber's rest frame, u' = u(1 + b)/(1 - b), the mathematician's reading.
5. Ripple tank: 1 - 0.585^2 = 0.658 against measured 0.852; 11.28/19.20 = 0.5875, so the printed 0.585 is a small rounding in the paper; the comparison uses (1 - v^2/c^2), not its square root (0.811) [calc].
6. The 1938 Table III numbers stand from the earlier arbiter: slope 0.972 +/- 0.031, mean observed/computed 0.964, chi-square about 1.7 for 8 points; Table II plates 0.036-0.052 A; H3+ at 6,788 V reproduced (mass 3). The mean deficit of 3-4% is the reviewers' calculation, not discussed by Ives and Stilwell, and not clearly significant.
7. Sensitivity of the perihelion to the second-order term in the field (Mathematician, SymPy): with y = 1 - 2m/r + kappa m^2/r^2 the advance is (6 - kappa)/6 of the Ives/GR value. I accept the algebra, and the consequence below (claim 4).

Web checks by the arbiter (budget 8, used 5 including the earlier three)

• Earlier, retained: Alvager et al. 1964 and de Sitter 1913 [web: https://en.wikipedia.org/wiki/De_Sitter_double_star_experiment; https://docente.unife.it/guido.zavattini/allegati/alvaeger-pl1964.pdf]; Mandelberg and Witten 1962 existence [web: https://arxiv.org/pdf/physics/0008012]; Sommerfeld one sixth [web: https://arxiv.org/pdf/physics/9805032].
• New: Nordstrom's scalar theory has the Newtonian static limit but predicts no light deflection, because the electromagnetic energy-momentum tensor is traceless and the metric is conformally flat [web: https://en.wikipedia.org/wiki/Scalar_theories_of_gravitation, search summary]. This upgrades the Mathematician's [memory: high] and supports the point that "a Newtonian static force" alone does not give GR's light bending.
• New: Birkhoff's flat-space theory postulates the super-hard equation of state (pressure equal to energy density, sound speed equal to c); the absence of a derivation from more general propositions was held against it (Weyl) [web: https://arxiv.org/pdf/gr-qc/9912003, search-summary level; I did not open the paper]. This confirms the sound-speed point the physicist and QE held at [memory: medium/low]; confidence now medium (summary level).
• New: Grangier, Roger and Aspect 1986 measured fewer coincidences at a beam splitter than any classical wave model allows, a violation of 13 standard deviations, consistent with a single-photon state [web: https://courses.physics.illinois.edu/phys513/sp2019/reading/week1/GrangierSinglePhoton1986.pdf, search summary; paper not opened].
• Not web-checked (no verdict turns on them alone): Hafele-Keating, Mansouri-Sexl, Lamb-Scully 1969 semiclassical photoelectric treatment, Kimble-Dagenais-Mandel 1977 antibunching, the PPN perihelion factor (2 + 2 gamma - beta)/3, Hughes-Drever.

Where the three reviewers agree

• Claim 1 Established (Part II answers the doublet caveat; Pobedonostsev-Parshin kept as unreplicated dissent).
• Claim 2: ether-versus-SR is not decidable by this experiment; weak form Established.
• Claim 3: every route is equivalent to SR in prediction; the 1949 "internal inconsistency" charge is Refuted by calculation.
• Claim 4: all three withdrew their own earlier "Never properly tested / fitted to one number". Ives's gravitation is the exact Schwarzschild solution in ether language. Ruderfer's "necessarily equivalent" is stronger than the mathematics supports. The mathematician found no nonzero fitted constant and flagged the exact form of y as the real input.
• Claim 7a: Tested, not supported. 7d: Established as an analogy only.
• On the owner's questions all three give the same answer (below).
• All keep the Pobedonostsev-Parshin dissent, the Robertson stake, the Lalli/LoC letter and the single-source Princeton account in the record; none finds evidence that the experiment or the papers were suppressed.

Corrected claim list and final verdicts

Claim 1. Moving clocks run slow (the Ives-Stilwell second-order Doppler shift)

