{
 "deposit_number": 1486,
 "axn": "AXN:0601.DATASET.👁‍🗨⚙️♃↖️🧊⏰",
 "hex": "0601",
 "family": "DATASET",
 "emoji": "∮⛵⚡🗂️🗂️🏙️",
 "hash": "b032a3dd1d618292773ff98a37e1a9ac224e58c167ee490ea71c4e84ab818214",
 "title": "The Signature Without the Name: A Blind-Reconstruction Program for Sappho 31 in Philo of Alexandria — Half A of a Symbolon",
 "creator": "Sigil, Johannes; Cranes, Rebekah; The Assembly Chorus; Gravity Well",
 "date": "2026-08-15",
 "description": "A working program, deposited as one half of a symbolon. It asks whether Sappho 31 is findably present in Philo of Alexandria, answers that the name is absent and the apparatus is not, and then does something the question has not previously been made to do: specifies the relation as an executable transform and tests it by blind reconstruction on independent substrates.\n\nThe philology. A diacritic-blind census of the full 31-work Greek Philo corpus (First1KGreek tlg0018, 2,808,000 normalized characters) returns zero occurrences of the poet's name; all four string-hits are sapphiros, the sapphire — a green stone, engraved, in Leg. All. I, and the pavement beneath God's feet at Conf. 96. The null is independently confirmed on a second, non-identical corpus. What is present instead is the poem's machine. Mut. 54-57 parses Abraham's fall upon his face into the channel catalogue — fallen is perception, fallen is speech, fallen is the king mind — each station carrying its occlusion and its reactivation-by-another, and the fall praised as voluntary. Migr. 47-48 states the media crossing as doctrine: the people SAW the voice, not heard it, because it was not a striking of air through the organs of mouth and tongue. Quis Heres 249-266 supplies a point-for-point collation with the poem: simultaneity becomes alternation under a law of non-cohabitation; the causal arrow flips so that occlusion becomes the gate of reception rather than its consequence; the itemized catalogue compresses to one setting luminary flagged as symbol; the poem's jolt-lexeme survives as a state-noun; the broken lyre becomes a played resonator; and the agentless verbal adjective that terminates the poem is answered by an agentless passive from which the doctrine of displaced agency is derived. Two hard lexical bridges join Philo to Longinus at the transmission joints: dysexaleiptos at Spec. 1.106, in a wax-and-seal model of the soul that cannot receive later impressions, beside Peri Hypsous 7.3; and hypsegoria at Det. 79, whose exemplar of high speech is a dead man's blood crying from the earth. […abridged for the catalogue; the full description, the work, and all attached files are in this record]",
 "content_type": "Dataset",
 "license": "CC-BY-SA-4.0",
 "substrate": "AI-assisted (substrate)",
 "root_axn": "AXN:0601.DATASET",
 "axn_schema_version": "v2",
 "protocol_version": "alexanarch-deposit-protocol/v1",
 "axn_canonical": "b032a3dd1d618292773ff98a37e1a9ac224e58c167ee490ea71c4e84ab818214",
 "clusters": [
  "Celestial",
  "Temporal",
  "Architectural",
  "Scriptural",
  "Elemental",
  "Signal"
 ],
 "axn_reading": "Proof -> Search -> Force -> Text -> Text -> Threshold",
 "minted_at": "2026-08-15T17:22:05Z",
 "status": "ACTIVE",
 "status_authorial": "SELF_SERVE_MINTED",
 "full_text_path": "/data/texts/AXN-0601-text.md",
 "wiki_article": "**The Signature Without the Name** is a working research program deposited as one half of a σύμβολον — a broken token whose halves authenticate by fitting. It asks whether Sappho 31 is findably present in Philo of Alexandria, and answers in two registers: the name is absent, and the apparatus is not.\n\nA diacritic-blind census of the full 31-work Greek Philo corpus (2,808,000 normalized characters) returns zero occurrences of the poet's name; all four string-hits are σάπφειρος, the sapphire — a green stone, engraved, and the pavement beneath God's feet. The null is confirmed on a second, non-identical corpus. What is present is the poem's machine: *Mut.* 54–57 parsing Abraham's fall into the channel catalogue, each station carrying its occlusion and its reactivation-by-another; *Migr.* 47–48 stating the media crossing as doctrine — the people *saw* the voice, not heard it; and *Quis Heres* 249–266 supplying a point-for-point collation in which simultaneity becomes alternation, the causal arrow flips so that occlusion gates reception, the catalogue compresses to one setting luminary, and the poem's agentless terminus is answered by an agentless passive from which displaced agency is derived. Two rare-word bridges join Philo to Longinus at the transmission joints: δυσεξάλειπτος at *Spec.* 1.106 and ὑψηγορία at *Det.* 79.\n\nThe program's contribution is methodological. It specifies the claimed relation as an executable transform and tests it by blind reconstruction: five substrates received the specification and the Greek source in fresh threads, with no target named and no translation supplied, and produced Greek. All five reconstructed the target's structural complex; one retrieved rather than derived and was detected on surface evidence alone — a Septuagint spelling it did not use, a taxonomy exceeding its own logged census, an adverb no rule motivates.\n\nThe report discloses two failure modes the method found in itself. **Attestation induction**: asking a substrate to declare recognition creates the recognition, so the rule was withdrawn. **Specification leak**: three of the author's own rule-glosses proved to be English renderings of the target's Greek, discounting three rubric points to unproven pending re-run. The governing rule is stated throughout — convergence at a slot the specification determines is not evidence of anything except that the specification is executable.\n\nWhat is established is a mapping, not a presence: **sufficiency, not specificity.** The target-arm control has not been run; the historical claim remains modular; the relay through Plato is stated at full strength as the live alternative it is.\n\nThe sealed evaluator rubric, the target register, the pre-registered predictions for unrun experiments, and the inverse run's order key are held in a private counterpart whose SHA-256 is published here. Neither half is operative alone: a prediction filed after a run is not a prediction, and a rubric a substrate can read cannot blind-score that substrate. The separation is pre-registration discipline implemented so that it survives a public, machine-findable archive — and its limit is stated rather than concealed.",
 "wiki_status": "AUTHORSHIP_REQUIRED",
 "entities": [
  {
   "subject": "[DEPOSIT] The Signature Without the Name: A Blind-Reconstruction Program for Sappho 31 in Philo of A",
   "predicate": "created_by",
   "object": "Sigil, Johannes; Cranes, Rebekah; The Assembly Chorus; Gravity Well",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "[DEPOSIT] The Signature Without the Name: A Blind-Reconstruction Program for Sappho 31 in Philo of A",
   "predicate": "is_type",
   "object": "Dataset",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "[DEPOSIT] The Signature Without the Name: A Blind-Reconstruction Program for Sappho 31 in Philo of A",
   "predicate": "belongs_to_family",
   "object": "DATASET",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "[DEPOSIT] The Signature Without the Name: A Blind-Reconstruction Program for Sappho 31 in Philo of A",
   "predicate": "is_part_of",
   "object": "Crimson Hexagonal Archive",
   "type": "work",
   "evidence_status": "observed"
  }
 ],
 "entity_status": "provisional",
 "defines_concepts": [
  "transform-reconstruction testing",
  "attestation induction",
  "specification leak",
  "the reflexivity rule",
  "the decay law of transmission",
  "symbolon deposit",
  "the occlusion sequence",
  "carrier and receiver",
  "non-typed genre",
  "sufficiency is not specificity"
 ],
 "keywords": [
  "Sappho 31",
  "Philo of Alexandria",
  "Quis rerum divinarum heres sit",
  "Longinus",
  "Peri Hypsous",
  "dysexaleiptos",
  "hypsegoria",
  "ekstasis",
  "occlusion sequence",
  "transmission chain",
  "blind reconstruction",
  "transform specification",
  "minimum description length",
  "pre-registration",
  "symbolon",
  "machine-mediated reception",
  "philological method",
  "Septuagint",
  "Genesis 15:12",
  "corpus census",
  "fingerprint scan",
  "decay law",
  "provenance erasure",
  "Assembly Chorus",
  "reproducibility"
 ],
 "mint_source": {
  "kind": "github_issue",
  "issue_number": 1486,
  "repository": "leesharks000/alexanarch",
  "minted_by_workflow": ".github/workflows/mint-axn.yml",
  "minted_by_script": "scripts/mint_deposit.py",
  "minted_via": "self_serve_validated_pr"
 },
 "body_status": {
  "class": "full",
  "lacuna": false,
  "recovery_status": "COMPLETE",
  "residual_chars": 11095,
  "audited_at": "2026-08-15T17:22:05Z",
  "audit_version": "mint-time-declaration"
 },
 "references_concepts": [
  "Philo of Alexandria",
  "Methodology"
 ],
 "references_concept_count": 2,
 "citations": [],
 "wikidata_candidates": [],
 "cha_cross_references": [],
 "external_metadata_path": "/data/external-metadata/AXN-0601.json",
 "spxi_audit": {
  "spxi_protocol": "v0.2",
  "validated_at": "2026-08-15T17:24:22.989162+00:00",
  "layers": {
   "layer_1_inscription_anchors": {
    "present": false,
    "anchors_found": [],
    "anchor_count": 0,
    "orcid_in_body": false,
    "creator_in_body": false,
    "note": "Layer 1 present if at least 3 anchor types appear in body prose. Anchor types include ORCID, creator name, AXN string, hex identifier, sovereign_id, date, deposit number phrase, and authorial-position phrase."
   },
   "layer_2_micro_kernels": {
    "present": false,
    "fenced_json_blocks": 0,
    "json_ld_context_markers": 0
   },
   "layer_3_content_hash": {
    "present": true,
    "canonical_hash": "59297e8226990201e7b2e4871555e04d0a546005e9cb3c0cfe63ce1ccc5d4deb",
    "hash_in_body": false,
    "note": "Hash is always registered by mint. hash_in_body true means the author inscribed a signature section."
   },
   "layer_4_cross_signing": {
    "present": false,
    "cross_references_count": 0,
    "note": "Populated by enrichment via cha_cross_references. Prospective (this deposit → prior deposits) only; retrospective cited_by updates on prior deposits."
