Science Contrarian review: Frank W. Very, The Luminiferous Ether (1919)

Reviewer: Science Contrarian (Step 2). Judges by experimental evidence, not consensus. Builds the strongest honest case for each claim, and applies the same rigour to Very. Tags: [read] = from the project document or a document I opened; [read: page image p.N] = rendered page; [web] = URL given; [memory] = recalled, with confidence. Page numbers: PDF page = printed page + 4.

Source note: one document, 64-page Google Books scan, 63 pages OCR. No source gap for the central text. Fessenden's papers are not in the project folder. I read his 1900 and 1914 notes from the Internet Archive text the Investigator located (fetched to a scratch directory, not saved into the project).

0. Corrections to Step 1

I checked Step 1 against the page images (PDF pp.37-38 rendered by me) and the extracted text for pp.42-52.

1. Confirmed. The deep read's page-image values are right [read: page image p.37, p.38]: Fessenden "about 3/4 x 10^-13 cms. in diameter", "2 x 10^32 dynes", Very's "more nearly 10^33 dynes", "10^73", "10^-25 sq. cm.", "1.660 x 10^22 revolutions per second". The digest lines for pp.37-38 (10^10, 1.660e21, "3 x 10^-13") are OCR errors and should not be used.
2. Confirmed by my own arithmetic (NumPy, CGS-Gaussian units: cm, g, s, esu, dyn): - c/(2 pi r) with c = 2.999e10 cm/s and r = 1.9e-13 cm gives 2.512e22 rev/s, not 1.660e22 (printed figure low by a factor of 1.513; it would need r = 2.875e-13 cm). - Surface electrostatic stress P = E^2/(8 pi), E = e/a^2: 2.0e32 dyn/cm^2 only if Fessenden's diameter 7.5e-14 cm is used as the radius (e = 4e-10 esu); 3.2e33 with the true half-diameter 3.75e-14 cm; 7.0e30 with Millikan's 1.9e-13 cm and e = 4.774e-10. - Thomson's radius 2e^2/(3 m c^2) = 1.85e-13 cm, matching Millikan's 1.9e-13 as quoted [read].
3. Investigator's open point settled in part. The Investigator could not tell whether the "diameter" and "dynes" wording was Fessenden's or Very's. I read the 1900 Science text [web] https://archive.org/stream/jstor-1628439/1628439_djvu.txt: "the corpuscle is about 3/4 x 10^-13 cms. in diameter ... we find for the electrostatic tension and pressure at the surface of the corpuscle, about 2.10^32 dynes" (the exponent is OCR-garbled as "10 82"; the page image was not checked). So the wording is Fessenden's, and Very quoted it faithfully. Whether Fessenden's 2e32 is a diameter-as-radius slip is my reconstruction, not something either text states.
4. Fessenden's own volume elasticity is unreadable. The same page says "we can calculate the volume elasticity, and we find it to be about 10" (exponent lost in OCR). Step 1 and the Investigator both imply Very's 10^73 is derived from Fessenden. The text does not show it: Very's (pressure)^(3/2)/area formula is his own, and I could not confirm what Fessenden obtained.
5. Dimensional check, stated precisely. (dyn/cm^2)^(3/2) / cm^2 = dyn^(3/2) cm^-5. A bulk modulus (volume elasticity) is dyn/cm^2. The expression cannot be a modulus, so "10^73" has no unit and the formula cannot be repaired by choosing different numbers.
6. facts.json "mainstream_conflict" over-states one point. It says bound-ether shells conflict with stellar aberration "under mainstream treatment". Very does not use a wave ether for aberration at all. He uses a corpuscular theory (Stokes 1845) and says aberration is "entirely independent of the bound ether" [read: p.52]. The conflict is with emission-theory tests (claim 4 below), not with wave-ether aberration. Also, "no intergalactic repulsion is established" needs a caveat: the cosmic expansion and its observed acceleration are, loosely, a repulsive effect at intergalactic scale [memory, high]. Very's repulsion is something else (local, from non-collision), but a reader should not think the word is meaningless in modern cosmology.
7. The "1,000 times atomic frequency" check (deep read 3.2). Correct as far as it goes, but one possible rescue was missed. 2.5e22/1000 = 2.5e19 Hz is about 100 keV, which is the K X-ray line region of the heaviest elements (uranium K-alpha is near that) [memory, medium]. Moseley's X-ray frequencies were known to Very, who cites Moseley numbers. So the "1,000" could mean heavy-atom X-ray lines, but the text says "the lighter atoms", where it fails by 10^4 to 10^7. I cannot tell which he meant.
8. Investigator's Fessenden 1914 lead needs a qualifier. I read the note [web] https://archive.org/stream/jstor-1640236/1640236_djvu.txt: "the work done in upsetting the gyroscope varies directly as the angular velocity of rotation of the body ... directly proportional to the frequency." That supplies an E proportional to nu route. But Fessenden's "body" rotates at the radiated frequency. In Very's model the electron's own surface rotation is fixed at c/(2 pi r), so Very's overturn energy would be one constant, and he indeed writes of "the fixed value of the energy-constant" [read: p.40 per deep read]. A fixed energy per overturn cannot give E = h nu for all nu. So the Fessenden bridge is real but is not Very's.
9. Unit slip in Step 1 and in Very. Planck's h has units of action (erg s), not energy. Very writes "Planck's least quantity of energy". The energy quantum is h nu. Step 1 carried the phrase unchallenged.