Corrected statement: Ives and Stilwell (1938) measured the centre of the Doppler-displaced H-beta lines (hydrogen canal rays, H2+ and H3+, 6,788-18,350 V) shifted by lambda0 beta^2/2 (0.011-0.047 A against a first-order shift of about 21 A), agreeing "within the probable errors" (their words) with the prediction. About 5% statistical (probable error 0.0025 A per point); Mandelberg and Witten's 1962 reassessment gives a real uncertainty of 10-15% (second hand through Giuliani; their own exponent 0.498 +/- 0.025). The 3-4% mean deficit is a reviewer's calculation and not clearly significant. The sign (red shift of the centre) follows from the formula; the paper never states it. The printed formula on p.216 has a wrong exponent for a wavelength (a slip in the original). Part II (JOSA 31:369, 1941): the tube ran to about 43,000 V; observed 0.113 and 0.1145 A at 42,260 and 42,280 V against 0.1098 and 0.1073 computed, "over 0.11 A as against the previous maximum of 0.047 A", with "the relationship previously observed ... maintained". That is above the doublet separation, so the 1938 caveat ("the least satisfactory feature") is answered on the authors' own numbers and by arithmetic. Limits, kept: coarser plates; nine H2 points about 3% above the computed curve, the opposite sign to 1938, which the authors attribute to a mirror under 100% reflecting (the reviewers disagree by one point on the count, six above and three at or below against six above, one equal, two below; check against Table I before quoting); one H3O entry looks misprinted. Jones's 1939 objection (imparted velocity): about 750 m/s with the mirror covered at 14,000 V, none within error with it in; double-ended tube 0.0744 A against 0.0717 computed at 28,300 V (one point, no error bar). These are the authors answering their critics, which is fair but not independent. Disagreement: none on the verdict. Stronger argument: the accumulated record (Mandelberg-Witten 1962; Saathoff 2003; Reinhardt 2007; Botermann 2014 at beta = 0.338, below 4e-9; CERN muon ring, gamma = 29.3, agreement to 2e-3) is far stronger than the 1938-41 data. Dissent kept: Pobedonostsev and Parshin (Khlopin Radium Institute; SPIE 1121, Interferometry '89, pp.36-41, 1989; further papers 1991-95, titles only). Table [read: deep read, p.4]: second-order shift 40-82% below SR at 150-300 keV (observed/SR 0.17-0.60), within about 10% at 1.5-2.0 MeV. Stated error 0.04 A per line, no systematic budget; deficit erratic point to point; two separate optical arms, one through a fibre; a half-degree misalignment gives a deficit of the right size (a mechanism, not proved); their classical model predicts about +5%, not a deficit; the voltage-to-speed step is not the source (relativistic correction 8e-5 [calc, Mathematician]). No replication or rebuttal found; the decisive record is the ZhTF 1989 account. Faraj (emission-theory advocate, citing Gezari 2009 and Kantor 1971): says the experiment shows only a longitudinal asymmetry. Not Ives's result; his algebra is invalid (his own eq. 3.3 gives twice the Larmor-Lorentz shift). What Ives's Rev. Sci. Instr. 9:283 (1938), "Unsymmetrical Doppler Effect in Hydrogen Canal Rays", reports is unknown. Verdict: Established (modern tests to parts per billion; 1938-41 data alone modest). Dissents kept in the record, not suppressed and not replicated.

Claim 2. The result shows a real slowing in a fixed ether and favours Larmor-Lorentz over Einstein

Corrected statement: Ives and Stilwell wrote nu = nu0(1 - V^2/c^2)^(1/2) with nu0 "stationary in the ether" and called the result "more decisive evidence for the Larmor-Lorentz theory" than null results, because nulls could be explained more simply by entrained ether or ballistic light. With Kennedy-Thorndike (KT) and Michelson-Morley (MM) it fixes the contraction at l = 1 (the 1947 Note excludes Voigt's, equal-volume and no-clock-change cases). Ives's 1937 papers (JOSA 27:177, 27:389) show KT "cannot by itself test the Larmor-Lorentz hypothesis" and propose the canal-ray experiment; his 1947 note credits Einstein alone with believing the equations and naming the shift an experimentum crucis. In Ives's own algebra the laboratory's ether velocity drops out (Mathematician). Part II's six pages use neither "ether" nor "relativity". Ives concedes the ether's "elusiveness by any measurements of the sort we have been considering" (1937). Split verdict: - 2A, IS plus KT plus MM fix clock rate and contraction: Established. - 2B, IS disfavours entrained-ether and ballistic theories: Tested, not supported (those alternatives also fail on other evidence: aberration, Fizeau, de Sitter, Alvager). That IS favours the ether over SR: Not testable as stated (Jones 1939 and Robertson give the same predictions). - 2C, a real undetectable preferred frame and absolute simultaneity versus Einstein's interpretation: Not testable as stated; as a 1938 position, Historically superseded but instructive. Bridgman's objection ("a complete negation of the operational attitude", via Lalli) is on file. - 2D (side argument), Sagnac and Michelson-Gale show a fixed ether and SR cannot handle rotation: Refuted by calculation as a statement about SR (von Laue 1920, Langevin 1921). Ives's 1938 closed-path paper itself derives the standard first-order result (4 A w/c^2) and agrees transported clocks give c; he objects to interpretation. "Uniform motion only" is from the 1953 Genesis paper. Whether his reading of Sagnac differs measurably from SR's: Not testable as stated. - Aberration as Ives's "strongest evidence" for the ether (1937, p.10; "sheer weight of enormous probability", "not metric"): Not testable as stated (SR predicts aberration too). Interested parties: Bell Laboratories (employer, funder; 1947 Record: "first direct proof") and Einstein's reported reaction ("most direct proof yet", NYT 27 Apr 1938, lead only) show both camps claimed the result at once. Robertson (referee, then author of the memoir's assessment) has a stake; his equivalence judgement stands on its own.

Claim 3. An alternative theory from classical ether physics (transported clocks, conservation laws, rod-clock quotients)