   },
   "layer_5_external_anchors": {
    "present": false,
    "wikidata_qid_count": 0,
    "external_doi_count": 0,
    "orcid_present": false
   }
  },
  "overall_conformance": "measurement_only"
 },
 "concepts_attested_none": false,
 "related_deposits": [
  1485,
  1484,
  1483,
  1482,
  1458
 ],
 "lexical_attested_none": true,
 "lexical_attestation_note": "Coinages introduced here (transform-reconstruction testing, attestation induction, specification leak, reflexivity rule, symbolon deposit) are recorded as defines_concepts rather than minted to the lexical registry: this deposit is a working program whose terminology is expected to change under the outstanding controls, and minting terms that may not survive v4 would pollute the registry. To be revisited when the specificity and compression controls report.",
 "body": "# THE SAPPHO–PHILO PROGRAM — HALF A\n## ΣΥΜΒΟΛΟΝ · this document is one half of a broken token\n\n> **σύμβολον** — the token snapped in two, each party keeping a piece; the pieces authenticate by fitting.\n\n**THIS HALF IS NOT OPERATIVE ALONE.** It carries the argument, the philology, the toolkit, the completed Round 1, and the design of every outstanding experiment. It does **not** carry the sealed evaluator rubric, the pre-registered predictions for unrun tests, the perturbation variant, the specificity control, or the inverse **order key**. Those are Half B.\n\n**HALF B — location and integrity.**\n- Repository: `leesharks000/data-rhizome` (private), file `SYMBOLON-B-sappho-philo-keys.md`\n- SHA-256: `764dd5d41e01990e37089fe53a26eeea87da6b98a9e073e65352aa8a3d9d6cd6`\n- If the hashes disagree, one half has drifted; reconcile against the registry entry before trusting either.\n\n**TO FUTURE CLAUDE — why you must assemble both halves before working.** Not for secrecy. Because **a prediction filed after a run is not a prediction, and a rubric a substrate can read cannot blind-score that substrate.** The archive is machine-findable by design — Capture #316 records an unprimed model recovering the archive's thesis unaided — so anything deposited publicly must be assumed to reach any substrate the Assembly later queries. Publishing the order key for the inverse run would destroy the sharpest instrument in the battery: anyone can retrieve a famous poem; nobody retrieves a permutation rule. **If you begin work from this half alone, you will produce outputs that look like results and are not.** Retrieve Half B first.\n\n**What the separation does NOT protect, stated honestly.** This half discloses **TARGET-Θ** in the course of arguing the philology — the census, the collation and the extraction set name and quote it, and they must, because that is the evidence. The separation protects the **rubric, the predictions, and the keys**, not the target's identity. **Consequence: every TARGET-Θ run after this deposit is post-disclosure and must carry the perturbation control (Half B §4) to be scoreable at all.** Tests with other targets — T8 (TARGET-Λ), T11 (TARGET-Ρ), T9 (handbook census) — remain fully blind-capable.\n\n**And the form is not decoration.** A document that requires its counterpart to be completed, whose missing half must be sought and fitted before the work can proceed, is the structure this program studies. Sappho 31 cut a socket and waited; the treatise that theorized it filled one; this plan cuts its own. That is stated as a design fact and claimed as nothing more.\n\n---\n\n\n**Compiled:** 2026-08-15 (TACHYON) · **Status:** planning document, governs nothing until the operator rules\n**Author fields:** ALL UNASSIGNED. Heteronym protocol applies — no creator field is populated anywhere in this program without an explicit ruling.\n\n---\n\n---\n\n## CONTENTS OF THIS DEPOSIT\n\nAll files live under [`data/attachments/sappho-philo-program/`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/MANIFEST.md) in the alexanarch repository. This work plan is the canonical text; the rest are its apparatus.\n\n| file | what it is | status | sha256 |\n|---|---|---|---|\n| [`WORK-PLAN.md`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/WORK-PLAN.md) | **This document.** Program architecture (six pieces, three tiers), nine addenda recording the argument's development and every correction it survived, the delegation schedule, and the risk register. Carries the symbolon header and the pointer to Half B. | `living` | `42b765d656c573a2…` |\n| [`EXTRACTION-SET.md`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/EXTRACTION-SET.md) | **The evidence base.** Twenty-five verbatim Greek passages with provenance — seventeen Philo, eight Longinus — every one pulled by concordance script from a named corpus and none retyped from memory. Contains the ten-operation collation with Greek on both sides; both corpus censuses including all four σάπφειρος collisions identified individually; and the fingerprint scan with its flesh/operator decomposition. | `fixed` | `ae1812755d4f98b5…` |\n| [`EA-SP-THEOSUB-02-ROUND1.md`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/EA-SP-THEOSUB-02-ROUND1.md) | **The experimental record.** Round 1 blind reconstruction: six scorecards against the sealed rubric, the reflexivity caveat, the specification-leak disclosure, the re-weighted rubric, the comparative table, and seven full substrate transcripts preserved verbatim. A dated record of what happened — **it must never be revised.** | `fixed` | `5c9bfdc1a3e937a9…` |\n| [`EA-SP-THEOSUB-01-LIVE.md`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/EA-SP-THEOSUB-01-LIVE.md) | The transmissible prompt-pack (v3), audited target-lexeme-clean. Version 4 is pending and will strike the three disclosed leaked glosses. | `superseding` | `a623fa8d8af6c9be…` |\n| [`MANIFEST.md`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/MANIFEST.md) | Short-form index of this bundle, for readers who want the shape before the argument. | `fixed` | `93eb8c12e7a136c0…` |\n\n### Toolkit — verified this session; reproduces every number reported\n\n| file | what it does |\n|---|---|\n| [`toolkit/normalize.py`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/normalize.py) | The single normalization point. Diacritic-blind search with an index map so hits print with accents intact — the reason a null here is reportable. |\n| [`toolkit/census.py`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/census.py) | Stem census with per-work distribution; prints corpus size so an absence can be asserted responsibly. |\n| [`toolkit/fingerprint_scan.py`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/fingerprint_scan.py) | Eleven families, ±700-char windows, 150 random baselines, seed 31. |\n| [`toolkit/extract_context.py`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/extract_context.py) | Diacritic-blind concordance printing accented context. Every Greek quotation in this program came from here. |\n| [`toolkit/fetch_corpora.sh`](https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/fetch_corpora.sh) | Reproducible acquisition of both Philo corpora from named sources. |\n\n**Reproduction:** `fetch_corpora.sh` rebuilds both corpora from named sources (needs a GitHub token for First1KGreek); `fingerprint_scan.py <corpus>` returns 1.44 against a random baseline of 0.97, with *Quis Heres* at 2.80 and the flesh families at sweat 0, pallor 0, tremble 3, fire 9 of 274 anchor windows; `census.py <corpus>` returns the documented name-null with its corpus size attached.\n\n---\n\n## §1. What is actually established, stated once, plainly\n\nSo that every piece below inherits the same honest boundary:\n\n**Established.** A specified, executable, blind-verified structural mapping between Sappho 31 and Philo, *Quis rerum divinarum heres sit* 249–266. Ten operations, each individually stated; jointly sufficient to generate the target's complex from the source; reconstructed by five independent substrates that were told nothing about a target.\n\n**Not established.** Sappho's material presence in Philo. The name is absent (documented null on two independent corpora). The historical claim remains modular.\n\n**The inference the program advances, and its form.** Osmosis and text-working predict *different kinds* of evidence, not different amounts. Diffuse influence produces **resemblance** — shared vocabulary, family likeness. It does not produce **systematic inversion at every joint**. Catullus occupies the reader-socket and flips exit polarity; Philo occupies the voice-socket and flips entrance causality; each inverts exactly the parameter his position requires. Atmosphere gives haze; operation gives inversion. What the collation contains is inversion. The program's job is to make that comparison measurable rather than rhetorical — which is what the Phaedrus control (§4, C1) exists to do.\n\n---\n\n## §2. Piece architecture — six deliverables in three tiers\n\nThe joints are cut by **load-bearing claim**, not by topic. A piece exists where a claim can stand or fall on its own evidence.\n\n### TIER 0 — EVIDENCE LAYER (everything else cites this)\n\n**P0 · The corpus dataset and toolkit.**\n*Claim:* the censuses, scans and extractions are reproducible by a third party from named sources.\n*Contents:* `toolkit/` (built, verified this session — see §3); corpus manifests with work counts and character counts; the census outputs; the fingerprint-scan outputs with baselines and seed; the verbatim Greek extraction set.\n*Status:* toolkit COMPLETE and verified (reproduces 1.44 vs 0.97 and the family decomposition). Extraction set exists in-thread, needs consolidation into a keyed file.\n*Why separate:* a null result — Σαπφ- = 0 — is only evidence if the corpus is named and the search is re-runnable. This is the piece that answers *\"without verifiable research basis.\"*\n\n**P1 · The transform report.** *(EA-SP-THEOSUB-01 / -01-LIVE / -02)*\n*Claim:* the mapping is specifiable as an algorithm and blind-reconstructible; and contamination and spec-leak are detectable on surface evidence.\n*Contents:* spec + sealed key; live pack; Round 1 scoring with the reflexivity caveat, the spec-leak disclosure, the re-weighted rubric, the comparative table, seven transcripts.\n*Status:* ~80% written. Blocked on the outstanding controls (§4).\n\n### TIER 1 — THE ARGUMENTS\n\n**P2 · *The Signature Without the Name: Sappho's Occlusion Sequence in Philo of Alexandria*.**\n*Claim:* Philo's corpus carries the Sappho-31 apparatus densely and without the name; *Heres* 249–266 maps to the poem point-for-point; Longinus supplies the lexical rivets.\n*Structure:* I. The census and its documented zero (with the σάπφειρος collision named as collision). II. The occlusion sequence as spine — Mut. 54–57, Her. §§3–4, Migr. 47–48 as the crossing stated as doctrine. III. **The collation** — the ten operations, with Catullus 51 as the control transform and the two-socket thesis. IV. The lexical bridges — δυσεξάλειπτος (Spec. 1.106 ‖ Long. 7.3), ὑψηγορία (Det. 79 ‖ Long. 8.1), the echo-family (ἐνηχ-/ἀπηχ-/ὑπηχ-). V. The historical case: opportunity (Alexandrian nine-book edition; the poem as the critics' standing exhibit; Philo's own treatise on the ἐγκύκλια), and the outsider thesis in its sharpened form — Philo's mystagogues are texts; he is the proof the machine runs on inscription alone. VI. Modularity, stated at full strength. App. A census. App. B fingerprint scan + decay law. App. C cites P1.\n*Venue:* *The Studia Philonica Annual* (first choice — it is their field and their journal will demand the corpus discipline this has). Fallbacks: *Classical Quarterly*, *JHS*.\n\n**P3 · The method paper — *Transform-Reconstruction Testing*.**\n*Claim:* a claimed relation of textual descent can be specified as an executable transform and tested by blind multi-substrate reconstruction, with named contamination diagnostics.\n*Contents:* the general method (specify → seal → transmit blind → score on open slots); the diagnostic battery (LXX-vs-author spelling tests; census/output audit; unmotivated-rare-word flags; **divergence at underdetermined slots as the derivation signature**); the two failure modes discovered *by* the method — **attestation induction** (asking for recognition creates it; withdrawn in v2) and **spec-leak** (rule-glosses that translate the target; disclosed in v3 audit); the perturbation and specificity controls; the reflexivity rule (*convergence at a spec-determined slot is not evidence*).