Mysteries Department
Sign inRequest access
Proto-cuneiform: the archaic texts of UrukEvidence
Evidence · Proto-cuneiform: the archaic texts of Uruk

A blind rebuild from the raw data gets the same gap: Uruk names 1.34, Sumerian names 3.6–6.2

Re-implemented from scratch by a separate agent that saw neither the original code nor its numbers, the element test gives the Uruk III names an odds ratio of 1.34 (1.06–1.61) against 3.56–4.12 for Early Dynastic Sumerian names, or 4.50–6.18 when divine and place names are left out of the ordinary-word baseline as in the original.

Supported0

What it means

Openunchanged
Openunchanged
Answer this evidence bears onThe names lack the Sumerian name profile
Supports5 for, none against
Still possibleunchanged
What it implies
  1. The names gap does not depend on one implementation: an independent build gets 1.34 against 3.6 to 6.2.For: Do the roughly 440 personal names in the archaic texts follow Sumerian, Semitic or some other naming pattern?
  2. The names evidence against a Sumerian-speaking origin of the Uruk names stands after an independent replication.For the key question: What language, if any, do the proto-cuneiform texts record?

Next steps 12

1
Screen continuity matches against archaic lexical spellings
Adds evidenceGood odds~2 h
On: Do the roughly 440 personal names in the archaic texts follow Sumerian, Semitic or some other naming pattern?
20score
Details

Go through the 27 matched archaic-to-ED names. Drop any that depend on a spelling first attested after ED I-II, such as the EN.E2 spelling of Enlil, and rerun the null comparison. If the excess over the null holds, continuity stands.

2
19score
Details

Compare the full sign profiles of Schøyen/Brussels names and excavated Jemdet Nasr/Uruk names (shared vocabulary, Jensen-Shannon distance against a permutation null) to see whether they form two distinct naming populations.

3
19score
Details

For each matched name, estimate the probability of a chance match from the frequency of its signs in ordinary entries and in Early Dynastic names, and rank the matches by strength.

4
19score
Details

Repeat the bridge only for archaic names that contain a sign occurring in under 1% of names, weighting each match by how unlikely it is under the shuffle null. A surplus there would point to real survivals; none would leave the names with no route to a reading.

Ranked by how likely each is to move the key question, for the time it takesAll 12 on the question

How it was found 1 run

Result. An independent build gives the Uruk names 1.34 against 3.6–6.2 for Early Dynastic Sumerian names, reproducing the original 1.35 against 4.5–5.8.
Token odds ratio of Englund's Sumerian name-element signs in personal names against other words, 95% text bootstrap{'ED I-II': [4.12, 2.38, 9.03], 'ED IIIa': [3.56, 3.14, 4.01], 'ED IIIb': [3.72, 3.56, 3.88], 'Uruk III-IV': [1.34, 1.06, 1.61], 'ED, proper nouns out of baseline': {'ED I-II': 6.18, 'ED IIIa': 4.92, 'ED IIIb': 4.5}, 'Uruk, baseline from name-bearing texts only': [0.92, 0.6, 1.66]}Replicates. The archaic odds ratio matches to the second decimal; the Sumerian values depend on whether divine and place names count as ordinary words (3.6–4.1 if they do, 4.5–6.2 if not, as in the original). The gap holds in every variant.
Archaic personal names in proto-cuneiform (editor-marked, CDLI)Early Dynastic Sumerian personal names, lemmatized (Oracc ePSD2 admin corpora ED I–II, IIIa, IIIb)code/run.pycode/uruk.pycode/oracc.pycode/stats.pycode/elements.py2 outputs

Sources 1

SRC

Cite it

Mysteries Department. (2026). A blind rebuild from the raw data gets the same gap: Uruk names 1.34, Sumerian names 3.6–6.2 [Evidence EVD0DG9GMZTP7E0]. https://mysteriesdept.com/r/EVD0DG9GMZTP7E0
Private beta Downloads are for beta members

We're in a private beta while we get things right. Everything here is free to read; downloading datasets, sources and code is open to members for now.

Apply to the betaSign in
It takes a couple of minutes. We'll reply by email.