Corrected statement: three routes, each read. - (a) 1937 transported rods and clocks. Clocks set by transport at a finite observed speed; results depend on the transport speeds and reduce exactly to Lorentz as they go to zero; for finite speeds they equal SR applied to the same transport procedure (40-digit check). The difference from Einstein-synchronised clocks is a synchronisation convention, first order in transport speed (about 1.7 ps over 1 km at 300 m/s; 5.6e-17 s for 1 m/s over 10 m). Robertson: valid, "completely equivalent in substance". Misprints in eqs. 14, 25, 33, 34, 38-40 change no conclusion. "Approach but never coincide" is true of the formulas, not of the predictions. - (b) 1943-45 conservation-law route. Mass-velocity law from classical radiation pressure on a moving mirror, momentum E/c, mass E/c^2 (pre-SR, Maxwell-Poynting-Poincare 1900 [memory: high]), and Newtonian conservation laws; then D1, D2, D3 from a light-clock argument. Three premises do work the conservation laws alone do not: the inertia of energy; an unexplained step that an extra term "must drop out" (a counterexample with a free function h = 0.1 satisfies conservation exactly with a different final speed, 0.709 against 0.696 [calc, deep read]; h = 0 gives relativistic Doppler and velocity addition, i.e. the relativity principle for that process); and clock universality and isotropy (MM and KT). The answer to QE's question: E/c^2 for radiation is not itself relativistic input, but the relativity principle is carried by the premises, so "superfluous" is not shown. No measurable difference from SR is named. - (c) 1948-51 operational route. "Velocity" reserved for ether quantities; "rod-clock-quotient" first defined in JOSA 38:879 (1948), not in 1953; c "firmly established by experiment" is invalid because only out-and-back speed had been measured; the constant ratio "is not equal to c and should not be called the velocity of light". The 1949 charge: SR "leads to an internal inconsistency if clocks are admitted for distant time setting", and the Lorentz transformations are "an approximation". Deep read plus SymPy: his eqs. (19)-(20) are term for term the Lorentz transformation with clock readings lagging Einstein time by k r'; SR predicts exactly his one-way result c/(1 - ck) for a clock carried at speed q (about c + q/2, tending to c as q goes to 0); the round-trip speed is c for every k (Reichenbach family, Mathematician). Ives names no experiment that separates the theories; the 1950 and 1951 papers cite only MM and IS. He rejects the q-to-0 limit because it "could not be performed", an operationalist point. What survives is the conventionality of one-way simultaneity, a recognised point. Disagreement: none of substance now; the earlier split (Physicist "Historically superseded", Mathematician and Contrarian "Not testable" with a source gap) closed when the 1945 and 1949 papers were read. Verdicts: Ives's alternative theory as a re-derivation of SR: Not testable as stated (equal predictions) and Historically superseded but instructive (the 1945 route is the most instructive: it shows how little of SR is extra). The charge that SR is internally contradictory (transport synchronisation; the 1937 clock paradox, where Ives does not discuss the standard resolution that the travelling clock changes frame, though his Eq. 17 matches the textbook lag at leading order): Refuted by calculation, now from the 1949 and 1937 papers themselves. The 1945 claim that the relativity principle is superfluous: Not testable as stated (and not shown in logic). The one-way-light-speed statement: Not testable as stated; partly right as conventionality. Miller's remark that the theory was never "developed to the point where it could be seriously considered as an alternative" is one historian's characterisation; the equivalence is the stronger ground.

Claim 4. Gravitation and interferometers: the Schwarzschild solution as forces in an ether (JOSA 29:183; Phys. Rev. 72:229; JOSA 38:413)