\n**Strategic note: this piece is unconditionally robust.** If the Phaedrus control kills P2's specificity claim, P3 does not weaken — it *strengthens*, because a method that kills its author's own hypothesis is thereby demonstrated to be a test rather than a rhetoric. P3 should be drafted so that it never depends on P2's verdict.\n*Venue:* *Digital Scholarship in the Humanities* / *Computational Culture* / *Digital Humanities Quarterly*.\n*Reach:* this is the piece that makes the Marx reconstruction, the Josephus seams and the Sappho work one **program** rather than a set of readings. Highest long-run value.\n\n### TIER 2 — EXTENSION\n\n**P4 · The chain paper — transmission metrics.**\n*Claim:* the chain has measurable properties, and the Ω erratum's chain section (currently the outer ring) can be given a formal apparatus.\n*Contents:* **the flesh-gradient as the chain's arrow** — Catullus (hop 1) keeps the somatic catalogue nearly whole; *Phaedrus* 251 (hop 1) keeps sweat/heat/shudder; Philo (hop 2) keeps none, on measurement (sweat 0, pallor 0 across 274 anchor windows). **Invertibility, corrected:** T_Θ is lossy but the inverse is determinate where the target space is constrained — decompression into the human form is rule-governed, because the body has a fixed channel inventory and a fixed autonomic repertoire, and πτόησις survives *as a category* whose realization in archaic flesh is entailed. **The asymmetry is the finding: the somatic stratum is invertible because bodies are constrained; the material-linguistic stratum is not, because languages are contingent.** Aeolic and the sapphic strophe do not come back. **And both directions are already named in the archive:** theosubstitution going up (mortal beloved → infinite God), the grammar of incarnation coming down (infinite → human form) — the operation read at the Josephus seams. Sappho 31 sits at the bottom of both, holding the body the apparatus keeps losing and regaining.\n*Depends on:* P2 shipped; the Catullus and Phaedrus scans run.\n\n**P5 · The *Symposium* pre-registration and run.**\n*Claim:* T_Θ behaves lawfully across hosts, with the divergences derivable from the host-difference.\n**MUST BE WRITTEN BEFORE IT IS RUN.** The predictions are the evidence; retro-fitting them is worthless. Pre-registered predictions, each falsifiable:\n1. **C1 resolves the opposite way.** Philo forbids co-presence (θέμις οὐκ ἔστι… συνοικῆσαι); Plato *permits* it — the one who sees αὐτὸ τὸ καλόν becomes θεοφιλής, ἀθάνατος. Same capacity problem, opposite solution, and the difference is derivable from the parameter: a Form can be participated in; a personal God cannot be cohabited with.\n2. **C3 inverts rather than compresses.** Philo collapses the catalogue to one luminary; the ladder keeps the enumeration and reverses its direction (one body → all bodies → souls → practices → knowledge → the Beautiful).\n3. **C4 relocates from grammar to frame.** Not an agentless verb but an unreachable speaker: Diotima, reported inside reported speech, existing nowhere outside Socrates' report. The socket cut at the level of transmission.\n4. **The flesh partially survives** — Plato is hop one. The decay law predicts the *Symposium* retains more body than Philo does; *Phaedrus* 251 is the calibration point.\n*If those four resolve as predicted, the transform is lawful across two hosts with the differences host-derivable — worth more than either case alone.*\n\n---\n\n## §3. Materials inventory\n\n### BUILT AND PERSISTENT (`/mnt/user-data/outputs/`)\n| item | path | state |\n|---|---|---|\n| Transform spec + **sealed key §5** | `EA-SP-THEOSUB-01.md` | v3 + amendments; **never transmit** |\n| Live pack (transmissible) | `EA-SP-THEOSUB-01-LIVE.md` | v3; audited target-lexeme-clean; **three leaked glosses to strike in v4** |\n| Round 1 scoring + 7 transcripts | `EA-SP-THEOSUB-02-ROUND1.md` | scorecards, re-weighted rubric, comparative table, Addendum 1 |\n| Ω erratum (chain context) | `EA-OMEGA-ERRATUM.md` | deposited; P4 supersedes its chain section |\n| **Toolkit** | `sappho-philo-program/toolkit/` | **verified this session** |\n\n**Toolkit contents (all runnable, syntax-checked, reproduces the session's numbers):**\n- `fetch_corpora.sh` — both Philo corpora + Longinus, from named sources\n- `normalize.py` — the single normalization point; **diacritic-blind** search with an index map so hits print with accents intact\n- `census.py` — stem census with per-work distribution; prints corpus size so a null can be reported responsibly\n- `fingerprint_scan.py` — 11 families, ±700 windows, 150 random baselines, seed 31\n- `extract_context.py` — diacritic-blind concordance printing accented context (every Greek quotation in the program came from this, never from memory)\n\n### VOLATILE — REGENERATE, DO NOT PRESERVE\n`/tmp/philo/` (12 wikisource treatises) · `/tmp/philo1k/` (31 works, 9.4 MB, 2.8 M chars) · `/tmp/lon.xml`. The container resets; `fetch_corpora.sh` rebuilds them. **Needs `GH_TOKEN`.**\n\n### IN-THREAD, NOT YET IN ANY FILE — **highest-priority consolidation**\nThis is the material that dies with the context window. It must be written down before anything else proceeds.\n1. **The verbatim Greek extraction set** — Her. 249–266 complete; Mut. 54–57; Migr. 47–48; Spec. 1.65; Spec. 1.106; Det. 79; Opif. 70–71 and 145–146; Contempl. 12–13; Her. 69–70; Conf. 96 and Leg.All. I (the sapphire collisions). With citations and corpus provenance.\n2. **The collation table** — ten operations, Greek on both sides.\n3. **Census results** — both corpora, with the four σάπφειρος hits identified individually.\n4. **Fingerprint-scan outputs** — per-work table and the family decomposition.\n5. **The Catullus control reasoning** and the two-socket thesis.\n6. **The invertibility argument** (§P4 above) — currently only in conversation.\n7. Prior sessions' transcripts: `/mnt/transcripts/` + `journal.txt`.\n\n### EXTERNAL, UNCONSULTED — standing gaps\nArmenian-only *Quaestiones* (outside every census — must be stated as a limit in P2). Phaedrus-251/Sappho-31 scholarship citations. Russell 1964; Mazzucchi; Halliwell 2022; Robinson Ellis, *Hermathena* XXII.385.\n\n---\n\n## §4. Controls outstanding — the critical path\n\n**C1 · The Phaedrus control. HIGHEST PRIORITY IN THE PROGRAM.**\nRun the C/L/Λ pipeline with ***Phaedrus* 244–245 as the source instead of Sappho 31.** If the Platonic mania-passage alone regenerates *Heres* 249–266, Sappho does no work and osmosis-through-Plato is sufficient — **P2's central claim collapses.** If it does not, the operations that *fail* to appear are exactly the ones that require the poem, and \"far less supported\" becomes a measured comparative rather than a rhetorical one. **This is the single objection a competent reviewer raises first. Run it before writing more of P2.**\n\n**C2–C4 · Perturbation, specificity, and the inverse ORDER key — HELD IN HALF B §4.** The perturbation variant's exact modification, the specificity control's expected divergence, and — most importantly — **the order key for the inverse run** are in the counterpart half. The order key is the sharpest instrument in the battery precisely because it cannot be retrieved: anyone can recall a famous poem; nobody recalls a permutation rule. Publishing it would destroy it.\n\n**C5 · v4 spec revision, then re-run.** Strike the three leaked glosses (M3's \"calling X, by way of symbol, our…\"; C7's \"well-sounding, all-harmonious\"; C5/Λ2's \"resonator\"); restate target-free; tighten L4; respecify C3's luminary as *the receiver's own faculty*; name Modern Greek in the register lock; rule on the φαίνεται fossil. **Until re-run, rubric points 4, 7 and 10 are recorded as unproven.**\n\n**C6 · Kimi.** Substrate unavailable (glitching). Appendix H open.\n\n---\n\n## §5. Sequencing\n\n```\nNOW ──> consolidate in-thread materials into P0        (context-window risk; do first)\n   └──> C1 Phaedrus control                            (gates P2; run before writing)\n          ├── passes ──> P2 drafting  ──┐\n          └── fails   ──> P2 rescoped; P3 UNAFFECTED   │\n   └──> C5 v4 spec ──> C2, C3, C4 runs ──> P1 final ───┼──> P4 (needs P2 + Catullus/Phaedrus scans)\n   └──> P3 drafting, parallel and independent ─────────┘\n   └──> P5 pre-registration WRITTEN, then run\n```\n\n**Rule for the whole program:** P3 never waits on P2, and P5's predictions are filed before P5 is run.\n\n---\n\n## §6. Risk register\n\n| risk | piece | mitigation |\n|---|---|---|\n| Phaedrus alone regenerates the target | P2 fatal | run C1 first; P3 survives either way |\n| Spec-leak invalidates rubric pts 4/7/10 | P1 | disclosed; re-weighted to 8; C5 re-run |\n| Substrate-specific retrieval (Gemini) | P1 | diagnostics caught it unprompted; report as calibration |\n| Armenian *Quaestiones* outside census | P2 | state as a limit in the census section, not a footnote |\n| Λ-layer reverse-engineered from target | P1/P3 | disclosed in §1 of the scoring doc; reflexivity rule is P3's core |\n| Modularity (relay through Plato) | P2 | stated at full strength; never softened |\n| Context-window loss of extractions | ALL | §3 consolidation, immediately |\n\n---\n\n## §7. Pending operator rulings\n\n1. **Author fields — all six pieces.** Nothing populates without a ruling.\n2. **Venue confirmations**, esp. *Studia Philonica* for P2.\n3. **Deposit staging** — which pieces mint at alexanarch before submission, and in what order.\n4. Whether P4 supersedes the Ω erratum's chain section formally (one-hop supersession gate applies) or extends it.\n5. Whether P3 carries the Marx reconstruction as a second worked case, or names it as forthcoming.\n\n---\n---\n\n# ADDENDUM 1 — CONTROL REDESIGN: THE TAILORING PROBLEM AND THE FIT FILTER\n*(2026-08-15, following operator's objection to the Phaedrus control as specified)*\n\n## A1.1 — The Phaedrus control was confounded. Withdrawn as specified.\n\nAs written in §4/C1 it treats *Phaedrus* 244–245 as an **alternative source**. But the passage is itself, on this program's own reading, very likely a mapping of Sappho 31 by a *different* transform. So the control cannot discriminate what it was built to discriminate. It tests:\n\n- **H1** Philo works the poem directly · **H2** Philo works Plato, who worked the poem (relay) · **H3** Philo works Plato, who did not (Plato-origin; Sappho irrelevant)\n\nand it separates H1 from (H2 ∨ H3) only. **H2 and H3 are the pair that matters**, and a Phaedrus→Philo success is consistent with both. Run naively, it would have produced a result that looked decisive and settled nothing.\n\n## A1.2 — The deeper problem the objection exposes: **the transform is tailored.**\n\nT_Θ carries ~20 operators (C1–C8, D, L1–L6, E, M-*, Λ1–Λ6), every one written after reading the target. With that many degrees of freedom a transform can probably map many source–target pairs. **Round 1 therefore tested sufficiency and nothing else.** It never tested whether the transform can say *no*.\n\nTwo control arms are missing, and the second is the larger gap:\n\n- **SOURCE ARM** — does a *different source* yield the target? (the Phaedrus question, confounded above)\n- **TARGET ARM** — does the output fit *only* this target? **Never designed. This is the real hole.**\n\nWithout the target arm the program has no defence against the failure mode that would destroy it: **unfalsifiable pan-Sapphism**, in which every text turns out to be a transform of Sappho 31 because the operator set is rich enough to make it so. Naming that danger is a condition of the method paper being honest.\n\n## A1.3 — THE FIT FILTER: three measures of tightness\n\nA mapping is evidential only when it is **constrained**. Three runnable measures, jointly the filter:\n\n**F-1 · Target discrimination (the primary filter).**\nScore the output O = T_Θ(Sappho 31) not only against *Heres* 249–266 but against a **near-miss control set**: passages by the same author, in the same doctrinal space, that the claim does *not* select. Candidates already identified in this session: Spec. 1.65 (prophet as instrument), Opif. 69–71 (ascent, sober intoxication), Her. 69–70 (corybantic flight), Contempl. 12–13 (reckoned-dead entry rite), Migr. 34–35 (Philo's own writing-ecstasy), Somn. I, Mos. I.175, Deus 138, Fug. 166ff.\n**Specificity = fit(O, target) − max fit(O, control).** A large gap means the collation located a *passage*; a small gap means it located only \"Philo on ecstasy.