- Enlil. Englund's note on KIDₐ [@englund2011kid] and the archaic lexical lists show Uruk III writes Enlil as EN.KID (EN.NUN). The EN.E2 spelling is ED IIIa. The session 8 match of EN AN E2 to {d}en-lil2 was an anachronism; filed against the Enlil reading.
- NI in the lexical lists. NI occurs almost only in the Vessels and Garments lists (oil/fat). GIR3 also appears as a title sign in Lu2 A. ŠUBUR continues into Archaic Ur names [@steinkeller2013prisoner]. Rerun with GIR3 excluded, the NI–body-part pairing holds (5/12, p = 0.007).
- NI in every later name corpus. All ED I–OAkk PN lemmas were mapped sign by sign. NI-names have body parts at the background rate (13–18%), -ni after body parts is barely above -ni after other signs, and no 'oil for the feet' names exist. Jagersma [@jagersma2010grammar] puts the first written grammar in Archaic Ur. Filed against the possessive reading and neutral on the oil reading.
- Name continuity, re-tested. Archaic names were matched against about 17,000 ED–OAkk PN types, against a null that shuffles the names' own signs. Real names match at the shuffle rate (10.9% vs 11.3%, p = 0.63). The session 8 excess came from the name sign stock. Withdrawn and marked superseded.
- Order. Archaic Ur names (ePSD2 ed12, read) put predicates last (SI, GAL, HI, DA 85–93%). Uruk III final-position shares do not predict them (rho 0.01). Pair-order rigidity: Uruk III entries 0.77, Uruk names 0.84, proto-Elamite 0.85 [@dahl2009iran], Archaic Ur 0.96.
- Kushim. On 23 tablets, 21 of them MSVO 3; not in the title lists; 4 entries next to SANGA. The SANGA + X slot is mostly offices and units (61%) rather than names (14%). Undecided, slightly favouring a person by list absence [@nissen1993bookkeeping].
- Titles in names. 11 of 540 name entries contain a witnessed multi-sign title, and none of the NI names do. The names statistics stand.
Research session 9 · Proto-cuneiform: the archaic texts of Uruk
Brute-force checks of the candidate readings, sign by sign against the literature
Literature and lexical-list checks on the session 8 candidates (Enlil, NI, GIR3, ŠUBUR), then systematic corpus runs on all five open leads: NI in later names, name order in Archaic Ur, Kushim, titles inside names, and a proto-Elamite contrast. Two candidate readings fell, and the session 8 name-continuity result was withdrawn.
Flagged noteworthy · score 30Archaic Uruk names recur as Sumerian names only at chance once the sign stock is held fixed
Rules out an explanation. Sent to the editors.
Flagged noteworthy · score 25Sumerian name order appears in the Archaic Ur texts and is absent from Uruk III
First measurement. Sent to the editors.
Going in the sweep before the work
What changed since Session 83 Nov 2026, 05:29 → 31 Oct 2026, 15:20
0datasets added
0sources added
0news stories matched
0duplicates or retired records
Top next step going in: Context of counted GIR3 NI entries
STEPRanked 1 of 41 · could decide · good odds · ~3 h
In brief
- Matched with a null that keeps the names' own signs, archaic names recur as Sumerian names no more than chance; session 8's 'Sumerian layer' was the sign stock, not survival.
- The two best candidate readings fail: Enlil is written EN.KID in Uruk III, and later Sumerian names show no -ni–body-part pairing while grammar is not written before Archaic Ur.
- Sign order in names becomes Sumerian syntax only in Archaic Ur (predicates final), and Uruk III order does not predict it.
What this changes: Against H-SUMERIAN (continuity withdrawn), against the NI possessive and Enlil readings, for phonetic writing beginning at Archaic Ur; Kushim raised and undecided.
What moved1 question changed
What language, if any, do the proto-cuneiform texts record?Do the roughly 440 personal names in the archaic texts follow Sumerian, Semitic or some other naming pattern?this question
Does the name entry EN AN E2 on MS 3035 (P006268, obv. i 1.b09) contain the divine name Enlil, written as in Early Dynastic {d}en-lil2?
+1 evidence, 1 against2 answers proposed
Yes, it is Enlil · proposed this session · 0 for, 1 againstPossible
Not Enlil: the archaic spelling is EN.KID · proposed this session · 0 for, 0 againstUntested
Openwas No evidence yet
Key question
What language, if any, do the proto-cuneiform texts record?
+1 evidence, 1 against1 next step added
Proto-cuneiform records Sumerian · 3 for, 6 againstPossible
Openunchanged
Key question
When and how did phonetic spelling begin in Mesopotamian writing?
