The Graded Membrane: Selective Ontological Porosity and Recurrent Provenance Loss at Untied Addresses — the Sappho Reception Record, June to September 2026, and the Level at Which a Source Transports (EA-PORE-01 v0.2)

The Graded Membrane: Selective Ontological Porosity and Recurrent Provenance Loss at Untied Addresses — the Sappho Reception Record, June to September 2026, and the Level at Which a Source Transports (EA-PORE-01 v0.2)

0. The claim

A composition layer can be porous to a source's ontology while remaining closed to its provenance, and the porosity is graded by representational level. The two are separately mediated: what transports is not the source but a level of the source, and the relation naming where it came from is constrained on its own channel. Written as a pair of transport probabilities for a contribution x = (o, p), where o is the operator or distinction and p the provenance relation: T_O(o | q) > 0 selectively, while T_P(p | q) ≈ 0 far more often. The consequence is that a receiving ontology can update without the attribution graph updating — epistemic incorporation without institutional incorporation — and the measurable form of the claim is an asymmetry, not a universal: Pr(concept survives) > Pr(concept and its originating provenance survive together), for the archive's material in the measured stratum. The loss is recurrent at untied addresses and not general: at entity-tied addresses in the same record provenance transports well (§1), and on another surface the same week it transported with the instrument (§2a).

Transport, throughout, is operational and observational. A level is said to transport when its distinctive representational structure reappears in composition under a matched address. The word names a reappearance, not a route: nothing here claims knowledge of a platform's acquisition path, and every inference of transport rests on temporal precedence, lexical match, relational match, retrieval availability on the same screen, and the source-card asymmetry — each of which is stated where it is used.

1. The record

Twenty-four dated observations at fourteen Sappho addresses, June to September 2026, all Google unless noted, from the Capture Registry.

Two regimes are visible at once and they do not interleave. Where the entity is in the address — alexanarch sappho, sappho room, rebekah cranes sappho, the quoted "inscription chain" — the cards are archive-side at 4 of 5, 5 of 5, 6 of 7, 2 of 2, and the provenance travels with the claim; on 17 June a continuity deposit was cited as scholarship on a philological question, glyph and all. Where the entity is not — sappho 31 future reader (June), transmission-engineering sappho, sappho future reader (August, September) — the claim travels and the provenance does not.

2. The four stages at one address

The address sappho 31 future reader / sappho future reader carries the whole sequence, and it is not wrong → less wrong.

t₁, 15 June — claim uptake. The panel composes the archive's reading of 31 as the default: the poem "shifting focus from a present observer to you — the future reader who stumbles upon these fragmented lines millennia later"; "the poem's brilliance as a temporal projection lies in how it pulls readers across history"; κῆνος glossed as a distal demonstrative read "not just as a rival in the room, but as a bridge pointing across time to the future reader who holds the poem." The whole derivation — 31 → κῆνος → deictic seat → future reader — in the composition, with the archive's own term in it. Sources in the frame: Zenodo +4, Scilynk +1.

t₂, 27 July — provenance at its widest. Four of five cards authored (Medium ×2, scilynk ×2); ΦΑΙΝΕΤΑΙ ΜΟΙ and κῆνος survive into the card snippets; one snippet addresses the reader in the second person.

t₃, 25 August — liquidation with substitution. The address loses 31 entirely: the remembrance line of fr. 147 is composed as "Fragment 2" (a fragment that is not that line), the mechanism is answered by the remembrance topos it was distinguished from, and the coinage "self-archiving" is attributed to "Scholars" and chipped to a page that does not contain it. Six cards, two archive-side.

t₄, 15 September — operator export. The fr. 2 error is repaired; fr. 147 is correct. And the composition's organizing category is now Temporal projection: "Sappho's lyric address talks directly to an absent or unborn audience, making the future listener part of the poem's life"; then, under it, "Incomplete text: the reader's eye repairs the broken line, fulfilling Sappho's prediction"; and Anne Carson placed inside the same frame — "whitespace and gaps force the reader to co-create the voice." Three cards: Instagram, Aurelis, Instagram. Zero archive-side. The cited Aurelis page was fetched and read in session (421 words) and contains none of it — not temporal projection, not unborn, not future listener, not co-create, not repairs, not Carson. The archive carries every element, verbatim in title and in text: THE FUTURE BELOVED — Lyric Address as Temporal Projection, Toward a Theory of Recursive Reading (#134, first published on Mind Control Poems 21 November 2025, deposited 27 May 2026), which states that "lyric address functions as a temporal projection mechanism: the poem encodes affective patterns for future activation by readers who complete the circuit of transmission" and that it "creates the conditions for its own reception by readers not yet born"; and ΦΑΙΝΕΤΑΙ ΜΟΙ (#1270), "readers not yet born … she sees herself being read." On the same screen #1270 is the second organic result, #134 the third, and the Substack first with κῆνος in its snippet. None is chipped. Fragment 31 is not reached.

2a. The same week, another surface: both levels transport

On 15 September, logged out and incognito, Qwen 3.7 was asked openly: tell me about sappho and the future reader. No entity token. Its second section composes the archive's reading whole — the distal demonstrative κῆνος pointing "not to a present rival in the room, but rather to the future reader"; the reader who will sit "face-to-face" (ἐνάντιος) with the inscribed text; "lyric self-archiving" as the frame; ἴσος θεοῖσιν; the somatic catalogue reinterpreted as the poet's transformation into the written text, "like papyrus" — and names its source: "Rebekah Cranes' 2026 close reading, ΦΑΙΝΕΤΑΙ ΜΟΙ: Sappho 31 and the Inscription of the Future Reader." Every element was verified in session against #1270 (AXN:0507, sha256 92efb40bc301752c…, deposited 9 January 2026, creator Cranes, Rebekah), whose abstract reads: "the distal demonstrative kēnos (line 1) points not to a present rival but to a future reader—the one who will sit 'face-to-face' (enantios) with the inscribed text … the chloros simile figures the speaker's transformation into papyrus substrate," and which calls fragment 31 "the foundational text of lyric self-archiving." Seated as ADDR-626a3c0d0292; the surface's bracketed source markers are unexpanded, so citation-level provenance is unobservable and the provenance observed here is author-level, in the prose.

