Mills-Randal: Arbiter (Step 3)

Project: R. L. Mills, The Grand Unified Theory of Classical Physics (GUT-CP), project copy = "December 10, 2018 Edition" PDF (1,876 pp; file name "2020 Ed Web"). Date of arbitration 2026-10-09. Inputs weighed: claims.md, facts.json, physicist.md, mathematician.md, contrarian.md, deep reads, sites.md and research/ notes. Tags: [read] from the project documents or reviewer files that quote them; [web] with URL; [memory] with confidence; [calc] recomputed by me in the project venv (mpmath) from printed numbers. Interested party: Mills founded and runs BlackLight Power / Brilliant Light Power (BLP), which sells the hydrino technology. The book, the company reports and most cited experimental papers come from that side. That is a fact about the sources, not a judgement of their data. Several "validation" authors were hired by or co-authored with the company (Booker, Tse, Nansteel); the sole outside-academic datum with a declared absence of company funding is the EPR measurement by W. R. Hagen (TU Delft), on a sample the company supplied.

Source gap note (repeated, as required)

• The deep reads cover about 40 percent of the 1,876 pages. Not deep-read by anyone: Ch.3-4 (free electron, photon), Ch.9, Ch.11 (the H2 molecule and H2(1/p): the source of the 4.478 eV bond and 31.68 eV numbers), Ch.15 pdf 801-1082, Ch.17-21, Ch.23, Ch.24-30, Ch.31, Ch.33-35 and the first half of Ch.32 (Schwarzschild metric from the wave equation, perihelion, light deflection).
• Claim 7 rests on digests and [memory] alone. Its verdict below is provisional. Any statement about Ch.11, the first half of Ch.32, and the GR part of Claim 6 is unassessed, not "passed".
• Equations are image-only in the text layer; every number the deep readers checked came from page images. I did not re-render pages. I recomputed the reviewers' numbers from printed formulas where the formulas were given in the review files.
• Not in the project folder or not read by any reviewer: the November 1995 edition (the alleged prediction of cosmic acceleration and top-quark mass); Mills's 165-page 2008 reply to Rathke (only the sign-slip point and the delta-shell admission were read, by the mathematician); the Booker/Nansteel theory reviews; the company calorimetry reports and the 194-page Jansson report; Phelps-Clementson and Lawler-Goebel (abstracts only); the Hagen-Mills supplementary file (controls); the Harvard CfA continuum report.
• Where a conclusion rests on [memory] alone I say so in the claim sections.
• Update 2026-10-09 (QE revision, below): the QE-ordered deep reads of Ch.3-4, Ch.9, Ch.11, Ch.15 (pdf 801-1082, prose only), Ch.24-30, Ch.31, the first half of Ch.32, Ch.34 (from Eq. 34.10), Ch.35 and the Introduction (pdf 1-86) are now done. The lists above are out of date for those chapters; where the QE revision and an earlier section differ, the QE revision governs. Still not deep-read by anyone: Ch.17-21, Ch.23, Ch.33 and the start of Ch.34 (pdf 1661-1665), the numerical tables of Ch.15, and the amino-acid and phosphate sections (not found in pdf 801-1082).

Verdict scale used

Established; Plausible, untested; Never properly tested; Tested, not supported; Refuted by calculation; Internally inconsistent; Not testable as stated; Historically superseded but instructive. Split verdicts are stated part by part.

Recomputation where the reviewers differ or a number carries a verdict [calc]

I could not recompute the hydride-ion reduced-mass sensitivity (0.754104 vs 0.754175 eV) because the formula is image-only; I carry the mathematician's [calc] with that caveat.

Web checks done (budget 8; 6 used)

1. Planck 2018: H0 = 67.4 +/- 0.5 km/s/Mpc [web: https://arxiv.org/abs/1807.06209]. Retagged from [memory, high].
2. BICEP/Keck BK18: r(0.05) < 0.036 at 95 percent [web: https://arxiv.org/abs/2110.10902]; tighter (r < 0.032) with Planck PR4 and BAO. Retagged from [memory, high]. Supports "BICEP2's r = 0.2 withdrawn" only indirectly (not checked here; stays [memory, high]; it does not drive a verdict because r = 0.16 vs <0.036 stands on BK18 alone).
3. PDG top-quark mass, direct measurements: 172.57 +/- 0.29 GeV (scale factor 1.5) [web: https://pdglive.lbl.gov/view/Q007TP]. Retagged from [memory, medium-high]. The PDG notes this is the Monte-Carlo mass parameter, not a pole mass; the 8 percent gap to Mills's 187 GeV is far larger than that ambiguity (about 0.5-1 GeV [memory, medium]).
4. 3.5 keV line: XRISM Collaboration stacked ten clusters (3.75 Ms): no unidentified line; 3-sigma decay-rate limit about 1.0e-27 /s, 3-4 times lower than Hitomi's Perseus limit but still 5 times above the rate implied by Bulbul et al. (2014); ApJ Letters 994, L28 (2025). Dessert, Rodd and Safdi blank-sky XMM analysis (Science 2020) is inconsistent with a dark-matter origin; a 2025 talk by Dessert argues the remaining detection shows mismodelling (conference abstract only) [web: https://arxiv.org/abs/1812.06976; XRISM stacked-cluster search listing via https://shogo-benjamin-kobayashi.owlstown.net/publications/64647-xrism-constraints-on-unidentified-x-ray-emission-lines-including-the-3-5-kev-line-in-the-stacked-spectrum-of-10-galaxy-clusters]. So the line is unconfirmed, not excluded at the original flux by XRISM. The physicist's "Tested, not supported" overstates what XRISM alone does; Dessert-Rodd-Safdi is the stronger test and is disputed (Abazajian, per the physicist).
5. Phelps and Clementson, Eur. Phys. J. D 66, 120 (2012): argue the Mills EUV continua come from ions sputtered or evaporated from the electrodes (O V, O VI lines present; background shape varies with Ta, W, Mo cathodes), "removing the justification for the introduction of hydrino particles" [web: https://link.springer.com/article/10.1140/epjd/e2012-30114-5, abstract and search summary; full text not read]. It is an interpretive critique, not a new experiment.
6. [2008] EWHC 2763 (Pat), Floyd J: the appeal was ALLOWED and the applications REMITTED to the Hearing Officer, to ask "whether there is a reasonable prospect that on a full investigation with the benefit of expert evidence" GUTCQM will turn out to be a valid theory (paras 49-52) [web: https://caselaw.nationalarchives.gov.uk/ewhc/pat/2008/2763]. A search summary of the later UK IPO decision (BAILII O/170/09) says the Hearing Officer, on remittal, found no reasonable prospect and refused the applications again; I could not open the BAILII page (HTTP 403) and did not find a further appeal [web: https://mansfield.bailii.org/uk/cases/UKIntelP/2009/o17009.html, via search summary only]. This corrects the physicist (see Corrections).

Not re-checked, still [memory], none decisive: Bullet Cluster argument; Benson group-additivity precision; Tonomura AB; ab initio atomic ionization energies; Hess-Philipp/Gill et al. 2002 reply; BICEP2 withdrawal. Aspect 1981 values (delta_QM = 5.8e-2 +/- 0.2e-2, measured 5.72e-2 +/- 0.43e-2) are [web via the contrarian's search summary, https://arxiv.org/pdf/2212.05535]; I did not open the PRL, so I carry them as [web, secondary]. The Bell conclusion does not depend on them (Hensen et al. 2015, S = 2.42 +/- 0.20 [web: https://arxiv.org/abs/1508.05949], read by the physicist).

---

Claim 1. Classical electron shell; closed-form atoms; g factor

Reviewers agree. Arithmetic reproduces everywhere checked (Bohr energy, Table 1.5, Table 7.1, helium r1 = 0.566987 a0, Eq. 1.228). Postulates, not derivations, supply hbar (circular Eqs. 1.35/1.37), spin 1/2 (hemisphere construction with the target stated first), the full Bohr magneton (extra Larmor-photon postulate) and N = 3/2. Nonradiation for l >= 1 needs lambda_n = r_n; the correct factor j_l(k r_n) is not zero (j_1(2 pi) = -0.159). The self-force is removed by postulate; with it, the same balance gives r = 2 a0. Eq. 1.228 is 2.87e-10 (about 2,200 sigma) from the measured g with modern alpha, and its second and third coefficients differ from QED's (-1/3 vs -0.3285; pi^2/3 = 3.29 vs 1.18) with errors that nearly cancel. The muon-g "prediction" returns its input.

Where they differ. - Mathematician adds a stronger result: a rotating l >= 1 spherical-harmonic charge wave satisfies the book's own two-dimensional wave equation only if v = m omega R / sqrt(l(l+1)), never v = omega R (sympy, with a clean argument: l(l+1) = m^2 has no solution with l >= 1, |m| <= l). It is independent of the radial delta function. Physicist and contrarian did not have this; it is not contradicted by anyone and I adopt it. - Multi-electron atoms: mathematician and physicist say "Tested, not supported" as "exact"; contrarian says the integer choices are "Never properly tested" (no blind test). Both are true of different statements: as a parameter-free exact solution the 2-4.6 percent errors and chosen integers do not support the claim; as a semi-empirical scheme it has never been tested on ions it was not tuned to. - Fairness items kept in the record (contrarian's "strongest honest case"): nonradiating extended sources exist in classical electrodynamics (Goedecke 1964; Devaney and Wolf 1973 [memory, high]); a single-constant closed form that reaches the hydrogen ground state, Schwinger's term and 1-20-electron ionization energies to a few percent is not trivial; the work is explicit and checkable. Against that: one-electron cases are Bohr/Dirac relabelled, and the one sharp precision test goes the wrong way.

Stronger argument. Mathematician and contrarian on the postulates; all three agree on g. The g-factor failure is arithmetic and needs no judgement: CODATA alpha is independent of g-2 (Rb and Cs recoil), and the result is 2.9e-10 off. The old "eleven-figure" agreement was a coincidence with a poorly known 1970s-80s alpha and two nearly cancelling wrong coefficients.

Settling calculation (done). Eq. 1.228 against the modern alpha: failed. A further calculation that would settle the nonradiation question: evaluate the l = 1 Fourier component at k = omega/c numerically with the correct j_1 factor, and solve the shell force balance with the Purcell self-pressure included.

Final verdict (split). - Derivation of spin, hbar, Bohr magneton and nonradiation (l >= 1, self-force) from Maxwell and Newton: Internally inconsistent. - g factor, Lamb shift, muonium, 21 cm as closed-form predictions: Tested, not supported (g: about 2,200 sigma; Lamb 80 sigma; muonium 10 sigma; arithmetic correct; per-case factors). - Ionization energies of 2-20-electron atoms as exact, parameter-free: Tested, not supported; as a post-hoc scheme with chosen integers: Never properly tested on unseen cases. - One-electron energies (Bohr/Dirac, Table 1.5): Established as Dirac's result, not independent support. - Muon g "explanation" (App. III Eq. 26): Internally inconsistent (circular).

