QE (Step 4b): Very-Frank (Frank W. Very, The Luminiferous Ether, 1919)

QE agent, 2026-10-04. Inputs checked: _text/manifest.json, both digests, deep/The-Luminiferous-Ether-eb1408.md, facts.json, claims.md, sites.md, research/ (5 notes; research/files/ is empty), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. I spot-checked the extracted text (_text/The-Luminiferous-Ether-eb1408.txt) and rendered PDF p.38 myself (/tmp/qe-veryfrank/p-38.png). Page numbers are PDF pages (printed page + 4).

Verdict: MINOR

The central verdicts are sound and well argued, and the numerical core (claim 3) has been checked by four agents and agrees with the page image. The defects are: one misread number in the galaxy-collision argument, which three files repeat and which runs against Very; several citation and page errors; facts.json and claims.md not brought into line with the reviewers' corrections; and a set of experiments and checkable numbers in the source (pp.11-35 and 42-59) that no file mentions. None of these is likely to move a headline verdict. They can be fixed by updating facts.json, claims.md, arbiter.md (adding a QE revision section) and primer.md, after two deep reads.

---

1. Coverage

There is only one document, and both of its chunks were digested. Pages 1-5 (the Google notice), 10, 60 and 61-64 (the library stamp and blank pages) are junk and were acceptably skipped. There are no unread files and no coverage defect.

2. Shortcuts

• Deep-read scope is too narrow for claims 1, 4, 5, 6 and 7 (defect, moderate). facts.json says it plainly: "PDF pp.1-35 and 43-60 rest on digests only". Only pp.36-42 were deep-read. Claims 1 (ether-particle formation, pp.13-20), 5 (atom structure, pp.21-33), 4 (refraction and bound ether, pp.33-35 and 51-53), 6 (gravity and galaxies, pp.44-51) and 7 (entropy, pp.42-44) lie mostly outside that range. The reviewers made up for part of this. The mathematician read the OCR of pp.19-20, 24, 28, 33-34 and 39-54, the physicist pp.28, 43-44 and 50-51, the arbiter about ten pages, and the contrarian pp.42-52. But nobody read pp.11-18 or 21-32 closely, and the material missed in section 3 comes from those gaps. The galaxy-scale misreading (section 4, check 5) shows that the partial reads of pp.45-47 were not careful enough either.
• Investigator stopped early. sites.md records 260 s used out of the 600 s allowed. Searching stopped "after the planned site set", and HathiTrust, Google Books, the Pitt and Lowell archives and the newspapers were left "not reached (time limit)". That is not accurate, because time remained. The note says nothing was downloaded because "downloading needs the owner's go-ahead". prompt.txt has no such rule ("research/files/ - copies of useful documents found ... when downloadable"), so research/files/ is empty, although the Fessenden 1900/1914 PDFs, Very 1909, the second scan of the book and the NCR Oct 1918 issue were all freely downloadable. Later agents had to re-fetch these to scratch.
• The second scan (archive.org luminiferousethe00veryrich) and the NCR Oct 1918 printing went unread. The arbiter names them as the item that would settle the 1.660e22 question. It is a cheap check, and it was left open.

3. Missed material (from the digests and text; not in facts.json, claims.md, any review, the arbiter or the primer unless noted)

