Arbiter: Gerry Vassilatos, Secrets of Cold War Technology: Project HAARP and Beyond (Step 3)

Arbiter, 2026-10-08. Inputs read: claims.md, facts.json (via the reviews), physicist.md, mathematician.md, contrarian.md, deep/ (through the reviews' quotations), sites.md "Leads for review" and the research/ notes (through the reviews). Reader: self-taught, undergraduate level. Tags: [read] project files or the book; [web] URL; [memory] with confidence; [calc] my own arithmetic (script [scratch file], run with the project venv).

Source note (repeated from claims.md; this is not a source gap)

The folder holds one work: a popular, secondary, interested account (about 1996; the publisher sells the author's other books). Its text is fully readable (OCR text layer, 162 PDF pages), but it rarely cites sources and its bibliography (pp.154-160) lists patents and articles that nobody in this review read. So every claim below is "what the book claims". The author states his own method: "qualitative and visionary", set against "quantitative science" [web: https://www.aetherforce.energy/redisovery-of-vril-an-interview-with-gerry-vassilatos/, via the Investigator], and he reaches the HAARP thesis "in the absence of all other evidence" and "on logic alone" [read: book pp.134-135 of the text file]. Where a conclusion below rests on [memory] alone I say so. Outside facts come from the Investigator's notes, the deep reads and the web checks listed in section 6.

1. How the reviewers split

All three reviewers agree on the large pattern: the book is accurate on some hardware and history (prototype element counts, Raytheon's ownership of the contractor 1995-98, real patents, real aurora and echo observations, Tesla's own 1,100 A), and fails on the energy and attenuation numbers that its central claims need. The differences are in labels and in a few numbers. I settle each below.

2. Corrected claim list

Final verdicts use the Step 2 scale. Each claim lists: agreement, disagreement, the stronger argument, the final verdict (split where needed), and the single experiment or calculation that would settle what is left.

Claim 1. Tesla's "Radiant Energy" (neutral ether gas released by short unidirectional impulses, not electromagnetic, passes shields) (PDF pp.9-35) [corrected 2026-10-09 after QE: was "book pp."]

Agreement. All three: the book gives no operational definition, no equations for the claim, and the "neutral gas" reading does not match Tesla's own 1901 patents (US 685,957 and 685,958 describe a stream of minute, positively electrified particles from the Sun charging an elevated plate against ground [web via Inv.: https://patents.google.com/patent/US685957A/en]). The pulse-duration effects match real nerve strength-duration behaviour (threshold current I(t) = I_rh (1 + tau_c/t), where I_rh is the long-pulse threshold current and tau_c the chronaxie, about 0.1 to 0.7 ms for motor nerves [memory, medium-high]; with tau_c = 300 microseconds the threshold is 4x I_rh at 100 microseconds and about 300x at 1 microsecond [calc]).

Disagreement. The shield-penetration verdict. The mathematician says a sealed copper wall stops such pulses. I verified the number: skin depth of copper delta = 1/sqrt(pi f mu0 sigma) is 0.17 mm at 150 kHz, so 1 mm of continuous copper gives exp(-5.9) = 0.0027, about 51 dB [calc]. The physicist and contrarian note that Tesla's "shield" was an arrangement of glass and metal in a room full of sparks, with no instrument inside and no statement that it was closed and grounded.

Stronger argument. The two views are about different things. The mathematician is right that a closed, grounded, thick-enough copper box would stop pulses of this bandwidth. The contrarian is right that Tesla's reported sensations through his actual shields were a first-hand report by a first-class experimenter and no neutral, instrumented repeat is on record. The physicist's conventional explanations (low-frequency magnetic and capacitive leakage through gaps, ionised air, ozone, ultraviolet and x-rays from his tubes) are all plausible and untested against this specific report. Nobody has shown the effect, and nobody has shown it absent in Tesla's arrangement.

Final verdict. (Revised by the QE revision below; the bullets here are the first-pass verdicts, with the revisions given there.) - Pulse-duration and transient physics (short high-voltage pulses give strong, duration-dependent effects): Established. - Penetrating a continuous, closed, grounded copper shield at these pulse widths: Refuted by calculation (conventional skin-depth attenuation, about 51 dB at 1 mm). - What Tesla actually felt through his arrangement of shields, and Dollard and Theroux's reported reproductions (no publication cited): Never properly tested (no instrumented shield test on record; the mainstream moved on by theory, so it was dismissed without a neutral repeat). - A neutral, non-electromagnetic ether gas: Not testable as stated (and at odds with Tesla's own 1901 patent text).

Settle it. A sealed, grounded 1 mm copper box with a calibrated 1 to 6 microsecond impulse source of Tesla's type outside and a field probe plus a detector inside, with the same test using a box with a 1 mm seam. Cheap and not on record.

Claim 2. Colorado Springs "ether engine" (PDF pp.72-93) [corrected 2026-10-09 after QE: was "book pp."]

Agreement. 100 MV and 10,000 hp are Tesla's 1904 Wardenclyffe proposal, not Colorado measurements; Tesla's own 1917 figure for his highest potential is "something like 20,000,000 volts" and for antenna current "from 1,000 to 1,100 amperes. That was in 1899" [web via Inv.: http://www.tfcbooks.com/tesla/1917-05-08.htm]. The correction of Step 1 stands: 1,100 A is Tesla's number, not an invention of the book. The 150,000/s and 0.0062 ms figures have no source; they imply a 0.467 microsecond rest, which the book calls "lost" [calc: 1/150,000 = 6.667 microseconds minus 6.2]. The "stationary versus standing wave" distinction contradicts Tesla's 1900 definition of a stationary wave as one with fixed nodal and ventral regions [web via deep read: https://en.wikisource.org/wiki/The_Problem_of_Increasing_Human_Energy].

Recomputed numbers (the reviewers differed only in framing). - Displacement current I = omega C V with omega = 2 pi x 150 kHz: 1,100 A at 20 MV needs C = 58.4 pF, a sphere of radius 0.52 m with surface field 38 MV/m; at 100 MV it needs C = 11.7 pF, radius 10.5 cm, surface field 953 MV/m [calc]. Tesla's two numbers (20 MV, 1,100 A) fit a metre-scale top load; the book's 100 MV and 1,100 A do not. (Surface fields above about 3 MV/m break down air, so a bare 0.5 m sphere cannot hold 20 MV either; Tesla's real terminals sprayed streamers, which is what limits voltage. This weakens the neatness of the fit but not the point that 1,100 A is ordinary circulating current.) - Product V x I = 110 GW is reactive power (voltage and current 90 degrees apart in a resonator), not delivered power. - Disconnectable: a resonator with quality factor Q rings with energy time constant of order Q/omega, about 106 microseconds for Q = 100 at 150 kHz [calc]. Tesla's own statement that with no receivers "the plant consumes only a few horsepower necessary to maintain the electric vibration" [web via Inv., tfcbooks 1917] is the conventional high-Q reading. The book's "twenty minute or more aether avalanche" from long photographic exposures is an inference; the simpler reading is that the apparatus kept running while the shutter was open.

Disagreement and recomputation of the 26-mile lamp test (200 lamps x 50 W = 10 kW received at r = 41.8 km, a hemisphere of area 2 pi r^2 = 1.1e10 m^2) [calc]:

The mathematician's figure is the most generous possible to Tesla, and still needs 100 to 230 MW from a station whose supply was of order hundreds of kW at most [memory, low-medium; the book gives no supply power]. The physicist's TW figure assumes an ordinary receiver and is the realistic end. So: by radiation, the claim fails by at least three orders of magnitude on the most generous antenna. The reviewers did not compute a non-radiative (earth-conduction) route; Tesla's own proposal was a conduction and earth-resonance scheme, and the book supplies no quantification of it. The lamp story originates in O'Neill, Prodigal Genius (1944), assembled "by piecing together the fragmentary material" [web via Inv.: https://pupman.com/listarchives/2007/Oct/msg00342.html]; no measurement record was found.

Savo car (1931; Savo to Ahlers 1967 to the book). Recomputed required broadcast power for 80 hp = 59.7 kW [calc]: a 1 m^2 aperture needs 60 kW/m^2 (E = 4.7 kV/m), 38 TW at 10 km (physicist; the 1 m^2 is arbitrary); a best-case matched whip needs S = 5.5 W/m^2 (46 V/m rms) at 1 MHz, 3.5 GW at 10 km, or S = 0.125 W/m^2 at 150 kHz, 78 MW at 10 km. For scale, a 50 kW broadcast station gives E = sqrt(30 P)/r = 1.2 V/m at 1 km [calc]. No transmitter of that size existed in 1931 [memory, medium]. The ambient atmospheric field (about 130 V/m dc, air-earth current density about 2 pA/m^2 [web: https://arxiv.org/pdf/physics/0506077]) gives nanowatts. So a radiated or ambient source is excluded by the energy budget; a hidden onboard source is not excluded by anything in the account. The account itself (Ahlers's 1967 summary [web via Inv.: http://www.tfcbooks.com/special/savo_account.htm]): one witness, 36 years later, one interviewer who noted Savo "lacks a formal education", no second witness, the object left in a barn and not recovered, the 12 V battery "for the lights only", and tubes named "70-L-7" that date to about 1939 (Smithsonian sample) or 1947 (Valve Museum) [web via Inv.: https://americanhistory.si.edu/collections/object/nmah_702597 ; https://www.r-type.org/exhib/aaa1317.htm]. It stays in the record as a documented single-witness account with a dating anomaly; it is neither confirmed nor shown to be false.

