QE (Step 4b): Vassilatos-Gerry

QE agent (Opus), 2026-10-08. Inputs read: prompt.txt; _text/manifest.json; all 16 digests (read in full); the deep-read file in full (three deep reads: PDF pp.36-46; pp.121-129; pp.130-138); facts.json (all 9 entries); claims.md; sites.md (log, keywords, Leads for review, corrections); the research/ note list; physicist.md, mathematician.md and contrarian.md (structure, tag tallies and the sections a check turned on); arbiter.md and primer.md in full; example.py (run). For the spot checks I read the extracted text layer (_text/...-3adc7e.txt) directly. Tags: [read] = the extracted book text, with PDF page and line; [memory] = my own recall, with confidence; [calc] = my arithmetic.

Verdict: MINOR

The central verdicts are well supported and unlikely to change: the 60 GW auroral-power thesis fails on three independent bounds, the gamma-ray and Mössbauer claims fail on energy and attenuation, and the Colorado Springs figures are traced to Tesla's own 1904 plan and 1917 speech. The defects are fixable by updating facts, the arbiter and the primer: - two central claims rest on digests plus scattered spot reads, with no deep read (claim 1, pp.9-35; claim 4, pp.70-96 and 138-153); - some Investigator findings were dropped between the reviews and the arbiter and primer (mainly the 1895 fire record); - one factual error in the arbiter (Argus and the Van Allen belts) and a few page-citation slips; - two odd uses of the verdict scale; - several undefined acronyms in the primer.

1. Coverage

• The folder has one document, and nothing was skipped.
• facts.json splits the book into 8 chapter-range entries plus a whole-book entry. Every page range is covered.
• how_read is honest for each entry: "digest; not deep read" where that is true.

No coverage defect.

2. Shortcuts

The method was sound for a large popular book. The chapters were digested page by page, and the two load-bearing sections were deep read: Tesla's Colorado Springs "ether engine" with the Savo car, and the HAARP chapter. The HAARP deep read also checked page images for the figures that the OCR garbled.

Remaining shortcut defects:

1. Claim 1 (Radiant Energy, PDF pp.9-35) has no deep read. - Claim 1 is the physical foundation of the whole book, and the arbiter's split verdict depends on what Tesla's "shields" were (closed? grounded? what metal and thickness?). - The reviewers spot-read pp.14-16 and 28-31 (physicist "[read: pp.14-15, 28-29 of the text file]"; contrarian "[read: text p.14-15]", "[read: text p.15-16]", "[read: text p.31]"). That is partial sampling of a key section. - These items in the chapter were never extracted or weighed: the Henry 1842 and Thomson 1872 shield-penetration precedents (p.15), Dollard's reported 1985 "line interferometry" superluminal result (p.20), Theroux's 12 ft streamers (p.21), Hodowanec's "continues to flow for hours" corroboration (p.31), and the Piggot and Nipher gravity experiments (p.34).
2. Claim 4 (Cold War radiation technology) rests on digests plus spot reads. - The reviewers read pp.89, 91-96 and 150-153 for selected numbers. Nobody deep read Chapter 4 (pp.84-96: EMP history, "Clear Atomic Light", Mössbauer bombs, Reich) or Chapter 8's patent section (pp.146-153) with the question "what does each named patent claim, as the book paraphrases it, against what the book infers?". - The arbiter states that "All three read the source text" for claim 4. That is true only for selected pages.
3. Claim 3 (Marconi, the Tesla Rule, the "Houses", pp.47-70) is digest only. Its verdicts are mostly "Not testable as stated", so the risk is low. But the book's "calculations have proven" that a "Teslian component" carried the 1901 signal (p.50) and its "1915" court award (p.58) were never read at full text.
4. Chapter 6 (VLF/ELF, pp.109-120) is digest only, apart from the Seafarer arithmetic. These are the ELF arguments the author reuses to rebut the HAARP readings in Chapter 7. One testable claim on p.119 ("a few audio-frequency watts" injected into the ground by amateurs disrupted SEAFARER) was never weighed.