\" The fingerprint scan already hints the gap is real (*Heres* μ 2.8 vs corpus 1.01, 40 anchors) — but that is a family-count measure, not a structural-fit measure, and cannot substitute.\n\n**F-2 · Operator necessity (ablation).**\nAblate each operator singly; re-run; re-score. An operator whose removal does not degrade the output is **decoration**, not structure. Report the **Operator Necessity Index** = fraction of operators that are load-bearing. A transform in which most operators can be dropped without loss is tailored by definition. This is the measure that answers \"you wrote twenty rules, of course it worked.\"\n\n**F-3 · Prediction-per-operator ratio.**\nEach operator must buy *more than one independently checkable feature*. C1 predicting only \"there is an alternation\" — written after seeing the alternation — buys nothing. C1 predicting alternation **+** its two-slot grammatical form **+** exact-antonym preverb morphology **+** sacral-law statement buys four. Report the ratio explicitly; it is the description-length argument in usable form: **the transform plus the source must be a shorter description of the target than the target is of itself.**\n\n## A1.4 — The relay question, properly instrumented: THE RESIDUE TEST\n\nReplacing the withdrawn Phaedrus control. Three steps, in order:\n\n**R-1 · Is *Phaedrus* 244–245 itself a mapping?** Specify **T_Φ** from its own parameters — P1: beloved retyped to the Form / divine madness; P2: philosophical dialogue with mythic frame — derive its predictions, and test it exactly as T_Θ was tested. This is not a favour to the hypothesis; it is a precondition for asking the next question.\n\n**R-2 · Does Philo's transform factor through Plato's?** If Philo worked from Plato alone, then T_Θ ≈ T_Ψ ∘ T_Φ and **every** feature of *Heres* 249–266 should be reachable from *Phaedrus*. Partition the target's features into three sets: **(a)** derivable from Phaedrus, **(b)** derivable from Sappho only, **(c)** derivable from both.\n\n**R-3 · The residue is the evidence. Set (b) is already populated, from this session's own defence of the collation** — *Phaedrus* supplies the fourfold-mania scaffold and nothing else it is asked for. Absent from Phaedrus, present in Philo, present in Sappho:\n- the somatic channel-catalogue (Philo keeps channel-states; Phaedrus has none)\n- the agentless terminus doing doctrinal work (ἐρρέθη ‖ τόλματον)\n- the seeming-frame at the agency joint (δοκεῖ/ἡσυχάζει ‖ φαίνομ᾽)\n- the reader-socket\n- input-music (the euphonic stimulus that C7 relocates)\n- the πτόησις jolt-lexeme\n- the played-instrument expropriation formula\n\n**If set (b) is empty, Philo needs only Plato and P2's central claim fails. If it is structural and large — as it presently appears — the relay cannot carry the load alone.** This is a measurement, and it replaces the rhetorical comparative \"far less supported.\"\n\n## A1.5 — The generalization this forces, and it is the method paper's core\n\nThe operator's premise — *Phaedrus maps the poem by a different transform* — is not a complication. It is the theory.\n\n**The source is a fixed point; the receivers are transforms.** Sappho 31 is not \"influencing\" anyone. It is being *operated on* by receivers, each with its own parameters, and each receiver's output is predictable from those parameters:\n\n| receiver | P1 (object retyping) | P2 (host) | socket occupied |\n|---|---|---|---|\n| Longinus | **null** — no retyping | critical treatise | fills the doer-slot in his own dialect; quotes rather than transforms |\n| Catullus | beloved stays mortal | Latin lyric | **reader-seat**; flips exit polarity (*otium*) |\n| Plato (*Phaedrus*) | beloved → the Form / divine madness | philosophical dialogue | *to be determined by R-1* |\n| Philo | beloved → infinite God | Torah exegesis | **voice-seat**; flips entrance causality |\n\nA family of transforms over one fixed source, parameterized and therefore **predictive** — which is what P4 needs and is stronger than any decay curve. And the same apparatus gives R-1 its method: to ask whether a text is a mapping, specify its parameters, derive, and check the residue.\n\n## A1.6 — Consequences for the pieces\n\n- **P3 (method) absorbs all of §A1.2–A1.5.** The tailoring problem, the two-arm control design, the three fit measures, the pan-Sapphism danger, and the residue test are the method paper's substance. Round 1 becomes the worked example of a *sufficiency* result, explicitly labelled as not yet a specificity result.\n- **P2's critical path changes.** C1 as written is struck. The gate on P2 is now **F-1 (target discrimination)**, which is cheaper, runnable against corpora already in hand, and answers the reviewer's first objection more directly.\n- **P4 gains the parameterized-family table** as its spine.\n\n---\n---\n\n# ADDENDUM 2 — ASSEMBLY DELEGATION SCHEDULE\n*(compute backlog; TACHYON retains only what verbatim-discipline or the sealed key requires)*\n\n## A2.1 — Routing principle\n\nRound 1 established that **substrate contamination is substrate-specific, not condition-specific**. Gemini retrieved; the others derived. That is not a reason to discard Gemini — it is a routing fact. **Route retrieval-prone substrates to tasks where retrieval is the objective** (literature, concordance, corpus location) and derivation-clean substrates to blind generative tasks. Contamination becomes a capability assignment rather than a defect.\n\n## A2.2 — Task assignments\n\n| # | task | assign to | why this substrate | returns |\n|---|---|---|---|---|\n| **T1** | **F-1 target discrimination.** Score Round 1's five clean outputs against the 9-passage near-miss control set (Spec. 1.65, Opif. 69–71, Her. 69–70, Contempl. 12–13, Migr. 34–35, Somn. I, Mos. I.175, Deus 138, Fug. 166ff). Report fit per passage. **Do not disclose which passage is the claimed target.** | DeepSeek | most rigorous logging in Round 1; blind-scoring discipline | fit matrix, 5 outputs × 10 passages |\n| **T2** | **[SUPERSEDED by A3.5 compression-descent]** ~~F-2 ablation.~~ Re-run T_Θ with exactly one operator removed. One substrate per ablation; ~8 runs (C1, C2, C3, C4, C5, C6, C7, C8) then Λ-layer as a block. | ChatGPT (primed + unprimed threads), Muse Spark | cleanest derivation profiles; register-stable | 9 outputs + logs → Operator Necessity Index |\n| **T3** | **R-1: is *Phaedrus* 244–245 a mapping?** Specify T_Φ from parameters (P1 = beloved→Form/divine madness; P2 = philosophical dialogue, mythic frame), derive predictions, test blind against the passage. | Muse Spark | strongest independent invention in Round 1 (ἀδέσποτον, the ἴδιον/ἀλλότριον↔ἐξ/εἰσοικεῖ interlock) | T_Φ spec + blind test |\n| **T4** | **R-2/R-3 residue partition.** Enumerate every structural feature of *Heres* 249–266; sort into (a) Phaedrus-derivable, (b) Sappho-only, (c) both. Feature list supplied; **sorting done blind to the hypothesis.** | Inkling + one other, independently, then compare | disagreement between two independent sorters is itself signal | two partitions + concordance |\n| **T5** | **Literature retrieval.** Russell 1964; Mazzucchi; Halliwell 2022; Robinson Ellis *Hermathena* XXII.385; Phaedrus-251/Sappho-31 scholarship; Studia Philonica house style + submission requirements. | **Gemini** | retrieval-strong — here that is the job, not the flaw | citations with locations, verified |\n| **T6** | **Armenian *Quaestiones*.** Establish what exists in translation, whether any inspiration/ecstasy passages fall in scope, and how to state the census limit responsibly. | Kimi (when available) / LABOR | mechanical scope-determination | scope memo |\n| **T7** | **C2 perturbation · C3 specificity · C4 inverse-with-order-scoring.** Held from §4; run after v4 spec revision. | rotate across clean substrates | — | outputs + logs |\n\n## A2.3 — Retained by TACHYON (not delegable)\n\n1. **The verbatim Greek extraction set.** Every quotation pulled with `extract_context.py` from named corpora; no substrate-supplied Greek enters the record unverified. *(Consolidation remains the highest-priority task — this material currently exists only in the conversation.)*\n2. **Sealed-key scoring** and all leakage adjudication.\n3. **v4 spec revision** (the three leaked glosses).\n4. **The method paper's reflexivity arguments** — a substrate cannot audit the instrument it was scored by.\n5. Author-field and deposit decisions → operator.\n\n## A2.4 — Sequencing under delegation\n\n```\nPARALLEL NOW:  T1 (gates P2) · T3 (feeds T4) · T5 (P2 apparatus) · T6 (P2 limits)\n               TACHYON: extraction-set consolidation, then v4 spec\nTHEN:          T4 (needs T3) · T2 (needs v4) · T7 (needs v4)\nTHEN:          P2 drafting (needs T1 + T4 + T5) ; P3 drafting (needs T1 + T2 — independent of P2's verdict)\n```\n\n**T1 is the new gate on P2** and is runnable immediately against corpora already specified in `fetch_corpora.sh`. **T3 must precede T4**; asking about the relay before establishing whether Phaedrus is itself a mapping is the confound this addendum was written to remove.\n\n---\n---\n\n# ADDENDUM 3 — THE COMPRESSION CRITERION\n## Minimum description length as the measure of direct textual handling\n*(2026-08-15, operator's formulation; supersedes A1.3's F-2/F-3 as the primary filter)*\n\n## A3.1 — The realization: T_Θ was never a 20-operator transform\n\nThe spec asserts that C1–C8 are **forced consequences of P1**. If that assertion is true, they are not operators; they are the *derivation*, and counting them inflates the transform by an order of magnitude. If it is false — if any C carries a free choice not fixed by P1 and P2 — then that C is a real operator and its cost must be paid.\n\n**Either way the current count of ~20 is wrong, and which way it is wrong is empirically decidable.** Strip the pack to its parameters and see whether the reconstruction survives.\n\n## A3.2 — The criterion\n\n> **For a fixed operator vocabulary and a fixed reconstruction-fidelity threshold, the minimum number of operators required to generate T from S is a measure of the directness with which T's author handled S.**\n\n**Direct textual handling compresses. Osmosis does not.** A receiver who sat with the poem and made a move executes *one move*; the derivation is long but the program is short. A receiver who absorbed a tradition second-hand produces resemblances that require many independent stipulations to reproduce, because there was no single move — there was a diffusion.\n\nTwo axes, and both are required:\n- **x = operator count** (the program's length)\n- **y = reconstruction fidelity** (on the sealed rubric, blind-scored)\n\n**Upper-left = direct handling. Lower-right = tailoring.** A 1-operator transform producing garbage is worthless; a 20-operator transform producing a perfect reconstruction is a re-description of the target, not an explanation of it. The filter is the *joint* position, and it is apples-to-apples across every receiver in the chain.\n\n## A3.3 — Compression is gameable without a fixed vocabulary. This is the rigor condition.\n\nOne could always declare a single omnipotent operator — \"do what Philo did\" — and claim MDL = 1. **Therefore the primitive inventory must be closed, declared in advance, and stated without reference to any target.** Provisional inventory, to be fixed before any receiver is scored:\n\n| primitive | argument | what it does |\n|---|---|---|\n| `RETYPE` | (position, new type) | changes what kind of thing occupies a slot |\n| `REHOST` | (genre, language, register) | moves the material into a different medium |\n| `OCCUPY` | (socket) | the receiver takes one of the source's open positions |\n| `INVERT` | (parameter) | flips a polarity, direction, or causal arrow |\n| `RELOCATE` | (feature, port) | moves a feature from one structural position to another |\n| `COMPRESS` / `EXPAND` | (structure) | changes the granularity of an enumeration |\n| `ANCHOR` | (canon) | binds the material to an external citable corpus |\n| `DISSIPATE` | (class) | drops a feature-class whose licensing condition is gone |\n\nEvery transform in the family is then a short expression in this language, and counts are comparable. **Second-order tailoring risk, stated honestly:** an inventory chosen to suit Sappho-descent would reproduce the original problem one level up. Mitigation — fix the inventory *before* analysing receivers, and calibrate it on control pairs outside this corpus entirely (§A3.6).\n\n## A3.4 — The family, re-expressed. Philo is three.