Stronger argument. The energy-budget argument is decisive against any wave or ambient-field supply, and conventional resonance explains the numbers Tesla himself gave. The contrarian is right that no one ever did an energy accounting at Colorado Springs, so "energy from space" is not tested; it is also not needed to explain anything reported.

Final verdict (split). - High-Q resonant transformer with large circulating current, streamers, near-field lamp lighting, small running power with no load: Established. - 100 MV and 10,000 hp as Colorado Springs achievements: Internally inconsistent with the book's own 10 kV per inch rule (100 MV needs 833 ft of coil surface [calc]); also a misattribution of Tesla's 1904 plan to 1899 measurements. (Reworded in the QE revision; the earlier label "Historically superseded but instructive" did not fit a documentary misattribution.) - 150,000/s with 0.0062 ms and a "lost" rest period: Internally inconsistent. - 200 lamps at 26 miles by radiation: Refuted by calculation (at least 100 MW needed on the best antenna). As a non-radiative earth-conduction effect, and as a historical event: Never properly tested (1944 reconstruction, no measurement record). - Energy "drawn from space", disconnectable and self-sustaining: Never properly tested (no energy accounting by anyone); the strong form that a resonator keeps running for seconds to minutes without supply is Refuted by calculation (Q/omega time constants of microseconds to milliseconds). - Savo car: Never properly tested (one witness, no object, tube dating anomaly); any radiated or ambient supply Refuted by calculation.

Settle it. Tesla's Colorado Springs Notes (the 1978 edition) for supply power, frequency, capacitances and the source of 150,000/s and 0.0062 ms; then an energy-metered replica (input power measured at the wall, output measured in a calorimeter) run through a disconnect.

Claim 3. Marconi plagiarised and inferior; the "Tesla Rule"; ruling "Houses" (PDF pp.47-153) [corrected 2026-10-09 after QE: was "book pp."]

Agreement. A historical and political claim, outside what physics can test. The "Tesla Rule" is unfalsifiable as used, because the same book calls HAARP a frail copy of Tesla (p.122, "HAARP seeks the evocation of forces which Tesla successfully secured") and also "the very antithesis of systems developed by Tesla" (p.127) [read via deep read]. The Eastlund patent's only non-patent citations are Tesla items (Liberty 1935; NYT 1915 and 1940) [web via Inv.: https://patents.google.com/patent/US4686605A/en], which sits beside the book's statement that the patent has nothing to do with Tesla. No named entities, dates or records are given for the "Houses".

Disagreement and web check. The reviewers differed on what the 1943 Supreme Court decision (Marconi Wireless Telegraph Co. v. United States, 320 U.S. 1) says. I read the opinion text [web: https://www.law.cornell.edu/supremecourt/text/320/1] (first 100,000 of 163,541 characters). It held Marconi's broad claims of US 763,772 invalid as anticipated by Stone's patent (the Court of Claims "was right in deciding that Stone anticipated Marconi"), considered Lodge's variable inductance, and vacated and remanded claim 16 (Pupin, Fessenden). It also says of Tesla's Patent 645,576 that Tesla "recognized that his apparatus could, without change, be used for wireless communication", that "Tesla thus anticipated the following features of the Marconi patent", and that a feature "not shown by Tesla was the use of a variable inductance". It stated it need not decide whether Marconi's patent involved invention over "Lodge, Tesla, and Stone". So: the physicist's memory (Tesla among the anticipating art) is supported; the Court did not award Tesla priority over Marconi; the popular "Court found Tesla invented radio" is a reading beyond the holding (the Mercurians article makes the same point [web: https://www.mercurians.org/1998_fall/misreading.htm, snippet]). The author's priority strand therefore has real but limited support. Marconi's 1901 reception rested on one listener and is doubted by some historians, which supports the author's scepticism of that event, not his explanation (a "Teslian component"), for which "calculations have proven" appears with no calculation shown [read via facts.json].

Final verdict. - Marconi plagiarism or priority: Not testable as stated (legal and historical; partly supported by the 1943 ruling on anticipation of particular claims; overstated as plagiarism). - "Teslian component carried the 1901 signal": Not testable as stated (no calculation given). - "Tesla Rule": Internally inconsistent (unfalsifiable in the way it is used). - Oligarchic "Houses" directing suppression: Not testable as stated; the book cites no sources for it. Tesla's loss of funding (Morgan, about 1903-04) and Wardenclyffe's foreclosure are documented history [memory, high]; whether that was design or ordinary commercial failure is not settled by anything the book cites.

Settle it. Not an experiment. Documents: the full opinion text and the Court of Claims record (for what Tesla's patents were held to disclose), Marconi Company and Morgan correspondence, and Wardenclyffe foreclosure papers.

Claim 4. Unacknowledged Cold War radiation technology "documented in patents" (PDF pp.70-153) [corrected 2026-10-09 after QE: was "book pp."]

Agreement. All three read the source text and give consistent splits. The wording "documented in patents" conflates the existence of a patent with the existence of working technology; the patents named (Eerkens 3,430,046, DeLoraine 2,546,984, and others in the bibliography) exist as bibliographic entries, and a patent records intended operation, not performance.

4a Nuclear EMP. Established (Starfish Prime, 1962, damaged equipment in Hawaii about 1,400 km away [web: https://en.wikipedia.org/wiki/Starfish_Prime via Math]). The book's Argus altitude "12,900 miles" is wrong by about 26x to 117x (the three shots were at 170, 310 and 794 km, 110 to 493 miles [web: https://en.wikipedia.org/wiki/Operation_Argus]); the book is consistent with itself about the belts (12,900 miles is about 20,800 km, inside the outer belt) but not with the record. [CORRECTED in the QE revision: my first draft said 12,900 miles lay outside the belts; that was wrong.] The unsupported "destroyed powergrid integrity" belongs to Starfish, not Argus; the book's Starfish "27,000 miles near the North Magnetic Pole" is also wrong (about 250 miles over the central Pacific [web: https://en.wikipedia.org/wiki/Starfish_Prime]). The contrarian's Rainier flag is confirmed: Rainier was a 1.7 kt underground shot of Operation Plumbbob on 19 September 1957 [web: https://en.wikipedia.org/wiki/Operation_Plumbbob], not a Hardtack (1958 Pacific series) shot as the book lists it.

4b "Clear Atomic Light". The book describes buried nuclear or chemical charges in lined tunnels giving "curtain cruciform beams", sourced by the companion volume to a "De Ment" patent [web via Inv.: https://archive.org/details/gerry-vassilatos-declassified-patents-of-the-cold-war-and-sdi-1997-borderland-sc]. Real kernel: explosive-driven light sources and the nuclear-pumped x-ray laser concept (Livermore) are real [memory, high/medium]. The patent text was not obtained. The mathematician adds that x-rays reflect efficiently only at grazing incidence, so "plastic sheets" as x-ray mirrors is doubtful [memory, medium]. Verdict: weak form (a nuclear or chemical explosion as a one-shot intense source of light and radiation): Plausible, untested. The book's beams and its figure of 10 TW per square metre (no area or duration given): Never properly tested (no open record; classification cannot be excluded, but the only evidence offered is the patent text).

4c Mossbauer "gamma-ray bomb". Agreed split. Recomputed [calc]: 1 kg of cobalt-57 holds N = 1.06e25 atoms; gamma energy at 122 keV per decay is 2.06e11 J (about 49 tons TNT; 20 kg per kiloton); the mathematician's 2.3e11 J includes the 0.136 MeV all-emission figure, the contrarian's 2.1e11 J/kg the 122 keV gamma alone, and they agree. Fission releases about 8.2e10 J/g, so the gamma store is about 400x lower per mass, which reverses the book's claim that gamma bombs make fission obsolete. A kilogram of Co-57 radiates about 6.8 kW continuously (half-life 272 days), so a big store self-heats. The book also says each Mossbauer gamma "is accompanied by ... phonons"; the effect is the opposite (recoil-free, so no phonon is created, which is why the line is sharp) [memory, high]. Hafnium-178m2 triggering (Collins et al., 1998-99) is the nearest real analogue: not reproduced by Argonne at 100,000x the x-ray intensity [web: https://en.wikipedia.org/wiki/Hafnium_controversy, via Physicist and Contrarian]. Verdict: acoustic modulation of Mossbauer lines: Established. Release of "the entire sum of gamma emissions" in one pulse and "complete conversion of matter into gamma rays": Refuted by calculation (energy store, baryon-number conservation). Mutable half-life by sound or other non-nuclear means: Tested, not supported (constancy to better than the percent level; the known exceptions, 7Be and fully ionised 187Re, need atomic-state changes, not sound [web: https://en.wikipedia.org/wiki/Isotopes_of_rhenium via Math]). Hf-178m2 x-ray triggering: Tested, not supported.