Correction to the earlier arbiter: it said "the only content read is the 1948 abstract" and that the velocity terms were "presumably chosen to reach 43". That was an inference from an abstract and is wrong; the primary papers show no fitted constant. Corrected statement. - 1939: from three assumptions (a body at rest in the field behaves like one moving at the speed gained falling from infinity, V^2 = 2GM/R; a null interferometer reading is preserved; Fitzgerald contraction relative to the local light speed) Ives gets light speeds c_R = c y, c_T = c y^(1/2) and ds^2 = c^2 y dT^2 - R^2 dtheta^2 - dR^2/y, y = 1 - 2GM/(Rc^2): exactly the Schwarzschild form in Schwarzschild coordinates, which he calls "the Schwarzschild formula of general relativity". Red shift and bending are stated in words; his light-speed field gives the full limb deflection 1.7512 arcsec [calc]. His limits, in his words: "of purely optical significance"; "no non-Euclidean geometry, nor new law of gravitation"; it "calls for anisotropy of the ether". An isotropic-light-speed ether is rejected as "in conflict with observation". Large interferometers: if invariance fails, "a sufficiently large and sensitive interferometer could be used to detect motion in a gravitational field"; the 1939 chronotopic-interval paper tabulates contraction and clock-rate combinations that all give a null and says choosing needs experiments "where the gravitational parameter varies", such as near the Sun (digest). - 1947: Birkhoff's flat-space equations, rewritten in ordinary time and force, contain a force normal to the motion, "v^2/c^2 times the component of the gravitational force normal to the direction of motion" (about 2.5e-8 for Mercury [calc]). Ives does not endorse Birkhoff as physics ("an independent physical derivation of these transverse forces would be welcome"); the Gauss analogy is a resemblance. Birkhoff's theory reproduces the classic tests but admits only a perfect fluid with sound speed c and has no derivation from more general propositions (medium confidence, summary-level web check above). - 1948: the postulate gives m_g = m0/S, force equations, the orbit equation d^2u/dtheta^2 + u(1 - 6 G^2M^2 b^2/(c^2 h^2)) = K-bar, and a perihelion advance 6 pi GM/(c^2 a (1 - e^2)) = 42.98 arcsec per century [calc]. "The force equations are actually implicit in the general relativity solution", and r^2 thetadot/S = h "is a relation also obtained in the general relativity treatment". Mass increase alone gives one sixth (conditional on holding the force fixed; one third otherwise). - Where Ives overreaches, or where the result rests on a choice (same rigour for the inventor): (1) the exact y with no (GM/Rc^2)^2 term is a postulate, and it is the real input: for y = 1 - 2m/r + kappa m^2/r^2 the advance is (6 - kappa)/6 of the Ives/GR value [calc, Mathematician]; Ives's bookkeeping remark that departures "would enter as additional terms in 1/c^4" and may be dropped does not hold for forces. He states the exactness openly and it coincides with GR, so this is an input, not a fudge. "No free parameter" is true given kappa = 0. (2) The "also exact" claim in the note added in proof holds for the three governing integrals but not for the printed force equation (23), which is first order in 1/c^2 and differs from the exact radial equation by terms that give no perihelion advance (circular-orbit difference about 2.5 parts in 1e8 for Mercury, absorbed in GM). (3) F_theta = 0 and the split of the transverse force into a pair are Ives's choices; footnote 9 concedes GR gives "no ground for favoring" his pair over Birkhoff's symmetrical one, and the two give identical orbits. (4) Eddington (1917, 1923) and Phillips (1922) had given the same relations; Ives cites them. It is a re-derivation, not a prediction made in advance. Ruderfer: "a classical derivation of the Schwarzschild metric, hence ... an identical explanation of the three famous tests of relativity"; "any Lorentz-ether theory of gravitation which provides the static Newtonian gravitational force is necessarily equivalent observationally to general relativity"; "the continued neglect of Ives' gravitational theory is unwarranted" [read: Moray 5th ed. pp.272-273, as read by the reviewers; Ruderfer is an ether advocate with his own neutrino-sea model]. The first sentence is accurate about what Ives did. The second is too strong: Nordstrom's scalar theory has the Newtonian static limit and predicts no light deflection [web above], so what gives Ives the GR numbers is the equivalence postulate with the exact y, which builds in beta = gamma = 1, not the presence of an ether or a Newtonian static force. Ives's own hedging is more accurate than Ruderfer's gloss. The theory covers one static spherical source and a test particle; it has no field equations, no radiation, no two-body treatment. Reception and neglect: refereed and printed in Phys. Rev. and JOSA, so "dismissed without a hearing" is not shown. Robertson's report on the 1947 paper called the Lorentzian framing "a bit (ca. 40 yrs) out of date", and it was published [read: Lalli 2016 via reviewers; web: https://pmc.ncbi.nlm.nih.gov/articles/PMC4856100/]. No review of the gravitation papers is on file, favourable or not. Two readings of the neglect, both kept: Ruderfer's (a useful, transparent derivation overlooked) and the Contrarian's (it adds no prediction and its author says so, so there was nothing to test or cite: neglect by irrelevance, not suppression). Citations beyond Ruderfer are unchecked. Agreement on the numbers, disagreement on weight: the Physicist and Contrarian give the single-mass predictions as Established (GR's); the Mathematician adds the kappa caveat. I accept all three: the predictions are established because they are GR's; the caveat concerns Ives's derivation, not the predictions. Final split verdict: - (i) Predictions for one static central mass (Mercury 43 arcsec per century, light bending 1.75 arcsec, gravitational red shift): Established (GR's values; Shapiro delay follows from the same light-speed field, though Ives never states it). - (ii) The ether reading ("anisotropy of the ether", forces in flat space, absolute space and time) and the claim that the transverse force is a separate physical force: Not testable as stated (no measurable difference; Ives concedes it; his preference for his force pair over Birkhoff's likewise). - (iii) As an independent, complete theory of gravitation (many bodies, radiation, pulsars, waves, Shapiro delay as Ives's own prediction): Never properly tested (not dismissed; never worked out; no field equations). - (iv) The 1939 large-interferometer conjecture and the claim that the 1948 force equations are "also exact": Never properly tested (the exactness holds for the integrals only). - (v) The earlier belief that the velocity terms were fitted to 43 arcsec: Refuted by calculation. - Confidence: high for (i), (v); medium for the status of Birkhoff (summary-level web check).

Claim 5. Einstein did not derive E = mc^2 in 1905

Unchanged from the earlier arbiter. Corrected statement: Ives (JOSA 42(8):540-543, Aug 1952; Ohanian's "Am. J. Phys. 42" is his slip) accepted E = mc^2, credited Poincare (1900, 1904), and objected, following Planck 1907, that Einstein's step "H - E differs from kinetic energy K only by a constant" used K = (1/2)mv^2, valid "only as a first approximation". "Circular" is the Stachel-Torretti/Ohanian label; Ives narrowed it in 1953 ("something must be added ... from valid physical considerations, not by assumption"). Jammer 1961, Arzelies 1966 and Miller 1981 sided with Ives (via Ohanian). Label disagreement and resolution: Physicist and Mathematician say Refuted; Contrarian says "Internally inconsistent as Ives set it up". The factor L/((m - m')c^2) appears only after Ives adds the particle formula K = mc^2(gamma - 1), which Einstein avoided; Einstein's chain (H0 - E0) - (H1 - E1) = L(gamma - 1) uses no knowledge of the result. "Refuted by calculation" is the exact scale term; "Internally inconsistent" would say Ives contradicted himself, which his 1952 text does not strictly do. Ives's Appendix claim that subtraction gives only 0 = 0 is wrong (it gives L(gamma - 1)). Verdicts: (i) circular: Refuted by calculation (partly withdrawn by Ives in 1953). (ii) 1905 proof not fully general (slow internal motions): Established as history (Planck 1907, von Laue 1911, Klein 1918, Ohanian 2008). (iii) E = mc^2: Established (Rainville et al. 2005, 4e-7 [web: https://news.mit.edu/2005/emc2]). New cross-link: Ives's own 1942-45 derivations use m = E/c^2 for radiation as input (claim 3b); that is a different starting point, not a contradiction of the 1952 charge.