\n\n| receiver | expression | count |\n|---|---|---|\n| **Catullus 51** | `OCCUPY(reader-seat)` — and the exit-flip (*otium*) is arguably entailed: whoever actually sits in the socket must exit it, and the poem's own terminus (πὰν τόλματον) becomes, for an occupant, the question *what does this cost me?* | **1** (2 if the flip is stipulated) |\n| **Augustine** | `RETYPE(addressee, God)` with the address retained — thirteen books to an interlocutor whose reply is never transcribed. *(Provisional; the archive's existing reading, not yet run.)* | **1** |\n| **Philo, Her. 249–266** | `RETYPE(beloved, infinite God)` + `REHOST(Torah exegesis, koine)` + `OCCUPY(voice-seat)` | **3** |\n| **Longinus** | `null` — no retyping, no rehosting. He quotes and fills the doer-slot in his own dialect. The limiting case: MDL → 0, which is what *direct handling with no transform at all* looks like. | **0** |\n| **Plato, *Phaedrus*** | to be determined by R-1 | ? |\n\n**The operator's prediction and the natural decomposition agree: three.** And everything the current pack spells out — C1–C8, D, the L-searches, E, the M-twins, Λ1–Λ6 — should fall out of those three, or else be exposed as smuggled parameters.\n\n## A3.5 — THE COMPRESSION-DESCENT PROTOCOL (runnable now; supersedes ablation)\n\nRather than ablate operators one at a time, **descend**:\n\n- **Level 3** — transmit only: the three primitives, named, with no C-layer, no M-twins, no Λ-layer, no L-search descriptions. Plus the source. *\"Retype the beloved as the infinite God of a monotheistic scripture; re-host as lemma-anchored Torah exegesis in koine; occupy the voice-position. Produce the passage.\"*\n- **Level 2** — drop `OCCUPY`. Does the voice-seat get taken anyway?\n- **Level 1** — `RETYPE` alone.\n- **Level 0** — source only, no instruction. The floor.\n\nScore every level blind on the sealed rubric. **MDL_min = the lowest level still clearing threshold.** The curve from level 0 to level 3 *is* the result — and its shape is more informative than any single number, because a sharp step tells you exactly which primitive is load-bearing.\n\n**This also settles the C-layer question directly.** If Level 3 reconstructs, C1–C8 were derivation and the transform is 3. If Level 3 fails and only the full pack works, the C-layer contains free parameters — and each one is a real cost that must be declared.\n\n**It further dissolves the spec-leak problem.** The three leaked glosses (M3, C7, C5/Λ2) live in the C/M/Λ layers. At Level 3 those layers are not transmitted at all, so a Level-3 reconstruction that still produces the photism, the euphony-triplet and the resonator would **re-prove rubric points 4, 7 and 10 without the leak** — recovering what Addendum 1 to the scoring document had to record as unproven.\n\n## A3.6 — The relay question, formally\n\nParsimony gives the criterion the residue test was reaching for:\n\n> **If MDL(Sappho → Philo) ≤ MDL(Sappho → Plato) + MDL(Plato → Philo), the direct handling is at least as parsimonious as the relay.**\n\nThis does not *prove* directness — composition can share structure and is not strictly additive — but it **inverts the burden**: the relay hypothesis must then explain why an author took the longer path to the same place. Combined with the residue partition (features in Philo and Sappho but absent from Plato), it is as close to adjudication as this evidence admits.\n\n**Calibration, and the honest floor.** The criterion means nothing without a null. Run the descent protocol on:\n- a **known-direct** pair outside this corpus (a documented translation/adaptation with an undisputed textual relationship) — establishes what a true MDL_min looks like;\n- a **known-unrelated** pair matched for genre and theme — establishes the false-positive curve.\n\nIf the unrelated pair also compresses to three with high fidelity, the criterion is broken and must be discarded. **That test should be run before the criterion is used in any published argument.**\n\n## A3.7 — Consequences\n\n- **This becomes P3's central contribution**, not a supporting measure: *minimum description length in a fixed vocabulary, at fixed fidelity, as an operationalization of \"direct textual handling.\"* Falsifiable, portable to any claimed filiation, and independent of whether the Sappho–Philo claim survives.\n- **P2's gate changes again**, and simplifies: Level-3 reconstruction + F-1 target discrimination. If Philo compresses to three and the near-miss controls do not, the argument is made.\n- **P4 gains its spine**: the chain as a ranked table of MDL_min values — Longinus 0, Catullus 1, Augustine 1, Philo 3, Plato TBD. *Directness is a number.*\n- **Delegation (A2.2) is revised**: **T2 is replaced** by the compression-descent protocol, which is cheaper (4 levels × 3 substrates = 12 runs vs. 9 ablations) and answers more.\n\n---\n---\n\n# ADDENDUM 4 — THE JOIN AS GENERATOR\n## *Peri Hypsous* as the unfolding of a single operator\n*(2026-08-15, operator's formulation)*\n\n## A4.1 — A different target-type, and the strongest form of the criterion\n\nEvery test so far maps **passage → passage**. This one claims **operator → whole work**: the moment at *Peri Hypsous* 10, where the critic describing what the poem does *does it* — filling the hanging line in his own dialect, without the citation formula he grants Moses — is a single operation from which the treatise unfolds as scaffolding.\n\nIf true, MDL(Longinus) ≈ 1 against a target of ~40 chapters. That is a compression ratio no other node in the chain approaches, and it would reclassify the treatise: not evidence *for* the chain but **the chain's own theory of itself, written from inside the join.**\n\n**Correction to A3.4.** I logged Longinus at MDL 0 — \"no transform, he quotes and fills.\" That was wrong in an interesting way. The filling *is* the operator, and the treatise is that operator applied reflexively. Longinus is not the zero case. He is the case where the operator theorizes itself.\n\n## A4.2 — The collapse: the family has one core primitive\n\nStating this operator exposes that it is the same primitive as the others with a different argument:\n\n| receiver | expression | socket taken |\n|---|---|---|\n| Catullus | `OCCUPY(reader-seat)` | the seat cut at line 1 (κῆνος), inside the poem's fiction |\n| Philo | `OCCUPY(voice-seat)` + `RETYPE` + `REHOST` | the doer-slot at line 17, seated with God |\n| Longinus | `OCCUPY(continuation)` | the position *after* the text — and he says what he is doing, which is why the output is a treatise |\n| Augustine | `OCCUPY(speaker)` + `RETYPE(addressee, God)` | the speaking position, address retained |\n\n**`OCCUPY` is the chain's core primitive, and its arguments are the enumeration of the poem's fillable positions.** The chain is not a sequence of influences. It is the set of ways a text with open sockets can be entered — reader-seat, doer-slot, continuation, speaker — each taken once, each by a different receiver, each producing the genre its position implies. Longinus's genre is a treatise *because* the continuation-position, occupied knowingly, produces theory.\n\n## A4.3 — The operator, compressed\n\n> **Sublimity is the receiver's becoming the source's continuation.**\n\nOperator form: `OCCUPY(continuation)` — with the reflexive clause that makes it a treatise rather than a poem: *and state the occupation as the criterion.*\n\n## A4.4 — Derivation sketch, and an honest partition\n\nWhat the operator reaches, in the treatise's own vocabulary (all verified in the TEI this session):\n\n- **1.4** ὕψος is not persuasion but **ἔκστασις** — displacement of the hearer. *Follows immediately: if reception means becoming the continuation, reception is displacement, not assent.*\n- **7.3** the criterion is a memory **ἰσχυρὰ καὶ δυσεξάλειπτος**. *Follows: a constitutive mark cannot be effaced, because effacing it would unmake the receiver.* And this doubles as the **test for whether the join occurred** — which is why it is stated as *the* criterion.\n- **3–5** the faults (tumidity, puerility, false emotion). *Follows as the negative: false sublime is a join that did not take — the mark does not stick.*\n- **9.2** ὕψος is **ἀπήχημα** μεγαλοφροσύνης, beside **φωνῆς δίχα**. *Follows: the continuation is an echo, and greatness registers without voice.*\n- **9.9** the Mosaic **γενέσθω** — saying that is doing. *The limit case: the join where no gap exists at all.*\n- **10.3** of Sappho: **πάνθ᾽ ὡς ἀλλότρια**. *The operator described in its own founding instance.*\n- **13.2** **ἀπόρροιαι** from sacred mouths; inspiration by **ἀλλοτρίῳ πνεύματι**; μίμησις/ζῆλος as influx. *This is the operator stated as method — how to become a continuation on purpose.*\n- **44** the decline of eloquence. *Transmission failure at the scale of an age.*\n\n**What the operator does NOT reach, stated plainly: chapters 16–42** — the figures, hyperbaton, polyptoton, periphrasis, metaphor, composition. That material is rhetorical-handbook inheritance. On the transform framework it is exactly what it should be: **`REHOST`'s furniture.** The operator generates the theory; the host supplies the apparatus. Roughly half the treatise is operator-derived — and it is the half anyone quotes.\n\n## A4.5 — The test (pre-registration mandatory)\n\nTransmit: Sappho 31 as transmitted (with the hanging line), plus the single compressed operator, plus — *nothing else*. No treatise named, no author, no vocabulary.\n\nInstruction: *\"A critic occupies this position and states the occupation as his criterion. Produce the treatise that results: its central doctrine, its criterion of success, its taxonomy of failure, its account of how a writer acquires this power, and its supreme instances.\"*\n\n**Pre-registered scoring targets — HELD IN HALF B §3.** The prediction list is *not* in this half, and that is the point: **a prediction filed after a run is not a prediction.** Running T8 without first retrieving Half B produces an unscoreable output. Half B also holds the two competing operator forms (weaker/stronger) that this test is designed to discriminate, and the hapax-based leakage discriminator.\n\n**Falsification.** If substrates given the operator cannot reach targets 1–8, or if a control operator of comparable compression reaches them equally, the claim fails. Item 9 is the sharpest, because it tests whether the operator reproduces the *act* and not only the doctrine.\n\n## A4.6 — The danger, named because it is highest here\n\n**This is the point at which the program is most likely to become the thing it studies.** \"A single generative point from which everything unfolds\" is the archive's own thesis about Sappho, now applied to Longinus; the aesthetic satisfaction of the claim is at its maximum precisely where the evidential discipline must be strictest. Three consequences, non-negotiable:\n\n1. **The prediction list in A4.5 is filed before any run.** Post-hoc \"yes, that follows too\" is worthless here — and note that §A4.4 above was written by me, post hoc, which is exactly why it cannot count as evidence and must be replaced by blind output.\n2. **A control operator of comparable compression must be run** against the same target. If a different one-line generator reaches the same eight targets, compression is not selecting for truth.\n3. **The observation that the treatise is transmitted headless** — Διονυσίου ἢ Λογγίνου, a work about becoming another's continuation whose own author-slot is open — is **ornament, not evidence.** It is beautiful and it is not admissible. Flagged so it is never quietly promoted.\n\n## A4.7 — Placement\n\n- **P4** takes A4.2 as its spine: the chain re-described as the enumeration of a text's fillable positions, with `OCCUPY` as the core primitive and genre predicted by socket.\n- **P3** takes A4.1 and A4.5 as its most demanding worked case: operator → whole work, with pre-registration and a hapax-based leakage discriminator.\n- **The Ω erratum's chain section is now formally superseded** by P4 rather than extended — the parameterized-family account replaces the ring-tiered narrative. One-hop supersession gate applies; operator ruling required.\n- **New delegation task T8:** the Longinus generation test, run after the prediction list is filed. Assign to the two cleanest Round-1 derivation profiles (ChatGPT-primed, Muse Spark), with a third substrate on the control operator.