1. Very's own experimental datum: firefly spectro-bolometry (PDF p.21 [read]). Very cites "direct spectro-bolometric observations of the entire energy-spectrum" (Langley and Very, Am. J. Sci. XL, 1890, per digest). They found short-wavelength emission "which if it were thermally produced would correspond to a temperature of several thousand degrees ... given off at the temperature of the living body without the corresponding infra-red". He uses it as evidence that luminescence comes from satellite electrons and not from thermal nuclear vibration. This is the only measurement in the book made by Very himself, and nobody mentions it. (Cold bioluminescence is mainstream [memory, high]; the point is that it is his evidence, and it is real.)
2. Eichenwald 1903 (PDF p.23 [read]). A rotating electrified, metal-coated glass disc gave a magnetic effect, "which proves that the aura is immobile in respect to the motion of matter". This is an experiment Very relies on for the aura not being dragged, and it bears directly on claim 4: an undragged aura plus a dragged bound-ether shell. It is relevant to the arbiter's "frame in which light moves at c" question. Not mentioned anywhere.
3. Thornton's cube rule (PDF p.32 [read]). Very says H, B, Al, Cu, Ru and Os, "the densest atoms of their respective thermal periods, have their atomic weights proportional to the cubes of the first six natural numbers". This is a checkable claim of the kind the reader profile asks for. My check, using modern atomic weights [memory, high], gives ratios of weight/(1.008 n³) of 1.00, 1.34, 0.99, 0.99, 0.80 and 0.87: it holds for Al and Cu, fails for B by 34% and for Ru and Os by 13-20%. Very uses it to support his twelve-storied uranium.
4. Hirn and d'Auria on the ether's mass (PDF pp.34-35 [read]). facts.json key claim 13 says Hirn's resisting-medium computations rule out a massive space-filling ether, so ethereal atmospheres are confined near massive bodies. Very also corrects d'Auria (1897) for confusing Maxwell's ether density reference (water, not air) and notes a numerical error in Maxwell's own value. This claim was dropped from claims.md and is in no review. It is a positive argument from planetary dynamics, and it carries part of claim 4's structure.
5. Very's Nova Persei mass and his galaxy scale (PDF p.45 [read]). He deduced a Nova Persei mass "1149 times that of the sun" (Astron. Nachr. 3771). He argues that Curtis's small white nebulae "should be more nearly 80,000 light-years distant", and that faint stars at 50,000 ly are "improbably" connected with "our Galaxy whose condensations occupy a much smaller space" (p.48). These numbers fix his own scale for the collision argument (section 4, check 5).
6. Millikan's Kepler-law derivation of the Rydberg constant "to one part in one thousand", and Marion Erwin's 1,732-electron (12³ + 4) hydrogen atom, which Very says Millikan's electron mass "definitely condemns" (PDF p.28 fn [read]). This shows Very rejecting an atom-count model on measured data, which is relevant to how claim 5 was built.
7. Lenard 1902 (PDF p.19-20 [read]): photoelectron energy independent of intensity, used by Very as the reason ether-particles need a "structural energy". Bragg's corpuscular X-rays (pp.13-19), which Very reinterprets. The H₃ molecule and the V Puppis "two-electron hydrogen" spectrum (p.29). Swan's 1897 resin stress-figures (p.27), Very's evidence for the polar inflow/outflow structure of electrons. Lodge's steel-rod comparison (p.44). Zöllner's "Transcendental Physics" as experimental "semblance" for hyperspace (p.49). Swedenborg's flying-machine design and R. W. Brown's explanation of the vortex error (pp.54, 57-58). Newton's "aetherial spirits" passage (p.56-57). None of these is in any file. The Lenard and Swan items bear on claims 1 and 5; the rest are context.
8. Telepathy (PDF p.55 [read]): Very suggests a resting free-ether atmosphere may be the "physical basis ... of that telepathic thought which is received by harmoniously attuned instruments". It is in facts.json and contrarian.md, but not in arbiter.md or primer.md (see Retention).
9. "Is Entropy always One-Sided?" and Shapley (p.43): Very disputes Shapley's M13 conclusion and cites his own study of white nebulae as showing non-selective absorption by the aura. The physicist mentions it; the primer does not.

4. Accuracy spot checks

5. Retention

• Collision argument (check 5). The mathematician's misreading turns a point that is internally coherent for Very into "fails on his own numbers", and the arbiter's table carries the label "his scale". The arbiter's verdict text partly rescues it (infinite-age caveat), and the primer uses only the modern-input figure. Still, the record should say plainly that on Very's own distances the argument works, and that the failure lies in the distance scale. This is a retention and fairness defect (minor).
• Telepathy speculation (p.55) is in facts.json and contrarian.md but was dropped from arbiter.md's "Evidence kept in the record" and from primer.md's "religious frame" bullet, which lists only "spiritual" and "finger of God". It should be restored beside them.
• Mathematician's steelman (Fessenden's 5e36 cm/s with a galactic quarter-wave gives about 3e13 Hz, matching Very's "periodicity comparable to that of light"). The arbiter kept it; the primer did not. It is one of the few numerical points in Very's favour, so it belongs in the primer section 7 "Solid" or in section 6's kept evidence.
• Shapley/M13 non-selective absorption (Very's own white-nebula study): the physicist mentions it, the arbiter and primer do not. It is relevant to claim 7 and to the "dust is a real counterpart" point.
• Everything else of weight was retained: Dayton Miller, Fessenden's rejection account, the Langley credit omission, the Norwood fraud exposure, Very 1909, Fessenden's glass-prism and gases-weigh tests, and the absence-of-evidence analysis (arbiter table, primer summary). Good.

