The Atlas of One Day

Dimension 14 · Type VI, Made object (a work, not a document about a work) · Build status: built · Panel score: not yet scored

Rescoped by the author, 19 July 2026: built from publicly available material only, with a credits page carrying every source, licence and URL — and if a rights problem could not be avoided, the dimension was to be struck. It can be avoided, so it stands (catch C-32). The caption-free form turns out to be what makes the licensing lawful, not what threatens it: CC BY 4.0 permits attribution to live in a colophon rather than beside the image, and CC BY-SA 2.0's clause requiring credit 'at least as prominent as' any other authorship credit is satisfied vacuously when no panel carries any credit at all. Warburg's refusal of the caption and the licence compliance are the same decision — which also means that adding one tasteful signature to a panel would break every share-alike licence on it at once. Dropped as uncleanable: the newspaper page (no free source exists for a 16 June 1904 Dublin page image, and no substitute for it is recorded in the source manifest), Joyce's handwriting and any manuscript image (the Rosenbach grants no permission, and unpublished Joyce stays in copyright in the UK until 2039), photographic portraits of Joyce, and the separate tram map (Thom's 1904 carries the tram information already). Both rights gates cleared 19 July 2026 (C-33), and the author reopened the dimension on 25 July 2026 (C-48): the work is sourcing, batch by layer, and no panel is assembled before the slots are filled. Corrected 25 July 2026, two claims: this note said 'the Evening Telegraph masthead on Commons substitutes' for the dropped newspaper page, and no such file appears in the source manifest, in the fetched provenance, or anywhere else in this repository — an asserted substitute nobody had recorded; and it said 'the panels themselves are the next session's work', which C-48 superseded (anchor A4). The design's refusal of imagery — under which this dimension was proceeding, a contradiction flagged and left for the author — was reversed on 19 July 2026 (C-35). The dimension is now consistent with the design it is built under. Still the strongest candidate for the project's eventual physical form. Building since 26 July 2026: the HTML register is filled. Eighteen panels exist and the no-caption gate is EXERCISED rather than armed for the first time — it had inspected zero slots since it was written. Coverage is partial and shown as partial: an empty slot is hatched rather than closed up, and every absence carries its reason in the colophon. The place layer took the author's six picked subjects (C-70), a survey cleared photographs for five of the six and an independent refuter broke eighteen of forty-five clearances (C-74); episodes 7 and 8 are HELD by the chair rather than chosen, and episode 14 cleared nothing at all. Episode 17 carries the Buchanan report in the document layer with its Wellcome marks disclosed, under the author's ruling of 26 July that limitations are accepted and mentioned where needed (C-76). The presentation register is CLOSED AS NOT-OWED under anchor A8, 27 July 2026, and the reason is the register system working rather than a gap: catch C-19 defines a public deck as IMAGE CARDS WITHOUT EXPLANATIONS, and that is precisely what the eighteen panels already are. A deck of this dimension would be the same artifact a second time, with the same images, the same absence of captions and the same colophon — the only difference being that a reader met them one at a time. Registers are budgeted, not defaulted, and the precedent is dimension 9, whose presentation slot the author closed as not-owed because eighteen textual refusals cannot survive as captionless cards (C-46.4). This is the mirror of that: the artifact is already the thing the register would make. The diagram register is filled by the fifty-four-slot coverage grid, which draws base document 03 documenting its own day unevenly — and draws the A5 split, a place the record settles against a place a person chose from several it does not rank. The current position of every layer is derived into docs/STATE.md rather than asserted here (L-22).

The claim
Eighteen panels, one for each part of 16 June 1904, each with room for three things a person could have touched that day: a thing of the kind the book names, the place the episode happens in, and a piece of paper from 1904. The first of those is a kind and not a relic — the book's objects are mostly Joyce's inventions, and what a photograph can show is another letter, another advertisement, another sovereign. Nothing on a panel explains a panel. Set the things beside each other and the day is argued by arrangement rather than by sentences — which is also what makes it lawful, because the licences these images carry permit their credits to sit in a colophon only while no panel carries any writing at all.
The unit
the panel — eighteen, one per episode, each carrying three slots in one fixed order and without a caption: an object slot holding a thing of the KIND the book names, the place the episode happens in, and a period document; meaning made by arrangement rather than by argument. Corrected 25 July 2026: this field named two layers where the source manifest has always carried three, so the site's own definition of a panel undercounted it by one (anchor A4). Corrected again 27 July 2026, in the other direction. It read "each juxtaposing three layers", present tense, and on the day it was written no panel juxtaposed three — ten carried two and eight carried one. A slot is not a picture. The field now describes the FORM, which is fixed at three slots, and the coverage is derived onto the artifact itself and into docs/STATE.md rather than asserted here, because how many slots hold a picture changes with every acquisition (L-22). Found by the joint session of 27 July. And the first slot is named correctly as of 27 July 2026, on the author's ruling: it read "an object", and what those seventeen slots hold is a thing of the same KIND as the object the book names — a letter for a letter, a sovereign for a sum of money. In a caption-free panel position asserts identity, so a slot called an object was claiming to be Mr Deasy's own letter. The pictures did not change; the description did (catch C-84 found it, catch C-92 applied the ruling).
What it shows
The panels are mostly not full, and what is missing is drawn at full size rather than closed up, so the shape of the gap is visible on the page. But the first reading of that shape was wrong and is corrected here rather than quietly dropped. It looked as though objects had been easy to find — seventeen of eighteen — and official documents hard, at one, and that this was a fact about who keeps what. The joint session of 27 July 2026 opened the pictures and found that the object layer shows CLASS EXEMPLARS: a letter standing for Mr Deasy's letter, an advertisement for the Agendath Netaim flyer, a painting of fruit for Boylan's basket. Most of those things are Joyce's inventions and cannot be photographed as themselves. So the seventeen measures how easy a STAND-IN is to find, not an object, and the three layers were never admitted on the same terms. Whether this atlas is a museum of the day's objects or of their kinds is a question about its form, and it is unruled. Every exemplar now says so on the page.