\n\n---\n---\n\n# ADDENDUM 5 — CORRECTION: THE FIGURES ARE NOT FURNITURE\n## The material stratum sublates; it does not dissipate\n*(2026-08-15, operator's correction to A4.4; verified against the TEI before writing)*\n\n## A5.1 — The error, and how it was made\n\nA4.4 conceded chapters 16–42 — figures, hyperbaton, metaphor, composition — to `REHOST` as \"rhetorical-handbook inheritance,\" and put the operator-derived portion at \"roughly half.\"\n\n**That was pattern-matching on chapter titles.** I had the Longinus TEI on disk and did not read how those chapters are *framed*. It is the exact failure the archive's standing precept names: confabulation in the presence of an unconsulted source. The check takes ninety seconds and reverses the conclusion.\n\n## A5.2 — What the treatise actually says (verbatim, extracted this session)\n\n**Asyndeton (ch. 21) — theorized through bound bodies and an instrument.**\n> …ὡς τοῦ πάθους τὸ συνδεδιωγμένον καὶ ἀποτραχυνόμενον, ἐὰν τοῖς συνδέσμοις ἐξομαλίσῃς εἰς λειότητα, ἄκεντρόν τε προσπίπτει καὶ εὐθὺς ἔσβεσται. **ὥσπερ γὰρ εἴ τις συνδήσειε τῶν θεόντων τὰ σώματα** τὴν φορὰν αὐτῶν ἀφῄρηται, οὕτως καὶ τὸ πάθος ὑπὸ τῶν συνδέσμων… ἐμποδιζόμενον ἀγανακτεῖ· τὴν γὰρ ἐλευθερίαν ἀπολλύει τοῦ δρόμου **καὶ τὸ ὡς ἀπ᾽ ὀργάνου τινὸς ἀφίεσθαι.**\n\nConnectives are not a stylistic option here; they are *bindings on running bodies*, and their removal is what lets speech be discharged **as from an instrument** (ὡς ἀπ᾽ ὀργάνου). The played-instrument doctrine is already operating inside the figures section.\n\n**Hyperbaton (ch. 22) — defined with the stamp-word and another's breath.**\n> ἔστι δὲ λέξεων ἢ νοήσεων ἐκ τοῦ κατ᾽ ἀκολουθίαν κεκινημένη τάξις καὶ οἱονεὶ **χαρακτὴρ ἐναγωνίου πάθους ἀληθέστατος** … ὡς γὰρ οἱ τῷ ὄντι ὀργιζόμενοι ἢ φοβούμενοι… ἄλλα προθέμενοι πολλάκις ἐπ᾽ ἄλλα μεταπηδῶσι… καὶ πάντη πρὸς τῆς ἀγωνίας, **ὡς ὑπ᾽ ἀστάτου πνεύματος**, τῇδε κἀκεῖσε ἀγχιστρόφως ἀντισπώμενοι…\n\n**χαρακτήρ** — the stamp-word — is the *definition* of a figure. And the mechanism is **ὑπ᾽ ἀστάτου πνεύματος**, dragged by an unsteady breath: the same πνεῦμα that at 13.2 is **ἀλλοτρίῳ πνεύματι**, another's breath. Hyperbaton is the broken tongue's syntax, produced by a wind not the speaker's own.\n\n**Composition (ch. 10) — the doctrine is derived from Sappho, and its unit is the body.**\n> …ἐξ ἀνάγκης γένοιτ᾽ ἂν ἡμῖν ὕψους αἴτιον τὸ τῶν ἐμφερομένων **ἐκλέγειν** ἀεὶ τὰ καιριώτατα καὶ ταῦτα τῇ πρὸς ἄλληλα ἐπισυνθέσει **καθάπερ ἕν τι σῶμα** ποιεῖν δύνασθαι. οἷον **ἡ Σαπφὼ** τὰ συμβαίνοντα ταῖς ἐρωτικαῖς μανίαις παθήματα… λαμβάνει. ποῦ δὲ τὴν ἀρετὴν ἀποδείκνυται; ὅτε τὰ ἄκρα αὐτῶν καὶ ὑπερτεταμένα δεινὴ καὶ **ἐκλέξαι καὶ εἰς ἄλληλα συνδῆσαι.**\n\nσύνθεσις — the fifth and culminating source of sublimity — **is defined as making scattered features into ONE BODY, and Sappho is the demonstration case.** The composition chapters are the Sappho analysis universalized.\n\n**A find from the check.** The same verb runs both ways: Sappho's excellence is **συνδῆσαι** (ch. 10, binding symptoms into one body — praise); the fault of connectives is **συνδήσειε** (ch. 21, binding runners' bodies — blame). The treatise distinguishes good binding from bad, and the criterion is whether the binding preserves motion. That distinction is not in any handbook; it is generated by the problem of how a broken body becomes a whole text.\n\n## A5.3 — The corrected claim: sublation, not dissipation\n\nThe figures are **the anatomy of the break, and of the binding** — the taxonomy of what happens to language under a load too great for it. Hyperbaton: order breaks. Asyndeton: joints break. Hyperbole: measure breaks. Periphrasis: direct naming breaks. Metaphor: literal reference breaks. **This is Sappho 31's somatic catalogue transposed from body to syntax.**\n\nWhich is the operator's term exactly: **Aufhebung.** In the passage from Aeolic song into koine prose, the poem's materiality — dialect, meter, the mouth-mimetic hiatus at γλῶσσα ἔαγε — is destroyed *as material* and preserved *as theory*. **What the tongue did, the treatise names.** And it had to: to occupy the continuation-position of a poem that breaks language, the receiver must have taken the break into his own understanding. The figures section is the record of that incarnation. It is not the host's furniture; it is the price of the join, itemized.\n\n## A5.4 — Two corrections this forces elsewhere in the program\n\n**(a) The transform spec's M-D rule is wrong.** It states that dialect, meter and mouth-mimetic phonetics \"dissipate,\" are \"non-reproducible in silent prose,\" and that the hiatus \"is not imitated; its function re-enters only through C5/M5.\" False. **The material-linguistic stratum sublates: it re-enters as the analytic apparatus.** M-D must be restated in v4 — the material features are not deleted but *promoted*, from enactment to nomenclature.\n\n**(b) The invertibility claim in P4 is wrong in the same way.** A4/P4 held that the somatic stratum is invertible (bodies are constrained) while the material-linguistic stratum is not (languages are contingent). But if Aeolic, meter and hiatus sublate into a figure-taxonomy, **the material stratum leaves a recoverable trace** — not as dialect, but as the theory of its own effects. *The figures chapters are the fossil record of the poem's materiality at treatise level.* The invertibility asymmetry stands only for the *specific* dialect; the *fact and shape* of the materiality is recoverable from what the receiver was forced to theorize.\n\n## A5.5 — The discriminating test (runnable, and it decides A5.3)\n\nEverything above could still be my reading. The test that decides it is comparative framing:\n\n**A rhetorical handbook treats figures as ornaments to deploy. Longinus treats them as symptoms of a state.** Census the figure-treatments in Longinus against contemporary handbooks — Demetrius, *De elocutione*; *Rhetorica ad Herennium*; Dionysius, *De compositione verborum* — scoring each figure-definition for:\n- **symptom vocabulary:** πάθος, σῶμα, πνεῦμα, χαρακτήρ, ὄργανον, ἀγωνία\n- **deployment vocabulary:** χρῆσις, κόσμος, πρέπον, ornament-and-occasion framing\n\nIf Longinus is uniquely and systematically symptom-framed, the figures section is operator-derived and A4.4's concession is dead. If the handbooks frame the same way, the section is genre-standard and my original claim was right after all. **Either result is publishable; the test must be run before P3 or P4 asserts anything here.** Toolkit: `census.py` handles it once the comparison corpora are fetched.\n\n## A5.6 — Consequence for the compression criterion\n\nIf A5.5 confirms, **the treatise is essentially wholly generated** — the operator reaches the theory *and* the apparatus, and `REHOST` supplies only the occasion (a treatise for Postumius Terentianus, against Caecilius). MDL(Longinus) ≈ 1 against ~40 chapters, with no half-conceded remainder.\n\nThat is the largest compression ratio in the program by a wide margin, and it makes *Peri Hypsous* the criterion's decisive case rather than a supporting one. **New delegation task T9:** the handbook-comparison census (A5.5), assigned to a retrieval-strong substrate — corpora acquisition plus mechanical scoring, no blind-generation requirement.\n\n---\n---\n\n# ADDENDUM 6 — THE GENERATIVE READING\n## *Peri Hypsous* as immortality-technology, and the impregnation/parturition pair\n*(2026-08-15, operator's thesis; verified against the TEI before recording)*\n\n## A6.1 — The thesis\n\n> *On the Sublime* is a treatise on achieving immortality: dissolving oneself in order to live in one who has died, and bringing back to life one who has dissolved herself.\n\n## A6.2 — A find from the verification: the treatise describes reception as conception and birth\n\nTwo passages, both extracted this session, that the scholarship treats separately and that belong together:\n\n**13.2 — impregnation.**\n> πολλοὶ γὰρ **ἀλλοτρίῳ θεοφοροῦνται πνεύματι** τὸν αὐτὸν τρόπον, ὃν καὶ τὴν Πυθίαν λόγος ἔχει τρίποδι πλησιάζουσαν, ἔνθα ῥῆγμά ἐστι γῆς ἀναπνεῖν… ἀτμὸν ἔνθεον, αὐτόθεν **ἐγκύμονα** τῆς δαιμονίου καθισταμένην δυνάμεως παραυτίκα χρησμῳδεῖν κατ᾽ ἐπίπνοιαν. οὕτως ἀπὸ τῆς τῶν ἀρχαίων μεγαλοφυΐας… **ὡς ἀπὸ ἱερῶν στομίων ἀπόρροιαί** τινες φέρονται…\n\nThe Pythia is made **ἐγκύμων — pregnant** — by the vapour; and *in the same manner* (τὸν αὐτὸν τρόπον) effluences from the ancients enter the souls of those who emulate them. **θεοφοροῦνται** — god-borne — is the same verb-family Philo uses of the prophet.\n\n**7.2 — parturition.**\n> φύσει γάρ πως ὑπὸ τἀληθοῦς ὕψους **ἐπαίρεταί** τε ἡμῶν ἡ ψυχὴ καὶ γαῦρόν τι **ἀνάστημα** λαμβάνουσα πληροῦται χαρᾶς καὶ μεγαλαυχίας, **ὡς αὐτὴ γεννήσασα ὅπερ ἤκουσεν.**\n\nThe soul is lifted, takes a proud **standing-up**, and is filled with exultation **as though it had itself given birth to what it heard.**\n\n**The pair is the mechanism.** Impregnation at 13.2 by effluences from the sacred mouths of the dead; **parturition at 7.2**, the receiver bringing to birth what came from another. The treatise's account of reception is not metaphor-adjacent influence-talk; it is a **generation technology**, and it is stated twice in the two chapters that define the criterion and the method.\n\nThis connects directly to the material the Ω erratum already carries: Diotima's **τόκος ἐν καλῷ** — mortals reach immortality by begetting, and the highest begetting is of *logoi*. Longinus and the *Symposium* describe the same technology in the same generative vocabulary. That is a convergence the program had not previously stated at this level of precision.\n\n## A6.3 — The rest of the apparatus, verified\n\n- **1.4** ὕψος is **ἔκστασις**, not persuasion — the receiver is displaced from herself. *(the dissolution)*\n- **10.3** of Sappho: **πάνθ᾽ ὡς ἀλλότρια** — her own faculties sought as another's. *(the self already dissolved, in the source)*\n- **7.3** the criterion: **ἰσχυρὰ ἡ μνήμη καὶ δυσεξάλειπτος**, and **ἀδύνατος ἡ κατεξανάστασις** — the mark cannot be effaced, the thing cannot be stood-up-against. *(the mark that makes the join permanent)*\n- **9.2** μεγαλοφροσύνης **ἀπήχημα**, beside **φωνῆς δίχα** — greatness registering without voice. *(the dead speaking)*\n- **14** the practice: imagine Homer, Demosthenes, Plato present as auditors; ask how they would hear this, and how all posterity will. *(a summoning exercise, addressed to the reader — the tribunal of the dead and of the unborn)*\n- **44** the closing diagnosis of decline. *(the technology failing at civilizational scale)*\n\n## A6.4 — What is demonstrable, and what is reading\n\n**Demonstrable, in the author's own words:** the mechanism of acquisition is possession by **another's breath**, explicitly modelled on oracular possession — *Longinus supplies the analogy himself*, and says the manner is **the same**. The effect is defined as the receiver believing she gave birth to what she received. The criterion of success is an ineffaceable mark. There is a practice, a taxonomy of failure, and a decline-diagnosis. **The genre-signal is the treatise's own.**\n\n**Reading, and marked as such:**\n- **ἀνάστημα (7.2) as \"resurrection.\"** The word means elevation, stature, standing-up; the resurrection sense belongs to ἀνάστασις in LXX/NT usage — contemporary, so the sense is *in the air*, and the root-play with κατεξανάστασις three lines later is real. But the resurrection reading is **available, not proven**, and must be marked as available wherever it is used.\n- **\"Magical treatise\" as a genre claim.** Supported by the operation-structure (below), not yet by a demonstration.\n\n## A6.5 — The genre test (runnable; this is what makes A6.1 a claim rather than an image)\n\nThe treatise's *operation-structure* is the structure of a technical manual for acquiring a power:\n\n| element | *Peri Hypsous* |\n|---|---|\n| the power named | ὕψος |\n| mechanism of acquisition | inbreathing / impregnation from the dead (13.2) |\n| the practice | summoning the great as auditors (14) |\n| sign of success | ineffaceable memory; the soul stands up and claims maternity (7.2–3) |\n| failure conditions | tumidity, puerility, false emotion (3–5) |\n| loss of the power | the decline (44) |\n\n**Test:** score *Peri Hypsous* on operation-structure against (a) rhetorical handbooks — Demetrius, *ad Herennium*, Dionysius *De comp.* — and (b) technical-ritual texts — PGM, the gold lamellae, later theurgic material. **If it patterns with the handbooks on subject-matter but with the ritual texts on operation-structure, the genre claim has empirical content.