---

Claim 2. Hydrinos and the BlackLight process

Reviewers agree. The energy-hole algebra reproduces: edges 13.6, 54.4, 122.4 eV (91.2, 22.8, 10.1 nm) for p = 1, m = 1-3; H -> H(1/4) releases 204 eV vs 1.48 eV per H atom for combustion. The force balance (Eq. 5.28) is an identity for any real Z_eff, so integers come from a separate "trapped photon" postulate (Eq. 5.27) and the 27.2 eV unit from a multipole-counting assumption. Standard QM has no state below n = 1; the formal route (Naudts, Khelashvili-Nadareishvili) uses singular solutions that vanish for a finite nucleus (Dombey). The hydride affinity (0.75418 vs 0.754195(19) eV, about 1 sigma) is ordinary physics and does not discriminate. Every Ch.5 experimental citation has Mills as an author. The index limit moved from p <= 137 (book) to p <= 126 (2025 paper, a different stated reason). Hagen-Mills 2022 is the strongest outside datum and its limits are in its own words ("semi-quantitative", "0.004 +/- 0.550 gauss", fluxon separation "not a constant", "presently not understood", a broad underlying signal of "origin and nature unknown", a conventional two-radical fit that "approximately reproduces" it). The Bruker duplicate is reported by Mills and cites Mills's own preprint; no Bruker-authored record.

Where they differ. 1. What to call the theory. Physicist: "Tested, not supported" (QM theorem excludes it; spectroscopy sees no need). Mathematician: "Internally inconsistent" for Mills's derivation and "Refuted by calculation" for the standard-QM version with a finite nucleus. Contrarian: "Never properly tested" for the strong form. Ruling: these answer different questions. (i) Mills's own derivation is internally inconsistent (integers not selected by the force balance; delta-shell and rotating harmonics fail his wave equation; self-force). (ii) A sub-ground state in the hydrogen Hamiltonian with a finite-size proton is excluded by calculation; the mathematician's caveat is correct and I keep it: for a point proton, the l = 0 radial problem is limit-circle and admits self-adjoint extensions with an extra deep state, so "no such state exists" is a statement about the physical boundary condition (Dombey; Khelashvili-Nadareishvili: non-conserved norm), which is why the verdict applies to physical protons. (iii) Whether something at the lab level behaves as Mills says is an experimental question and has not had a neutral test. Calling (iii) "Tested, not supported" because of (ii) would mix theory-based rejection with experiment; the physicist's own text keeps them apart for kW energy. 2. Was it dismissed without testing? Physicist and mathematician: the patent rulings were institutional dismissals on theory; contrarian: "dismissed on theory, without a neutral experimental test", but notes the one examiner who saw the data and samples allowed the claims. The correct statement: at the patent offices it was dismissed on theory without a neutral test; elsewhere it has been tested by affiliates and by a few outsiders (NASA 1996, EarthTech 1997, Hagen 2022), with mixed results. Both are in the record. 3. The UK ruling. The physicist wrote that Floyd J's appeal was "dismissed"; the contrarian wrote "remitted [memory, medium]". Web check above: allowed and remitted with a new test (reasonable prospect on a full investigation with expert evidence); the Hearing Officer then reportedly refused again. The physicist's statement is wrong; see Corrections. 4. Phelps and Clementson. Contrarian cites it with a web summary; physicist marked it [memory, low-medium] and declined to rely. Now [web] (abstract). It offers an ordinary source for the 10.1 and 22.8 nm continua (electrode ions); it has not been shown to account for every spectral feature, and Mills and Lu argued no conventional explanation was plausible; I have not read the reply. Both sides' positions stay in the record.

Stronger argument. On theory: mathematician and physicist (they agree and the mathematician's evidence is a calculation). On experiment: the contrarian is better calibrated (no neutral controlled protocol exists; positive outside results name their own ordinary explanations; the one clear independent replication, EarthTech, was negative for the light-water cell). The physicist's point that a real 100 kW hydrino source would be easy for outsiders to authenticate within days (mass spectrometer, a loaned calorimeter) is an argument about plausibility, not a test.

Mathematician's EPR analysis kept. The predicted fine-line grid is dense: with finest spacing s = 0.31 G, chance mean miss is s/4 = 0.078 G, so the book's 0.097 G is at chance level unless the assignment table is shown; the paper's main-line comparison (11 lines, mean absolute 0.43 G on a grid 2-6 G apart, where random placement gives 0.5-1.5 G) is better than chance but "semi-quantitative" in the authors' words. The physicist's ordinary-explanation point (a fixed-g multi-line pattern from 69Ga/71Ga, I = 3/2, or a defect centre) is [memory, medium]; the authors' counter-argument (fixed g, hyperfine spacing independent of frequency, no anisotropy, no N or transition ions by ToF-SIMS, 11 samples) is real and not refuted by a blank from a hydrogen-free run (none found in the files).

Absence of evidence (central to this claim). See the section below.

Settling experiment. (a) EPR on a Ga(O)OH/FeOOH product from a neutral lab's own reactor run, plus a matched hydrogen-free control run through the same KOH purification, at several microwave bands with spin counting, and independent measurement at Bruker with a Bruker-authored report; (b) Raman/IR of the product at the predicted H2(1/4) lines (rotational spacing 16 x the H2 constant; about 1950 cm^-1 in Table 16.52) by a lab with no stake; (c) a heat run with a neutral party supplying its own calorimeter and a recombiner-equipped light-water cell (the EarthTech design) and a BLP SunCell run with outside instrumentation; (d) mass-spectrometric or gravimetric identification of the claimed bulk hydrino gas. The IIT Kanpur results are the first outside datum due.

Final verdict (split). - Mills's derivation of states below n = 1 (force balance, trapped photon, wave-equation solutions, self-force): Internally inconsistent. - Sub-ground states of the hydrogen Hamiltonian with a physical finite-size proton: Refuted by calculation (boundary-condition caveat stated above). - Hydrino identification of the lab anomalies (EUV continua, EPR, DSC exotherms): Never properly tested (outside measurements exist; no neutral controlled test of the p-scaled signatures; for the EPR, one company-linked replication claim; ordinary alternative fit not excluded by a blank). - Energy release at kW scale: Never properly tested (company-hosted, hired or co-author observers, single runs; not dismissed without testing at the claim level; dismissed at the patent offices on theory). - Light-water Ni-K2CO3 cell as a hydrino heat source: Tested, not supported for the EarthTech 1997 repeat (negative, with recombiner); NASA 1996 ambiguous (authors could not exclude recombination); Thermacore 1994 contractor "in cooperation with" the inventor. Overall for this earlier form: Never properly tested by a neutral party; the EarthTech result is stated beside NASA's. - Algebra of energy release, Table 5.1/5.2 arithmetic: reproducible (no verdict needed); the 27.2 eV resonance table is uninformative without the count of candidates tried (above).

---

Claim 3. Hydrinos as dark matter and cosmic emission

Agree. Cascade arithmetic reproduces; 3481.6 eV = 256 x 13.6 eV; the power and rate examples (8e21 /s; 100.7 and 144.1 kW) hold only because coupling factors are set to 1 and an efficiency of 1e-4 is chosen; the p <= 137 limit used in the chain is withdrawn in the author's 2025 paper (p <= 126). The 10.1/22.8/91.2 nm pairing with m = 1/2/3 is reversed on p.263. The ladder of available energies (p up to 137, several m) is dense, so one-number matches are weak without a pre-fixed line list. Hydrinos are baryonic by the book's account, which sits against the CMB and BBN baryon budget; the book does not address the Bullet Cluster [memory, medium; not checked].

Differ. Physicist: dark-matter identity "Never properly tested", 3.48 keV "Tested, not supported". Mathematician: "Never properly tested". Contrarian: "Never properly tested"; 3.48 keV match "Not testable as stated" as evidence, and the 3.5 keV line itself "Tested, not supported" by Hitomi/XRISM. Ruling: the claim contains three things. (i) The line as a dark-matter decay: Hitomi (Perseus) and XRISM (ten-cluster stack) have not confirmed it; XRISM's limit is still 5x above the original rate, so XRISM alone does not exclude the original flux; the XMM blank-sky analysis (Dessert, Rodd, Safdi 2020) is inconsistent with a dark-matter origin but is disputed (Abazajian). Verdict for the dark-matter-decay interpretation: Tested, not supported (contested); for the line's existence: unresolved. (ii) The hydrino assignment of a single energy, 1-2 percent from the reported value, drawn from a dense ladder, with no predicted shape or flux: Not testable as stated. (iii) The identity of dark matter with hydrinos: Never properly tested, and cannot be until hydrinos are independently produced (Claim 2). The physicist's "Tested, not supported" for the assignment is not adopted because the evidence against is about the line, not about the hydrino.

Final verdict. Dark matter = hydrinos: Never properly tested (conditional on Claim 2). 3.48 keV assignment as evidence: Not testable as stated. 3.5 keV as decaying dark matter generally: Tested, not supported (contested). 511 keV and gamma-ray-burst assignments: Never properly tested (cited by reference only; not deep-read in detail).

Settle it. A predicted continuum shape, flux and line list written before examining Hitomi/XRISM/XMM data; a look-elsewhere count over the whole p, m ladder; independent production of hydrinos.

---

Claim 4. One method for molecules and matter

Agree. Arithmetic reproduces (hundreds of numbers). 431 summary rows: mean absolute relative error 0.172 percent, median 0.056, 317 within 0.1 percent; but 1-3 percent for 13 of 20 amino acids (12 compared with crystal-phase data against gas-phase calculations), cytosine, trimethylphosphine, D-ribose, noble-gas van der Waals; nine Table 15 rows whose error does not follow (exp - calc)/exp. "No fitted parameters" is not true in the sense of no case-specific inputs: experimental vibrational energies and sub-radical bond energies enter, and hybridisation factors (0.91771, 0.93172), charge splits (0.75/0.25), the CH2 magnetic term (0.14803 eV) and "energy-minimum" integers are chosen group by group; boron-compound errors exceed the 0.87 eV bridge bond energy. No blind or pre-registered test exists. Booker's checks are company-hosted arithmetic checks.

Differ. Contrarian: out-of-sample accuracy "Plausible, untested"; "no fitted parameters" "Internally inconsistent in part". Physicist/mathematician: strong form "Never properly tested", weak form "Established as arithmetic". Ruling: "Plausible, untested" is reserved for claims consistent with known physics with no real test yet. The strong form claims to replace quantum chemistry; it is a testable claim with a testable procedure and no neutral test. Never properly tested. The sentence "no fitted parameters" is Not supported as stated by the book's own derivations (I use the scale name Internally inconsistent for that sentence only, per the contrarian, since the book's text and its derivations conflict). The 0.17 percent is comparable to group-additivity schemes that use fitted values (Benson) [memory, medium]; it would be remarkable only if the choices are fixed in advance, which the text does not show. Ch.11 (H2 itself) was not deep-read, so this verdict is provisional there.