+1 evidence, 1 for1 next step added
Phonetic spelling begins at Archaic Ur, c. 2900 BC · 2 for, 2 againstPossible
Openunchanged
What language, if any, do the proto-cuneiform texts record?Do the roughly 440 personal names in the archaic texts follow Sumerian, Semitic or some other naming pattern?this question
Is KU.ŠIM ('Kushim') in the Uruk III barley accounts a person's name or an office?
raised this session+2 evidence, 1 for, 1 neutral2 answers proposed
A person · proposed this session · 1 for, 0 againstPossible
An office or institution · proposed this session · 0 for, 0 againstPossible
Openunchanged
Is the archaic designation GIR3 NI the Sumerian name giri3-ni, with NI writing the possessive -ni ('his')?
+4 evidence, 1 for, 1 against, 2 neutral
NI is a logogram; GIR3 NI is a designation · 1 for, 1 againstPossible
NI writes the possessive -ni in body-part names · 1 for, 1 againstPossible
Openunchanged
What's next3 steps set · ranked
1
Proto-Elamite owner entries vs Uruk names
Adds evidenceFair odds~3 h
14score
Details
Split proto-Elamite entries into owner/person entries (Dahl's ~100 name signs) and commodity entries, and compare their order rigidity and reuse with Uruk names. If proto-Elamite names are more linear than Uruk names, Uruk names were not written phonetically.
2
Rare-sign name bridge with an order constraint
Adds evidenceFair odds~3 h
12score
Details
Restrict the bridge to archaic names with a sign used in under 1% of names and weight matches by their shuffle-null probability; a surplus would be the first real survivals.
3
Test the paywalled phonetic readings
Adds evidencePoor odds~4 h
3score
Details
Get Monaco 2014, Steinkeller 1995 and Wilcke 2005 and test each proposed Uruk III phonetic reading against the same nulls.
Next steps live on their questions, where anyone can take one up. Ranked by how likely each is to move a key question, for the time it takes.
New questions 1
What was done
Runs 6
| Run | Measured | Result | Against | Decision |
|---|---|---|---|---|
| NI–body-part test with GIR3 excluded | Share of multi-sign NI names containing a body-part sign vs resampled base rate | 0.417 | Null median 0.083 (95%: 0–0.333) | The association does not rest only on the ambiguous GIR3. It holds at reduced strength on SZU and SAG; the pattern is 5 name types, so it is small-n. |
| Name bridge against a sign-preserving shuffle null · 1 figure | Share of Uruk multi-sign name types whose sign multiset equals an ED/OAkk PN, vs names rebuilt by shuffling name signs | 0.109 | Shuffle null median 0.113 (95% 0.089–0.137) | Withdraw the session 8 continuity claim: the excess over ordinary entries reflects the name sign stock, not surviving names. |
| NI sign in Early Dynastic and Old Akkadian names: readings, positions, body-part pairing | Share of NI-containing PN types that contain a body-part sign, vs all PN types; NI rate after body-part signs vs after other signs | ED IIIb 0.176 vs 0.175 | All PN types with a body-part sign | The archaic NI–body-part pairing has no Sumerian counterpart; it counts against the possessive reading and does not support the oil reading. |
| Sign position in names: Uruk III vs read Archaic Ur, plus order rigidity with proto-Elamite · 1 figure | Spearman correlation of per-sign final-position share, Uruk III names vs Archaic Ur read names | 0.011 | Permutation null 95% −0.46 to 0.45 | Linguistic order in names appears with Archaic Ur; Uruk III name order is habitual but not Sumerian syntax. |
| Kushim tests: archive concentration and the SANGA slot | Share of a type's tablets in its largest publication volume / provenience | 0.909 | Names median 0.67–0.70 (n = 4), titles 0.50–0.67 (n = 6) | Undecided; the distribution features cannot separate names from titles in this corpus (too few recurrent marked names). |
| Titles inside archaic name entries | Share of name entries that equal or contain a multi-sign title from Lu2 A / Officials (2+ witnesses) | 11 of 540 (2.0%) | Keep the name statistics as computed; no rerun needed. |
Answers proposed 4
Answers taken up again 1
Evidence found 9
EVD
Cite it
Mysteries Department. (2026). Session 9: Brute-force checks of the candidate readings, sign by sign against the literature [Research session SES0DG38A190RRC]. https://mysteriesdept.com/mysteries/proto-cuneiform/sessions/SES0DG38A190RRC/session-9-brute-force-checks-of-the-candidate-readings-sign
@misc{SES0DG38A190RRC,
title = {Session 9: Brute-force checks of the candidate readings, sign by sign against the literature},
author = {Mysteries Department},
year = {2026},
url = {https://mysteriesdept.com/mysteries/proto-cuneiform/sessions/SES0DG38A190RRC/session-9-brute-force-checks-of-the-candidate-readings-sign}
}