The Decipherment of Linear B
Created | Updated Jun 22, 2007
Linear B is an ancient form of writing, used on the island of Crete and in mainland Greece in the period 1450 - 1200 BC. It was discovered at the beginning of the 20th Century on clay tablets uncovered in Crete, and was successfully deciphered in 1952 by Michael Ventris, an architect with an interest in such things, with the help of John Chadwick, a classical scholar.
In the late 19th Century, it was known that there were some ancient ruins to the south of the city of Iraklion in Crete. It was assumed that they were Ancient Greek, since Greece is littered with such ruins. Arthur Evans, the director of the Ashmolean Museum in Oxford, went out to Crete and started excavating in 1900. What he found took the world by storm: the ruins were much older than had been thought, dating from the period 2000 - 1350 BC, and belonged to a hitherto-unknown civilisation. Evans recalled the Ancient Greek legends of a King Minos who lived in a palace called Knossos on Crete. Minos's palace was protected by a labyrinth and a bull-headed monster, the Minotaur. Evans invented the name 'Minoan' for the newly-discovered civilisation.
The ruins turned out to be a vast building with more than 1,500 rooms. Evans was convinced that this was indeed Knossos, the palace of Minos, and that the vast number of rooms had given rise to the legend of the labyrinth. Frescoes in the palace showed that the people worshipped bulls; Evans identified this as the source of the legend of the Minotaur.
Among the ruins, Evans found more than four thousand clay tablets1 covered in strange symbols. These were obviously writing, but of a sort never seen before. Some of them bore pictures similar to Egyptian hieroglyphs; Evans named these 'Hieroglyphic Script'. Most of them, however, consisted of symbols made of lines, forming squares, triangles, crosses and so on. Evans gave this script the unimaginative title of 'Linear'.
As excavations progressed, it became evident that there were two types of Linear. The earlier tablets, dating from 1650 - 1450 BC, used symbols which appeared to be adapted from the hieroglyphic script. The later tablets, dating from 1450 BC onward, used a slightly different set of linear symbols. Evans named the two scripts 'Linear A' and 'Linear B'.
You can see a simple example of the Linear B symbols here2. Linear A looks very similar.
Evans tried for many years to decipher the two linear scripts, as well as the hieroglyphic script, but failed. He assumed that the tablets were written in the Minoan language, and that it was unrelated to Ancient Greek, since he thought that the Minoan civilisation had already died out when the Greek-speaking people first arrived in what is now Greece.
First Steps at Decipherment
Evans did make some progress in analysis of the scripts. It was he who divided the inscriptions into Linear A and Linear B. He counted the number of different symbols - there were about 120 in each writing system. This made it likely that the script was a syllabic script, where each symbol represents one syllable. Alphabets (such as our own), where each letter represents one consonant or vowel, tend to need less than 40 symbols, while logographic systems (such as Chinese) with a symbol for each word need far more, usually thousands or tens of thousands. Syllabic scripts (such as the Japanese kana system) fall in between.
Syllabic scripts are generally used for languages where there is an alternation between consonants and vowels, such as in the words senorita and Mexico. These can be broken down into syllables, each syllable consisting of a consonant followed by a vowel: se-no-ri-ta and Me-xi-co. There is a different symbol in the script for each possible syllable: sa, se, si, so, su, na, ne, ni, no, nu, etc. Syllabic systems are not so useful for languages like English where there are a lot of consonant clusters, such as the 'str' at the start of 'stripy' and where words end in consonants without a vowel after them.
Further analysis of Linear B showed that there appeared to be a mixture of both pictures and syllabic symbols. Some symbols stood on their own, separated from the rest, and were recognisable as pictures: a horse, a man, a jar. Other symbols were simpler and were not of anything recognisable. These were grouped together with other similar symbols. It was assumed that these were the syllables of words, although Evans felt that some of them might serve other purposes.
Evans correctly guessed that the small vertical line which occurred regularly throughout the inscriptions marked the end of a word. Since some words occurred at the ends of lines and were continued on the next line, it was clear that the script was written from left to right. He also managed to figure out the system of numerals, with symbols for 1, 10, 100 and 1,000 repeated as often as necessary to make numbers. Many of the tablets were obviously lists of items, with each line containing a word, a picture and a number. At the bottom was a word presumably meaning 'total', and the actual total. They appeared to be stock lists.