Settle it. A list of molecules and groups published with all rules frozen before the experimental values are looked up (the pre-registered blind test); then compare with DFT/CCSD(T) on the same list. The ordinary Hobbs-catalogue match of H2(1/4) Raman lines to diffuse interstellar bands needs a chance-coincidence estimate (none in the book).

Final verdict (split). Weak form (a rule-based group scheme reproducing tabulated totals to about 0.2 percent): Established as arithmetic (not as physics). Strong form (first-principles closed form with no fitted choices, accuracy as prediction): Never properly tested. "No fitted parameters" as worded: Internally inconsistent with the case-specific inputs.

---

Claim 5. Quantum mechanics is wrong or unnecessary

Agree. The "Schroedinger charge must radiate" argument uses a time-harmonic density that a stationary state does not have; the transform of r exp(-r/a0) (not the 1s density) is correct up to a factor 2 and is generic for any smooth density; the algebra is fine, the premise is the error. The "classical" Aspect curve (Eq. 42.23/42.29) is Aspect's QM prediction formula (now [web, secondary] delta_QM = 5.8e-2); the words-only mechanism (shared random polarisation, Malus law) gives half the modulation (0.1089 vs 0.2178 cos 2 phi) and satisfies the inequality (delta = -0.096). The book's own statement of Bell's theorem uses deterministic outcome functions A(a, lambda), contradicting "Bell's theorem does not apply to deterministic classical physics". Jackson's Eqs. 17.70/17.71 labelled reversed. No Bell test after 2007 is cited; loophole-free tests (2015) are not addressed. Muon g: circular.

Differ. Contrarian: radiation argument "Internally inconsistent (premise not derived)"; physicist and mathematician: "Refuted by calculation". Ruling: the premise (a density oscillating at omega_n) is contradicted by the stationarity of the standard state; accepted electrodynamics applied correctly rules out the conclusion. That is Refuted by calculation (a physical statement about QM's stationary state), and the book also contains the circular steps. The contrarian's helium-mobility peaks (15 peaks, r1 = 6.7 A, mean deviation 3.4 percent, max 8.9, about 69 candidate (n, l, m) values) were counted as a free-choice problem by the deep reader; nobody has compared with alternative explanations. The contrarian's fair point is kept: foundational criticism of the Born rule and of the lack of a physical picture of the electron is legitimate; superdeterminism is a position some physicists hold, though not Mills's version.

Final verdict (split). "Schroedinger charge density must radiate": Refuted by calculation. "Aspect/Bell results are classical": Internally inconsistent (the 'classical' curve is the QM formula; Bell's functions are deterministic) and Tested, not supported (loophole-free Bell tests, Delft 2015 S = 2.42 +/- 0.20 and others). "Uncertainty principle and Born rule are refuted": Not testable as stated for the interpretation (no measurable difference proposed); Tested, not supported where Mills offers observable differences. Superfluid-helium electron-mobility peaks: Never properly tested against alternatives (lack of a chance count). Stern-Gerlach, electron diffraction and SQUID-cat comparisons: single-number matches with fitted-by-choice quantum numbers; SQUID energy reproduces only with the book's own alpha factor (otherwise about 120 K vs 0.14 K): Not testable as stated as evidence.

Settle it. Run the Eq. 42.23 model through the 2015 loophole-free data; count the free (n, l, m) choices in the helium-mobility fit and compare with an alternative bubble model.

---

Claim 6. Spacetime, gravity, cosmology, particle masses

Agree. Reproduced: T = 2 pi G m_U / c^3 = 3.11e19 s (book 3.10e19), H0 = 78.3-78.5, the lepton ratios, Z/W/Higgs, mu_p = (8 pi/9) mu_N. Errors listed by all: 1 Gyr = 3.1358e16 s (correct 3.156e16), prefactor 1229 vs 12,290, "km/s^2 = km/s/Mpc" (Eq. 32.155), "sec = 0.9975 s" fitted from m_e, box 32.3 ratio mislabelled, neutron mass 5.2e-6 off. Observations: H0 7.5-16.5 percent high; T_CMB 3.08 K vs 2.7255 K (13 percent); CMB peaks 197/541/824 close to 220/540/810 only with t = 10 Gyr (13.8 Gyr gives 143, 369, 555); r = 0.16 vs r < 0.036; top 187 vs 172.57 GeV (+8.4 percent); Higgs +2.8 percent; deuteron -2 percent; proton moment -1.1e-4 relative.

Differ. Only in emphasis on the muon/electron ratio. The mathematician ran a look-elsewhere count: 57,600 two-factor corrections, 24,148 within 1 percent of the needed correction, one family within 2.2e-8, chance expectation about 0.02. That count is built around Mills's own formula, with free base exponent 2/3 and placement of 2 pi, so it does not remove the hindsight problem; the tau ratio (miss 5.6e-5 vs 5e-5 uncertainty) is an easy target. The contrarian calls it "unreasonably good if the corrections are pre-specified"; the physicist calls it possible numerology. My recomputation confirms the miss -1.6e-8 inside the uncertainty (2.2e-8). Ruling: a striking, unexplained, retrospective match; no prospective statement exists.

The mathematician's H0 point is important and new to the other two: H0 is not a prediction of the model but follows from the chosen m_U (2.0e54 kg "for internal consistency", [read]); no single m_U gives both H0 = 67.4 and T_CMB = 2.7255 K (78.27 vs 61.1 at the T-matching mass). So the cosmology numbers are fitted and still miss.

The 1995 "prediction". All three agree that the November 1995 edition is not in the project folder or found online, so "predicted acceleration and top-quark mass in 1995" is unverified as a prediction. Facts that bear on it: the book cites its own Nov. 1995 edition, Ch.22; the model accelerates from t = 0 with no earlier decelerating epoch, whereas data show deceleration then acceleration (z about 0.7) [memory, high]; the top-quark discovery was announced 2 March 1995 [memory, high], so a mass in a late-1995 edition is not blind. The cited supernova references are news items.

Final verdict (split). Cosmology numbers (H0, T_CMB, CMB peaks, B-mode, density): Tested, not supported. Top quark, Higgs, nucleon moments, deuteron: Tested, not supported. Muon/electron mass ratio: Never properly tested (retrospective, corrections justified in words). Absolute lepton masses with fitted "sec": Not testable as stated. The claim of having predicted acceleration and the top mass in 1995: Never properly tested (primary edition not obtained). Gravity from the wave equation, Schwarzschild metric, perihelion and light deflection (first half of Ch.32): not assessed (not read).

Settle it. Obtain the November 1995 edition (Library of Congress / WorldCat / archive) and the earliest dated statements of H0 = 78.5, the CMB-peak formula, the top mass, the X17 mass (17.11 MeV; the book postdates the ATOMKI reports, and X17 is unconfirmed [memory, medium-low]); write down the correction factors 1/(1 + alpha/2) and (1 + pi alpha^2) from first principles and predict the tau/electron ratio before looking.

---

Claim 7. Collective phenomena are classical (this section is the pre-QE, digest-only version; SUPERSEDED by "QE revision", claim 7)

Agree. Ch.24 re-derives Bose-Einstein, Fermi-Dirac, Planck and Debye results using h and photon counting in cavity modes, i.e. it reproduces the standard results rather than deriving them classically. Ch.25 gives Tc = 30.8 K for conventional metals (Nb3Ge 23.2 K) and 7, 22, 180 K for oxides depending on chosen dimension (a free discrete choice, factor 25); Ch.26 R_H = h/(n e^2) with n entering by a band-pass condition and fractions by "linked wells"; Ch.27 treats Aharonov-Bohm as a magnetoresistance effect (the mathematician notes this cannot fit the vacuum electron-interferometry version [memory, high]); Ch.30 quotes g/2 = 1.001159652137 vs 1.001159652188(4) (contrarian, from the digest, about 5e-11 off, page image not read; corrected in the QE revision: this is the free-electron g re-quoted from Ch.1, not a positronium quantity). Differ. None. All three say not assessed beyond the digests. I add no new analysis. Final verdict (provisional). Never properly tested (provisional, digest only). The one stated reason to doubt it that survives from the digests is that h and the integers are supplied as inputs. The Opus deep read of Ch.25-27 ordered in claims.md has not been done and should be run before this verdict is relied on.

---

The Investigator's findings (kept in the record whatever the verdict)

• Newer edition: February 2025 (1,888 pp) at the same file name; p <= 126 (not 137); the changes are evidence of where the theory is moving.
• Hagen-Mills 2022 (read by physicist, contrarian and mathematician), Hagen-Mills 2024/26 broad-line EPR (anti-Curie behaviour, transition near 107 K), the cosmic-neutrino-background EPR paper (not read).
• NASA Lewis TM-107167 (Niedra et al., Feb 1996): excess up to 11 W on 60 W in, gains 1.06-1.68 in a cell on loan from Hydrocatalysis; recombination "an ordinary explanation" the authors could not exclude; further study "warranted". EarthTech 1997: 164 h, Pd recombiner, no excess within about 10 percent. Thermacore SBIR (USAF Wright Lab, 1993-94): 25 W; Lehigh ESCA reported the predicted 55 eV peak; no Phase II located. NIAC Phase I (Rowan, 2002). IIT Kanpur ECS 2026 abstract (replication programme announced, no results; the group's abstract treats hydrinos as already supported, so it is not neutral in advance).
• Company-hosted calorimetry: Nansteel 2021 (one 16 March 2021 test, 169-210 kW, local gain about 4.2), Booker 2020 (194-218 kW), Tse 2020 (Rutgers, hired observer), Glumac 2014 DSC (UIUC; 101-148 J/g exotherms "far more exothermic than the predicted thermochemistry"), Crouse DSC, 2016 summary claiming "1 MW+", Jansson 194-page redacted image-only report. Presented by an interested party; the authors are named scientists; none was read in full by a reviewer.
• Booker's chapter-by-chapter checks: arithmetic confirmation, consistent with our deep reads; they do not address the alpha dependence of g, the circular muon input, the 16 pi^2 factor or the nonradiation proof.
• Related mainstream-formalism work: Naudts 2005, Khelashvili and Nadareishvili 2012/2015, Giri 2008; Dombey 2006, de Castro 2007 (against). Rathke 2005 (New J. Phys. 7, 127: "severe inconsistencies in the deterministic model") and Mills's 2008 reply: the mathematician found Rathke's Eq. (9) has the wrong sign on the time-derivative term, and that Mills's own reply concedes the delta function "is not a radial solution of Eq. 4; rather, it defines a constant charge density on a spherical shell" (a boundary-condition choice). Both points stand: Rathke's sign slip and the surviving objections.
• Phillips 2008, Loureiro-Amorim 2010 on fast H; Selke 2025 (force-balance ionization energies to heavy ions; not read).
• Speculative relatives recorded as found: Zhigalov's placement of the hydrino beside Baranov-Zatelepin "dark hydrogen", Nevolin's subatoms and erzions, Surmont 2025, Wulf 2026, Magnitskii 2025 (not weighed).
• USPTO 2000: application 09/009,294 allowed after the examiner reviewed experimental results and samples, withdrawn from issue on 17 Feb 2000 after "an outside inquiry"; the Group Director's concern was "the electron going to a lower orbital in a fashion that I knew was contrary to the known laws of physics and chemistry"; "undisputed that no one involved in the withdrawal had reviewed the '294 patent application before issuance of the Notice" (quote restored in full per the QE revision) [read: research/2026-10-09-findlaw.md; web: https://caselaw.findlaw.com/court/us-federal-circuit/1362587.html]; Federal Circuit 2002 upheld the Director's authority, not the physics. UK: IPO BL O/114/08 (17 April 2008) refusal; Floyd J allowed the appeal and remitted (web check above); the Hearing Officer then reportedly refused again (O/170/09, secondary only).