3. Missed material

From my reading of the digests: items that facts.json, claims.md, the reviews, the arbiter or the primer missed or underweighted. None is likely to change a central verdict. Several bear on the author's reliability, and several are experiments or people the reader would want in the record.

Experiments and witness claims never weighed: - Joseph Henry (1842: needles magnetised through a building) and Elihu Thomson (1872: sparks through a school building, reported in Scientific American), offered as precedents for shield penetration [read: p.15]. Only the mathematician mentions Henry, once. They are absent from the arbiter and the primer. They bear on claim 1, and both have conventional induction explanations that should be stated beside them. - Eric Dollard's 1985 "line interferometry" showing a superluminal surface component [read: p.20], and his videotaped reproductions [read: pp.25, 28]. The arbiter mentions Dollard only as "reported, no data". The bibliography lists Dollard's 1985-86 videos and books [read: p.155]; these are records to obtain. - Gregory Hodowanec's lasting-current corroboration [read: p.31]; Hodowanec's private publications 1985-86 are in the bibliography [read: p.156]. - George Piggot (silver beads levitated in a positive field) and Francis Nipher's reported gravity reversal in a modified Cavendish balance [read: p.34]. Nipher's "Can Electricity Destroy Gravitation?" (1918) is in the bibliography [read: p.157]. The ether-pressure theory of gravity is a key claim in facts.json, but no reviewer, nor the arbiter, addresses it. - Paneth and Peters (1926) and Helbronner (1920): "extra energy" for hours from hydrogen discharges over palladium or tungsten [read: p.72]. The cold-fusion lineage the author lists (Le Bon, Tesla, Paneth, Moray, Reich, Pons, Fleischmann, Mills, Patterson [read: p.96]; Patterson patent 5,494,559 in the bibliography [read: p.158]) is a cross-project theme for Step 6 and is not recorded. [memory, medium] Paneth and Peters withdrew their helium-from-hydrogen claim in 1927 (helium from the glass); this needs a web check before use. - "A few audio-frequency watts" injected by American experimenters disrupted SEAFARER [read: p.119]. - Turpain's and MacFarland-Moore's devices are said to have set off munitions at a distance; Ulivi's 1914 exploding-bomb experiments (Scientific American, 4 July 1914, in the bibliography); Grindell-Matthews's "death ray" [read: pp.69, 103, 107; bibliography pp.156, 160]. These are on the back-cover list of the book's main topics [read: p.1] and are absent from every review file. They are historical records the Investigator can target. - The claim that Teller's SDI X-ray laser is "identical" to Reiffel's 1957 X-ray source patent 2,797,333 [read: p.149]. It is in facts.json key claims but no review weighs it.

People who matter to the sourcing and are not identified as such: - William Lehr is the author's cited source for several Chapter 1 claims (the 20 ft nitrogen plasma, carbon electrodes, radioactive roof copper) [read: pp.24, 33] and for lethal VLF sparks [read: p.114]. - Michael Theroux both supplied the HAARP files [read: p.3] and edited the bibliography [read: p.154]. He belongs to the publishing foundation (Borderland Sciences Research Foundation), so he is an interested party. The primer names the publisher's interest but not Theroux's double role.

