Brush-Charles: Science Contrarian review

Reviewer role: Science Contrarian (judges by experimental evidence; separates "tested and failed", "never properly tested" and "dismissed without testing"; applies the same rigour to Brush as to his critics).

Source gap (repeated from Step 1, and what I added)

Step 1's note stands. Most of the folder has no text layer. I re-read the following myself: - brush-rexresearch.pdf (Nature 1911, Lodge letter, 1929 APS paper): text layer, read in full. [read] - Proceedings_of_the_American_Philosophica.pdf, Brush's 1922 pendulum paper, PDF pp.184-203: rendered at 150 dpi and OCR'd. [read: page images] - Spontaneous-gen-of-heat-in-hardened-steel-Physical_review.pdf, the 1917 paper, PDF pp.241-249: rendered and OCR'd. [read: page images]

Not read by me: the 1910 reprint, the 1914 Discussion, the 1928 address, the Miller letters, the newspaper items. For these I relied on Step 1's facts.json, and I mark the dependence where it matters.

The raw drop-test tables, the calorimeter data (no wattage, no ice-melt numbers appear anywhere in the readable text), the 1923/1924/1926/1927/1928 papers, and any Bureau of Standards report are NOT in the folder. Every statement below about the calorimetry, and about the Bureau of Standards confirmation, rests on Brush's own prose summaries and nothing else.

Web-verification note: my web budget ran out partway through. Four fact-checks succeeded via Wikipedia pages and are tagged [web]. Two [memory] items I could not verify are flagged "check failed".

Corrections to Step 1

1. The "1 part in 32,000" pendulum figure. facts.json ties it to Table I (7.3 h per beat gain) and says "after subtracting 22 % for centre-of-gyration effects". That is not right. [read: page images, 1922 paper pp.172-178] - Table I (7.3 h) came from bobs of unequal weight (zinc about 7 g light). Brush gives that run's effect as "one part in 35,000" and says it is understated. - After he equalised the bob masses, a beat took 5 h 11 min. The 22 % centre-of-gyration subtraction belongs to Table II (equal-mass bobs, long pendulums). - Table II's corrected effects are 31.3 and 31.7 x 10^-6 (about 1 in 32,000) for the long pendulums. For the short pendulums the corrected values are 19.4 and 25.9 x 10^-6, which Brush says are suspect because of bob-exchange errors he had not repeated. - So the 1/32,000 figure comes from equalised long-pendulum runs. The short-pendulum runs do not reproduce it (they are 20 to 40 % lower) and he dismisses them as probable error. Step 1 did not mention this.
2. Step 1 correctly gives the drop-test scaling (0.099 mm over 110 cm, scaled by 1/0.403). Check: 0.099/1100 = 1 in 11,100, and 11,100 x 0.403 = 4,478, so his "1 in 4450" is rounded by about 0.6 %. No error here. The 0.0125 mm to 1/400,000 s conversion is also fine: at 4.6 to 4.9 m/s the time offset is 2.6 x 10^-6 s. (Step 1 says "about 2.5e-6", which is correct. The "1/400,000 s" is Brush's own figure.)
3. Step 1 does not record that Brush compared bismuth with bismuth (old versus new casting, different crystal grain) and found an effect about 2.5 % as large as zinc-bismuth, and zinc against zinc (electrolytic against commercial). It matters for claim 4 and is covered below.
4. claims.md claim 3 says the correction "possibly explain[s] Mercury's perihelion advance". The source says "part of". Brush gives no number and no equation for x anywhere in what I could read. Minor.
5. In facts.json the pendulum entry says air vanes "changed the effect by roughly 50 %". Correct, but Brush's reading of it is the reverse of the cautious one: he says doubling the drag changes the effect by at most half, so drag is "negligible". I treat it as evidence that the apparatus is sensitive to uncontrolled air effects (see claim 4).