Reading of the case, once

Very was a working astronomer (Allegheny Observatory, bolometric lunar and planetary temperatures; MIT 1873) [web] https://en.wikipedia.org/wiki/Frank_Washington_Very. The pamphlet is a mix of three things: (a) concrete, checkable physical statements (numbers, experiments cited); (b) a mechanical-model speculation with no predictions; (c) a Swedenborgian and devotional frame ("the finger of God", telepathy, p.45 and p.55 [read]). I judge (a) on experiment, (b) on whether it ever risks a measurement, and set (c) aside as not a physical claim, but I do not remove it from the record: it is Very's stated reason for the aura.

Claim 1: Light is discrete electron-sized "ether-particles" at c, each with one Planck quantum; a separate "aura" fills space

a. Mathematics. No equation for the particles. The text states a size (electron-sized, about 1e-13 cm), a speed c, and a "period of vibration" that "appears to depend on some function of the radius of the electronic orbit" [read: p.39]. A light wave of 5e-5 cm wavelength carried by objects 1e-13 cm across is not connected to anything numerical. Internally, "each carries one Planck quantum" and "light of different frequencies" conflict unless quantum = h nu (see claim 2).

b. Experiments. Supporting: the particulate side of light is real. Einstein's 1905 light quantum and the photoelectric effect; Compton scattering (1923) [memory, high]. Very was early in 1918-19 to say discrete quanta make a continuous luminiferous medium unnecessary, and the 1905 relativity route said the same by another road. Against: interference and diffraction need wave-like phase. Laue/Bragg X-ray crystal diffraction (1912) was known before the pamphlet; single-photon double-slit interference is later [memory, high]. A particle with a fixed electron-sized body, no phase and no wavelength cannot reproduce fringes spaced by wavelength. The ballistic (source-dependent) version of corpuscles is excluded by de Sitter's 1913 argument from binary stars [memory, medium; web check not done]; Very's version does not say whether the particle speed is relative to the source, the Earth or the aura.

c. Tested? Never tested as a specific model, because it makes no prediction that differs from existing theory. Nobody tested it and nobody rejected it by name: I found no contemporary review (Investigator, 6 Google queries and 5 Internet Archive queries; silence there is not evidence of how it was received). Its parts were tested against other people's results (interference, Compton), not against it.

d. If true. Wave optics and quantum electrodynamics would need replacing by a mechanical particle model that still yields fringes, polarisation and Maxwell's equations. Very gives none of that.

