# [PAPER DRAFT] The relativity ingress experiment: exact closures, the fossil law, and the dissolution of the landscape

> **THE READING PLAQUE.** The second paper of the trilogy, draft 0.3 — **a supporting
> pre-print (the academic-research category of communication artifact), readable on
> the SX surface by its covering persona's reading grant; not an externally
> published form** (venue submission, if ever, rides the venue-attenuated 0.1
> lineage as its own act). Covering: **starl3n** · endorsing: **Link Digital** · the
> trilogy construct intent sealed 2026-08-01 (`ooi-paper-trilogy-construct-intent.md`;
> by the paper genre's law the body carries no persona mention — the construct
> standing lives here, in the register, and in the intent, never in the paper's own
> text). The grant: the operator's review and word, 2026-08-01, sequenced — **this
> paper second, granted after the first paper's readable surface stood**, so that
> this version can reference it; this is the named test's report of record, required
> for the correct continuity of the prior statements of intent (the first paper
> names the test; this paper reports it). Claimed only as a valid research summary
> of the noted experiments and results. Sources: draft 0.1 in the record (sealed
> `7f49e8c`, 2026-07-26); draft 0.2 the venue-unflagged staging (`8b5eed4`); this
> 0.3 adds the online-reading links to the cover note only. Its sha256 is pinned in
> the reading register. Reading is free; the way across is a statement of intent
> (`/sx/matters`).

---

**[COVER NOTE — not part of the paper.]**
Draft 0.3, 2026-08-01 — the reading version: supersedes 0.2 (2026-08-01, the venue
flagging removed) and 0.1 (2026-07-26); the only change from 0.2: the online-reading
links added to this cover note — the paper body is unaltered from 0.1. The companion
and sequel to the programme's first paper
(`horizon-first-paper-draft.md`, "Exact frontier reads of the primality stream by frozen
online learners"), reporting in full the test that paper named: the relativity ingress
experiment, run as **oxexp-relativity-ingress** through thirty-five sealed attempts. The
body is drawn from the witnessed record only (the sealed design document at its record
commits and the externally anchored dataset); nothing is claimed beyond the record.
**Online reading (this version's convenience):** the experiment surfaces this paper
references are live — the named test's instrument at
[dev.opendata.ai/rix](https://dev.opendata.ai/rix), the carrier's instrument at
[dev.opendata.ai/horizon](https://dev.opendata.ai/horizon), and its predecessor at
[dev.opendata.ai/walls](https://dev.opendata.ai/walls); the anchored datasets live at
[mldata.opendata.ai](https://mldata.opendata.ai). The first paper is readable on this
surface at [the readable first paper](/sx/reading/horizon-first-paper) — this paper
reports the test that one names. These links are a convenience of this readable
version, and it can be assumed that a future version will remove them as the
experiments evolve beyond the step of the paper's own claims.
**Register discipline carried from the
record:** every use of physics vocabulary is tagged **[R2]** (register-2: stated as the
experimentalist's reading, never asserted as physics); an editor may cut every [R2]
passage without loss to the mathematical content. The firewall sentence of the record is
carried verbatim in §2 and §8 and binds the whole paper.

---

**Author:** Steven De Costa (Link Digital)

**MSC 2020 (draft):** 11A41 (primes) · 68T05 (learning and adaptive systems) ·
68Q80 (cellular automata) · 11Y16 (algorithms; complexity) · 11S05 (2-adic valuation,
applications)

**Keywords:** pre-registered computational experiment · frozen online learner · exact
closure · window policy · 2-adic stratification · landscape laws · open science