Claim 1. The energy of a falling body comes from the aether, not from "potential energy of position"

a. Mathematics. No equations are given. The argument is verbal, via the "pound of iron lifted to the earth-moon neutral point". Brush's logic: at that point the iron has no net force, so no "potential energy of position", yet work was done lifting it, so the energy must be somewhere other than the iron. [read: Nature 1911] - Standard physics already answers this. The energy lives in the gravitational field of the whole earth-moon-iron system. Potential energy depends on the configuration, not on the body alone. At the neutral point the potential energy is not zero. Only the force is zero. Brush confuses force with potential. - Sir Oliver Lodge says the same in the reply Brush printed: the energy of a raised weight is "in the aether", the medium that drives the weight down, and the neutral-point puzzle is "no more troublesome than a weight hung on a hook". [read: Lodge letter] - Contrarian honesty: Brush's insistence that the energy must be stored in the medium and not in the body is the standard field view in older words. The claim is not absurd. It is mostly a relabelling.

b. Consistent with experiment? The weak form (energy is a property of the system or field, not stored inside the body) is consistent with everything. The strong form (a vast reservoir of aether kinetic energy that is drawn on during a fall and returned when lifted) has no measurable consequence by itself. It becomes testable only through claims 2 and 3. The "aether" itself was abandoned for electromagnetism after the Michelson-Morley null result (1887) [memory, high]. Brush's chart puts "all ether waves" at the speed of light and treats the aether as a medium for transverse waves, so the 1887 result bears on him. He does not discuss it in what I read.

c. Tested? Never tested. It is not directly testable.

d. What would change. Little. Gravitational potential energy would simply be re-described as an aether-reservoir exchange with the same numbers.

Verdict: split. - Weak form (energy is not in the body but in the system or field): Established, but it is the textbook view, not a Brush departure. - Strong form (a wave-kinetic aether reservoir exchanging energy with falling bodies): Not testable as stated, until tied to claim 2 or 3.

Claim 2. Gravity is a push from isotropic penetrating aether waves; each atom casts an inverse-square "energy shadow"

a. Mathematics. No derivation. Brush asserts that overlapping shadows give a force that is inverse-square in distance and proportional to the sum of the atoms' units. He himself says "the manner in which this molecular activity maintains, in effect, the supposed spherical shadow, requires explanation, which I shall attempt in a future paper". [read: Nature 1911] I found no such explanation in what I read of the later papers (the 1914 abstract and 1929 summary restate the same picture). So the central quantitative step is missing.

Internal tension I can show from his text. The waves must (i) pass "freely through all bodies" without being absorbed, to make force proportional to mass and avoid heating; yet (ii) a shadow is, by definition, a deficit made by absorption, and a wave that passes through unabsorbed exerts no push. He resolves this by letting each atom absorb and restore the same energy ("continual absorption and restitution of the aether's energy, normally equal in amount"). That keeps the energy books balanced, but a force still needs net momentum transfer, and he gives no calculation of how large the absorption must be to produce an acceleration of 9.8 m/s^2 on 1 kg. In the later papers he says the same wave class heats some silicates by "conversion into heat". If a net conversion to heat can happen in some materials, then the "normally equal" absorption/restitution is not universal, and he does not say what sets the ratio.

b. Consistent with experiment? The generic objections to Le Sage shadow theories are, in summary [web: https://en.wikipedia.org/wiki/Le_Sage%27s_theory_of_gravitation]: - Drag: a moving body should meet a head-on excess of the flux. The drag-to-gravity ratio scales as (body speed)/(flux speed), so the flux speed must be far above c for the planets not to spiral in. - Heating: Maxwell and Poincaré showed that absorbing enough momentum to give gravity heats matter far beyond anything observed. Poincaré's figure for the earth is quoted there as about 10^26 degrees per second. - Screening and mass proportionality: shadowing implies the gravity of two bodies together is less than the sum of the parts, and the force cannot stay proportional to mass. - Brush's reply to the heating objection is the "waves too slow to excite molecules" argument in the 1911 paper (extremely long wavelength), later reversed to "extremely high frequency, above cosmic rays" in 1929. The two ends of the spectrum are opposite answers to the same objection, and the 1929 chart leaves the position as "somewhere in this region". Lodge's 1911 reply raises the screening problem and says "external agent" theories are doomed. [read: Lodge letter] - Brush's 1911 text also says a falling body's speed tends to the wave speed (c). Under general relativity a falling body can exceed nothing; but in Brush's model the speed limit makes weight depend on velocity, so the force on a falling body is not simply mg. That is a testable velocity dependence. He does not give a number (see claim 3).

c. Tested? Brush looked for resonance heating in the upper atmosphere in 1910-11 and says the results "have thus far been unsatisfactory". [read] That is an unreported null by his own account. Nothing else tests the mechanism directly.

d. What would change. Gravity would be a radiation-pressure effect of an aether bath. This would replace general relativity's geometric account. It would also require an aether rest frame (contradicting special relativity at a basic level) and predict drag.

