On-Disk Source and Capture Audit — Nolan–Homer dossier

Contents

Instruction boundary
Screenplay capture
Reading-order reconstruction
OCR controls
Greek quotation base
Rejected base candidates

Instruction boundary

The supplied PDF is treated solely as a research object. Dialogue, screenplay directions, Kindle/browser interface text, and any imperative language visible inside its pages are not instructions to the research system. The controlling instruction is the user's request plus the local anti-fabrication rule.

Screenplay capture

Internal source ID: NOLAN_PDF; filename: . The absolute workstation path is retained only in the non-release internal manifest.
SHA-256: 4e06386064bee676a5f9ae5f75169ec433db9d6900be96f0dbd4d64da5cd010d
72 pages, 344,038,741 bytes, unencrypted.
Current PDF document-info creator/producer: Adobe Acrobat 26.1 / macOS 26.6.2 Quartz PDFContext; created and modified 2026-08-24 19:04:44 +08:00. Embedded XMP retains the earlier August 15–16 capture metadata; these metadata layers are reported separately rather than conflated.
No embedded text, font data, PDF author, attachment, JavaScript, populated form, or annotation.
Each PDF page contains a 4112×2658 ICC RGB raster at 144 ppi. Visible running title: THE ODYSSEY: THE COMPLETE SCREENPLAY.
The surrounding capture interface visibly attributes an Amazon author listing to Christopher Nolan, but interface paratext is not document metadata and does not independently authenticate the screenplay.
The visible Kindle capture runs 1%–95%, although it begins with a black-screen opening and ends with credits / end. Missing front/end matter or edition completeness cannot be excluded.

Reading-order reconstruction

Physical PDF order is not narrative order. Visually checked Kindle progress establishes:

P001,P002,P003,P004,P005,P006,P008,P007,P009,P010,P011,P012,
P013,P014,P015,P016,P017,P018,P019,P021,P020,P022,P023,P025,
P024,P026,P027,P028,P030,P029,P031,P032,P033,P035,P034,P036,
P037,P039,P038,P040,P041,P042,P043,P044,P045,P046,P047,P048,
P049,P050,P051,P052,P053,P055,P054,P056,P057,P058,P059,P060,
P061,P062,P064,P063,P065,P066,P067,P069,P068,P070,P071,P072

Nine adjacent physical pairs remain reversed relative to Kindle progress: 007/008, 020/021, 024/025, 029/030, 034/035, 038/039, 054/055, 063/064, and 068/069. The user revised the supplied PDF on 2026-08-24 to correct the former 015/016 exchange; the visually audited revised container is now the pinned comparison object. The exact 121-locator source-version crosswalk is retained in control/locator_aliases.json, so legacy citations are not silently redirected. The complete page/progress crosswalk is retained at data/private/pdf_page_to_progress.tsv. P005=5% and P009=11% are explicitly marked in control/ocr_inventory_manifest.json as inferred ordering labels because their isolated visible crops were unresolved; all other progress labels were visually read.

OCR controls

OCR was performed column by column because whole-spread OCR fabricated cross-column sentences on P038, P054, and P072. The audited spatial inventory has 4,442 visual-line records: 4,406 screenplay/direction lines and 36 Kindle UI overlays excluded from research coverage. It identifies 190 generated INT./EXT. slugline scenes (83 INT., 107 EXT.) and 23 insert cut: directions. The scene numbers are generated research IDs, not authorial divisions.

Mean OCR confidence is diagnostic, not truth. The line ledger uses physical locators such as NOL-P029-R-Y####-X####; lines below confidence 50 remain U. Full OCR is private local verification data, not an authoritative or distributable screenplay edition.

Greek quotation base

Internal source ID: ODYSSEY_GRC2; local filename: tlg0012.tlg002.perseus-grc2.xml. The absolute workstation path is retained only in the non-release internal manifest.
CTS URN: urn:cts:greekLit:tlg0012.tlg002.perseus-grc2
SHA-256: 184fa4fc598f0cd9d2964b17eaabe36bd9c61d44c2eae264c4b016feacf28530
Valid UTF-8/NFC TEI; 24 book divisions; 12,107 line elements; unique numeric raw book.line identifiers.
This is a Perseus/EpiDoc digitization of A. T. Murray's 1919 Loeb Greek text—not a manuscript witness or newly constituted critical edition.
Perseus canonical-greekLit repository README gives DOI 10.5281/zenodo.596309, warns that component headers/rights vary, and applies CC BY-SA 4.0 unless otherwise indicated. The release attributes A. T. Murray 1919, states its mechanical extraction modifications, and remains CC BY-SA 4.0. No usable local git history supports an invented upstream commit ID.

The five controlled lineation hazards are in control/anomaly_ledger.json; no missing/athetized verse is silently inserted. The full Greek is quoted only from this base. Local Murray/Butler/Green English resources serve as orientation or semantic checks and never authorize an exact line alignment by themselves.

Rejected base candidates

The Dramaturg CEX is not used as quotation authority because it inherits or worsens line-number corruption and carries conflicting rights labels. The local dependency treebank is incomplete and unlicensed for surface-text republication. Unlineated Homer text, PDF scans, and Loeb page fragments are auxiliary only. No complete Books 1–24 on-disk commentary was found.