SilverTooth: QE (Step 4b)

Date: 2026-10-04. Inputs checked: _text/manifest.json, _textprep.log, digest/ (1 file, read in full), deep/ (1 file), facts.json, claims.md, sites.md, research/ (8 notes, 2 saved PDFs), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. I also read the extracted text of the project paper in full, rendered page 3 of the source PDF as an image (Fig. 2 and the D2 sentence), and checked key passages in the saved 1976 Applied Optics paper (research/files) against its text layer.

Verdict: MINOR. The work is complete and accurate on everything load-bearing. The central calculation (Claim 3: the first-order signal cancels under the paper's own eq. 5) holds up when checked against the drawn geometry of Fig. 2. No verdict needs to change. The defects are retention gaps (pertinent items from the Investigator's findings left out of the arbiter and primer), one missed internal inconsistency between the 1976 and 1986 papers, stale entries in facts.json and claims.md, and a few tagging slips.

1. Coverage

There is one document in the manifest, it has one chunk, and nothing was skipped (_textprep.log). There is no coverage defect. The two PDFs the Investigator saved (1976 AO paper, 1983 AO sensor paper) are not in the manifest. The reviewers read the 1976 paper directly, with tags and page numbers, which is acceptable.

2. Shortcuts

• The only document has a full deep read, with page images where the equations were garbled. No central claim rests on the digest alone. There is no sampling. No defect.
• The 1989 Electronics & Wireless World article (the inventor's own statement of "within ±5%" repeatability and the GE Syracuse sensors) was read only by the Investigator, from screenshots in Chrome's viewer. It was not saved. Every later step depends on that transcription. This is acceptable for a supporting item, but the PDF should be saved (owner approval) so that it can be re-read.
• The arbiter checked Marett's replication only through a search summary. All the detail (wrong sign, solar-time pattern, temperature correlation) is the Investigator's reading. The arbiter and primer both say this openly. That is acceptable, but Marett's page is the most important outside evidence on Claim 4. A direct read of it should be the first web task in the next run.

3. Accuracy spot checks (against the extracted text and page images)

example.py runs cleanly in the .venv ("All checks passed"). Its outputs match the numbers in the primer: 377.7/378 km/s, 793.65 cycles, a 205.00 nm residual shift, 0.125 mm and 189 km/s for the same-sign geometry, 2.9°, 3.2e-6, and 3,777 times the Krisher limit. One small mismatch: the arbiter's "acceleration turns about 3.5 times in flight" comes from the Mathematician's parameters (d = 10 kpc, v = 200 km/s). example.py uses ω = 1e-15 rad/s and gets 5.3. Both are right for their own inputs, but the primer's evidence table states neither number.

