The Inflectional Grid

Alice Kober, Michael Ventris, and the Decipherment of Linear B

Stacked inside empty cigarette cartons and biscuit tins in a Flatbush, Brooklyn apartment between 1943 and 1950, 180,000 two-by-three-inch index cards cut from wartime scrap paper held the internal grammar of Bronze Age Crete. Their creator, Brooklyn College classics professor Alice Elizabeth Kober, carried a full teaching load by day and translated textbooks into Braille for blind students. At night she taught herself Hittite, Akkadian, Sanskrit, Basque, and Old Irish to attack a script that had defied every cryptanalyst on Earth for half a century.

In the spring of 1900, British archaeologist Sir Arthur Evans dug into the Bronze Age palace of Knossos on Crete and uncovered more than three thousand clay tablets dated to around 1400 BCE. Scratched into the clay were inventories of sheep, chariots, bronze cauldrons, and workers written in an unknown script that Evans named Linear B. In 1939, American archaeologist Carl Blegen unearthed six hundred more Linear B tablets inside the Mycenaean palace at Pylos on the Greek mainland. Yet Evans, who controlled access to most of the Knossos tablets until his death in 1941, enforced a strict dogma. He insisted that the Minoans of Crete had colonized mainland Greece and that Linear B recorded an indigenous Minoan language that could not be Greek.

To a cryptanalyst, Linear B represented the hardest class of problem in information theory: an unknown script writing an unknown language with zero bilingual inscriptions. Worse, the tablets were terse accounting lists containing no royal genealogies or narrative prose. Kober realized that guessing phonetic values by comparing Linear B shapes to Cypriot or Phoenician letters was futile. Before anyone could ask how a sign sounded, one had to map how the signs behaved mathematically in relation to one another.

Between 1943 and 1948, sorting her 180,000 scrap-paper cards by hand, Kober extracted the structural skeleton of the unknown language:

• - The Syllable Count Bottleneck: Kober established that Linear B used roughly 87 active phonetic signs alongside commodity ideograms. Because an alphabetic script uses 20 to 30 signs and a logographic script uses thousands, a repertoire of 87 signs proved mathematically that Linear B was an open-syllable consonant-vowel ($CV$) syllabary combining five vowels with roughly fifteen consonants.
• - Kober's Triplets (Inflectional Paradigms): In a landmark 1945 paper, Kober proved that the language behind Linear B was heavily inflected, changing word endings to mark grammatical cases and genders. She isolated groups of three words (Kober's triplets) that shared the exact same root signs at the beginning (A-B-...) while varying their final signs across three recurring grammatical patterns: Case I (A-B-C-D), Case II (A-B-C-E), and Case III (A-B-F).
• - The Bridge Syllable and the 2D Phonetic Grid: In an open $CV$ syllabary, when an inflected root ending in a consonant takes different vowel case-endings, the bridge syllable spanning the end of the stem and the start of the suffix must change its sign while keeping the same consonant. Thus Sign C (in Cases I and II) and Sign F (in Case III) shared the same consonant but carried different vowels. Chaining these intersecting triplets together in 1948, Kober built a two-dimensional grid of consonants and vowels without knowing a single sound value.

Kober died of cancer in May 1950. In London, British architect Michael Ventris expanded Kober's grid using Blegen's newly published Pylos tablets. In Work Note 20 (June 1, 1952), Ventris noticed that five prominent inflectional triplets appeared on the Knossos tablets from Crete but were absent from Pylos on the mainland. They had to be Cretan towns (Case III) along with their masculine (Case I) and feminine (Case II) ethnic adjectives. Substituting the Cretan city names Amnisos (A-mi-ni-so), Knossos (Ko-no-so), and Phaistos (Pa-i-to) into Kober's grid triggered a chain reaction.

When Ventris read off the resulting words—po-me (poimēn, shepherd), ka-ke-u (khalkeus, bronzesmith), ko-wo (boy), and ko-wa (girl)—the language proved to be archaic Mycenaean Greek, written seven hundred years before Homer. Ten months after Ventris announced the decipherment on BBC Radio on July 1, 1952, with philologist John Chadwick, Carl Blegen dug up tablet Ta 641 at Pylos stamped with pictograms of three-legged cauldrons and four-handled goblets. Decoded with Ventris's grid, the syllables spelled out the Greek words ti-ri-po-de (tripods) and qe-to-ro-we (four-eared).