Observation F-20

Tested plant-label links remain unsupported, with a corrected evidence base

D-086 records a research decision. The code s53 identifies the earlier tests discussed below.

The retained plant-label and repeated-label tests do not support the particular text links they tested. D-086 removes the erroneous s53 text-partner tests; the separate s53 image-matching run is unaffected.

I compiled this story with the team (the AI agents I direct) from the versioned export below.

Research status and checks

Current status CORROBORATED; standing What this means

What was tested Specific partner-page and label tests; not a test of all possible text–picture relationships.

What to look for, and why it matters

Notice the short inscriptions beside plant parts and containers. If these were repeated plant names, they might appear again beside related drawings. The tested links have not established that connection.

Folio 99r: painted jars beside rows of small plant parts, each with a short label
Folio 99r. Containers stand beside rows of plant parts and short inscriptions. These are the kinds of picture-and-label arrangements that motivate the search for a written name. An unannotated context scan; the measurements are reported in the study below. Yale University, Beinecke Rare Book and Manuscript Library, MS 408. Public domain. Larger scan of folio 99r at Yale

Other explanations

  • Labels may index something other than plant names, use a transformation not tested, or refer to a different textual level.

Limits

  • The s128 power estimate applies to its planted naming-link model. A null result does not show that labels or the manuscript lack meaning. Visual partners share manuscript and folio dependencies.

How independent the evidence is

Transcriptions, catalogue entries and copied summaries are not independent confirmations. Repeated use of the manuscript and pipeline is shared evidence.

Corrections

  1. Replaces c-377.text-side-s53

    Eight of twelve partner mappings were mis-specified; these text-side lines and duplicate-herbal test stop counting. F-20 is CORROBORATED, not TRIANGULATED. Fully qualified artifact paths and hashes distinguish the unaffected image program.

    Recorded in Main research record: .inventory/registers/DECISIONS.md, line 92 (| D-086 |), sha256 9b0f6a628e53. Read the cited line

The earlier claim stays in the research record unchanged; this card shows what replaced it and why.

Sources and evidence

Credit Simone Systems Research: Jon directing AI-assisted research; retain original authors and run owners in the cited artifacts. Published plant-resemblance judgments are credited to René Zandbergen and Jorge Stolfi; the Atlas claims no plant identification. Petersen supplied the historical marginal-note comparison included in the source catalogue; it is not an independently confirmed identity.

Research records behind this card (10, 10 you can open)

Canonical finding or disposition for F-20; use the recorded locator and version.

Governing correction or disposition: | D-086 |. This governs interpretation of earlier results.

Preserved supporting artifact upstream:results_s54.txt; exact version is identified by its digest.

Preserved supporting artifact upstream:results_s128_20261019T090659.txt; exact version is identified by its digest.

Exact implementation version (upstream:s53_p4a1_p4e1_anchors_and_errors.py); consult its associated protocol and frozen inputs before reproduction.

Preserved supporting artifact upstream:results_s53_p4a1_p4e1_anchors_and_errors.txt; exact version is identified by its digest.

Frozen protocol or digest record for the cited run; retained to establish the planned scope.

Exact implementation version (upstream:s53_same_plant_parts.py); consult its associated protocol and frozen inputs before reproduction.

Preserved supporting artifact upstream:results_s53.txt; exact version is identified by its digest.

How to check it

Mode read-only source inspection; no experiment run by this package

Open the linked evidence copies or excerpts to check the cited wording and measurements. File details distinguish exact copies, path-normalized copies and excerpts, and retain source and copy hashes. Some cited files remain private and are marked as such. These selected evidence files are not a complete experiment-reproduction package: inputs, code and environment are not uniformly supplied. No experiment was rerun to prepare this export.

The files to read are the 10 research records listed under Sources and evidence; 10 of them can be opened there.

Cite and reuse this record

Cite this story

Simone Systems Research. “Tested plant-label links remain unsupported, with a corrected evidence base.” Voynich Evidence Atlas, record F-20, export atlas-public-v1.0, 2026-10-19. https://simoneresearch.com/research/voynich/atlas/f-20/

Report a correction to this story. Your email draft includes its public URL and record ID. Add the change you suggest and a supporting source.

This card is built from record F-20 in the export named below. The downloads contain the same record.

No redistribution license asserted for laboratory artifacts. Third-party content is linked, not bundled.

File details on this page

Format, size and a short SHA-256 identifier, measured when this page was built.