4. Missed material

1. The 1976 and 1986 papers contradict each other on premise A1. No step noticed this. In 1986 the paper says frequency "cannot change", so a change of light speed appears only as a change of wavelength [read: p.1]. In 1976 Silvertooth wrote: "The cause of the phase shift has been experimentally demonstrated by the writer to be a change in the frequency and the velocity, the change in wavelength being a resultant effect" [read: 1976 saved PDF, journal p.1101]. That was his reading of his own two-laser null. He therefore reversed his position on what changes between 1976 and 1986, and the 1986 paper does not mention the earlier experiment. The Contrarian paraphrases the 1976 sentence, but no one sets it beside A1. This belongs in arbiter Claim 1/2 and in primer section 4 or 7 as an instructive point.
2. The 1976 null was a different setup. It used two free-running lasers whose beat frequency was "held under 3 MHz", read at two photocathodes 50 cm apart [read: 1976 text layer]. It was not a single laser split on a slide. The primer's sentence "His 1976 paper reported exactly such a null" overstates the match. The Mathematician correctly hedges ("unless the 1986 apparatus differs in an essential way"). The primer should say "a related null" and name the difference.
3. Turner-Hill (Mössbauer) and Cole (pulsar). The paper says both are "defeated through the second order by the argument of Tyapkin" [read: p.2-3]. This is in facts.json key_claims, but no reviewer, nor the arbiter or primer, assesses it. Points worth one sentence: the paper gives no argument. Tyapkin (1973) is the conventionality paper that Marett uses against a one-way wavelength measurement, so the citation cuts against Claim 3. Turner-Hill was a first-order one-way-type null.
4. The Whitney & Silvertooth 1992 Physics Essays experiment (a standing-wave nodal spacing inside one Michelson arm departing from λ/2; fitted v0 ≈ 378 km/s with θ fixed to the stars [web: naturalphilosophy wiki abstract, via Investigator]). This is the inventor's own later positive experimental claim. The arbiter and primer list it only as "to obtain" or under further reading. The Mathematician's remark belongs beside it: in the model of eq. 5 the node spacing is exactly λ/2. That is the identity of primer Check-yourself Q2, so the 1992 claim also conflicts with the 1986 model. The point is directly instructive and is missing from the primer.
5. Silvertooth's 1972 JOSA "Isotropy of C": a one-way test with frequency-doubling crystals that Roberts & Schleif record as a null consistent with SR [web: https://math.ucr.edu/home/baez/physics/Relativity/SR/experiments.html]. This is his own earlier result against the effect. The arbiter mentions it only in the "obtain" list. The primer has it only in further reading, without saying it was a null.
6. Marett's companion single-laser one-way standing-wave experiment (2011, null) [web: conspiracyoflight.com, via Investigator]. It appears in the Physicist, Mathematician and Contrarian reviews, but not in the arbiter or primer.
7. Gagnon, Torr, Kolen & Chang, Phys. Rev. A 38, 1767 (1988): a guided-wave first-order one-way test, cited by the Contrarian as the strongest mainstream evidence against Claim 3. The arbiter's web checks list it as "title only". The primer leaves it out. It should be stated with its [web] tag, and the result should be marked as not read.
8. Smaller leads absent from both arbiter and primer: "The Silvertooth effect: TETET-99" (1999, UNIAM, ISBN 978-0968308547; authorship unknown); the 1989 statement that groups in the USA, West Germany, UK, Italy, France and Austria worked on explaining the result; Wireless World Nov 1986 and Nov 1989 coverage and the Essen items; the unsourced wiki statement of classified Air Force think-tank work (Aerospace Corp.). The last one bears on the sponsorship absence-of-evidence analysis, and the arbiter mislabels its source as a "Wikipedia-type biography" when it is the Natural Philosophy Wiki.

5. Retention check

Question: was pertinent material from the sources or the Investigator left out of the arbiter or primer because a reviewer judged it negatively?

• Kept (good): Marett's result with both halves (the signal reproduced; the thermal attribution); Marinov 1987 and Wesley 1987, with their stakes stated; the 1976 null; the 1983 sensor; the Academy Award and Caltech record; the 1989 "±5%" repeatability statement; the RADC/DARPA absence-of-evidence analysis with four competing explanations and the records that would decide between them; the Arizona OSC experiment "under way" with no outcome; the critiques of Sherwin, Phipps, Newburgh and Grøn; the unsigned suppression sidebar, reported as its writer's characterization; Navia 2006; the point that laser frequency drift is far too small to fake the effect (in the inventor's favour); and the geometry fit of the E-W maxima at Leo rise and set.
• Left out (defects, minor): the items in section 4, numbers 4, 5 and 6 (the 1992 Whitney-Silvertooth positive claim, the 1972 JOSA null, Marett's single-laser null), plus Gagnon 1988 and TETET-99. None of these omissions looks driven by a negative judgement. Items on both sides were condensed (a positive claim by the inventor and two nulls). They are condensation losses, not suppression. Still, the rule is that pertinent material stays in the record, so each one should be restored with one line.
• Not a retention issue but worth noting: the arbiter gives no verdict-scale label for the middle part of Claim 4 ("signal partly reproduced"). It should either carry a scale label or be stated explicitly as an observation, not a verdict.

