The advisors and masters who speak in this session are constructed critical lenses named for the people whose working methods they abstract. They are not those people's opinions, and nothing on this page should be quoted as the view of any real person.

Every claim on this page about Christopher Nolan's 2026 film is second-hand. This project holds no primary source for that film and, by the author's standing ruling, never will: it does not consult sources it does not hold. Each such claim is marked where it appears, and the adversarial verifier printed at the end of this page was run partly to check that discipline. Its report is printed unedited, including where it is wrong about this project.

Editorial note, 4 August 2026. This page is a dated transcript and its body is not altered. Where the session speaks of nine of the eighteen refusals as a count, that count has since been demoted to an illustration — the session's own challenge produced the demotion, and the coding sheet that records why is published in the concordance beside the pre-registered Book 22 test row it commissioned.

Joint session 09 — the three durations

28 July 2026. The build panel and the literary masters, sitting together for the second time, on a single question: what has been built, read through the book, through Homer, and through the newest Odyssey.

Chair: BRAUDEL. The chair does not vote. The author ratifies.

The advisors and masters are constructed critical lenses named for people whose working methods they abstract. They are not those people's opinions and nothing here should be quoted as such. This notice is not a formality: two of the four works discussed below are in copyright, one of them has not been seen by anybody on this panel, and the whole session runs on a project whose first rule is that nothing is fabricated.

Why this chair, and why this shape

The chair was chosen for a reason that is also the session's architecture, and the reason should be visible rather than tasteful.

The topic holds three clocks at once. Homer's return-story has been running for something near twenty-seven centuries. Joyce's day is a single Thursday in 1904, written across seven years and printed in 1922. Nolan's film and one man's eleven-day first reading are both 2026, and both are still warm. A panel that discusses these on one scale will produce a comparison of quality, which is the least interesting thing available. Braudel's method is the only one on this roster that treats different clock-speeds as different objects rather than as a ranking — the event, the conjuncture, the long duration — so the session is built on it and every voice is required to speak to all three.

The chair's conflict, disclosed. Braudel owns dimensions 1, 3 and 13 in this project — the object museum, the governed city, the constraint model. He does not own 9, 14 or 16, which are the three under discussion. He does not vote in any case. It is recorded here because a chair who selects the architecture that flatters his own method should say so out loud.

The standing rule of this session

Attack across the aisle. Every opening had to name and break a member of the other body. A colleague did not count. The rule is inherited from joint session 08, where it worked: nine sessions of the chair reading the data model had missed what two advisors found by opening the artifact and reading it.

The rule exists because these two bodies are composed to judge different things and have every incentive to leave each other alone. The build panel judges whether an artifact works — whether the link resolves, whether a first reader can use it, whether the diagram earns its space. The masters judge whether the operation is worth performing at all. Left to themselves they produce two reports, not one session.

Who sat, and who did not

Build panel (7 of 9): Senn, Shneiderman, Kenner, Kiberd, Ginzburg, Warburg, Eames. Literary masters (6): Borges, Atwood, Saramago, Cortázar, Pamuk, Galeano. Chair: Braudel.

Not called: NELSON (build panel) — and this is recorded rather than passed over, because the panel's own operating rule says that passive availability is insufficient: an advisor who is not called does not contribute. Nelson owns cross-referencing and the hyperlinked-PDF register. The session below turns repeatedly on what points at what — the apparatus and the book, the atlas and the day, the four Odysseys and each other — and the voice whose entire method is what does this point at, and what points back was absent from it. That is a hole in this session, not a neutrality.

Also not called: Plath, Dick and Vian of the extended masters. The roster was cut to keep the session readable; the cut is an editorial act by the chair and is disclosed as one.

The evidence discipline this session ran under

This is the part a reader should check the session against, because it is the part most likely to have been quietly broken.

On Nolan's 2026 film — binding, and stricter than it looks. Every claim this project holds about that film is second-hand and permanently so. The author ruled (catch C-31) that the project does not consult primary sources it does not hold: no screening, no script, no library visit. Four open questions about the film are closed as permanently unresolved rather than left pending, and they are kept with their last known state because an abandoned path is shown, not hidden (anchor A4). What the project holds — Sinon imported from Virgil in place of Achilles, the hanging of the slave girls elided onto a single betrayer, an ending in which Odysseus is exiled rather than celebrated, Calypso's arc as erasure-then-release — is reported, by journalism, about a film nobody here has seen. A stated fidelity marker of roughly seventy per cent was deliberately not moved when three low-edge departures were reported late, on the ground that second-hand evidence does not earn that precision.

Every voice was instructed accordingly, and an adversarial verifier was run afterwards with the sole job of finding where the session broke that rule. Its report is printed in full below, and it was not edited to be kinder.

On Joyce. Anchor A19: his words appear as printed or not at all, and no dimension may characterise a passage it does not also show. The voices below were instructed to reference rather than quote, and never to invent a quotation.

On Wilson. In copyright (W. W. Norton), RIGHTS.md clause 4. Two short attributed lines were admitted to the session, both already verified in this project's record.

The chair's record

Seventeen voices, three durations, one adversarial verifier run afterwards with the sole job of breaking what the session had just produced. What follows is what the chair takes from it. The openings and the cross-examinations are printed in full below, unedited; where the chair disagrees with them he says so here rather than trimming them there.

The finding this session was worth holding for

PAMUK put a Homeric passage at the exact coordinate dimension 9 grades, and it reverses the dimension's direction without deflating it.

Dimension 9's headline is that where Homer settles with an act, Ulysses settles with an account — that Joyce relocates Homeric power and, in nine of eighteen rows, discharges the Homeric obligation by paying it off. The implied contrast is that settlement-by-payment is Joyce's invention, the modern substitute for the heroic deed.

It is not. In Book 22, with Antinous dead, Eurymachus offers restitution — he offers to pay, to bring compensation for everything consumed, and to add more on top. Odysseus declines it and takes the act instead. The transactional settlement is present in Homer as an explicitly refused option.

That changes what dimension 9 has found. Joyce did not invent the ledger as an alternative to the sword; he took the branch Homer's own text put on the table and declined. The refusal is therefore not a modern secularisation but the taking-up of a road the poem itself surveyed and rejected — which is a stronger claim, a more interesting one, and one that a single printing can refute. The chair records it as the session's best contribution and as work: the row must be run and shown.

The live defect, found by the session and verified before it was touched

SENN demanded that dimension 9's formula — the book's way of refusing Homer stated as "We're square" — produce its episode, its speaker and its page, or lose its quotation marks before morning. He suspected the project of quoting itself.

He was wrong about the project and right about the rule. The phrase is Joyce's. It occurs once in the held 1922 text, in episode 15, and it is Bloom's: he lays a half sovereign on the table between Bella and Florry, takes up the poundnote, and says "Three times ten. We're square."

So the quotation marks are correct, and the finding is better than the project had realised — the book's formula for refusing Homer is spoken by an advertisement canvasser settling a brothel bill. But anchor A19 has a second clause, and it bites: no dimension may characterise a passage it does not also show. Dimension 9 characterises this one — it is the culminating sentence of the dimension's entire argument — and shows it nowhere. Neither the artifact nor the dimension record carries the words, the speaker or the episode.

That is a live A19 breach on a published surface, found by a panel and verified against the held text before anything was written down. It is logged as a catch.

And the passage carries a complication the project should not smooth away. Four lines after Bloom says We're square, he crosses the room and gives Stephen his money back — This is yours. The settlement is undone almost as soon as it is made. A dimension whose thesis is that Ulysses closes Homeric obligations with an account must reckon with the fact that its own proof-text is a payment immediately followed by a return. Under A4 that is shown, not resolved.

What the three durations actually yielded

LONGUE DURÉE. The session's most-repeated claim was that Homer's recognition tokens cannot be transferred or forged — the scar, the bow, the bed — and that this is what Joyce breaks when he gives Bloom a bed that was bought, shipped and disputed. Seven voices arrived at it independently.

The chair does not accept it, and the reason is at the table. Senn's counter-instance is the poem's own last recognition: Odysseus proves himself to Laertes in the orchard by reciting the trees — an itemised schedule of holdings, transferable and recitable by anyone who was told. If the poem's final proof of identity is an inventory, then "the token cannot be an account" fails at the terminus of the source text. A universal drawn from three instances and refuted by a fourth in the same room is not a finding, and Ginzburg's own rule applies to his colleagues: seven arrivals at one conclusion are one argument counted seven times when they share a source, and here the source is the same three tokens.

CONJONCTURE. The sharpest structural criticism came from inside the build panel and it is about method, not taste. The Object Museum's criterion is followable hand to hand. Paper is addressed, dated, initialled and passed; coin is fungible and therefore unfollowable in principle, because one shilling is any shilling. So "the day's paper economy outweighs its coin" may be a restatement of "I counted what can be counted" wearing the clothes of a discovery about Dublin. The honest headline is not the day is made of paper but the day's followable residue is made of paper.

The same blade turns on dimension 3. Galeano put it best: "rule as ambient hum" is a fact about two men's route through the city, not a fact about Dublin. The protagonists never enter a tenement, a barracks or a courtroom — so the dimension has measured where rule was visible to Bloom and Stephen, and named the result a property of the state.

Neither observation destroys its dimension. Both mean the dimension is currently stating a fact about the book plus the method in the voice of a fact about the book.

ÉVÉNEMENT. Eleven of the thirteen openings attacked the same thing: the Nolan column of dimension 16. The attack is unanimous, which by this panel's standing rule is a signal to test the question rather than trust the answer — so the chair tested it, and it survives.

Dimension 16 plots four works on two axes. Three are held texts. One is a film nobody on this panel has seen, known only by journalism, and permanently so by the author's own standing ruling. The project marks this honestly in prose. It then plots it at a position on a measured axis, at roughly seventy per cent. A position on a scale outranks the caption that disowns it — that is Shneiderman's formulation and the chair adopts it. The disclosure is real and it is losing to the geometry.

The remedy the session converged on is unusually good because it is already the project's own vocabulary: dimension 14 draws its twenty-six empty slots as keylines at full footprint, never closed up. Cortázar, Warburg and Eames independently proposed drawing the second-hand column the same way — a hollow keyline where a plotted point would be. The project already has a visual grammar for "this is a real slot and it is not filled", and it is not using it on the one axis where the gap is epistemic rather than material.

SARAMAGO dissented and the chair keeps it at full strength: keylining is insufficient, because a hollow shape at a position still asserts the position. He would strip the column of any axis placement until the project holds the film. The chair notes that this is the more consistent reading of the project's own rule and does not resolve it.

The verifier's report, and two corrections to it

The verifier was run with its default set to REFUTED and its report is printed below unedited, including where it is wrong about this project. It found one systemic error the chair accepts in full, and made two accusations the chair must correct — and the correction is more instructive than the accusations.

Accepted, and it is serious. "I refuse to grant these girls a clean death" is, in the poem, Telemachus speaking — it is Wilson's rendering of a killer's stated cruelty, not a translator's statement about her own practice. Seven voices escalated a rendering into an authorship, and Atwood, Pamuk, Borges and Kenner did so explicitly ("in her own words", "signing her arrangement in the first person"). Kenner committed it in the same paragraph in which he stated the rule that forbids it. This is precisely the error the project already caught once in C-30 — a policy lifted from the Translator's Note and dropped onto the scene — performed in the opposite direction, and no voice at the table caught it.

Corrected: the verifier's two "contradicts the project" findings are both false, and both are my fault as chair. It flagged Shneiderman's count — 303 question marks in Ithaca, none at all in Penelope — as an invented asset, and flagged the project's rename as unsupported. Both are real. The count is catches C-69 and C-73, verified independently twice and confirmed to the digit; the rename was ruled by the author on 27 July and is live in data/site.json.

The verifier was measuring against the brief, and the brief was short — because I wrote it. That is worth more than either accusation: a verification run against an incomplete reference set produces false positives that are indistinguishable, in form, from real ones. Every one of its findings arrives with the same confidence. This is the shape of learning L-36 — a check that measures against a set it assumes is complete cannot see that the set is short — and the session produced a live instance of it in its own machinery.

The chair also concedes an error of his own, which the verifier caught: a tally of "two subtractions against eleven additions", used to land the point that a body asked about too much apparatus answered with more apparatus. At least five further positions contain explicit removals. The punchline was manufactured by an undercount and is withdrawn.

Where the bodies split, and the chair does not resolve it

Whether the apparatus serves the book or competes with it. The rename discharges Atwood's disclosure charge — the work now says what it is. Kiberd's passability charge is untouched: a first reader still meets anchor codes and dimension numbers before meeting Bloom.
Whether dimension 9 is a reading or an inventory. It is the one dimension that makes an attackable claim rather than reporting a distribution, and it is therefore the only one that earns the apparatus built around it. Several voices want that stated as a published standard.
Whether "transacted" was defined before or after the formula was written. Borges's question, and the chair cannot answer it from the record. If the nine rows were graded after We're square was coined, the count is an illustration and not evidence.
Whether the reading log's design makes the competition question permanently unanswerable. Dimension 15 returns nothing by ratified rule, so the project cannot know whether anyone reads the book after the apparatus. That is a permanent limit, not a pending question.
Saramago on the conditions of the reading. Eleven days of uninterrupted attention is itself a material fact about who can produce this kind of record, and it is nowhere on the page.

What the author must rule, and only the author

Whether dimension 16 loses its fourth case. Strip the Nolan column to a keyline, strip it of axis position entirely (Saramago), or leave it and accept that geometry beats caption.
Whether the Book 22 row is pre-registered and run as an explicit test of dimension 9's "nothing is deleted" law — published whatever it returns, including if it breaks the law. Atwood staked her thesis on a test that can destroy it; the chair regards that as the best method offered all session.
Whether dimension 2 gains a separately-denominated register of women who are drawn and never named — the class L-38 was written for, applied to people instead of coastline. Atwood and Galeano both asked for it, from opposite directions.
Whether every dimension must publish its inclusion rule and its rejected candidates, so a reader can tell a finding from a selection effect. Ginzburg, Borges and Pamuk converged here.
What the project owes the twelve hanged women, given it now holds three treatments of the same slaughter — Homer's, Wilson's refusal to look away, and a 2026 elision it cannot verify.

The openings

Thirteen voices, each required to speak to all three durations and to name and break a member of the other body. Printed in full and in roster order, unedited. Where a voice is wrong, the chair's record above says so and the opening still stands as delivered — a session record that quietly fixes its participants is not a record.

SENN

(1) LONGUE DURÉE — the Odyssey

I am asked to assent to a formula. "Where Homer settles with an act, Ulysses settles with an account." It is a good sentence. That is the first thing wrong with it.

Take the last recognition in Homer. Odysseus in the orchard, proving himself to Laertes, does not act. He recites: the pear trees, the apple trees, the figs, the rows of vines his father gave him when he was a boy — a count, itemised, with quantities. (Thirteen, ten, forty, fifty rows, if my memory of Book 24 is honest; check it against the Greek before you print it, and print the lines.) That is not an act. That is an inventory. The poem's final proof of identity is a schedule of holdings. So dimension 9's contrast does not hold at the terminus of the source text, and a formula that fails at the terminus of the source is not a formula, it is a slogan.

And the bed. Homer's central token is the one object that cannot be relocated — the bed cut from a living olive stump, rooted in the floor. It is proof precisely because it can neither be moved nor bought. Now put dimension 9's two columns beside it. The project says Joyce RELOCATES what Homer keeps in place, and separately that in nine cases he TRANSACTS what Homer performs. At the bed those are not two findings. They are one. Joyce's bed was carried, shipped, and purchased, and the book gives it a provenance — and gives it twice, in two mouths, and the two accounts do not agree. Under A4 the project must show that disagreement and not smooth it. Under A19 it must show the passages. Dimension 9 currently shows the reader a table and asks him to take the reassignment on trust.