Verdict: split. - Light carries quantised energy and acts in discrete units: Established (Einstein, Compton), but by others' work, not Very's. - Light as electron-sized mechanical ether-particles, with no wave aspect: Tested, not supported (by diffraction and interference results that he knew of; no test aimed at him). - The aura as a separate universal medium: Not testable as stated (no property is given that any instrument could measure). [Context: space is not empty, with gas, dust and the 2.7 K background radiation [memory, high], but none of those is Very's aura.]

Claim 2: The Planck quantum is the energy to gyroscopically overturn an electron and reverse its sign, shedding the sheath that becomes the ether-particle

a. Mathematics. No equation. "Planck's least quantum of energy represents the resistance which an electron offers to a gyroscopic uptilting" [read: p.41, per deep read]. Units: h is erg s, so "least quantity of energy" is a category slip. With a fixed overturn energy, E = h nu for every nu cannot hold (note 8 above). My arithmetic: h x 1.66e22 s^-1 = 1.1e-4 erg = 69 MeV, about 130 electron rest energies; with the corrected 2.51e22, 104 MeV. If the overturn energy were h x (rotation rate) it would be a gamma-ray photon, not the visible quanta he wants it to explain.

b. Experiments. Photon energies span radio to gamma, 15 orders of magnitude, all with the same electron. A single fixed overturn cannot produce them. Atomic emission lines come from transitions between atom states (Bohr, which Very cites), not from reversing electron sign. Pair production and annihilation do reverse sign, but at energies of 1.02 MeV and up, a different phenomenon [memory, high].

c. Tested? Never properly tested. Fessenden's gyroscope idea (1914) was a published note; I found no test. No one engaged it in the record I found.

d. If true. Emission would need an electron-sign flip every time light is emitted, which means net charge exchange between atoms and radiation fields. Observed charge conservation in atomic emission would have to be re-explained.

Verdict: Not testable as stated (no number connects the overturn to any frequency); with the Fessenden bridge added it becomes a different, still untested proposal ("Never properly tested"). As written by Very, the fixed-energy statement and E = h nu are Internally inconsistent.

Claim 3: The electron is a rotating vortical sphere of aura, surface at c, 1.660e22 rev/s, pressure 1e32-1e33 "dynes", volume elasticity 1e73

a. Mathematics. This is the one claim with numbers, so it is the most checkable. Results of my checks: - Rotation: stated 1.660e22; correct c/(2 pi r) = 2.512e22 (error of 1.513, source unknown). Worked step: 2.999e10 / (2 pi x 1.9e-13) = 2.999e10 / 1.1938e-12 = 2.512e22. - Very says the surface speed is c at every point so "the angular velocity increases from the equator to the poles". With speed c at latitude lambda the angular velocity is c/(r cos lambda), infinite at the poles. He does not address it [read: p.38]. - Pressure: the revised Millikan radius (1.9e-13) gives a lower stress, 7e30 dyn/cm^2, not "more nearly 10^33". The rise only occurs with a smaller radius. The 10^73 calculation itself uses 10^32 not 10^33. - Unit: dyne is force; stress is dyn/cm^2. Fessenden wrote "dynes" in 1900 [web], Very copied it. - 10^73 is unit-free (dyn^(3/2) cm^-5) and not a modulus. - Additional check, mine: a solid sphere of electron mass 9.109e-28 g and radius 1.9e-13 cm with surface speed c has spin angular momentum 0.4 m r c = 2.1e-30 erg s. The measured electron spin is hbar/2 = 5.3e-28 erg s [memory, high]. So a classical-radius electron spun at c carries about 1/250 of the measured spin. To match, the radius must be about 250 times larger (about 5e-11 cm), the size ruled out by scattering. The rotational energy (0.5 I w^2) is 1.6e-7 erg, about 20% of m c^2 (8.2e-7 erg), which breaks the Thomson mass bookkeeping Very relies on. This does not refute his claim as he wrote it (spin was not known in 1918), but it shows the rotating-at-c sphere cannot also be the electron's spin.