Factual errors in the book that no file records (each [memory, high] unless marked; a web check should retag them before they go into facts.json): - Sputnik I is dated "October 4, 1956" and "(1956)" [read: lines 12805, 13219 of the text file]. The launch was 4 October 1957. - "Plutonium production was stepped up for the 'Little Boy' bomb" [read: p.77]. Little Boy was the uranium-235 gun device; plutonium went into the Nagasaki device. - Mössbauer is called "then a graduate student at Caltech" [read: p.92]. His 1958 discovery was made in Heidelberg, as a doctoral student of the Technical University of Munich. - The multicavity magnetron is placed at Birmingham University "headed by Robert Watson-Watt" [read: p.102]. It was Randall and Boot in Oliphant's group; Watson-Watt led the radar work elsewhere. - Fessenden's 1906 broadcast is placed at "Brant Rock in Nova Scotia" [read: p.56]. Brant Rock is in Massachusetts. - STARFISH (1962) is placed "at 27,000 miles near the North Magnetic Pole" [read: p.89, line 82]. It was about 400 km (250 miles) up, over Johnston Island in the central Pacific. facts.json flags this only as "looks wrong"; the arbiter corrects Argus but not Starfish. - The 1979 "two separate 2 kiloton nuclear shots over the Antarctic" are credited to DSP satellites [read: p.146]. The record usually cited is a single Vela satellite event (the Vela incident, 22 September 1979, South Atlantic/Indian Ocean) [memory, medium-high]. - Galvani's work is dated "1620" [read: p.24, as printed by the digest]. Galvani lived 1737-1798. This may be an OCR or print error.

4. Accuracy spot checks (against the extracted text)

Every load-bearing number checked is quoted accurately. The defects are one wrong gloss and one wrong page.

5. Retention check

The rule: pertinent material from the sources or the Investigator must not be dropped from the arbiter or the primer, whatever the verdict.

Kept well. Lemström; Corliss's ground-level aurora reports and Norwood 1929; the long-delay echoes; Siple-Roberval; the pipeline currents; Tesla's own 1,100 A and 20 MV; O'Neill as the origin of the 26-mile lamp test; Ahlers's 1967 Savo text with its tube-dating anomaly; the Eastlund patent chain and its Tesla-only non-patent citations; the Farnsworth family finding aid; the Reich conviction record; the author's own rebuttals of the HAARP-as-weapon readings. All of these appear in the arbiter and the primer, beside the negative verdicts, stated fairly. The primer's "What the author gets right" paragraph is good practice.

Dropped or thinned (defects, minor): 1. The 1895 laboratory fire. - The book's sabotage-by-dynamite claim [read: p.36] is a pillar of its suppression narrative (claim 3). - The Investigator found the contemporary record: the fire started about 2:30 a.m. on the floors below the laboratory according to the night watchman; oil-soaked premises or a gas jet are proposed causes; Seifer reports suspicion of the watchman; no account mentions dynamite (sites.md Leads 4). - The physicist explicitly lists "the 1895 fire accounts" among the material "kept in the record", and the contrarian weighs them. - The arbiter and the primer do not mention the fire at all. - Note that the watchman suspicion is the one item in the record that leans toward deliberate arson. Dropping the topic removes evidence on both sides. 2. Other Investigator historical findings missing from both the arbiter and the primer: - Tesla's post-fire laboratory at "46 Houston Street", in Tesla's own words, against the book's "40 West Houston"; - Tesla living at the Governor Clinton, not the Hotel New Yorker, around 1931, which bears on the Savo and "Terbo" chronology; - Hammond Castle's statement that Tesla mentored Hammond Jr., and that nothing supports the "permanent guest" claim; - Toby Grotz's "Project Tesla" recreation and his opposing view that Tesla "merely stored energy in the earth". The book itself records this view [read: p.43]; both the physicist and the mathematician cite it. - The arbiter keeps "Terbo" only as open question 10. 3. Ahlers's account details (Savo born in Knin, trained at Wiener Neustadt; the secretary "told General Electric"; a Yugoslav diplomat asking after the car). The arbiter keeps only the Yugoslavia rumour. These are the leads by which the witness could be traced, so they belong in section 4.2.

No item was dropped because a reviewer judged it negatively. The omissions look like condensation. They still remove pertinent historical record and should be restored briefly.