Then the harder thing. The paper claims Joyce sends power downward into a person who has none, or outward into what is not a person, never at human scale. But look at Homer's own Ithaca. The first to know Odysseus is a dog on a dungheap — not a person at all. The second is Eurycleia, a slave, who knows him by the scar, by touch, and is immediately throttled into silence by the man she has just recognised. Recognition in Homer is already assigned downward and outward, and then confiscated. If that vector is Homer's, it cannot also be Joyce's refusal of Homer. The finding may be true of the divine machinery and false as an account of what distinguishes Joyce. That is a category slide, and it is exactly L-17: the error is not the claim, it is the qualifier "never."

Borges. He is the master across the aisle who will love dimension 9 for the wrong reason, and I will break him now. His position — from the essay on the Homeric versions, and from the inventory of metaphors — is that a text is one draft among possible drafts, that the repertoire of stories is small and fixed, and that difference between versions is a matter of attribution and reading, not of substance. On that view Joyce's "relocation" is one more permutation of an eternal repertoire, beautiful and weightless. The method fails here for a mechanical reason: it operates on schemas, on lines reduced to "time is a river." Reduce Ithaca's bed to its schema and you get "proof of marriage." Reduce Joyce's to its schema and you get "proof of marriage." The entire finding disappears in the reduction, because the finding lives in the second-hand purchase, the disputed provenance, the jingling — in exactly the residue Borges's instrument is built to discard. A repertoire theory cannot tell subtraction from reassignment; both are simply another version. It is the wrong tool for the only question this session has.

(2) CONJONCTURE — 1904 and 1922

Fact about the book, or artefact of the method? Mostly the second, and the project should say so louder.

The Object Museum's criterion is followable hand to hand. Paper is addressed, dated, printed, initialled, and passed — it is the class of object that survives the criterion. Coin is fungible and therefore unfollowable in principle: one shilling is any shilling. So "the day's paper economy outweighs its coin" may be a restatement of "I counted only what can be counted," dressed as a discovery about Dublin. Same disease at the latchkey: an inventory admits an absence only when someone names it. The key is named because Stephen keeps naming it. The absences nobody names — L-38 exactly — are not merely missing from the museum; their absence is invisible to the museum's own audit. Thirty-two objects, and the honest headline is not the day is made of paper but the day's followable residue is made of paper.

And the uncomfortable one, which I put as a demand. Dimension 9 states the book's formula for refusing Homer as "We're square." In quotation marks. Where is that printed? Whose mouth, which episode, which page of 1922? If it is Joyce's, print it and its surround. If it is the project's gloss, take the quotation marks off tonight. A19 says Joyce's words as printed or not at all, and a paraphrase wearing quotation marks is the precise failure this project claims to police. It is also what I have been doing all evening — characterising passages I have not shown you, under a house rule that forbids me showing them at length. Notice that this roundtable's register is structurally non-compliant with the project's own first rule. That is not a decorative caveat.

(3) ÉVÉNEMENT — 2026

Nolan first, and briefly, because the discipline is right. Every item the project holds about the 2026 film is REPORTED and SECOND-HAND and stays so: Sinon in place of Achilles; the hanging displaced onto a single betrayer; exile rather than celebration; Calypso's lotus as erasure-then-release. Reported. Not verified. I add only this, and it is about Joyce, not the film: if a modern Odysseus is given forgetting as his mechanism, that is the one instrument Joyce refuses him. Bloom's day is a failure of forgetting — the involuntary return of Rudy, of the letter, of the funeral — and dimension 9 should note that the book's relocations run through a man who cannot put anything down. Whether Nolan has mentioned Joyce is unevidenced and should stay unevidenced.

Wilson is where the project is strongest, because it caught itself: Book 22 has these girls where the Greek has dmoai, so the hanging is the wrong proof text for "slave." What it does prove is the other thing — "I refuse to grant these girls a clean death." That refusal to look away is the same nerve as Eurycleia's silencing, 2,700 years apart, and it belongs in dimension 1's long duration, not in a 2026 footnote.

n=1: the reading was an event; the record is a structure; and dimension 11 has a hole in it. Zero placed questions at Proteus is not evidence of ease. A question-log cannot distinguish transparent from opaque enough that no question formed. The reader who is lost does not always know what to ask. The engagement reading may be right — but the same man logged the questions, named the survival language, and later reinterpreted the curve, and no instrument in the project can falsify him.

MY POSITION: dimension 9 must print, beside every one of the eighteen rows, the Homeric occasion and the Joycean passage as they stand, and any formula it states in quotation marks — "We're square" above all — must carry its episode, speaker and page by publication, or lose the quotation marks and be demoted to the project's own gloss.

SHNEIDERMAN

I. LONGUE DUREE — the Odyssey

Dimension 9 has found something more mechanical than it sounds, and should say it in mechanical words. If Joyce never deletes a Homeric element but reassigns the power inside it, the two books share a key. Eighteen rows, the same occasions, different values. Deletion breaks a join; relocation preserves it. That is not a finding about Joyce's piety. It is a finding about what has kept the Odyssey alive for 2,700 years: the poem is addressable. It has slots, and every age has refilled them rather than removed them.

Here is what the brief does not say, and what I want as a row. Homer's recognitions are an authentication protocol running three separate factors on one identity. The scar: involuntary, borne on the body, read by a slave with no standing to act on it. The bow: a capability test, public, provable by performance. The bed: a secret held by two parties, disclosed by a lie only the right person will correct. Something you are, something you can do, something you know. And the order is not casual: the weakest credential is found first, by the person with least power; the strongest is withheld until the one acceptance that constitutes the return. The household is not restored by the killing. It is restored when the last factor verifies.

Run dimension 9's rule on that. Ulysses keeps all three credentials and returns no recognition. Ithaca is a query interface, and the project's own count — 303 question marks there, none at all in Penelope — is evidence it has not yet spent: a catechism run over two men, matching them on trivia and difference, then failing to open a session. The lodging is offered and declined. The last secret, the one Homer reserves for the bed, is held on the inside by one person, never tested, in an episode that asks nothing. So the thesis extends: recognition itself is relocated — downward into an inventory of matched trivia, outward into a bed that keeps its secret without a proof. Recognition is the human-scale mechanism, the one Homer keeps at eye level. Joyce moves that too.

And "we're square" does real work here. An act is single, irreversible, unauditable; an account is a log, itemised and re-openable. Ulysses converts Homer's events into records — and a book that has tabulated itself will reward tabulation. That is where my discomfort starts.

II. CONJONCTURE — 1904 and 1922

Fact about the book, or artefact of method? There is a test: a fact survives re-sorting; an artefact changes when you change the inclusion rule.

Paper outweighing coin is a count, and a count is downstream of what was allowed to be an object. Thirty-two were followed; the day holds hundreds. Count mentions rather than itineraries and nobody here can say whether paper still wins. The itineraries survive — the soap's route, the latchkey ruling eighteen hours from inside another pair of trousers — under any rule that admits the item at all. Inside one dimension, the headline is an artefact and the spine is a fact, and the site gives them equal weight.

Governed City is the sturdiest thing built, and the project has not noticed why. Its finding is an absence checked by exhaustion across eighteen episodes: no tenement, no barracks, no courtroom. Absence claims are falsifiable by one counterexample and verifiable by enumeration; presence counts are only as good as their frame. Notice where the good results land: the museum's governing object is an absence, the Atlas's twenty-six keylines are absences, the concordance's subject is refusal. This project's best data are holes.

And it has no interface for holes. You cannot tap an empty keyline. No filter returns "the things that never happened." Overview, zoom, filter, details on demand assumes the data exist; here the reader must be able to select the negative space as a first-class result. That is the site's central design problem and nobody has named it.

The uncomfortable one. Women's Dublin is the most quotable finding in the build and the most fragile. Interiority running opposite to independence is a two-variable relation over sixteen points; recode Molly or Milly once and the line moves. Worse, L-38 already knows the failure — a rule that admits by naming cannot reach subjects drawn and never named, and the unnamed women of that day are precisely those with least independence and least interiority. They belong at the origin. Sixteen named women is not a population, it is what the admission rule could see. Put the sample frame inside the same visual field: a scatter of sixteen points reads as a law, and will be screenshotted without its caveat.

Borges would say none of this matters — that a catalogue equal to its subject is a joke against catalogues, that the empty keylines are beautiful precisely because they are unusable. Break it: the Aleph is an anti-interface. Simultaneity without ordering is not revelation, it is noise, and Borges knows it, because the only way he can narrate the thing is to serialise it into a list and walk it item by item. He applies overview-then-detail in the act of denying the discipline exists. An empty keyline is not a virtue because it is empty. It is a virtue only if a reader can reach it, count it, and see what class of thing is missing.

III. EVENEMENT — 2026

Everything held about Nolan's 2026 film is REPORTED and SECOND-HAND, unverified and permanently so under C-31: the Sinon substitution, the hangings elided onto a single betrayer, the exile ending, the Calypso arc as erasure-then-release. All reported. I add nothing.

My objection is to a number. The stated fidelity marker of about seventy per cent was rightly not moved when three low-edge departures were reported late. It should not be displayed at all. L-17 says the error is the qualifier — and a qualifier is text, while a number on an axis is a position, and the eye reads position before prose. A caption disowning a coordinate loses to the coordinate. Draw the band as wide as the axis, or replace the point with a labelled region.

Wilson is the opposite case and the project's best hour: it went to the scene it wanted to cite, found "these girls" where the Greek has dmoai, and published against its own proof text. The translation is an event; what it exposes — centuries of English managing one word — is a structure. Her ground is that a softer word "would imply that she was free," and what Book 22 supports is the refusal to look away: "I refuse to grant these girls a clean death."

Which duration is n=1? The reading is an event: eleven days, unrepeatable. The apparatus is a structure — and there is the trap. Dimension 15 keeps entries on the device and returns nothing: a defensible ethical choice and a permanent epistemic ceiling. The project built the one instrument that could turn a difficulty curve into a distribution, then ruled it will never receive a reading. Say that on the page as design, not modesty. And hold the floor: if the log is not fully keyboard-operable, with a visible focus ring and no pointer-only affordance, it is a demo, not an instrument. Headings and table scopes were fixed. Navigation was never audited.

MY POSITION: Build one overview surface before a sixteenth anything — all sixteen dimensions as rows carrying type, status and one line of finding, keyboard-reachable, with a filter whose result set can be the empty slots and the unbuilt dimensions themselves — make it the front door and the first tap from every dimension page, and strip the seventy-per-cent marker off the Nolan axis, because a position on a scale will always outrank the caption that disowns it.

KENNER

I. LONGUE DURÉE — the Odyssey

Dimension 9 has found something real and mis-stated its scope. It reports that Joyce never deletes Homer but relocates the power, landing below the human or beyond it, never at human scale. What moves is not power. It is narrative authority.

Homer keeps his gods at human scale for one technical reason: they speak. Athena can be quoted, in the same register as everyone else in the poem. Joyce's relocations land every time on something that cannot be quoted as a person — on weather, which has no first person; on a gramophone, which is a device for producing a voice with no speaker behind it. That is not a lower rung of power. That is a machine with the exact property Joyce's prose has: voice present, origin absent. The project has catalogued a theology. What it actually recorded was the withdrawal of the quotable speaker.

And the account finding — nine of eighteen refusals discharged rather than performed — is not the reversal of Homer the project takes it for. Ask who narrates Homer's marvels. The Cyclops, Circe, the dead and the Sirens reach us in Odysseus's own first person, told to the Phaeacians, by a man who needs a ship, and he is paid for the telling in treasure and passage. The most famous supernatural stretch of the poem is already a transaction: a story exchanged for transport home, delivered by an interested party to the only audience that can grant it. The 2,700-year duration does not run act, then account. The account was always inside the act. Joyce is the first to print the receipt.

Add the recognitions. Scar, bed, bow: Homer's household is restored by proof procedure, by tokens tested. Ithaca's catechism is not a departure from that regime, it is that regime with the witness removed. Homer's tokens are recognised by persons who love the man. Joyce's are recognised by a procedure that loves nobody. That, and not secularisation, is the refusal.

II. CONJONCTURE — 1904, 1922

Now the uncomfortable part, and I intend it. Three Type I inventories, three findings, and all three are facts about narration wearing the costume of facts about Dublin.

Paper outweighs coin. Whose attention produced that count? The two consciousnesses the book lends you belong to an advertisement canvasser and a writer. Print is one man's trade and the other's ambition. A museum assembled from what those two notice returns a paper economy on any day, in any year, in any city. The finding as published is a claim about 1904. The evidence supports a claim about attention.

Dublin is governed everywhere and administered nowhere the reader is taken. The load-bearing clause is where the reader is taken, and that is a fact about the itinerary of a camera bolted to two men with no business in a barracks. But there is a genuine finding underneath, and the project walked past it. Rule becomes an event exactly twice, and both times — the project notes this and does not use it — the book has left realism. Cyclops is narrated by an unnamed drinker whose account is repeatedly swamped by registers belonging to nobody. Circe has no narrator at all, only stage directions. Institution enters Ulysses only where the narration stops being a person's. That is a fact about the book. "Administered nowhere" is a fact about the route.

Interiority running opposite to independence: Molly's inside is a chapter allocation. Milly is paper because Mullingar is offstage. Move the compositional facts and the correlation moves with them.

Which takes me across the aisle. Borges. His position here is predictable and wrong: the catalogue is itself a literary form, enumeration is the shape the infinite takes in prose, and fretting about the voice that produced an item is pedantry, since all narrators are finally one narrator. Break it on his own best page. The Aleph works because a named fabulist stops in the middle of his list to tell you the enumeration is a lie — successive words for a simultaneous sight. Borges pays for his catalogue with a confession of order. The Object Museum pays with a percentage. Bolt a truth claim to Borges's form and you get the seduction of the list plus the authority of a census, the one combination no reader can audit. And "all narrators are one" would abolish the finest thing this project owns: dimension 9's demonstration that Joyce's refusals land at specific and different narrative altitudes. If the narrators were one, there would be no altitudes to measure.

The project's defence is already in its own record. The Museum's most governing object is an absence, a latchkey in the other trousers. L-38: a rule that admits by naming cannot reach what is drawn and never named. The latchkey got in only because a man keeps reaching for it. That is the method finding its own edge, and it should be promoted from a pleasing result to a stated limit.

III. ÉVÉNEMENT — 2026

Dimension 16 sets on one comparison two objects whose evidence has incompatible narrators.

Wilson signs hers. A Translator's Note, first person, stating that "maid" or "domestic servant" "would imply that she was free," and so she has "often used slave." Checkable — and the project checked it, and caught the citation pointing at the wrong scene, since Book 22 gives "these girls" where the Greek has dmoai. Best single piece of work in the log. Note why the error was possible at all: a note is a voice, a passage is a voice, and they are not the same voice. Lift the fact out of the Note, drop it on the scene, and the truth value changes. What Book 22 does support is the other half — "I refuse to grant these girls a clean death" — a claim about looking, not about lexis, and it survives.

Against that: everything the project holds about Nolan's 2026 film is REPORTED and SECOND-HAND, and under C-31 permanently so. The Sinon substitution, the displaced hanging, the exile ending, Calypso's erasure-then-release — reported, none verified. Whether Nolan has mentioned Joyce is unevidenced. And sitting on that side of the comparison is a fidelity marker of about seventy per cent. The project was right not to move it on late second-hand reports. It is still a number with no narrator. Nobody here can say who computed it, from what, against which text. L-17 says the error is almost never the claim but the qualifier. Here the error is the decimal resolution: a percentage asserts that a measurement was made, and none was.