---

Absence-of-evidence analysis

What goes dark. About 35 years (1991-2026) of company existence (HydroCatalysis -> BlackLight Power -> Brilliant Light Power); Form D sales of $18.3M by 2018 [read: EDGAR note] and "over $150 million" by the company's own 2025 statement (the company's characterization; more than $25M by 1999 and more than $60M by 2006 per press); demonstrations, validation reports and a 100 kW prototype claimed since 2020; and no product sale found, no independent demonstration of a running cell with the outsider's own instruments, only one non-company group (IIT Kanpur, 2026) announcing replication, and no results from it yet. Specific disappearances: the "13 laboratories" the company lists as confirmations (AECL Chalk River, Westinghouse, MIT Lincoln, INEL, Hokkaido, Moscow Power Engineering Institute, Brookhaven, with Lehigh and Harvard CfA) whose own reports were not found; Thermacore Phase II; the November 1995 edition; the redacted Jansson report; the Hagen-Mills supplementary controls.

Competing explanations, evidence for and against. 1. Real effect, hard to engineer/commercialise. For: repeat investors and a stable board over decades; positive outside measurements (NASA 1996 excess, Glumac DSC, Hagen EPR) that the measurers did not explain; anomalies across several techniques. Against: no sale or independent demonstration in 35 years of claimed readiness; NASA's authors themselves named recombination; EarthTech negative; the changing index limit (137 -> 126) and EPR metric (0.097 G -> 0.55 G scatter) show a theory still being adjusted. Deciding record: outside engineers running the cell with their own instruments; the Kanpur results. 2. Trade secrecy / IP protection. For: the heavily redacted Jansson and summary reports; patents are a main asset; commercial interest in not releasing details. Against: the company publishes the book, validation reports and the "urgent invitation to academia at large to repeat and extend" (Hagen-Mills 2022 p.8); secrecy would not prevent a mass-spectrometer test of a sample under a non-disclosure agreement. Deciding record: unredacted reports; whether any NDA tests exist. 3. Absorption/renaming. For: the successive names (HydroCatalysis, BlackLight, Brilliant Light) and the move of the reports to company-hosted sites. Against: the continuity of the same inventor and the same claims; no evidence of a separate absorbing company. This explains the trail, not the silence. 4. The effect is not real (artifacts: recombination, electrode ions, calorimetry error, ordinary paramagnetic centres). For: NASA's own caveat; EarthTech negative; Phelps-Clementson; the conventional two-radical EPR fit; the Hagen paper's own "contaminated" sample and the single-source sample production; interested-party calorimetry with hired observers; no independent replication in 35 years. Against: the anomalies are multiple and measured by multiple named scientists; a ready artifact for every measurement is not yet shown (Phelps-Clementson address the EUV only); the effect has not had a neutral negative result for the SunCell either. Deciding record: independent replication with its own samples and a hydrogen-free control. 5. Institutional refusal to test after the 2000 USPTO and 2008 UK rulings. For: the USPTO withdrawal (no one involved in the withdrawal had reviewed the '294 application "before issuance of the Notice", i.e. before the allowance notice, not "ever"); the Group Director's physics objection; UK refusal on "contrary to accepted physical law" grounds, remittal then second refusal; Rathke wrote that discussion of the theory had been "restricted to the sweeping statement that the hydrino model is in contradiction to quantum mechanics" [read via the contrarian]. Against: NASA/NIAC funded a Phase I (2002); Rowan, Rutgers, UIUC, UNC Asheville academics did measurements; a patent office is not a laboratory. Deciding record: USPTO file wrappers, identity of the "outside inquiry", the O/170/09 text.

My weighing. From the record alone, 4 and 1 remain open, and 2 is plausible for details; 3 and 5 explain parts of the trail but not the lack of a sale or an independent run. No single explanation is excluded. The cheapest decisive step is one neutral laboratory running (a) a hydrogen-free control and (b) the claimed signatures on a sample it did not prepare; the company has invited exactly that and the first outside group (IIT Kanpur) is the record to watch. The silence is therefore an anomaly not a verdict: it lowers the odds of explanation 1 but cannot be read as proof of 4.

---

Corrected claim list (final)

1. Classical electron shell; closed-form atoms. The derivation is internally inconsistent; the g-factor and spectroscopic predictions are tested and not supported against modern constants; the arithmetic and Dirac-equivalent one-electron results are established as arithmetic. (Split as above.) QE revision: helium excited states (Ch.9) as an "exact solution within experimental error": tested, not supported (mean relative error 1e-3, 61 independent levels, not 104); the free-electron disc and its "KEK confirmation": not supported by the cited experiment; the free electron absorbing a photon ("space drive"): refuted by calculation within accepted conservation laws.
2. Hydrinos (states below n = 1; catalytic release). Mills's derivation: internally inconsistent (QE revision adds: the molecular-hydrino EPR activity rests on a doublet state that contradicts the paired-electron state used for its own energy); physical-proton QM version: refuted by calculation; lab identification of the anomalies: never properly tested; kW energy release: never properly tested; early light-water cell: EarthTech tested, not supported, NASA ambiguous (NASA's authors say the data "marginally admit" the recombination explanation, which needs an efficiency above 0.55). The EPR is the strongest outside datum, and it is semi-quantitative with an ordinary alternative fit. Fast-H Balmer broadening and hydrino-catalysed fusion are carried as distinct, never properly tested sub-claims.
3. Hydrinos as dark matter and cosmic emission. Never properly tested; 3.48 keV assignment not testable as stated; 3.5 keV as decaying dark matter tested, not supported (contested).
4. One method solves molecules and matter. Weak form established as arithmetic (QE revision: not for every printed number; ion vibrations, Table 11.1 distances and rotational energies do not follow from the stated recipe); strong (blind, no-fit) form never properly tested; "no fitted parameters" is internally inconsistent with the case-specific inputs (measured atom energies, rounding, a0 versus aH by species, imported Morse, Moessbauer and Lavrov-Weber formulas, an E_T chosen from a 19-value lattice per group).
5. QM is wrong or unnecessary. Radiation argument refuted by calculation; Bell/Aspect "classical" internally inconsistent and tested, not supported; interpretation not testable as stated; helium-mobility peaks never properly tested.
6. Spacetime, gravity, cosmology, particles. Cosmology numbers and top/Higgs/nucleon moments tested, not supported; muon/electron ratio never properly tested; absolute masses not testable as stated; 1995 predictions never properly tested (edition not obtained). QE revision (GR half now assessed): the Schwarzschild metric is GR's own, with its reciprocal radial factor assumed (perihelion and deflection tables therefore do not separate Mills from GR); "conservative versus not" is refuted by calculation; "no far-field gravitational waves" and "the Higgs will not be observed at the LHC" are tested, not supported; the "absolute space confirmed experimentally" argument is tested, not supported; speed of gravity = c agrees with GW170817; free-electron zero gravitational mass never properly tested; pseudoelectron fifth force not testable as stated.
7. Collective phenomena are classical. QE revision (deep-read, no longer provisional; verdict changed from the single "never properly tested"). Superconducting Tc formula: tested, not supported (fails on Cu, Ag, Au, Na, Al; chosen dimension and density); band-pass step: internally inconsistent; quantum Hall: reproduces h/(n e^2) by construction, no plateau positions, no fractional selection rule (not testable as stated); Aharonov-Bohm: internally inconsistent (units); alpha "derived": internally inconsistent (identity); statistics: standard results with h supplied (not testable as a classical derivation). No chapter avoids supplying h and the integers.

What would settle it (one test per disagreement)

Open questions

1. Are the Hagen-Mills EPR lines produced by a hydrino or by an ordinary centre? What would a matched hydrogen-free control show?
2. What do the 13 listed laboratories' own reports say, if they exist (AECL, Westinghouse, MIT Lincoln, INEL, Hokkaido, Brookhaven, Moscow Power Engineering Institute, Lehigh, Harvard CfA)?
3. Will IIT Kanpur's replication be published with a control?
4. Does the 2025 edition change the book's derivations (p <= 126, EPR statistic, alpha-dependence of g)? The project PDF is 2018 text with a 2022 citation.
5. How were the molecular-group parameters fixed: before or after seeing the data? How many ionization-sum candidates were tried for Table 5.2?
6. Is the muon/electron match to 1.6e-8 independent of the target value once the correction factors are derived?
7. What did the November 1995 edition say?
8. What did the "outside inquiry" at the USPTO consist of, and what did the O/170/09 hearing officer weigh as "expert evidence"?
9. Chapters never read: Ch.3-4, 9, 11, 17-21, 23-31, 33-35, first half of Ch.32.