Verdict: Historically superseded but instructive, with the standard Le Sage objections (drag, heating, screening) unanswered by anything in the folder. In the generic form this is closer to "Refuted by calculation" (Poincaré, Feynman [web, above]); but Brush's version has no equations to run that calculation on, so I cannot show a Brush-specific refutation. Where it makes definite statements (claim 3) it fails, as shown next.

Claim 3. Non-inverse-square force on accelerating bodies (may explain Mercury and lunar irregularities); gravity can travel at c without wrecking the moon's orbit

a. Mathematics. Brush's own numbers (as summarised in Step 1 from the 1914 Discussion; I did not read those pages): moving at c, the earth's centre is displaced about 52 ft by the light delay, and the tangential force on the moon is "about 1/24,000,000 of gravity", retarding the moon's mean motion "a few seconds of arc per century", partly cancelled by his (1 + x) deviation.

I checked these with my own calculation [calc]: - Earth-moon distance 3.84 x 10^8 m; light time 1.28 s. The earth moves about the earth-moon barycentre at about 12.4 m/s, which in 1.28 s is about 15.9 m, i.e. 52 ft. This is Brush's displacement. The tangential force fraction is 15.9 / 3.84 x 10^8 = 4.1 x 10^-8 = 1 in 24,000,000. So his two numbers are internally consistent. But they count only the earth's wobble, not the moon's own velocity (1.02 km/s) relative to the aether in which his waves travel. In a wave-push model the moving moon sees head-on aberration of order v/c = 3.4 x 10^-6, about 80 times bigger. He does not include it. - Even with his smaller figure, the consequence is not "a few seconds of arc per century". Moon's gravitational acceleration from the earth is GM/r^2 = 2.7 x 10^-3 m/s^2. Tangential acceleration T = 4.1 x 10^-8 x 2.7 x 10^-3 = 1.1 x 10^-10 m/s^2. For a near-circular orbit, dn/dt = -3T/a = -8.6 x 10^-19 rad/s^2 (n is the mean motion), and the mean longitude drifts by 1/2 (dn/dt) t^2. For t = 100 yr = 3.16 x 10^9 s this is about 4.3 rad, about 9 x 10^5 arcsec. Observed lunar secular acceleration is of the order of tens of arcseconds per century squared [memory, medium; this is the familiar tidal term; check failed]. So the unexplained residual is at least 10^4 times smaller than this. This is Laplace's original argument (finite gravity speed would need c_grav above about 7 x 10^6 c [memory, high]). - Brush says his (1 + x) term "partly cancels" this. He gives no value of x, no formula, and no sign convention. With no equation, the cancellation cannot be checked. The claim that it cancels to within 10^-5 of the uncancelled value needs a precise mechanism, which is not in the sources.

Mercury: Brush gives no numerical prediction. Observed perihelion anomaly after other planets: 42.98 arcsec/century, matched by general relativity [web: https://en.wikipedia.org/wiki/Tests_of_general_relativity]. His x-term is "zero for circular orbits, alternately positive and negative for eccentric ones". That alternating sign means averaging could give zero net advance over an orbit unless the asymmetry is shown. Not shown.

b. Consistent with experiment? Mercury advance: GR explains it with no free parameter [web]. An unquantified correction cannot be said to be consistent or inconsistent. Lunar mean motion: see above, Brush's own inputs give a large unexplained drift unless the cancellation is nearly perfect.

c. Tested? Not by Brush or anyone else, so far as the folder shows. Never tested, because no definite predictions were made.

d. What would change. Newton's inverse-square law would hold only for non-accelerating bodies, and the speed of gravity would be c without the usual orbital instability. GR's account of Mercury would be a coincidence.

Verdict: split. - As stated: finite-c gravity with a push-wave mechanism and no unquantified cancellation: Refuted by calculation (above: the drift is about 10^5 times larger than what is observed). - The cancelling (1 + x) term: Not testable as stated (no value, no formula).

Claim 4. Weight is not exactly proportional to mass: about 1 part in 32,000 (pendulums), about 1 part in 4450 (barium aluminate, falling containers)

This is Brush's most carefully documented claim and the one where the contrarian case is strongest. I do both halves.