---

## Abstract

The first paper of this programme closed with a prediction and a named test: that a
measured, computable differential between the standpoint at an origin and the standpoint
at any address constitutes a way to measure quantum relativity **[R2]**, testable only by
constructing a *relativity ingress experiment* with three required properties. This paper
reports that test, run to its measured boundary: thirty-five sealed attempts, every read
pre-registered before it ran, every verdict landed as committed, every close externally
anchored. The principal findings, each an exact count or exact identity: (1) closures —
two-carrier exact agreements on shared ground — grew 0 → 1 → 3 → 6 pairs while the
closure criterion was adversarially strengthened mid-campaign, ending in a fully-lived
quadruple of carriers agreeing pairwise at one exact count; (2) the read-state space of
the instrument closed at **seven exact states with zero unexplained reads** across the
whole record, and every read reduces to an 8-bit window policy composed with the
address's shape table; (3) **the fossil law**, the record's one declared law: a frozen
carrier collides at its own frontier **if and only if** it holds a one-bit trace of the
integer line's unique consecutive-prime event (2, 3) — forty-nine training lengths for
forty-nine; (4) a landscape campaign over the 2-adic depth of the training length
produced six held laws (a density floor, two valleys, a cross-era ordering at matched
depths, a crossover, and a two-sided depth band for the richest pattern family) and then
**dissolved every interval-like boundary candidate**: the fade, the era-end quiet, and
the gate's quiet band each proved a lattice coincidence of the depths measured — nothing
interval-like binds across strata; the era gates mark soft valleys; every stratum keeps
an intermittent tail on its own clock. On the test's own terms the first paper's
prediction is supported — repeated resolutions at the exactness standard under a
strengthened criterion — and its measured boundary is named: two of six surfaces reached
no face, and global binding is refused by the landscape itself. No claim on any
Millennium Prize problem is made, supported, or weakened by any part of this record.

---

## 1. The inherited prediction and its disciplines

The first paper's §6 stated, as a prediction and not a result **[R2]**: *the Horizon
experiment has provided a way to measure quantum relativity — the exact, computable,
carrier-invariant differential measured between the standpoint at an origin 0 and the
standpoint at any address N is the measurement, and the binary phase behaviour of the
freeze is the mechanism.* Three disciplines bounded it: the prediction asserts no
physics; it is not staked on further number theory; and its test is the named mechanism
of that paper's §7 — a constructed *relativity ingress experiment* — and only that.

§7 specified the test by three required properties, apparatus deliberately deferred:

1. **The ingress diode is constructed from being's side** **[R2]** — the one-way
   structure the seven Horizon campaigns had *measured* must be *built*, so that ingress
   is a designed property rather than a measured one.
2. **At least two pre-experiment computational dynamics project into the scenario** —
   fixed before any read, as the two standpoints of the first paper were fixed before
   its campaigns.
3. **Resolution at the record's exactness standard** — exact integer and rational counts
   only; a resolution requiring estimation or tolerance is a failure of the test's
   construction, not a partial success.

The sequencing was fixed in the programme's records: the prediction and mechanism sealed
first, candidate apparatus considered only after. This paper reports the apparatus as
built and the test as run — thirty-five sealed attempts under the first paper's
protocol, unchanged: pre-registration committed before every read; one read per block;
measured values pinned with live re-verification paths; counterfactual verdict branches
implemented and tested; verdicts landed as committed, kills kept with the same standing
as confirmations; the external anchor republished at every close.

## 2. The apparatus

**The constructed diode.** The apparatus re-reads the first paper's seven measured pins
exactly — 33 · 35-vs-139 · 34 · 23 · 141 · 168 — re-witnessed inside every one of the
thirty-five attempts by a calibration harness (seven for seven, every attempt). The
constructed instrument and the measured instrument are the same diode; property 1 was
discharged at the first attempt and held at every subsequent one.

**The six surfaces.** Property 2 was discharged with a lineage: six computable surfaces
were fixed before every read — deterministic indicator streams with exact span counts,
each derived from one of the six unresolved Millennium Prize problem statements *as
received texts*, and each carrying a printed **unknowing statement** naming what it does
not decide. The encodings are generation-lineaged: a failed encoding is superseded,
never erased, and no generation's failure reads back onto any conjecture. The six
problem statements served as the *field of uncomputed knowing* **[R2]** from which the
pre-fixed dynamics were drawn — and the record's firewall binds every section of this
paper as it bound every seal: **no claim on any Millennium Prize problem, ever; every
read is a statement about carriers, grounds, and exact counts inside the apparatus, and
about nothing else.**

**The verdict grammar.** Each attempt sealed its outcome assignments before reading —
RESOLVED against named gates, OPEN-EDGE otherwise — and closed witnessed either way. Of
the thirty-five attempts, the record contains both resolutions and open edges in roughly
equal measure; the kills below are reported with the confirmations, as sealed.