Corrections to earlier documents (for the orchestrator)

1. physicist.md, section 3 (Historical record) and section 11 item 2: "the High Court appeal [2008] EWHC 2763 (Pat), Floyd J, was dismissed and is reported for the standard of proof"; also section 10. This is wrong. Evidence: the judgment text on the National Archives site states the appeal was allowed and the cases remitted to the Hearing Officer with a new test ("whether there is a reasonable prospect that on a full investigation with the benefit of expert evidence" GUTCQM will turn out to be a valid theory, paras 49-52) [web: https://caselaw.nationalarchives.gov.uk/ewhc/pat/2008/2763]. On remittal the Hearing Officer is reported (secondary summary only) to have refused again, O/170/09. The contrarian's "remitted" was right. Propagate to: any primer, entry or card that repeats "dismissed"; search the review files and work/rankings/entries for "dismissed". The wording of sites.md ("appeal not yet read") should be updated to say it was read.
2. claims.md, claim 2: "p up to 137" and "replicated by Bruker". Add: the 2025 Mills-Hagen paper gives p <= 126; the Bruker statement appears twice (book; Hagen-Mills 2022 naming a Billerica MA lab and citing Mills's own preprint), both Mills-co-authored, no Bruker-authored record; the paper's own EPR agreement is "0.004 +/- 0.550 gauss over the 11 lines" (mean absolute 0.43 G [calc]), "semi-quantitative", not the book's 0.097 G; the paper concedes an approximate conventional two-radical fit and an underlying broad signal of unknown origin.
3. claims.md, claim 4: "no fitted parameters". Mark as the book's statement qualified by case-specific factors and experimental inputs listed above (contrarian and mathematician).
4. claims.md, claim 6 / deep reads: H0 comparison. Add that H0 = 4 pi c^3/(G m_U) follows from the chosen m_U (78.27 for 2e54 kg; 2.32e54 kg would give 67.4); T_CMB needs a different mass (2.56e54 kg), giving H0 = 61.1 [calc, mathematician; recomputed].
5. claims.md and all reviewers: "eleven-figure agreement" in g/2. Keep, with the reason: the 1980s alpha limited the test, and the three-term formula has two wrong coefficients that nearly cancel; against modern alpha it is 2.87e-10 away, about 2,200 times the experimental uncertainty (reviewers wrote "about 2,000").
6. Deep read (Ch.42), medium-confidence Aspect numbers. Upgrade to [web, secondary]: delta_QM = 5.8e-2 +/- 0.2e-2 (contrarian). Not opened at the PRL; the Bell verdict does not depend on it.
7. facts.json: where the 3.48 keV match is listed as support for claim 3, add Hitomi (Perseus) and XRISM (ten-cluster stack, limit 3-4 times tighter than Hitomi, still 5 times above the Bulbul rate), and Dessert-Rodd-Safdi (contested by Abazajian) with the note that the line is unconfirmed rather than excluded at the original flux.
8. physicist.md Claim 3 verdict "Tested, not supported" for the 3.48 keV assignment. Replaced by "Not testable as stated" for the assignment and "Tested, not supported (contested)" for the dark-matter-decay interpretation of the line; the physicist's reading of XRISM is not what XRISM shows (it does not reach the original flux).
9. Mathematician's note on Claim 2 (Table 5.2). Retained and quantified: the table is above chance for one candidate per element and ordinary chance for about 16 candidates; the count is not in the book.
10. Digest and deep-read corrections already listed in claims.md (digest .28, .110, .111, .121; truncated "Corrections" bullet of the pdf 1214-1252 deep read) stand; I found nothing that contradicts them. Research Square preprint date: Crossref gives 2021-02-10; the book prints 2020.

---

QE revision (2026-10-09)

Arbiter, after qe.md (verdict MINOR) and the deep reads it ordered. Where this section and the sections above differ, this section governs. Tags as above: [read] project documents and deep-read files; [web] with URL; [memory] with confidence; [calc] recomputed by me or taken from a deep reader's stated calculation (I say which).

0. What was weighed

• QE: qe.md (sections 1-8).
• New deep reads, all in deep/GUT-CP-2020-Ed-Web-b290db.md: Ch.35 Fifth Force (pdf 1680-1703); Ch.9 helium excited states (333-350); Introduction (1-86); Ch.3-4 free electron and photon (201-247); Ch.15 functional groups (801-1082, prose and a few table pages); Ch.31, first half of Ch.32 and Ch.34 (1581-1620, 1666-1679); Ch.24-30 (1521-1580); Ch.11 (445-504).
• I re-read these in full. I did not re-render page images; where a deep reader read an equation or table from a page image I carry that as [read: page image, via deep read] and say so when a verdict turns on it.
• Not yet deep-read by anyone: Ch.17-21, Ch.23, Ch.33, pdf 1661-1665 (start of Ch.34), the numerical tables of Ch.15 (error columns; the deep reader saw four table pages), amino-acid and phosphate sections (not found in pdf 801-1082), the Jansson, Booker and Nansteel reports, Mills's 165-page reply to Rathke.

0.1 Web checks done in this revision (budget 8; 6 used; the earlier 6 were spent on the first pass and are not re-counted)

1. GW150914: signal observed 14 September 2015 by both LIGO detectors, peak strain 1.0e-21, matches the GR waveform for inspiral, merger and ringdown, significance above 5.1 sigma [web: https://arxiv.org/abs/1602.03837]. Retagged from [memory, high]. The 11 February 2016 announcement date remains [memory, high].
2. GW170817 and GRB 170817A: time delay (+1.74 +/- 0.05) s; the difference between the speed of gravity and the speed of light constrained to between -3e-15 and +7e-16 of c [web: https://arxiv.org/abs/1710.05834]. Retagged from [memory, high].
3. Higgs: ATLAS and CMS combined mass measurement 7 and 8 TeV [web: https://arxiv.org/abs/1503.07589, abstract opened; I took the numerical value 125.09 +/- 0.24 GeV from memory, high, because the abstract text I fetched was cut off before the number]. ATLAS ten-year coupling map, Nature 2022 [web: https://arxiv.org/abs/2207.00092]. Mills's 128.75 GeV is +2.8 percent from 125.20; at the 2015 combined uncertainty (0.24) that is about 15 sigma, at the PDG-type 0.11 about 30 sigma [calc].
4. Quasar time dilation: Lewis and Brewer, "Detection of the cosmological time dilation of high-redshift quasars", Nature Astronomy, 3 July 2023 [web: Crossref record, DOI 10.1038/s41550-023-02029-2; title and date only, I did not open the paper]. Hawkins, MNRAS 405 (2010), "On time dilation in quasar light curves" [web: Crossref, DOI 10.1111/j.1365-2966.2010.16581.x]. So the quasar premise on p.1678 (Hawkins, no dilation) has a later published detection of the (1 + z) dilation; the numerical result I take from the deep reader [memory, medium].
5. Witteborn and Fairbank, PRL 19, 1049 (30 October 1967), "Experimental Comparison of the Gravitational Force on Freely Falling Electrons and Metallic Electrons" [web: Crossref, DOI 10.1103/physrevlett.19.1049; title only]. The title itself states the experiment compared free electrons with electrons in the metal. Fairbank, Witteborn and Knight proposed the experiment in Science 144 (1964) [web: Crossref, DOI 10.1126/science.144.3618.562-b]. A Rev. Sci. Instrum. 1977 apparatus paper by the same authors exists [web: Crossref, DOI 10.1063/1.1134860]. I did not open the PRL, so the "0.09 g" figure is the book's own words [read: p.1680, 1686] and the Schiff-Barnhill physics stays [memory, medium].
6. Fast-H literature: Loureiro and Amorim, "Anomalous Doppler broadening caused by exothermic reactions: application to hydrogen Balmer lines", Eur. Phys. J. D (published 28 June 2011 per Crossref) [web: DOI 10.1140/epjd/e2011-20028-1; title only, not read]. The Investigator's date (2010) should read 2011 unless an earlier online date exists. Rozema et al., PRL 109, 100404 (2012), title "Violation of Heisenberg's Measurement-Disturbance Relationship by Weak Measurements" [web: Crossref, DOI 10.1103/physrevlett.109.100404]. The title concerns the measurement-disturbance relation, which is not the Kennard-Robertson preparation uncertainty relation; what that means for the book's "QM is proven wrong" is a Physicist question and I do not decide it here [memory, medium-high for the distinction].

Not checked, still [memory]: Schiff-Barnhill physics, Hulse-Taylor decay rate, Taylor-Weisberg, SN 1997ex redshift, ejecta speeds, Cassini gamma bound, helioseismic J2, the "26.3 ppm" absolute shielding of gaseous H2, Ramsey 1950 quadrupole convention, Loch-Stengler-Werth g value, Ritter/Ishida positronium hyperfine values, MgB2/hydride Tc values. None of these is the sole support for a verdict below; I mark each use.

1. Changes at a glance (verdict table)

2. Claim 1 additions (Ch.3-4, Ch.9)

Ch.9 helium excited states [read: deep read Ch.9; recomputed there with numpy, which I did not repeat]. - Facts the deep reader established: the energies come from E = -13.6038/(n r2) eV with four closed forms for r2 (singlet or triplet, l = 0 or l != 0). No continuous parameter is fitted. About six discrete choices are fixed once (charge e/n for the outer electron, the 2/3 factor, s = 1/2, the doubling for triplets, the sign and orientation choice for l != 0, Z = 2) plus the derived constant a_He. The book does not say whether the factor 2 and the orientation sign were fixed before or after comparison with NIST. - "Over 100 states" is 104 table rows but 61 independent (n, l, spin) values from 4 formulas [read: deep read section 3]. - Mean absolute relative error 1.0e-3 (about three figures). The n = 2 triplet S row is 13 meV off. This does not support "five significant figures" or "within experimental error" [read: p.345 quoted by the deep read]. For l >= 3 the rows equal the hydrogenic -13.6/n^2 (errors 1e-4 to 1e-6), so agreement there is not a test. - R^2 = 0.9999936 is reproduced by the deep reader; a parameter-free -13.6038/n^2 already gives R^2 = 0.967 and one global defect 0.982, so the metric hardly discriminates. A one-number-per-series quantum-defect fit beats the book's formulas in 11 of 12 series, the exception being triplet S. - In favour, and kept: the fixed factors give S-series defects 0.1443 (singlet) and 0.2887 (triplet) against fitted values about 0.145 and 0.305 [calc, deep reader]. That is a non-trivial hit for a zero-fit formula. It is also what a post-hoc choice of the factor 2 would produce, and the book is silent on the order. - The M8 point in qe.md (high-n agreement is cheap) is confirmed by these numbers. - Verdicts. Exact solution agreeing with experiment to measurement error: Tested, not supported. Zero-parameter scheme reproducing the helium quantum defects: Never properly tested (on atoms and ions it was not built on). The sidebar "hydrino states are strongly supported by the helium calculation": Not testable as stated as evidence, because the chapter contains no state with a central field of magnitude n (below ground); it is an analogy [read: p.346]. The sidebar is also an interested party's characterisation (author of the theory and founder of the company).

