QE: Linerman-Peter (Peter A. Lindemann / Edwin V. Gray)

Step 4b. Inputs read: _text/manifest.json, all 10 digests, both deep reads, facts.json, claims.md, sites.md, all research/ notes (and spot reads of the saved research/files/rexresearch-evgray-1gray.htm), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. I checked quotes against the extracted text _text/free-energy-secrets-45a60f.txt (line numbers given) and rendered one page image (PDF p.85) to /tmp/qe-linerman/. Page numbers below are PDF pages (the book's printed page numbers run 8 lower).

Verdict: MINOR. The central verdicts are well argued and unlikely to change. The defects are one repeated overstatement of a source, items missed from the folder and from the Investigator's saved files, several pertinent witness and history items dropped between the reviews and the arbiter and primer, the central Lindemann chapter resting on reviewers' ad hoc text reads instead of a deep read, and some process slips. Fix them by updating facts.json, claims.md, an arbiter "QE revision" section and the primer.

1. Coverage

No documents were skipped. facts.json splits the book into its parts (chapters, patents, articles, references), as STEP 1 asks for a bundled PDF. Coverage: PASS.

2. Shortcuts

• Defect (central claim on digest only): Chapters 3-4, pp.39-70. Claims 1 and 2 rest on these chapters: the Electro-Radiant Event list (p.51), the Dollard photograph caption (p.50), the "probable schematic" (Fig. 34, pp.64-66), the "nothing to do with recharging" passage (p.65) and the 99%/100:1 accounting (pp.67-68). facts.json marks them "digest only". The mathematician, physicist and primer re-read parts of the extracted text, which reduces the risk. But nobody checked the schematic figures (pp.53-54, 57, 61-66) on page images, and the reviewers' page citations for this chapter disagree (see section 6). A proper deep read is needed (deep read 1).
• Defect: Gray circuit patents US 4,595,975 and 4,661,747, pp.110-125, digest only. Lindemann's critique of Gray's drawing ("#42 shorted, #28 grid unattached, #36 no coil symbol, #38 no discharge path", p.58-59) is accepted in facts.json, but nobody checked it against the patent drawing (p.115, garbled OCR). Claim 1's "Gray's schematic is a block diagram" depends on that check (deep read 3).
• Defect (moderate): the Vassilatos "Rosetta Stone" chapter, pp.22-38, digest only. It is the stated basis of claim 2 and carries specific claimed experiments: zero secondary current, "10,000 volts per inch", the U-shaped copper bar that lights shorted lamps, pulse-duration thresholds, Henry 1842, Thomson 1872. Nobody reviews these individually (deep read 4).
• Acceptable with caveat: the 1994-95 Borderland articles, pp.141-152. These carry claim 4. The mathematician quoted the 1994 text directly; the 1995 "Self-Acting Engine" article (Doczekal, Schaeffer/Serogodski DE 42 44 016, Wiseman) is digest only (deep read 6, lower priority).
• Acceptable: Tesla patents, pp.125-139, digest only. These are peripheral, except for one missed statement (section 3, item 2; deep read 5).
• Acceptable: Tattler article p.7 (garbled OCR). The same text is reprinted legibly in Chapter 1, pp.9-10 [read: lines 545-594], so the digest's reliance on p.7 is not a gap. facts.json should say so.

3. Missed material

Items in the folder or in the Investigator's saved files that facts.json, claims.md, the reviews, the arbiter or the primer missed or got wrong:

1. The "earlier application" behind US 4,661,747 is printed in the folder. The patent's front page reads "[30] Foreign Application Priority Data: Dec. 16, 1983 [GR] Greece ... 24388" [read: p.117, line 7797-7798]. sites.md, the research notes, the mathematician, the contrarian and the arbiter's open questions all treat the 1983-12-16 priority as "an earlier application not named in the folder". It is a Greek application. This is a new Investigator lead: the Greek patent office (OBI) for application 24388. Gray filing in Greece is unexplained in the record.
2. Tesla's own patent states a faster-than-light speed. US 787,412 (reprinted pp.135-139) gives a wave-train time of "0.08484 of a second ... over the earth's surface with a mean velocity of about four hundred and seventy-one thousand two hundred and forty kilometers per second" [read: p.137, lines 9776-9782]. That is 471,240 km/s, about 1.57 c. Claim 2 (instantaneous or superluminal propagation) is reviewed without this primary Tesla statement. 471,240 ≈ (π/2) × 300,000 km/s, which fits a geometric reading: a disturbance crossing the Earth's diameter at c has a surface-projected speed of π/2 times c [memory, medium; the reviewers should check]. The point is instructive for the primer.
3. Gray's 1980 application is in the saved file at more than abstract length. research/files/rexresearch-evgray-1gray.htm has its description (phases A-C, a 24 V battery, a capacitor charged to "about 2000 volts", the discharge passing through the inductive load and then through "energy recovery battery number 2") [web: https://www.rexresearch.com/evgray/1gray.htm, saved copy]. In the passage I read, the recovery battery sits in the discharge path itself, not only on the overshoot. I saw no measured results there. The mathematician, arbiter and primer list "whether the 1980 application contains measured data" as open, but the file is already on disk and should be read in full.
4. The Crosby/Maize "203 hours" is a projection, not a run. The saved text reads: "The system will operate continuously for 203 hours ... since the total battery power consumed is only 26.8 watts per hour" [web: same, saved copy]. The mathematician ("run time 203 hours") and the primer, Step 9 ("for 203 hours") read it as a measured duration. The arithmetic stands, but the wording should say it was extrapolated from the claimed 26.8 figure.
5. Witness accounts of the 1973 demonstrations. C. V. Wood Jr., president of McCulloch Oil, "looked around for hidden outlets"; Richard Hackenberger ("only 1% of the energy was used - 99% went back into the battery"); Fritz Lens; "two knowledgeable physicists"; and a 15-amp motorcycle battery running six 15 W bulbs, a TV and two radios, with a hand in water beside a lit 40 W bulb [read: pp.9-10, lines 545-640]. The physicist analyses the magnet demonstration. The arbiter has none of it, and the primer reduces it to "showed the press". The arbiter's absence-of-evidence table and the primer's evidence table have no row for the 1973 demonstrations, which are the most widely cited evidence for Gray.
6. Gray biography items dropped after facts.json. None of the five reviews, the arbiter or the primer mentions any of these [read: p.21]: the third EMA prototype "successfully tested for 32 days straight" (an endurance-test claim, untested and unsourced); the February 1976 "Inventor of the Year" nomination by the Los Angeles Patent Attorney's Association; the Reagan "Certificate of Merit"; the letters to every member of Congress offering the technology for SDI with "not a single reply"; the car built with Paul M. Lewis. The Tattler/NewsReal headline "Gasless Auto Gets $6-Million Backing" (in the saved file) is also absent.
7. KeelyNet informant details not carried into the arbiter or primer: a claimed rating of "1,000 HP" at 10,000 rpm (with the page's own 1 hp = 760 W error); "the FBI and others suppressed anything they could"; "one particular guy who gathered lots of investor money for Gray, with as much as 60 million being placed offshore"; Gray "worked for NASA as a mechanic" with Dr. Alexei Popov [web: research/2026-10-06-keelynet-mirror.md]. The $60 million claim bears on claim 5 both ways (fraud by a promoter, and money that disappeared).
8. Records named in the book but not obtained: Borderland Labs videos (1987-88) and "unpublished photographs and unreleased video footage of experiments (1986-1989)" [read: pp.153-154]; Brian Desborough's "incoming spikes over 1000 amps" testimonial [read: p.156]. None is in the arbiter's list of deciding records.
9. Eyewitness bulb-lighting test, 1986. Lindemann held a filament bulb near the coil while Tom Brown touched it, and the filament lit with no shock [read: p.50-51, lines ~2290-2300]. The mathematician explains it as near-field coupling. The arbiter and primer omit the account.
10. The ebook's checkable fraud claim and Lindemann's own anti-charlatan stance. Lindemann names a con man who raised "more than $100,000,000, has been barred from doing business in the State of Washington, has been jailed in California" and lists Keely, Hubbard, Coler and Hendershot as undecidable [read: ebook p.5]. Only the contrarian mentions this. For balance the primer should note that Lindemann himself warns against fraud. The arbiter should list the con-man claim as checkable in public records.
11. Kona, Hawaii heat-pump plant [read: ebook p.3]: by description an OTEC (ocean thermal energy conversion) plant, an ordinary technology Lindemann cites as "free energy". It is an easy example for the primer of a real "control and release" system; optional.

4. Accuracy spot checks

5. Retention

The reviewers kept the Investigator's leads well, including those against Gray (Cronin, AP counts) and those for him (Crosby/Maize, Chalfin and Wester, the anonymous 12-hour run). The arbiter and primer keep both sides on Crosby and Cronin. Defects: - Witness accounts and supportive history dropped. The 1973 demonstration witnesses (C. V. Wood Jr., Hackenberger, the two physicists), the 32-day endurance claim, the "Inventor of the Year" nomination and the SDI letters appear in no review, arbiter or primer (section 3, items 5-6). Most of these items favour Gray, so the omission tilts the record. This looks like condensation, not deliberate exclusion, but under SOURCE HONESTY ("Keep pertinent information") it is a defect. - KeelyNet details (1,000 HP claim, FBI, $60 million offshore, Popov) and the unreleased 1986-89 footage were dropped between sites.md and the arbiter (section 3, items 7-8). - One-way overstatement. The "own build" phrasing (section 4, item 6) strengthens evidence against the claim beyond what the source says. The arbiter calls it "the book's strongest negative evidence". - Single-source evidence is mostly described, not labelled, as required. The primer's "known only from reprints on rexresearch.com" is accurate and acceptable.