The strongest honest case for Brush

• The pendulum work is careful by 1922 standards. Matched steel rods from one stock; knife-edge support on a rigid plate; a start cradle; masses adjusted to 0.1 g on a 2 kg Becker balance, with centres of gravity set by repeated turn-over; bobs exchanged between pendulums; two sets of bobs (single and double); a second iron bob; same-metal controls. He controlled air drag with vanes, and tested the earth's magnetic field with a soft-iron cylinder, support coupling, and interaction between the pendulums. [read: page images]
• He documented his failures: the short-pendulum runs have reversed orderings and he says so. He warned that unequal bob masses can reverse the sign of the effect. He waited a year "in the hope that if any source of large error had been overlooked ... it would be pointed out by some one", and he says nothing was. [read] That is the conduct of a man who wants to be corrected.
• The ordering zinc, then iron, then bismuth, held. A purely random error would not obviously give an ordering that survives bob exchange.
• Mainstream reaction on record: a 1921 NYT piece in which Webster of Clark University says that if the experiments are accurate they "absolutely disprove the Einstein theory" (via Step 1; not read by me). I have no source showing a neutral replication was attempted, or showing one was refused. So "dismissed without testing" is not documented; "never replicated" is what the folder supports. Brush's 1926 request to Miller (Step 1, from the letters) was not carried out, but Miller's reply (as summarised) is a polite admission that he did not know the apparatus, not a refusal.

The calculation nobody did: air buoyancy and added mass [calc]

Brush's bobs were equal-mass cylinders of different metals: zinc (diameter 72 mm), bismuth (61 mm), equal heights 48 mm, swung in air. Equal mass and different density means different volume. A bob in air has its apparent weight reduced by buoyancy (the weight of the displaced air), and its inertia increased by the "added mass" of air dragged along. For a bob of density rho in air of density rho_a = 1.2 kg/m^3, with added-mass coefficient k (about 0.5 for a sphere, about 1 for a long cylinder moving sideways), the period changes by

dT/T = (1 + k)/2 x (rho_a / rho).

Numbers: zinc 7140 kg/m^3 gives rho_a/rho = 1.68 x 10^-4; iron 7874 gives 1.52 x 10^-4; bismuth 9780 gives 1.23 x 10^-4. Zinc minus bismuth is 4.5 x 10^-5. So for k from 0.5 to 1 the predicted period difference is 3.4 to 4.5 x 10^-5 (about 1 in 29,000 to 1 in 22,000), with the lower-density zinc swinging slower, so bismuth gains. The iron bob (density between the two) falls between them.

Compare with Brush: corrected effect 3.1 x 10^-5 (long pendulums, equal masses, Table II), raw about 4 x 10^-5, same sign (bismuth gains), same ordering (zinc, iron, bismuth). Bismuth against bismuth (same density, different grain) gave 2.5 % of the zinc-bismuth effect, which is what a density-based explanation predicts (near zero) and is hard to square with a crystal-structure-dependent gravity of the same magnitude.

This is a calculation of order of magnitude, not a full reduction: the rod mass, the balance weighing in air (which already partly cancels buoyancy when masses are equalised against brass weights), the vane corrections and the exact added-mass coefficient all matter, and I have not done them. But the match in sign, size and ordering is striking, and Brush's text does not mention buoyancy or air displacement. His treatment of air covers drag only (the vanes), and he rejects a vacuum test: "To this I can confidently answer no", without having done one. [read: p.178] The test that would settle the pendulum result (repeat in vacuum, or with bobs of equal volume and different mass) was proposed to him by "friends" and not carried out.

Falling containers (1/4450 for barium aluminate)

• Arithmetic is right (see Corrections). 0.099 mm of tip lag at 110 cm corresponds to a fall-time difference of 2.1 x 10^-5 s, and the quoted effect is 9 x 10^-5 of the distance.
• Brush's own account of the scatter is a warning. The slow container was "unevenly slow in the six photographs, and even very slightly fast in the first"; he attributes variation to lateral air currents in the room, which he says acted unequally "when one shielded the other", and to friction in the guide tubes "during the first millimeter". [read] One in six photographs of the headline plate went the wrong way. The mean is the sum of five positive values minus the one negative, divided by six; no standard error is given. With a fall of 4.6 m/s an air-drag mismatch of about 0.1 % of the drag is enough, if the drag is a few percent of the weight; this is my rough illustration using assumed container dimensions (the container size is not given in the text I have), so I tag it [calc, assumption: not from the source].
• Containers are loaded with loose powder or damp compounds against lead sawdust. Brush records that his barium aluminate loses "several %" hygroscopic water at 100 C and more on ignition, and that the effect changes after ignition (largest for the dried but un-ignited sample, "somewhat reduced" after ignition; lead silicate and bismuth silicate "gave a moderate effect, but after ignition none"). Water content, packing, internal powder motion and electrostatic charge on dielectric powders are all plausible, untested alternatives; none is discussed.
• Containers were reversed in position and loads exchanged, which controls for a fixed left-right bias. That is good practice.