b. Experiments. High-energy scattering shows the electron has no size down to about 1e-18 m (i.e. 1e-16 cm), well below the 1.9e-13 cm classical radius [memory, high]. The classical radius is a scale where electromagnetic and rest energy match, not a measured body size. Experimentally unsupported: any inner structure of the electron, and any rotating surface.

c. Tested? No one has tested the rotation claim; I found no sign anyone computed it before this review. The numbers have never been through a referee that I can find.

d. If true. Electrons would have measurable size and structure; the standard model's point electron (checked by g-2 to parts per trillion [memory, high]) would be wrong.

Verdict: split. The numeric chain (1.660e22 rev/s; 10^73) is Internally inconsistent (arithmetic error, unit error, wrong direction of revision). The underlying idea, an extended electron held together by a medium's elasticity, is Never properly tested; extended-electron models (Abraham, Lorentz) were a serious line before 1920 and Very is working in it. At the experimental level a size of 1.9e-13 cm is Tested, not supported (scattering limits).

Claim 4: Michelson-Morley null, Rayleigh and Brace nulls and aberration explained by a bound-ether shell carried by each atom, plus corpuscular light

a. Mathematics. No calculation given. "The result is at least in harmony with the supposition that a shell of bound ether is attached to ... every atom and thus ... the earth as a whole" [read: p.36]. This is the old Stokes complete-drag idea limited to a shell. Very says the bound ether is not what carries light: it retards light by induced reaction [read: p.52-53]. So light is carried by a "free" ether-particle stream that does not feel the shell's motion. Then why should an interferometer arm see a null? The text does not derive it; if the free particles are not dragged, an Earth-bound shell does not remove an ether wind from the light's propagation. The statement is a verbal promise, not a calculation.

b. Experiments. Supporting: Michelson-Morley (1887), Rayleigh (1902) and Brace (1904) really gave nulls at the sensitivity then achieved [memory, high; a web check for Rayleigh and Brace failed: the page fetched did not carry them]. A drag-along interpretation does fit those at sea-level laboratories. Strongest honest case: they cannot rule out partial dragging, and Dayton Miller reported a small positive drift at altitude on Mt Wilson (0.12 fringe, about 9 km/s; Shankland et al. 1955 attributed it to statistics and uncontrolled temperature effects) [web] https://en.wikipedia.org/wiki/Dayton_Miller. That was a serious experiment, careful and large (5.2 million readings), and one with a published reanalysis, which is the sort of rejection I look for; the reanalysis is plausible but it is the critics' own reading of the data, and Miller's altitude pattern is the only respectable positive on record. Against: later tests tighten the null to far below any shell effect, including Kennedy-Thorndike (1932) and Ives-Stilwell (1938), both after Very [memory, high]. Sagnac (1913) found a light-path shift on a rotating platform with respect to the lab, as a non-dragging ether predicts [memory, high], and the Michelson-Gale experiment (1925) detects Earth rotation optically [memory, high], which is not what a fully Earth-carried medium would do. Aberration by a corpuscle: Stokes (1845) could explain the aberration angle if the source medium moves; the ballistic form is excluded by de Sitter (1913) [memory, medium; not web-checked].

c. Tested? The nulls were tested by others, carefully. The bound-shell mechanism was never tested as distinct from Lorentz-FitzGerald or relativity: it makes no distinct prediction I can find. So not rejected, simply not engaged.

d. If true. Special relativity's role would be reduced to a coincidence of nulls; the Sagnac and Michelson-Gale rotation results would have to be accounted for by the aura.

Verdict: Never properly tested (the mechanism has no prediction that distinguishes it); the null experiments themselves are Established data. The corpuscular-aberration piece in its ballistic form is Tested, not supported [de Sitter, memory medium].