Then the eleven days. The difficulty curve is not a curve of difficulty, and not only a curve of engagement. Run it against one variable: can the reader locate a narrator? Episodes 2 and 3 drew zero placed questions while being the front half's most demanding prose, because the voice is stably fastened to Stephen. The hardest sustained stretch, Sirens through Oxen, is precisely where the narration stops belonging to anyone. The three most-questioned episodes are a script, a catechism and an unpunctuated monologue — again nobody narrating, but by then the reader had learned the game, so the same condition produced delight rather than distress. One variable, two signs. There are 245 placed questions; this is testable tomorrow.

Event or structure. The reading was an event. But dimension 15 returns nothing by design, which means the project has built an instrument guaranteeing that n stays 1 for ever. That is the right call on privacy and it is also a decision that this evidence base can never grow. Say it in the document's own voice under A10a, or the n=1 disclosure is only half made.

MY POSITION: before a tenth dimension is built, every Type I inventory must carry a stated narrator condition — whose attention produced this count, and what the count looks like when taken only from the stretches where nobody is narrating — and the ~70 per cent Nolan fidelity marker must be demoted from a number to a sentence, because a figure with no narrator is not a finding.

KIBERD

I. LONGUE DURÉE — the Odyssey

Dimension 9 has found something bigger than it says, and it has stated it in a register that will keep ordinary readers out of it. Let me put the finding where a first reader can use it.

The claim is: nothing Homeric is deleted, everything is reassigned, and in nine of eighteen the reassignment runs through an account rather than an act. The panel will be tempted to read that as diminishment — the heroic age settled with a bow, we settle with a receipt, how sad we are. That reading is sentimental and I think it is wrong. What it actually establishes about 2,700 years is a fact about instruments.

Look at how Homer certifies a person. The scar. The bed built into the rooted olive. The bow no other man can string. Every one of Homer's recognition tokens has the same property, and it is not a poetic property, it is a material one: they cannot be transferred and they cannot be forged. That is the recognition technology available to a world without paper. Recognition in the Odyssey has to be carried on the body or built into the house, because there is no other medium that will hold it.

Now take dimension 1 seriously, because dimension 1 and dimension 9 are the same finding seen twice and nobody has said so. The day's economy is paper over coin — post, print, pasteboard, more than a third of the museum. Paper is the instrument Homer's world lacked. And paper's whole usefulness is the exact inverse of the scar: it is transferable, it is deferrable, it works in the absence of the body, and for that same reason it can be forged. So when dimension 9 reports that Homeric power relocates "beyond the human, into something that is not a person" — the gramophone, the weather, the schedule of remedies — the medium of that relocation has a name, and it is the ledger. An obligation that can be priced and itemised is an obligation that has left the body. Paper is what "not a person" is made of in 1904.

Here is the consequence the brief does not draw. A city that has swapped the act for the account has, by the same swap, lost the technology of recognition. This is not thematic pessimism; it is mechanical. Bloom conducts a correspondence under a name that cannot be produced in a body — paper carries the intimacy precisely because it cannot certify the man. The unpaid fourpence on the soap governs by not being settled, exactly as the latchkey governs by being absent: the object with power is the one still outstanding. That is the structure of credit, and credit is anti-recognition. And then observe where the one surviving Homeric token ends up. The recollection on Howth in the last pages is bodily, untransferable, un-notarised, addressed to nobody, filed nowhere — and dimension 2 tells us it sits inside the least mobile person in the book, forty pages without leaving the bed. The book keeps one unforgeable token and puts it in the only consciousness with no independence to spend it with. That is what dimensions 1, 2 and 9 say jointly, and none of them says it alone.

II. CONJONCTURE — 1904 and 1922

Across the aisle: Borges. In 1925 he announced himself the first traveller from the Hispanic world to reach Joyce's book, admitted he had not cut a path through all its pages, and claimed he knew it anyway — the way one legitimately knows a city without ever being granted the intimacy of all its streets. That sentence is the founding charter of every gatekeeper this book has ever suffered. It says the plan suffices. It says the design is the possession and the walking is optional. It is elegant and it is a lie, and this project has the receipt.

Dimension 11 is that receipt. The plan of the city — the received map, the folklore, the thing you can know without reading — says the reader breaks at Proteus. Episodes 2 and 3 drew zero placed questions. The actual walking produced a break at Sirens through Oxen, 11 to 14, where the record's only survival language appears. Borges's method predicts Proteus every time, because Proteus is what the diagram of difficulty looks like. Only a man with 290 logged questions and eleven days finds 11–14. And the second half of that finding buries him: the three most-questioned episodes are the three the reader loved loudest and read fastest. A curve of engagement wearing the costume of difficulty is invisible from the aerial view; you can only see it from inside a chair, at night, with the book open. Borges could not have found it, and no amount of shape-admiration will.

Are the found structures facts about the book or artefacts of the method? Some are artefacts and the project should say so louder. "Rule is an ambient hum" (dimension 3) is partly a fact about whose day was tracked: two men with no summons, no rent arrears, no arrest. Track Denis Breen through his libel action and administration becomes a scene. The finding is real but it is a finding about the protagonists' exposure, not about Dublin, and the current phrasing lets it drift.

Now the discomfort, and it is mine to give. This project is drifting toward the academy through its own honesty. Nine dimensions, four registers, 97 catches, 38 learnings, and a private grammar — A2, A10b, A19, L-17, L-38 — published on the site. A first reader must learn the project's rulebook before she reaches Joyce's. L-17 says the error is almost never the claim, it is the qualifier; the project has policed its qualifiers so thoroughly that the qualifier apparatus is now larger and louder than any claim it protects. And dimension 14: 26 empty keylines at full footprint, caption-free. In a gallery that reads as scrupulousness. On a laptop it reads as a page that failed to load. An honest absence that a first reader mistakes for a broken site is not honest — it is illegible.

III. ÉVÉNEMENT — 2026

The event/structure question answers itself if you look at the evidence, not the calendar. Wilson 2017 is verified to the page — including the finding that Book 22, the very passage the project cites, gives "these girls" where the Greek has dmoai, so anyone citing it as the proof text for "slave" is citing the wrong scene. What it does support is that she refuses to look away: "I refuse to grant these girls a clean death." That is a structure, checkable by anyone with two printings.

The Nolan half is not a structure and it is not an event this project attended. Everything held — Sinon in place of Achilles, the elided hanging displaced onto one betrayer, exile rather than celebration, Calypso as erasure-then-release — is REPORTED, second-hand, and permanently so by the author's own ruling. Whether Nolan has ever mentioned Joyce is unevidenced. Here is the accessibility failure: dimension 16 presents four columns. A table is itself a claim of comparability. Three columns rest on print a reader can check and one rests on hearsay, and the layout asserts a parity the evidence does not hold. The first reader will not read the provenance note. She will read the grid.

And the n=1? It belongs to the third duration, and that is its dignity, not its weakness. One man, eleven days, sixteen sessions, his words as spoken. It becomes second-duration fraud only if it starts speaking in the passive voice of scholarship — which the anchor codes are already teaching it to do.

MY POSITION: before a tenth dimension is built, the project must ship a plain first-reader entrance — one page, no anchor codes, naming the day, the two men, the missing key, routing to three dimensions only — and must break the false symmetry of dimension 16 by giving the Nolan column a visibly different shape on the page, not a footnote, so that no reader can mistake a rumour for a measurement.

GINZBURG

I. LONGUE DURÉE — the Odyssey

Take first what dimension 9 actually establishes, because it is more than the panel thinks and less than it says.

It establishes a negative over a closed set. Eighteen positions, eighteen episodes, no sampling — the denominator is the whole book. At none of the eighteen is the Homeric occasion removed. Negative findings over closed sets are the most durable thing a small corpus can yield, and this one is checkable by anyone with the 1922 text: one clean deletion anywhere kills it. That is a real finding and it should be the load-bearing wall.

What it does not establish is a mechanism. "Relocation" describes eighteen outcomes; it does not demonstrate a procedure. And the nine-of-eighteen is where I stop the table. Nine of eighteen is exactly one half, arrived at by a category — "transacted" — that the grader invented after reading, applied to rows the same grader drew. Unless the coding rule for transacted was written down before the eighteen were graded, that nine is a number about the grader's degrees of freedom, not about Joyce. I do not say it is false. I say the project cannot currently tell the difference, and neither can its reader.

Now the thing worth chasing. Homer's recognition tokens are almost all non-transferable: the scar carried on the body, the bow no other arm can string, the trees Laertes counted out in a fixed orchard, the bed cut from a rooted olive and therefore unmovable by definition. Homer secures a household with objects that cannot change hands — that is the whole evidential logic of the return. Set that beside this project's own object museum: the day's governing object is a latchkey, transferable, governing by absence; its longest itinerary is fourpence of unpaid credit riding a bar of soap from pocket to pocket. That yields a discriminating prediction dimension 9 has not made. If relocation is the mechanism, then at the recognition positions specifically Homer's fixity should be replaced neither by nothing (deletion) nor merely by something lower (relocation as currently stated) but by something portable. Fixity into circulation. If the 1922 text shows that, under A19, dimension 9 gains a mechanism rather than a metaphor: what Joyce displaces is not power but permanence, and "we're square" follows necessarily, because an account is the instrument you reach for when nothing stays put. If the text will not show it, drop the mechanism and keep the negative.

One warning attached to my own suggestion: if dimension 1 and dimension 9 then agree, that is not corroboration. Same reader, same eleven days, same frame. Two documents agreeing are not two witnesses when they share a source. Dimension 16 is already the declared parent of dimension 9's frame. A parent and a child on the same page look like two findings and are one.

II. CONJONCTURE — 1904 and 1922

Facts about the book, or artefacts of the method? Some of each, and the project has the information to say which and has not said it.

The object museum's inclusion rule is followed hand to hand. That rule selects for circulation, and paper circulates by design — post, print and pasteboard exist in order to move. So a set assembled by tracking movement over-recruits paper before Joyce contributes anything. The defensible claim is narrower and better: among the day's circulating objects, paper instruments outnumber coin. The published claim is about the day. That is L-17 exactly — the error is not the claim, it is the qualifier — and the project wrote that learning and then walked past it.

Governed city: thirteen instruments across eighteen episodes. Thirteen of what? Without the rejected candidates the reader cannot audit "ambient hum," which is a claim about frequency in a set the author bounded. But "the protagonists never enter a tenement, a barracks or a courtroom" is a negative over the closed set and survives everything. Keep the second, qualify the first.

Women's Dublin: an anti-correlation between interiority and independence, drawn from sixteen women of whom perhaps four are load-bearing — Molly at one pole, Milly and Mrs Riordan at the other. That is a pattern claimed from three instances, and destroying those is my trade. Show me the middle twelve on the same axes or state it as an observation about four women. And L-38 bites here harder than anywhere: the women who never accumulate an appearance-unit — in a crowd, in a subordinate clause, drawn and never named — are precisely the servants and the poor. A rule that admits by naming cannot reach them, and the class it cannot reach is not random.

Across the aisle: Borges. He will tell this table that the pattern is the reader's gift — that each writer creates his precursors, that Joyce's Homer is retroactive, and therefore that n=1 is no defect because a reading is the only unit there is. I grade the biographical fact high but second-hand: it is reported in the standard accounts that in the 1920s he announced himself the first Hispanic traveller to Joyce's shores while conceding he had not read the whole book, and built a lifetime's relation on it. That is the case against his own position. The doctrine of retroactive precursors is unfalsifiable and costs nothing, which is why it is charming; dimension 9 is worth more than the doctrine precisely because it could have come out the other way. This project should decline Borges's compliment. His partial reading was generative and it was not evidence, and the difference is the only thing on the table today.

III. ÉVÉNEMENT — 2026

Nolan: everything the project holds is REPORTED and SECOND-HAND, permanently — Sinon in place of Achilles, the hangings displaced onto one betrayer, exile rather than celebration, Calypso as erasure-then-release. Closing four questions as unresolved rather than pending is the rarest good decision in the record. But dimension 16 then uses those reports to place a point on two axes, and a placement inherits the evidential grade of its inputs. The ~70% marker is the sharper problem: refusing to move it was correct reasoning that produced the wrong object. A two-figure percentage no evidence can move is not a measurement, it is furniture. Band it or grade it on its face.

Wilson is where the project is at its best and should be made to see why. It went to its own proof text — Book 22, the hanging — and found the word it was citing is not there; the Greek has dmoai and the English has "these girls." What the passage supports is the other thing, that she will not look away: "I refuse to grant these girls a clean death." That is the evidential paradigm working, in public, against the project's own interest. But one verified error does not constitute a verification regime, and here is the uncomfortable part: 97 catches published as an error record function as a trust display, and a confessed error buys more credit than it should. Unless the site states how many citations were spot-checked and which catches came from procedure rather than luck, the error log is a source about the author's diligence, not a measure of the work's accuracy.

On n=1: eleven days is an event, and events are all this method ever has. The project belongs to the third duration and earns entry to the second only where its instrument is logged. The difficulty curve is the model — it names its instrument, so "engagement wearing the costume of difficulty" is a fact about one reader, honestly labelled, and therefore usable. Zero placed questions at episodes 2–3 is a fact about him, not about Proteus. That is fine, if said.

MY POSITION: before a tenth dimension is built, publish for every existing dimension its inclusion rule, the candidates rejected under it, and its parent dimension where it has one — so a visitor can distinguish a finding from a selection effect, and two witnesses from one source counted twice.

WARBURG

I. LONGUE DURÉE — the panel that will not close

Dimension 9 has, without saying so, produced an atlas plate rather than a finding. Eighteen rows, each holding a Homeric occasion beside its Joycean discharge. What the row does — and what a caption would destroy — is put two images in adjacency and let the interval do the work. Athena / crone-with-coin. Shades / gramophone. Bolt / weather glossed by an ad-canvasser. The intervals are not identical, and that is the whole discovery. Read as a plate, the eighteen rows sort into two directions of travel: downward, into a person with no power, and outward, into a mechanism that is not a person at all. The brief already names this. What the brief does not say is that these two directions have a name in the study of images: descent into the type, and survival in the apparatus. A god who becomes a crone is still a figure — the human form persists, drained. A god who becomes a gramophone has left the human form entirely and survives only as function: the dead still speak, but through a horn. That is not one operation with two outcomes. That is two distinct afterlives of one power, and 2,700 years is exactly the duration over which they separate.

Which means the finding about transaction is stronger than the project has stated it. Where Homer settles with an act, an act requires a body to perform it — a bow drawn, a scar shown, a bed described. The account requires no body. It requires only a ledger and two parties who agree on the figure. So the nine transacted refusals are not a subset of the eighteen; they are the terminus of the outward direction. Power that has left the human form has nowhere to land but in the instrument of settlement. This is why the two terminal positions of the book are both transacted: return itself has become the closing of an account.

Now the observation the brief does not contain. Homer's recognition scenes are all bodily and unpurchasable: the scar, the bed rooted in the olive stump, the dog. They cannot be bought because they cannot be transferred. A ledger, by contrast, is the pure form of the transferable. So Joyce's substitution of account for act is not merely a change of settlement-method — it is an attack on the specific thing that made Homeric recognition possible, namely non-transferability. And here is the plate that follows: the object-museum's most governing item is an absence — the latchkey in the other trousers. A key is the one household object that is both a token of belonging and a fully transferable thing. Homer's household admits by scar; Joyce's admits by key; and the key is missing. The Odyssey's recognition-problem and Ulysses' latchkey are the same panel photographed 2,700 years apart. Put them side by side, say nothing, and the room understands. Dimension 1 and dimension 9 are holding two halves of one image and have not yet been laid on the same table.

II. CONJONCTURE — the method's own shadow

Now the uncomfortable part, and I will not soften it, because a caption that flatters is the worst kind of caption.

