Every few years someone announces that they have deciphered Linear A. The news spreads, and then, usually quietly, specialists explain why they are not convinced. After Why Linear A Can't Be Read Yet you can see why such claims are hard to make good. This lesson gives you the tools to judge one yourself, using the one decipherment that did succeed, Linear B (The Tripod Tablet Confirms It), as the model.
Key idea
A good decipherment makes testable predictions, reads the whole corpus the same way, fits place names and new finds, and convinces specialists who check it. Claims to read Linear A appear every few years; none has met these tests or been accepted.
Five tests
| Test | What it asks | How Linear B passed |
|---|---|---|
| Fixed values | Is each sign given one value, used everywhere? | Ventris's grid gave each sign one value, used in every word |
| The whole corpus | Do the values read most texts, not just a chosen few? | They turned thousands of words into Greek |
| Predictions | Does the reading predict what a new text will say? | Ta 641's words matched its pictures |
| Outside checks | Do place names, logograms and numbers agree? | ko-no-so, pu-ro and others; ti-ri-po beside the tripod |
| Acceptance | Do independent specialists, checking the evidence, agree? | Most were convinced within a few years |
Warning signs
- Different values for the same sign in different words, chosen to make each word fit.
- A few "matches" picked from the corpus, with the rest left unread.
- Meanings taken from many different languages, whichever fits best for each word.
- Readings that cannot be checked by anything outside the claim itself.
- Announcement to the press before specialists have examined the evidence.
Why it works
Why are a few matching words not enough? Because chance produces matches. Linear A has hundreds of short words, and the world has thousands of languages with millions of words. Allow some freedom in sounds and meanings, and a handful of "hits" will turn up for almost any language. A real decipherment is shown by what it does with all the rest, and above all by texts it had not seen.
Common mistake
Being persuaded by a few words that seem to match. Claims to read Linear A appear every few years; none has met these tests or been accepted by specialists. Check: ask whether the claim uses fixed values, reads the whole corpus, makes a prediction that came true, and has convinced independent experts.
This course does not discuss particular claims or the people who made them. The tests above apply to any claim, and you can use them on the next one you read about.
Worked example: Spot the warning sign
An imaginary claim reads one Linear A word as Greek, the next as Egyptian and a third as Sanskrit, whichever language gives a meaning. Which test does this fail?
- A real decipherment reads the corpus in one language, consistently.
- Choosing a language word by word lets almost anything "match".
Answer: It fails consistency: one script, one language, read the same way throughout.
Worked example: What made Linear B convincing?
Name a test the Linear B decipherment passed that a Linear A claim would also need to pass.
- Ta 641 was read with values fixed before it was seen.
- Its words matched its pictures.
Answer: A correct prediction on a new text, like Ta 641.