Ch.3-4 free electron and photon [read: deep read Ch.3-4]. - The free electron is a disc of radius rho0 = hbar/(m v) with density proportional to sqrt(1 - (rho/rho0)^2). The book says this is "the projection of the atomic orbital into a plane". The deep reader's argument: a uniform spherical shell projected onto its equatorial plane has density proportional to 1/sqrt(1 - (rho/rho0)^2) (rising to the rim), the reciprocal of the book's profile. Equation 3.3 was garbled in the text layer, so this is the deep reader's reasoning, not a page-image check. It is simple enough that I accept it [calc, deep reader]; the profile is effectively stipulated. - A free electron at rest has infinite size and zero angular velocity yet carries L = hbar and one Bohr magneton; the size depends on velocity relative to the proton's frame (footnote 2). The book accepts both. - KEK "confirmation". The cited paper (Levine et al., PRL 78, 424, 1997, TOPAZ detector at TRISTAN, KEK) measures how the electromagnetic coupling strengthens with momentum transfer, by muon-pair production [web, via the deep reader's search snippets: https://link.aps.org/doi/10.1103/PhysRevLett.78.424; I did not open the paper]. Its published reading is standard QED vacuum polarisation [memory, medium]. The book's sentence is a reinterpretation (charge rises toward the core), with no alpha(Q) value, no figure and no derivation from Eq. 3.8. Verdict on the sentence "recently confirmed experimentally": Tested, not supported (the experiment's subject differs, and no quantitative comparison is offered). The interested-party reading is the author's. A real test would be a disc-model prediction of alpha(Q) against LEP and TRISTAN data. - Space drive. The book states that a 5 GHz photon (momentum 1.1e-32 kg m/s) gives a free electron 1.73e-27 kg m/s, five orders of magnitude more, with "spacetime" conserving momentum [read: p.245-246]. I reproduce both numbers from the printed formula [calc, deep reader]. Energy and momentum cannot both be conserved when a free electron absorbs a single photon [memory, high]. Refuted by calculation within accepted conservation laws; the book's remedy (spacetime as momentum reservoir) is Not testable as stated. The book itself adds "Photons and free electrons each have zero gravitational mass". - h put in by hand (the QE question): the photon angular momentum is set equal to hbar as a boundary condition, then E = hbar omega is read off; the photoelectric and Compton derivations are the textbook ones with h nu and h/lambda supplied; the Compton calculation does not use the book's photon or disc at all [read: deep read section 5]. The "classical Doppler misconception" it opposes is a partial straw man (standard classical scattering is Thomson scattering, which gives no shift) [memory, medium]. - Stern-Gerlach. The book explains g = 2 by adding the Larmor photon's hbar/2 to the electron's hbar/2 projection. No inhomogeneous-field force, no beam deflection and no splitting is calculated; the free-electron application meets the Bohr-Mott objection that the book does not address [memory, medium]. Not testable as stated ("completely predictable from first principles" is not backed by a calculation in these pages). - g/2 correction (applies to claim 1 and 7). p.1578 (positronium chapter) and p.60 (Introduction) quote the same pair: calculated g/2 = 1.001 159 652 137, experiment [Van Dyck, Schwinberg, Dehmelt 1987] 1.001 159 652 188(4). This is the free-electron g, not a positronium measurement. The difference -5.1e-11 is 12.75 sigma at the printed uncertainty, 11.9 sigma at the published 4.3e-12 [calc, deep reader]. Against the 2023 Northwestern/Harvard value 1.001 159 652 180 59(13) [web: https://arxiv.org/abs/2209.13084] it is -4.36e-11, 335 sigma. With present-day alpha the formula gives 1.00115965247, +2.9e-10, about 2,200 sigma (my earlier recomputation, unchanged). The value ...137 corresponds to alpha^-1 = 137.0360387, not the printed 137.03603 (deep reader). Claim 1's g verdict is unchanged.

3. Claim 2 additions

• Dihydrino derivation adds postulates beyond Ch.5 [read: deep read Ch.11, section 9]. The hydrino evidence in Ch.16 (p^2 scaling of bond, vibration, rotation; NMR; EPR) uses: (i) a two-centre integer field p, asserted not derived; (ii) the same field multiplying the proton-proton repulsion (needed for exact p^2 scaling; with Vp unscaled the p = 2 total energy would be -165.2 eV instead of -126.71 eV [calc, deep reader]); (iii) the Doppler term divided by p^2, while the anharmonicity equations and Ch.16 keep a p^3 Doppler term (bond energy at p = 4: 71.64 or 87.31 eV depending on which form; at p = 2, 17.91 or 19.22 eV [calc, deep reader]); (iv) for EPR activity, a doublet state (half of each electron's current paired) imported from Ch.16, which contradicts the paired-electron force balance (Eq. 11.200) that yields the same H2(1/p) energies, sizes and NMR shift; (v) the Raman and rotational "signatures" use measured H2 constants times p^2 [read: Ch.5 Eqs. 5.21-5.22 and Ch.16 Table 16.52 per the deep read]. The NMR photon term reuses a Ch.5 postulate.
• Consequence for the verdict: the molecular-hydrino EPR and Raman signatures are not independent calculations; "predicted lines" for H2(1/4) are rescaled measured H2 lines. That is one more reason the Raman and EPR matches are weak as tests of the derivation, although they remain observations that need an ordinary explanation (kept in the record).
• Mills's derivation: Internally inconsistent (verdict unchanged, one more instance). Standard QM: a two-electron system has total spin 0 or 1, never 1/2 [memory, high].
• The Introduction's evidence statements [read: deep read Introduction].
• The headline "over 100 published journal articles and over 50 independent test reports and articles" cites only ref [22], the company website. Within refs [22]-[42], Mills is an author of 19 of 21; ref [29] (Bykanov, "Validation ...", file name GEN3_Harvard.pdf) is a web document on the company's server; ref [30] is the Hagen-Mills joint paper; no laboratory is named and no test-report author is named anywhere in pp.1-86. No negative result or failed replication is mentioned.
• This is the company's and inventor's own characterization (M5 in qe.md), to be set beside the outside record: the Investigator found NASA 1996, EarthTech 1997, Hagen 2022 and the IIT Kanpur announcement. The Introduction's affiliations for refs 37, 41 and 42 (Scharer, Phillips, Chen, Mayo, Nansteel) are [memory, medium-low] from the deep reader and unverified; all are co-authors with Mills.
• Hydrino statements in the Introduction give no power figure, run condition or spectrum. The NMR "definitive test" paragraph gives the ordinary-H2 shift -28.0 ppm and no hydrino-gas shift. Isolation: "Hydrinos have been isolated in the laboratory" (p.83) sits against Mills's reported 2014 Scientific American remark that they had not been isolated (M10; tertiary pointer in research/2026-10-09-wikipedia.md, original not read). Both statements are kept in the record.
• The celestial claims (white dwarfs, 11-16 nm background, diffuse H-alpha) cite only Mills's own refs [23-29, 31]; the third-party dark-matter papers (Bournaud, Jee) are cited as "matching", and whether those authors say anything about hydrinos is [memory, medium] "no" and unchecked.
• The file's title page is "December 10, 2018 Edition" but its Introduction cites a 2022 paper and ref [31] is "submitted for publication". Date any claim by its reference, not by the edition line.
• Fast-H / Balmer broadening (M7). Seven papers (refs 36-42, 2005-2010), all with Mills as an author; Scharer (ref 37), Phillips and Chen (ref 41), Mayo and Nansteel (ref 42) are co-authors with possible non-company affiliations [memory, medium-low]. A mainstream rival explanation exists: Loureiro and Amorim, "Anomalous Doppler broadening caused by exothermic reactions: application to hydrogen Balmer lines", Eur. Phys. J. D (2011) [web: DOI 10.1140/epjd/e2011-20028-1; title only]. I have read neither side. Never properly tested as a hydrino signature (an ordinary exothermic-reaction mechanism is proposed in the literature; no neutral controlled comparison is on record). Settling record: the Loureiro-Amorim paper and the Mills-Phillips papers read in full, and a broadening measurement with and without the catalyst by a laboratory with no stake.
• Hydrino-catalysed fusion (M4). pp.272-274 and Table I.7: deuterium or tritium hydrinos fuse more readily; silver-shot detonations cited as comparable to nuclear weapons in power density [read: digest p.272-274, via qe.md]. The observable it implies (neutrons, tritium, 3He above background) is not reported in the pages I have. Never properly tested. The NASA TM itself names fusion of shrunken atoms as one hypothesis [read: NASA TM p.1, via qe.md].
• NASA TM-107167 nuance (M6), added to the record. The authors state that explaining all the excess by recombination "requires a recombination efficiency of quite a high magnitude" (above 0.55, rising to about 1 at 5 A), which the data "marginally admit"; they also quote the loaned cell's maker for "50 W steady excess heat for a continuous period exceeding hundreds of days" (the company's claim, quoted by NASA) [read: research/files/NASA-TM-Niedra-1996 p.1, p.4, via qe.md]. This leans slightly toward the claim compared with "could not exclude recombination", and the verdict wording above is amended: NASA 1996 remains ambiguous, with the authors' own weighing stated.

4. Claim 4 additions (Ch.11 and Ch.15)

Ch.11 (H2, H2+, D2, D2+, NMR, quadrupole) [read: deep read Ch.11; numbers recomputed there with numpy]. - Reproduced from the printed formulas: H2+ and H2 raw energies (-16.2802, -31.6351 eV); H2 vibration 0.5176 eV (book 0.517) and D2 0.3713 (0.371); H2 and D2 anharmonicity (120.4 and 60.93 against 121.33 and 61.82 cm^-1); NMR -28.01 ppm; the quadrupole integral as printed. - Not reproduced or not derived in the text: - H2+ and D2+ vibration: the stated recipe gives 0.280 and 0.200 eV, the book prints 0.270 and 0.193. The printed values are closer to experiment (0.271 and 0.196) than the recipe's. Source of the printed numbers unknown. - Table 11.1 "calculated" internuclear distances 0.7411 and 1.0577 A and rotational energies 0.01511 and 0.00742 eV are not computed anywhere in the text; the derived values are 0.7484 A (+0.94 percent against 0.7414 A [memory, high]) and 0.0148 eV; the table cites Ch.12 equation numbers that do not match. - H2 bond energy 4.478 eV: the exact match depends on writing 2 x 13.59844 = 27.19688 eV as 27.20 (27.19688 gives 4.4809, +0.06 percent), and on using the infinite-mass a0 for H2 while using the reduced-mass aH for H2+ (a consistent treatment gives 4.461-4.464 eV, -0.3 percent). Without the "Doppler" term the answer is 4.435 eV (-0.96 percent). The D2 value 4.556 is a 1-2 meV arithmetic slip against the book's own 4.5549. - Quadrupole moment: the integral (Eq. 11.421) omits the surface element; Mills's own surface density integrated with it gives 1.3333 a0^2 = 0.373e-16 cm^2, not 1.70127 a0^2 = 0.476e-16 [calc, deep reader]. Both lie inside Ramsey's 0.38 +/- 0.15; the corrected one is closer. What quantity and convention Ramsey's 1950 value refers to is [memory, medium] and unchecked. - NMR "experiment": -28.0 ppm is Mills's composite of -28.5 (gas-phase TMS absolute) plus 0.48 (H2 relative), from two papers by other authors; the same paragraph uses a second TMS value (-31.5 ppm). A commonly cited absolute shielding of gaseous H2 near 26.3 ppm [memory, medium] would be 6 percent from the prediction. Unresolved; web check due. - p^2 versus p^3 Doppler scaling is inconsistent inside the book (see claim 2). - Effect on verdicts. "No fitted parameters" stays Internally inconsistent and is better supported: no continuous parameter is fitted in Ch.11, but measured atom energies (27.20 eV), rounding, species-dependent a0/aH choice, imported Morse (Karplus-Porter), Moessbauer (Gibb) and Lavrov-Weber formulas, and Table 11.1 entries not computed in the text all enter. The weak form's "Established as arithmetic" is qualified accordingly.