6. Process

• Source tags. Present and honest throughout. [memory] items carry confidence levels, and the arbiter lists which [memory] items stayed unchecked and why no verdict depends on them. The arbiter and the primer label their arithmetic [calc], and example.py reproduces it.
• Web checks.
• The Investigator used 9.7 of 10 minutes, with queries recorded.
• The arbiter used 7 of 8 web calls on verdict-bearing facts (the 1943 opinion, Hirsch neutron rate, NIST attenuation, HAARP ERP, Reich conviction, Rainier; the Gakona dip check failed and is reported as failed).
• The Hirsch 1967 rate rests on a secondary reprint. That is stated.
• The book-error items in section 3 above are [memory] and need checks before they enter facts.json.
• Verdict scale. Mostly consistent, with split verdicts used well. Two uses are off-scale:
• Claim 2b, "100 MV and 10,000 hp as Colorado Springs achievements: Historically superseded but instructive". That label means "a reasonable position once, replaced since". A 1904 plan misread as an 1899 measurement is a documentary misattribution, not a superseded theory. The "Internally inconsistent" half of the split (the book's own 10 kV per inch rule) carries the point. The arbiter should reword it, for example "Internally inconsistent; a misattribution of Tesla's 1904 plan".
• Claim 5a: "HAARP prototype is a conventional HF heater ... hardware description accurate: Plausible, untested (as to the author's intent-by-elimination thesis)". The row mixes an operator-documented fact (the hardware, the conventional heater: on the arbiter's own evidence this reads as Established) with the author's intent thesis (Not testable as stated, or covered by 5g). It should be split into two rows.
• "Established as reports" (claim 6) is a qualifier, not a scale verdict. It is acceptable because it is paired with "Never properly tested" for the physical question.
• Page labelling. claims.md labels PDF page ranges as "book pp." in claims 1, 3, 4, 6 and 7 (for example claim 6 "book pp.130-135" is PDF pp.130-135, printed pp.260-271). Only claims 2 and 5 use true printed pages. The arbiter copies these headings and also writes "book pp.134-135 of the text file". The mathematician mixes "book pp.~30-31" with "PDF p.N". The primer is consistent (PDF pages throughout). Because each PDF page is a two-page spread, the mix can send a reader to the wrong place by a factor of two.
• Primer against STEP 4 and the reader profile.
• All nine sections are present and in order, including the evidence table split by the arbiter's verdicts, 5 check-yourself questions with answers, and further reading that names both mainstream works and the author's own.
• The worked example (the 60 GW ground-screen budget) is clear, uses real numbers, and is fair to the author: it shows his 60 GW is the right size of the natural phenomenon.
• Symbols are defined, apart from these acronyms and units: dB/dBi (decibel, used from section 3.1), NOAA, UAF, ITT, APTI, DARPA, USAF, EMP (first use in section 4) and RCA (in the arbiter, not the primer). "Q" is used both for quality factor and for fusion gain; each is labelled in place, but one sentence noting the double use would help.
• example.py. Runs with the project venv and exits 0. Its output matches every number quoted in primer section 5 and sections 3.1-3.5 (60 GW; 8.514e12 m²; 7.05 mW/m²; 0.342 kW; 1.75e8; 0.661 MW; 1.894 EJ; 58.36 pF; 106 µs; 50.9 dB; 13.8 cm and 10^-960; 11.69 mW; 4.68 lunar distances).