6. Process

• Source tags: present in every file. Page-number inconsistency: the digests, deep reads and primer use PDF pages. The physicist and contrarian mostly use printed book pages without saying so (e.g. "p.8" for PDF p.16, "p.59-60" for PDF p.67-68). The mathematician writes "pp.56-60 of the PDF" for printed pages, and the physicist has at least one page that matches neither numbering ("p.66-67" for the Fig. 24 caption, which is PDF p.50, printed p.42). The primer's note on the 8-page offset is good; the arbiter's QE revision should switch to PDF pages.
• Web checks: physicist 5, mathematician 3, arbiter 8, contrarian about 3. These are within budget. The medium/low-confidence [memory] items left unchecked (Wimshurst current, battery voltage recovery, nerve strength-duration) are stated as not load-bearing, and that holds.
• Investigator time rule: sites.md records 472 s of web time (7.9 min), below the 9.5-minute minimum, yet the unreached sites are marked "not reached (time limit)". Either keep searching to 9.5 minutes or mark them "not reached (stopped early)". Fix this in the next run's sites.md entry. The research/files folder holds only the two pages; no patents were downloaded (acceptable).
• Verdict scale: used consistently overall. Departures from the scale: "Contested" (cold fusion; the scale offers it only to the Curator); "Established" for historical record claims (stated openly by the arbiter); the primer's evidence table has "Error of reasoning" for the 99% row. The contrarian's "Internally inconsistent" for the Freedman alloy misuses that label; the arbiter corrected it. The split verdicts are clear.
• Primer: follows STEP 4 sections 1-9 in order, with answers at the end. It defines symbols and acronyms for the reader profile, gives refreshers with textbooks, and includes a worked energy ledger. It adds no new verdicts and repeats the limits of the record. Good.
• example.py: runs cleanly with the venv and ends "All checks passed". Every number quoted in primer section 5 matches its output (15.31 J, 2.9% arc, 6.05 J returned, 278:1, 43.9 min, 495 firings/s, 54 mJ).

7. Remediation (one round)

Deep reads (most important first; slug free-energy-secrets-45a60f for all): 1. pp.39-70, Opus: the central argument of claims 1-2. 2. pp.7-21, Sonnet: Chapter 1 and the biography, for the "build" wording and the demonstrations. 3. pp.110-125, Sonnet: the circuit patents and their drawings. 4. pp.22-38, Sonnet: the Vassilatos chapter's experimental claims. 5. pp.125-139, Sonnet: the Tesla patents, including the 471,240 km/s statement. 6. pp.141-152, Sonnet: the two Borderland articles.

facts.json: - Rewrite the "Lindemann built" wording in the book entry and the Chapter 1 entry to match p.16 (see deep read 2). - Add the Greek priority (24388, 16 Dec 1983) to the US 4,661,747 entry. - Add Tesla's 471,240 km/s statement to the Tesla patents entry. - Note that the p.7 Tattler text is reprinted legibly on pp.9-10. - Add the contents-page "1923" birth year and the 1990 VA death date beside Lindemann's 1925-1989. - Update how_read for the documents deep-read in this round.

claims.md: - Claim 5: replace "grand-theft charge dropped" with Lindemann's account beside the AP account (12 counts filed May 1975; outcome unknown). Give three death years (1989 Lindemann, 20 April 1990 VA record, 1993 KeelyNet). - Claim 1: state plainly that the radiant-energy reading is Lindemann's, and that Gray's patents and 1980 application describe recovery of inductive energy.

arbiter.md, new "QE revision" section: - (a) Correct the "own build" statement wherever it is used (agreement list, absence-of-evidence table, item D). - (b) Answer the open question on the 4,661,747 priority (a Greek application) and add the Greek patent office as a deciding record. - (c) Report what the saved 1980 application text says, and whether it contains any numbers. - (d) Say that "203 hours" is a projection. - (e) Add the 1973 demonstrations and witnesses (Wood, Hackenberger), the 32-day test claim, the Inventor of the Year nomination, the SDI letters, the KeelyNet 1,000 HP, FBI and $60 million statements, and Desborough, each with its source and stake. - (f) Add the unreleased 1986-89 footage and the Borderland videos to the deciding records. - (g) Under claim 2, weigh Tesla's own 471,240 km/s statement. - (h) Under claim 2, weigh the Vassilatos experiments item by item once deep read 4 is done. - (i) Change any verdict that deep reads 1 and 3 overturn (e.g. if Gray's drawing is a working schematic after all). - (j) Use PDF page numbers throughout.