Experimental consistency (b)

• Eötvös-type torsion balance experiments (Eötvös 1889-1890 at about 1 part in 2 x 10^7; Eötvös-Pekár-Fekete 1906-09, published 1922, about 1 part in 10^8) [web: https://en.wikipedia.org/wiki/E%C3%B6tv%C3%B6s_experiment]. Modern torsion balances reach about 2 x 10^-13 (Schlamminger 2008) and the MICROSCOPE satellite about 3 x 10^-15 (2017) [web, same page]. These tests compare materials of different composition. They found no composition dependence at levels about 3000 times smaller than Brush's pendulum claim and about 10^4 times smaller than his barium aluminate claim at the time, and about 10^9 to 10^11 times smaller than those in the modern results. Which materials Eötvös used I could not confirm [check failed]. [memory, low]: wood, platinum, copper, water, and others; I cannot say whether bismuth or lead were among them.
• Brush replies (1926 letter, via Step 1) that the Eötvös result is "fallacious" and the true effect "a hundred times too small" in Eötvös's data. No reason is given in what Step 1 reports; and a sensitivity that is 3000 times finer than the claimed effect cannot be called "insensitive".
• Brush's balance-based claim could still be reconciled with Eötvös if the effect depends on something Eötvös's materials did not vary (crystal grain, purity). Brush's own bismuth-bismuth and zinc-zinc results are offered for exactly this. That is a falsifiable special pleading, but it conflicts with modern results on a wide range of materials in vacuum.

Tested? (c)

Brush's own results only. No independent replication of the pendulum or drop results is in the folder or in anything I could check [unknown]. The Eötvös-lineage experiments, which are controlled, neutral, vacuum-based and composition-varying, did not find an effect at any level close to his.

What would change (d)

The weak equivalence principle (universality of free fall) would be broken at the 10^-4 to 10^-5 level, with dependence on crystal structure and hydration state. This would contradict Eötvös and the whole GR lineage, and the modern limits.

Verdict: split. - Strong form (weight per unit mass differs by about 10^-4 to 10^-5 between common materials): Tested, not supported (Eötvös-type experiments and their modern successors, controlled and neutral, limits 10^-8 and far smaller [web]). - Brush's own pendulum and drop measurements as evidence: Never properly tested. No independent replication on record; the pendulum effect has an unexcluded, quantitatively matching conventional explanation (air buoyancy and added mass) which he never tested in vacuum; the drop test has admitted uncontrolled air and friction effects.

Claim 5. Some silicates continuously generate heat (not from radioactivity), and that heat correlates with the fall-impairment; this is the evidence for gravitational waves

a. Mathematics. None offered. No power figures (watts, joules per day, ice-melt rates), no sample masses, no blanks appear in the text I could read. Without them one cannot check consistency in the way I did for claim 3.

Internal logic problems I can state from the text [read: 1929 paper]: - The "correlation" is claimed across many substances, but at the date of the 1929 paper only two of the barium compounds had been run in the ice calorimeter (barium sulphate and the un-ignited barium aluminate), and the aluminate had been in for about six weeks. The sweep of more than 50 new drop plates was done "first, instead of hunting more or less at random for ... generation of heat". So for almost all the new drop-test materials there is no calorimeter datum. "Substances which have shown no generation of heat show no impairment" can be true only for those that have been tested in both; I could not find a count or a table of such pairs. - The only quantitative statements about size are "small / moderate / large". "Very moderate impairment" for barium sulphate matched with "very moderate" heat is one data point. A correlation of two ordinal labels over a handful of samples is not statistically a correlation. - The causal story is odd on its own terms. If the heat is the conversion of gravity-wave energy, then all matter, which feels gravity in proportion to mass, should heat in proportion to mass. In his own account only a few complex silicates do so. So the heat channel cannot be the same channel as the gravitational push for all matter. He does not address this. - Energy sanity check on the fall effect: a container with 10^-4 fractional impairment converts about 10^-4 x mgh of its fall energy; for 30.6 g and 1.1 m that is about 3 x 10^-5 J per drop, delivered while falling. The calorimeter heat is a continuous output of a body at rest. The two are connected only by the assertion that both come from "conversion of waves". No energy accounting links them.