4d Gamma communications to submarines at any depth, "unjammable". Water mass attenuation at 1 MeV is 0.07072 cm^2/g [web: https://physics.nist.gov/PhysRefData/XrayMassCoef/ComTab/water.html], giving a 14 cm e-folding length in seawater (density 1.025) [calc]. At 10 m the fraction is exp(-72) = 5e-32; at 305 m (1,000 ft) it is 10^-959. At the attenuation minimum (mu/rho about 0.017 to 0.022 cm^2/g at 10 to 50 MeV [memory, medium]) 305 m still gives 10^-231 to 10^-299 [calc]. The book's "neither dispersions nor diminutions of strength" through "several thousand feet of seawater" is wrong by hundreds of orders of magnitude. Verdict: Refuted by calculation. A gamma-ray laser has never been built [memory, high]: Never properly tested (asserted, no record). The patents exist.

4e Reich "neutralisation". Reich was convicted on 7 May 1956 of criminal contempt for violating the 1954 injunction on accumulators, not of "malpractice" [web: https://www.encyclopedia.com/law/law-magazines/wilhelm-reich-trial-1956]. The book's account of neutralised radium and cobalt-60 has no source; the Reich museum's own account of the 1951 Oranur experiment (a 1 mg radium needle in an accumulator, high readings, sick staff) points to heightened excitation, not removal [web: https://old.wilhelmreichmuseum.org/oranur_experiment.html, via Contrarian; interested party]. Verdict: generic half-life change by non-nuclear means: Tested, not supported; orgone neutralisation and biosensors "going dead": Never properly tested (single-source, 70-year-old, no measurements; the biosensor is defined as the only detector, so as written it is also close to Not testable as stated).

Stronger argument. The contrarian's attenuation and energy-density arguments and the physicist's and mathematician's agree to the same numbers, so the performance claims fail on calculation without needing any reading of classified material. The contrarian is right that classification cannot be excluded for the CAL and orbital items, and that the sole evidence offered is the patent text.

Settle it. Obtain the De Ment, Eerkens (3,430,046) and DeLoraine (2,546,984) patent texts and read them for stated numbers; for the half-life item, measure the Co-57 or Fe-57 decay rate under strong ultrasound (cheap, clean). The attenuation result needs no further test.

Claim 5. HAARP prototype as an industry-owned test of drawing auroral power, about 60 GW (PDF pp.242-276) [corrected 2026-10-09 after QE: was "book pp."]

Agreement. Verified: 300,000 A x 200,000 V = 6.0e10 W [calc]. The book's prototype hardware is correct and matches the operator: 18 energised elements at 360 kW (each element two orthogonal dipoles with one 10 kW transmitter each, so 36 transmitters; the same 20 kW per element gives 960 kW for 48 and 3.6 MW for 180 elements, the official stages [web: https://haarp.gi.alaska.edu/faq via Inv.]). The Investigator's "48 versus 18 conflict" is not a conflict (48 towers stand, 18 are energised); the book's one slip is a 1996 sentence that runs stages together. The Eastlund patent chain: Atlantic Richfield 1985, APTI 1987, E-Systems bought APTI 1994, Raytheon bought E-Systems 1995, APTI spun out 1998, BAE Systems 2003 (patent assignment recorded 2006) [web via Inv.: https://patents.google.com/patent/US4686605A/en ; https://en.wikipedia.org/wiki/Advanced_Power_Technologies]. So "Raytheon-owned contractor" was true 1995-98; "industry owns HAARP" is unsupported (HAARP was a USAF, Navy and DARPA programme [memory, high] and the book itself says at p.124 that operation is "obviously being directed by the military" beside "military has no ownership"). The book is wrong on: "HF cannot make ELF" (measured otherwise, see below); 1/lambda (not 1/lambda^2) gain scaling; 180-degree feed giving circular polarisation (it gives linear; 90 degrees is needed [memory, high]); a 12-acre screen on a 9.2-acre site (500 x 800 ft = 9.2 acres [calc]); Gakona on the "North Slope" (it is in the Copper River basin).

Differences in the power bounds, recomputed [calc] (they address different mechanisms, which is why the numbers differ): - Aperture bound (needs no circuit model). If 60 GW were spread over the whole auroral oval of about 7.6e12 m^2 [memory, medium for the oval geometry], that is 7.9 mW/m^2, within bright-aurora energy flux of order 1 to 100 mW/m^2 [memory, medium]. A 12-acre (4.9e4 m^2) screen intercepts 0.4 kW at that flux and about 5 kW at 100 mW/m^2. Local Joule heating is similar: Sigma_P E^2 = 10 S x (0.05 V/m)^2 = 25 mW/m^2, or 3.3 kW over the 33-acre site [calc; Sigma_P and E are memory, medium]. - Conduction bound. Even if 200 kV lay across 100 km of air (2 V/m) at fair-weather conductivity of 2e-14 S/m, J = 4e-14 A/m^2 and I = 2 nA over the screen, against 3e5 A wanted (a factor 1.5e14) [calc; conductivity from https://arxiv.org/pdf/physics/0506077, treated as order of magnitude, since ionised columns and streamers would raise it]. The whole Earth's fair-weather current is about 1.8 kA at about 250 kV, about 0.45 GW [calc], so 3e5 A is 167x the global circuit. - Induction bound. The measured counterpart is geomagnetically induced current: about 10 V/km ground field, pipeline currents of order 50 A [web via Inv.: https://www.gi.alaska.edu/alaska-science-forum/aurora-makes-waves-ground-too]. The generous version (full 3e5 A through the 11 V that a 240 m screen would see in a 0.045 V/m field) is 3.2 MW, 1.9e4 times below 60 GW [calc]. All bounds fall between 2 nW-scale and a few MW; the generous ones are 4 or more orders of magnitude short. The conduction number is the weakest (it assumes un-ionised air) and the aperture number the most robust. Economics: 60 GW continuous is 1.9 EJ per year, about 0.3% of world primary energy [calc; world figure memory, medium], so even if real it would not "replace oil".

ELF. The contrarian and mathematician add the experimental point that counts: modulated HF heating of the lower ionosphere in the presence of the electrojet generates ELF/VLF waves, measured at HAARP and elsewhere (Moore, Inan and Bell 2006; Cohen et al. 2012; DEMETER detections at watts-level [web: https://angeo.copernicus.org/articles/22/2643/2004/angeo-22-2643-2004.pdf, snippets via reviewers]). So the book's "impressed ELF" impossibility is wrong. This is also the one place where "a small transmitter controls a large natural current", a version of the author's valve idea, is a measured effect; the output is waves at watts-level, not delivered dc power.

Effective radiated power (ERP). The "1 Gigawatt" is gain-multiplied. The aperture-limited maximum for 3.6 MW over 33 acres at 30 m wavelength is gain 1,871 (32.7 dBi), ERP 6.7 GW [calc: mathematician]; reported values for the completed array are 0.36 to 4 GW for 20 to 30 dB gain, with one popular article giving about 5.8 GW [web: https://archive.aps.org/dpp/2011/bm10/1 ; https://newatlas.com/haarp-operations-on-hold/28383, snippets]. I take "a few GW" as the physicist's statement and 6.7 GW as the ideal ceiling. The book's "1 GW" for the planned array is within that range. The magnetic dip angle at Gakona (physicist: about 75 degrees [memory, high]) could not be confirmed on the web in this pass (check failed); it does not affect any verdict.

Absence of evidence: see section 4.

Stronger argument. All three agree. The 60 GW failure is by the book's own unspecified "funnelling" step: a real, large natural current whose coupling to a ground terminal has no mechanism and no measurement. The contrarian is right that no one tested drawing electrojet current to a ground screen, so the strong "never worked" claim cannot be made on a direct test; the case against rests on the bounds above and on 29 years of the open record.

Final verdict (split). - HAARP's Gakona prototype is a conventional HF ionospheric heater, and the hardware description is accurate: Established (operator's record). The author's thesis that its purpose is electrojet power by elimination, "in the absence of all other evidence": Not testable as stated (split in the QE revision). - "Industry owns HAARP": Never properly tested as an ownership claim; Raytheon's ownership of the contractor 1995-98 is documented. - "Not an ELF system" / "HF cannot make ELF": Tested, not supported (measured). - 60 GW (about 3e5 A x 2e5 V) drawn to a 12-acre ground screen at no fuel cost: Refuted by calculation (aperture, induction and global-circuit bounds). The mechanism ("bunching", "tornadic stream", "valve"): Not testable as stated (no equations). - Scaling argument ("1000 times Arecibo", 1.6 MW equivalent): Internally inconsistent. Correct 1/lambda^2 scaling gives (18/33) x (30/0.03)^2 = 5.5e5, about 870 GW [calc: mathematician; verified], the opposite of the book's conclusion; and a 10 GHz radar is hardly absorbed by the ionosphere at all, while 2.8 to 10 MHz is, so "equivalent work" mixes transmitted and absorbed power. - "Major announcements soon": Tested, not supported (see section 4).

