To understand properly what was at stake, we need to go back ten years, to the last few years of Evans’s life. Notwithstanding his great four-volume publication, The Palace of Minos – with regard to the Minoan script, Evans left a disorganized legacy. In 1909, he had published (with Oxford University Press) Scripta Minoa, but this contained the two other scripts discovered in Crete, the Hieroglyphic and Linear A, hardly any Linear B tablets. Of the more than 3000 Linear B tablets and fragments (there were about 1600 tablets) excavated by Evans and others at Knossos, only two or three hundred had been published by 1941, the year of Evans’s death, most of them only in the 1930s, along with a sign list seriously flawed by Evans’s falling for the pictographic fallacy mentioned in chapter 1 (many of the signs he had read as pictograms/logograms were really phonetic signs). The arduous task of completing their publication now fell to Myres, who was already well into his seventies. He would labour at Scripta Minoa, volume 2, for ten years with fading eyesight and help from a few others, chiefly Kober and Bennett, but he was in an impossible position: trapped between a loyal desire to keep faith with the faulty sign list prepared by Evans and the clear perception of Kober, Bennett and Ventris that a more logical, scientific approach to classifying the signs was required.
There was also the startling fact that more Linear B tablets had recently been discovered, this time not in Crete but in mainland Greece, in the western Peloponnese. In 1939, the American archaeologist Carl Blegen, having completed a famous dig at Troy, had struck luckywith his first trial trench at a place he believed to be the site of ancient Pylos, the city made famous in Homer’s Iliad as the seat of King Nestor. The result was almost 600 new pieces of Linear B – and a serious embarrassment to the Evans theory that Linear B was exclusively the writing of the Minoans. For if this were so, what were Linear B tablets doing in large quantity in mainland Greece? The 88-year-old Evans did not respond to Blegen’s find, but his followers rapidly came up with explanations, such as that the tablets at Pylos were ‘loot from Crete’ or that an illiterate Greek ruler had raided Minoan Crete and carried off its scribes to work in his own palace at Pylos. Whatever proved to be the truth, Blegen’s discovery was bound to have a profound effect on all scholars working on the decipherment of Linear B.
There was just time before the second world war intervened for the new pieces to be cleaned, mended and photographed, and then deposited in the Bank of Athens, where they remained intact during the next few turbulent years. The photographs were taken to the United States on the last American ship to leave the Mediterranean in 1940, after Italy declared war. Blegen entrusted their analysis to Bennett, his doctoral student at the University of Cincinnati, but he could get down to the task only after doing his war service as a cryptanalyst.
Throughout the 1940s, therefore, the situation for active research on Linear B – as opposed to scholarly speculation – was complicated and unsatisfactory. The tablets themselves were mostly inaccessible, in storage in Athens and Crete. Not very clear photographs and probably somewhat inaccurate drawings by Evans were under scrutiny in Britain and America by Myres, Ventris, Kober and Bennett, at first independently of each other (though keeping in touch by correspondence). Myres in Oxford was examining the entire set of Knossos tablets, but not those from Pylos, and would not publish them in Scripta Minoa, volume 2, until 1952; Ventris in London could work only with the few tablets published by Evans and others; ditto Kober in New York, until she began to help Myres in 1947; Bennett at Cincinnati and then Yale University had the Pylos archive, but comparatively little access to the Knossos tablets, which were obviously essential for comparison in compiling a definitive Linear B sign list. (Eventually, in 1949, Bennett was shown the Knossos tablets, in exchange for showing the Pylos tablets to Kober.) Overall, the situation was a mess, though not a hopeless one.
Nevertheless, during this time, both Bennett and Kober were able to carry out analyses that would be as vital to Ventris as the ground-breaking work of earlier scholars of the ancient Egyptian script (especially Thomas Young) had been to Champollion in deciphering the hieroglyphs. But their analyses were different in kind. Bennett’s work may be likened to clearing the terrain of jungle and straightening the path; Kober’s was more in the nature of proposing a methodology that would enable the decipherment to move forwards along a path of progress.
First, Bennett proved that while the numerical systems of Linear A and Linear B were very similar, the systems of measurement were not. Linear A has a system of fractional signs, e.g. ½, ⅔, ¾, while Linear B records fractional quantities in terms of smaller units, like pounds and pence or feet and inches. Besides being useful information, this added further weight to the suspicion that Linear B represented a language different from that of Linear A – probably from outside Crete since Linear B had been found in mainland Greece, unlike Linear A. (Evans had of course believed that both scripts wrote ‘Minoan’ – a view shared by Myres and Ventris, but not by Bennett and Kober, during the 1940s.)