6. Process

• Source tags: present and mostly honest throughout. Slips:
• The primer tags the 1989 EWW content "[read via Investigator]" (evidence table, Claim 4). It should be [web: https://spirit-science.fr/ArchivesScientifiques/1989Silvertooth.pdf], via Investigator.
• The arbiter calls Speculations in Science and Technology "a fringe venue [web via Investigator]". That is a characterization, and no source states it in those words. Drop it or attribute it.
• The Mathematician tags Muller's "about 390 km/s" as confirmed [web], but the Contrarian found the number only in a search summary. The arbiter resolved this conservatively (it stays [memory, medium]), and the primer avoids the number. Acceptable.
• Reviewer error caught downstream: the Contrarian says the 1976 Doppler factor (1/γ)(1 ± v/c) "equals" the standard √((1−b)/(1+b)). It does not: 0.433 against 0.577 at b = 0.5, per the Mathematician and arbiter item 5. The arbiter's recomputation implicitly supersedes the Contrarian, and the primer does not repeat the error. No action beyond noting it in the arbiter QE revision.
• Deep read error caught downstream: the deep read's reconstruction of δ = λ(v/c) subtracts the phases, which is wrong for the described geometry. The Mathematician flagged it, and the arbiter and primer (step 7) handle it correctly. facts.json does not repeat it. No action needed.
• Web checks: done with URLs. These were checked: CMB dipole, Krisher 1990, Gunn-Tinsley, the existence of the Muller article, and the existence of Marett's page. These remain [memory, medium] and are stated as such: the CMB dipole declination of about −7° (the geometry check depends on it, but no verdict does), Riis 1988 and Muller's numerical value. Acceptable.
• Verdict scale: applied consistently, apart from the unlabelled middle part of Claim 4 (section 5). The split verdicts are well motivated.
• Primer vs STEP 4: all nine sections are present and in order. Symbols are defined, the reader profile is respected (no tensors; SR introduced from the Lorentz transformation), the worked example uses real numbers, and the answers are at the end. Further reading is good. Pass.
• example.py: runs, and its assertions pass.
• facts.json and claims.md are stale:
• The journal is now known (Speculations in Science and Technology 10(1) 3-11 [web: Marett page via Investigator]).
• Ref. 8 is p.454 [web: https://www.nature.com/articles/257454a0].
• Muller's article is verified to exist [web: ADS].
• v̄ is the acceleration v̇ per the 1976 paper.
• claims.md Claim 2 says "(first order in v/c)", but eq. 5 is exact by construction (Mathematician correction 3).
• claims.md does not mention the inventor's later material (1976, 1989, 1992).

7. Remediation (MINOR)

Deep reads: none required. The only manifest document is fully deep-read, and the Fig. 2 geometry is confirmed above. Optional, not a deep read of a manifest document: the person updating the arbiter should read journal p.1101 of the saved 1976 PDF for item 4.1.

facts.json - Add the journal (SST 10(1) 3-11, [web] via Investigator, single secondary report). - Correct ref. 8 to p.454 [web]. - Retag Muller as existence verified [web: ADS], value still [memory, medium]. - Note v̄ = v̇ (radial acceleration) per the 1976 paper [read: research/files]. - Note the 1976 vs 1986 reversal on whether frequency or wavelength changes.

claims.md - Claim 2: replace "(first order in v/c)" with "exact by construction from eq. 3; only eq. 4 is first order". - Add one line after the source note: the inventor's related work (1972 JOSA null as recorded by Roberts & Schleif; 1976 AO; 1989 EWW; 1992 Physics Essays) exists outside the folder (Investigator).

arbiter.md: add a "QE revision" section 1. Claim 3: the Fig. 2 page image confirms the opposed-direction geometry (QE spot check 4), so the "geometry caveat" is narrower. The verdict stays Internally inconsistent. Keep the revert condition. 2. Claims 1-2: record the 1976 statement that frequency and velocity change, with the wavelength change "a resultant effect", against 1986's fixed ν (A1). It is an internal inconsistency across the inventor's own papers. No verdict change. 3. Claim 4: give the middle part a scale label, or mark it explicitly as an observation. Add Marett's 2011 single-laser null and the 1972 JOSA null (as recorded by Roberts & Schleif) to the evidence. Add the 1992 Whitney-Silvertooth nodal-spacing claim as the inventor's own later positive claim, with the note that in the eq. 5 model the node spacing is exactly λ/2. 4. Claim 2/3: one sentence on the Turner-Hill and Cole dismissal (no argument given; Tyapkin's conventionality argument cuts against Claim 3). One line on Gagnon et al. 1988 [web], result not read. 5. Fix the source label "Wikipedia-type biography" to Natural Philosophy Wiki (unsourced). Drop or attribute "fringe venue". Note that the Contrarian's statement on the 1976 Doppler factor is superseded by recomputation 5. 6. Retention note: add TETET-99 (1999) and the 1989 list of countries as open leads.