Ch.15 functional groups [read: deep read Ch.15, prose of pdf 801-1082; four table pages only]. - Each group's E_T(atom-atom) is picked from a lattice of 19 values built from 0.72457, 0.92918 and 1.13379 eV and their sums, with a verbal "energy match" reason. About 85 groups have a freshly specified choice; 22 hybridisation numbers are rule-computed (a ratio of energies times 0.91771), so those are not free. - CCl4, CBr4, CH2I2 and CHI3 each have their own group with their own E_T (0.92918, 0.36229, 0.18114, 0). The iodide groups (ii) and (iii) each serve exactly one compared molecule; a lattice-chosen constant used once reproduces that one total by construction. The text never says a value was adjusted until the total matched, and it never states a rule fixing E_T before the experimental total is seen. I therefore describe it as "selected from a small lattice with verbal justification", neither fitted nor derived; the author may have followed an unrecorded rule. - The same "equivalent" Ca-Cd group has E_T = 0 in guanine and 0.92918 in cytosine; cytosine's total is 1.7 percent off, against 0.03-0.08 percent for adenine and thymine and 0.006-0.44 percent for the seven iodides. The three nitroaniline isomers use the same E_osc, "for simplicity". Pyridine takes benzene's vibration energies. - The deep reader lists many source slips (wrong table cross-references, mislabelled ortho/meta, "chloroform with four C-Cl bonds"; section 7 of that read); the deep reader found no arithmetic error in the numbers it checked (115.70989 eV for sulfur, benzene totals, several c2 ratios). - Amino acids and phosphates are not in pdf 801-1082; the earlier 1-3 percent amino-acid error pattern comes from summary tables elsewhere and is not re-checked. - No held-out test appears. Strong form: Never properly tested (unchanged). Counting: about 14 single-molecule heterocycle and base classes, each carrying 3-11 group types against one experimental total.

5. Claim 5 note

Compton and photoelectric derivations, the Stern-Gerlach treatment and the "HUP experimentally disproved" citation (Rozema 2012, Isinger 2017, both about measurement or timing, not the preparation uncertainty relation [web title; memory, medium-high for the distinction]) are Physicist territory; I record the verdicts in section 2 and do not change the claim 5 verdicts above. The HUP statement stays Not testable as stated as an inference to "QM proven wrong" until the Physicist weighs the cited papers.

6. Claim 6: the GR half, gravitational waves, Higgs, fifth force

Where the metric comes from [read: deep read Ch.31-34, section 1, with the deep reader's SymPy check; I did not repeat it]. - The wave equation contributes one statement: a change in the gravitational field spreads as a front f(t - r/c) at speed c. The metric then comes from (a) asserting that g_rr is the reciprocal of g_tt (Eqs. 32.11-32.12, "therefore" does not follow), (b) a light-clock argument with the Newtonian escape velocity v_g = sqrt(2GM/r) in place of the relative velocity, which gives g_tt = 1 - 2GM/(c^2 r), and (c) superposing particles. The result (Eq. 32.38) is the textbook Schwarzschild metric. The escape-velocity route to g_tt is the known heuristic that textbooks treat as a coincidence of these coordinates [memory, medium: Gruber et al. 1988, unchecked]. The reciprocal in (a) is exactly what GR's field equations supply, via Birkhoff's theorem, and it carries half of the light deflection and the gamma-dependent part of the perihelion advance. - Verdict: Schwarzschild metric "derived from the wave equation": Internally inconsistent as a derivation (the decisive step is asserted, and the stated route does not yield it). The metric itself is GR's. - "Conservative versus not conservative", and avoiding "singularity issues": substituting R = r + GM/c^2 turns Fock's Eq. 32.41 into Eq. 32.38; both have zero Ricci tensor and the same coordinate singularity [calc, deep reader, SymPy]. Refuted by calculation. - Perihelion and deflection tables. The "theoretical" columns are GR values: Mercury 43.03 against 42.98 recomputed from GR by the deep reader; Venus 8.6, Earth 3.8, Icarus 10.3 (GR recomputed 10.06); light grazing the Sun 1.75 arcsec [calc, deep reader]. The observed columns are 1970s compilations (Fomalont-Sramek 1.761 +/- 0.010 arcsec, a ratio of 1.006 +/- 0.006 to GR [calc, deep reader]; the name is misspelled "Fomaleont"). Established as GR's results. They cannot distinguish Mills's theory from GR, and the book's claim that they show the metric "derived from Maxwell's equations is the correct theory of gravitation" does not follow. The solar J2 paragraph (Dicke-Goldenberg 5e-5, "recent" 1.84e-6) has not been updated since helioseismology [memory, medium]. - Photon has zero gravitational mass. Same deflection number as GR (which uses test-body independence of mass, not zero gravitational mass); GR has light's energy sourcing gravity [memory, high]. Never properly tested by the deflection data; the Physicist should weigh Eotvos-type tests with electromagnetic-energy fractions [memory, medium]. - Speed of gravity (p.1592): rests on a January 2003 New Scientist item about the 8 September 2002 Jupiter-quasar measurement, c_g/c = 1.06 +/- 0.21 [memory, medium], disputed at the time as a speed-of-light measurement [memory, medium]. The best test is GW170817: speed difference between -3e-15 and +7e-16 of c [web: https://arxiv.org/abs/1710.05834]. Established (agreement), made with the waves the book says do not exist.

Gravitational waves (M1). The book says gravity cannot form a transverse light-speed wave, "consistent with the absence of the direct experimental observation of gravity waves [128,129]" (p.1642) and "no transverse time-dependent gravity wave effect in the far field consistent with the absence of the observation of gravity waves [32-33]" (p.1678); the references are LIGO 2009 and Shawhan 2010 [read: deep read Ch.31-34, section 5]. p.1642 also cites 2017 papers (refs 134-136), so the page was edited after 2017 and this sentence was left. GW150914, 14 September 2015, observed by both detectors, 5.1 sigma or more, matches the GR waveform [web: https://arxiv.org/abs/1602.03837]; GW170817 followed in 2017 [web: https://arxiv.org/abs/1710.05834]. The Hulse-Taylor orbital decay (1974 onward) was available as indirect evidence [memory, high]. The prediction was specific and testable, the test was done, and it went against the book. Tested, not supported. Open: whether the 2025 edition (research/files/GUT-CP-online-2026-04-28.pdf) changed the passage; not yet read.

Higgs (M2). p.1678: "Such a mass conveying particle will not be observed in experiments performed at Fermilab, the LHC, or any future collider [31]" (ref 31 = Cho, Science 2010); the next sentence, added after the 4 July 2012 CERN announcement, reinterprets the 126 GeV excess as a "high-energy neutron resonance" predicted at 128.75 GeV (Eq. 37.47, p.1720) [read: deep read section 6]. The 2018 text keeps both. A boson at 125.09 +/- 0.24 GeV was reported by ATLAS and CMS combined [web: https://arxiv.org/abs/1503.07589; numerical value memory, high], and its couplings have been measured [web: https://arxiv.org/abs/2207.00092]. "Will not be observed": Tested, not supported. The 128.75 GeV neutron resonance: Tested, not supported on mass (+2.8 percent, about 15 to 30 sigma depending on the uncertainty used); its decay pattern is not checked here. The reinterpretation after the fact lowers the weight of the book's "predicted in 1995" statements; the 1995 edition is still not obtained, so those stay Never properly tested. Mills's remark that a short-lived resonance is "not a real particle" would equally apply to the Z, W and top and is an interpretation, not a measurable difference.

Absolute space "confirmed experimentally" (p.1678) [read: deep read section 7]. - Quasars show no time dilation (Hawkins 1993-2010): a later paper reports detection of the (1 + z) dilation (Lewis and Brewer, Nature Astronomy 2023) [web: DOI 10.1038/s41550-023-02029-2, title and date]. Tested, not supported. - SN 1997ex dilation attributed to ejecta moving "close to light speed": Type Ia ejecta move at about 0.03-0.07 c [memory, high]; the Lorentz factor at 0.07c is 1.0025 [calc]; the observed dilation matches 1 + z = 1.36 at z = 0.361 [memory, medium]. Refuted by calculation for the mechanism offered. - Twin paradox and Hafele-Keating are predicted by standard special relativity plus gravitational redshift and do not distinguish absolute space [memory, high].

Fifth force, pseudoelectrons, free-electron gravity (M3) [read: deep read Ch.35]. Interested party: the book and the patents are the author's and the company's. - Claims, quoted: a free electron has inertial but "not gravitational mass" (zero in the limit of absolutely free, spin cancelled, fields excluded); a negatively curved "pseudoelectron" has negative gravitational mass and feels a repulsive "fifth force" from a gravitating body, "with far greater magnitude" [read: p.1680, 1686]. - Experimental basis offered: Witteborn and Fairbank PRL 19, 1049 (1967). The book itself reports "less than 0.09 m_e", an upper limit, and reads it as supporting zero [read: p.1680, 1686]. Earlier statements that the experiment "measured zero" should read "upper limit, less than 0.09 m_e". The book rejects the Schiff-Barnhill shielding explanation with a comparison of m_e g h (about 5.6e-21 eV [calc, deep reader]) against a 5 eV binding energy, which the deep reader judges does not answer the effect (the physical argument concerns the lattice field, not binding energy) [memory, medium]. The 2013 CERN Courier quote (Amsler, March 2014) the book uses as support describes the shielding field as the reason drop tests give g consistent with zero [read: p.1686, 1702]. The paper's context was not fetched. - A neutral test of free-electron gravity would need shielding eliminated or independently calibrated. None is on record in the project files. Free electron zero gravitational mass: Never properly tested. I note that the controlling explanation in the mainstream is the shielding effect, and that I have not opened the primary papers. - Pseudoelectron and threshold. The chapter gives no number for the repulsive acceleration, the curvature or the negative gravitational mass. The threshold photon energy 44.2 eV is correct as evaluated from Eq. 35.96 [calc, deep reader]. The condition on the massive body is GM/(Rc^2) >= 8.65e-5, i.e. M/R >= about 1.17e23 kg/m [calc, deep reader]; the book prints 9.39e17 kg/m twice, once as Earth's ratio ("there is no solution using Earth") and once as the "reiterative solution", which cannot both hold. The gravitational mass in the inequality is taken as the magnitude of the inertial mass although Eq. 35.24 says it is negative and curvature-dependent. Several equation cross-references are dangling. Threshold calculation: Internally inconsistent. The fifth-force effect as stated: Not testable as stated (no predicted quantity, and the book's own conditions exclude Earth). The proposed laboratory route (H3+ collisions with spin-cancelled electrons) needs a gravitating body with M/R above the threshold, which the chapter does not show a laboratory can supply. Black-hole jets and the IC 5063 fast H2 (Tadhunter et al., Nature 2014) are reassigned to pseudoelectrons with no quantitative prediction: Not testable as stated; mainstream jet models are not discussed in the chapter. - Patents: "antigravitational force" (1989, 1994), "Fifth-force apparatus and method for propulsion" (2007), and a 2025 "Reactionless propulsion" paper [read: research/2026-10-09-google-patents.md, via qe.md]. These are the inventor's own descriptions of what the principle is for. A patent is not a test. - Kept in the record as a point of interest for the contrarian reader: the claim is distinct and testable in principle, and rests on a real, if disputed, experiment (Witteborn-Fairbank).