Claim 5: One kind of electron, no positive electron; all mass electrical (H = 1,844 electrons; uranium nucleus about 439,000; 92 satellites)

a. Mathematics. 1,844 is the hydrogen-atom to electron mass ratio, so the count is just that ratio, and "439,000" is 238.x x 1,844 [read: digest p.28-33; not rechecked in detail]. Charge bookkeeping: if every particle were negative and the atom is neutral with 1,844 electrons, a net positive charge of 1,843 electrons is needed. Very's answer (nucleus of many electrons with opposite "aspect") is not worked out.

b. Experiments. Positron: discovered by Anderson in 1932 (cosmic rays) [memory, high]; pair production and annihilation are routine; the proton has charge +e and mass 1,836 electron masses, a separate particle; neutron discovered 1932 [memory, high]. Rutherford (1911) gave the nucleus as a small positive charge centre, and the 1919 proton knock-out [memory, high]. X-ray scattering by Barkla (1911) gave an electron count per atom of about half the atomic weight, not the full weight [memory, medium], which already was in tension with H = 1,844 electrons. Fessenden's version has the same claim (positive nucleus made of negative electrons) [web, 1914 note], so this is a shared 1900-14 view, not Very's alone.

c. Tested? Tested by others across 1911-1932 and not supported. Not tested by Very.

d. If true. Radioactivity and nuclear structure, and all of positron physics, would need rewriting.

Strongest honest case: the idea that opposite signs are two orientations of one object has a faint echo in Wheeler-Feynman's idea of the positron as an electron moving backward in time [memory, high]. That is a resonance in spirit only; Very says handedness of rotation, and there are still particles of distinct mass (protons) with which the model cannot cope.

Verdict: Tested, not supported (strong form: only negative electrons, all mass from electron counts). The weaker point that charge sign is a reversible property of a creation or annihilation event is Established in pair production, but for a different reason.

Claim 6: Gravity is synchronous pulsation of electrons in the aura, limited to a "galactic cell"; galaxies repel at intergalactic distances

a. Mathematics. Very's argument for repulsion is an estimate [read: p.46]. Scale a galaxy to 1 mm; neighbours "only a few centimeters apart"; with velocities "upwards of 1,000 km per sec ... frequent collisions are inevitable unless there are mutual repulsions". My check of the collision claim: take galaxy diameter d = 30 kpc, spacing s = 1 Mpc (about 33 d, which matches his "few centimeters" for millimetre dots), v = 500 km/s = 0.51 kpc/Myr, time t = 10 Gyr. Collision probability per galaxy ≈ pi d^2 v t / s^3 = 3.14 x 900 x 0.51 x 1e4 / 1e9 ≈ 0.014. About 1% per galaxy in 10 Gyr in a typical field. His "frequent collisions inevitable" is a factor of about 100 too high for the field. (Inputs: mine, from [memory, medium]; the Hubble time was unknown in 1918 but the geometry is his.) The other numeric piece: his Andromeda distance of 1,600-8,000 light-years is too small by a factor of about 300-1,500 against 2.5 million light-years [web] https://en.wikipedia.org/wiki/Andromeda_Galaxy (revised in 1920s, after Hubble [memory, high]); he is right that it is of "truly galactic dimensions" [read: p.45], which was a live dispute until 1923-24. - Bjerknes pulsating spheres: in-phase spheres attract, out-of-phase repel. Hence his need for exact synchrony [read: p.48]. That is correct as a statement of Bjerknes's hydrodynamic analogy [memory, medium] and Very admits no means is known to produce it.