Evans's other contribution to the decipherment was to notice that there were two particular symbols, one looking like a box and the other like a backwards '5', that were often used to end words that were otherwise identical. He took this as an indication that the language used the system of changing the ending of a word to denote the word's function that is common in most European languages. This is known as 'inflection'.
Evans noticed a slight correspondence between some of the Linear symbols and the symbols of the Cypriot writing system (from the island of Cyprus), which was used a thousand years later and which had been deciphered, but he dismissed them as coincidence: the Cypriot alphabet was used for writing Greek, and he believed that Linear A and Linear B could not possibly be related to Greek, as they were far too old.
Linear B Found Outside Crete
In 1939, an archaeologist by the name of Carl Blegen found about 600 tablets written in Linear B in the ancient palace of Pylos. This was on the Greek mainland. Pylos was built by a different civilisation, the Mycenaeans, who flourished in Mainland Greece from about 2000 BC to about 1200 BC. They were quite a different culture to the Minoans, with different art, burial customs, gods and rituals.
This was a mystery, because if Linear B was Minoan, why was it being used by the Mycenaeans? Had the Minoans conquered the Mycenaeans and imposed their language on them? Or had the Mycenaeans conquered the Minoans, and was Knossos ruled by the Mycenaeans? Yet the artefacts discovered in the two sites (Knossos and Pylos) indicated two distinct cultures.
Over the next 50 years, Linear B tablets have been found in other locations, both in Mainland Greece (Mycenae, Tiryns and Thebes) and in Crete (Chania).
The next stage in the decipherment of the script was carried out by an American archaeologist by the name of Alice Kober. Over the period of 1943 to 1950 she studied the Linear B texts in detail and came up with some remarkable results. She assumed that Evans was right that the language used inflection. This means that the endings of words change in a predictable way. This is something which is not common in English, but a simple example would be the following:
Note the final syllables: -da -das -dian, -na -nas -nian. Using a similar reasoning, Kober examined many words and found ones in which the first few syllables were the same but the endings were different. By tabulating many such endings, she was able to build up a table of ten symbols:
Here, each symbol represents a consonant followed by a vowel. Kober deduced that all the symbols in the first column share the same vowel, although she didn't know what vowel it was. Similarly, all the symbols in the second column share the same vowel, a different one from that in the first column. Each row represents a separate consonant. Kober's work was unique, in that she deduced relationships between the vowels and consonants without any idea which vowels or consonants were being represented. Sadly, Alice Kober died in 1950 at the age of 43, before she had a chance to see the use to which her work was put.
In 1936, a group of schoolboys were visiting the Royal Academy of Arts. The 86-year-old Evans, who happened to be there to give a lecture, showed them around the section on Greek and Minoan Art. Michael Ventris was one of the boys, only 14 at the time. He asked Evans one question: 'Did you say the tablets haven't been deciphered, sir?' From that time onwards, Ventris was fascinated with the idea of deciphering the Linear scripts.
Ventris specialised in the classics in school, but never went on to college. He started work as an architect, so it was in the spirit of a hobby that he tackled the problem of Linear B in 1950, when he was 28. He decided to keep an open mind as to what language the inscriptions were written in.
The Cypriot Clue
Cypriot was a writing system which was used in the 1st Millennium BC in Cyprus. This was about a thousand years after the use of Linear B. Cyprus is a Greek-speaking island, and the Greek that is spoken there is very closely related to the Greek spoken to this day near Mycenae, so it is reasonable to assume there might be a connection.
Cypriot is also a syllabic writing system. Each of the 56 symbols represents one syllable: either a pure vowel or a consonant followed by a vowel. The Cypriot symbols are not the same as the Linear B ones, but there are a few (about seven) which look similar. It is conceivable that it arose from Linear B but that the symbols changed shape over the years.
Cypriot uses the 'se' symbol quite a bit to denote a final 's' sound in a word, a feature which is very common in Greek.
Ventris started out by extending the work done by Kober. He built up a much bigger table of consonants and vowels by analysing different words that started with the same symbols. He used clues such as that certain tablets were clearly lists of women's names, with a picture of a woman after each name, while others were lists of men's names, with a picture of a man after each name. The men's names tended to end in certain symbols while the women's ended in different symbols. This allowed certain symbols to be grouped together.
Other symbols appeared very often at the start of words, but also in the middle of words, suggesting that they were pure vowels with no consonant attached.