primer.md - Section 5, "What the example teaches": change "reported exactly such a null" to "reported a related null" and name the difference (two free-running lasers, beat frequency under 3 MHz, two detectors 50 cm apart). - Section 6 evidence table, Claim 4 row: retag the 1989 item [web: spirit-science.fr URL]. Add Marett's 2011 single-laser null and, under Claim 2 or 4, the 1972 JOSA null (as recorded by Roberts & Schleif, not read). - Section 6 notes or section 7 "Genuinely open": add the 1992 Whitney-Silvertooth claim of a departure from λ/2 nodal spacing, and link it to Check-yourself Q2 (the eq. 5 model predicts exactly λ/2). - Section 7 "Wrong, and why it is instructive": add one bullet on the 1976 to 1986 reversal (frequency changes vs frequency fixed). - Section 4 or 7: one sentence on the Turner-Hill and Cole dismissal. - Optionally, in the Claim 5 row, state the number of orbits during the light flight with its parameters (example.py: 5.3 at ω = 1e-15 rad/s).

Next Investigator run (not part of this remediation) - Read Marett's page and PDF directly. - Save the 1989 EWW PDF (owner approval). - Obtain the Nature 322:590 letter, JOSA 1972, Physics Essays 1992, Galilean Electrodynamics 1990 and Sherwin 1989. - Run a direct DTIC search for the RADC/DARPA report trail.

Recheck

Date: 2026-10-04. I re-read facts.json, claims.md, arbiter.md (including the new "QE revision" section) and searched primer.md for each remediation item.

facts.json - Journal (SST 10(1) 3-11, [web] via Investigator, single secondary report): fixed. - Ref. 8 corrected to p.454 [web]: fixed. - Muller retagged (existence verified [web: ADS], value still [memory, medium]): fixed. - v-bar = v-dot per the 1976 paper [read]: fixed. - 1976 vs 1986 reversal on frequency vs wavelength, with the different 1976 setup: fixed. - Also added: the Turner-Hill/Cole/Tyapkin note and the 1972 JOSA null, both tagged.

claims.md - Claim 2 now reads "exact by construction from eq. 3; only eq. 4 is first order": fixed. - Line on the inventor's related work (1972, 1976, 1989, 1992) outside the folder: fixed.

arbiter.md (QE revision, items 1-6) - Item 1, Claim 3 geometry caveat narrowed, verdict and revert condition kept: fixed. - Item 2, 1976 vs 1986 reversal recorded, no verdict change: fixed. - Item 3, Claim 4 middle part labelled as an observation (cause: Never properly tested); Marett single-laser null, 1972 JOSA null and 1992 Whitney-Silvertooth claim (with the exact-lambda/2 note) added: fixed. - Item 4, Turner-Hill/Cole/Tyapkin sentence and Gagnon 1988 ([web] via Contrarian, result not read): fixed. - Item 5, source label corrected to Natural Philosophy Wiki, "fringe venue" removed, Contrarian Doppler-factor statement marked superseded, and the 3.5 vs 5.3 orbit figures reconciled with their parameters: fixed. - Item 6, TETET-99 and the 1989 list of countries added as open leads: fixed.

primer.md - "Exactly such a null" replaced by "related null" with the setup difference: fixed (no occurrence of the old wording remains). - 1989 item retagged to [web: spirit-science.fr URL]; the "[read via Investigator]" tag is gone: fixed. - Evidence table additions (Marett single-laser null, 1972 JOSA null as recorded by Roberts & Schleif): fixed. - 1992 Whitney-Silvertooth claim linked to the exact-lambda/2 point: fixed. - 1976-to-1986 reversal bullet: fixed. - Turner-Hill/Cole/Tyapkin sentence (section 4): fixed. - Gagnon 1988 in further reading with "result not read": fixed. - TETET-99 and the 1989 country list kept as open leads: fixed. - Optional orbit-count parameters in the Claim 5 row: present (5.3).

Not remediation items (unchanged by design): the 1989 EWW PDF is still unsaved and Marett's page is still read only through the Investigator. Both are listed as next-Investigator-run tasks in the arbiter. No deep reads were required.

Remaining defects: none.

Final: PASS