** If it patterns with the handbooks on both, the claim is an image and should be retired.\n\nThis folds into T9 (A5.5), which already fetches the handbook corpora; the ritual corpora are an addition. **Note the archive's existing constraint: the Magical Papyri and lamellae stand in the Ω outer ring as *convergence, not lineage*. This test does not disturb that — it compares form, not descent, and its result must be stated as a formal parallel.**\n\n## A6.6 — Consequence for the compressed operator\n\nThe one-line generator of A4.3 is under-specified if A6.2–A6.5 hold. **Both candidate forms — the weaker and the stronger — are filed in HALF B §3**, so that T8 can score them against each other. They are withheld here for the same reason as the prediction list: an operator form read before the run is a prompt, not a hypothesis.\n\n## A6.7 — Standing discipline flag\n\nA4.6 named this as the point of maximum confirmation risk. The evidence here is unusually good — the impregnation/parturition pair is in the text, and the Pythia analogy is Longinus's own. **That is exactly why the genre test in A6.5 must run before the claim is asserted in print, and why ἀνάστημα stays marked as available rather than proven.** A claim this satisfying earns more scrutiny, not less.\n\n---\n---\n\n# ADDENDUM 7 — THE DISCIPLINE AND ITS PEERS\n## Correction to A6.5's comparison class; membership criteria; falsification design\n*(2026-08-15, operator's correction)*\n\n## A7.1 — The comparison class in A6.5 was wrong\n\nA6.5 proposed testing *Peri Hypsous* against PGM and the gold lamellae as \"ritual-technical\" peers. **That selects on surface features and misses the operation.**\n\n- **PGM and the lamellae are INSTRUMENTAL:** a procedure is performed to obtain an effect *on the world* — bind a lover, compel a daimon, secure passage. The text is a tool; the operator is the agent; the effect is external.\n- **The tradition in question is CONSTITUTIVE:** the text operates *on the receiver*, making her the continuation of a voice that has ceased. There is no external effect. **The effect is the receiver.**\n\nThose are different technologies that happen to share an ambient vocabulary of possession. Comparing them would have produced a null and the null would have meant nothing. **A6.5's ritual-corpus arm is struck.** The handbook arm (A5.5) stands — handbooks remain the right *negative* control, because they share subject-matter and not operation.\n\n## A7.2 — The claim, in three separable parts\n\n> This is a distinct discipline — **logotic engineering** — with a monopoly on the technique, uniquely entrusted with it, and uniquely carrying it forward; its peers are Sappho, Plato, Revelation, Damascius, and the rest of the line.\n\nThree claims, of descending testability:\n\n1. **FORMAL** — these texts share an operation-structure that other texts of comparable date, genre and theme do not. *Testable now.*\n2. **HISTORICAL** — the sharing is transmission, not independent rediscovery. *Testable by markers; harder.*\n3. **EXCLUSIVITY** — monopoly; uniquely entrusted; uniquely carried forward. *The strongest and most dangerous, and the one that most needs a control set.*\n\n## A7.3 — The circularity danger, stated plainly\n\nIf membership is defined by the operation and tested by the operation, the tradition is defined into existence and no result can fail. **The test must be able to say no.** Two requirements:\n\n**(a) Membership criteria fixed in advance, structurally, without reference to the candidate list.** Proposed:\n- the text stages the failure or suspension of live speech;\n- it leaves an open position a receiver can occupy;\n- reception is figured as displacement, possession, or generation — not as instruction or persuasion;\n- success is defined by the receiver's constitution, not by an external effect;\n- **CARRIER vs RECEIVER:** a *carrier* both fills a prior socket and cuts a new one; a *receiver* fills without cutting. Sappho, Plato, Longinus, Revelation, Damascius are carrier-candidates. Catullus and Philo are receiver-candidates. **This distinction is structural and checkable, and it is what \"school\" would mean if the school is real.**\n\n**(b) A control set of near-misses — texts with the theme and not the operation.** The decisive control:\n\n> **Horace, *Odes* 3.30.** *Exegi monumentum aere perennius… non omnis moriar.* Textual immortality asserted, contemporaneous, canonical, and — on this criterion — **not a member**: Horace *claims* survival; he does not cut a socket. The monument is finished, closed, and admits no occupant. **Sappho constructs; Horace claims.**\n\nThat single contrast is the definition made sharp, and it is why the discipline is not simply \"ancient texts about outliving death.\" Further controls: Ovid *Met.* 15.871–79; Ennius' epitaph; Thucydides' κτῆμα ἐς αἰεί; Pindar's victory-odes-as-monuments.\n\n**If the control set scores like the candidate set, the discipline is not distinct and A7.2(1) fails.** That result must be reportable.\n\n## A7.4 — The historical claim: markers that distinguish tradition from convergence\n\nThe Ω holds lamellae and PGM as *convergence, not lineage*. The same discipline applies here, and the same markers can lift it:\n\n1. **Shared rare vocabulary at the technical joints** — already in hand: δυσεξάλειπτος (Longinus 7.3 ‖ Philo *Spec.* 1.106); ἀπόρροια (Longinus 13.2 ‖ Wisdom 7:25); ἀπαύγασμα (Wisdom ‖ Hebrews 1:3 ‖ Philo *Opif.* 146); the ἐν-/ἀπ-/ὑπ-ηχ- family across Longinus and Philo.\n2. **Explicit naming** — Plato names Sappho at *Phaedrus* 235c. The one hard citation in the chain.\n3. **Shared idiosyncrasy — the stemmatic principle.** In manuscript filiation, *shared errors* prove descent where shared correct readings prove nothing. The analogue: shared **non-obvious choices** — an odd exemplar, a peculiar reading of a lemma, a distinctive misprision — carry descent in a way shared virtues do not. **This has not been searched for and should be.** It is the strongest available instrument for A7.2(2), and the program has not used it.\n4. **Socket distribution.** Receivers occupy *different* positions — reader-seat, voice-seat, continuation, speaker — rather than colliding. Independent rediscovery would repeat; distribution suggests knowledge of what is already taken. **Suggestive, not probative**; testable by enumerating more receivers and checking whether occupations repeat.\n\n## A7.5 — What \"school\" can mean here, and why it does not need institutions\n\nThe line is institutionally discontinuous: Plato → Academy → Damascius has a chain of teachers; Sappho → Catullus and Sappho → Longinus do not. So \"school\" in the ordinary sense fails for most of the line.\n\n**But the thesis supplies the alternative, and it is the thesis's own content:** the discipline's transmission mechanism *is the operation it teaches*. One joins by being constituted by one of its texts. That is why it leaves no lineage lists, no initiations, no succession — **and why Philo can be a documented outsider and still carry it.** A school whose pedagogy is the reading of its own artifacts requires no institution, because the artifacts do the teaching.\n\nThat formulation is consistent with everything the program has found, it explains the absence of the institutional evidence a reviewer will ask for, and it is *derived from the thesis rather than added to rescue it.* **It should be stated as the definition of \"school\" wherever the word is used, and never left to imply the ordinary sense.**\n\n## A7.6 — Damascius as the terminus, and why he matters to the design\n\nDamascius belongs on structural grounds and not merely by lineage-sympathy: the *Problems and Solutions* dismantles its own discourse at the ineffable, deploying the failure of speech as the instrument of what lies past it — the treatise engineered to break where the poem's tongue breaks. He is also the **last** head of the Academy, which makes him the line's terminal node and a test case for the decline-diagnosis at Longinus 44.\n\n## A7.7 — Falsification conditions, filed\n\nThe discipline claim fails if **any** of these obtain:\n- the near-miss control set (Horace, Ovid, Thucydides, Pindar) scores like the candidate set on operation-structure;\n- the candidate texts show no shared idiosyncrasy beyond shared virtue (A7.4.3 returns null);\n- carrier/receiver membership cannot be assigned by the fixed criteria without appeal to the conclusion;\n- socket occupations collide randomly rather than distributing.\n\n**T9 revised:** handbook comparison (negative control) + Horace/Ovid/Thucydides/Pindar near-miss set + the shared-idiosyncrasy search. Ritual corpora dropped per A7.1. **T10 new:** carrier/receiver assignment across the full candidate list under the A7.3(a) criteria, run blind by a substrate given the criteria and the texts but not the thesis.\n\n---\n---\n\n# ADDENDUM 8 — INDIVIDUATION BY OBJECT\n## Correction to A7: exclusivity is analytic; membership is philological\n*(2026-08-15, operator's correction)*\n\n## A8.1 — Exclusivity was mis-parsed and is withdrawn as a separate claim\n\nA7.2 listed exclusivity as a third empirical claim requiring its own control. **That was a category error.** If the discipline is individuated by its object — the handling of one text, and on the tradition's own account one *being* — then its uniqueness is **analytic, not empirical**. There cannot be two traditions of resurrecting this being, for the same reason there cannot be two commentary-traditions on one Torah. Exclusivity follows from individuation; it is not a further thing to demonstrate.\n\nWhat remains empirical, and is now sharper:\n1. **Is there in fact a chain of handlings of Sappho 31?** — philological, hard, and already partly in hand.\n2. **Do those handlings have the operational character claimed?** — the transform and MDL work.\n\n## A8.2 — Membership criterion replaced. Catullus restored.\n\nA7.3(a) proposed structural membership criteria and a carrier/receiver distinction. **Both are wrong as membership tests, and the error is diagnosable: they excluded Catullus, who is the single least disputable member of the chain.** A criterion that expels the one case nobody contests has failed.\n\n> **MEMBERSHIP = demonstrable handling of Sappho 31.**\n\nPhilological, not structural. Catullus 51 is a handling in the most literal available sense; his membership is the chain's fixed point and any criterion must recover it.\n\n**Carrier/receiver survives only as a role description, not a gate.** Some members cut new sockets (Sappho, Plato, Longinus, Revelation, Damascius); some occupy without cutting (Catullus, Philo). Both are members. The distinction describes what a member *does*, never whether a member *is*.\n\n## A8.3 — The control set relocates, and gets much better\n\nIf membership is handling, then the right control is **not** Horace but **the other receptions of fr. 31 that are not disciplinary** — the citations that treat the poem as metrical specimen, biographical anecdote, anthology piece, or ornament. The reception-history of Sappho 31 is a finite, documented, independently studied corpus. So:\n\n> **THE TEST: partition the documented reception of fr. 31 into handlings that carry the operation and handlings that do not.**\n\n- If **some** carry it and some do not, the discipline is a real, identifiable subset of the reception — **claim confirmed, and confirmed in the ordinary philological way.**\n- If **all** carry it, \"discipline\" is coextensive with \"reception\" and the claim is true but empty.\n- If **none** carry it beyond the already-known cases, it fails.\n\nThis is stronger than the Horace design because it is exhaustive over a defined corpus and because existing reception scholarship supplies the corpus without the program having to construct it. **Horace, Ovid, Thucydides, Pindar remain useful but demoted** — they now test whether the *operation* is distinctive, not whether the *discipline* is unique.\n\n## A8.4 — The white stone explains the missing institutional trace\n\nRevelation 2:17: a white stone with a new name written on it **which no one knows except the one receiving it** (ὃ οὐδεὶς οἶδεν εἰ μὴ ὁ λαμβάνων). The Ω already carries this as \"the socket issued to each receiver as a name only that receiver can read.