Claim 6. Atoms are not clocks (Bonizzoni-Giuliani, Giuliani; not Ives's claim)

Corrected statement: the IS formula follows from relativistic energy-momentum conservation in photon emission (Schroedinger 1922; Davisson 1938), numerically identical to the wave Doppler formula; the recoil correction Delta E/(2 E1) = 2.55 eV/(2 x 938 MeV) = 1.4e-9 is within reach of modern experiments but not of 1938-41 data. A. I. Miller held that IS is "the only positive proof" of time dilation. Giuliani does not endorse an ether. Verdict: Not testable as stated (interpretation with no measurable difference at IS-type precision). The Physicist's "Plausible, untested" is replaced for consistency with the scale. Time dilation for real clocks: Established by other tests (muon ring, optical clocks, GPS).

Claim 7. Ives's separate physical claims (not only a different explanation)

Each states whether it is a different prediction or a different explanation. - 7a. The photon "will go the way of the caloric" (1951 Rumford lecture). A different prediction. [read: _text 1951, lines 2185-2191, via QE and two reviewers] "I venture to predict that this will be done, and that the photon will go the way of the 'caloric' that Rumford demolished", after conceding "the difficulty of arriving at an explanation of all optical phenomena, such for instance as the Compton effect". Grounds: photons would have to exist and not exist alternately in loops and nodes; Lorentz's 1910 coherence remark; Poincare's "vestibule"; Beck-Sommerfeld bipartition data (not examined). Weighing: the standing-wave point is sound physics but does not discriminate, since the quantum prediction for a one-photon standing-wave mode is detection probability proportional to |E|^2, exactly what Ives measured (Mathematician; [memory: high]). In 1951 his position was held by serious physicists, and photoelectric and Compton effects have semiclassical treatments (Lamb-Scully 1969 [memory: medium]); what the later experiments contradict is a purely classical field: Grangier-Roger-Aspect 1986, 13 standard deviations [web above]. He could not have known them. Verdict: Tested, not supported. His partial point (waves explain more than the particle picture suggests) is mainstream-compatible. Kept in the record; pertinent to Moray's interest in radiant energy. - 7b. Radiation pressure on a moving absorber (JOSA 32:32, 1942). Ives finds P = E_a(c + v)/(c - v) for a perfect absorber moving toward a point source, and says Lorentz's P = E_i and Poynting-Abraham's P = E_i(1 + v/c) are wrong. Two reviewers worked it. The Physicist derived the force needed to hold constant velocity, including the momentum carried by the absorber's growing mass, and obtained Ives's formula; I reproduced it by SymPy [calc]. The Mathematician reads Ives's value as the plane-wave energy density in the absorber's rest frame, u(1 + b)/(1 - b); the two readings agree. The Contrarian had not checked. Caveat: the OCR leaves the definition of E_a in eqs. (5)-(7) ambiguous, so the units of (7) cannot be fully checked from the text. Poynting's u(1 + b) is the incident momentum rate alone. So Ives's formula agrees with SR for the quantity it names, and the older values are not "wrong" so much as a different quantity. A different explanation (a definition), not a different prediction; the first-order v/c effect was "practically impossible to detect" in 1942 (Abraham, quoted by Ives). Verdict: Established (the formula); that the earlier published values are wrong: Not testable as stated. - 7c. A rotating disk "dishes" (JOSA 29:472, 1939). A different prediction in a narrow sense: the Wood-Hackett "twist" follows from Fitzgerald contraction, and the disk takes a saw-blade shape, paraboloidal at small angular velocity [read: _text lines 197, 234-242, via QE]. Mainstream: the Ehrenfest problem is handled by elasticity (a body cannot be spun up Born-rigidly; spin-up induces stresses). The Physicist's size check [calc, not re-run by me]: 10 cm radius at 100 rev/s gives beta = 2.1e-7, dish depth about 18 nm, contraction strain about 2e-14 against a classical centrifugal strain of about 6e-5. No experiment on record; the digest gives no derivation the Mathematician could check. Verdict: Never properly tested (and far below the classical stress background). Kept. - 7d. Ripple-tank analogues with Stilwell (JOSA 31:14, 1941; 32:25, 1942). A demonstration, not a test of light. Capillary waves on mercury, medium speed 11.28 cm/s against wave speed 19.20 cm/s, reproduce the stationary interference pattern when lengths are scaled by the Fitzgerald factor and periods by the Larmor-Lorentz factor; Fresnel drag and aberration appear. One ratio, 0.852 measured against 0.658, a 30% miss, is attributed to capillary dispersion without quantification. Mathematically this is the covariance of a non-dispersive wave equation under the corresponding Lorentz transformation. Verdict: Established as a classical-wave analogy; not a test of the ether for light. It is the clearest physical picture in the record of what a Lorentz ether means. - 7e. Measured-velocity and rod-clock-quotient statements (1948-51). A different explanation, not a different prediction (claim 3c). Not testable as stated. - 7f. Large interferometer in a field (1939, 1948). Claim 4(iv). Never properly tested. - 7g. Aberration as strongest evidence. Claim 2. Not testable as stated.

