Linear B

Linear B is the collection's clearest demonstration that a script and a language are two different things, and that the join between them can be visibly bad.

The tablets were found at Knossos from 1900 and later on the mainland, baked hard by the fires that destroyed the palaces holding them. For half a century the assumption was that they recorded the language of Minoan Crete — something pre-Greek, unknown, probably unrelated to anything surviving. In 1952 Michael Ventris, an architect working outside the academic establishment, showed that the language was Greek: an archaic Greek, five hundred years older than any other written Greek, but Greek.

The script does not fit the language it writes. That misfit is the evidence that the Mycenaeans borrowed it.

What makes Linear B worth a plaque is not that it was deciphered but what the decipherment exposed. The syllabary is a poor instrument for Greek, and systematically so — which is precisely what you would expect from a writing system designed for another language and taken over wholesale.

Linear B signs are mostly open syllables — a consonant followed by a vowel. Greek is full of things that shape will not hold, and the script copes by simply not writing them.

Final consonants are dropped, and Greek ends words in -s and -n constantly. Consonant clusters must be padded with vowels that are not spoken, or left out. The script does not distinguish l from r — one series covers both, which is unremarkable for some languages and destructive for Greek. It does not mark vowel length, which is grammatically significant. It does not reliably separate the voiced, voiceless and aspirated stops that Greek contrasts: one series serves where Greek needs three.

The result is an orthography where a written word can correspond to several spoken ones and the reader recovers the difference from knowing the subject matter. That is workable for a palace clerk counting sheep and chariot wheels against a known inventory. It would be close to useless for poetry, and no poetry was written in it.

When Greek is written again, five centuries later, it is in a wholly different system — the alphabet borrowed from Phoenician, with vowels added. See Greek.

Most of the syllabary is shown here as forms only, for the same reason as elsewhere in this building: the values are established but the transliterations are conventional, and a table of eighty-seven signs invites a false sense of precision about a script recovered by inference. The five pure-vowel signs are the exception — they are secure, and they are the corner of the grid from which Ventris worked outwards.

Roughly six thousand tablets survive, and they are almost without exception administrative: inventories of grain, oil, livestock, textiles, bronze, chariot wheels, personnel. There is no literature, no correspondence, no narrative, no law. The tablets were not meant to last a year — they are unfired clay, notes for the palace's annual accounting — and they survive only because the buildings burned.

So the corpus preserves the Mycenaean world as a warehouse ledger: names of officials and gods, place names, quantities. That the earliest Greek we have is an accounting record rather than an epic is a fact worth sitting with. Compare the parent script, which has the same shape of corpus and remains unread — the language beneath it is still unknown. See Linear A in the Undeciphered Vault.