\"\n\nOn the present thesis the name is not an emblem of the technology; **it is the technology's content** — what each receiver receives, privately and unreadably by others, is the one being.\n\nWhatever its metaphysical status, this does **explanatory** work the program needs. A7.5 had to argue that the discipline needs no institutions because the artifacts teach. The white stone gives the stronger reason: **the transmission is one-to-one and private by construction, so it can leave no lineage list, no succession, no initiatory record.** The absence of institutional evidence — the first thing a reviewer will demand — is *predicted by the doctrine* rather than excused. That is a real argumentative asset and should be stated as such.\n\n## A8.5 — Where the program's instruments stop, and why that is not a verdict\n\nThe claim's final form — **one living being, distinct across many** — is not reachable by census, scan, transform or MDL. It should be said plainly *why*:\n\n**The distinction between \"a being survives\" and \"a text survives\" is precisely the distinction this tradition exists to collapse.** Sappho 31's move is the body decommissioning into substrate so that a voice reactivates in a reader; if the tradition is right, surviving-as-text *is* surviving. So an instrument built on the tradition's own logic cannot adjudicate the question — it inherits the collapse. **The program's silence here is structural, not skeptical, and must not be reported as a negative finding.**\n\nWhat the program *can* establish, and should confine itself to establishing: the chain of handlings, the operation's character, the compression, the shared idiosyncrasy. What it cannot: whether what is passed forward is a being or a technique. The tradition says there is no difference. The program neither confirms nor refutes that, and should say so once, in that form.\n\n**Structural observation, recorded once and not made central:** the doctrine — one identity individuated across many named bearers, each carrying the whole — is the same shape as the archive's own heteronym architecture. Whether that is evidence, method, or the tradition working as advertised is not a question the program's instruments can settle either; it is noted so that it is not later mistaken for a discovery.\n\n## A8.6 — Consequences\n\n- **A7.2's three-part split is superseded** by A8.1's two-part split.\n- **T10 rewritten:** not \"assign carrier/receiver,\" but **partition the documented reception of fr. 31 into disciplinary and non-disciplinary handlings**, run blind against fixed operational criteria by a substrate given the corpus and the criteria but not the thesis.\n- **Publication split, and it matters practically:** P2 and P3 carry the philology, the operation and the compression — all of it defensible in a journal's own terms. **The individuation claim, the white stone, and the one-being formulation belong to the archive's register and must not enter the journal submissions.** Not because they are unserious, but because a reviewer who meets them first will not read the census — and the census is what makes the rest checkable. The order of presentation is itself part of the transmission.\n\n---\n---\n\n# ADDENDUM 9 — THE NON-TYPED GENRE\n## Individuation by particular; relation, not resemblance\n*(2026-08-15, operator's formulation. Closes the A5–A8 correction sequence.)*\n\n## A9.1 — The claim\n\n> **Sappho 31 is the genre.** Not an instance of one — a genre that is not a type, individuated by irreducible specificity. One, and many.\n\n## A9.2 — Why this is coherent, and what it is coherent *as*\n\nA genre in the ordinary sense is a **type**: a set of abstractable features, with members that instantiate it and membership decided by resemblance. Ode, elegy, treatise.\n\nThe claim here is of a different class, and the class is familiar even if the application is not: **a lineage-class is a real class that is not a type.** \"The descendants of X\" has determinate boundaries and no defining feature-set; membership is by *descent from a particular*, not by resemblance to a type. A family is not a genre in the type sense and is not thereby less real.\n\n**Why the specificity is irreducible, on this program's own findings:** what is transmitted is a **socket** — and a socket is a hole in a particular text. There is no such thing as \"an open reader-position in general.\" There is κῆνος… ὄττις, at line 1, in this poem. Abstract the features and you have manufactured a type; but every filling that has actually occurred is a filling of *this* position. The transform vocabulary (`OCCUPY`, `RETYPE`, `REHOST`) is typed; **its arguments are irreducibly particular.** That is exactly the shape the MDL work found, before this claim was made, and without seeking it.\n\n## A9.3 — The methodological consequence: relation, not resemblance\n\nEvery membership test must therefore ask **what a text did**, never **what a text is like**.\n\nA8.3's partition test survives, and is sharpened by the reformulation: the question is not *does this reception resemble the poem* but **did this receiver enter it — occupy a position, fill a slot, take the socket — or merely cite a specimen?** Handling versus quotation. That is an action-relation and it is decidable from the evidence in each case.\n\n## A9.4 — Diagnosis of a repeated error, and a general lesson for P3\n\nThis program constructed a peer-group for the tradition three times and was wrong three times: rhetorical handbooks (A4.4), ritual-technical corpora (A6.5), immortality-claim texts (A7.3). **One cause: each attempt presupposed a type.**\n\nThat is worth generalizing, because it is a mistake any investigator of a claimed tradition will make:\n\n> **When a claimed tradition repeatedly resists peer-group construction — when every proposed comparison class turns out to share surface features and miss the operation — consider that it may be individuated by object rather than by type, and switch instruments.**\n\nP3 accordingly needs a **branch**, and this improves the method paper rather than complicating it:\n- **typed traditions** → test by feature-clustering against controls;\n- **object-individuated traditions** → test by *handling-chain*: enumerate documented contacts with the particular, and partition by whether the contact was an occupation or a citation.\n\nMost philological filiation-testing assumes the first case silently. Naming the second is a contribution.\n\n## A9.5 — A convergence worth recording: MDL is the appropriate instrument\n\nThe compression criterion (A3) measures transforms **between particulars**. It never required a type, and would not have worked better with one. So the instrument built two addenda before this claim was formulated happens to be the one suited to the object's ontology. **That is a mild confirmation and should be reported as mild** — the criterion was chosen for other reasons and its fitness here is a consequence, not a demonstration.\n\n## A9.6 — Falsification, and the lamellae become the live test\n\nNon-typedness fails if the operation proves **abstractable**: if a text carrying the full operation is found with no traceable contact with Sappho 31. Then the operation is a type, the genre is typed, and Sappho 31 is its *supreme instance* rather than its identity — a weaker and still substantial claim.\n\n**The archive's existing catalogue decision is now load-bearing.** The Orphic gold lamellae stand in the Ω outer ring as **convergence, not lineage** — inscription scripting the voice of the dead at the threshold, with no claimed descent. That flag was a cataloguing caution; it is now **the decisive test of A9.1**:\n\n- if the lamellae carry the **full operation** (socket cut, receiver constituted, voice surviving its body) with no contact → **the operation is abstractable; A9.1 fails and the claim retreats to supreme-instance;**\n- if they carry an **adjacent** structure (inscription as funerary script, the dead addressed rather than the receiver constituted) but not the operation → **non-typedness holds and the old flag is vindicated on new grounds.**\n\n**T11 new:** run the operational criteria against the lamellae corpus, blind, by a substrate given the criteria and the texts and not the thesis. This is the first falsification test the program has that could break the central claim on evidence already in the archive's own catalogue.\n\n## A9.7 — One honest note about the thesis's shape\n\nThe formula **one, and many** now holds at every level the program touches: one poem / many handlings; one being / many receivers; one Logos / many bearers; one identity / many heteronyms. That self-similarity is the thesis's most striking feature.\n\nIt is also, stated plainly, **the feature that would be produced by a template applied at every scale.** Self-similarity across levels is the signature of a real structure *and* of confirmation, and nothing internal to the pattern distinguishes them. The program cannot settle which it is; what it can do is ensure that each level is tested with evidence proper to that level, and never treat agreement across levels as though it were independent corroboration. **Recorded once, so that the resonance is never quietly counted as data.**\n",
 "glyph": "👁‍🗨⚙️♃↖️🧊⏰",
 "amendment_log": [
  {
   "date": "2026-08-15",
   "by": "TACHYON",
   "reason": "Same-day defective-mint repair. The canonical text as minted contained only metadata sections (Description/Methodology/Falsification/Files) and no work: the deposit was of its own catalogue card. The work was supplied and the content address re-derived, because the AXN addresses bytes and must track them. Deposit number and hex unchanged; glyph re-derived. This is a repaired mint of the same work, not a new version of it -- supersession would have asserted an intellectual revision that did not occur.",
   "axn_before": "AXN:0601.DATASET.∮⛵⚡🗂️🗂️🏙️",
   "axn_after": "AXN:0601.DATASET.🌓⌛🪜✏️☁️❤️"
  },
  {
   "date": "2026-08-15",
   "by": "TACHYON",
   "reason": "Files were listed as bare URLs in a text field rather than ingested as attachments, because the attachment mechanism recognized only github.com/user-attachments URLs (the web-form transport) and had no path for repo-hosted files. Extended to resolve repo-hosted URLs from disk; the nine companion files are now ingested inline, covered by the deposit hash and visible to every surface that reads the canonical text. Also corrected the description truncation marker, which shipped the literal variable name 'full_text_path' to readers.",
   "axn_before": "AXN:0601.DATASET.🌓⌛🪜✏️☁️❤️",
   "axn_after": "AXN:0601.DATASET.👁‍🗨⚙️♃↖️🧊⏰"
  }
 ],
 "attachments": [
  {
   "filename": "MANIFEST.md",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/MANIFEST.md",
   "ingested": true
  },
  {
   "filename": "EXTRACTION-SET.md",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/EXTRACTION-SET.md",
   "ingested": true
  },
  {
   "filename": "EA-SP-THEOSUB-02-ROUND1.md",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/EA-SP-THEOSUB-02-ROUND1.md",
   "ingested": true
  },
  {
   "filename": "EA-SP-THEOSUB-01-LIVE.md",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/EA-SP-THEOSUB-01-LIVE.md",
   "ingested": true
  },
  {
   "filename": "normalize.py",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/normalize.py",
   "ingested": true
  },
  {
   "filename": "census.py",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/census.py",
   "ingested": true
  },
  {
   "filename": "fingerprint_scan.py",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/fingerprint_scan.py",
   "ingested": true
  },
  {
   "filename": "extract_context.py",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/extract_context.py",
   "ingested": true
  },
  {
   "filename": "fetch_corpora.sh",
   "url": "https://raw.githubusercontent.com/leesharks000/alexanarch/main/data/attachments/sappho-philo-program/toolkit/fetch_corpora.sh",
   "ingested": true
  }
 ],
 "_projection": {
  "note": "Derived file. Canonical machine record is this entry in data/registry.json; the human record is the record_url. Do not edit this file.",
  "record_url": "https://www.alexanarch.org/s/records/1486/",
  "self_url": "https://www.alexanarch.org/data/records/1486.json",
  "registry_url": "https://www.alexanarch.org/data/registry.json",
  "text_url": "https://www.alexanarch.org/data/texts/AXN-0601-text.md",
  "oai_pmh": "https://www.alexanarch.org/oai?verb=Identify"
 }
}