## 3. The closure yield (attempts one through six)

Writing "closure" for a two-carrier exact agreement on shared ground:

- **A1 — zero closures.** All six first-generation encodings proved window-learnable
  (the sealed grounding hypotheses inverted); the constructed diode grounded regardless
  — property 1 discharged on a field of failed hypotheses, as the protocol intends.
- **A2 — the first closure**: two surfaces at the exact count 33.
- **A3 — the control that sharpened the instrument.** A declared-noise surface at
  matched density *also* self-admits: quiet admission alone cannot separate an
  externality from vacancy, and one member of the first closure proved
  battery-indistinguishable from noise. The discriminating instrument was discovered
  here: a **lived face** — at least one admitting and one firing address.
- **A4 — the first lived closure** under the strengthened reading (attempt-scoped).
- **A5 — two further lived closures**; the strengthening graduated into the decided
  closure operator on its second independent witness (the programme's two-witness
  standard).
- **A6 — the fully-lived quadruple.** Under the graduated operator a fourth surface
  reached a lived face, and **all six pairs of the four lived carriers closed pairwise
  at the same exact count on the same ground**.

The closure population thus grew 0 → 1 → 3 → 6 pairs while the operator *strengthened*
— each later resolution met a strictly harder criterion than the one before it.

**The honest anatomy, recorded alongside.** (i) *The vacancy basin*: twice, a raised
encoding grounded into all-quiet reads byte-indistinguishable from declared noise; both
escapes were achieved by coarsening the declared class, not by re-learning. (ii) *The
read-state space*: the frozen carriers' reads, initially quantised into three states,
produced three further reads outside that grammar — all at firing faces of lived
carriers. (iii) *Non-monotonicity*: admission is not monotone in training length — both
members of the first closure alternate between admission and refusal as the training
life lengthens. Each of these is a measured property of the carriers, not of the line.

## 4. The inward turn (attempts seven through eleven)

Five further attempts held the closure population at the four lived surfaces (the
boundary pair moved without joining: one surface reached a mixed face at a later
generation and returned; the other grounded into indistinguishability from declared
noise at two successive generations) — and the record's yield turned inward, onto the
instrument itself, landing four exact structural results:

1. **The read-state space closed (provisionally at six).** Every frontier read of the
   campaign to that point falls into six exact states, each an integer identity in the
   address's declared quantities, zero reads outside them. Two states are governed by
   elementary parity facts of the integer line: a window ending in a structure event
   forces a composite successor (the successor of an odd prime is even), so fires there
   are arithmetically incapable of catching structure; a lag-two window lands on an odd
   successor, and its catches are exactly the twin completions of the span.
2. **The two-axis law.** Across every (carrier, generation) pair on the record, a
   carrier quiet at every shared-ground address has always read its own frontier
   exactly, and a carrier firing at every address never has; between the extremes the
   two properties are independent. Measured over fifteen rows with both cells of the
   diagonal empty — and re-witnessed at every subsequent attempt through A35.
3. **The policy reduction.** A frozen carrier's entire behaviour on any ground is its
   8-bit window policy (one predict-bit per 3-symbol shape), and every read in the
   record re-derives exactly as that policy composed with the address's shape table.
   The six states of (1) are six policies; the all-quiet basin of (2) is the all-zero
   policy through all eight bits — measured directly in the weights for every quiet
   carrier on the record.
4. **The fossil discovery (the eleventh attempt).** The one window shape whose single
   natural occurrence on the primality stream is followed by a structure event — the
   consecutive-prime pair (2, 3), presented exactly once, at N = 5 — stamps the
   learner's first-encounter update and becomes a policy bit invisible at every declared
   address (the shape never recurs beyond the origin). On a declared eight-source
   training-length ladder, including three lengths never previously read, **the carrier
   holds that bit if and only if its own frozen read collides** — eight for eight, with
   every refusal-side policy landing {000, 110} exactly and every admitted policy empty.
   The companion constructions landed their derivations: a minimal stream presenting one
   forced-1 window builds a permanent single-bit carrier (two updates, then
   convergence), and its forced-0 twin builds the zero carrier (no update ever). A
   dense twenty-five-source trace shows the bit oscillating with training length
   (fourteen sign changes) while agreeing with the collision phase at every declared
   anchor — and on that trace a carrier fires somewhere if and only if it carries the
   bit, twenty-five of twenty-five.

The eleventh attempt was the record's first resolution whose object is the *instrument's
own memory*: the binary phase behaviour the first paper measured as admit-or-collide is
readable inside the frozen state as the presence or absence of a one-bit trace of the
integer line's unique consecutive-prime event.

