README

Lilly core admission packet

Date: 2026-10-22. Status: research packet complete for review; new executable doctrine not admitted.

Author/edition: William Lilly, Christian Astrology, London, 1647, selected general doctrine in Books I-II. Western horary only. Code inspection baseline: d6d43a1c9f52e5349eff4f8286dacb1bc1446332. The inspected horary files are unchanged from the preceding horary-fix commit 8c427532591c7901c8ec2d8424869b617bcb4634; subsequent generated-maintenance work on main is outside this packet.

This packet establishes the source inventory, candidate rules, unresolved interpretations, and worked-case denominator for the next Lilly tranche. It does not redefine the existing evidence-only v1 contract or claim that a complete Lilly judgement engine has been implemented.

Read the packet

File	Purpose
Source inventory	Edition, witnesses, rights, page-level inspection, scope and acquisition receipts
Rules and ambiguities	Source propositions, operational questions, current owners and admission gates
Worked-case catalogue	Original figures, reused questions, raw historical inputs and fixture promotion requirements
Structured index	Stable IDs, source locators, acquisition hashes, rules, ambiguities and case records
Integrity checker	Read-only reference, count, link and optional local witness-hash checks

The JSON index is research data, not an engine input or normative policy file. The Markdown explains qualifications that must accompany its compact records.

Decision at this checkpoint

The original 1647 photographed pages govern. OCR and modern editorial material are navigation aids. A second scan of the same edition can resolve a damaged glyph; it cannot provide independent doctrinal corroboration.

The principal findings are:

• Exact astronomical contact and a successful source-conditioned judgement need distinct representations. Lilly qualifies conjunctions, easy aspects, squares and oppositions differently.
• Reception has direction, basis, scope and epoch. Lilly's general discussion also includes a route without an aspect; the current six-form classifier does not make that route available.
• Moon/VOC qualifications, contextual cautions, same-sign combustion and turning instructions need explicit treatment before a broader profile can be admitted.
• Source language leaves genuine choices: alternative orb tables, the meaning of application within a sign, relative planetary weight, and reception at different moments. These remain visible in the ambiguity register.
• The catalogue inventories 36 Book II case groups using 35 distinct printed figures. The p.456 office question reuses p.455; p.473 is an event/inception example. Book I contributes one further printed figure and two additional erection exercises without new figures. These are different denominators.

No new rule or case in this packet is admitted as an executable or predictive golden. That is an admission result, not a missing status field. Visual heading and diagram checks do not constitute double transcription of every coordinate, complete judgement analysis, or independent confirmation of reported outcomes.

Finite core and deliberate deferrals

The core covers chart/time provenance, house and role foundations, the named dignity table, application/reception and solar condition, considerations, Moon/VOC, completion/blocking mechanisms, and qualitative timing prerequisites. It includes the general house significations as a source map, not twelve implemented topic modules.

The complete planetary/sign description chapters, all topic-specific clauses, antiscia, lots, fixed-star conditions, symbolic timing tables and every unillustrated narrative example require later source tranches. The fortitude table is inventoried as a dependency; no universal judgement score is admitted. Book III is auxiliary for house-policy context, not a way to import natal doctrine into questions. Other Western authors need separate packets.

Next admission gates

1. Resolve the blocking decisions in A01-A25 for a named new profile. Preserve unresolved alternatives explicitly when the source cannot choose one.
2. Double-transcribe the selected dignity table and historical fixture data; collate uncertain glyphs against a second photographed witness. Record question wording, entities, roles, judgements and outcome reports separately.
3. Specify rule inputs, epochs, conditions, exceptions and precedence. Separate astronomical event evidence, author-qualified propositions and conclusions.
4. Create independently specified positive, negative and boundary fixtures. The catalogue defines the promotion contract. Do not derive expected doctrine from current output.
5. Implement a separately admitted surface or policy after the above decisions; preserve v1 compatibility. Verify every candidate route, interrupted applications with no later exact contact, simultaneous events, and partial search coverage.

Recommended first fixture tranche: the didactic examples for blocking and reception, Book I clock exercises, the existing p.442 assignment case, one transaction case (p.219) and one marriage case (p.385). This is a research sequence, not a claim that those topics already have complete rules.

Research assembly receipt (before publication)

At research assembly, only this new directory was added. Runtime, native code, tests, old dossiers, protected validation baselines, generated wiki and unrelated work were left unchanged. No commit, push, wiki publication or release was performed.

Checks: the structured index was parsed; IDs, references, page locators, case counts, reused-figure links and non-admission fields were checked; source-cache SHA-256 values were verified; local Markdown targets and git diff --check were checked. These are documentation integrity checks, not engine tests or independent source review. Acquisition and inspection details are in the source inventory.

Integrity receipt: 203 unique IDs, 598 references, 18 local Markdown links, 87 cached source/image hashes verified. The packet checker also passed Ruff. A separate read-only review found no concrete corrections in the four Markdown documents; that review was not independent historical transcription.

Repeat the portable integrity check from the repository root:

.\.venv\Scripts\python.exe wiki\05_research\horary\lilly_core_2026\validate_packet.py

Add --cache C:\dev\outputs\lilly-core-admission-20261022 to verify the local acquired bytes as well. The checker never downloads, changes doctrine, or promotes fixtures. Run without Python's -O option so its assertions execute.