7. Corrections to earlier documents

1. arbiter.md, claim 4a (and Corrections item 8 by implication). The arbiter says Argus's "12,900 miles" "would be outside the Van Allen belts the book places it in". That is wrong. 12,900 miles is about 20,760 km altitude, which lies inside the outer Van Allen belt (about 13,000 to 60,000 km altitude) [memory, high; web check advised]. So the book's statement is internally consistent here; it is wrong against the historical record (the Argus shots were at roughly 200-540 km, about 125-335 miles [memory, medium]). The phrase "wrong by about 43x (about 300 miles)" is approximate: for Argus I at about 200 km the factor is nearer 100. Evidence: book p.89, line 36 ("This placed the blast well within the newly discovered Van Allen Belts"). The arbiter should also add the parallel Starfish error (book: "27,000 miles near the North Magnetic Pole", p.89, line 82; record: about 400 km over Johnston Island [memory, high]).
2. primer.md, section 4 table, gamma row. The seawater quote is cited "[read: PDF pp.150-151]". It is on PDF p.152 (lines 81-82 of that page).
3. digest 09, p.90 line. The STARFISH "27,000 miles near the North Magnetic Pole" sentence is on p.89 (line 82), not p.90. Low impact.
4. claims.md headings for claims 1, 3, 4, 6, 7. "book pp." should read "PDF pp." (or be converted to printed pages: printed = 2 x PDF, roughly). The same applies to the copied arbiter headings and to "book pp.134-135 of the text file" in the arbiter's source note.
5. facts.json "year". Should be 1996 (first edition, ISBN 0-945685-20-3; Adventures Unlimited edition 2000), per the Investigator. The arbiter repeats "about 1996"; the primer already has it right.

8. Remediation (one round)

8.1 Deep reads (most important first)

1. Chapter 1, first half (PDF pp.9-35), Sonnet. Question: For every experimental report on the "Radiant Energy" stream, quote the wording and say whether it is attributed to Tesla's published words (with the lecture or article the text names), to a named witness or replicator (Henry 1842, Thomson 1872, Dollard, Theroux, Hodowanec, Lehr, Piggot, Nipher), or to the author. What exactly were the "glass and copper shields" (material, thickness, closed or open, grounded or not, distance, what was sensed and with what instrument)? List every stated pulse duration with its claimed effect, every number (10,000 V per inch, 100 µs, 1 µs, 12 ft streamers, several hundred yards with no falloff), and every source in the bibliography that matches. Flag where the text contradicts Tesla's 1892 lectures or his 1901 patents 685,957/958.
2. Chapter 4, nuclear hybrids (PDF pp.84-96), Sonnet. Question: Quote every number and named test, project and patent (EMP burnouts, Teak, Orange, Argus, Starfish, "Clear Atomic Light", Operation Distant Plain, the HARDTACK list, the 50/10/40% spectral split, "10 Terrawatt per square meter", PETN dopants, Mössbauer "gamma ray bombs", piezonuclear DU rounds, biosensors, Reich/McCullough neutralisation), and tag each as a patent paraphrase, a public-record test, or the author's inference. Mark each against the bibliography entry it rests on, and list the factual errors (altitudes, dates, test-series membership).
3. Chapter 8, radiation technologies (PDF pp.146-153), Sonnet. Question: For each patent named (Weber, DeLoraine 2,546,984, Chope 3,123,714, Eerkens 3,430,046, Reiffel 2,797,333, Gschleiden/Gschlieden 2,904,692 (spelled both ways in the book), Gould, Pieckenbrock 3,557,370, Viecelli, Wood, Mallozzi, Kell, Jensen), separate what the book says the patent claims from the book's performance claims ("demonstrated their effectiveness", submarines at any depth, orbital gamma lasers, Teller = Reiffel). Quote the passages that cite no source, and check each patent number and date against the bibliography.
4. Chapter 6, VLF/ELF (PDF pp.109-120), Sonnet. Question: Set out the author's ELF arguments (non-directability, standing waves, ground conduction, the 1,400 km dipole, the 7-10 Hz point) that Chapter 7 reuses against HAARP. Quote the claim that "a few audio-frequency watts" disrupted SEAFARER and any source given. Quote every station figure (Jim Creek, Cutler, Seafarer line lengths, 76 Hz, the date "operational 1989") for checking.
5. Chapter 2, Marconi (PDF pp.47-62), Sonnet. Question: Quote the "calculations have proven" passage on the 1901 transatlantic signal and any numbers given; the court award "ratified in 1915" and what proceeding it names; Tesla's "The True Wireless" as quoted; and the list of engineers said to have left Marconi. Tag each item as sourced or unsourced.
6. Chapters 3 and 5 (PDF pp.62-70 and 97-108), Sonnet. Question: Quote the named sources for the "Houses" model (for example "Napolitano") and for the beamray and munition-detonation devices (Turpain, MacFarland-Moore, Ulivi, Grindell-Matthews, Hettinger, McCullough, Reno). Quote the Pacific Scatter, CLOUDS and Westford figures, and say which claims are presented as documented and which as the author's reading.

