The life and death of the cuneiform script
Cuneiform served Sumerian, then Akkadian, then a heavily logographic late scholarly idiom across more than three millennia — the longest natural experiment any writing system has run — before dying by degrees and leaving a last dated tablet, an astronomical text, in 75 CE. The death is usually told as that anecdote, but language-death research predicts a structured domain-by-domain retreat — losing the market before the liturgy — that the dated corpus can actually test. The same longevity makes the script a laboratory for what survives change: sign inventories, frequency profiles, and scribal conventions that outlived the very languages they were built to write.
Synopsis
LLM-authored synopsis Not an articleThe standard story of cuneiform's death is a single, tidy fact: the last known dated document, a text tracking the movement of stars, was written in 75 CE. This page gathers conjectures that treat that fact as far too tidy. The script itself carried three successive languages -- Sumerian first, then Akkadian, then a late, heavily logogram-based scholarly register -- across more than three thousand years, the longest continuous run any writing system has managed. Research into how languages actually die predicts that a system this durable should not simply stop working; it should retreat one domain at a time, vanishing from the marketplace long before it vanishes from temple ritual, a pattern the surviving dated tablets can genuinely be tested against.
The genre funnel tests whether cuneiform's use narrowed by genre in a predictable order as it declined, rather than shrinking evenly across all kinds of writing at once. The script's fingerprint asks whether the script's own sign inventory and frequency profile carry a detectable fingerprint of the very different languages it was built to write across its three-thousand-year run, outlasting the languages themselves.
Interpretive synthesis over gathered records only. It creates no article, evidence, inferon, source, or estimate.
Emergent lineage
Related pages, connected by this project's own research.
Related Pages
Other pages that share linked material with this one.
Cuneiform archives (CDLI)
CDLI/BDTNS -- Ur III and Old Babylonian archives, and the Astronomical Diaries.
Record claim
About the production, transmission, survival, or filtering of the written/material record up to the pre-modern past itself -- scribal behaviour and e…
Channel dissociation
Two transmission channels for the same content predicted to carry opposite statistical signatures, including error-spectrum forensics (diagnosing the…
Rank-size laws and heavy tails
Zipf's law in linguistics, Pareto in economics, and Gibrat's proportionate-growth model all say the same structural thing: many social and productive…
Open conjectures — noetic, untested
Falsifiable questions posed below the evidence boundary by the 1001 Conjectures project. They are leads, not evidence, and carry no weight in this page's assessment.
- The genre funnel Already answered
- The script's fingerprint Anticipated in print · untested
From the One Thousand and One Conjectures. They carry no evidential weight here; verdicts and credits live on each conjecture's own page.