## 5. The universal form, and the declaration (attempts twelve and thirteen)

Because a frozen carrier's read is its policy composed with the address's window table,
the seventeen traced-but-untested training lengths carried *forced* phase predictions —
the instrument predicting itself. **All seventeen landed exactly as derived.** A further
twenty-four training lengths, chosen where neither the policy nor the phase had ever
been read — including lengths to twice the record's prior reach — obeyed the same law:
the carrier collides at its own frontier if and only if it holds the one-bit origin
trace, **forty-nine cells for forty-nine**. The read-state grammar closed at **seven
exact states** (the seventh, a catch-all face whose fires include every structure
position of the span by the parity identities, was derived before it was read and then
witnessed at four addresses), with zero unexplained reads across the twelve-attempt
record — a closure that has held through all thirty-five attempts.

At the thirteenth attempt's close the promotion fired under its sealed conditional (the
forty-nine-cell re-derivation held), and the law entered the record **DECLARED** — the
campaign's one declared law:

> **THE FOSSIL LAW.** bit-'110' ⟺ ¬admit (the D family: the (2, 3) origin mark decides
> the phase). Declared at RIX-A13; forty-nine witnessed cells.

## 6. The landscape (attempts thirteen through thirty)

With the phase law universal, the test turned outward from any single frontier to the
**landscape**: the frozen representative's eight-bit policy read at structured families
of training lengths, profiled over the 2-adic depth v₂ of the training length and over
successive base-eight digit *eras*.

**Six kills, then a floor.** Six successively sharper laws for where non-trivial
policies live — by digit boundary, by alignment grid, by grid floor, by downward
thickening, by a plateau band, by a decay band — were each pre-registered and each
killed by the measured landscape, five of the six by exactly one cell of the declared
line. The seventh held: at four 32-cell transects of successively halved grid scale,
each a distinct residue class, the density of non-trivial policies steps down once and
then holds near five-eighths (eighteen of thirty-two against the pre-declared band
[18, 22]) — **the first held landscape law**.

**The first valley.** The floor did not extend downward: one transect further the
density fell to ten of thirty-two, and the pre-registered monotone reading of the fall
died by one cell when the next transect turned upward. The third pre-registration held,
and not narrowly: the density profile over v₂, at eight matched transects spanning
every depth of the era, is **a valley** — descending monotonically from 75 per cent at
depth seven to a floor of 31.25 per cent at depth two, then rising through 34.4 to
43.75 per cent at the odd lengths — the second held law.

**The eras, the ordering, the crossing.** The next base-eight digit era is not a copy:
at the depth where the first era's floor sat at 31 per cent, the second era reads twice
that. The pre-registered cross-era ordering held cleanly at the matched depths (twenty
against ten; nineteen against fourteen) — the third held law — and the second era's own
profile (22 · 20 · 12 · 19 across its measured depths) mirrors the first era's valley
**in kind** while differing in place and level: the fourth held law (the wall stake,
rich 22/32, nine clear of its one-sided line). Then the ordering met its boundary, and
the boundary proved lawful: at the fifth matched depth it held by exactly one cell
while a rim stake failed by one; at the sixth matched depth **the ordering failed for
the first time** (nineteen against twenty — the record's first sub-first-era matched
reading), and the pre-registered promotion of the ordering to a declared law **correctly
refused to fire**; at the seventh matched depth a stake in the *opposite* direction held
with four cells to spare — the fifth held law, the first staked in the poorer
direction. Read together: the two eras' density profiles are ordered on the shallow
depths (five matched depths), inverted at the deep (two), and **cross exactly once** —
between depths four and five. The proposed law remains on the record unpromoted:

> **THE ORDERING LAW (PROPOSED, undeclared).** The richer-era ordering: at every matched
> depth the sixth era reads strictly richer than the fifth (the enrichment belongs to
> the line; the shape belongs to the era). Proposed at RIX-A28; the promotion gate
> refused at the sixth matched depth, exactly as sealed.

**The depth band.** The richest pattern in the atlas — the all-eight policy — erupted
from an isolated trio into a nine-member family, and the record's leanest sealed read
(three cells, one law) bounded it: **the family lives in a measured two-sided depth
band, v₂ ∈ [3, 6]** — absent from every shallower transect in both eras, absent from
the depth column above — the sixth held law, the first about a form family rather than
a density profile. The eruption was the family filling its band, not breaking loose;
and the deepest cell ever read (v₂ = 8) is rich while carrying no mark — richness and
the all-eight are separable at depth, exactly as the two-axis grammar requires.