More important, though, was Bennett’s wrestling with the thousands of text characters in the Pylos tablets, to produce a sign list in which some 89 signs – presumably (but not yet provably) phonetic in function – were logically distinguished from each other. To do this, he had to identify allographs (equivalents) of the same sign, e.g. ‘a’ and ‘’, ‘k’ and ‘’, ‘s’ and ‘’, if we take the roman script, and also the same sign written by two and more different scribal hands, i.e. handwriting discrepancies. When we know what the signs and sign groups mean in a language, such identification is quite easy, even if we are sometimes foxed by someone’s illegible handwriting. Imagine, though, that you have to read some hand-written, and possibly semi-scribbled text in an unknown script and language – how would you know that the sign group ‘’ was the same word as ‘’? In addition, Bennett had to findwaystodistinguishthe 89 ?phonetic signs from a second class of signs, pictographic/iconic, which were apparently used as logograms (for instance, signs such as and on page 25). Evans and others had already done this, but there was absolutely no guarantee they were correct in their identications. By dint of careful analysis, Bennett managed to classify correctly signs such as the ‘double-axe’, , and ‘throne-and-sceptre’, , – which might have been thought to be logograms from their appearance – among the 89 ?phonetic signs, not among the logograms.
Besides painstaking visual comparisons of endless sign forms, Bennett achieved his sorting largely by using two techniques. One was a laborious comparison of the contexts of all the characters on the tablets; for example a lone sign which occurred only with numerals and was clearly iconic was almost certainly a logogram. (You could easily guess that ‘£’ or ‘$’ were logograms from a long list of goods with their prices.) The other technique involved frequency analysis of signs and sign combinations – counts of how frequent or infrequent each sign was, both in itself and in combination with other signs, and also in relation to its position within sign groups – which we shall discuss properly later in connection with Ventris’s use of the same statistical technique. To give just one example here, Bennett calculated that the ‘double-axe’ and ‘throne-and-sceptre’ signs were very frequent signs in the Pylos tablets, often found at the beginning of sign groups but also regularly found within sign groups, and seldom found alone like the obvious logograms. If the two signs really were logograms (or determinatives) with some kind of religious or royal significance, as Evans suspected, their high frequency would suggest an unusually large number of persons at Pylos possessing religious or royal honours. And even if one accepted this explanation for the occurrence of the signs at the beginning of sign groups (where a priest’s or ruler’s designation would naturally go), how would one explain their occurrence within sign groups? Thus Bennett ruled in favour of the two signs being more likely to be phonetic than logographic, though he could not entirely exclude the possibility that they could act as logograms at the beginning of sign groups but as phonetic signs within sign groups.
If Bennett was dedicated, the somewhat older Kober, one senses, was driven; and it is hard to write about her without a certain pathos, for she died of cancer in 1950 at the age of only 43, a mere two years before Ventris announced his decipherment. It seems reasonable to compare her with Rosalind Franklin, the unlucky competitor of Crick and Watson in the DNA story (who died young and probably missed sharing their Nobel prize), for Kober was theoretically well placed to have ‘cracked’ Linear B, besides sharing Franklin’s caution and determination (even obstinacy). However, on the evidence of Kober’s published work, it seems doubtful she would have succeeded, as we shall now see.
Ventris’s drawing of Emmett Bennett Jr’s Linear B signary.
(From Work Notes – see Further Reading for publication details)
Superficially, her career was that of a typical classicist of her time. She studied Latin and Greek and took a PhD from Columbia University with a dissertation on ‘Color terms in the Greek poets’, then she began teaching at a college. But as Bennett wrote after her death, in her mid-twenties Kober had developed a ‘consuming interest’ in the undeciphered scripts of Crete. It would do her academic status no good (ditto for Bennett), yet as soon as she could, she set about learning as many ancient languages as possible, chiefly in order to be able to eliminate them as candidates for the languages of Linear A and B, while also studying archaeology in the field – in New Mexico and Greece – and, even more determinedly, mathematics (for its use in statistics) and physics and chemistry (for their methodology). Everything, including marriage, was sacrificed to her pursuit of the Minoan scripts.
The result was a series of important, ruthlessly logical papers on the Cretan scripts published between 1943 and her early death, notably in the American Journal of Archaeology. (Not surprisingly, she firmly ignored Ventris’s speculative, ‘Etruscan’, 1940 paper in the same journal.) Their distinctive feature was Kober’s conviction that with enough material available, there was no absolute need for a bilingual inscription like the Rosetta stone: it should be possible, simply by an intelligent search for patterns in the unknown Linear B characters, to determine the nature of the ‘Minoan’ script and its language, and hence, if the language was in fact related to a known language, to decipher Linear B.
Her most important practical contribution to the decipherment came from a suggestion originally made by Evans: that there was evidence of inflection (declension and conjugation) in Linear B. She took it up in a paper, ‘Inflection in Linear Class B’, published in the AJA in 1946. Kober was of course familiar with declension in Latin and Greek, where nouns are inflected according to their number (singular/plural/dual) and gender (masculine/feminine/neuter), and their case (nominative/accusative/genitive/dative, e.g. dominus/dominum/domini/domino), and verbs are inflected as they conjugate (e.g. amo/amas/amat/amamus/amatis/amant). There is relatively little declension/conjugation in English (e.g. potato/potatoes or the Latin-derived hippopotamus hippopotami, I love/ she loves they love), more in French (e.g. j’aime/tu aimes/il aime/nous aimons/vous aimez/ils aiment). In Linear B, Kober identified five groups of words taken from various published Knossos tablets, with three words in each group – dubbed ‘Kober’s triplets’ by a slightly teasing Ventris – which suggested to her the presence of declension. She could not know what the words meant, but their contexts in the tablets seemed to be the same, making them likely to be nouns, maybe personal names or place names. That they shared the same context was of course essential, otherwise she might have been comparing groups of three words that were visually similar but, unknown to her, were grammatically dissimilar, which would have rendered a comparison invalid and potentially misleading. (A very rough equivalent might be a comparison of three bus timetables for the same route, one for daytime, the second for night-time and the third for Sundays. This would be valid because the context is the same, whereas a comparison of daytime timetables for three related but different routes would be potentially misleading, even though they might easily share some bus-stop names in common.)