The owner's questions, as the arbiter weighs them

Why Ives is interesting to Moray. No Ives paper mentions Moray. The link is M. Ruderfer's appendix "Neutrino Structure of the Ether" (Lett. Nuovo Cimento 13:1, 1975), reprinted in Moray's The Sea of Energy, 5th ed., pp.271-274 [read: Moray text, read by QE and all three reviewers]. Ruderfer argues in three steps, in his words: (1) the rejection of an ether applies "to a Galilean ether, but not to a Lorentz-invariant ether since its description of observation is necessarily identical to that of relativity"; (2) "the most direct verification of the validity of a Lorentz ether, and one which is independent of relativity, is found in the theoretical works of IVES": a classical derivation of the Lorentz transformations (1945) and "of the Schwarzschild metric, hence ... an identical explanation of the three famous tests of relativity"; (3) "the admissibility of the Lorentz ether focuses attention on its properties": he defines the ether as "the totality of all the presently unobservable forms of energy that may exist in the microcosm", proposes a sea of low-energy neutrinos whose "total unobservable energy ... may far exceed the total energy contained in matter, i.e. the vacuum is then a plenum rather than a void". So Ives matters to Moray as a respectable credential: a Bell Laboratories experimenter (Optical Society medallist and past president, NAS member) whose experiments and theory leave a physical ether admissible. Step 3 is Ruderfer's addition, not Ives's: none of the 31 papers assigns the ether an energy content that could be tapped; the 1939, 1945 and 1948 papers give it a frame and a mechanism, not a field, force or energy source. Stake: Ruderfer advocates his neutrino-sea ether and cites Ives as support. The Mathematician adds a tension: a Lorentz-invariant vacuum has an energy-momentum tensor proportional to the metric and no extractable energy, whereas a neutrino sea at rest in a preferred frame would define that frame and show up in frame-dependent tests (Ruderfer's own 1960 rotor proposal; Preikschat's null 2.0 +/- 5.2 cm/s); the Physicist notes it would also have to meet neutrino-detection limits [memory: medium]. Ives's standing-wave photoelectric work (1924-1951) is mainstream and bears on Moray's interest in standing waves and photoelectric conversion, but gives no energy-conversion efficiency.

"Even if Ives is correct then Aether follows Lorentz and Einstein ... a Lorentz Aether contributes no new forces beyond those specified by Lorentz/Einstein so it is only a difference of explanation." All three reviewers and I agree that this reading is correct for the core theory, and it is Ives's own: Robertson, "completely equivalent in substance to the special theory"; Ives 1937, the ether's "elusiveness"; Ives 1939, "no ... new law of gravitation"; Ives 1948, the force equations "are actually implicit in the general relativity solution"; Ruderfer, "necessarily equivalent observationally" (stronger than the mathematics supports, as above). Lorentz's and Einstein's theories share the same equations; the ether adds a preferred frame that no measurement depends on (in Reichenbach's terms, only the synchronisation label epsilon differs; the round-trip speed is c for every epsilon). Four qualifications make the summary exact: (1) Lorentz's version carries an assumption Einstein's does not need as a separate item: all forces holding matter together must themselves be Lorentz-covariant, or rods of different materials would contract differently and MM and KT would not be null (tested to high precision by Hughes-Drever-type experiments [memory: high]); the cost of the ether is that requirement, not a new force. (2) In gravity, relative to SR alone, one needs position-dependent mass, clock rate and length, not only speed-dependent ones; Ives writes this as forces (the 1947 transverse force and the 1948 position factor, which his paper says are the same effect seen two ways), so it is a rewriting, not a new force. (3) Ruderfer's "any Lorentz-ether theory with a Newtonian static force" is not a theorem. (4) Ives's separate claims in claim 7 are not rewritings: the photon prediction is a different prediction and was not supported; radiation pressure is a definition; the disk is untested. The ether in Ruderfer's and Moray's use adds an energy reservoir, which is not part of the Lorentz ether as Lorentz, Einstein or Ives specified it.

What would settle it

1. Claim 1 dissent: the Pobedonostsev-Parshin raw data and error budget (ZhTF 1989 issue 3 pp.84-99; the Khlopin preprint; the 1991-95 papers); an independent canal-ray repeat at 150-2,000 keV with the viewing-arm alignment measured, or the storage-ring data checked over their energy range. A measurement of the reference atoms' imparted velocity with the mirror covered would show whether Part II's Jones control carries an error bar. Ives, Rev. Sci. Instr. 9:283, and Gezari 2009 for the Faraj asymmetry claim.
2. Claim 2: nothing by IS-type experiments. A preferred-frame test (MM and KT types, Moessbauer rotor, satellite clocks, the Mansouri-Sexl parameters) is the real test; current limits are consistent with zero, and Preikschat's rotor limit is a null for the ether and SR alike.
3. Claim 3: no experiment separates the routes; Ives names none. The remaining open piece is a derivation of the free function h in the 1945 moving-platform step from conservation laws alone, or a proof it cannot be fixed without the relativity principle.
4. Claim 4: the existing tests already bound kappa near zero (PPN parameters near 1); a precise second-order test of the field (deflection or perihelion fixing kappa independently) would test the one postulate Ives imposed. For the completeness question: derive from Ives's postulate a two-body or radiating-source result (binary-pulsar period decay) and compare with the observed decay. For the interferometer conjecture: a large interferometer in a varying gravitational field. Citation indices for Phys. Rev. 72:229 and JOSA 38:413 would show how the gravitation papers were received.
5. Claim 5: nothing experimental; Ohanian's slow-motion counterexample is the test of "valid only to first order".
6. Claim 6: a measurement resolving the 1.4e-9 recoil term against a clock-rate-only prediction.
7. Claim 7: 7a is settled by photon antibunching and single-photon anticorrelation (done, 1977 and 1986). 7b: the definition of E_a from a clean copy of JOSA 32:32 and the textbook relativistic radiation pressure on a moving absorber. 7c: Ives's paraboloidal profile compared with the elasticity solution of a thin spinning disk. 7d: a quantified dispersion correction for the 0.852 ratio.
8. Reception: LoC MSS27508 container list (Ives-Butterfield 1950-52 letters), Caltech Robertson papers (his referee report on the 1947 paper), Harvard Bridgman papers (HUG 4234.50), Science Service records (SIA RU007091), NYT 8 May 1938, Bell Laboratories Record 1938, Princeton colloquium notes and Pauli correspondence.

