written.institute
𐘀𐘁𐘂 𐘀𐘁𐘂𐘃𐘄𐘅𐘆𐘇

Linear A

The administrative hand of Minoan Crete — a script we can partly sound out, over a language we cannot read. A sealed case in the Undeciphered Vault.

ISO 15924Lina · 400
Typologyundetermined — apparently syllabic, with logograms (like its descendant Linear B)
Status Undeciphered
RegionMinoan Crete (and the Aegean)
Datesc. 1800–1450 BCE
Corpus~1,400+ inscriptions, mostly short administrative texts on clay (~7,400 sign tokens)
Underlying language"Minoan" — unknown, with no securely identified relatives
Bilingual keynone known
Encodingencoded in Unicode 7.0 (2014), block U+10600–U+1077F
DirectionLeft to right →
Noto fontNoto Sans Linear A

Curatorial note

Linear A is the tantalising case — the one undeciphered script we can partly read aloud yet cannot understand. It was the administrative hand of Minoan Crete, and its younger relative Linear B — which reused many of the same signs — was famously deciphered by Michael Ventris in 1952 as an early form of Greek.

Borrowing those sign-values back, scholars can give many Linear A signs a probable sound, so a tablet can be voiced. But the language underneath, "Minoan", is unknown and unrelated to any language we can use as a key, so the sounds carry no meaning.

"We can pronounce it. We simply have no idea what we are saying."

The corpus is also small and almost entirely bookkeeping — inventories and totals, not the varied text a decipherment feeds on. The full case file sits downstairs in the Undeciphered Vault.

Why it resists

A decipherment needs three things
A large, varied corpus~1,400 mostly short administrative texts — too little, too repetitive
A known underlying languageMinoan is unknown, with no identified relatives to lean on
A bilingual keyno Rosetta-style parallel text has ever been found
𐘀𐘁𐘂 Linear B (deciphered, 1952) Mycenaean Greek The cruel twist: the signs are partly known through Linear B — so Linear A is pronounceable. Only the language is missing. Ventris solved the chain from the far end, but its first link opens onto a tongue with no living or recorded descendant.

What survives

The corpus

Roughly 1,400 inscribed objects survive, carrying about 7,400 sign tokens in all — a corpus so small that, typeset at 12 point, it fills under two pages of letter paper. Most texts are extremely short: a typical entry is a sign-sequence (a personal or place name) followed by a logogram for a commodity and a numeral. Very few inscriptions run to more than a handful of words.

The finds cluster on Crete. About half come from Haghia Triada (some 1,269 objects), with further groups from Zakros, Khania and Phaistos, and only around 32 from Knossos. Almost everything is accounting: inventories, receipts and totals kept on clay tablets and roundels — administration, not literature.

In the building

In the Reading Room

WI·DEC · Decipherment Dossiers · Open file

The Minoan Wall

Ventris broke Linear B in 1952 and it spoke Greek. Its parent, Linear A, uses many of the same signs — so why, seventy years on, can we still only pronounce it?

Plaque sources: The Unicode Standard (Linear A block U+10600–U+1077F, added in Unicode 7.0, June 2014) · ISO 15924 (Lina · 400) · standard scholarship on Linear A and on Ventris's 1952 decipherment of Linear B · Noto Sans Linear A · ScriptSource (SIL). Corpus figures (~1,400 inscriptions, ~7,400 sign tokens; ~1,269 objects from Haghia Triada, ~32 from Knossos) are approximate. Compiled 2026-07.

The Register New acquisitions and new chapters — once a month, nothing else.