StClairJohn: Quality Engineering (Step 4b)

Date: 2026-10-04. QE agent (Opus). Inputs read: _text/manifest.json, _textprep.log, all five digests, all five deep reads (searched in full; 0164824 and 0197093 read in their relevant sections), facts.json, claims.md, sites.md, all research/ notes (research/files/ is empty), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py (run).

Tags: [read] = the extracted text or a page image I rendered (pdftoppm, written to /tmp/stclair_qe); [calc] = output of example.py; [memory] = recalled, with confidence.

Verdict: MINOR

The five documents were all digested and deep-read. The reviews are careful and agree with each other, and the arbiter and primer are sound and follow the prompt. The five load-bearing facts I checked against the source are correct. No central verdict is likely to change. The defects are in missed material and retention. Several stated claims in the inventor's own text never got past the digest: the 19.47° planetary-volcano claim, the "negative mass cools a hot surface" property, and the list of applications he claims. The Investigator also missed one related-work lead (Hoagland's 19.5° "hyperdimensional physics") and one family-member lead (the "Dual Rotor Propulsion System" application). Some Investigator leads were dropped from the arbiter and primer. Corrections that the arbiter asked for have not yet been applied to facts.json and claims.md. All of these can be fixed by updating facts, claims, arbiter and primer.

1. Coverage

Nothing was skipped. The single OCR page (0164824 p.1, header) is garbled in the text layer [read]. The deep reader repaired it from the image.

Outside the folder: the sister applications US 2004/0200925, 2004/0102810 and 2007/0285325 hold the inventor's only experimental narratives. They are known only through the Investigator's notes from FreePatentsOnline. No PDF was saved to research/files/, because downloads needed the owner's go-ahead. This is not a coverage defect, since they are not manifest documents. But every later file now reports the experiments second-hand (see Remediation R7).

2. Shortcuts

No unacceptable shortcuts were found. Every central claim rests on a deep read, and the equations on figure sheets were read from page images. The mathematician did not re-render images and relied on the deep readers' transcriptions, which that review states openly. I confirmed two of those transcriptions from images myself (section 4).

Digest errors were caught downstream, as the process intends. I record them here so nobody relies on the digests for these points: - Digest of 0164824 p.17 gives the planar tetrahedron sides as {√1, √2, √5}. The page image reads {√1, √2, √3} [read: page image p.17, para 0003]. The second deep read, facts.json and the physicist all use √3 correctly. - Digest of 0145019 p.10 gives a force formula "f = (1/2c²)ε₀(E·E_W)E_W". No such formula is in the document. The page image shows only *d(E ∧ (B_w + E_w ∧ dt)) ε/c = force/volume [read: page image p.10, para 0009]. The deep read, mathematician and arbiter all say so. The digest invented the formula. - Digest of 0164824 p.19 gives "Planck wavelength (2√5·λ)". The text layer has "275A", which is most likely 2πΛ (the mathematician's reading). The reviews used 2πΛ. - Digest of 0164824 p.1 names the inventor "Sal (John) St. Clair". This is OCR garble of "St. Clair" [read: text p.1]. facts.json is correct.

3. Missed material (from my own reading of the digests and text)

M1. 19.47° planetary claim and Hoagland (related work). 0197093 para 0028 and claim 16 say the tetrahedron's corners touch the sphere at 19.47122063°, "the location where all the large volcanoes and vortices occur on Earth, Mars, Jupiter, Uranus and Neptune" [read: text pp.10, 12]. The digest has this, but it is absent from facts.json, claims.md, all three reviews, the arbiter and the primer. It matters for three reasons: - It is the one factual, checkable claim about the world in the 0197093 filing. A reader can check it from published latitudes of Olympus Mons, Hawaii, the Great Red Spot and the Great Dark Spot. I do not give the latitudes from memory as a finding, so the check is left to the remediation. - It is the signature claim of Richard C. Hoagland's "hyperdimensional physics" (19.5°, circumscribed tetrahedron, planetary "energy upwellings"), from the 1990s [memory, high that Hoagland made this claim]. That is plainly priority-3 related work, and it may also be a source for St. Clair's tetrahedron diagram (1996). The Investigator did not search for it. - The 2004 copyright "Mars Experimental Planetary Institute--MEPI" points the same way (a Mars interest). This is a lead, not a finding.