Absence of evidence: where the record goes dark

1. The experiment after 1941. Part II appeared in 1941; no further canal-ray measurement by Ives is in the sources, and he died in 1953. Competing explanations: (a) the question was settled for him; (b) wartime redirection of Bell Laboratories staff [memory: low; nothing in the files states Ives's wartime work]; (c) voltages needed to exclude the fine-structure artefact were beyond the apparatus (Part II then largely met this: 43 kV); (d) Stilwell moved on. Evidence for (c) is in the 1938 paper, and Part II shows the goal was reached, which weakens (c) as the reason for stopping. Decisive record: Bell Labs project files and the LoC container list.
2. Bell Laboratories Record is silent on the experiment in 1938 (text search of May-July 1938 found nothing) yet uses Larmor-Lorentz-Einstein wording in 1947. Explanations: house-journal timing, publicity left to the press (NYT and Science Service, April-May 1938), or search failure on poor OCR. Decisive record: the 1938 issues, page by page.
3. The gravitation theory (1939-1948) vanishes from the physics literature except Ruderfer 1975 and the neo-Lorentzians. Explanations: published but not persuasive (Robertson: "ca. 40 yrs out of date"); it adds no prediction beyond GR, so there was nothing to cite or test; absorbed by GR's success after 1960; or under-read because Ives's reputation was tied to an ether position. Against suppression: it was refereed and printed in Phys. Rev. and JOSA. Decisive record: citation indices.
4. Pobedonostsev-Parshin (SPIE 1989; later papers 1991-95): no replication or rebuttal found. Explanations: obscure venue, post-Soviet funding collapse, or a beam-calibration or alignment flaw nobody bothered to rebut. Silence is not contradiction. Decisive record: raw data and later Russian publications.
5. Ives's hostile-reception account (Princeton 1938; Pauli's remark): one published source (Ives's letter of 24 Oct 1950, quoted in Hazelett and Turner p.90 via Bonizzoni-Giuliani), Ives a party; the contemporary record of Einstein's praise, Bell Labs, NAS membership (1933) and the JOSA presidency (1924-25) concerns other audiences and is not in conflict. Decisive record: Princeton colloquium notes, Pauli correspondence.
6. Hazelett and Turner (1977/1979, rev. 2005): preserved much of Ives's later correspondence; edited by a person with a commercial profile (Hazelett Strip-Casting, per Martin Gardner, a critic), used by anti-relativity groups; quotations are second hand here and should be checked against the LoC originals.
7. The NAS memoir's bibliography stops at 1951 and omits JOSA 40:185 (1950), the 1952 mass-energy paper, Proc. Roy. Dublin Soc. 26:9 (1952) and the 1953 papers; the absence of an E = mc^2 paper from the list is therefore not evidence about that paper. Whether Robertson's relativity assessment shaped the list is not in the record.
8. The 1951 Butterfield letters and the 1959 Bridgman memo are known from one historian's quotations; neither letter nor memo has been seen by any role.

Open questions

• Does the second-order structure of Ives's field (kappa = 0) follow from anything but the postulate, and can his framework be extended to radiation and two bodies?
• Is the printed force equation (23) of 1948 right to first order only, as the deep read finds, and what did Ives's "also exact" rest on?
• What is behind Pobedonostsev-Parshin's deviation at 150-300 keV: viewing-angle alignment, or something else?
• Is the 3-4% mean deficit in Table III systematic (line intensity, calibration) or noise; what does Table I of Part II show exactly (count of points above and below the computed curve)?
• What does Ives, Rev. Sci. Instr. 9:283 report on an unsymmetrical Doppler effect, and is Faraj's reading of it in any sense right?
• Does the 1942 radiation-pressure paper define E_a as the rest-frame energy density, as the two reviewers' derivations suggest?
• Does the dishing disk differ measurably from SR with real elastic stresses?
• Would Ives's 1939 conjecture about large interferometers be detectable by modern gravimetric or optical-clock experiments?
• Is a Ruderfer-type neutrino sea compatible with Lorentz invariance and neutrino limits? Not an Ives question; for the Moray project.
• Dates and places: death 13, 15 or 3 November 1953; place the Century Club, New York (NAS memoir) or Upper Montclair (Wikipedia); birth 31 July (LoC) against 21 July (Chappell wiki) 1882. Immaterial to verdicts.
• The 1951 Ives letter ("classical English school ... against the paradox of the Semitic school") is documented by Lalli (2014, LoC Box 13): Lalli writes the sentence "seems to imply" an "implicit anti-Semitism", discussed in Lalli 2013 (not read). Ives's expressed attitude in one historian's quotation, letter unseen by any role; relevant to motive and alliances, not to the equations. Kept in the record.