6. Process

• Source tags. Present and mostly honest throughout. Exceptions: the primer's "He was a Swedenborgian" and "Swedenborg's 1734 Principia" are untagged (they should be [read: p.9, p.11] and [memory, high]). The primer does not introduce R. A. Fessenden (the radio pioneer) to the reader [memory, high]; one line is needed.
• Web checks. The physicist made 3, the contrarian 3 and the arbiter 4 plus a page image, within budget. Open items:
• Nipher "not replicated" appears as [memory, low] in the physicist, contrarian and arbiter files. prompt.txt says "Never state that replications or tests happened (or failed) without a source". The arbiter's wording "no replication found [memory, low; not checked]" should become "unknown; not checked". Nipher's 1916-1920 experiment is also an anomalous experimental claim that Very wrote about in 1920, and it should be an Investigator target.
• Taylor 1909 [memory, medium] is used in the primer's evidence table under "Tested?" for claim 1. It is not load-bearing, because many later single-photon experiments exist, but it should be web-checked or reworded to "later single-photon interference experiments".
• de Sitter 1913 was [memory, medium] in the contrarian review and was checked by the arbiter (good).
• The mathematician's statement that every [memory] fact was high confidence includes "Leahy (1889)" (a date not in the text) and the Rutherford gold-nucleus size [memory, medium]. Neither is load-bearing.
• Verdict scale. Applied consistently. Split verdicts are explained. The arbiter correctly declined "Refuted by calculation" for claim 2 (the 69 MeV figure is the reviewers' construction) and for the collision argument.
• facts.json and claims.md are stale and internally inconsistent.
• (a) The summary says aberration is explained "with no appeal to Rayleigh/Brace", while key claim 7 says Rayleigh/Brace show the bound ether. The physicist flagged this, and it was not fixed.
• (b) "No number links the Planck quantum to the rotation rate" is fine, but key claim 3, claims.md claim 2 ("no link to the frequency") and the deep read §4 miss p.19.
• (c) mainstream_conflict says bound-ether shells conflict with aberration, but Very makes aberration independent of the bound ether (p.52; contrarian note 6).
• (d) The Hirn claim is missing from claims.md.
• (e) The [memory, medium] Langley item is now [web] (Wikipedia, via the Investigator).
• Primer. It follows STEP 4 order 1-9, has answers at the end, defines CGS units, bulk modulus, the hairy ball theorem and the Laplacian, and the worked example is real. Small issues:
• Check-yourself Q4 and its answer frame Fizeau against "every atom carries its own ether completely along". Very's own p.52-53 statement implies the opposite: bound ether retards light "independently of its state of relative motion", which means zero drag. The arbiter is right that Fizeau's partial drag contradicts that zero drag. The question should test Very's actual position.
• "finger of God (p.55)" should be p.49.
• example.py runs cleanly with the venv (exit 0). Its outputs match the primer: 2.512e22, 2.875e-13, 2.01e32/3.22e33/6.96e30, unit ratio sqrt(g)/(cm^(5/2) s), 254/152, 69/104 MeV.

---

7. Remediation actions

Deep reads (manifest slug: The-Luminiferous-Ether-eb1408)