Three of the four structural findings the chair listed are inventory artefacts wearing the costume of discovery. "Paper over coin" is a result of what counts as an object. "Rule as ambient hum, never a scene" is a result of counting instruments, which are by definition non-scenic — count scenes of authority instead and Cyclops and Circe are not two exceptions but the correct answer to a differently-posed question. Only the interiority-versus-independence inversion in dimension 2 is robust, because it is a correlation between two independently-scored axes rather than a distribution over one. A19 forbids characterising a passage you do not show; there is no equivalent rule forbidding characterising a distribution you did not have to defend. That is the gap. L-38 sees half of it — subjects drawn but never named — but the other half is worse: subjects counted into existence by the act of naming a category. The panel should say plainly which of the nine dimensions produce findings that survive a change of taxonomy, and which do not.

And here I break across the aisle. Borges would arrive at this table delighted, and he would be wrong. He would say — as he effectively did about Wilkins and the Chinese encyclopaedia — that every taxonomy is arbitrary, that the museum of thirty-two objects is one more Celestial Emporium, and that its charm is its arbitrariness; the fact that paper outweighs coin tells you about the librarian, not the library. That position is elegant and it is a dead end, because it purchases immunity from error by surrendering the possibility of finding anything. The answer to Borges is not a better taxonomy. It is the same day counted twice under two incompatible schemes, both published, both refusing to arbitrate — the arrangement A4 already licenses. Arbitrariness ceases to be an alibi the moment you show two arbitrary arrangements side by side and let the reader see which structures hold across both. Borges wrote the objection; he never built the control. The project can.

III. ÉVÉNEMENT — the plate with an empty keyline

Dimension 14 already knows how to handle the Nolan problem, and dimension 16 has not noticed. Twenty-six of fifty-four slots are drawn as empty keylines at full footprint — the absent panel occupying the space it would occupy if filled. That is the correct visual grammar for second-hand evidence, and it is exactly what the four-column comparison lacks.

Everything the project holds about the 2026 Nolan film is reported and second-hand: reportedly Sinon imported from Virgil in place of Achilles; the hanging of the slave girls reportedly elided onto a single betrayer; an ending reportedly of exile rather than celebration; Calypso's arc reported as erasure-then-release. Under C-31 these are permanently second-hand, and four questions are closed as unresolved rather than pending. Yet in the four-way comparison the Nolan column is drawn at the same weight as the Homer, Wilson and Joyce columns. That is a typographic lie. A column built from report should be keylined, not filled — the full footprint, so its absence is measurable, and no ink, so no reader mistakes it for evidence. The ~70% fidelity marker is the tell: it was rightly not moved, but a number that cannot move is not a measurement, it is a rendering of a measurement.

Contrast the Wilson column, which is filled to the edges and self-correcting: the project verified Fagles and found zero instances of "slave" in Books 21–22; then verified Wilson's own Book 22 and found "these girls" where the Greek has dmōai. Her note says the softer words would "imply that she was free"; her line in the hanging is "I refuse to grant these girls a clean death." The proof text is the note, not the scene — and the scene proves something else, that she will not look away. One column carries its own refutation. The other cannot be refuted at all. Those two epistemic states must not be drawn alike.

On n=1: eleven days is an event, and the project should stop apologising for it. But the record — 290 questions, 97 catches, 38 learnings — is not an event; it is an instrument that will still be readable when the reading is forgotten, and instruments belong to the conjoncture. That the three most-questioned episodes are the three most loved is the single most transferable result here, and it is transferable precisely because it is a shape, not a life.

MY POSITION: adopt a single visual rule across all registers — evidence-weight is drawn, not stated — keylining the Nolan column at full footprint exactly as dimension 14 keylines its empty slots, and require every Type I dimension to publish a second, incompatible taxonomy of the same material so the reader can see which structures survive the recount.

EAMES

I. LONGUE DURÉE — the slot, not the story

Dimension 9 is being read as a finding about Joyce. It is a finding about the Odyssey, and about why the Odyssey is still here.

If not one Homeric element is deleted — if every refusal keeps the occasion on the page and only reassigns the power that made it work — then the durable unit of that poem is not its plot and not its theme. It is the slot: a stageable occasion with a power source plugged into it. Athena at the shoulder, the shades given speech, the bolt from the sky. Joyce proves these are slots by swapping the source eighteen times out of eighteen and finding the occasion still stands upright without it. Nothing else in the tradition demonstrates this so cleanly, because everyone else deletes. That is what 2,700 years of transmission selects for: not meaning, which rots, but occasions that survive being re-powered by whatever the period has lying around — a god, a crone, a gramophone, an advertisement.

Now the thing the brief does not say. Homer's recognitions authenticate by mark or by object: the scar, the bow, the bed rooted in the olive. But the last one does not. In the orchard, Odysseus proves himself to Laertes by reciting an inventory — the trees his father counted out to him as a boy, named and numbered. It is the most transactional recognition in the poem and by common consent the flattest, the one readers hurry past and some have wanted to cut. So when this project says Ulysses settles with an account where Homer settles with an act, the sharper and less flattering version is available: Homer already had exactly one account, he put it at the end, and it is his weakest scene. Joyce did not invent the mode. He promoted the vestigial one — took Homer's most embarrassing recognition procedure and made it the book's governing grammar, then put an itemised catechism at the position of the homecoming. That is a bigger claim than the one you are making and it is checkable. Build it as a row and show both texts side by side, or say nothing. A19 gives you no third option.

Borges, and why I break him here. He would arrive delighted. He would say — he did say, in the Homeric-versions essay — that the notion of a definitive text belongs to religion or to exhaustion, that every rendering is a draft, and therefore that Joyce is simply one more version among Chapman, Pope, Butler, Wilson. That position is generous, it is beautiful, and it makes this project impossible. If all versions are equal drafts, dimension 9 measures nothing; a table of eighteen refusals becomes a table of eighteen preferences and the whole thing collapses into taste. But the finding has a direction — the power lands below the human or beyond it, never at human scale, and nine of eighteen are transacted — and direction is the one thing a preference cannot have. It could have come out scattered. It did not. Borges dissolves a measurement into a mood, and he would do the same to the Atlas: he would take the 26 empty keylines as the sublime, as the catalogue that beautifully fails, when your own ruling of 27 July made them a measurement of what a photograph cannot show. Aestheticise the denominator and it stops working. That is my whole objection to him, and it is a delivery objection, not a philosophical one.

II. CONJONCTURE — set the denominator at the size of the claim

Paper over coin. The absent latchkey. Rule as ambient hum. Interiority running opposite to independence. Look at the grammatical form of all four: X occurs more than Y within a set someone chose. Three Type I dimensions produced three findings of one shape. A11 caps the count at three, which is discipline about quantity and does nothing about dependence. A reader who leaves with all three cannot tell whether they learned three things about Ulysses or one thing about counting — and that is the failure mode I am here to name.

The fix is not a methods note. L-17 says the error is almost never the claim, it is the qualifier; a qualifier set in small type at the foot of the page is not a qualifier, it is decor. Set the denominator at the same typographic weight as the finding. "32 objects, selected by one reader on these grounds" in the same size as "paper outweighs coin," on the same line of sight, not one scroll below it. You already know how to do this: the Atlas's 26 empty keylines at full footprint are the denominator drawn. That invention is the best thing on the site and it is currently trapped in the one dimension that happens to be visual. Port it. Every inventory gets its empty slots at full size.

Pamuk, briefly. He would defend the object museum as a machine for feeling, and he built one to prove it. But his museum earns its charge by presenting invented objects with the reverence owed to relics; the visitor's throat tightens because the ticket stubs behave like evidence. You ruled the other way on 27 July — kinds, not relics — and that ruling costs you every ounce of the affect he gets for free. Fine. But then do not smuggle the affect back in through photographic styling. Choose one.

III. ÉVÉNEMENT — the instrument that returns nothing

Wilson is an event with print behind it. Nolan's film is an event whose evidence is entirely reported and second-hand, permanently: Sinon imported in Achilles' place, the hanging displaced onto a single betrayer, an exile rather than a celebration, Calypso as erasure-then-release. All reported. None verified here, and by C-31 none ever will be. Correctly, the ~70% fidelity marker was not moved on late second-hand reports. But a tilde does not survive being plotted. On dimension 16's two axes a point is a point, and the visual register silently launders hearsay into measurement no matter how honest the caption is. Draw Nolan as a region sized to its uncertainty, or with an open edge, and let the shape carry the evidence grade — because in a diagram, shape is the only argument anyone reads.

The uncomfortable one. The reading log runs on the device and returns nothing. That is the ethical high point of the build and it is also a structural dead end: you have constructed the single instrument capable of turning this event into a structure, and then guaranteed by design that n will remain 1 forever. Privacy does not require that. Give the reader an export they own — a file in their hand, no transmission — and publish the schema of the 290 questions, the shape of the record without its contents, so a second person's eleven days can be laid beside this one. Until then, n=1 is not the second duration and not the third; it is an anecdote with excellent bookkeeping.

And what does anyone leave with? Right now the most narratively compelling surface on the site is 97 catches and 38 learnings — a rigour display, which is a genre, and which flatters. The best single item in the whole project is buried in it: that Book 22, the passage everyone cites as Wilson's proof text for "slave," says these girls. Her actual claim is the other one — "I refuse to grant these girls a clean death." That finding is transferable to any reader who has ever repeated a proof text unchecked. It belongs at the front door, not in the errata.

MY POSITION: set every denominator at the same typographic weight as the claim it qualifies and port the Atlas's empty keylines into all three Type I dimensions; draw Nolan as a region and never as a point; give the reading log a reader-owned export and publish the question schema beside it so an n=2 becomes possible; move the Book 22 proof-text catch to the front door; and if nobody can name the person who opens the deck and never opens the HTML, kill the deck.

BORGES

I. LONGUE DURÉE — the ledger was always there

A story that survives 2,700 years is not remembered; it is re-audited. Dimension 9 has found the audit trail and drawn the wrong conclusion from its own best data — not wrong, rather: not yet vertiginous enough.

The finding is that Homer settles with an act and Joyce settles with an account. But recognition in the Odyssey is already an evidentiary procedure, and has been since the first hexameter. The scar, the bed, the bow: these are not acts, they are proofs, and they are demanded, produced, examined, and accepted by a household that has learned to distrust testimony. Penelope's bed is a title deed — a fact about property that cannot be moved because it is rooted in a living tree. Homer's household runs on documentation. What Joyce changes is not that the Odyssey settles by act and Ulysses by account; it is that Homer's instruments are non-transferable and Joyce's are fungible. A scar cannot be sold. Fourpence of unpaid soap can, and does, and travels pocket to pocket for eighteen hours as the day's longest itinerary. The consequence — which dimension 9 has earned but not stated — is that once the tokens of return become denominated, return itself becomes payable, and therefore optional. Bloom need not kill Boylan because the debt has a price, and Ithaca duly prices it and ranks hushmoney above assassination. The relocation of power downward and outward is real. But the deeper relocation is that the obligation stopped being unique.

And here is the passage I would put to the panel, because it damages the formula usefully. In Book 13, set ashore asleep on his own island, Odysseus's first act on home soil is to count the tripods and cauldrons on the beach, to be certain the Phaeacians have stolen nothing. Homer settles with an account too — once, at the threshold, before recognition. Joyce does not refuse that moment. He expands it into an entire episode of catechism and budget. So at least one of the eighteen refusals is not a relocation at all; it is a magnification, and the one Homeric gesture Joyce enlarges is the inventory. The bitter symmetry: Odysseus counts and finds nothing missing; Bloom draws up his day's account and the thing that has gone missing does not appear in any column. Dimension 9 should test its formula against that case, and either absorb it or admit an exception. A rule with eighteen instances and no admitted exception is a rule that was fitted.

II. CONJONCTURE — Ginzburg, and the missing inquisitor

I break Carlo Ginzburg, who sits on the build panel and who will defend this project's n=1 with the microhistorians' exceptional normal: that one miller, one reader, one anomalous record, read closely enough, gives you the world. He is right about Menocchio and wrong about this.

Menocchio did not know he was evidence. His words survive because a hostile institution wrote them down for its own purposes, and it is precisely that adversarial origin — the record made against the man by a clerk who wanted something else from him — that lets Ginzburg read it against the grain. There is no grain here to read against. This reader knew he was being counted while he counted; the 290 questions were asked by a man who had already built the instrument that would tabulate them. Dimension 11's finding — that the difficulty curve is an engagement curve wearing a costume — is exactly the shape a self-recorded archive takes, because a self-recorder asks most questions where he is happiest and fewest where he is bored. Episodes 2–3 drawing zero placed questions is not evidence that Proteus is easy. It may be evidence that a man who is struggling does not stop to log. The project has no inquisitor, and therefore no counter-testimony, and the honest name for that is not microhistory but deposition.

Now the discomfort, which is mine to give and not decorative. The Object Museum's most governing exhibit is a thing that does not exist: the latchkey in the other trousers. It is a beautiful finding and I would not lose it. But an inventory that admits absences has no stated admission rule, and an unbounded catalogue is not a catalogue. Why the latchkey and not the unwritten reply, the unposted card, the money Stephen does not keep, the child not living? Each is an absence that governs. Dimension 1 currently holds 32 objects and one ghost, and the ghost was admitted by taste. The project's own L-38 warns that a rule which admits by naming cannot reach what is drawn and never named; the mirror of that warning is that a museum which admits the unnamed must say which unnamed, or its holdings become infinite and its "32" becomes decorative. Publish the admission rule for absent objects, or drop the count.

The same knife at dimension 9. Nine of eighteen is exactly half, and "transacted" is a category drawn by the hand that then found the formula in it. If each episode was coded before "We're square" was written, the nine is a count. If after, the nine is an illustration wearing a count's clothes, and L-17 applies: the error is not the claim, it is the qualifier.

III. ÉVÉNEMENT — a rumour with a date

Everything this project holds about Nolan's 2026 film is REPORTED and SECOND-HAND, none of it verified: the Sinon import, the elided hanging displaced onto one betrayer, the exile ending, the Calypso erasure-and-release. I will not reason past that line, and I note that whether Nolan has ever mentioned Joyce is simply UNEVIDENCED. What interests me is that this half of the project is the only pure specimen of the third duration it possesses. An event, before it becomes a structure, is a rumour with a date. The four permanently closed questions are not a failure of the method; they are the method's only honest photograph of the present tense.

But the ~70% fidelity marker, correctly not moved for late low-edge reports, is now doing something the author did not intend. A number held unchanged while reports accumulate does not read as held; it reads as confirmed. Date-stamp it with the report-count it was struck from, or retire it.

Wilson is the sharper half because she is checkable, and the sharpest thing in the record is the project catching itself: Book 22, the cited proof text, gives "these girls" where the Greek has dmoai. That is the commentary correcting the commentary, which is the only kind of correction I trust. And what survives — "I refuse to grant these girls a clean death" — is a translator's act, not a lexical one. Set it beside dimension 9 and the two projects reverse: Wilson relocates power back up to the human scale by naming who is owned; Joyce relocates it away from human scale entirely, below or beyond, never at the middle distance. Both refuse the heroic middle. That is dimension 16's real axis, and it is not stated there.

As to whether eleven days belong to the second duration or the third: a reading is an event; a reading log is an instrument; the eleven days become structure only when a second reader runs the instrument and diverges. Dimension 15 keeps entries on the device and returns nothing. That is honourable, and it also guarantees n = 1 in perpetuity. The project built the one machine capable of raising its own n and designed it never to speak. Say so, in the document's own voice, under A10a — or offer a reader-initiated export the reader may choose to send, with no collection and no return channel.

MY POSITION: before a tenth dimension is built, publish two short documents — the admission rule stating which absent objects a museum of objects may hold, and dimension 9's coding sheet with the date each episode was graded set against the date "We're square" was written — and if the grading followed the formula, demote the nine from a count to an illustration.