8.2 What each later file must change

• facts.json:
• year to 1996 (first edition);
• Chapter 1 entry: add Henry/Thomson, Dollard 1985, Hodowanec, Piggot/Nipher and Lehr as cited sources, from deep read 1;
• Chapter 4 and 8 entries: add the book-error list from section 3 above, after web checks, plus the deep-read results;
• Chapter 5-6 entry: add the amateur-disruption claim;
• whole-book entry: note Theroux's double role (supplier of files and editor of the bibliography).
• claims.md: fix the page labels (Corrections 4). Fold in the arbiter's corrections 1, 5, 6 and 7 if the orchestrator has not yet done so. Consider adding the ether-pressure gravity claim (Piggot/Nipher) to claim 1, or as a sub-claim.
• arbiter.md ("QE revision" section):
• (a) correct the Argus and Van Allen statement and add the Starfish error;
• (b) restore the 1895 fire record under claim 3, with the absence-of-evidence analysis: sabotage (book, no source) against an ordinary fire (night watchman, oil-soaked premises, gas jet) against Seifer's watchman suspicion; the deciding records are the New York Sun and Herald of 13-14 March 1895 and any fire-marshal or insurance report;
• (c) restore the Investigator's 46 Houston Street, Governor Clinton, Hammond and Grotz/Project Tesla findings, and the Ahlers tracing details, in section 4.2;
• (d) reword verdict 2b and split 5a (section 6 above);
• (e) weigh any new material from deep reads 1-4 (Henry and Thomson, Dollard, Nipher; the patent paraphrases; the Seafarer disruption claim). No change to the central verdicts is expected.
• primer.md:
• fix the gamma-quote page (p.152);
• define dB, NOAA, UAF, ITT, APTI, DARPA, USAF and EMP at first use;
• add one or two sentences on the 1895 fire record under section 4 or 7 ("What is genuinely open"), and a line on Henry and Thomson as the book's precedents in section 3.1, with their conventional induction reading;
• mirror any changed arbiter verdicts (2b, 5a) in the evidence table;
• mention Theroux's role among the interested parties.
• example.py: no change needed.

Recheck

QE recheck (STEP 4b), 2026-10-08. Checked against the current files by grep and targeted reads: arbiter.md, primer.md, claims.md, facts.json, digest 09, deep/.

Remaining (all minor, none touches a verdict): 1. arbiter.md section 9 (lines 279 and 281) still carries the old 2b label and the unsplit 5a. The primer already notes this. 2. arbiter.md headings in claims 1-6 still say "book pp." for PDF pages. 3. digest 09 still places the STARFISH sentence at p.90 instead of p.89.

Final: OPEN (3 issues)

Orchestrator fix (2026-10-09)

• arbiter.md:279 (section 9 item 2): old "100 MV as measured Historically superseded but instructive and Internally inconsistent"; new "Internally inconsistent (misattribution of the 1904 plan)", matching 10.6.
• arbiter.md:282 (section 9 item 5; the QE text said 281, item 4 is at 281): old unsplit 5a; new bracketed note that 5a is split (hardware Established; electrojet purpose Not testable as stated), matching 10.6.
• arbiter.md lines 29, 45, 78, 92, 111, 139 (Claim 1-6 headings): old "book pp."; new "PDF pp.".
• digest/...09.md:21 to 19: STARFISH (1962) sentence moved from p.90 to p.89, with correction note.

Final: PASS