Testing readings
Recorded results, laid beside their controls.
Seven batteries were run and their outputs committed. This page replays those stored values: you choose the battery, the measure and the pair to compare, and every number opens the file, pointer and hash it was read from. The measurements come from the recorded runs.
Chapter 10 / Where the Keys Stop
The search described in the book.
A restricted distinct-letter mapping was checked on constructed cases with known answers, then applied to selected Voynich text under three language objectives.
First: check the search
45 matched checks.
The search recovered every observed key and every character of the withheld plaintext in these constructed positive cases.
Chapter 10 gives the mapping restrictions, corpus selection, two shuffle controls and stability requirements. The older recorded comparison tools follow below.
Return to the chapter exploration →01 / The bench
One measure at a time
| witness | class | value | relative | compare |
|---|
Recorded comparisons
These buttons set the bench to a pair that was measured and stored. They replay recorded values; they do not run a test.
02 / Reading attempts
What the key searches produced
Each run built a key from a language's lexicon, then read held-out folios with it and measured the same reading against shuffled-glyph controls given the same freedom.
03 / What follows from this
The scope of these results
Outcome of the restricted search
The restricted inverse-key search produced no candidate meeting its requirements, and the separate dependence detector stopped at its own controls before any manuscript test was licensed. Neither outcome excludes encipherment in general, and no residual probability is offered here for systems that were not tested.