C1 · Allah ﷻ — Domain 1: The Names — SOURCES & LICENCES

Points 1–10 of the ledger. Written before any citing, as the block demands.

1 · Canonical source list (point 1)

Every source this domain may cite, named before use:

2 · Acquisition (point 2)

All open texts already stand acquired in the sealed main corpus

(00-CORPUS/MANIFEST.sha256) — the inheritance law: nothing re-acquired.

4 · Licence record (point 4)

Text: Tanzil open licence (v1 point 9 record, 09-MATH pillar I,

audit-passed). Morphology: Quranic Arabic Corpus open licence (v1 point 3

record). Ḥadīth/tafsīr corpora: rights recorded per-work in

07-SITE/register.html and 07-SITE/hadith.html. No source behind a

sign-in; no licence unrecorded.

5 · Edition register (point 5)

quran.json meta carries the text edition; morphology.json meta carries the

QAC version; every translation cited carries its edition string from

translations/_manifest.json. Citations in this domain name the edition,

so a citation means one thing.

6 · Scope declaration (point 6)

This domain registers the Names of Allah ﷻ. The **seed scope** (Block 2)

is mechanical and text-attested: the Names as they occur in the canonical

Names passages — the basmala and Sūrat al-Ḥashr 59:22–24 — counted at

lemma level across the whole text. The **full scope** (Blocks 3–5)

extends to every Name attested in the text and to the 99-Names reports

with their collection+number citations. Nothing enters the registry

without its attestation class stated.

7 · Exclusion log (point 7)

Excluded from this domain, by name:

- Modern devotional compilations without classical sourcing.

- Any Names list that cannot be tied to a text attestation or a cited

report — a beautiful list with invisible sources is a defect (L8).

- Kalām adjudication of what a Name entails — positions are DOCUMENTED

with carriers in the Attributes domain, never adjudicated

(Founding Principle 3).

8 · Two-source cross-check protocol (point 8)

Every occurrence count is produced twice, independently:

(a) lemma-level from morphology.json segments;

(b) surface-level from the **Uthmānī** text of quran.json — the

tokenization the morphology is aligned to.

The builder asserts per-word agreement: every segment position claimed by

(a) must contain the surface form in (b). Divergence fails the build and

is documented, never smoothed. Implemented in build_names_layer.py;

proven by _selftest.py.

**Documented finding (2026-09-01, first run):** the imlāʾī (simple)

word-split drifts from the Uthmānī split (one Uthmānī word such as

يٰٓأَيُّهَا becomes two imlāʾī words), so imlāʾī word indices cannot be

used against QAC positions. The first build was correctly REFUSED by

this gate (1,029 apparent divergences, all index drift, zero real count

errors) — recorded here so no future window rediscovers it.

9 · UNSOURCED policy (point 9)

Every field of every registry row carries its source class:

quran-direct (with āyah refs) · hadith-report (collection+number) ·

translation (edition string) · or the plain word **UNSOURCED**. An

honest gap outranks an invented fact, forever (Founding Principle 2).

3 · Hash & seal (point 3)

The exact source files above are hashed into _C1-SOURCES-SEAL.sha256

beside this file; the selftest re-verifies the seal on every run.

The seed layer as built - every count tool-printed

Occurrences are lemma-level: every use of the word in the text, including uses not addressed to Allah ﷻ (al-Rahim in 9:128 describes the Messenger ﷺ). The referential tagging is Block-3 work.

The seal, live

machine-built and machine-audited - independent human review is invited and never waited for · rendered 2026-09-01 18:50 · AL-WAQF C1 D1