Here are two of Kober’s ‘triplets’:
We can see the inflection more clearly if we highlight the word endings:
An English parallel might be:
Ca-na-da Ar-ge(n)-ti-na
Ca-na-di-a(n) Ar-ge(n)-ti-ni-a(n)
Ca-na-di-a(ns)
Ar-ge(n)-ti-ni-a(ns)
If such parallels were right (assuming that Linear B was syllabic, like the Cypriot script), and would have different consonants (C) but share the same vowel (V), like da and na in Cana-da/Argenti-na, i.e.:
So would and like di and ni in Cana-di-a(n)/Argenti-ni-a(n):
By the same token, using the other three ‘triplets’, Kober arrived at what she called ‘the beginning of a tentative phonetic pattern’:
The phonetic values of these syllabic signs were as yet undetermined, but their interrelationships – on the model of the blank square in a crossword where two words meet in which the shared letter is unknown yet must fit the two words – were (tentatively) established. This analytical principle, called a ‘grid’ by Ventris and others, was seminal in organizing the bewildering mass of inscriptions for decipherment. The concept of a grid, though not the actual word, was not new – it was used in the 19th century in the cuneiform and other decipherments – but it came into its own with Linear B.
Original as the above insight was, Kober somewhat spoilt it by giving a strong hint in one of her papers that she thought the ‘triplets’ were cases of a noun on the Latin model (e.g. dominus/domini/domino). She was wrong about this, as we shall see. On the other hand, careful scholar that she was, in a later paper she did not repeat the Latin parallel but instead contented herself with the observation that, ‘There is enough evidence to make it necessary to investigate the inflection theory thoroughly, and without prejudice. If it is right, more evidence will appear; if more evidence is not found, it is wrong.’
One other result by Kober must be mentioned, without going into her detailed reasons. She demonstrated that the two Linear B words for ‘total’ (page 25), and , are masculine and feminine variants of the same root word: the first appears with the ‘man’ logogram and male animals, the second with the ‘woman’ logogram and female animals. Since both words contain two signs, of which the first is the same, the variant was clearly formed by a change of vowel (or conceivably consonant) rather than by the addition of an extra syllable. The importance of this deduction lies in the fact that Indo-European languages are almost alone in this formation. Hence the language of Linear B was very likely to be Indo-European (which included Greek), and not Semitic or similar to Etruscan, the candidate favoured by Ventris. As for the language of Linear A, Kober had no truck with the lazy consensus that it must be the same as that of Linear B because the two scripts looked quite similar and were found in similar contexts; she pointed out, perfectly logically, that modern roman scripts look similar but are used to write quite unrelated languages. Linear A, she thought, probably wrote a different language from Linear B.
During 1948, an already-ailing Kober virtually bowed out of the Linear B battle with these words, in her article ‘The Minoan scripts: fact and theory’: ‘When we have the facts, certain conclusions will be almost inevitable. Until we have them, no conclusions are possible.’ Ventris read the piece and cr
iticized it to Myres for ending on ‘a rather pessimistic note’. In correspondence with Kober, and when they met in Oxford in 1948, Ventris and she developed little rapport. Although he undoubtedly learnt from her work and would later praise some aspects, he never went back on his basic view that she was too negative about the prospects of a decipherment.
Surely, she did go too far in her article. No science, and certainly no archaeological decipherment, proceeds on such an arid, all-or-nothing basis. The scientist and the decipherer never have all the facts they need, but when they have sufficient to form sensible hypotheses, they can hope to test these hypotheses against existing knowledge and against new facts as these become available. This is where the element of creativity and courage comes in. Alice Kober was probably too restrained a scholar to have ‘cracked’ Linear B. In the published words of Ventris written after the decipherment, her approach was ‘prim but necessary’ (privately, he told another American woman scholar that Kober’s logic was ‘a shade too frigid and destructive for my taste!’). To go further would require a mind like his, that combined her perseverance, logic and method with a willingness to take intellectual risks.
To return to Ventris in August 1948, he did indeed ‘stick to his own last’ – at least for a while. Abandoning Myres’s transcription project, in September he and Graeme Shankland started a year-long course in town planning, a relatively new subject which had acquired real importance with the crucial Town and Country Planning Act of 1947. But Ventris was distinctly dissatisfied with the course. The subject was rather nebulous and ill defined and the course was poorly presented, according to Shankland. ‘Michael was deeply furious about the sloppy and illogical way that most of the lectures were put together.… He found it irritating to be taught by men who were ignorant of logic.’
The Man Who Deciphered Linear B Page 6