b. Experiments. Galaxies attract: Andromeda is blue-shifted at -301 km/s (measured by Slipher in 1912, six years before the pamphlet) [web] https://en.wikipedia.org/wiki/Andromeda_Galaxy; it is approaching the Milky Way, which is attraction (and a collision with the Milky Way is considered possible [same source]). Colliding and merging galaxy pairs are observed (e.g. the Antennae) [memory, high], and clusters of galaxies are bound. Gravitational waves from mergers travel at c to within about 1e-15 [memory, high], not "enormous velocity" through a nearly incompressible medium. The only intergalactic "repulsion" now accepted is the accelerating expansion [memory, high]; it acts everywhere as a cosmological effect, not as a push between neighbours, and not through aura cells.

c. Tested? Very's premise (no observed collisions) was not a controlled test; it was an argument from absence of observed collisions in photographs of 1918, which is weak for objects whose collisions last 1e8 years. The pulsation model has never been tested: no prediction that differs from Newton or general relativity. He also cites his own paper on a "possible limit to gravitation" [read: p.44 footnote]; I did not find it. Very's later 1920 Science paper on Nipher's gravitational experiment is paywalled and unread [Investigator]. Nipher's claim of electrical effects on gravity is a genuine anomaly claim of the time, unreplicated as far as I know [memory, low]; whatever it showed, it is not Very's aura.

d. If true. Gravity would not be universal; galaxy dynamics, cluster mass estimates and lensing would be wrong at the large scale.

Verdict: split. Intergalactic repulsion and galactic-cell cutoff: Tested, not supported (Andromeda approach, mergers, bound clusters). The collision argument for it: Refuted by calculation (about 1% per 10 Gyr, shown above; assumptions stated). The pulsation mechanism for gravity: Never properly tested (no distinctive prediction).

Claim 7: Matter is converted to radiation and re-originates from the aura; entropy increase is not one-sided over the universe

a. Mathematics. None. Argument by Lodge's dilemma: if the universe is infinitely old, the second law demands it be dead already [read: p.42-43]. This is a logical premise ("age infinite") not an equation.

b. Experiments. Matter does turn into radiation (nuclear reactions, mass defect, the sun) and radiation can make matter in pairs above 1.02 MeV [memory, high]. These are Established. Matter condensing out of a "universal aura" in regions of "building-up" is not observed: no region with new atoms forming from field without radiation input has been found, though the steady-state cosmology (1948) later proposed continuous creation and lost to the microwave background evidence [memory, high]. Very's "non-selective absorption of light by the aura" [read: p.43 footnote] happens in a different form: interstellar dust causes general extinction (Trumpler 1930) [memory, medium], so a space-filling absorber exists, though it is dust, not aura. Interstellar medium in general is the part of his "universal medium" that survived.

c. Tested? His own 1909 paper [web] https://archive.org/stream/jstor-1635576/1635576_djvu.txt used Kaufmann, Lebedew and Nichols & Hull as supports (the mass-velocity and light-pressure measurements), which are real measurements, but they do not bear on re-origination. Not tested directly.

d. If true. The second law would fail for the universe as a whole, with no local mechanism given.

Verdict: Historically superseded but instructive. The dilemma he answers (infinite age) is answered differently: a finite-age, expanding universe (Hubble 1929, then the microwave background) [memory, high]. The re-origination piece is Never properly tested and has no mechanism.

Did the mainstream reject it, and how?

• No rejection is documented. I found no review, reply or refutation. The Investigator found none either, and its search was thin (see Investigator's open leads: HathiTrust, Google Books, the AAS 1918 abstract in Popular Astronomy, The New Philosophy). So "dismissed without testing" cannot be asserted: there is no record of dismissal. The accurate label is not engaged in the record located so far.
• Very was not shut out: he presented to the American Astronomical Society in August 1918, had papers in Science up to 1920 (the Nipher paper), and the pamphlet appears as Westwood Occasional Paper No. 2 [Investigator; read]. The pamphlet was published in the New-Church Review and by his own observatory's series, so the route avoided physics journals. Venue of publication is a competing explanation for silence.
• Record of the person: Langley's 1890 paper used Very's Moon observations but "for unknown reasons omitted his name" [web] https://en.wikipedia.org/wiki/Frank_Washington_Very. That concerns credit in another work, not the ether, but it is the one documented instance of exclusion.
• Competing explanations for the silence after 1919: (i) no testable predictions to answer; (ii) relativity and quantum theory took over in the 1920s; (iii) the Swedenborgian frame deterred physicists; (iv) loss of patron (Percival Lowell died in 1916 [Investigator, Dollar Bank snippet]) and Very's death in 1927. Records that would decide: Ogden's 1928 memoriam (Popular Astronomy 36:391, ADS scan not rendered), Lowell Observatory archives, the Allegheny Observatory file "Very, Frank (Westwood Astrophysical Observatory)" at Pittsburgh, 1927 obituaries. Silence in the sources searched is not contradiction.