M2. The inventor's claimed properties and applications (0197093 paras 0003, 0029-0031 and claims 6-13) [read]: - negative mass "projected onto a hot surface, the surface will cool rather than heat up" (para 0003), and a refrigeration claim (claim 8); - power supplies from "differential spacetime curvature on carbon electrode ends", like a chemical battery (claim 9); - a wormhole attached to a magnetic monopole that gives a radial field and lift through G_zz (claim 13); - surgery-less medical tables, cold-welded crystals, folding-space waveguides, teleportation over light-years, and inertia-less craft at "hundreds of thousands of g's" (claims 6, 10-12).

facts.json lists these in key_claims. claims.md, the arbiter and the primer leave them out. The physicist and contrarian mention only the g-load and teleportation in passing. These are the inventor's stated claims and belong in the record next to the verdicts. The hot-surface cooling claim is also a concrete, cheap test the reader could imagine, which the reader profile and the Curator value.

M3. "Dual Rotor Propulsion System". 0230675 para 0008 says: "From my previous patent application Dual Rotor Propulsion System ..." [read: text line 286]. This title is in neither the Investigator's 16-item family list nor the Justia list. It is a missing family member, or one that was never published. Investigator lead.

M4. The vector potential as the primary field. 0197093 para 0024 and 0230675 FIG. 5 and Point 3 rest on the claim that the vector potential A is "a more important field than the magnetic field". They cite Bohm-Aharonov and Feynman, and the design uses non-linear winding "to preserve the vector potential" [read]. The Aharonov-Bohm effect is real and measured [memory, high]. That makes this a real teaching point on which the inventor is partly right, and the primer does not discuss it. Missing from the arbiter and primer. The magnetron and Bell System Technical Journal inspiration (0230675 para 0007) and the SuperMalloy saturable-choke drive circuit (0197093 paras 0038-0040) are also missing from the primer. They are minor, but they are the only hardware description given.

M5. Examiner citations. One of the six examiner citations on 0197093 is Paramount Pictures' design patent "Toy spaceship" (USD260789S) [web, per research/2026-10-04-google-patents.md]. It is minor, but it is part of the historical record of how the application was treated. The arbiter lists the other examiner citations and leaves this one out.

4. Accuracy spot checks

The worked-example numbers in the primer match example.py output line for line: 79.7 and 6.34 N/m², 8.27e-42 m⁻², 1.117e9, 895, 4.45e8 J, 3.34 µN/kW, 2.94e12 W, the cgs base -84.40, and 2,780,590 [calc].

5. Retention check

Rule: pertinent material from the sources or the Investigator must not be dropped from the arbiter or primer because a reviewer judged it negatively.

Retained well: - The sister-application experiments: the Chakra Vortex Accelerator arm report, the bifilar coil foot sensation and rocking, and the pendulum data at 20-120 kHz and 2.78 MHz. Both the arbiter and the primer keep them, quoted, with conditions, competing explanations (Kohnstamm, electrostatic force) labelled as such, and a cheap blind test proposed. This is good practice. - The Horsburgh observation, the copyright series, the lightning anecdotes, Jefimenko, Pais, examiner prior art, commentary labelled as commentators' characterisations, and the absence-of-evidence analysis with five explanations. All handled correctly.