Other Ch.31-35 items (record only): critique of GR's equivalence principle and cosmological constant ("all cosmological solutions predict a decelerating Universe" holds only after setting Lambda = 0, which the book assumes); Lieu-Hillman and Ragazzoni phase-coherence papers used against the big bang; historical assertions (Poincare 1904, Hilbert before Einstein); hydrino-adjacent "fast H" in Ch.35; Q = c^3/(4 pi G) = 3.22e34 kg/s [calc 3.21e34].

Claim 6 revised final verdict (split, replaces the paragraph in the original claim 6 section). - Cosmology numbers (H0, T_CMB, CMB peaks, B-mode, density): Tested, not supported (unchanged). - Top quark, nucleon moments, deuteron: Tested, not supported (unchanged). Higgs: Tested, not supported (prediction of non-observation and the 128.75 GeV replacement). - Gravitational waves in the far field: Tested, not supported. - Schwarzschild metric as derived from the wave equation: Internally inconsistent; "conservative versus not": Refuted by calculation; perihelion and deflection numbers: Established (as GR; no discrimination); speed of gravity: Established. - Absolute space "confirmed experimentally": Tested, not supported (quasars) and Refuted by calculation (supernova mechanism). - Photon zero gravitational mass: Never properly tested. Free-electron gravitational mass zero: Never properly tested. Pseudoelectron / fifth force: Not testable as stated; its threshold arithmetic: Internally inconsistent. - Muon/electron ratio: Never properly tested. Absolute lepton masses: Not testable as stated. 1995 predictions: Never properly tested (primary edition not obtained). "Not assessed" for the GR half is withdrawn.

7. Claim 7 revised (deep read of Ch.24-30)

All of the earlier "provisional, digest only" wording is withdrawn. Inputs the deep reader lists for each chapter [read: deep read Ch.24-30]:

Does any derivation avoid supplying h and the integers? No [read: deep read section 8, item 7]. hbar is declared an axiom (p.1566); integers n, j, n_1, n_2 and f are postulated.

Claim 7 final verdict (split, replaces the provisional one). - Standard statistical results: Established (textbook); their classical derivation: Not testable as stated. - Superconducting Tc formula as a predictor: Tested, not supported; band-pass step: Internally inconsistent. - Quantum Hall (integral and fractional): Not testable as stated (reproduces h/(n e^2) by construction, no selection rule). - Aharonov-Bohm: Internally inconsistent (units). - Fine-structure constant "derived": Internally inconsistent (identity). Positronium hyperfine: Never properly tested. - The earlier "factor 25" in the dimension choice is a factor of exactly 27 (Tc goes as 1/f^3 at fixed n); the printed 7 K is a rounded 6.64 K.

8. Absence-of-evidence addendum (items from QE section 3, M5 and M10)

• The company's and inventor's count, "over 100 published journal articles and over 50 independent test reports and articles", cites only the company website (ref [22]). Beside it: the evidence refs [23]-[42] are almost entirely Mills-authored (19 of 21), the one single-author validation (Bykanov) is hosted on the company server, and the independent outside datum with a declared absence of company funding is the Hagen EPR measurement on a company-supplied sample. These two statements go together; neither is offered as the neutral fact.
• Scientific American 2014: Mills said hydrinos had not been isolated [read: research/2026-10-09-wikipedia.md, tertiary pointer; original not read]. Book: "Hydrinos have been isolated in the laboratory" [read: p.83].
• The Group Director's declaration that the application "appeared to involve cold fusion and perpetual motion" [read: research/pages/wiki-blp.txt, citing Rimmer 2011; Rimmer and Bob Park 2002 are named leads, not read]. This is the institutional party's characterization.
• Harvard Crimson 2000: "over $25 million" from investors; a board with Shelby Brewer (former NRC official), Vice Admiral Kalleres and Aris Melissaratos; Mills's claim of Navy research trials; critics Georgi and Chu; law-firm letters to four physicists [read: research/2026-10-09-press-crimson-crossref.md, via qe.md]. The Navy-trials claim has no document in the project folder; it is the company's statement as reported by a student newspaper.
• Jansson 1997 Rowan thesis, done under a HydroCatalysis service contract (the earliest academic calorimetry for the company) [read: research/2026-10-09-rowan-digital-works.md]. Mills's patent family (115 members) includes pre-hydrino medical patents and the 1989 "Energy emission process"; NJEDA grant of about $1.1M (2013) [read: research/2026-10-09-google-patents.md, 2026-10-09-wikipedia.md, via qe.md].
• Hagen-Mills 2024/26 broad-line EPR, Selke 2025 and Phillips/Loureiro-Amorim: already in the Investigator section above; the primer should carry them.
• The weighing in the absence-of-evidence analysis above is unchanged: no explanation is excluded; the cheapest decisive step is still one neutral laboratory running a hydrogen-free control and the claimed signatures on a sample it did not prepare.

9. Retention and process notes

• No evidence for Mills was dropped by a reviewer's negative judgement in this revision. The items listed in QE section 5 are restored above: the NASA recombination-efficiency nuance and "50 W" quotation (section 3), the fast-H class (section 3), the 2014 Scientific American remark (section 8), the Maris point on helium mobility peaks (contrarian: standard bubble models fail on the mobility peaks; carry into primer row 5d), and the patent record (section 8).
• Against-evidence that was dropped and is now carried: gravitational waves, the Higgs prediction, the Scientific American remark.
• A8 fixed in place: the USPTO sentence is quoted in full (see the edit in the Investigator's findings and absence #5 above).
• Biography tag: the book's own statement of Mills's training ("MD from Harvard Medical School", chemistry degree at Franklin and Marshall, electrical engineering study at MIT) is [read: p.1875, also p.439, 1518] and is the publisher's description, an interested party's characterization. This is a primer retag, not an arbiter verdict.
• Medium-confidence [memory] facts that a verdict above depends on, and their check status: GW150914 and GW170817 [web]; Higgs mass [web, number from memory]; Lewis-Brewer [web, title only, result memory medium]; Witteborn-Fairbank [web, title only]; Schiff-Barnhill physics [memory, medium, unchecked and not decisive]; SN 1997ex [memory, medium, unchecked; the arithmetic of the Lorentz factor is mine]; MgB2 and hydride Tc values [memory, high]; positronium lifetimes [memory, high]. Web budget for this revision: 6 of 8.

10. Corrections to earlier documents (QE revision, for the orchestrator)

1. arbiter.md, claim 7, "positronium g/2". Wrong label: p.1578's g/2 is the free-electron g re-quoted from Ch.1 (Van Dyck et al. 1987). Fixed in place here. Propagate to contrarian.md claim 7(a), claims.md, facts.json (the pdf 1521-1580 entry), and any card.
2. arbiter.md, claim 7, "factor 25". Exactly 27 at fixed n.
3. arbiter.md, claim 6, "GR part not assessed". Withdrawn; see section 6. Propagate to primer.md evidence table row 6f ("Not read / Not assessed"), physicist.md claim 6 note, facts.json Ch.32 how_read.
4. facts.json and claims.md, claim 6, "Gravity and the Schwarzschild metric come from the classical wave equation". Keep as Mills's claim and add: the wave equation contributes only the light-speed front; g_rr = 1/g_tt is asserted; the result is GR's metric.
5. claims.md and facts.json, Ch.9. "Over 100 states" is 104 rows, 61 independent; "agreement to five significant figures" / "within experimental error" should carry the table's own mean relative error 1.0e-3.
6. facts.json, Ch.3-4. "Confirmed experimentally [at KEK]" is the book's wording; the cited PRL 78, 424 is a measurement of the running coupling, read as vacuum polarisation.
7. facts.json, Ch.25-29. "Aharonov-Bohm via flux linkage" should read: ring magnetoresistance plus a damped oscillation of assumed period h/2e, Eq. 27.8 dimensionally inconsistent. "Fine-structure constant derived" should read "identified". "Positronium levels E_n = -6.795/n^2" should note that 6.795 eV is 0.11 percent below the book's own formula (6.8028 eV).
8. Any file saying Witteborn-Fairbank "measured zero". Should read: upper limit "less than 0.09 m_e" in the book's words.
9. Any file giving the Ch.35 Earth gravitational potential as "3.56e4 eV". The book prints -3.56 x 10^-4 eV (the text layer dropped the minus sign). Any file giving the required M/R as 9.39e17 kg/m should carry the flag: that is Earth's ratio; the inequality needs about 1.17e23 kg/m.
10. claims.md, claim 4. "No fitted parameters": accurate only in the narrow sense that no continuous parameter is fitted; add the Ch.11 and Ch.15 inputs listed in section 4. Also the NMR -28.0 ppm "experiment" is Mills's composite of two values (-28.5 and +0.48 ppm) with a second TMS reference (-31.5) in the same paragraph.
11. sites.md Leads 4 and physicist.md ("appeal dismissed"). Applied by the orchestrator 2026-10-09 (arbiter correction 1 above and qe.md section 6): the UK appeal was allowed and remitted.
12. Investigator's Loureiro-Amorim date. Crossref lists 28 June 2011 (EPJD); sites.md/arbiter say 2010. Check for an earlier online date before correcting.
13. claims.md "Orders for further deep reads". Mark done: Ch.24-30, Ch.11, Ch.31-34 (to 1679), Ch.35, Ch.9, Ch.15 prose (801-1082), Introduction, Ch.3-4. Open: Ch.17-21, Ch.23, Ch.33 and pdf 1661-1665, Ch.15 numerical tables, amino-acid and phosphate sections.

11. What would settle it: updated rows