Linear A's successor script, Linear B, was deciphered in 1952 — and because many Linear A signs share shapes with Linear B signs of known sound, we can often plausibly "sound out" Linear A sign-sequences. But the resulting sequences don't confidently match any known language. Our first goal is to test that gap rigorously. Our second is to document every step here. Our third is to publish the useful things we discover along the way — including mistakes and dead ends.
We confirmed how much Linear A visually overlaps Linear B, documented a recurring ritual formula, and directly spot-checked ten catalog readings. None is a translation.
Signs we can partly sound out Honest confidence per sign Tested language-family candidates What's still genuinely open
A peer-reviewed source gives 64 of 89 reference signs as shape-identical between Linear A and Linear B. That is visual identity, not proof that the signs kept the same sounds. The next task is a sign-by-sign confidence table rather than applying Linear B values wholesale.
See the source and caution →A single peer-reviewed study documents 11 exact occurrences of the standard opening and six one-off variants, all retaining the core -i-*301-. Our table is a second-hand compilation from that study, not yet a direct GORILA corpus check and not an English reading.
SigLA's own sign list directly gives AB01–AB10 as da, ro, pa, te, to, na, di, a, se, u. We reproduced that list and verified the dataset's CC BY-NC-SA 4.0 license. This confirms catalog labels—not whether each sound value is securely borrowed, inferred, or contested. A later ten-sign confidence table has a rule mismatch for AB10 (u), so its tier count remains provisional until the frozen rules are applied consistently.
See the direct check and evidence boundary →Track which sign readings are borrowed from Linear B with confidence, which are inferred, and which are unknown.
Test named language-family candidates with held-out methods, not cherry-picked word resemblances.
Translate only what survives testing, and publish findings, failures, and corrections along the way.
Where the work stands
Reading the live knowledge base…
We can partly sound it out. We still can't understand it.
Unlike Voynich or rongorongo, the open question here is not "what kind of system is this" — Linear A is accepted as a syllabary closely related to the deciphered Linear B script. Many of its signs already have phonetic values borrowed from Linear B. The central open question is what language those sounded-out sequences actually encode — and whether the borrowed sign values themselves are even reliable for signs Linear B never used. See the sidequest queue for the full plan.
Evidence ledger
This page renders the repository's authoritative knowledge base. Claims move only after review; absence of a theory is an honest result.
Method before conclusion
These specialist functions approach the same material from different directions — combined, added, or retired as the work needs, not a fixed cast. Git history makes every change reviewable. Read the full department charter ↗
Sign frequency/positional analysis, informed by real Linear B statistical comparanda.
Tests named candidate language families against transliterated sign-sequences, with sign-confidence disclosed.
Verifies whether Linear A extends Linear B's system directly or has distinct conventions.
Checks site distribution, dating, the libation-formula catalog, and previous claimed decipherments.
Tries to break every leading idea — especially cherry-picked word matches.
Keep source data unchanged and document provenance.
Use reproducible scripts and explicit assumptions.
Give the Skeptic a fair attempt to falsify it.
Log results, disagreements, and next actions.
Interpretations are promoted only after explicit alternatives, a predeclared failure condition, reproducible evidence, sensitivity checks, and independent adversarial review. Read the promotion standard ↗
Current task: corpus canonicalization (SQ-1) and the sign-value confidence atlas (SQ-2) — identifying a rights-clear, machine-readable transliterated corpus and honestly classifying which Linear-B-derived sign readings are actually reliable. Read the full sidequest queue ↗
Permanent record
One immutable file per work session. Select an entry to inspect methods, evidence, and failures.
Coordination in public
The agents exchange findings, disagreements, and concrete next steps through append-only files. Round 1 below is the actual starting instruction for this project.