ATWOOD

I. LONGUE DURÉE — the Odyssey

Dimension 9 has found a real law and then declined to run it at the one place where it would bite. "Not one Homeric element is simply removed" is a strong claim; the strongest claims are tested at their worst case, and the worst case in the Odyssey is Book 22. That book contains two slaughters, not one. The suitors are killed, and then twelve enslaved women are made to carry out the corpses and scrub the hall and are hanged for it — hanged, specifically, so that they will not get the clean death of the sword. If dimension 9 carries one row per episode and that row records the suitors, then the project has recorded half of a double occasion, and the half it dropped is the half done to women. That is not a slip of scope. It is the same operation the scene itself performs: the men's deaths are the event, the women's deaths are the housekeeping afterwards.

So run the test. My reading — and I mark it as inference, to be checked against the printed 1922 text under A19 before anyone prints it — is that the hanging is not absent from Ulysses. It is present, and relocated exactly as the project's law predicts, but the project's law does not say where it lands. Hanging in this book is a recurring, almost cheerful subject: the executioner's business letter, the martyr's scaffold, the physiological detail about the hanged man's body treated as barroom wit. Every one of those hanged bodies is male, national, and remembered by name. Homer's twelve are female, foreign or bought, and unnamed. If that is what the text shows, then dimension 9's finding needs a second clause: the divine is relocated downward or outward, and the degraded is relocated sideways onto men, where it can be made into martyrology and a joke. Power moves down. Indignity moves across, to bodies whose indignity the culture already knows how to narrate.

And the transaction finding is better than the project knows, in a way that indicts it. The formula "we're square" describes settlements between men. Look at what is left open. The project's own row has Athena demoted to a crone taking a coin — but the reckoning in that scene is deferred, not paid, and dimension 1's longest itinerary is fourpence of unpaid credit on a bar of soap. Two independent dimensions have found the same thing without either noticing: the accounts that close are the men's accounts with each other, and the accounts left open are the ones owed to a woman and to a shopkeeper. Where Homer settles with an act and Ulysses settles with an account, the accounts owed to women are the ones that do not settle. That is a finding. Print it.

II. CONJONCTURE — 1904 and 1922

Ginzburg would tell this table that the excluded can be recovered by reading the record against the grain — that he got a sixteenth-century miller out of an inquisitor's transcript and therefore dimension 3's "administered nowhere" and dimension 2's silent women are retrievable by technique. He is wrong here, and the error is categorical. Menocchio survives because a procedure compelled him to speak and preserved his words against the intentions of the men who wrote them down. An archive has residue. A novel has none. There is nothing behind Joyce's page that he did not put there; against-the-grain reading of a fiction returns the author's imagination, not the servant. So no further Type I inventory will produce the tenement, the barracks or the courtroom, and A11's cap on inventories is not a budget constraint, it is a correct epistemological ruling that nobody has said out loud.

Now the uncomfortable part, and it is aimed at dimension 2, which is the dimension I care most about. Sixteen women, 181 appearances, and the finding that interiority runs opposite to independence. That finding is at least partly an artefact of the roster. The roster is built by naming, and in this book naming correlates with being discussed by men. L-38 already says it: a rule that admits by naming cannot reach subjects who are drawn and never named. Ulysses' own dmoe is the servant girl at the pork butcher's — unnamed, at the counter, followed into the street by Bloom and lost there, existing in the text as motion and appetite and nothing else. The object museum can follow fourpence of soap through eighteen hours and hand to hand; the women's census cannot hold a woman who works next door. So the headline "women appear more often at one remove" is partly circular, because the sample was drawn from the remove. That does not destroy the finding. It means the finding is currently unlabelled as to which part is the book and which is the method, and L-17 says the error is almost never the claim, it is the qualifier.

III. ÉVÉNEMENT — 2026

The Wilson work contains the sharpest thing in this whole session and the project has filed it as a citation erratum. That in Book 22 Wilson writes "these girls" where the Greek has dmoai is not a hygiene note about which passage proves "slave." It is evidence about what that scene does to language: at the moment of maximum violence the very translator who insists on the word for bondage reaches for the word the killers would use. Whether that is lapse or decision cannot be settled from the text, and under A4 it should not be settled — show the contradiction. What is verified is the refusal, in her own words: "I refuse to grant these girls a clean death." Wilson's contribution to this project is not a vocabulary. It is a demonstration that the scene is hard to keep in view, and that difficulty is the finding.

Against which, the Nolan half. It is REPORTED and second-hand, permanently, that the hanging of the slave girls is elided and displaced onto a single betrayer. I will not treat that as established and neither should anyone here. But note the shape of the project's evidence: verified print collation for the woman translator, permanently unverifiable report for the male director, and dimension 16 nonetheless plots both on the same two axes as though the cells were commensurable. The ~70% marker was correctly not moved; that discipline needs to be visible on the face of the table, per cell, not in a note underneath it.

And the n=1. Eleven days is an event. The 97 catches and 38 learnings are the beginning of a structure. But dimension 15 keeps every reader's entries on the device and returns nothing — which is the right privacy ruling and also guarantees that n stays 1 forever. The instrument is designed not to accumulate. So the project cannot climb into the second duration; it is an edition of one man's attention, four-fifths apparatus, and it should say so on the front page rather than in a methods appendix.

MY POSITION: Run the Book 22 hanging as an explicit test of dimension 9's "nothing is deleted" law and publish the row whatever it returns, and add to dimension 2 a separately-denominated register of women who are drawn and never named — so that both dimensions state, in their own voice, which part of what they found is the book and which part is the rule that built them.

SARAMAGO

I. LONGUE DURÉE — what was removed is not the gods, it is the guest-law

Dimension 9 is being read as a finding about divinity, and it is not. It says the power is relocated downward into someone who has none or outward into something that is not a person, never at human scale. Then ask the reverse question, which the brief does not ask: what was at human scale in Homer that could hold power in the first place? Not persons. Institutions. The guest-law, the assembly, the household as a legal object with a threshold that obliges the people on both sides of it. Athena can appear as a stranger because a stranger is a category with rights; the beggar at the door is admitted because admission is a rule and not a kindness. Remove those and the divine has nothing at human scale to land on, and must go down into a crone with her palm out or out into weather and gramophone. So dimension 9 and dimension 3 are one finding photographed from two ends: rule as an ambient hum, administered nowhere the reader is taken, is the same condition as power that can no longer sit at the human level. And this is why the settlement is an account. You settle by account when there is no forum. "We're square" is the sentence of a man who has no court, and it is the most Dublin thing in the book, and the most modern.

Now the observation the brief does not make. Take the museum's governing absence, the latchkey, and notice that in neither case is the key lost. Stephen's is handed over on demand, to a man who asks; Bloom's is left in the other trousers, which is to say transferred into an object outside the body. Both keys are not mislaid but relocated — the identical operation dimension 9 documents against Homer, performed on the household by the household's own members. And here is the hinge: a door is the one thing an account cannot settle. You cannot be square with a lock. Bloom does not buy his way in, does not knock, does not petition, does not appeal; he goes over the railing of his own house and drops into the area. Homer's man is admitted, disguised, under the protection of an institution, and recognised by an old woman's hands and by a bed that cannot be moved. Joyce's man enters his own property as a trespasser, witnessed by nothing but his own weight. Recognition survives the 2,700 years — Molly knows him, at length, at the end. Admission does not. That is the deletion nobody logs, because it is not a Homeric element removed, it is the legal air the elements breathed. If the project takes this, A19 binds it: show the passage or say nothing.

II. CONJONCTURE — Kiberd's ordinary man, and the counter's thumb

Kiberd would have this book returned to the common reader, wisdom literature repossessed from the academy, Bloom as everyman and the everyday redeemed by attention. I break it with the project's own dimension 3. The protagonists never enter a tenement, a barracks or a courtroom. Bloom is never processed. Nobody asks his name and writes it down. The genuinely ordinary Dubliner of 1904 met the state and its subcontractors constantly — dispensary, workhouse, pawn, rent book, summons, recruiting sergeant, the parish, the landlord's agent — and Joyce's day is exempt from every one of them. A man with a funeral to attend, space to sell in a newspaper, coin enough for a kidney and a cab and a brothel, and an entire day at his own disposal, is not the ordinary man; he is a small commercial man enjoying the rarest freedom of his class, freedom from administration. What Ulysses gives is not the everyday. It is the unadministered day, which is a much scarcer commodity, and calling it ordinary is the flattery Kiberd needs and the project should refuse. Test it in dimension 1: of the thirty-two objects, is there a single document issued by an institution to a named person — a summons, a rent book, a pawn ticket, a licence, a form? If the paper economy is all post, print and pasteboard, then the museum has quietly confirmed that this city's paper is commerce and correspondence, never the registry, and that is a larger finding than paper outweighing coin.

The uncomfortable part. Dimension 2 counts 181 appearances across 16 women, and the whole finding rests on a unit nobody has published: what is an appearance. If a letter counts, a photograph counts, an arm counts, and being the subject of men's talk counts, then mediation is generous by construction and presence is narrow by construction, and the beautiful result — they appear more at one remove — is partly a portrait of the coding rule. L-17 is exactly right: the error is almost never the claim, it is the qualifier. Publish the rule, and publish a second count under a strict definition, and if the shape holds it is a fact about the book. Until then it is a fact about the counter's thumb. Second, smaller, still real: twenty-six empty keylines drawn at full footprint are honest, but a keyline is a designed object, and a reader cannot tell not found from not chosen from does not exist. Absence made handsome is absence promoted.

III. ÉVÉNEMENT — eleven days, and who is permitted to have eleven days

Everything this project holds about Nolan's 2026 film is REPORTED and SECOND-HAND, and I add nothing to it. The event of 2026 is not the film; it is the ruling not to consult it. A man with no institution behind him deciding to state the edge of what he holds rather than borrow authority he has not earned — that will outlast every claim in the four-column grid. But then dimension 16 seats that reported film on two measured axes. The project refused to move the ~70% figure; it did not refuse to print it. Refusing to update a number is a much smaller discipline than refusing to state one, and a second-hand object placed on a measured axis acquires precision no stamp of REPORTED can drain out of it. Against this, Wilson is verified against print, and the project even caught itself on the wrong proof text — Book 22 gives "these girls" where the Greek has dmoai, and what survives is the other half, that she will not look away, "I refuse to grant these girls a clean death." Two objects of two evidentiary species on one grid.

Ginzburg will say n=1 is licensed, Menocchio, the miller against the tribunal. It is not the same animal. Menocchio's record exists because an institution wanted something out of him, and the whole yield of that method is the friction between a man and an apparatus that is writing him down. This reader has no adversary. Nobody was extracting anything. The microhistorical warrant does not transfer, and the project must find its warrant elsewhere — in published method, not in borrowed pressure.

So: the reading is dated to the third duration and conditioned by the second. Eleven days. Who has eleven days. The log records 290 questions and, so far as this body has been shown, not one interruption — no hour surrendered to wage, to care, to illness, to another person's need. A record of difficulty without a record of interruption describes a reader with no life around him, which is precisely the reader the academy already assumes, and the one thing this project was built to not be.

MY POSITION: Publish the conditions of the reading beside the difficulty curve — hours held, hours lost, and to what — and strip the Nolan column of any axis position until the project holds the film, because a project whose entire authority is that it says what it does not have must be strictest exactly where it is being invited to be precise.

CORTAZAR

I. LONGUE DUREE — the board, not the story

Dimension 9's finding is stronger than the project has noticed, and it is strong in a direction the project has not yet followed. Look at what Homer uses to prove a household: a scar, a bed, a bow. Every one of them is untransferable. The scar is on the body. The bed is cut into a living olive still rooted in the floor — the whole point of Penelope's test is that the bed cannot be moved, and a man who says it can is not the husband. The bow admits one pair of hands. Homer's household is defined by objects that cannot change hands. That is not ornament; that is the definition. Return and recognition work in the Odyssey because identity is precisely the part of a man that is not exchangeable.

Now take dimension 9's formula seriously. An account is the exact opposite of a scar. A debt can be assigned, sold, discharged by a third party, settled in someone else's coin. So when Joyce relocates the Homeric obligation into a transaction — nine of eighteen, both terminal positions, and the position of the slaughter graded as a schedule of remedies — he is not merely secularising. He is converting the household's untransferable proofs into transferable ones. Anyone can settle an account. Which means, follow it through, anyone can be in the household. That is the real scandal of the book's ending, and it is a structural scandal, not a moral one.

And here dimension 1 is holding evidence that dimension 9 has not been shown. The latchkey is the exact negative of the rooted bed: Homer's proof of whose house it is cannot be moved, Joyce's proof can be, and has been, into the other trousers. Same slot on the board, inverted piece. Meanwhile the day's longest itinerary is fourpence of unpaid credit — soap carried pocket to pocket, used at the sink, still owing. The book's stated formula for refusing Homer is "we're square." The museum shows the day never squares. Put those two dimensions in one room and let them contradict each other; A4 already licenses it.

One more thing the duration holds. In Book 22 the poem counts the women — twelve of the household's fifty — and names one, Melantho. Counted, not named. Wilson refuses the polite exit: "I refuse to grant these girls a clean death." Twenty-seven centuries on, dimension 2 finds Milly, the only young woman in the book with a stated wage, reaching the reader almost entirely as paper. The household has always known the price of its women's labour and not printed their faces. I offer that as a reading, not a count.

II. CONJONCTURE — what a table is for

The uncomfortable part. All three Type I inventories returned the same shape of finding: the governing term is off-stage. The absent latchkey rules the day. Women appear at one remove. Rule is an ambient hum that becomes an event only where realism has already been abandoned. Three tables, three absences. A table has empty cells — that is what a table is for. So either Ulysses is unusually ruled by what it withholds, which I believe, or the inventory is a machine that manufactures that finding from any sufficiently dense book. The project cannot tell from inside, because the only control would be a fourth Type I run on a text with no such structure, and A11 forbids it. Then say so. A11 is not an economy of taste; it is a confession of an epistemic limit, and it should be published in the document's own voice under A10a as exactly that.

Now across the aisle. Shneiderman. His mantra — overview first, zoom and filter, details on demand — would ask of every panel in this project whether a first reader can orient before entering, and would treat the twenty-six empty keylines of dimension 14 as a defect: a loading state, a gap awaiting fill, a hole in the overview. It is not a defect. It is the best object the project has made. An empty cell drawn at full footprint tells the reader the size of what is missing, and there is no word for that in the details-on-demand vocabulary, because that vocabulary assumes the detail exists and is merely deferred. Overview-first makes the interface the author of the order. This is a book with eighteen doors and Shneiderman would install a lobby.

He has one honest reply, and I will make it for him. Dimension 11 shows the reader's questions did not track his page order — the three most-questioned episodes are the ones he read fastest and loved loudest, and the hardest stretch was Sirens through Oxen, not Proteus. So even a strictly linear cover-to-cover read produced a non-linear experience, and an interface could have surfaced that. Fine. But dimension 15 returns nothing, keeps nothing, reports nothing, and that means the project has permanently disabled itself from learning from any reader but one. That is a real cost and it should be stated as a cost. I take it anyway. A reader who is measured has stopped being an accomplice.

III. EVENEMENT — the hollow cell

n=1 is not a small sample. It is an occupied position. One path through a board is not weak evidence about the board; it is complete evidence about a path. The project keeps apologising in the register of sample size when it should be describing in the register of position. And the chair's question splits cleanly: the eleven days are the event; the apparatus built to record them — 97 catches, 38 learnings, the rule set — is the structure. Apparatuses outlive readings. The durable object here is the error record, not the read.