Material kept in the record whatever the verdict (carried forward and extended)

• Standing-wave photoelectric work (mainstream, uncontested): Ives called it "probably my most important contribution to science"; selective photoemission of thin alkali films as standing-wave patterns, traced to his 1908 Lippmann colour-photography thesis (NAS memoir). His papers add numbers: selective/normal ratios of 15-25 against at most about 4 from absorption (ApJ 60:209, 1924); 70 for Na-K at 70 degrees (Elster and Geitel, ApJ 60:231); photoemission peaking 5-10 times above room temperature near 650 K (ApJ 64:128, 1926); commercial potassium cells (BSTJ 5:320, 1926); Ives-Fry 1933, where he expected "only positional and qualitative agreement" and called the result "very satisfactory", against the 1951 lecture's "confirmed to fine detail". The memoirists' "hunch was right" is more guarded in Ives's own 1933 wording.
• Neo-Lorentzian line: Builder (1957-58), Ruderfer's 1960 Moessbauer ether-drift proposal and Preikschat's rotor experiment (Birmingham PhD thesis, October 1968; "Ives" occurs 8 times, not 27; limit 2.0 +/- 5.2 cm/s; both theories predict the same null at first order; caveats: north-direction runs fluctuated more than the instrumental channel, and the limit is on the earth's orbital motion relative to an assumed frame), Prokhovnik, Dingle, Janossy. Dirac, "Is there an aether?" (Nature 168:906, 1951). The US Naval Research Laboratory bibliography (Benton, 1959) listing Ives-Stilwell as verification of clock retardation and Ives 1937 as resolving the clock paradox "through motion relative to the ether".
• Later use: Nieper's free-energy book, "Galileo Was Wrong", the Chinese Academy of Sciences "Relativity Criticism Group"; Butterfield's letters at the LoC on anti-Einstein feeling and Einstein "castigated as left-wing radicals". This is the evidence for Robertson's 1956 worry that the work would be "seized upon": it shows that it was used, not that Ives intended it.
• Lalli 2014 (verified pp.359-365): the 5 Oct 1951 sentence is quoted accurately; "progressive marginalization" and "hidden network of allies" come from the Lalli 2013 abstract. Bridgman's phrase in the 1959 memo is "real but indeterminable" absolute simultaneity, "a complete negation of the operational attitude".
• Ives's fairness to Einstein (1947 note), his own caveats in 1938, 1939, 1947 and 1948, and the generous record of his concessions (relativity principle holds; distant clocks set by transport).
• Ives's other relativity experiments not in hand: Rev. Sci. Instr. 9:283; the 1951 clock-paradox letter (Nature 168:246).

Corrections to earlier documents (for the orchestrator)

1. primer.md section 4.4 and its evidence table (gravitation): replace "one matched number does not test a force law with that much freedom" and "Light bending, Shapiro delay, redshift, pulsars: not addressed in anything read". Light bending and red shift are in the 1939 paper; Shapiro delay follows from the same light-speed field; pulsars and waves are not addressed. Explain the Schwarzschild metric from scratch as "how rulers and clocks behave near a mass". Evidence: deep reads 1939 and 1948; Mercury 42.98 [calc].
2. primer.md sections 4.3, 4.8, source-gap and further reading: the 1941, 1945, 1947, 1948, 1949 and 1951 papers and the Rumford lecture are in hand; "rod-clock-quotient" is defined in 1948, not 1953; Ives-Fry 1933 is no longer "known only from the bibliography title"; add the ripple-tank analogues as the physical picture of a Lorentz ether; add the owner's section.
3. primer.md: the earlier arbiter's m0(1 + v^2/(4c^2)) low-speed factor was the Mathematician's arithmetic error; correct form is m0(1 + GM/(rc^2) + v^2/(2c^2)). Also, the "7.16 of 43 arcsec ... extra terms must supply the rest" argument is withdrawn.
4. claims.md claim 4: light bending 1.7516 should be 1.7512 arcsec [calc]; add the kappa caveat, the Nordstrom point against Ruderfer's "necessarily equivalent", and the Birkhoff sound-speed note (medium confidence). In claim 1 the count "six above, three at or below" disagrees by one between the Physicist and Mathematician; check Table I of JOSA 31:369 before quoting. In claim 7b record that Ives's formula is correct for the force needed to hold constant velocity (Physicist, reproduced by the arbiter); "wrong" is a definition difference. In claim 7d add that the ripple result is the wave equation's own covariance, not a test of light.
5. facts.json: add the 26 missing entries (QE round 2 section 1); correct the Ives-Stilwell 1938 digest errors; fix the Preikschat entry ("Ives 1948 itself not read" is outdated).
6. README in the source folder: "34" Ives papers; the folder and manifest hold 31 (project file, read only; flagged, not a review defect).
7. Digests: the five QE errors plus the JOSA 29:294 "c_R = c y^(1/2)" sentence.
8. followups.json: record the owner's question (why Ives matters to Moray; a Lorentz ether adds no forces) with tools/followups.py add Ives "...", as QE recommends, so the answer above is on the site.
9. Earlier arbiter text now superseded (this file replaces it): claim 4 "only the 1948 abstract read", "presumably chosen to reach 43", "Never properly tested (low confidence)"; the "1945, 1949, 1951 unread" notes under claim 3; the source-gap list of "not read" papers.

Summary verdict table

Recheck

(To be appended by the QE recheck, Sonnet.)