Defects (minor; none appears to be a deliberate suppression): - R-a. The inventor's applied claims (M2) are dropped from the arbiter and primer. They are pertinent claims, and they were left out without a verdict or a mention. - R-b. Investigator leads not carried into either file: - "Mars Experimental Planetary Institute (MEPI)" (2004 copyright): a name lead for the absence-of-evidence analysis; - the "tet0565 ... stored in the Library of Congress" statement (the inventor's own pointer to the deposits); - the 2005 HRI news issue "Planck box corner at the Cabibbo angle" (a further tetrahedron-diagram claim); - Egely (2024) and Díaz (2016): in the arbiter only, not in the primer's further reading or record.

The Investigator's caution that same-name patents (port security, mailbox, sash clip) may be another person is properly absent from the attributions. That is fine. - R-c. The primer's "When the record goes dark" omits the arbiter's note that 0197093's abandonment date and its "non-response" reason come from a fetch-tool summary of Google Patents, a single source. The arbiter states this; the primer drops the caveat by omitting the item. It is acceptable, but the arbiter's wording "[web: patents.google.com]" for a reason that the contrarian says came from a small model's summary of the page needs a direct check (see R6).

6. Process

• Source tags. Present and honest throughout. The arbiter and primer add a [calc] tag and define it; this is acceptable. The second-hand status of the sister-application material is stated everywhere.
• Web checks. These were weak, through no fault of the reviewers: the physicist and contrarian report the search quota exhausted, the mathematician made 2 checks and the arbiter 3 fetches. Medium- or low-confidence facts left unverified: the Witten attribution, the content of the Horsburgh thesis, the old CODATA h value, lifter and EM-drive outcomes, and the abandonment reason. The arbiter correctly keeps the lifter and EM-drive outcomes out of every verdict. None of the open items carries a verdict alone. One error: the arbiter's fetch summary said Hau's slow light used strontium. The arbiter's own [memory, medium] sodium Bose-Einstein condensate is right [memory, high: Hau et al., Nature 397, 594 (1999), sodium BEC], and the primer cites it correctly.
• Verdict scale. Applied consistently. Split verdicts state each part. The arbiter reconciled the three label differences (Claim 3 electron = proton; Claim 4; Claim 5) with reasons.
• Arbiter corrections not yet applied. The arbiter's "Corrections to earlier steps" list three changes. facts.json/claims.md should note the sister-application experiments and the Ω value. The G_tt spike location should be corrected (the spikes are near θ = π/4 and 5π/4; at those angles G_tt = 0 for r > 0). claims.md still says 0164824 holds "the one experimental statement in the corpus", and facts.json still says G_tt "spikes at theta = pi/4, 5pi/4".
• Primer and STEP 4. All nine sections are present, in order, with answers at the end. The primer runs example.py and defines the general-relativity pieces from scratch (2.7). It meets the reader profile with small gaps. These are used without definition:
• "Faraday tensor" (Section 4 table, 0145019 row);
• "deep inelastic scattering", "lepton number" and "baryon number" (Sections 4 and 6);
• "Reissner-Nordström";
• "Einstein tensor component G_zz / G_tt" as specific entries (only the table as a whole is defined).

Textbook chapters (Griffiths 2, 4, 7, 8, 9; Taylor 5; Feynman II-27) match my recollection [memory, high]. - example.py. It runs cleanly in the venv (exit 0) and reproduces every number in the primer.

7. Remediation actions

Deep reads (most important first; slugs from manifest.json):

1. US20030197093-09173a, pp.8-12, Sonnet. Quote exactly every stated property and application claim: hot-surface cooling (para 0003), refrigeration (claim 8), the carbon-electrode power supply (claim 9 and para 0030), monopole lift through G_zz (claim 13), medical tables, cold welding, folding-space waveguide, teleportation, inertia-less g-loads, and the 19.47° "volcanoes and vortices" statement (para 0028, claim 16). For each, say whether any number, measurement or test is given, and what observation would test it.
2. US20030230675-ed3d2e, pp.10-12, Sonnet. Quote the "previous patent application Dual Rotor Propulsion System" reference (para 0008) and any other cross-references to the inventor's filings. Set out the inventor's argument that the vector potential is the primary field (Bohm-Aharonov, Feynman formula, the magnetron cavity analogy), separating what is textbook from what is added.
3. US20030197093-09173a, pp.9-11 (paras 0024-0026, 0038-0040), Sonnet. Give the non-linear winding "to preserve the vector potential" reasoning and the drive circuit (SuperMalloy choke, step-up transformer, <20 MHz, quantized hyperspace frequencies) as a buildable description. List the quantities that are missing.
4. US20030209637-f6bf67, pp.6-8, Sonnet. List every cross-reference to "related patent applications" and to the "enclosed reference calculation", and check whether any filing number or title is given (feeds the Investigator's family list).