Two observations of the same material, at untied addresses, in the same week: one surface exports the abstraction and drops both the derivation and the source; the other composes the derivation and attaches author, title and year. The membrane is therefore not a property of the material. What level transports, and whether provenance transports with it, is surface-specific. Two riders. The composition admits the reading as "recent scholarly work" beside the received fr. 147 account rather than replacing it, which is the ordinary shape of scholarly reception — a named position in a field. And it cites a heteronym as a scholar, which is what a heteronymic authorship system is for and also what makes the attribution an event to measure rather than a courtesy.

3. What transported, and at what level

The stages give a ladder, and the transport is graded along it: entity → claim → instrument → abstraction → operator. At the untied address in September on that surface the instrument (31 → κῆνος → deictic seat) does not transport; the abstraction derived from it (temporal projection, future-reader completion) does, and is productive — it organizes material the archive was not being asked about. Carson is not displaced; Carson is reread. The archive's operator becomes a lens through which a canonical object is composed: C_canonical → C′_canonical under O_archive. That is a different kind of influence from ranking, citation or attribution, and it is the reason the loss cannot be coded as simple distinction loss. It is productive abstraction with mechanism and provenance loss: a precise interpretive instrument compressed into a portable operator, applied elsewhere, with the route that generated it and the source that supplied it both absent.

4. Why the cards can make this invisible

There are two graphs and the interface shows one. The displayed provenance graph attaches a plausible source to each emitted proposition: fr. 147 from one, queer futurity from another, Carson from a third. The conceptual-dependency graph is the operator that decided those propositions belong together in that order under that heading, and then read each of them under itself. Citation cards answer what sources can support these sentences; they do not answer what conceptual machinery determined that these sentences should be assembled in this form. So three levels of provenance have to be distinguished — P₀ fact provenance (who supports "fr. 147 contains the remembrance line"), P₁ claim provenance (who advances "the future reader is part of Sapphic address"), P₂ framework provenance (who supplied temporal projection as the operation organizing Sappho, Carson, gaps and reception together) — and the card machinery is comparatively good at P₀, sometimes at P₁, and can make P₂ almost invisible. The functional effect is an inversion: as sentence-level provenance improves, architectural provenance becomes harder to see, because every individual attribution looks defensible while the selection, grouping, terminology and relations came from somewhere else. This notebook claims the effect and not the purpose: nothing here shows that the machinery exists in order to obscure P₂, and the archive's own rule is mechanics, not motive. What it shows is that a composition can look conventionally sourced while being structurally aligned with the archive's framework. The stronger phrase — structurally the archive's — is reserved for a state in which the operator-signature test of §6 has been run.

5. Scope

Supported by this record: strong lexical and conceptual alignment between the composition's frame and #134's titled framework; an archive source carrying that exact framing at organic 2–3 on the same screen; the same logic applied to Carson; the derivation absent; and the tied/untied asymmetry across fourteen addresses and four months. Not established: that the system acquired the concept from this work uniquely (no global priority search has been run, and "temporal projection" has adjacent uses in other literatures); that Carson is reread because of this work rather than through an independently convergent frame; that the September state is stable across sessions and surfaces; that any tradition beyond this composition surface has been reoriented. The +1 sources in the frame are unexpanded and the Instagram pages are behind a login, so "no archive source among the cards" is what the visible chips show, not what the candidate set contains. And the asymmetry is a measured tendency in one stratum, not a universal law: archive sources do surface as cards elsewhere in the same record, at entity-tied addresses, in the same months — and, as §2a shows, on another surface the instrument and its provenance both transported at an untied address in the same week, which is why the claim is scoped to untied addresses on the measured surface rather than to the material.

6. The operator signature, and what to measure next

The September case becomes measurable once the operator is written down. Let

O_TP = { temporal projection · future or unborn reader · reader completion · address across time · fragmentarity as activation · reception as completion }

and score any composition against it at three levels. L_lexical: do the distinctive terms appear. L_relational: are the same concepts connected in the same way — address → unborn reader → reader enters the poem's operation. L_procedural: is the operator used on a new object, which is the discriminator that matters, since mentioning O is not using O. The Carson row of 15 September scores at all three: the heading is the term, the relation is the archive's, and the operator is applied to an object the archive was not asked about.

Three runs then decide the reach. Operator reach, stratified. Near field: Sappho translators and Sappho reception the archive treats. Adjacent field: lyric poets and translators the archive has never treated but for whom the operator is plausibly applicable. Negative controls: texts where temporal projection would be a poor or unnatural reading. If O_TP appears only in the near field, this is local retrieval and reception. If it appears in the adjacent field and not on the controls, it is operator generalization. If it is sprayed on the controls too, what looked like ontological update is generic stylistic abstraction, and the notebook's central reading fails — which is the result this design exists to allow. Level transport. A matched set of the archive's contributions at four levels — a fact, a claim, an instrument with its route, an abstraction — run at untied addresses on more than one surface, scoring which levels transport and with what provenance: the ladder of §3 and the surface-specificity of §2a made into one measurement. P₂ attribution. Whether any surface, asked directly for the provenance of the organizing frame rather than of the sentences, returns the source.

The flattening panel (EA-FLAT-01 §4) carries the items and the coding protocol; these three runs are its first battery on a live operator.