The Riddle of the Labyrinth (38 page)

Ventris kept his word at first, pursuing little but architecture for the next several months. Then, in the summer of 1950, he had the opportunity to meet Emmett Bennett in London: Bennett had come to England from Yale to help Myres prepare the
Scripta Minoa II
inscriptions for publication, a task he assumed after Kober's death. Ventris and Bennett, who shared a similar dry wit, developed an immediate rapport. The meeting also gave Ventris the chance to see some of the unpublished inscriptions from Pylos.

After that, Ventris was irretrievably in the grip of the script. He had made out handsomely in the stock market (in the course of a single transaction, Robinson reports, he earned more than he did in a full year at the ministry) and soon quit his job to devote himself entirely to decipherment. Now his real detective work begins, and within eighteen months—between the start of 1951 and the middle of 1952—he would solve the riddle of the script, a solution arrived at through a combination of genius, perseverance, and remarkably inspired guesswork.

DURING THIS PERIOD, Ventris chronicled his progress, with its attendant false starts, hunches, and minute advances, in a series of typed documents he called Work Notes. There would be twenty Work Notes in all, together comprising nearly two hundred pages; these, too, he duplicated and sent to his circle of scholars. The Notes are profuse and often hard to penetrate, filled with statistics, linguistics, etymology, geography, and Ventris's speculations on everything from the formation of words in the language of Linear B to the movements of populations in the ancient world. In the first Note, dated January 28, 1951, he flails more or less blindly. In the last, dated June 1, 1952, he solves the riddle.

Work Note 1 does have two notable features. First, Ventris independently replicates Kober's “slight discovery,” from late 1947, about the sign
, which functioned as “and” when attached to the end of a noun. Second, the note contains his own first attempt on paper at a phonetic grid for the script. (Ventris had previously built a three-dimensional “grid,” now lost, consisting of a board studded at regular intervals with hooks or nails; from these he hung paper tags bearing Linear B symbols. The tags, which almost certainly recalled those he had made for the orderlies' room in the RAF, could be repositioned at will, with symbols assigned to particular rows or columns according to the consonants or vowels they were believed to share.)

In Work Note 1, Ventris sets down his first written grid, modeled on the one in Kober's famous 1948 article. Her grid, repeated here, was small in scale. It was also refreshingly abstract, plotting only
relativ
e
values among the signs while refusing to assign hard-and-fast sound-values to any of them:

Ventris's grid was far more ambitious. Where Kober had confined herself to ten signs, he, in his crisp architectural hand, plotted twenty-nine. He also rearranged many of Kober's original ten, which was a mistake: As the decipherment would show, she had plotted the relationships among those signs with 100 percent accuracy. Worst of all—and this would have had Kober spinning in her grave—Ventris proposed sound-values for all twenty-nine of his signs. They seemed to derive largely from analogies to sounds in ancient Etruscan, for he was still committed to an “Etruscan solution” to the problem. It was the position he had put forth in his youthful article of 1940, and one to which he would hold fast until only weeks before his decipherment. Ventris's first grid “must be regarded as a failure,” Maurice Pope writes in
The Story of Archaeological Decipherment
. Nearly 70 percent of its values were wrong.

Then, in the spring of 1951, Ventris received a copy of Bennett's newly published book,
The Pylos Tablets
. Long awaited, it represented the first substantial published transcription of Linear B tablets anywhere;
Scripta Minoa II
, containing Evans's Knossos tablets, remained unpublished. (When Ventris went to the post office to pick up the precious package, as Robinson recounts, “a suspicious . . . official asked him: ‘I see the contents are listed as
PYLOS TABLETS
. Now, just what ailments are pylos tablets supposed to alleviate?'”)

The Pylos Tablets
contained the first established signary for the script, an inventory of its more than eighty syllabic symbols, carefully teased out from the tangle of signs drawn on clay. Having a signary made serious analysis of the script possible at last, for without it, investigators were little better off than our bewildered alien, adrift in Times Square. The Linear B signary has long been attributed to Bennett, and with ample justification, for he worked for years to compile a definitive list. But it is also clear from his long correspondence with Kober on the subject—a blizzard of back-and-forth about which signs were the same and which authentically differed—that she deserves as much credit as he.

Bennett chose to arrange his signary by character shape, with simpler, straighter characters coming first and more complex, curvier ones coming later; until true sound-values could be assigned it was as good an organizing principle as any. His signary looked much like this:

Note that some symbols, like the little pig in the last row, function both as logograms and as syllabic characters in Linear B. Other writing systems likewise conscript symbols for double duty: In the Roman alphabet,
K
can be phonetic, standing for the hard “c” sound, or (as used by a baseball-loving American) it can be logographic, denoting the word
strikeout
.

Now, with a bounty of inscriptions to work from, Ventris could start to make real progress. He began to count the Pylos characters, gradually compiling a set of statistics on their frequency. This let him impose order on the unruly mass of inscriptions, and this newly imposed order led him to his first truly significant discovery.

Ventris sorted the Pylos characters into three groups: Frequent, Average, and Infrequent. Much as Kober had done with the Knossos inscriptions, he also tabulated their use in various positions in Linear B words. As a result, he saw something striking: Five characters—
and
—appeared especially often at the beginnings of words.

There are different types of syllabaries, with the type depending on how large a syllable each character represents. The Linear B syllabary, as investigators had long known, was a “CV” syllabary, with each character standing for one consonant plus one vowel. A CV syllabary can spell most words adequately—as long as they start with a consonant. But what happens when a word starts with a vowel? Ventris realized that the five characters he had isolated were the exceptions to the one-consonant-plus-one-vowel rule: They were used to write the “pure vowels”—“a,” “e,” “i,” “o,” and “u”—found often at the beginnings of words.

He would eventually identify these characters as follows. (Note the presence of
, whose “throne-and-scepter” shape had seduced Evans into calling it a determinative):