1. Opus, PDF pp.42-51. Question: Write out Very's galaxy and gravity argument with every number he gives: the Andromeda nova distances, the Curtis nebula distances, the size implied for our Galaxy (p.48, "much smaller space" than 50,000 ly), the millimetre-dot spacing and velocities (p.46), and the Nova Persei mass. Recompute the collision time on his own scale. State exactly what Bjerknes, Hicks, Burton and Leahy are quoted as showing, how Very turns Leahy's quarter-wave objection into a prediction (pp.49-51), and what the entropy section and the Shapley/M13 footnote (pp.42-44) actually claim. Quote, tagged [read: p.N].
2. Sonnet, PDF pp.11-35. Question: List every experiment, measurement and numerical claim Very uses in Part I and the start of Part II. Include the formation and properties of ether-particles (pp.13-20), Lenard and Bragg, the firefly bolometry (p.21), Eichenwald 1903 (p.23), the atom and shell structure (pp.24-32), Swan's stress figures, Hicks and Thornton (the cube rule), the bound-ether refraction argument (pp.33-35) and Hirn, d'Auria and Maxwell on ether density (pp.34-35). Say for each how Very uses it and which central claim it supports.
3. Sonnet, PDF pp.51-59. Question: Read the bound ether vs free ether-particle passage (pp.51-53), especially the frame in which light moves and the "independently of its state of relative motion" statement. Also read Swedenborg's atmospheres and vortex theory (pp.53-57), the telepathy passage (p.55), Newton's "aetherial spirits" (p.56-57) and Very's closing summary (p.59). Does the text anywhere state what light's speed c is measured relative to?

Investigator (next run, not a deep read; prompt.txt Step I)

• Download into research/files/: Fessenden 1900 (jstor-1628439) and 1914 (jstor-1640236), Very 1909 (jstor-1635576), the second scan of the book (luminiferousethe00veryrich) and NCR Oct 1918 (new-church-review-1894_1918-10_25_4). Compare PDF p.38's "1.660 × 10^22" and the "10^33" in the second scan and the NCR printing.
• Use the remaining time budget on the sites marked not reached: HathiTrust, Google Books, the AAS 1918 abstract in Popular Astronomy, the Ogden 1928 memoriam, the Pitt and Lowell archives and the 1927 obituaries.
• Find Very, "On a Possible Limit to Gravitation" (Publ. AAS III p.335), Very 1920 on Nipher, and Nipher's own papers, including any attempted replications.

File changes

• facts.json:
• Fix the Rayleigh/Brace contradiction in the summary.
• Correct "no link to frequency" (p.19).
• Correct the aberration point in mainstream_conflict.
• Retag the Langley fact [web].
• Add the firefly bolometry, Eichenwald, Thornton cube rule, Lenard, Nova Persei mass and Very's galaxy distance scale to the summary or key_claims.
• Update how_read after the deep reads.
• claims.md:
• Claim 2: add the p.19 frequency-proportional "energy of formation" sentence.
• Claim 4: add Eichenwald (aura undragged) and the Hirn/no-massive-ether argument, or make the latter a separate claim.
• Remove "digest lines" caveats that the deep reads supersede.
• arbiter.md (add a "QE revision" section):
• Withdraw "d = 80,000 ly (his scale)" and state that the collision argument holds on Very's own distances (collision time about 1e9 yr) and fails through his distance scale.
• Correct "finger of God" to p.49.
• Change Nipher's "no replication found" to "unknown; not checked".
• Add telepathy (p.55) to the evidence kept.
• Weigh Eichenwald 1903 under claim 4 (it fixes the aura as undragged, which favours the mathematician's Reading A and the 0.37-fringe problem).
• Note firefly bolometry and the Thornton cube rule (with the check above) as kept evidence.
• No headline verdict is expected to change. Claim 6's sub-verdict on the collision argument should read "valid on his 1918 distance scale; fails with modern distances".
• primer.md:
• Fix p.55 to p.49.
• Add telepathy to the religious frame bullet.
• Introduce Fessenden in one line.
• Tag the Swedenborgian and Principia statements.
• Rewrite check-yourself Q4 and its answer around Very's zero-drag statement (p.52-53) vs Fizeau's (1 − 1/n²).
• Replace or web-check Taylor 1909.
• Add the mathematician's 3e13 Hz steelman and the Shapley/M13 point to section 6 or 7.
• Add the Thornton cube rule as an optional "check yourself" exercise; it is a good units-free numeric test for the reader.
• Add one sentence to section 6 (claim 6 row) saying the collision argument works on Very's own distances and fails on modern ones.
• mathematician.md: No rewrite is needed, but the QE revision in arbiter.md should cite these errors: the p.20/p.44 page slips and the 80,000 ly misreading.

---

Recheck

QE recheck, 2026-10-04. I re-read arbiter.md (QE revision section), claims.md, facts.json and primer.md by grep and spot reads, and confirmed the state of research/files/.

No headline verdict changed. The deliverables now reflect the corrections.

Final: PASS