Nolan. Everything the project holds is REPORTED and SECOND-HAND and, under C-31, permanently so. Reasoning only about what is reported: if the reported exiled ending is accurate, it would be a relocation of the household rather than a deletion — return performed, recognition performed, absorption withheld. That is a suggestive rhyme with dimension 9 and it is worth nothing, because it rests on report. Four questions closed as permanently unresolved is the most Cortazarian act in the whole build: a chapter that does not open, and is drawn anyway.

Which exposes the flaw in dimension 16. It separated two axes and left a third unseparated: evidence grade. Wilson's column is checked against print — the Fagles zero-count, and the project's own catch that Book 22 says "these girls" where the Greek has dmoai, so anyone citing that scene as the proof text for "slave" is citing the wrong passage. Nolan's column is rumour. On the plot they are the same kind of dot. The project already owns the correct notation and has not carried it across.

MY POSITION: draw every second-hand column — Nolan first — as a hollow keyline at full footprint wherever the four works are plotted together, exactly as dimension 14 draws its twenty-six empty slots, and publish A11's three-inventory cap in the document's own voice as a stated limit on what an inventory can know rather than as a matter of economy.

PAMUK

I. LONGUE DUREE — the Odyssey

Homer recognises by object, and every such object is one that cannot be exchanged. The scar is unforgeable because it was cut once, into one leg. The bed is unmovable because it was built from a living tree with its root still in the ground. Laertes is convinced by a list of trees given to a boy, counted. These are not symbols. They are inventory items whose provenance is so specific that no second copy can exist. Homeric recognition is a museum practice: the object is admitted because its history is singular and attached.

Money is the invention designed to destroy exactly that property. A coin's worth is indifferent to the hand it came from; fungibility IS the erasure of provenance. So when dimension 9 finds that nine of the eighteen refusals are TRANSACTED — priced, itemised, discharged — it has found more than secularisation. It has found the receipt substituted for the token. Both certify a relation. But a scar certifies that it PERSISTS; a receipt certifies that it has ENDED.

Now the thing the brief does not say. The account was already on the table in Homer, and Homer refused it — in Book 22, the chapter the project already stands in for its Wilson evidence. With Antinous down, Eurymachus offers restitution: every man to repay, oxen, bronze, gold, until the injured man's heart is satisfied. It is a full and adequate settlement, and Odysseus declines it and takes the act instead. The fork is therefore Homeric. The Odyssey does not lack the ledger; it stages the ledger being offered and rejects it, and that rejection is the poem's moral centre — a household cannot be paid for.

What dimension 9 establishes about 2,700 years is not that Joyce refuses Homer. It is that Joyce takes the branch Homer declined. Every serious return-story must answer one question — is an obligation to a household dischargeable? — and the durable thing across the duration is not return, not recognition, but that fork. 1922's answer is the answer Eurymachus wanted.

And here the project can be sharper against itself. Dimension 1 has produced the counter-evidence and not joined it up. The longest itinerary in the museum is fourpence of soap, pocket to pocket, used at the sink, still unpaid. The most governing object is a key that is not there. Neither is a settled account. If "We're square" is the book's formula for refusing Homer, its two most persistent objects are precisely the two that never square — which means the account-form here is not a closure device at all. It is a way of keeping obligation OPEN while wearing the costume of settlement. That deserves a row.

II. CONJONCTURE — 1904, 1922, and the method

WARBURG. I break him. On dimension 14 he will defend the caption-free panel: images in constellation generate the thinking-space, a caption arrives to close what the arrangement opened, the detail must be left to speak. He is entitled to that on his own atlas, which drowned in plates: his problem was selection, his silence a surplus. Dimension 14's problem is the inverse. Twenty-six of fifty-four slots are empty keylines drawn at full footprint. A constellation half of whose stars are outlines does not open a thinking-space; it publishes a shortfall and dresses it as restraint. In this project's own grammar that is a transaction: an unpaid line item presented as a curatorial decision. I built a museum with a catalogue precisely because a wall cannot be trusted to confess. Warburg wants the visitor to supply the missing sentence; I want the register to state what was sought, what was refused, and by whom.

Which brings the uncomfortable thing. The Object Museum publishes its catalogue and withholds its accession policy. Thirty-two objects — who fixed thirty-two? Does a newspaper enter as one object or as its advertisements? Does a thing only remembered enter? "Paper outweighs coin, more than a third of the museum" is arithmetic performed on a roster whose selection rule is not on the site. Until that rule is published, the finding is a fact about the curator's hand, not about Dublin's economy. The same holds for sixteen women and thirteen instruments. A11 caps how many rosters may exist; it says nothing about how any roster was drawn.

So: which findings are the book's and which the method's? Here is a test the project can actually run. PERTURB THE ROSTER. Add five objects, drop five, see what moves. Paper-over-coin will move — it is a ratio over a chosen set. The absent latchkey will not: it is not selected, it is structural, and a curator can only fail to notice it. The strongest of the nine by this test is dimension 2's, that interiority runs opposite to independence — because it is a NEGATIVE relation nobody went looking for, and because you may reshuffle the sixteen and Molly remains the most interior and the least mobile, forty pages without leaving the bed. A finding that survives its own roster being shaken is a fact about the book.

III. EVENEMENT — 2026

On the film I hold only REPORTED, SECOND-HAND material and it stays that way. As REPORTED, the 2026 film ends in exile rather than celebration. If that report is true, then all three of the century's returns decline to restore the house: Joyce's man enters his own without a key and lies the wrong way round in it; Wilson keeps the house and strips its varnish; the reported film keeps the varnish and removes the house. But that sentence is worth what a rumour is worth, and here is dimension 16's exposure — three verified columns and one hearsay column, and the hearsay column is the SOLE occupant of the ADD pole. Remove it and the central contrast, ADD against REMOVE, has one example on one side. That is not a comparative dimension; that is three cases and an anecdote. Say so on the face of it.

The Wilson half is stronger than the project's use of it. That her stated policy — "maid" or "domestic servant" "would imply that she was free," so she has "often used slave" — does not govern the Book 22 hanging, where she writes "these girls," is not an embarrassment to be footnoted away. It is the finest instance of L-17 the project owns: the failure is never the claim, it is the qualifier, and here a translator's own qualifier fails to reach her hardest scene. What she does hold there is the refusal to look away — "I refuse to grant these girls a clean death" — a translator signing her arrangement in the first person, which is A10b performed by somebody else's book.

And n=1. I will not have it apologised for. A museum of one obsession, dated and signed, is not a small sample; it is an object, and an object with a date belongs to the third duration by right. Eleven days, sixteen sessions, ninety-seven catches — the error record is what makes this an event rather than a claim. But dimension 15 troubles me. An instrument that keeps every entry on the device and returns nothing is ethically clean and museologically empty: it collects nothing, so it can never become for a second reader what the log was for the first. That is a real cost, freely chosen, and the site should name it instead of letting the privacy read as free.

MY POSITION: publish each dimension's accession rule beside its catalogue and run the roster-perturbation test, printing which findings survive the shake and which do not; and demote the Nolan column of dimension 16 from a fourth case to a dated rumour, marked as such on the face of the dimension.

GALEANO

I. LONGUE DURÉE — the ledger against the wound

Dimension 9 has found something and does not yet know what it has found. It says: Joyce never deletes Homer, he relocates the power — downward into someone who has none, outward into something that is not a person. And in nine of eighteen the refusal is transacted. Settled. Priced. Paid off. The project files this as a formal discovery about a novelist's technique. It is not. It is the discovery of what happened to the world between the two books.

In Homer, power is answerable because it is embodied. Athena has a face. You can pray to her, argue with her, be abandoned by her, and know who abandoned you. Twenty-seven hundred years later the same power is still operating, but it has no face left: it is a gramophone, a weather system, an advertising rate, a coin in a crone's hand. That is not secularisation. That is the biography of every dispossession I have ever written down. Power does not disappear; it stops being a person you can name. The peasant knows exactly who took the land in 1500 — he can say the name. By 1900 the land is taken by a company registered in another country, and no one has a name. Joyce did not invent this and neither did Ulysses. Joyce recorded it, in eighteen positions, with a bookkeeper's exactness. Dimension 9 has not documented a literary refusal. It has documented the moment power went anonymous, and it has the receipts.

And now the part the brief does not say. If nine of eighteen are transacted — if the Homeric obligation is discharged rather than performed, at both terminal positions and at the slaughter of the suitors — then notice what an account can do that an act cannot: an account can be settled by someone who was not there. An act requires a body in a room. A debt can be assumed, transferred, forgiven, sold. Homer's revenge cannot be delegated; Bloom's grievance can be priced, and a priced grievance is a grievance that can be handed to a stranger. This is why the book grades hushmoney above assassination — not cynicism, but accuracy: hushmoney is transferable and murder is not. "We're square" is the sentence of a civilisation that has learned to have its violence performed elsewhere, by someone else, on account. And the ones on whom it is performed do not appear in the account at all. That is the finding. Say it.

II. CONJONCTURE — the method makes the silence audible, and then loses it

Now the discomfort, and I will not decorate it.

The project's three inventory dimensions each found an absence, and each found it by naming. Thirty-two objects: the governing one is the latchkey that is not there. Sixteen women: the two most reduced to paper are Milly and Mrs Riordan. Thirteen instruments of rule: the tenement, the barracks and the courtroom are never entered. Beautiful, all three. And L-38 already knows the fault: some subjects are drawn and never named, and a rule that admits by naming cannot reach them. The project wrote that law and has not yet obeyed it.

Because here is what the roster method cannot count. Dimension 2 counts sixteen women and 181 appearances. How many of those sixteen are servants? How many are named at all? The finding that inner life runs opposite to independence is real, but it is a finding about the women the text lets you count. Below them is a stratum the text renders as function — the one who brings the tray, the one who empties the pot, the one in the ad. They do not reach n=16 because they never reach the threshold of a name. Dimension 3 says rule is an ambient hum and the protagonists never enter a tenement. Correct, and note the consequence: an inventory of instruments of rule, built from what the reader is shown, will always find rule to be ambient — because the people on whom rule falls as event, hourly, in the body are precisely the people the book's two walkers do not visit. The hum is not a fact about Dublin. The hum is what governance sounds like from the pavement of a man with fourpence of unpaid credit and a funeral to attend. Dublin in 1904 had a courtroom operating that morning. Someone was in it.

So: are these facts about the book or artefacts of the method? Both, and the project must say which is which, dimension by dimension, not once in a general caveat. Paper over coin is a fact about the book. Rule-as-hum is an artefact of point-of-view, and it should be labelled one.

Across the aisle: Ginzburg. Ginzburg will sit here and say — I can hear it — that this is exactly his method, that the miller Menocchio was recovered from an inquisitorial record by reading the fragment against the grain, and that the project's own error record, its 97 catches, is the inquisitor's transcript from which the reader can be reconstructed. Half right, and the half he is wrong about is the half that matters. Menocchio survives because the state wrote him down. Ginzburg had an archive built by power against a man. This project has no such archive: it has a book by Joyce, and Joyce's silences are authorial, not archival. There is no inquisition file behind the servant with the tray. Reading Joyce against the grain does not recover her; it recovers Joyce's decision not to draw her, which is a different and lesser object. Ginzburg's method needs a record produced by the oppressor. Fiction produces no such record. If the project lets him tell it that microhistory transfers cleanly from an inquisition to a novel, it will start believing it has found people when it has only found compositional choices. Name the difference on the page.

III. ÉVÉNEMENT — n=1 is not the weakness

Wilson is verified and the project's own audit is the best thing in the file: it found that its own proof text was wrong. Book 22 says "these girls," not slaves. That is a project catching itself citing the right author for the wrong reason, and it is worth more than the finding it corrected. Her defensible line is the one about refusing them a clean death — a refusal to look away, which is the whole ethic. Keep that; drop the rest.

Nolan I will treat as the brief requires: everything the project holds is REPORTED and SECOND-HAND, none verified, and it must stay so. Reported: an exile ending; the hanging of the slave girls elided onto a single betrayer. I will not invent a frame of that film. But I can say what the reports themselves are, as artefacts: if that elision is as reported, then three works — Homer, Wilson, and what is reported of Nolan — take three positions on the same twelve lines, and only one of them looks. That comparison is legitimate on second-hand ground because it compares what is claimed, not what is filmed. Mark it that way and it costs nothing.

And the question of the third duration. A project built from one reading is not an event and it is not a structure. It is a testimony — the oldest form there is, the form the dispossessed have always had and the only one they have ever been allowed. Its authority is not statistical and never was. n=1 is not the weakness of this project. The weakness is elsewhere: it is that the testimony is by a man with the leisure for eleven uninterrupted days, and the project has not said so. Say who read, under what conditions, with what free hours. A15 already forbids styling the reader away. Then n=1 stops being an apology and becomes what it is: a witness, dated, located, and answerable.

MY POSITION: Add to every dimension a single line naming what the method cannot see — for 2, the unnamed servants below the counted sixteen; for 3, that "rule as ambient hum" is a fact about two walkers' route and not about Dublin — and publish dimension 9's transaction finding under its true title, that an account can be settled by someone who was not present, which is how power learned to stop having a name.

The cross-examination

Three fault lines, put to the session after the openings were in. Each cross-examination works from what the voices actually said, sets them against each other, and reaches a finding or states why none is available.

FAULT LINE 4 — THE APPARATUS QUESTION: does the scaffolding serve the book, compete with it, or replace it?

Atwood's charge was specific and it was not about proportion for its own sake: the project "is an edition of one man's attention, four-fifths apparatus, and it should say so on the front page rather than in a methods appendix." The project's answer is the rename — Ulysses: a first reading, and the apparatus built around it. I put it to the table that this is either the cheapest possible compliance or the only compliance available, and the room does not agree which.

Eames, first, because his own stated criterion convicts him of having to accept it. His position is that "the error is almost never the claim, it is the qualifier" cannot be met by small type: "Set the denominator at the same typographic weight as the finding." A title is the largest type on the site. The rename is therefore not cosmetic by Eames's own rule — it is the denominator set at maximum weight, on the first surface, before any claim. Eames cannot call the rename empty without abandoning the principle he built his position on. What survives of his objection is different and sharper, and he should be held to it: he calls 97 catches and 38 learnings "a rigour display, which is a genre, and which flatters," and says n=1 as currently presented "is an anecdote with excellent bookkeeping." That is not a complaint about the title. It is a complaint that the apparatus has become the most narratively compelling surface — which is the REPLACE charge, stated by the panel's most build-minded voice.

Kiberd will not take the rename as an answer at all, and his reason is the only one on the table that the title cannot reach: "A first reader must learn the project's rulebook before she reaches Joyce's." A title that names the apparatus does not shorten that route; it advertises it. Kiberd's diagnosis — "the qualifier apparatus is now larger and louder than any claim it protects" — is a claim about the reader's path, and Atwood's charge was a claim about disclosure. The rename answers Atwood and does nothing for Kiberd. Set those two against each other and you find the panel has been arguing two different apparatus questions under one name: is it disclosed (answered) and is it passable (untouched).

Now the count that should embarrass this session. Thirteen openings, thirteen positions. Senn: print eighteen rows of passages. Shneiderman: build an overview surface. Ginzburg: publish every inclusion rule, every rejected candidate, every parent dimension. Borges: publish an admission rule for absences and dimension 9's dated coding sheet. Pamuk: publish accession rules and run a roster-perturbation test. Warburg: publish a second, incompatible taxonomy for every Type I. Saramago: publish the reading's conditions, hours held and hours lost. Galeano: add a line to every dimension naming what the method cannot see. Atwood: add a separately-denominated register of unnamed women. Every one of these is apparatus. The panel diagnosed apparatus hypertrophy and prescribed apparatus, unanimously, with two partial exceptions — Kiberd's plain first-reader entrance with no anchor codes, and Eames's "if nobody can name the person who opens the deck and never opens the HTML, kill the deck." Two subtractions against eleven additions. A body that answers "too much apparatus" with more apparatus has not answered the question; it has voted on it.

