The Riddle of the Labyrinth (18 page)

       • The verbs
and
, are (order unknown) “to blow,” “to cry,” “to breathe.”

The rest of the symbols can be unraveled in similar fashion, until the meaning of all eleven is established. It will help to note that many of the English words refer to parts of the body or bodily functions. It will also help to think about the multipart form that most of the symbols take: This form may well encode a meaning that is the sum of the individual components.

CONFRONTING THE LINEAR B inscriptions, Kober and the other analysts faced a similar deductive process, gridded up a thousandfold. But to her great disgust, most investigators persisted in looking at the problem through the wrong end of the telescope, seeking first to identify the language the Minoans spoke and only afterward to unravel the script.

Everyone, or so it seemed, had a theory about what language the tablets recorded. Michael Ventris, for instance, who had begun thinking fervently about the problem as a schoolboy, was convinced it was the lost Etruscan tongue. He clung steadfastly to the idea until weeks before his decipherment, when he was forced to contemplate a vastly different scenario. Others held even stranger notions. “It is possible to prove, quite logically, that the Cretans spoke any language whatever known to have existed at that time—provided only that one disregards the fact that half a dozen other possibilities are equally logical and equally likely,” Kober said at Yale in 1948. “One of my correspondents maintains that they were Celts, on their way to Ireland and England, and another insists that they are related to the Polynesians of the Pacific.”

Kober indulged in no such speculation, remaining firmly agnostic on the language of the script. (“I am interested,” she once said, “only in what the Minoans wrote.”) The tablets, she insisted, must be analyzed based on internal evidence alone, and the scripts must be allowed to speak for themselves, absent the decipherer's prejudice.

“We cannot speak of
language
, but only of
script
, because the system of writing cannot be read,” she once said. “Since all we know about the language or languages of the scripts is what the graphic remains show, a thorough understanding of these written documents is necessary before any linguistic theories are promulgated. Otherwise it is impossible to avoid reasoning in a circle.”

By the mid-1930s, with two hundred inscriptions at her disposal, Kober could begin to sift the teeming mass of symbols for the kind of internal clues she sought. And so, hour by hour and symbol by symbol, she began to count.

A LANGUAGE AND a writing system are not remotely the same thing, though each impinges on the other revealingly. In any given language, sentences can be deconstructed into words, and words further deconstructed into smaller component parts. These parts occur, importantly, “in patterns of selection and arrangement,” as the anthropologist E. J. W. Barber has said. The script used to write that language will display corresponding patterns.

A simple example: In any inscription, Barber writes, “each sign bears a relation to the signs adjacent to it.” In Italian, for instance, it is permissible for
s
to be followed by
f
or
g
at the start of a word:
sforza
(“force”),
sfumato
(“smoky”),
sgraffito
(“drawing, writing”; compare English
graffito
and its plural,
graffiti
). Not so in English, for no reason other than that is simply how English chose long ago to behave. As a result, certain characters will crop up side by side in certain positions in written Italian but not in written English. Tabulating such behavior helps the decipherer seeking the language of an inscription narrow the field of likely suspects.

When Kober began her work, she kept her tabulations in a series of notebooks. In just a few years, she would fill forty of them—the twentieth-century account books in which she itemized the Bronze Age account books of the Minoan kingdom. But during World War II and afterward, paper was scarce, and she had to resort to enterprise (and occasional genteel larceny) to keep her statistics going. When she could no longer get notebooks, she began hand-cutting two-by-three-inch “index cards” from any spare paper she could find: church circulars, the backs of greeting cards, examination-book covers, checkout slips from the college library, and whatever else she could lay her hands on.

On the front of each card, Kober inked statistics for various signs in the minutest of writing. On the backs of the cards, bits of residual text, like fragmentary inscriptions on broken tablets, gave evidence of their original use:

 

Before her death, Kober would cut and annotate more than 116,600 two-by-three-inch slips, as well as more than 63,300 larger slips—some 180,000 cards in all. The smaller ones she fitted neatly into empty cigarette cartons, the one paper product of which she appeared to have no short supply. Even now, more than six decades later, to open one of them is to be met with a faint whiff of midcentury tobacco.
FLEETWOOD IMPERIALS: A CLEANER, FINER SMOKE,
her ersatz file cabinets say.
HERBERT TAREYTON: THERE'S SOMETHING ABOUT THEM YOU'LL LIKE
.

A WRITING SYSTEM is a woven fabric, an interlaced network of sounds and symbols. As with real woven cloth, there are many different interlacements possible—many different ways, that is, in which sound and symbol may be bound up together. When a decipherer confronts an unknown language in an unknown script, she must hunt for a thread that, pulled, will begin to reveal the structure of the weave. She does this by searching for patterns, for every combination of language and script, as Barber writes in
Archeological Decipherment
, bears “a distinctive fingerprint . . . which may help us recognize the language involved.” It was for this reason that Kober had done fifteen years' preparation, learning languages that employ diverse array of scripts, from logographic (Chinese) to syllabic (Akkadian) to alphabetic (Persian).

Her task would have been so much easier if only the Minoans had used an alphabet. A syllabary is far harder to decipher, because in a syllabary nearly every character stands for more than one sound. As a result, frequency counts skew very differently for syllabic scripts than they do for alphabets, and the same language will behave very differently statistically written with the one than with the other. A two-syllable word written syllabically may entail precisely two signs. Written alphabetically, as Barber notes, the same word “may be represented by from two to a dozen signs”—think of the vast difference in length between two-syllable English words like
Io
and
strengthened
.

As Kober was painfully aware, there were no published frequency tables for syllabically written languages. That is where her homemade file cards came in, helping her corral the seemingly random mass of Linear B characters into orderly statistical sets. For every word on the tablets—the two hundred published inscriptions gave decipherers about seven hundred different words to work with—she cut a separate card. On each card, in her tiny hand, she recorded as much data as she could mine about the characters it contained.

She catalogued the frequency of each character, of course, but she catalogued a great deal more than that. She noted the frequency of each character in any
position
in a word (initial, second, middle, next-to-last, and final); the characters that appeared before and after every sign; the chances of a given character's occurring in combination with any other character; repeated instances of two- and three-character clusters; and much else.

In a letter to a colleague in 1947, Kober itemized the time it took her to compile a single statistic, the one that tallied a sign's frequency in combination with others: “You can figure out for yourself how long it will take to compare each of 78 signs with 78 other signs, at 15 minutes (with luck) for each comparison. Let's see, 78 times 77 times 15 minutes—that's about 1,500 hours. I did it on the little slide rule I just bought to hasten the arithmetic I'll have to do.”

After calculating her figures, Kober, using a dime-store hand punch, made holes in precisely determined spots on her cards. Each spot stood for a particular Linear B character: There was a designated spot for
, for instance, another for
, another for
, and so on. The location of each character's hole remained constant from card to card. In this way, by stacking two or more cards together, Kober could see which holes aligned. That told her instantly which characters those words had in common. “Making all these files takes time, and the files have considerable bulk,” she wrote in 1947, “but once they are finished, all the material is so arranged that anything, no matter how unexpected, can be checked in a few minutes.”