Settle it. The E-Systems and Raytheon HAARP contract technical reports (DTIC), USPTO assignment records for the Eastlund patents, and a direct measurement of current into a large ground screen during an auroral event (mA or A against the claimed 3e5 A).

Claim 6. The auroral energy record (PDF pp.130-135) [corrected 2026-10-09 after QE: was "book pp."]

Agreement. Every item is a real report and is kept in the record: Lemstrom's 1872-1884 programme in Finnish Lapland (Sodankyla 1882-84 [web via Inv.: https://pubmed.ncbi.nlm.nih.gov/38494601/ ; https://hgss.copernicus.org/articles/15/17/2024/]), which Corliss rates "widely reported but apparently never replicated" and which the book admits "did not produce consistent available outputs of the energy"; ground-touching aurora reports compiled by Corliss, including Norwood 1929 (Nature 127:108, 1931, not obtained); long-delay echoes, first heard by Hals near Oslo in 1927 and investigated in 1928 with Stormer and van der Pol, delays of about 3 to 15 s, "still considered anomalous, with several proposed explanations" [web via Inv.: https://en.wikipedia.org/wiki/Long_delayed_echo]; Siple-Roberval VLF amplification (growth rates near 100 dB/s, total gains up to 30 dB (1974), 30-50 dB (1988) [web via Inv., snippets]); and auroral currents in long conductors (pipelines about 50 A, ground field up to 10 V/km). Correction to claims.md: the echoes were first heard in 1927, tested 1928.

The long-delay-echo arithmetic: 12 s x c = 3.6e6 km = 89.8 earth circumferences, so "exceeding 84" holds; a 12 s round trip puts a reflector at 4.7 Earth-Moon distances (half the path), not "6 times further than the moon" [calc]. The mathematician's refinement stands: a mean delay of 7.7 s gives a path of exactly 6.0 lunar distances, so the author appears to have quoted path length as reflector distance (a factor-two slip, not an invented number).

Mainstream counterpart (physicist, new in the record): recent recordings place auroral "sounds" about 70 to 75 m above ground, preceded by magnetic pulses, with a temperature-inversion mechanism proposed [web: https://aalto.fi/en/news/the-aurora-borealis-can-be-heard-even-when-they-cant-be-seen]. That is a real near-ground geomagnetic-storm effect, not a conduction path carrying amperes. Auroral emission is at about 80 to 300 km, with the lowest instrumented triangulation on Corliss's list at 65 km (Tromso 1932) [web via Inv.].

Disagreement. None of substance. The physicist labels the observations "Established as reports"; the contrarian more carefully labels ground-touching aurora Never properly tested because no instrument measured it.

Stronger argument. The honest reading of the Norwood account is that it is a first-hand description from a technical witness that conventional discharge and radio-absorption explanations also fit; neither side has instrument data. Siple-Roberval is the strongest item for the author, but it amplifies a wave using radiation-belt electron energy; it does not deliver current to the ground.

Final verdict (split). - Siple-Roberval amplification (narrow form: natural plasma energy amplifies a weak injected signal): Established. - Lemstrom's glow, Corliss ground-level reports, Norwood 1929: Established as reports; whether aurora reaches the ground: Never properly tested (no instrument, not replicated). - Long-delay echoes: Established as an observation; the cause is open (the author's storage-in-the-electrojet idea is one of several): Never properly tested. - Atmospheric-electricity collection patents (Vion 1860, Ward 1872, Dewey 1889, Palencsar 1901, Pennock 1909): the patents exist; Plausible, untested as a power source but a 1 km^2 collector draws about 2 microamps (0.02 W at 10 kV) from the fair-weather field [calc: 2 pA/m^2 x 1e6 m^2]. - "A transmitter can pull auroral energy down to the ground": Never properly tested; see claim 5 for the Refuted by calculation power-scale result.

Settle it. Instrumented low-aurora observation (camera triangulation plus ground field mill) in a storm; the original Nature 127:108 (1931) note; a systematic long-delay-echo campaign with a controlled transmitter.

Claim 7. Philo Farnsworth's 1966 self-sustained fusion, suppressed through ITT (book p.137)

Agreement. The book cites no source. The fusor is a real device and the Farnsworth-Hirsch fusor produced D-D fusion neutrons; Hirsch's more carefully controlled 1967 trials gave more than 10^10 neutrons per second [web: https://iec.neep.wisc.edu/files/AmericanScientist.htm; secondary reprint; the 1967 J. Appl. Phys. 38:4522 paper was not opened]. Recomputed energy balance [calc]: 1e10 neutrons/s means 2e10 reactions/s (half the branches yield a neutron), at an average 3.65 MeV per D-D reaction (branches 3.27 and 4.03 MeV) that is 11.7 mW of fusion power. Against an input of 500 W the ratio Q is 2.3e-5; against 5 kW, 2.3e-6. Break-even would need Q of order 1, five or six orders of magnitude more. No fusor has come close to break-even [web: https://en.wikipedia.org/wiki/Fusor]. "Thermonuclear" is also a category slip (fusor ions are accelerated by grid voltage, not in thermal equilibrium). The family finding aid (University of Utah MS 0648, an interested archive) says "the necessary reaction lasted no longer than thirty seconds"; ITT's board pressed to end the work in December 1965, the 1966 budget was approved at Geneen's urging, the work ended a year later and continued at BYU in 1967 [web: https://archiveswest.orbiscascade.org/ark:80444/xv98312 via Inv.]. The companion volume words it more weakly: "contained nuclear fusion reactions with control" [web via Inv.].

Disagreement. The suppression part: the physicist says Not testable as stated, the mathematician and contrarian say Never properly tested. I side with the latter two. The claim is a statement about intent that records could decide (ITT board minutes, Farnsworth's notebooks); what is documented is a corporate funding termination, which has an ordinary commercial explanation, but its motive is unrecorded. It is not untestable in principle. Both ITT and the family have a stake in how the ending is told.

Final verdict (split). - A fusor can produce D-D fusion neutrons: Established. - "Self-sustained fusion with an electrical output of enormous value" in 1966: Tested, not supported (best-controlled runs at Q of order 1e-5 to 1e-6; 30-second reactions; no net gain by any fusor). - "Suppressed through ITT": Never properly tested (a documented end of funding; no document of intent either way).

Settle it. Farnsworth papers MS 0648 (Boxes 16-17, Mark II drawings and research notes) for neutron counts and input power; ITT board minutes.

3. Calculations I re-ran (reviewers differed or I wanted an independent check)

Two points the reviewers' numbers do not settle. (i) The mathematician's chronaxie, polar-cap field (0.045 V/m over 4,400 km), auroral oval area and Pedersen conductance are [memory, medium]; they set the order of the bounds but not the conclusion. (ii) The physicist's "few GW" for HAARP ERP and the mathematician's 6.7 GW are compatible: 6.7 GW is an aperture ceiling, a few GW the reported figure.

4. Absence-of-evidence analysis

Where the record goes dark and why it matters (SOURCE HONESTY).

4.1 HAARP and the predicted "major announcements ... soon" (1996-97)

The record: HAARP was built from 1993; first prototype winter 1994 at 360 kW; 48 elements and 960 kW by 1999; 180 elements and 3.6 MW by 2007; equipment transferred to the University of Alaska Fairbanks in August 2015; about 10 MW of diesel prime power on site [web via Inv.: https://haarp.gi.alaska.edu/faq]. No energy announcement of the kind predicted appears in the operator's public record over the 29 years since the book.

Competing explanations and evidence. 1. The thesis is wrong (no mechanism; bounds in claim 5; HAARP's published science is ionospheric heating, ELF generation and plasma physics). For: the calculations and the diesel dependence. Against: none found, but this explanation does not require absence of a result to be informative. 2. The work was classified, proprietary or moved. For: Raytheon's 1995-98 ownership of the contractor and the Eastlund patents is documented; the HAARP contract reports (E-Systems, Raytheon) are not in this project folder. Against: a GW-scale source would show in public grid data, and HAARP still burns diesel; a hidden extraction experiment would have to sit inside an open, visited facility that was handed to a university in 2015. 3. HAARP was never an energy project. For: the operator's description and sponsorship (USAF, Navy, DARPA) and the research record. Against: the Eastlund patents' own language (weather modification, missile effects, "energy"), which the author reads as intent. 4. HAARP tried and found nothing worth announcing (contrarian). Compatible with 1 and 3; it would explain silence without a cover-up. The deciding records: E-Systems and Raytheon technical contract reports (DTIC), USPTO assignment records for US 4,686,605 and relatives, HAARP's 1993-95 experiment plans, and the facility's power log. None was obtained. Not obtained is not the same as negative: the Investigator found no Raytheon assignment of the Eastlund patents (the recorded chain goes ARCO, APTI, BAE), which weakens the book's "Raytheon swept the patent floor" while leaving temporary 1995-98 ownership of APTI intact.