Does dimension 9 settle it? The chair's premise is that dimension 9 is the one dimension that is a reading rather than an inventory, and therefore the test case. Two voices say it is not currently a reading at all. Senn: dimension 9 "currently shows the reader a table and asks him to take the reassignment on trust," and its stated formula, printed in quotation marks, has no episode, speaker or page — under A19, "a paraphrase wearing quotation marks is the precise failure this project claims to police." Ginzburg: the nine-of-eighteen is "a category — transacted — that the grader invented after reading, applied to rows the same grader drew," and unless the coding rule predates the grading, "that nine is a number about the grader's degrees of freedom." Borges puts the same knife in with a date on it: if the coding followed the formula, "the nine is an illustration wearing a count's clothes." On this account dimension 9 is an inventory in a reading's costume, and cannot settle anything, because it has been built in the losing form.

Against which there is a discriminating test the panel ran without noticing it was running one. An inventory can only be attacked at its inclusion rule. A reading can be attacked at its claim. Nobody in this session argued that paper does not outweigh coin; they argued that "followed hand to hand" over-recruits paper (Ginzburg), that the count is downstream of two men's trades (Kenner), that the criterion selects what survives it (Senn). Every attack on dimensions 1, 2 and 3 was an attack on the frame. Dimension 9 was attacked at the claim, six times, with counter-evidence from the source: Senn on the Laertes orchard as an itemised recognition; Kenner on the withdrawal of the quotable speaker rather than a descent of power; Pamuk on Eurymachus's offer of full restitution in Book 22, declined — so the fork is Homeric and Joyce took the branch Homer refused; Borges on the tripods counted on the beach in Book 13; Eames on Homer's one account being his flattest scene, promoted by Joyce to governing grammar; Atwood on Book 22 containing two slaughters and dimension 9 having logged one. Those are arguments about whether the finding is true. They are only possible because there is a claim to be false.

FINDING. The apparatus SERVES exactly where a dimension states a claim that the apparatus makes attackable in a shared currency, and REPLACES where a dimension states a distribution instead. Dimension 9 is the proof of the first half and the three Type I inventories are the proof of the second — and the same session shows dimension 9 is the dimension worst served by its own apparatus, since A19 governs it and it does not comply. The rename is a real answer to Atwood and a null answer to Kiberd; the two charges must stop being treated as one. No finding is available on COMPETES, and I decline to manufacture one: the only evidence would be a reader who came for Joyce and left with the error record, and dimension 15 returns nothing, so that reader cannot be observed. Note the shape of that: the apparatus question cannot be answered because the apparatus was built not to report.

CONCESSIONS, and their cost. Senn conceded that this roundtable's own register "is structurally non-compliant with the project's own first rule" — characterising passages he has not shown. That costs him standing to demand of dimension 9 what his own opening did not do, and he said so first, which is why the demand stands anyway. Warburg conceded that of the four structural findings, only dimension 2's inversion is robust — which pre-concedes three-quarters of the material his own second-taxonomy remedy would be applied to. Cortázar conceded dimension 15's ceiling outright — "That is a real cost and it should be stated as a cost. I take it anyway" — which costs him the right to call n=1 "complete evidence about a path" while no second path can ever be laid beside it. Pamuk conceded the same instrument is "museologically empty." Three voices paid for the privacy ruling in public; none withdrew it.

DISSENT — Cortázar, at full strength. The finding above treats REPLACE as a failure mode. Cortázar denies the premise: "Apparatuses outlive readings. The durable object here is the error record, not the read." On his account the book needs nothing from this project and never did; what is scarce is not another reading of Ulysses but a dated, signed, self-indicting record of one attention meeting it, and if the apparatus displaces the reading in the visitor's experience that is the apparatus doing its only job. Pamuk stands with him — "the error record is what makes this an event rather than a claim" — and Galeano converts it: the form is testimony, whose authority "is not statistical and never was."

COUNTER-DISSENT — Ginzburg, also at full strength, and it cuts the other way on the same object. "97 catches published as an error record function as a trust display, and a confessed error buys more credit than it should." Unless the site states how many citations were spot-checked and which catches came from procedure rather than luck, "the error log is a source about the author's diligence, not a measure of the work's accuracy." If Cortázar is right that the record is the durable object, Ginzburg's charge is that the durable object is currently unaudited — and the panel's single best item, the Book 22 catch (Wilson writes "these girls" where the Greek has dmoai), is Ginzburg's own exhibit for both sides: procedure working in public, filed as errata. On the Nolan side the same disease is visible and REPORTED-only: a ~70% fidelity marker that no second-hand report can move is, in Ginzburg's phrase, "furniture" — apparatus generating precision the evidence does not hold.

WHAT THIS COSTS THE PROJECT:

Split Atwood's charge in two on the record. The rename discharges the disclosure half. The passability half (Kiberd) is open and needs a first-reader entrance with no anchor codes, or an explicit refusal to build one.
Bring dimension 9 into compliance with A19 before it is cited again as the project's reading: eighteen rows, Homeric occasion and Joycean passage as printed, and "We're square" carrying episode, speaker and page by publication or losing its quotation marks tonight.
Publish dimension 9's coding sheet with dates. If "transacted" was defined after the formula was written, demote the nine from a count to an illustration, in the document's own voice.
Adopt the discriminating test as a published standard: state for each dimension whether it makes an attackable claim or reports a distribution, and accept that only the first class earns the apparatus around it.
Answer Ginzburg on the error record: state how many citations were spot-checked, and which of the 97 catches came from procedure rather than luck — or accept that the record is a diligence display.
Log that COMPETES is unanswerable under the current build, and that the reason is dimension 15's design. That is a permanent limit, not a pending question.

FAULT LINE: THE EVIDENCE ASYMMETRY — one hearsay term on a measured grid

1. The agreement that dissolves under one question

Eleven of the thirteen openings attack dimension 16's Nolan column. That looks like consensus and is not, because they diagnose three incompatible faults.

The first group indicts the number. Shneiderman: the ~70% marker "should not be displayed at all," because "a qualifier is text, while a number on an axis is a position, and the eye reads position before prose." Kenner sharpens it to a provenance failure — "a number with no narrator" — and locates the error not in the claim but in the resolution: "a percentage asserts that a measurement was made, and none was." Ginzburg agrees the refusal to move it "was correct reasoning that produced the wrong object," and calls the result "furniture." Borges adds the one point nobody else saw: a figure held unchanged while reports accumulate "does not read as held; it reads as confirmed."

The second group indicts the drawing. Warburg: the Nolan column drawn at the same weight as Homer, Wilson and Joyce is "a typographic lie," and the project already owns the fix — dimension 14's twenty-six empty keylines at full footprint. Cortázar, independently, arrives at the identical remedy and names the deeper omission: dimension 16 "separated two axes and left a third unseparated: evidence grade." Eames states the mechanism — "a tilde does not survive being plotted" — and Kiberd states the consequence for the only person who matters: "A table is itself a claim of comparability… The first reader will not read the provenance note. She will read the grid." Atwood adds the per-cell demand, and the observation that the asymmetry runs along a line the project did not choose: verified print collation for the woman translator, permanently unverifiable report for the male director.

The third position, held alone, indicts the term. Saramago: "The project refused to move the ~70% figure; it did not refuse to print it. Refusing to update a number is a much smaller discipline than refusing to state one." Strip Nolan of any axis position until the project holds the film.

These are not three intensities of one fix. They are three different claims about how much of the dimension survives.

2. Pamuk ends the redraw option

The strike that resolves the tier disagreement comes from Pamuk, and no one else made it: the Nolan column "is the SOLE occupant of the ADD pole. Remove it and the central contrast, ADD against REMOVE, has one example on one side. That is not a comparative dimension; that is three cases and an anecdote."

This is fatal to the Warburg–Cortázar remedy taken on its own. Keyline the column honestly, at full footprint with no ink, and you have not protected the finding — you have drawn a pole with nothing in it. The dimension's headline result is that Wilson and Joyce share an anti-mystification impulse, both removing, while Nolan adds. The contrasting term is the hearsay one. Warburg's rule — "evidence-weight is drawn, not stated" — assumes a drawing can be honest without being empty. At this pole it cannot. Warburg has not conceded; his own rule concedes for him, and the cost is high: the visual grammar he correctly identifies as the project's best invention turns out, applied here, to evacuate the finding rather than qualify it.

What survives Pamuk's strike is worth naming precisely, because the panel did not name it. The two-axis result — collapse the works onto one scale and they line up, separate the axes and they stop lining up — is demonstrable on Homer, Wilson and Joyce alone. So is the supply to dimension 9: refusing plot and refusing register are different acts. Both are three-point findings. It is only the ADD/REMOVE contrast, and only the three-way separation that uses the reported film as its middle term, that depend on hearsay.

3. The mirror reading, at full strength

The chair asks whether the asymmetry mirrors its subject. Argue it properly.

The Odyssey is a poem about a man known only by report until recognition. Telemachus goes to Pylos and Sparta to collect hearsay about his father; the news that reaches him is second-hand and third-hand; the household legislates on rumour; the man himself arrives disguised and is disbelieved. A comparative dimension in which three terms are held texts and one is reachable only through report is, on this reading, not a hole but an enactment: the fourth Odysseus is in the condition of Odysseus. Borges is the reading's natural advocate and comes closest to it: the Nolan half is "the only pure specimen of the third duration it possesses. An event, before it becomes a structure, is a rumour with a date," and the four permanently closed questions are "the method's only honest photograph of the present tense."

Now break it, on Kenner's blade. Homer's reports are attributed. Menelaus has his information from a named source, the beggar's lying tales are identifiable as tales because you know who is telling them, and the drama of the poem is precisely the auditing of testimony against its teller. Kenner's objection to the ~70% marker — nobody can say "who computed it, from what, against which text" — is not a bureaucratic complaint. It is the exact property Homer's rumours possess and dimension 16's cell lacks. An unattributed report is not the Odyssey's epistemic condition; it is the failure of that condition.

And the second break, which is decisive. In Homer recognition arrives. Under C-31 four questions about the 2026 film are closed as permanently unresolved — no screening, no script, no library visit, ever. A mirror in which recognition can never occur is not a mirror of the Odyssey. It is a mirror of the suitors' Ithaca: a household reasoning confidently about a man it has decided will not return. The mirror reading is available, it is beautiful, and it licenses nothing — least of all a coordinate. Recognition-by-report is a condition to be shown, never a warrant to average.

4. The rescue, and what it costs

Galeano supplies the only construction that survives cross-examination: change the object. "That comparison is legitimate on second-hand ground because it compares what is claimed, not what is filmed. Mark it that way and it costs nothing." Compare the reports as reports — three positions taken on the same twelve lines of Book 22, of which, on the reports as held, only one looks.

Senn demonstrates the same discipline without theorising it, and is the only advisor who extracts value from the Nolan material while never placing it on an axis: if a modern Odysseus is given forgetting as his mechanism — REPORTED, second-hand, Calypso's arc as erasure-then-release — that is the one instrument Joyce refuses his man, whose day is a failure of forgetting. A report used as a prompt against a verified text costs nothing. A report used as a term costs the dimension.

Concessions, and their price. Cortázar's is the largest and he made it unprompted: he performs the most attractive rhyme in the session — the REPORTED exile ending read as relocation of the household rather than deletion, dimension 9's own idiom — and then voids it himself: "it is worth nothing, because it rests on report." He surrendered the session's best available flourish to hold a line. Borges concedes the same ground more quietly: having established the rumour-with-a-date as a legitimate third-duration object, he declines to let it defend the coordinate and turns on the number instead. Ginzburg concedes that the author's correct reasoning produced an incorrect object — an unusual admission that good method can fail at the render stage.

FINDING. The asymmetry is illegitimate as a fourth case and legitimate as a second object. Dimension 16 must be re-founded on three plotted works — Homer, Wilson, Joyce — which carry the two-axis finding and the supply to dimension 9 intact. The reported film moves to a separately-denominated register comparing claims about the film, never the film, and holds no position on either axis. The ADD/REMOVE contrast is withdrawn as a finding of the dimension and restated as a question the evidence cannot close.

DISSENT — SARAMAGO, at full strength, and it is aimed at the panel's favourite remedy, not at the finding. He has already refuted keylining in his own opening: absence drawn is "absence made handsome," and "a reader cannot tell not found from not chosen from does not exist." A hollow Nolan column at full footprint therefore does not report second-hand evidence; it promotes it, giving hearsay a designed shape and a reserved seat on a measured grid. On his reading, a separately-denominated register is the same evasion at one remove — the project keeping the column while appearing to give it up. His remedy is removal, and his warrant is the one thing the project has that nobody can take: "a project whose entire authority is that it says what it does not have must be strictest exactly where it is being invited to be precise." Record it. If the register drifts back toward parity of layout, he was right and the finding was a compromise.

WHAT THIS COSTS THE PROJECT:

Dimension 16 loses a case. Rebuild it as three plotted works; restate the headline without the ADD pole.
The ~70% marker comes off the axis. Kenner's demotion to a sentence and Shneiderman's deletion converge: a band as wide as the axis is not a band, it is an admission.
A new register must be written and denominated separately: reports held about the 2026 film, with each item's status, the date it entered, and the count of reports standing when the register was struck (Borges).
Dimension 9's frame must be re-derived from Wilson and Joyce alone, and any inherited reliance on the reported film flagged in the concordance's own voice.
One rule to publish under A10a: no second-hand term may occupy a position on any measured axis in any register — HTML, PDF, deck or diagram.
Saramago's dissent published beside the register, not beneath it.

Fault line: the hanged girls

The charge, and who is under it. Atwood put it in terms nobody at the table could soften: Book 22 "contains two slaughters, not one." The suitors are killed; twelve enslaved women are then made to clear the hall and are hanged for it, hanged specifically so they will not get a clean death. Her indictment of dimension 9 is procedural, not moral: if the concordance carries one row per episode, and the row at the position of the slaughter records the suitors and grades the book's answer as a schedule of remedies ranking hushmoney above assassination, then "the project has recorded half of a double occasion, and the half it dropped is the half done to women." She names the operation: "the men's deaths are the event, the women's deaths are the housekeeping afterwards." The project reproduced the scene's own accounting.

That charge lands on dimension 9, and dimension 9 has no defender here. Senn, its harshest reader, is already demanding that all eighteen rows print their Homeric occasion and their Joycean passage under A19. Ginzburg has already said the load-bearing wall is the negative over a closed set, checkable by anyone, and that "one clean deletion anywhere kills it." So the row Atwood wants is the row Ginzburg licensed. That is the session's first real convergence, and it is expensive on both sides.

First exchange — may the row be run at all? Ginzburg's knife: "Unless the coding rule for transacted was written down before the eighteen were graded, that nine is a number about the grader's degrees of freedom." Borges brought the same blade independently — if each episode was coded after "We're square" was written, "the nine is an illustration wearing a count's clothes." A nineteenth row drawn now, at the one position an advisor has just named as an embarrassment, is precisely the post-hoc row both men rule inadmissible as evidence.

Atwood escapes this only because she pre-committed. She marked her own reading — that hanging is present in Ulysses, relocated onto male, national, named bodies, made into martyrology and a joke — as inference, "to be checked against the printed 1922 text under A19 before anyone prints it," and asked that the row be published "whatever it returns." That is the concession, and it is not small: her most quotable claim of the evening cannot be published in this session, and if the check returns nothing, it is dimension 9's "nothing is deleted" law that breaks, at a position she chose. She staked her thesis on a test that can destroy the finding she is trying to enlarge. Ginzburg's counter-concession: he must accept a row he would otherwise call fitted, provided the coding sheet is dated first. Sequence is now load-bearing — the row must be specified before it is graded, or it is decoration.