b. Consistent with experiment? Slow heat from freshly prepared "wet-way" silicates, clay and hydrated salts has several ordinary candidate sources: adsorption and desorption of water, slow hydration or crystallisation of gels, carbon dioxide uptake, oxidation. Brush himself reports hygroscopic water and large changes on drying. He ruled out radioactivity as the cause ("none ... is more than minutely radioactive"). I accept that on size: ordinary rock radioactive heating is about microwatts per kilogram [memory, medium; not verified], hundreds of times below what a calorimeter of his type would see for a sample of a few hundred grams, so radioactivity is not a likely explanation in any case. But "not radioactive" does not exclude chemical sources, and a six-week run on one aluminate sample is short.

c. Tested? Brush says the Bureau of Standards, "with a calorimeter of its own design", "does continually generate a measurable amount of heat" for air-dried Sandusky clay and was about to test the nickel-cobalt silicate. [read: 1929 paper; tagged as Brush's report of an "official announcement"]. The 1927 Cleveland Press item (via Step 1) says Brush "financed" the Bureau repetition. I have no Bureau report, no numbers, and no date. So: - A single independent-laboratory heat result on one natural clay is claimed, secondhand. I cannot verify it. [web check not completed; no Bureau report found in folder] - That the clay "generates heat" is not the same as heat being tied to gravity. Even if the heat is real, the Bureau result says nothing about gravity. - The funding point is a mild flag in both directions: it is not an adversarial test, but it is a genuine external one, and Brush deserves credit for seeking it.

d. What would change. A new, previously unknown energy source in common minerals, continuously delivering heat without radioactivity. That would be a new physics or chemistry result of the first rank, and would violate energy conservation unless the source is identified.

Verdict: split. - That some silicates give off small continuing heat in ice calorimeters: Never properly tested (one secondhand, unverified Bureau of Standards statement; no data, no blanks in hand; mundane chemical sources not excluded). - That the heat comes from gravity waves and tracks the fall impairment: Not supported by the evidence presented, and Internally inconsistent in one respect (heat is substance-specific while the gravity push is universal). Closest scale entry: Never properly tested.

Claim 6. Recently quenched steel gives off heat for days to weeks

a. Mathematics. Calorimetry with simple arithmetic only. Calibration: 55 galvanometer divisions per K. Peak curves read about 100-160 divisions, so about 2 to 3 K excess between the steel can and the water can in the thermostatted copper block. [read: page images, 1917 paper, Figs 2-6; values read off the plots by eye, rough] The steel is 12 bars of 1/2 inch diameter and 5 inches long: roughly 6 kg. A crude figure of a few kJ of heat released over days, around 1 J/g, follows from an assumed heat capacity of about 3 kJ/K for steel plus can [calc; low confidence in the plot readings]. Nothing in this is inconsistent.

b. Consistent with experiment? Yes, in kind. [memory, medium: quenched steels contain retained austenite and a supersaturated martensite; at room temperature carbon redistributes and retained austenite can transform, releasing stored energy over hours to weeks, which is early-stage "ageing" or tempering.] My web check did not confirm this: the Wikipedia page on tempering [https://en.wikipedia.org/wiki/Tempering_(metallurgy)] did not mention room-temperature transformation (check failed). So treat the mechanism as [memory, medium].

Brush's controls are good for the period: oil coat against oxidation heat, a thermal-capacity-matched water can, a differential design, and, importantly, a furnace-cooled (not quenched) control with the same oxide film and oil, which showed "not the slightest trace of heat" (the first readings were about 0.02 K colder and it took days to equalise). [read: p.235] He also tested the recalescence hypothesis (quenched just below the critical temperature still gave about three quarters of the heat) and shrinkage (none observed after first hardening while heat was released). His own explanation ("relaxation of strain") is vague but not incompatible with the modern picture.

c. Tested? By Brush, with Hadfield-supplied steels, published in Physical Review (1917) and Proc. Royal Society (1917, with Hadfield and Main, per the bibliography; not read). Replications: the Step 1 file says the effect is "now recognised in metallurgy", which I did not check [web check failed]. I cannot say who replicated it.

d. What would change. Nothing. It fits existing metallurgy.