Investigator, next run (incremental): - R1. Search Hoagland "hyperdimensional physics" 19.5° tetrahedral and record it as priority-3 related work. Search "St. Clair" with Hoagland, "19.47", and "Enterprise Mission", and check for a link to the 1996 tetrahedron diagram. Search "Mars Experimental Planetary Institute". - R2. Search "Dual Rotor Propulsion System" St. Clair (Google Patents, Justia, the USPTO application number). - R3. Download the sister-application PDFs (0200925, 0102810, 0285325) into research/files/ with the owner's approval, so that the experiment narratives are read first-hand. Then a deep read of the experiment paragraphs, Sonnet. - R4. Check the planetary latitudes of the features St. Clair names (Olympus Mons, Hawaiian hotspot, Great Red Spot, Neptune's Great Dark Spot) against published values, so the 19.47° claim can be weighed [web]. - R5. Carry forward the not-reached sites: the USPTO Patent Center file wrappers, the Puerto Rico registry, the Horsburgh thesis, and Copyright Office name searches. - R6. Verify the 0197093 abandonment date and reason directly on the Google Patents page or in Patent Center.

facts.json / claims.md: - Apply the arbiter's three corrections (sister-application experiments, Ω = 1.346812891e27 kg/m from 0285325, G_tt spike location). - Add to the 0197093 key_claims: the 19.47° planetary claim and the hot-surface cooling property. - Add a claims.md note that the applied claims (M2) and the 19.47° claim are part of the inventor's case. They may stay outside the 3-7 central claims, but they should be listed under "other stated claims". - Record the "Dual Rotor Propulsion System" reference.

arbiter.md (add a "QE revision" section): - No verdict changes expected. - Add the M2 applied claims to the record. Each takes "Never properly tested" or "Not testable as stated" as the arbiter judges. Hot-surface cooling is testable as stated; the device was never characterised. - Add the 19.47° planetary claim, with its verdict once R4 is done, and Hoagland as related work. - Add MEPI, tet0565 and the Cabibbo-angle HRI news issue to the record, and the Dual Rotor lead to the absence-of-evidence records list. - Note the Hau medium correction (sodium BEC). - Restate the abandonment-reason source as single-source and unverified until R6.

primer.md: - Add a short "Other things the patents claim" list (M2) with the arbiter's labels. - Add the 19.47° claim as a "check it yourself" item with Hoagland as context. - Add a paragraph on the vector potential and the Aharonov-Bohm effect (something the inventor got partly right) to Section 2 or 3. - Define the Faraday tensor, deep inelastic scattering, and lepton and baryon number on first use. - Add Egely (2024) and Díaz (2016) as secondary mentions under Further reading, and MEPI to the dark-record section.

Recheck

Date: 2026-10-04. Re-read facts.json, claims.md, arbiter.md (with its new "QE revision" section) and primer.md against the defects above. Tags as before: [read] means checked in the current files.

Not done, and outside what this step can fix: the Investigator's incremental run (R1 Hoagland and MEPI search, R2 Dual Rotor search, R3 download of the sister-application PDFs into research/files/, which is still empty, R5 not-reached sites, R6 abandonment check). These are open leads that the arbiter and primer label as unverified or second-hand. No verdict depends on them alone.

Final: PASS