primer.md: - Correct "his own build" / "his own copy" (Safety box, section 4 "Where Lindemann is right", Step 7 note, evidence table, absence table) to what p.16 says. - Add a row to the evidence table for the 1973 press demonstrations (witnesses, what was and was not measured) and one for the 32-day endurance claim. - Add the Inventor of the Year nomination and the SDI letters to the history paragraph. - Change "for 203 hours" to "projected 203 hours". - Mention Lindemann's own warning about charlatans (ebook, Third Force) for balance. - Add Tesla's 471,240 km/s statement as a short worked check in section 4 or 8, with the π/2 × c reading marked [memory] until checked. - Mark "Error of reasoning" as not a scale verdict.

Next Investigator run (sites.md): - Mark unreached rows "not reached (stopped early)" for run 1. - Add the Greek patent office (OBI) for application 24388. - Read the saved rexresearch file in full: the 1980 application, the EMA-4 data sheet, the March 1974 Tattler and the "$6-Million Backing" article. - Open the KeelyNet 1990-91 files.

Then a QE recheck (Sonnet) appends a "Recheck" section here.

Recheck

Step 4b recheck. Checked the current facts.json, claims.md, arbiter.md ("QE revision" section), primer.md, sites.md and deep/ against the defects above, by searching the files for each item.

Deep reads. Fixed. deep/ now holds a deep read for the book, and facts.json how_read lists deep reads for pp.7-21, 22-38, 39-70, 74-81, 110-125, 125-139 and 141-152, with page images checked for Fig. 34, the circuit patent drawings and Tesla p.136. Only the reference and back-cover pages 153-156 stay digest only, as claims.md says.

Section 4 item 6, "Lindemann built a copy" overstatement. Fixed in arbiter.md (new section (a), agreement list, absence table) and in claims.md and primer.md (no "own build" wording remains). Not fixed in one place: facts.json line 178 (US 3,890,548 entry, "relevance") still reads "the one Lindemann built from in the 1970s". The physicist, mathematician and contrarian reviews also keep the old wording, but the arbiter's correction covers them.

Section 3 missed material 1. Greek priority 24388: fixed (facts.json, arbiter, primer, sites.md). 2. Tesla 471,240 km/s: fixed (facts.json, claims.md claim 2, arbiter, primer). 3. Gray 1980 application read in full: fixed (arbiter (c)). 4. "203 hours" is a projection: fixed (arbiter (d), primer). 5. 1973 witnesses (Wood, Hackenberger): fixed (facts.json, claims.md, arbiter, primer). 6. Biography items (32-day test, Inventor of the Year, SDI letters): fixed. 7. KeelyNet 1,000 HP, $60 million, FBI: fixed (arbiter, primer). 8. Unreleased 1986-89 footage, Desborough: fixed (arbiter, primer, facts.json). 9. Tom Brown bulb test: fixed (arbiter, primer, facts.json). 10. Con-man claim and Lindemann's anti-charlatan stance: fixed (arbiter, primer, claims.md). 11. Kona OTEC example: not added. It was marked optional, so it does not count as a defect.

Section 2 shortcuts, section 5 retention, section 6 process - Shortcuts (Chapters 3-4, circuit patents, Vassilatos, Tesla, Borderland): fixed by the deep reads. - Tattler p.7 reprint note: fixed (facts.json). - Birth year 1923/1925 and VA death date: fixed (claims.md claim 5, facts.json). - Claim 5 grand-theft wording and the three death years: fixed. - Claim 1 radiant-energy reading marked as Lindemann's: fixed. - Page numbers: arbiter's revision uses PDF pages: fixed. - Primer "Error of reasoning" row marked as not a scale verdict: fixed. - sites.md rows 17-22 still read "not reached (time limit)" although only 7.9 minutes were used. Remaining. The QE assigned this to the next Investigator run, so it is outstanding but not an error in the deliverables.

Remaining 1. facts.json line 178 still says Lindemann "built from" the motor patent. 2. sites.md still labels the unreached sites "not reached (time limit)".

Both are small. No verdict changes and no source misstatement reaches the arbiter, claims.md or the primer.

Final: OPEN (2 issues)