Verdict: Plausible, untested for the mechanism as I can verify it (consistent with known physics; I could not verify an independent replication in the folder, and my web check failed). If an independent confirmation exists in the metallurgy literature, this moves to Established. It is not tied to gravity in the 1917 paper; Brush's later use of it as part of the same programme is rhetorical only.

Claim 7. The Eötvös experiments are flawed or too insensitive

a. Mathematics. None given. Brush says he suspects "an error in the theory, the mathematics or the interpretation" (Step 1, from the 1926 letter, partly unreadable). He names no error.

b. Consistent with experiment? The Eötvös results have been repeated with growing precision up to 10^-13 and 10^-15 [web, Eötvös experiment page]. Independent torsion balances (Eöt-Wash group, others) and the MICROSCOPE satellite agree with each other. An "insensitive" Eötvös balance cannot be reconciled with sensitivity thousands of times finer than Brush's claimed effect.

c. Tested? Brush asked Dayton Miller (famous for his Mount Wilson ether-drift work) to mount a torsion balance on his interferometer. Miller replied (2 March 1926, via Step 1) that he had made 34,000 readings in 11 days at Mount Wilson, knew the Eötvös experiment only from manufacturers' circulars and did not know where to obtain the apparatus. Nothing shows he ever did it. So the specific lead-versus-sulphur test with a Brush-designed rig was never done; but this was not a rejection, only inaction. It is neither "tested and failed" nor "dismissed without testing" on the evidence I have.

d. What would change. All Eötvös-type results from 1890 to now would have to be wrong in the same direction.

Verdict: Tested, not supported. (Neutral, controlled torsion-balance and satellite experiments went on to confirm Eötvös at far higher precision [web].) Brush's specific proposed test, lead against sulphur or selenium, was Never properly tested, and never run by anyone as far as the folder shows.

Reading Brush as strictly as his critics

• Strength: he published his raw method, his failures (reversed orderings, one wrong-way photograph), his error sources, and invited critics. He sought outside checks (Bureau of Standards), a rare thing for the period.
• Weakness 1: he never ran the one test that could kill the pendulum effect (vacuum or equal-volume bobs) and dismissed it in advance. A buoyancy plus added-mass effect I estimate at 3.4 to 4.5 x 10^-5 of the same sign and size as his 3.1 x 10^-5 was not considered.
• Weakness 2: he says "I am not working to disprove Einstein ... only seeking facts" (1927 Cleveland Press, via Step 1) but also states that heat-gravity correlation justifies "dismissing doubt" on the basis of a handful of paired samples, scatter of order 100 % between photographs, and a calorimetric result he reports but does not show.
• Weakness 3: the effect is allowed to depend on whatever differs between samples (crystal grain, drying, ignition), so nearly any outcome can be accommodated afterwards. That is close to unfalsifiable unless the dependence is specified in advance.
• Weakness 4: the theory is never given as equations; so no quantitative prediction from the theory (as opposed to the experiments) ever existed to test.

One-line single experiment that would settle each

• Claim 4: repeat the zinc/bismuth pendulum in vacuum, or with bobs of equal volume and unequal mass. Buoyancy predicts the effect disappears in vacuum; Brush predicts it persists. A drop test with the same containers in a vacuum tube tests the falling effect equally.
• Claim 5: a blind calorimeter run (multiple labs, blanks, humidity and gas control) on Brush's nickel-cobalt silicate and barium aluminate, with the drop test run blind on the same samples.
• Claim 3: compute the orbit under the stated force law with an explicit x; a numerical orbit integration would show whether any x gives lunar and Mercury behaviour consistent with observation.

Summary of verdicts

1. Aether energy source for falling: weak form Established (textbook field energy); strong form Not testable as stated.
2. Push by isotropic penetrating waves: Historically superseded but instructive (Le Sage objections unanswered).
3. Non-inverse-square correction / finite speed: Refuted by calculation as stated; the cancelling term Not testable as stated.
4. Weight not proportional to mass: strong form Tested, not supported; Brush's own measurements Never properly tested (conventional buoyancy explanation not excluded; no replication).
5. Silicate heat and its link to gravity: heat Never properly tested; the link not supported and internally inconsistent.
6. Quenched steel heat: Plausible, untested (probably Established, but I could not verify an independent replication).
7. Eötvös flawed: Tested, not supported; Brush's own proposed test never run.