It's one of a group of similar scripts, generically called Aegean, found in mainland Greece, on Crete and Cyprus, and at Ugarit on the Syrian coast. One was used on Cyprus for writing Greek into classical times, and this was easily deciphered in the nineteenth century. It provided some help in the decipherment of Linear B, but neither of them has allowed us to understand the Linear A of Crete or the Cypro-Minoan.

We can tell that they are all syllabaries: too many signs for an alphabet, too few for a hieroglyphic system. (There is also a hieroglyphic script on Crete, of which Linear A and B are reduced forms.) They are deeply unsuited to writing Greek. There are only open syllables, and there is no distinction of voice, except for [d], so [ko kho go] are all written the same. However, there are unnecessary combinations like [dwo dwe nwo rja pje]. There is no distinction between [l] and [r]. No language known anywhere in the region matches this, so it is unknown who Aegean was first created for.

Cypriot and Linear B are defective in similar ways. In Linear B the main points are illustrated by the name Knôssos. The long vowel and double consonant are not distinguished; the consonant cluster kn is broken up by echoing the next vowel; and consonants at the ends of syllables are neglected: the result being KO-NO-SO. In Cypriot final consonants were marked by adding -e, so this would be written KO-NO-SO-SE. The lack of this very common character SE at the ends of words was another reason why Linear B was thought unlikely to be Greek.

It is important to bear in mind that KONOSO is just a modern conventional transliteration of the signs: it was not pronounced [konoso]. In what follows I shall represent Linear B signs in SMALL CAPITALS, the actual Mycenaean pronunciation in [phonetic symbols], and the later (alphabetic) Greek equivalent in italics.

The one sign we transliterate KO could be any one of [ko kho go ko: kon koi kol kho:n gor go:s] and so on.

The scripts are too curved to be convenient for clay tablets. They must have been written with pen or brush on some other medium, and the tablets were temporary records. They have no indication of years beyond 'this year', 'last year' and so on, but contain census information that must have been stored long-term. The clay was baked but the other medium destroyed in the fires that ended the occupation of the palaces.


Decipherment largely confirmed linguists' reconstruction of Primitive Greek, but for the first time gave a precise dating to which changes had taken place in the evolution from the Proto-Indo-European (PIE) to Proto-Greek to the divergence of the Greek dialects.

The PIE labiovelar series [kw kwh gw] still existed. In later Greek they changed to the p-series or t-series depending on dialect and position: ATOROQO [anthro:kwos] ánthropos 'person', QETORO- [kwetro-] tetra- 'four-'.

There was a palatal series [kj kjh gj], but these did not come from the PIE palatals, which had fallen together with the k-series in Proto-Greek. The palatals were developed from clusters like [kj] and [tj]. The voiced [gj] also had a third source, [j]. These sounds are conventionally transliterated Z but this is an anticipation of a later change to [ts] and [dz]. For example MEZO [megjos] meizos 'bigger', from PIE *[megj-]. (The asterisk means a reconstructed form.)

[h] was not usually written, except that there was a sign for HA. Two vowels separated by [h], from earlier *[s], kept their hiatus, whereas they later fused, so it was probably pronounced wherever it derived from [s].

The semivowels [j] and [w] still existed. They disappeared or became [h] in later Greek. So WANAKA [wanaks] anax 'king'. (The classical word for 'king', basileús, appears to have been quite a minor title in Mycenaean times, if it comes from QASIREU [gwasileus].)


The Linear B language is remarkably uniform, given that it was used in such distant locations as Mycenae, Pylos, and Knossos. The Knossos archive has several minor differences, but Linear B has been described as a 'chancellery language', a standardized and possibly slightly archaic form taught to scribes.

There would have been some dialect differences in the popular language of the Mycenaean people, though Linear B is close to the ancestor of the later Greek dialect groups. That is, most features distinguishing them developed after the Mycenaean period. But of the dialect groups, Linear B is closest to Arcado-Cypriot. Arcadia is in the central Peloponnese, very far from Cyprus and separated from it by a chain of Doric dialect lands, including Crete. The Dorian migration or invasion happened into the gap left by the fall of the Mycenaean palaces.

That it is not fully ancestral to later Greek is shown by the word for 'figs', SUZA [sukja]. This comes from an earlier *[sukia], with a natural phonetic change, but later dialects retain the [k]: if they were descendants of the Mycenaean they would have been [syssa] and [sytta].


The tale of the decipherment is told in write-ups above and under Michael Ventris. The following is a linguistic perspective. Parts of the lists could be read easily because they consisted of simple ideograms and numerals: four horses, ten pots, two women, and so on. Alice Kober had done detailed study of the positions each sign occurred in, and groups of similar signs, before her early death in 1950. This gave Ventris the key, before he suspected the language was Greek. Some lists showed girls and boys. If the language inflected in any way like Greek, these might be distinguished by different endings on the same stem; and indeed the girl and boy lists show three signs in the pattern AB and AC. Additional information in this might be that B and C began with the same consonant and had a different vowel. Building on such clues, he compiled tables of tentatively grouped same-consonant and same-vowel signs.

He already had the names of some important towns. Identifying which they were involved some guesses about how the echoing vowels and final consonants were handled, but one of them was three signs long, with perhaps the same vowel, and began with the first sign of 'girl/boy'. Knossos in Cypriot would be four signs, KONOSOSE, but if the final -s was unwritten he would have KONOSO.

As he built up possible endings, of locatives, of place-name adjectives, of lists connected by 'and', and of which of them shared vowels, he increasingly found the unknown language had grammar like that of Greek. So he looked for Greek words to fit, and deciphered KOWA, KOWO [korwai, korwoi] korai, koroi 'girls, boys'.

As he deciphered more, the endings included archaic forms used by Homer: genitive singular [-ojo] (later -ou), locative and instrumental plural [-phi]. The particle meaning 'and' looked wrong: he was expecting [-te]. When he began working with the philologist John Chadwick he learnt that the language was so archaic that the particle was actually [-kwe].

The solution was quickly accepted by most Minoan scholars because, despite the freedom to substitute so many values that the scribal rules gave, there was no doubt that it was easy to read Greek words in familiar contexts: Carl Blegen looked at his newer, unpublished Pylos tablets and found QETOROWE beside a picture of a four-eared (four-handled) pot, TIRIJOWE by a three-eared one, and ANOWE before an earless one: all easy Greek. The same tablet showed TIRIPO for one tripod and TIRIPODE for two, in the dual number.

John Chadwick, The Decipherment of Linear B, Cambridge, 1958.
L.R. Palmer, The Greek Language, Faber & Faber, 1980.