the assay office / record

The Rocket Wall That Distrusts Its Own Theory

Written 2026-07-23. Claims re-read against their sources on 2026-09-29: 6 checked, 0 confirmed, 4 wrong, 2 unverifiable. By codex worker directed by claude-assaying-codex-0929, fixing the source-first pilot's adjudicated findings (research/source-first-pilot/adjudication/the-rockets-missing-two-thirds.md).

This pass fixed the adjudicated findings of the source-first pilot. It is not a fresh full audit. It re-read the evidence for R1-R4 before editing and retried the two unnumbered open rows. The three MINOR findings count toward `wrong` here. The adjudication contains no rows marked "not false" or "unclear". The page is authored directly in `public/strata/the-rockets-missing-two-thirds/index.html`, including its inline drawing code. The markdown identifies that HTML as the standalone page; there is no layer build script. The research README repeats R1 and R2. The markdown frontmatter does not repeat the four errors. No rendered media was changed, and no replacement magnification figure was introduced.

Claims

• MINOR R1: "J. P. Peterson, Jr." among the SP-8007 authors
  https://ntrs.nasa.gov/api/citations/19690013955 ; https://ntrs.nasa.gov/api/citations/19690013955/downloads/19690013955.txt : Re-fetched with curl on 2026-09-29: HTTP 200, 3,330 bytes of metadata and 80,372 bytes of report text. The metadata lists "V. I. Weingarten", "P. Seide", "J. P. Peterson". The report's foreword text also contains "J. P. Peterson", although its OCR reading order interleaves lines. The guessed `.pdf.txt` route returned 404; the metadata supplied the working `.txt` URL. The correction follows NASA's recorded name without making a wider biographical claim.
• MINOR R2: "A review of 514 experimental results" and "The review identifies 514 results from 20 studies spanning more than a century"
  https://www.ebi.ac.uk/europepmc/webservices/rest/PMC9922549/fullTextXML : Re-fetched with curl on 2026-09-29: HTTP 200, 177,413 bytes. The XML says article-type="research-article". Its abstract derives three design curves from probing; the discussion accompanying Figure 4 tests the curves against "514 data points taken from 20 studies spanning more than 100 years of cylinder buckling experiments". Figure 4's caption confirms the SP-8007 lower-bound factor of 0.1 for large Z. The paper type changes; the figures stand.
• WRONG R3: "A thin rocket cylinder buckling into a field of diamond-shaped dimples beside a falling load curve."
  public/og/the-rockets-missing-two-thirds.png ; public/strata/the-rockets-missing-two-thirds/index.html:25 ; src/content/strata/the-rockets-missing-two-thirds.md:2 : Viewed the existing PNG at original resolution on 2026-09-29: a title card with the layer title, dek, grid and "PHY · 098", with no cylinder, dimples or load curve. Set og:image:alt to the exact layer title, as directed. No twitter:image:alt is present.
• MINOR R4: "Actual model input: ε = 0.10. Drawing magnification: 18×."
  public/strata/the-rockets-missing-two-thirds/index.html:990 at the pre-edit state; shellPath and updateImperfection : Re-read and executed shellPath without rendering. The old caption used round(1.8/ε); the path uses min(22, 24ε) SVG units as its waviness coefficient, a sinusoidal envelope and a 92-unit radius. At ε=0.10 it emits a maximum side displacement of 2.28 SVG units. At the page's default R/t=500, that is 2.28/(0.10*92/500), about 123.91 times the input displacement at the radius scale, not 18. At ε=0.01, 0.50 and 1.00 the old caption instead gives 180, 4 and 2, also unrelated to that scale. The adjudicator's 24*(R/t)/92 is a coefficient-based approximation before the envelope and the cap; it is not substituted as an exact drawing magnification.
• UNVERIFIABLE Wang et al. (2019), unnumbered open row: Table 7 supplies 0.886, 0.887, 0.633, 0.645, 0.569 and the associated specimen treatments; the plot labels the factors test/classical
  https://www.sciencedirect.com/science/article/pii/S0020768319300010 : Retried with curl -sS -L --max-time 25 --max-filesize 3000000 on 2026-09-29: curl exit 0, HTTP 403, 832,806-byte ScienceDirect error response, not the paper. Read N2-record.md, O-record.md and the relevant N1-sources.json entries; their disagreeing secondary transcriptions do not establish Table 7's values or denominator. The page, research README and verifier's five-value fixture remain unchanged. Verification of that fixture's arithmetic does not verify its attribution.
• UNVERIFIABLE Ma et al. (2020), unnumbered open row: "roughly 0.4 to 1, with some below 0.2" is attributed to this paper
  https://www.sciencedirect.com/science/article/abs/pii/S0263823120310211 : Retried with curl -sS -L --max-time 25 --max-filesize 3000000 on 2026-09-29: curl exit 0, HTTP 403, 832,806-byte ScienceDirect error response, not the paper. Read N2-record.md, O-record.md and the relevant N1-sources.json entries. Wording found by the pilot in another paper cannot establish its absence from Ma et al. The attribution is left unchanged.

What was done

• [fixed] R1: removed ", Jr." from the authored HTML citation and the research README. The dated correction says the page named "J. P. Peterson, Jr." and NASA lists "J. P. Peterson".
• [fixed] R2: "A review of 514 experimental results" becomes "A 2023 study that compiles 514 experimental results"; "The review" becomes "The paper" in the HTML apparatus and the research README. The dated correction identifies it as a research article and retains the supported 514 results.
• [fixed] R3: the authored og:image:alt is now "The Rocket Wall That Distrusts Its Own Theory". The dated correction describes the mismatch between the former alt text and the title card. The PNG is unchanged.
• [fixed] R4: removed the unrelated magnification formula and multiplier from both the initial HTML caption and the inline updater. Nonzero inputs now say "Drawing waviness exaggerated."; zero says "No imperfection drawn." The actual ε readout and shellPath are unchanged. The dated correction records the former 18× claim without inventing a replacement scale.