4.2 The Savo car and the vanished receiver (1931 to 1967)

The object was reportedly left in a barn and not recovered; one rumour (shipping to Yugoslavia, from the physicist's reading of the Investigator) has no source. Explanations: (a) the account is mistaken or embellished (supported by the tube dating, a recollection 36 years later, an interviewer's note about Savo's education, and the energy budget); (b) the object existed but was an ordinary hidden-source device or a different vehicle story (not excluded by anything in the account); (c) the object was real and removed (no evidence; Tesla's papers were taken after his death in 1943, a separate fact and not specific to this object [memory, high]). Deciding records: Savo's immigration and Knin records, Ahlers's original notes, any contemporary Buffalo press, the Pierce-Arrow registration. None found.

4.3 Tesla's Colorado Springs numbers

The Colorado Springs Notes exist (1978 edition) and were not read here. The 150,000/s and 0.0062 ms figures, the 26-mile test and the statement that the transmitter was run disconnected all lack a primary source in the files. Silence in the sources the Investigator reached is not contradiction. Deciding record: the Notes, and any surviving power-company records for the Colorado Springs Electric Company supply.

4.4 Farnsworth after 1966

ITT ended funding about 1966-67; work continued briefly at BYU; the fusor survives as an amateur neutron source. Explanations: ordinary corporate termination after a short-lived reaction (documented in the family's own finding aid); suppression (no document); a technical dead end in grid losses (the physics supports this). The deciding records are ITT board minutes and the Farnsworth notebooks (MS 0648). The documented story does not need suppression, and a suppression directive, if it existed, has left no trace in what was reached.

4.5 CAL and orbital gamma devices

No test record, no open patent drawing and no performance data beyond the book's assertions. Classification would explain the absence of tests, but the same explanation would apply to any claim, so it cannot support this one. The patents named but not obtained (De Ment, Eerkens, DeLoraine) are the nearest decisive records.

5. What would settle it (the single deciding item per claim)

1. Instrumented closed-shield impulse test (sealed grounded copper box, calibrated 1 to 6 microsecond source, probe inside; repeat with a seam).
2. Colorado Springs Notes for supply power and a metered replication with a disconnect; a measurement of the 26-mile conduction route is not possible now, but the Notes' stated receiver and field values would give it.
3. Not an experiment: the 1943 opinion and the Court of Claims record on what Tesla's patents were held to disclose; Marconi and Morgan correspondence.
4. De Ment, Eerkens and DeLoraine patent texts; an ultrasonic decay-rate measurement of Co-57; the attenuation result needs nothing.
5. DTIC contract reports and USPTO assignments; a direct current measurement into a large ground screen during an auroral event.
6. A low-aurora triangulation with a ground field mill; Nature 127:108 (1931).
7. Farnsworth papers MS 0648 Boxes 16-17 and ITT board minutes.

6. Web verification (budget 8; used 7)

7. Open questions

1. What did Tesla's "shielded" arrangement really consist of, and does any instrumented repeat of the sting-through-shield observation exist (Dollard's reproduction is reported with no data)?
2. What was the supply power at Colorado Springs, and how was the 26-mile lamp test actually conducted (if at all)? The Notes should say.
3. Could the earth-conduction route Tesla proposed deliver tens of kW at 40 km? The reviewers computed only the radiative budget.
4. What does the De Ment patent describe, and does any open record of explosive-driven "beam" performance exist?
5. Do the E-Systems and Raytheon HAARP contract reports say anything about electrojet energy contact? Why does the Eastlund patent cite Tesla items?
6. Are long-delay echoes explained? Mainstream sources still call them anomalous; the author's storage idea is one candidate among several.
7. What is the origin of the low-altitude ground-touching aurora reports (Norwood 1929): corona on house wiring in a strong ground field, auroral radio absorption for the dead receiver, or something else? An instrumented low aurora would decide.
8. Was Peter Savo a nephew of Tesla, what exactly did he see, and what became of the car?
9. Did Farnsworth's notebooks show anything beyond neutrons, and was there a documented ITT instruction to stop?
10. The book's source for "Houses" and for Terbo/RCA (a family name; Tesla's great-nephew William Terbo was an RCA executive [web via Inv.]; no record of Tesla at RCA found) is unstated.

8. Corrections to earlier documents

(Per prompt.txt; the orchestrator applies them.)

1. claims.md claim 2 and deep read pp.36-46 section 4 ("1100 A has no source"). Wrong. 1,000-1,100 A is Tesla's own figure (1917 Edison Medal speech; 1919 My Inventions). Same speech gives about 20 MV as his highest potential. At 150 kHz, 1,100 A at 20 MV needs a 58 pF top load [calc]. The book reads the current as "a measure of aether flow"; that interpretation remains the author's inference.
2. Deep read pp.36-46 section 3.7 ("editing seam" between two Savo receiver descriptions). Withdrawn: both descriptions are in Ahlers's 1967 summary (physicist, mathematician).
3. sites.md Leads and Investigator ("prototype element count conflicts: 48 versus 18"). Not a conflict: the book says 48 towers stand and 18 are energised at 360 kW, identical to UAF. Add that to the lead.
4. Deep read Ch.7 first half E8.3 (power chain ambiguity). Resolved: 18 elements x 2 dipoles x 10 kW = 360 kW; 20 kW per element reproduces 960 kW (48) and 3.6 MW (180).
5. claims.md claim 6 ("heard in 1928"). First heard 1927 (Hals); tested 1928.
6. claims.md claim 4 ("documented in patents"). Split: the patents exist; the performance the book describes is not documented by them.
7. claims.md claim 5 ("industry-owned (Raytheon)"). Raytheon owned the contractor APTI 1995-98 (documented); the chain never put the Eastlund patents in Raytheon's name in the recorded assignments (ARCO, APTI, BAE); ownership of the facility by industry is unsupported.
8. facts.json Chapter 4 and Chapters 5-6 entries. Add: Argus altitude 12,900 miles is wrong by about 43x (about 300 miles); the Argus grid-damage statement belongs to Starfish; Rainier was Plumbbob 1957, not Hardtack 1958 [web]; the Mossbauer phonon statement is reversed; Jim Creek NLK is 1.2 MW at 24.8 kHz (book: 10 MW) and NAA is near 24 kHz (book: 2 MW at 14.8 kHz) [web: https://en.wikipedia.org/wiki/Jim_Creek_Naval_Radio_Station via Math]; the Reich conviction was criminal contempt (1956), not "malpractice".
9. Physicist, claim 2 (Savo: "about 38 TW"). Overstated: it assumes a 1 m^2 aperture. The best-case matched antenna gives 78 MW (150 kHz) to 3.5 GW (1 MHz). Conclusion unchanged: no 1931 transmitter could supply that.
10. Physicist, claim 3 and mathematician: the 1943 ruling. The ruling names Stone and Lodge as the anticipating art for the broad claims and treats Tesla's 645,576 as having anticipated features of the four-circuit system; it does not give Tesla priority.
11. Mathematician, claim 5 ("1 GW" ERP scaling) and physicist ("few GW"). Compatible: ceiling 6.7 GW by aperture, reported 0.36 to 4 GW.
12. Contrarian, claim 2 ("1,500 times Tesla's published 1900 figure" for VI). I did not verify the 1900 figure; the ratio is not used here.
13. Digest 12 and claims.md (Seafarer). 5.035 is a digit slip for 5.085; the quantity counted is wavelengths, not nodal zones (10.17 zones).
14. Contrarian Step 1 correction 6 (Rainier flagged for checking). Confirmed as an error in the book (see 8).

9. Final verdicts in brief

1. Radiant Energy: pulse physics Established; a continuous closed copper shield stopping the pulses Refuted by calculation; Tesla's actual shield observations Never properly tested; neutral ether gas Not testable as stated.
2. Ether engine: resonance physics Established; 100 MV as measured Internally inconsistent (misattribution of the 1904 plan) [corrected 2026-10-09 after QE: was "Historically superseded but instructive and Internally inconsistent"; see 10.6]; 200 lamps at 26 miles by radiation Refuted by calculation (by conduction Never properly tested); "from space"/disconnect Never properly tested (strong persistence form Refuted); Savo car Never properly tested.
3. Marconi / Tesla Rule / Houses: Marconi and Houses Not testable as stated; Tesla Rule Internally inconsistent.
4. Cold War radiation: nuclear EMP Established; CAL weak form Plausible, untested, beams Never properly tested; gamma bomb and matter-to-gamma Refuted by calculation; mutable half-life Tested, not supported; submarine gamma links Refuted by calculation; Reich neutralisation Never properly tested.
5. HAARP [5a split, see 10.6: hardware Established; electrojet purpose Not testable as stated; added 2026-10-09 after QE, was unsplit 5a]: ELF impossibility Tested, not supported; 60 GW Refuted by calculation; scaling argument Internally inconsistent; "announcements soon" Tested, not supported; ownership Never properly tested.
6. Auroral record: Siple-Roberval Established; reports Established as reports, ground-touching aurora and long-delay echoes Never properly tested; "pull energy down" Never properly tested.
7. Farnsworth: fusor neutrons Established; "self-sustained" Tested, not supported; suppression Never properly tested.