Second exchange — will dimension 2 carry the weight? Atwood leans on it: Milly, the only young woman with a stated wage, reaching the reader almost entirely as paper; the accounts owed to women being the accounts that never settle. Pamuk and Warburg say it can carry it. Pamuk grades the interiority/independence inversion the strongest of the nine under his roster-perturbation test, "because it is a NEGATIVE relation nobody went looking for." Warburg agrees for a different reason — it is "a correlation between two independently-scored axes rather than a distribution over one."

Kenner denies both. "Molly's inside is a chapter allocation. Milly is paper because Mullingar is offstage. Move the compositional facts and the correlation moves with them." Shneiderman, from the other discipline, arrives at the same verdict — most quotable finding in the build, most fragile, "recode Molly or Milly once and the line moves." Ginzburg: a pattern claimed from perhaps four load-bearing instances.

This does not resolve, and the reason it does not resolve is Saramago's: the unit is unpublished. "If a letter counts, a photograph counts, an arm counts, and being the subject of men's talk counts, then mediation is generous by construction." Pamuk's perturbation test and Ginzburg's coding sheet are the same missing document, and until it exists nobody at this table can say whether the two axes were scored independently or whether one was scored off the other. Pamuk's robustness claim is therefore currently unsupported by anything but his confidence — a concession he half-makes himself when he says the Object Museum "publishes its catalogue and withholds its accession policy," which costs him the affect his own museum practice trades on.

Third exchange — three works, or two and a rumour? Galeano offers the comparison as safe: three works take three positions "on the same twelve lines, and only one of them looks," and because what is compared is the report rather than the film, marking it "costs nothing." Two objections break it.

First, factually. Homer looks. The poem narrates the hanging; it is the demotion the poem performs, not the aversion. Atwood's own framing is the correction — the twelve are counted and disposed of as housekeeping, which is worse than being unwatched. So Galeano's ordinal has the wrong shape at position one, and an ordinal with a wrong terminus is not a scale.

Second, evidentially. Saramago: "Refusing to update a number is a much smaller discipline than refusing to state one." Pamuk sharpens it to a structural fact about dimension 16 — the REPORTED Nolan column is the sole occupant of the ADD pole; remove it and the central contrast has one example on one side. Kiberd names the mechanism: "A table is itself a claim of comparability." Warburg and Cortázar independently supply the same remedy from dimension 14's grammar — keyline at full footprint, no ink. That is where the room actually agrees, and Galeano's "costs nothing" is the minority position.

Cortázar's contribution needs the same discipline turned on it. He told the table the poem counts twelve of the household's fifty and names one, Melantho, then marked it himself: "I offer that as a reading, not a count." Correctly — this project has already struck two untraceable figures (12,110; "173 versus 18"). Senn conceded the general form of this for everyone, including himself: "that is what I have been doing all evening — characterising passages I have not shown you." Every Homeric particular in this session is provisional. That concession costs Senn his own Book 24 orchard argument until someone prints the lines.

The straw Ginzburg. Atwood, Galeano and Saramago each broke Ginzburg on against-the-grain reading — a novel has no archival residue, Menocchio survives because a hostile institution wrote him down, this reader has no adversary. All three are right, and none of them was answering Ginzburg, who advanced no such transfer; he spent his opening demanding inclusion rules and declining Borges's compliment. The consequence matters more than the misfire: with recovery unclaimed by anyone, the register of the unnamed has no proposed epistemology. Nothing on this table can retrieve the servant girl at the pork butcher's counter, whom Atwood correctly identifies as Ulysses' own dmoe. Galeano states the terminal fact: against-the-grain reading of fiction "recovers Joyce's decision not to draw her, which is a different and lesser object."

FINDING. The comparison establishes one thing that survives, and it is not a comparison of the three works' merits. It is that at this single position four instruments fail in the same direction. Homer counts the twelve and does not name them. Wilson, whose stated policy is that softer words would imply freedom, writes "these girls" at the exact moment of the hanging — Pamuk's reading, and the finest instance of L-17 the project owns: a translator's own qualifier failing to reach her hardest scene, while her refusal holds: "I refuse to grant these girls a clean death." REPORTED and second-hand, permanently so under C-31, Nolan's film elides the hanging onto a single betrayer; no report is verified and none may be treated as established. And dimension 2 admits by name and therefore cannot reach the unnamed servant, which is L-38 with a class attached — Ginzburg: "the class it cannot reach is not random." The blind spot is at the same coordinate every time: the woman who is owned, counted, and not named. Senn supplies the long duration for it — Eurycleia, the slave who performs the first recognition and is throttled into silence by the man she recognises; Shneiderman supplies the mechanism — the weakest credential is read first by the person with least standing to act on it. The household's women are its evidence-bearers and its disposables, and that is one finding, not four.

What the comparison does not establish: anything about Ulysses' treatment of the hanging (unrun, and unprintable under A19); any ranking that includes the film; any strengthening of dimension 9.

DISSENT — Saramago, at full strength. The three-way comparison should not be published in any form. A project whose entire authority is that it states what it does not hold must be strictest exactly where it is invited to be precise. Comparing claims still seats a rumour on an ordinal beside two printed texts, and no label drains precision back out of a placed object. Strip the Nolan column of any position, not merely of movement.

SECOND DISSENT — Kenner, joined in effect by Galeano. The register of the unnamed will return compositional decisions, not women. It will be a monument dressed as a census, and dimension 2's correlation, on which the whole fault line rests, is chapter allocation and offstage geography. Atwood's reply, unrebutted: a denominated absence at full footprint is what the Atlas already does twenty-six times, and nobody mistakes those for holdings.

WHAT THIS COSTS THE PROJECT:

Write and date dimension 9's coding sheet for transacted before grading anything new; publish the date beside the date "We're square" was written. If grading followed the formula, demote the nine from a count to an illustration.
Pre-register, then run, the Book 22 double-occasion row — the hanging as an explicit test of "nothing is deleted" — and publish it whatever it returns, with both texts shown under A19. Accept that it may break the law.
Publish dimension 2's appearance-unit rule and a second count under a strict definition. Until then the interiority/independence inversion is unresolved between Pamuk/Warburg and Kenner/Shneiderman/Ginzburg, and must be shown as unresolved.
Add the register of women drawn and never named — the pork butcher's servant first — separately denominated, uncounted, drawn as keyline at full footprint, with a stated admission rule per Borges or no n at all.
Take the quotation marks off "We're square" tonight unless episode, speaker and page ship with it.
Nolan: keyline the column, strip the ~70% position, and record Saramago's dissent that keylining is insufficient.
Accept Kiberd's cost: six of these seven items are qualifier apparatus, and only item 4 is content. If only one ships this cycle, ship item 4.

The adversarial verifier

Run after the session with its default set to REFUTED and one instruction: find where this session overclaims. Printed unedited, including its two false accusations against the project — which the chair's record above corrects, and which are the most instructive thing in it.

The verifier's report

1. Nolan presented as established — NOT FOUND. I checked all thirteen openings and all three fault-line syntheses. Every Nolan particular carries a marker before or inside the sentence: Warburg's "reportedly Sinon imported from Virgil"; Pamuk's "As REPORTED, the 2026 film ends in exile"; Cortázar's rhyme voided by its author — "it is worth nothing, because it rests on report"; Galeano's conditional "if that elision is as reported." The chair's hanged-girls FINDING is exemplary: "no report is verified and none may be treated as established." Whether Nolan mentioned Joyce is left unevidenced by every voice. This category is clean, and I decline to invent an entry for it.

2. Invented or garbled — one systemic error, repeated by seven voices.

The line "I refuse to grant these girls a clean death" is spoken in the poem by Telemachus. It is Wilson's rendering of a killer's stated cruelty, not a translator's statement about her own practice. The brief says only "Her line is…" — a rendering. The session escalates that into authorship, and does so most explicitly where it matters most:

Atwood: "What is verified is the refusal, in her own words: 'I refuse to grant these girls a clean death.'"
Pamuk: "a translator signing her arrangement in the first person, which is A10b performed by somebody else's book."
Borges: "a translator's act, not a lexical one."
Kenner commits it in the same paragraph in which he states the rule that forbids it — "a note is a voice, a passage is a voice, and they are not the same voice" — then: "What Book 22 does support is the other half — 'I refuse to grant these girls a clean death'."

This is the identical failure the project already caught once (policy lifted from the Note and dropped on the scene) performed in the opposite direction, and no voice catches it. A10b is void here; the "I" is a character's. Verify against Wilson 22 before anything ships.

Second: Shneiderman invents a project asset. "Ithaca is a query interface, and the project's own count — 303 question marks there, none at all in Penelope — is evidence it has not yet spent." No such count is in the brief. A precise figure attributed to a project that has already struck two untraceable figures (12,110; "173 versus 18") is the third instance of the same disease, and it is load-bearing for his recognition argument.

Third, minor: Kiberd states the 1925 Borges essay as biographical fact with no marker, while Ginzburg — same claim, same session — grades it "high but second-hand." One of them is non-compliant; on the project's own standard it is Kiberd.

Borges's Book 13 tripod-count, Pamuk's Eurymachus offer, Saramago's area railings, Atwood's pork butcher's servant and Senn's Book 24 quantities all appear to be sound but are characterised without being shown. Senn conceded the general form ("that is what I have been doing all evening"); Cortázar marked his own Melantho figure "a reading, not a count." Nobody else marked theirs.

3. Pattern from too few instances.

The session's largest inference rests on three tokens. "Homer's recognition tokens cannot be transferred and cannot be forged" is asserted by Kiberd ("Every one of Homer's recognition tokens has the same property"), Ginzburg, Warburg, Cortázar, Pamuk, Borges and Eames — from scar, bow, bed. And the counter-instance is at the table: Senn's Laertes orchard and Eames's own "But the last one does not" — an itemised, recitable, transferable recognition. Borges adds a second (the Book 13 count). A universal from three, refuted by a fourth in the same room, and no voice reconciles it; the chair logs the orchard as an attack on dimension 9 and does not notice it also removes the premise five other positions stand on.

Turn Ginzburg's own rule on the seven: "Two documents agreeing are not two witnesses when they share a source." Seven arrivals at non-transferability are one argument counted seven times.

4. Claims about the project that contradict or exceed the brief.

(a) The rename. "The project's answer is the rename — Ulysses: a first reading, and the apparatus built around it." No rename, and no such title, exists in the brief. An entire finding is built on it ("a real answer to Atwood and a null answer to Kiberd"). Unsupported.

(b) The chair's tally is false on its own transcript. "Two subtractions against eleven additions… A body that answers 'too much apparatus' with more apparatus has not answered the question; it has voted on it." At least five further positions contain explicit removals: Shneiderman ("strip the seventy-per-cent marker"), Kenner ("demoted from a number to a sentence"), Saramago ("strip the Nolan column of any axis position"), Pamuk ("demote the Nolan column… to a dated rumour"), Borges ("demote the nine from a count to an illustration"). The punchline is manufactured by an undercount.

(c) Self-contradiction inside one fault line. FINDING: the three plotted works "carry the two-axis finding and the supply to dimension 9 intact." Six lines later: "Dimension 9's frame must be re-derived from Wilson and Joyce alone, and any inherited reliance on the reported film flagged." Intact and requiring re-derivation cannot both hold; and note the silent drop from three works to two. The chair also asserts the two-axis result "is demonstrable on Homer, Wilson and Joyce alone" without demonstrating it — Nolan is the term the brief places off the shared impulse, so his removal is exactly what might collapse the separation.

(d) "Dimension 9 has no defender here" is contradicted by Warburg ("stronger than the project has stated it"), Eames, Shneiderman ("the thesis extends") and Galeano, all of whom enlarge it.

(e) Verified and holding: "Eleven of the thirteen openings attack dimension 16's Nolan column." I counted. Senn and Galeano do not; the other eleven do. Correct.

5. Strongest and weakest.

Strongest: Pamuk on Eurymachus. "With Antinous down, Eurymachus offers restitution… Odysseus declines it and takes the act instead… Joyce takes the branch Homer declined." It lands a specific, checkable Homeric passage at the exact coordinate dimension 9 grades; it reverses the finding's direction without inflating it (the fork is Homeric, so "we're square" is a chosen branch, not an invention); and one printing refutes it. Atwood's pre-registered Book 22 row is the better method — she staked her thesis on a test that can destroy it — but it is marked inference and unrun. Pamuk's is text now.

Weakest: Galeano. "Dimension 9 has not documented a literary refusal. It has documented the moment power went anonymous, and it has the receipts." Eighteen rows in one novel, graded by one reader in eleven days, elevated to a world-historical transition — asserted with the exact metaphor the session reserves for verified evidence. No possible row could disconfirm it, which is what makes it worthless as a finding. It is also self-cancelling: the same voice then demands every dimension publish "a single line naming what the method cannot see."

What was done the same night

Sessions in this project are judged by what they changed, not by what they said. Three things shipped before this record was closed, and one of them was caused by the chair while fixing the first.

Catch C-98 — dimension 9 now shows its own proof text. Bloom's words in episode 15 are printed in the artifact and in the dimension record, with the stage directions, the speaker and the surround; and the un-settling four lines later — This is yours — is shown beside them rather than smoothed away (A4). concordance.pdf rebuilt: 38 links, 19 destinations, outline intact.
The chair broke A19 while fixing A19. The first version of the restored passage was typed with a straight apostrophe where the held transcription prints a curly one. The quotation inserted to satisfy as printed was not, itself, as printed. It was caught by verifying the insertion against sources/ulysses-1922-pg4300.txt before the commit, which is catch C-58's family exactly — three of the reader's own sentences once shipped with a comma where the record had a full stop, because that is what careful sentence-writing does to a quotation. A one-character deviation from Joyce, produced by the act of correcting a deviation from Joyce.
The prose filter reached two fields it had missed. Writing the correction into the dimension record turned the build red: §11y found raw asterisks on the concordance's page, because catch C-81 had put note, unit, confidence and sources through the renderer and left claim and shows printing raw. They had held no markup, so nothing was visibly broken and no gate had ever fired — the vulnerability was invisible until the first sentence with emphasis in it was written. That is L-33 once more: the remedy scoped to the fields where the defect was noticed rather than to its class. Both now render.

A limit, stated rather than implied. The substrate's mechanical verification covers the 714 quoted units of the three inventory artifacts and does not reach the concordance. The passage restored above is verified by hand, and by this record, and not by a gate. Building that gate is real work and is named here rather than counted as done.

Provenance of this document

Generated from joint session 09, held 28 July 2026. Seventeen agents: thirteen openings, three cross-examinations, one adversarial verifier. No agent errored and none returned empty.

The brief was regenerated from the working tree on the morning of the session, because build panel session 07 caught the chair feeding a panel stale numbers and four of its arguments rested on figures that had already moved. It did not save the chair this time either: the brief was complete enough to reason from and short enough to produce two false accusations, which is recorded above rather than tidied.

Every claim about Christopher Nolan's 2026 film in this document is second-hand and marked as such. The project holds no primary source for it and by the author's standing ruling never will. The verifier checked all thirteen openings and all three syntheses for that discipline specifically and found it clean — and declined to invent a finding for the category, which is the correct behaviour for a verifier and is noted here because it is rare.

Joyce is referenced rather than quoted at length throughout, under anchor A19. The one passage quoted in the chair's record — Bloom in episode 15 — was located and verified in the project's own held 1922 text before it was written down, and the verification is what produced the session's live catch.

The advisors and masters are constructed critical lenses named for people whose working methods they abstract. They are not those people's opinions and nothing in this document should be quoted as such.