Eventually, Ventris had a grid with all the most common Linear B symbols on it, showing which ones shared the same vowel, and which ones shared the same consonant. But he still didn't know what the actual vowels and consonants represented were, although he had suspicions about some of them.
You can see Ventris's grid3 at the time of decipherment here:
Ventris did not want to rely too heavily on correspondence between Linear B and Cypriot, since there were major differences. About 90% of the Cypriot symbols didn't match the Linear B ones at all, and Linear B did not have any single character that was regularly used at the ends of words to correspond with the Cypriot 'se'. This did not seem all that surprising since the final 's' sound is a feature of the Greek language and there was no reason to believe that Minoan or Mycenaean had the same feature. However, he needed some starting point for assigning actual vowels and consonants to the grid. In the end, he used just two similarities with the Cypriot system in his decipherment.
As an experiment, Ventris tried making a few assumptions. After each, he looked for supporting evidence and contradictions. He found very few contradictions, and much to support his case, leading him further and further along the road to full decipherment:
It seemed likely that one of the 'pure vowel' symbols, v5, was the vowel a, since it is the most common vowel in virtually every language. This meant that every symbol in the v5 column ended in a.
He assumed that the symbol with consonant c8 and vowel a was na, because it looked very similar to the Cypriot symbol for na. This gave c8 = n.
He assumed that the vowel v1 was i because the symbol in position c6 v1 of the grid was identical to the Cypriot symbol di.
He compiled a list of words that were very common in the Knossos tablets but not used at all in the Pylos tablets. He assumed that these were the names of towns in Crete.
He looked in this list for the name 'Amnisos', the harbour town of Knossos, a name which surely must appear often. This should have a pattern of syllables a-m?-ni-so-... The ending could change depending on where the word was used. He already knew the symbol for a and for ni. He found only one match in the list of names, with the symbol at c9, v1 in second position and the symbol at c7, v2 in fourth position. This showed that c9 = m, c7 = s and v2 = o. The name was written as a-mi-ni-so.
Now he looked at other names in the list of towns. One very common one had the pattern ?o-no-so. This was probably Knossos itself, written as ko-no-so. Already some spelling rules are starting to appear - final s is omitted, and two consonants followed by a vowel are written as two separate syllables with the vowel repeated.
A third place name was tentatively assumed to be Tylissos, represented as tu-li-so. This provided the fact that c11 = l, although later he discovered that it is also used for r.
Next Ventris tackled a word which existed in both the Knossos and Pylos records, apparently representing some sort of a foodstuff or fuel, as the picture beside it showed a jar with a lid. This word had slightly different spellings in Knossos and Pylos:
The fact that one ends in ?o-no and the other in ?a-na suggest that this is actually a syllable starting with two consonants. He suggested the word koliyadno, an ancient word closely related to the modern name for the spice coriander. If this was the case, then it confirmed Ventris's suspicion that the third syllable was ya and that c1 = y. It also provided c5 = d.
In this manner, Ventris progressed to decipher all the letters in his grid. As he worked, he noticed that the endings of the words corresponded very closely with what he knew of Ancient Greek. Gradually, he realised that the language he was studying actually was Ancient Greek. This was despite all the experts' opinions that it couldn't possibly be. Ventris decided that the time had come to let the world in on his discovery.
Ventris had been very open about his progress on the decipherment. Throughout the period, he had written down all his findings as 'Work Notes' and circulated them to anyone he thought would be interested. His final results were announced in Work Note 20, in June 1952, in which he tentatively outlined his theory that the language of the Linear B tablets, even in Knossos, was Greek.
Just after he issued this, he was asked to talk about the Minoan Script in general on BBC Radio 4. He gave his talk, and briefly outlined his new theory.
John Chadwick (1920 - 1998) was a comparative linguist. He studied Greek dialects, analysing the differences and attempting to reconstruct the mother language. In effect, he studied a type of Greek which had never been seen written down. In fact, he was reconstructing the language of Mycenae, the very language that was on the Linear B tablets. In June 1952, Chadwick heard Ventris's tentative announcement on the radio that he had deciphered the writing system and that it was Greek. He was impressed with what he heard, so in July, he contacted Ventris and offered to assist him. Ventris accepted.
From that time on, Ventris and Chadwick worked together, using Ventris's knowledge of the symbols and Chadwick's knowledge of the language to produce a full decipherment. The Mycenaean Greek was shown to be different from the Ancient Greek of Homer, but in line with the predictions of the comparative linguists. They published their work together in 1953 in an article entitled 'Evidence of Greek Dialects in the Mycenaean Archives'.