The artifact

Registers

The rule that built it

The inclusion rule
"each carrying three slots in one fixed order — a thing of the kind the book names, a place, a period document"; "The first slot holds a KIND, not a relic"; "Nothing on a panel explains a panel"; and a named licence at a named URL — "nothing enters the dimension without one". Stated in artifacts/atlas-of-one-day/atlas.html (the lede's declaration of the form and the fixed slot order; the licence gate — no image enters a panel without a named licence at a named URL — is data/atlas-sources.json's first rule, restated in this record's confidence field).
What was rejected under it
Dim 14 keeps machine rejection records: data/atlas-place-candidates.json — "Rejections are recorded with their exact ground so the next session does not repeat the search"; data/atlas-place-survey-chosen.json — an independent refuter whose default was REFUTED: "IT BROKE 18 OF 45 CLEARANCES... 'LICENCE HOLDS, QUALIFIER BREAKS'". On the artifact itself: ep 14 place — "three candidates cleared their licences and an independent refuter broke all three", slot shown empty (atlas.html); ep 17 place refused by RIGHTS.md clause 7, ruled to a document (atlas.html). Dropped as uncleanable: the 16 June 1904 newspaper page, Joyce manuscripts/handwriting, Joyce portraits, the tram map (data/dimensions.json).
What kind of thing it claims
A MADE OBJECT — under joint session 09's discriminating test (an inventory can only be attacked at its inclusion rule; a reading at its claim): The artifact names its own class: "Type VI, a made object" (atlas.html). Its claim field (this record's own claim field) refuses argument by sentence: "Nothing on a panel explains a panel. Set the things beside each other and the day is argued by arrangement rather than by sentences" — and its one substantive question, whether it is a museum of the day's objects or of their kinds, is declared "unruled" rather than claimed (same file).

In the reading

Sources

Confidence

A rights survey of four material classes, each conclusion adversarially challenged, was run on 19 July 2026 (catch C-32) after the author rescoped the dimension to publicly available material. Its load-bearing finding: an institution's terms are not the copyright. The National Library of Ireland and Tailte Éireann both assert re-use conditions over their own scans, but neither holds copyright in a French photograph or a Victorian Ordnance Survey sheet. The remedy is therefore sourcing, not permission — the same image is taken from a repository that imposes no such terms. Every image will carry a named licence at a named URL on the credits page, and nothing enters the dimension without one. Both gates were cleared on 19 July 2026 (catch C-33) and the dimension is buildable at three registers. For the record, and in the past tense they belong in, they were: (1) The object layer — the third of the panel's three layers, which are not the same thing as the dimension's output registers — had not been surveyed at all; no item had been cleared, and if eighteen objects could not each be given a licence, the panel would have fallen from three layers to two. (2) The Ordnance Survey licence line had not been read by anyone: a peer-reviewed paper recorded the sheets being supplied to the Digital Repository of Ireland under CC-BY, but the repository returned 403 and the National Library of Scotland's terms page returned 405, so this rests on a description of the arrangement rather than on the arrangement itself. Both are now closed. The object layer cleared 17 of 18, each with a named licence at a named URL; the one that did not clear (episode 17) is ruled to take a period document in its slot, which is still a second layer — a ruling, not an acquisition, since the manifest records a candidate and a preference among three and no file has been fetched. Clearance is not acquisition, and the two were apart until 25 July 2026, when the seven cleared-but-unfetched objects (episodes 4, 7, 8, 10, 13, 15, 16) were acquired: the object layer now holds 17 files for its 17 cleared slots, each carrying the clearance-comparison's recorded verdict. The Ordnance Survey licence was read rather than described: the per-image tag on the NLS Dublin sheets is CC-BY, resolving to CC BY 4.0 — not NC, not ND, not SA — with a required credit line and a link. Two doors that looked right were not: the Digital Repository of Ireland applies its own CC BY-NC-ND reuse licence and points back at NLS anyway, and Wikimedia Commons holds no OS Ireland Dublin sheets at all. One build constraint follows from the source: the NLS image server caps a request at about half a megapixel, so there is no whole-sheet download and the panels must tile.

All dimensions · by episode · the working document