10. QE revision (2026-10-08, after qe.md and six new deep reads)

Inputs: qe.md (verdict MINOR); the new deep reads in deep/ (Chapter 1 first half, PDF pp.9-35; Chapter 2 Marconi, pp.47-62; Chapters 3 and 5, pp.62-70 and 97-108; Chapter 4 second half, pp.84-96; Chapter 6 VLF/ELF, pp.109-120; Chapter 8 patents, pp.146-153); sites.md Leads. Page numbers are PDF pages (each is a two-page spread; printed page is about twice the PDF page), which also corrects the "book pp." headings of claims 1, 3, 4, 6 and 7 above. Web checks used in this revision: 4 (Argus, Van Allen belts, Starfish Prime, West Ford; results below). The central verdicts do not change. Revisions are listed claim by claim; the verdict table (section 1) and the claim list (section 2) are edited where a verdict or label changed.

10.1 Claim 1 (Radiant Energy): the source layer is thinner than I assumed, and one verdict is re-split

What the deep read shows [read: deep/ Ch.1, sections B-F]: - The shield reports have no instrument, thickness, closure or grounding stated. The sensor was Tesla's face and hands. The book's "glass shield" is singular on first mention, then "several materials", then "glass and copper" (pp.14-15). It is the author's narrative, with no Tesla text named. - The only primary Tesla text found that bears on it is the 21 December 1892 article in The Electrical Engineer [web: https://www.nikolatesla.ch/en/article/on-the-dissipation-of-the-electrical-energy-of-the-hertz-resonator]: "The experimenter may feel the impact of the air at distances of six feet or more from the bar ... These waves cannot be entirely stopped by the interposition of an insulated metal plate." That is one insulated plate, partial attenuation, six feet, and Tesla calls them "waves, which differ from ordinary sound-waves only because the air is electrified". The book's version (copper and glass "as if the shield were not there at all", "nearly fifty feet", "several hundred yards", p.14-17) is several times stronger and unsourced. (The deep reader read the 1892 article and the 1901 patents on the web; the February 1892 London lecture was not obtained.) - The pulse-duration scale (pain above 0.1 ms; none at 100 microseconds or less; heat at 1 microsecond; light and breezes below) and "10,000 volts per inch" appear in none of the Tesla texts checked and are given with no lecture, patent or date (p.17-18). The book's rule gives 24 in x 10,000 V = 240,000 V for the two-foot helix, which matches its own figure, so those are one claim stated twice [read: deep/ Ch.1 section E]. - The 1901 patents (US 685,957 and 685,958) again: a collector of "minute particles of matter which are strongly electrified" (positive from the Sun), not a neutral ether gas, and alternating currents are allowed with the rays admitted only during the right half-cycles, against the book's "alternations were absolutely useless" (p.20). - New items the first pass left out, now in the record: - Joseph Henry (1842), needles magnetised in a cellar vault by a Leyden-jar spark on an upper floor, and Elihu Thomson with Edwin Houston (dated 1872 in the book; the bibliography lists Houston, Scientific American 1872; [memory, medium] the Thomson "etheric force" sparks are usually dated 1875), both third-hand in the book with no matching entry for Henry. The conventional reading is induction and, for Henry, the oscillatory character of a Leyden-jar discharge [memory, high for Henry's oscillatory discharge; medium for the building story]. They are real nineteenth-century observations of action at a distance and are kept as such. - Eric Dollard's 1985 "line interferometry": a surface component said to travel superluminally. No velocity, line, instrument or venue is given; his 1985-86 videos are sold by Borderland Sciences Research Foundation (BSRF), which also published the Vril Compendium volumes the bibliography cites and employs the book's editor Michael Theroux. Dollard also reports charging of capacitors by impulse-driven vacuum lamps, mass repulsion and attraction, and bulbs that rupture and keep discharging (p.25-28). - Theroux's 12-foot white streamers from a 2 ft by 1 ft secondary (p.21): reported by the person who edited the book and works for the publisher; no publication, voltage or date. - Gregory Hodowanec: "continues to flow for hours" with capacitor detectors (p.31); four private or magazine titles, none on Tesla. - George Piggot (silver beads suspended in a positive field) and Francis Nipher (1918, Electrical Experimenter, "Can Electricity Destroy Gravitation?") on charge and gravity (p.34); the book says Nipher showed "complete reversal"; [memory, medium] Nipher reported a reduction and later accounts attribute it to electrostatic forces. I did not obtain the article. - William Lehr as the book's private source for the 20 ft plasma blasts and the radioactive roofing copper (p.24, 33): "one of my vital links" (Prologue); no document.

Verdicts (revised): - Nerve strength-duration behaviour (short pulses give duration-dependent effects): Established as physiology; the threshold scale is not attributed to any Tesla text and I record it as the author's, not as a Tesla finding. - Tesla's own 1892 report (felt at six feet or more; not entirely stopped by an insulated metal plate): Plausible, untested. A spark's blast wave, diffraction round the edges of a plate, and capacitive or magnetic leakage fit it, as does Tesla's own reading ("waves of electrified air"). No instrumented repeat is on record. - The book's strengthened version (glass and copper shields ineffective, 50 ft, several hundred yards): Never properly tested, and not traceable to a Tesla text. It has not been tested or refuted; it has not been shown that Tesla said it. - Penetration of a continuous, closed, grounded copper shield: Refuted by calculation (about 51 dB at 1 mm, section 2). The book never says its shields were closed, so this is a conditional result, not a refutation of the reported sensations. - Dollard 1985 "superluminal" interferometry: Not testable as stated (no velocity or setup). Phase velocities above c occur in ordinary dispersive structures and carry no signal [memory, high], so the word alone settles nothing. - Henry and Thomson as "shield" precedents: Established as observations, with conventional induction readings; they say nothing about a non-EM channel. - Piggot and Nipher charge-gravity reports: Never properly tested (articles not obtained). - Hodowanec "ether currents continue to flow": Never properly tested. - Neutral ether gas: Not testable as stated (unchanged), and now also against the 1901 patent text.

Settle it. Unchanged (the sealed-box test), plus two document tasks: Tesla's February 1892 London lecture and the 1893 lectures for the shield wording, and Dollard's 1985 videos and any measurement behind "superluminal".

10.2 Claim 2: smaller changes

• Label of the 100 MV item reworded (see claim list).
• Added to the Savo tracing record (section 4.2): Ahlers's summary (16 September 1967) gives Savo as born in Knin and trained at Wiener Neustadt; says Tesla's secretary "told General Electric"; and mentions a Yugoslav diplomat asking after the car [web via Inv.: http://www.tfcbooks.com/special/savo_account.htm]. These are the leads by which the witness could be traced. The identity is contested: the 1967 summary calls Tesla "Peter's uncle on his mother's side"; a secondary genealogy lists Angelina's son Petar (Trbojevic) and Marica's son Sava (Kosanovic) [web: https://treemily.com/blog/nikola-tesla-family-tree]; Open Tesla Research (a sceptical site) says Tesla had no nephew by that name. Tesla in 1931 lived at the Hotel Governor Clinton (the New Yorker from 1934) [web via Inv., snippets], not the New Yorker the book names for the RCA period. Grotz's Project Tesla (1980s recreation) and his view that Tesla "merely stored energy in the earth" is recorded in the book itself (p.43) and by the physicist and mathematician; it is one of the competing readings of the Colorado Springs experiments.
• Hodowanec's "obtain energy for hours" Tesla phrase (p.31) is unsourced; it does not change the "disconnect" verdict (Never properly tested; strong persistence form Refuted by calculation).

10.3 Claim 3: new material from the Marconi, Houses and 1895 fire records

Marconi chapter (deep read PDF pp.47-62) [read]: - "Calculations have proven" that the wave component could never cross the Atlantic and that a "Teslian component" did (p.50): no number, power, frequency, distance or date appears in the passage or the chapter. The year 1901 is not in the chapter text at all, and the book places the receiving site at Glace Bay, Nova Scotia (p.50). [memory, high] The widely reported reception was at Signal Hill, Newfoundland, on 12 December 1901, with the transmitter at Poldhu, Cornwall; Glace Bay's station dates from 1902. Not web-checked here. Verdict unchanged: "Teslian component": Not testable as stated, and now more clearly because no calculation exists to examine. - The court award "ratified in 1915" (p.58, page image read): "the predetermined Court awarded Marconi with sole priority of Radio"; no court, country, case, patent or judge is named, and "Financial base weighed and paid the court" is an assertion with no document. I cannot identify any 1915 proceeding of that description [memory, medium]. The documented ruling is the 1943 Supreme Court decision, checked in section 6, which went against the broad Marconi claims. Verdict: the 1915 award: Not testable as stated (unsourced narrative); it does not rebut the 1943 ruling. - "The True Wireless": the title only; no quotation (the words "heat" and "electrical rays" are the author's). [memory, medium] It is a Tesla article of May 1919 (Electrical Experimenter); if so it post-dates the 1915 scene the chapter places it after. Not checked on the web. - Engineers who "broke all affiliations with Marconi": De Moura, Hartmann, Gati, McCullough, Hettinger, Ulivi, Turpain, Reno, Moray, Farnsworth (p.57). Surnames only; the text does not say each worked for Marconi, and several are the book's own Tesla-technology subjects. Recorded as the book's roster, single source, with the author's stake; no second source sought. - Poulsen "adopted unpatented designs of Tesla", GE as a body "designed to break Teslian patents", Alexanderson alternators "lifted" from Tesla (pp.53-56): unsourced. - Book errors found in this chapter: Fessenden's 1906 broadcast at "Brant Rock in Nova Scotia" (Brant Rock is in Massachusetts [memory, high]); KDKA attributed to De Forest (it was a Westinghouse station [memory, high]).