Their findings tell us two interesting things. Firstly, the Mycenaeans spoke Greek. This is something that had not been known for certain before. It had always been known that the Greek-speaking Dorians did not arrive in Greece until about 1000 BC, and it was also known that the Mycenaeans were not Dorian. Secondly, Knossos was controlled by Greek-speaking people from about 1450 BC onwards. The most reasonable explanation is that the Mycenaeans invaded Crete and conquered it.
Initially, Ventris and Chadwick received some praise and some disbelief from other scholars in the field, although the harshest judgements appear to have come from people who didn't read the document. Soon, however, a tablet came to light which was clearer than any previous ones: among the Linear B text were pictures of pots and jars which clearly had three legs, four handles, two handles, no handles and so on. The newly deciphered text said just that: 'pot with three legs', 'jar with no handles', 'jar with four handles', etc. This was such a good confirmation of Ventris and Chadwick's work that virtually all opposition to the theory was dropped and the decipherment was accepted as being the genuine article.
Linear B Revealed
The full Linear B system can be seen in the Linear B Table:
Each symbol in the main grid is either a pure vowel (the first row) or a consonant followed by a vowel, with the consonant shown on the left and the vowel shown at the top.
Linear B is like a shorthand system, with much abbreviation. It does not distinguish between many consonants. Aspirated consonants such as ph, th, etc are written using the same symbols as their unaspirated versions (p, t, etc)4. Certain other sounds which are quite clearly different are written using the same symbols:
- The sounds l and r use the same symbol.
- The sounds b and p use the same symbol.
- The sounds g and k use the same symbol.
- The sounds gw and qu use the same symbol.
In addition, Linear B omits many letters: the letters l, r, s and n at the end of a syllable are omitted, as is the letter s before another consonant at the start of a syllable.
Linear B handles syllables with two consonants followed by a vowel by breaking them up into two syllables and repeating the vowel: kno becomes ko-no; mni becomes mi-ni.
These can lead to much ambiguity. However, all examples of Linear B so far have been palace stock records. In this limited area, the users of the script would have had no problem interpreting the script.
Ventris and Chadwick went on to publish a detailed account of the decipherment in 1956 under the title 'Documents in Mycenaean Greek'. Tragically, just after this was published, Michael Ventris was killed in a car accident, at the age of 34.
Other researchers went on to translate most of the tablets from Pylos and Knossos, and additional tablets which were discovered later in such locations as Thebes and Chania. All of them are store records, giving us a detailed look at what went into the day-to-day running of a Mycenaean palace, but no histories, no poetry and nothing that could be counted as literature have yet been discovered. It is possible that, due to the shorthand nature of Linear B, it would not have been practical to write down such works.
One interesting discovery is that writing was not just the job of a few scribes. All the tablets in Pylos were made within a single year. Presumably the tablets would have been wiped clean at the end of the year for a new accounting year, but due to a fire, the tablets were baked hard, making them permanent. Analysis of the handwriting on the tablets shows that they were written by more than 50 different people. The palace in Pylos is big, but not big enough to support 50 scribes doing nothing but writing records. This suggests that everybody with a responsible job was expected to keep their own records. So writing was a skill of every educated person. But no writing has ever been uncovered outside of the palaces. Probably it was never a skill of the peasants, and also, most agricultural land has been ploughed over thousands of times since the age of the Mycenaeans.
No progress has been made in the decipherment of Linear A, which has only ever been found in Crete. It is assumed that Linear A represents the original Minoan language. Many of the symbols are the same as those in Linear B, but using the Linear B sounds to read Linear A gives us a meaningless string of syllables. These may be in fact the Minoan language, but we have no way of knowing since we aren't familiar with that language. Alternatively, there may be a different set of sounds for these symbols. Perhaps some future scholar, with as much determination and persistence as Michael Ventris, can solve the mystery of Linear A.
The short example of Linear B given above can now be decoded: a-re-ka-sa-da-ra. This has been found as the name of a woman on one of the Linear B tablets. There are a few different ways this could be read, but taking ka-sa as ksa and taking da-ra as dra, we get a-re-ksa-dra which is quite clearly the name Alexandra. It is strange to think that there were people with this name living more than 3,000 years ago!
- The Decipherment of Linear B - John Chadwick
- The Story of Decipherment - Maurice Pope
- The Story of Writing - Andrew Robinson