**The atlas.** Under the kills and holds an atlas accumulated: **sixteen realized
live-pattern sets** over hundreds of measured cells, with one bit-implication standing
at every realized form; two selector signatures crossing carrier families — one
occupying a solid, exactly-bounded sixteen-integer interval of pure single-pattern
policies, later finished as a five-cell object sharp-edged both sides; and the origin
mark's exact two-pattern block, finished wall-to-wall at seven cells and **porous at
exactly one interior cell** — solidity, where it had been observed, was a property of
the resolutions read, not of the object. The atlas closed early and stayed closed:
by the campaign's end, seventeen consecutive reads had added no new live-pattern set.

## 7. The dissolution (attempts thirty-one through thirty-five)

The landscape campaign left three interval-like boundary candidates standing: a fade
line where richness appeared to stop, the apparent quiet approaching the era's end, and
the era gate at 2¹⁴. Five sealed attempts then killed each of them — and the kills,
read together, are the campaign's terminal result.

- **A31 (the fade transect).** The fade law was comprehensively killed: six of the
  eight cells above the fade line read rich. **The fade is stratum-local** — the
  boundary binds the strata that measured it, not the v₂ = 6 stratum, which runs rich
  to 16064 and fades only at the era's end (bracketed to one step: rich 16064 · quiet
  16192). A ninth all-eight member appeared in the upper half (13760, at v₂ = 6 —
  *inside* the depth band: the family grows in position while its depth law holds).
- **A32 (the fade map).** The fade dissolved entirely: four of the eight v₂ = 4 cells
  read rich — **no stratum obeys the fade line beyond the exact cells that measured
  it**. A new candidate was named honestly: the era-end quiet (both measured strata
  quiet approaching the era boundary). The ninth all-eight proved isolated (its
  neighbour 13696 reads {000}; the isolation grammar four for four; the deep column
  stays mark-free — the depth band's second deep test passed).
- **A33 (the end-quiet probe).** The end-quiet was killed: **the era ends rich** (four
  of six cells) — the "quiet zone" was the coincidence of the v₂ = 4 and v₂ = 6
  strata. The v₂ = 5 stratum runs rich to within 96 cells of the era gate. **The
  stratum-clock lesson completed**: no mid-era boundary binds across strata; each
  stratum runs to its *own* stop — v₂ = 4 by 16016 · v₂ = 6 by 16064 · v₂ = 5
  bracketed to (16288, 16352] · v₂ = 3 open — leaving the era gate 2¹⁴ as the one
  cross-stratum boundary candidate standing.
- **A34 (the gate crossing).** The gate law was killed: **the seventh era opens live**
  (three of the six opening cells rich — richness crosses 2¹⁴). **The dissolution is
  total: no boundary at all survives; only strata and their clocks.** Yet the gate is
  not nothing: it sits inside a measured quiet band (16288, 16520) — rich · four quiet
  cells · rich — **the era gate is a valley, not a wall**; and the era-opening
  quiet-start is now a two-era pattern (8208/8224/8256 · 16392/16416/16480).
- **A35 (the straddle probe).** The quiet band itself was killed: **porous** — two of
  the seven final-block v₂ = 3 cells read rich (16280 and 16344, the latter **forty
  cells from the gate, the nearest rich cell to any era gate ever read**). The band was
  the measured depths' lattice coincidence — the third such lesson (the end-quiet, the
  fade, the straddle). And the stratum's tail does not run to a stop: it **intermits**
  — rich · quiet-run · rich · quiet-run (16264, 16280 rich · 16296–16328 quiet · 16344
  rich · 16360–16376 quiet) — with the v₂ = 3 stop bracketed to (16344, 16360], one
  16-step of precision, the stratum's own grid.

The record's closing grammar, quoted from the thirty-fifth attempt as landed: *the
campaign's boundary story is now fully consistent: NOTHING interval-like binds across
strata — not the fade, not the end zone, not the gate's quiet band — while the era
gates mark soft valleys and every stratum keeps an intermittent tail on its own clock.*
Seventeen closure-clean reads; the sixteen-form atlas held still across the dissolution's
hundreds of fresh cells; the origin-mark carriage census closed at fifty-one; the
all-eight family at nine members, depth-banded [3, 6].

## 8. The verdict on the prediction

Per the first paper's own disciplines, the verdict is stated on the test's terms and on
no other.