Houses model (deep read PDF pp.62-70, 97-108) [read]: the one named source for the whole model is the parenthetical "(Napolitano)", which the bibliography gives as "[private conversation on geopolitical regions, 13 September, 1996]". No House, family or document is named anywhere; the regions are labelled WEO, CAO, SEAO, IO, CO, JO and NAO. The author presents it as "a model", then states applications flatly (Pearl Harbor as oligarchic "permission", p.103). It is the author's reading; I keep it in the record with its one private source. Verdict unchanged: Not testable as stated.

The 1895 fire (restored; qe.md found it dropped). The book (p.36) says the 13 March 1895 destruction of Tesla's South Fifth Avenue laboratory was sabotage by dynamite. The Investigator's record: - The fire began about 2:30 a.m. on the floors below the laboratory, according to the night watchman; oil-soaked premises of a pipe-cutter or dry-cleaner, or a gas jet (Cheney), are proposed causes; Seifer reports suspicion of the watchman; the New York Sun and Herald are quoted; no account found mentions dynamite [web via Inv.: http://teslapodcast.com/2022/03/24/show-notes-029-towering-inferno-1895/ ; https://teslauniverse.com/nikola-tesla/articles/inventor-teslas-loss]. An 1895 trade report gives "33 and 35 South Fifth Avenue" (the book's 8 East 40th St and 33-35 South Fifth Avenue are in its p.11 and p.16). Tesla called his post-fire laboratory "46 Houston Street" (Anderson, Guided Weapons and Computer Technology, via Open Tesla Research [web via Inv.]), against the book's "40 West Houston". - Competing explanations: (a) ordinary fire from the occupied floors below (the watchman's account, the oil-soaked premises); (b) arson, possibly by or through the watchman (Seifer's reported suspicion is the one item that leans this way; it names no motive or paymaster); (c) sabotage aimed at Tesla (the book; no document, and no mention of dynamite in any account found). The simplest reading fitting the documents reached is (a), but (b) is not excluded by anything in the record. Deciding records: the New York Sun and Herald of 13 and 14 March 1895, any fire-marshal, insurance or police report, and Tesla's insurance claim. None obtained. Verdict: sabotage by dynamite: Never properly tested (no document of the book's claim; ordinary-fire accounts exist). The wider "Houses" suppression thesis is unchanged. - Related: Hammond Castle says John Hays Hammond Jr. "would also come to be mentored by both Alexander Graham Bell and Nikola Tesla" [web via Inv.: https://hammondcastle.org/about-john-hays-hammond-jr/]; nothing found supports the book's "permanent guest" claim. - The Eastlund-patent citation of Tesla items and Theroux's double role stand (Theroux, the BSRF editor, supplied the HAARP files [read: front matter] and edited the bibliography [read: p.154]; BSRF also sells the author's other books).

10.4 Claim 4: patents, tests and errors, now read

Chapter 8 patent list (deep read PDF pp.146-153) [read]. Every patent number named matches the bibliography; where the text year differs from the bibliography the text year is the earlier, consistent with filing date against grant date (DeLoraine 1947 text, 1951 grant; Gould 1967/1968, the one case the book never explains; Viecelli, Wood, Mallozzi 1972/1977, Kell, Jensen). Titles differ in almost every case; Pieckenbrock's number is 3,557,370, not "3,587,370" (OCR error in the text layer, so a digest copying it is wrong). The reproduced drawings show Chope filed 1961 (granted 1964), Eerkens filed 1966 (granted 1969), Reiffel filed 1953 (granted 1957). The book's performance claims have no source: "demonstrated their effectiveness in connecting submarines, at extreme depths, with orbital communications satellites" (p.153), "an armada of super pulse Gamma Ray LASER systems" already in orbit (p.153), the Teller bomb-pumped laser being "identical" to Reiffel's rods (p.149-150; the only evidence on the page is a drawing of a plate, rods and tapered tip, and the book itself gives the two devices different power sources), and the 30 GW (Pieckenbrock) and 100 GW (Mallozzi) figures, which do not appear on the visible patent sheets. Two points favour the author and are kept: the patents are a public record that can be pulled and read, and the book's own pp.150 hedge ("seem to be pure fantasy, more like science fiction") sits beside its flat assertions on pp.152-153. The word "(LASER)" in the Reiffel bibliography entry is the book's addition; a 1953 filing predates working lasers [memory, high]. Mallozzi's patent text, as the book quotes it, says "weld, cut, and melt metals", an industrial use the book turns into a weapon. No patent text was obtained for this revision, so what each patent claims is still open.

Verdicts (revised or confirmed): - Gamma/X-ray submarine communication "at any depth", "unjammable": Refuted by calculation (unchanged; the book gives no attenuation number, and its own p.150 admits UV is absorbed in short air paths). - Orbital gamma-ray-laser armada already deployed: Never properly tested (assertion from patents and media impressions; no record; classification cannot be excluded but cannot support a claim either). - "Documented in patents": the patents exist and are listed; their performance is not documented by them (unchanged). - Teller = Reiffel identity: Not testable as stated (a shape comparison with different power sources; no Teller drawing cited). - Beamray devices of Grindell-Matthews, Hettinger, McCullough, Reno "tested and classified during World War II" (p.69): Never properly tested as to any WWII test (stated once, no document; the bibliography lists only patents and 1914-1936 periodicals, none a government record; the book's p.107 says the same devices delivered "scarcely sufficient penetrating energy", range "several hundred feet"; and p.99 treats McCullough and Reno as communications impulse transmitters, so the book classifies them two ways). Turpain's munitions explosion and MacFarland-Moore's exploder are "rumours"/"said to" in the book's own hedge. - Radar pulses "focus terrestrial dielectric energy" into an EMP ray (pp.104-107): Not testable as stated (no tests named, no equations). The mainstream account of high-altitude EMP is the Compton-electron and geomagnetic mechanism [memory, medium]; the book's "dielectric focussing" is an alternative with no data. - Pacific Scatter (8 links, 800 MHz, about 40 kW, 7,400 miles), CLOUDS (1963, barium, strontium, lithium, caesium clouds, Aerobee rockets, public events) and Westford (1962, 350,000 wires, "clumped together"): stated figures, uncited. West Ford: the first launch on 21 October 1961 "failed to disperse"; the second on 9 May 1963 succeeded in carrying radio transmissions; the original aim was 480 million dipoles; the page does not give 350,000 [web: https://en.wikipedia.org/wiki/Project_West_Ford]. So the book's "1962", its 350,000 and its "dismal failure" fit the 1961 launch only; the 1963 success is omitted. The "cover story" reading (a radar target veil for EMP weapons) is the author's. Pacific Scatter's "backscatter" is forward scatter in practice [memory, medium]. - Argus, Starfish, Teak, Orange, Soviet shots (pp.89-90): the book gives altitudes and dates but no sources. Corrections: Argus shots 27 August (170 km), 30 August (310 km), 6 September 1958 (794 km), 1.7 kt each, no grid damage reported [web: https://en.wikipedia.org/wiki/Operation_Argus]; the book gives 29 August and 12,900 miles. Starfish Prime (9 July 1962, about 1.4 Mt) was about 250 miles up, about 900 miles from Hawaii, where it knocked out about 300 streetlights and a telephone microwave link [web: https://en.wikipedia.org/wiki/Starfish_Prime]; the book says 27,000 miles "near the North Magnetic Pole", more than the Earth's circumference of about 24,900 miles. My first draft of 4a stated that 12,900 miles was outside the Van Allen belts; that was wrong. The outer belt extends well past 12,000 km, so 12,900 miles (about 20,800 km) is inside it [web: https://en.wikipedia.org/wiki/Van_Allen_radiation_belt]. The Soviet "16,000 miles, Pacific" is Kazakhstan, 59-300 km [memory, medium]. Nuclear EMP: Established (verdict unchanged). - Hardtack list (p.92): "five NX shots" followed by six names; Rainier is Plumbbob 1957; Gnome's sentence is cut off; yields for Neptune, Blanca, Gnome and Hardhat differ 15 to 25 percent from my memory [memory, medium; not web-checked]. The link of these tests to optoexplosives is the author's. "Operation Distant Plain": the name is not in the bibliography, and I could not identify it [memory, low]. The 10 TW/m^2 figure is an intensity; if 1 ton TNT left in one microsecond it needs only about 400 m^2 of cross-section [calc: 4.184e9 J / 1e-6 s = 4e15 W, divided by 1e13 W/m^2]. So the number is not absurd by energy count, but the book gives no area or pulse length; Never properly tested stands. The De Ment patent (3,715,596, 1973) is the source for the CAL system; not obtained. - Errors in the book that no earlier file listed (web checks pending unless tagged): Mossbauer "graduate student at Caltech" (his work was at Heidelberg for a Munich thesis; he moved to Caltech in 1961 [memory, medium-high]); "ERW" expanded as "explosive radiation weapons" (normally enhanced radiation weapon); the depleted-uranium round called APDS "ignited by piezonuclear stimulation" (modern rounds are APFSDS, and DU is pyrophoric [memory, medium]); the free neutron "12 minutes" (mean life about 14.7 minutes, half-life about 10 minutes [memory, high]); Sputnik I dated 1956 (it was 1957 [memory, high]); "Little Boy" plutonium (it was uranium-235 [memory, high]); the magnetron at Birmingham "headed by Watson-Watt" (Randall and Boot in Oliphant's group [memory, high]); Galvani dated 1620 (lived 1737-1798; possibly a print or OCR error); Hertz's wave experiments placed in Bonn (Karlsruhe 1886-89 [memory, high]); Persinger and Puthoff placed at "NRL and Phillips Corporation" while the book's own bibliography puts them at Laurentian and SRI; a 1979 Antarctic event credited to "DSP" satellites (usually told as a Vela double flash, 22 September 1979 [memory, medium-high]); Jim Creek and Cutler powers and frequencies (already in correction 8 above). These are errors of detail; none changes a verdict, but they bear on the book's reliability, and the reader should know the author states his figures without sources. - Reich: unchanged, plus the following. The neutralisation account rests on one first-hand interested source (McCullough's letter via BSRF) plus Clark (1982) and the Oranur papers; Reich died in federal prison [memory, medium: 3 November 1957, Lewisburg]. The book links the sentence to a warhead-deactivator proposal with no source. Verdict unchanged (Never properly tested for neutralisation; generic half-life change Tested, not supported). - Biosensors going "dead" after each nuclear test (p.94): single unnamed "consortium", no numbers; Never properly tested; the biosensor is its own detector, so close to Not testable as stated. - Paneth and Peters (1926) and Helbronner (1920) "extra energy" from hydrogen over palladium (p.72), and the cold-fusion lineage the author lists (p.96): kept as a cross-project theme for the synthesis step. [memory, medium] Paneth and Peters withdrew their helium claim in 1927 (helium from the glass); needs a web check before it is used.