Strongest honest case for Very

1. Direction of travel: discrete light quanta make a continuous luminiferous ether unnecessary. He says this in 1918, 13 years after Einstein's 1905 quantum paper [memory, high].
2. He asks whether an electron is extended and what holds it together. Extended-electron models were serious physics in 1900-1915.
3. He argued independently that Andromeda is "of truly galactic dimensions" [read: p.45] before it was settled, though his distances are far too small.
4. A universal interstellar medium that absorbs light has a real counterpart (interstellar dust and gas).
5. Very is careful with evidence in one documented case: he exposed the 1910 Norwood "meteorite" as a fraud after first reporting the fall (Investigator, jstor-1636137), and he says plainly he "could not verify" the Franklin quotation [read: p.42]. This is relevant character evidence for how he handles data, not a physics argument.
6. The Fessenden gyroscope-quanta route (E proportional to frequency) is a coherent idea about a vortex origin of quanta, in print in 1913-14 [web].

Strongest case against, applied to Very with the same rigour

• No claim except the electron numbers is quantitative, and the electron numbers fail on arithmetic and units (above).
• Every claim has an escape: the "aura" accounts for anything and has no measurable property.
• It argues from absence of observed galaxy collisions where the real figure is about 1% per 10 Gyr.
• It admits the key gravity mechanism (exact synchrony) has no known cause.
• It mixes scientific and religious claims ("the finger of God") without separating what is tested from what is faith. Both are in the record.

Investigator leads, kept in the record

These stay in the review whatever the verdicts; I weigh them briefly. - Very 1909 Science [web]: his first statement of claims 3, 4, 7, with an explicit mass equation dm/m0 = v/(V - v) for radiation-added mass. I did not check it. - Very 1913 New-Church Review "A Scientific Estimate of Swedenborg's Principia" (attribution via AbeBooks only), Oct 1918 NCR printing, a second scan at https://archive.org/details/luminiferousethe00veryrich: unread. The second scan would settle whether "1.660" is a misprint or Very's. - Fessenden 1900 "Determination of the Nature and Velocity of Gravitation": includes a stretched-glass vs. Kerr comparison (30e6 vs 25e6 dynes) which Fessenden himself calls "probably accidental"; it is not evidence for claim 4. His proposed test (gases weigh slightly more than solids) is a testable prediction; equivalence-principle torsion balances show composition independence to far smaller fractions [memory, high], so it was tested, not supported, but that is Fessenden's claim, not Very's. - Fessenden 1913 "Transformation of Gravitational Waves into Ether Vortices": parallel route to quanta; not tested. - Fessenden's account that his 1890-91 theory was withheld for about a year and printed with an editorial calling it "probably wrong": a single-source account of how a related idea was received, stated in Fessenden's own words; it concerns Fessenden. - Very 1920 "On Nipher's Gravitational Experiment": unread (paywall). - Reception, Westwood funding and archives: not reached; open.

Summary table

Web checks done: 3 (Andromeda; Miller; Rayleigh/Brace, the last failed). Medium-confidence [memory] items not web-checked: de Sitter 1913, Barkla 1911, Trumpler 1930, Bjerknes phase rule, galaxy diameter and spacing inputs for the collision estimate.