Beinecke MS 408 · computational analysis
A reproducible toolkit and shared benchmark for Voynich-Manuscript research
It doesn't claim a solution — and it doesn't wave the mystery away. It gives the research community a common, honest way to put a hypothesis about Beinecke MS 408 to the test — "is this a cipher of Latin?", "does my generator reproduce the manuscript?" — and to see, reproducibly, exactly where the evidence stands. Cold and logical in method; honest and even-handed about what it will and won't claim.
Point it at your hypothesis and it reports, against firewall-computed reference bands, which of the manuscript's statistical properties you actually reproduce — and which you only appear to. It is built to be unimpressed, including by its own authors: the public record includes the toolkit retracting its authors' own headline result after a concurrently-published cipher was run through it.
01 · Start here
Pick your path
Evaluate an idea
Test a hypothesis and read the verdict honestly — no code required.
The API & method
The evaluate() surface, the reproducibility contract, the discipline.
02 · Run it
Try it
pip install ms408
python -m ms408.acquire # pinned, checksummed reference data (consume-only)
python -m ms408 my_tokens.txt # ≥1000 word tokens; prints a per-axis verdictMatching the manuscript is necessary, not sufficient: an in-band result means your hypothesis is not excluded, never that it is the mechanism. Every axis ships with its caveat. See Limits.
03 · Read the work
The research
Preprints — the constraint-envelope program and the methods paper on adversarial self-correction. Every reported number is reproducible from the repository.