10.5 Claim 5: Chapter 6 ELF record, and the verdict-scale fix

• The five ELF arguments Chapter 7 reuses against HAARP (non-directability, standing waves, ground conduction, the 1,400 km dipole, 7-10 Hz versus 76 Hz) are all in Chapter 6 (pp.119-120) with no new sources: "(Bearden)" for triangulation being impossible, "(Hunt)", "(Grissett, deLorge, Wever)" for the neuro-entrainment accusations, nothing for the 76 Hz limit beyond assertion. The 1,400 km half-wave dipole is about 1,500 km on the book's own constants (7 percent low), and 5.035 zones is 5.085 (already corrected). The Murgas (50 kHz) and Rogers (70 kHz) "ELF" tests are low frequency, not ELF. [memory, medium] Standard treatment models ELF/VLF between ground and ionosphere as a waveguide, with Schumann-type global modes at these scales, which is not the "primarily ground conduction" the book asserts; I leave that to the physicist's file.
• "A few audio-frequency watts" injected into the ground by American experimenters "can sufficiently disrupt the supposed invincible SEAFARER System" (p.119): a claim in one published source (this book) with no experimenter, date, place, frequency, power figure or publication named; "brought government officials to their door" is likewise uncorroborated; the book's own closing question (p.120) uses it ("If SEAFARER is so weak that amateurs can interfere with its operation ..."). I record it as the author's report, neither confirmed nor dismissed. Verdict: Never properly tested; plausibility depends on where the signal is injected relative to the receivers, which the book does not say. Deciding records: the 1977 Seafarer environmental impact statement, Navy hearings, ARRL or earth-current amateur literature (leads for the Investigator).
• Book figures to check (all unsourced): Jim Creek NLK "10 MW" (already corrected: about 1.2 MW at 24.8 kHz per the mathematician's web check); NAA "2 MW at 14.8 kilocycles"; PANGLOSS 1959, SANGUINE 1962, SEAFARER 1973, operational 1989 at 76 Hz only ([memory, medium] two frequencies, about 45 and 76 Hz, and later dates for the names). Not web-checked here.
• Verdict-scale fix (qe.md): the old single row "conventional HF heater ... Plausible, untested" is split into two (see claim list): hardware and conventional heater Established; author's electrojet-by-elimination purpose Not testable as stated.
• Claim 6 and 7: no verdict changes. Theroux's role as editor and BSRF colleague is recorded in 10.3.

10.6 Changed verdicts in one table

Everything else in sections 1 to 9 stands.

10.7 Added decisive records

Tesla's February 1892 London lecture and the 1893 lectures; Dollard's 1985-86 BSRF videos and any line measurement; Henry's 1842 paper; the 1872 and 1875 Thomson and Houston items; Nipher 1918; the New York Sun and Herald, 13 and 14 March 1895, and any fire or insurance report; the patents De Ment 3,715,596, Eerkens 3,430,046, DeLoraine 2,546,984, Pieckenbrock 3,557,370 and the others in the Chapter 8 list (for the 30 GW and 100 GW figures and Mallozzi's "weld, cut, and melt metals"); the DOE nuclear test table (DOE/NV-209) for Hardtack II, Gnome and Hardhat; the 1977 Seafarer environmental impact statement; McCullough's letter in BSRF holdings and Clark (1982) for Reich's neutralisation; Westford reports (MIT Lincoln Laboratory).

10.8 Corrections to earlier documents (additional; the orchestrator applies them)

1. arbiter.md section 2, claim 4a (my own first draft). The Van Allen statement was wrong and the 43x factor was approximate. Fixed in place (see above). Evidence: Argus altitudes 170, 310, 794 km [web]; outer belt extends beyond 12,000 km [web]. The mathematician's phrase "more than the Earth's circumference" belongs to the Starfish figure (27,000 miles), not to Argus.
2. arbiter.md claim 2, 100 MV label ("Historically superseded but instructive"): reworded in place, as qe.md found.
3. arbiter.md claim 5, verdict row on HAARP hardware and purpose: split in place.
4. arbiter.md claim 4 "All three read the source text": true only for selected pages; the Chapter 4 and Chapter 8 deep reads now exist.
5. claims.md headings "book pp." in claims 1, 3, 4, 6, 7 mean PDF pages (printed page is about twice the PDF page); the arbiter's headings and its source note ("book pp.134-135 of the text file") carry the same labelling. Also add to claims.md claim 1 the note that Tesla's published 1892 and 1901 texts do not contain the shield, microsecond or 10,000 V per inch statements.
6. digests (from the deep reads): digest 02 p.15 line ("Tesla finds rays ... pass through glass and copper shields") is the author's narrative, not a Tesla finding, and calls an article a lecture; digest 02 p.20 reverses Mendeleev (the book says he anticipated the gases would be found); digest 01 p.11 drops "Few" (the book says few of his experiments were conducted without reliable witnesses); the Pieckenbrock number is 3,557,370; the Starfish sentence is on p.89, not p.90; the Teller and "30 Gigawatts" passages are on PDF p.149 and the Mallozzi "100 Gigawatts" on p.150; the Seawater quote is on p.152 (primer cites pp.150-151).
7. facts.json: year 1996 (first edition); chapter entries to add the sourcing findings above (Henry, Thomson/Houston, Dollard, Theroux's dual role, Hodowanec, Piggot/Nipher, Lehr, Napolitano as a private conversation, the Chapter 8 patent table, the book-error list). The "shield" entries should say these are the author's narrative of Tesla, not Tesla's published words.
8. Contrarian and physicist files (mentions of Dollard and Theroux reproductions as "reported"): add that both are published or edited by BSRF, an interested party, with no measurement given.
9. Deep read PDF pp.47-62, 1915 note and memory about the 1943 ruling: the memory that the Supreme Court held Marconi's patents invalid "in favour of Tesla, Lodge and Stone" is too strong; see section 6 (Stone anticipated Marconi; Tesla 645,576 anticipated features; no priority awarded).
10. Deep read Chapter 4 H13 / facts.json: the arbiter's "about 300 miles" for Argus matched only the third shot; restated here as 110 to 493 miles by shot.

10.9 For the primer (no verdict is hidden there)

Fix the gamma quotation page (p.152); mirror the changed rows in 10.6 in the evidence table; add a sentence on the 1895 fire accounts and one on Henry and Thomson with their induction readings; state that the book's shield and pulse-scale claims are not found in Tesla's checked texts; note Theroux's double role; define dB, dBi, NOAA, UAF, ITT, APTI, DARPA, USAF, EMP at first use; and note that Q is used for both quality factor and fusion gain.