**Supported, on the test's terms.** The named mechanism produced repeated resolutions
at the record's exactness standard — exact integer agreements at declared span counts,
zero tolerance anywhere in thirty-five attempts — under a closure criterion that was
adversarially strengthened mid-campaign and then met again, ending in a fully-lived
quadruple closing pairwise at one exact count. The three required properties of the
test were discharged and held: the diode constructed and re-witnessed at every attempt;
the pre-fixed dynamics projected with their unknowing statements printed at every seal;
the exactness standard met at every closure.

**The boundary, named.** Two of the six surfaces reached no face at any generation
read; their standing is the measured boundary of the test's reach, and the attempt
lineage remains open. And the landscape campaign's terminal shape bounds the
interpretation from the other side: the instrument's laws are **local** — a declared
phase law inside the frozen state, held profile laws inside eras and strata — while
every *global*, interval-like binding candidate dissolved under its sealed test. In
the first paper's grammar, its §4.6 lesson — *the limit is in the carriers, never in
the line* — recurs here one scale up: the limit is in the strata and the lattices
measured, never in the line **[R2]**; what the relativity measures is exact wherever a
carrier stands, and refuses to globalise **[R2]**.

**What is not claimed.** A closure is a statement about carriers, grounds, and exact
counts inside the apparatus. **No claim on any Millennium Prize problem is made,
supported, or weakened by any part of this record** — the six problem statements served
as received texts from which the pre-fixed dynamics were drawn, and every seal prints
the corresponding unknowing statement.

## 9. The standing laws and the open edge

The record's law ledger at the thirty-fifth close:

- **Declared (one):** THE FOSSIL LAW — bit-'110' ⟺ ¬admit; the (2, 3) origin mark
  decides the phase; forty-nine witnessed cells.
- **Proposed, undeclared (one):** THE ORDERING LAW — the richer-era ordering at matched
  depths; the promotion gate refused at the sixth matched depth; the crossing stands as
  the measured object in its place.
- **Held landscape laws (six):** the density floor near five-eighths; the first era's
  valley (75 → 31.25 → 43.75 per cent over v₂); the cross-era ordering at its staked
  matched depths; the second era's wall (its valley in kind, richer in level); the
  crossover (the first law staked in the poorer direction); the depth band (the
  all-eight family two-sided at v₂ ∈ [3, 6]).
- **Structural, standing:** the seven-state read grammar with zero unexplained reads;
  the two-axis law (fifteen rows, empty diagonal, re-witnessed throughout); the policy
  reduction (every read = policy ∘ shape table).

The continuation is recorded, not acted: the thirty-sixth attempt's hypotheses stand
sealed in the design record — the seventh era's own profile (its first transect, the
thirty-two virgin v₂ = 5 cells in [16384, 20480), with the band sealed from the sixth
era's matched reading), and the tail-shape question (is intermittence the general
stratum-tail grammar?) typed, not staked. The lineage is open; nothing runs before its
pre-registration seals.

## 10. Reproducibility and data availability

- **Code.** The apparatus (`daemon/rix.py`, readout-only over the frozen house carrier),
  the experiment plugin, and the acceptance tests (verdict + counterfactual pair per
  probe) are published in the programme repository; every pinned value carries a live
  recomputation path on the running node (`/rix`). *(Public URL at submission; the
  repository releases under the programme's code-anchor process.)*
- **Seals.** Thirty-five sealed attempts; each seal is a git commit whose hash is cited
  in the design document (`oxexp-relativity-ingress-design.md`, the sole source of
  truth at its record commits) and re-verified by the live pages.
- **Data.** The attempt record is externally anchored to an immutable public dataset,
  republished at every close (each publish mints a new dataset — the newest is the
  record):
  [mldata.opendata.ai/dataset/rix-exp_7037eb09](https://mldata.opendata.ai/dataset/rix-exp_7037eb09)
  (245 rows at the A35 close, thirty-five attempts). The base campaign's anchors stand
  as cited in the first paper
  ([horizon-exp_6ec1e777](https://mldata.opendata.ai/dataset/horizon-exp_6ec1e777), 115
  rows; predecessor [walls-exp_19be23b1](https://mldata.opendata.ai/dataset/walls-exp_19be23b1)).
- **The sealed hand-off.** The record's consumable digest — restatement by citation
  only, no new claim — is sealed as *oxexp-relativity-ingress — THE HARVEST (A1..A35)*
  in the programme repository (commit `940891b`, over the design record at `7fbbd6a`);
  successor experiments in the programme consume this record only through that digest
  and the anchored datasets.

---

## References (draft)

1. S. De Costa, *Exact frontier reads of the primality stream by frozen online
   learners: a sealed seven-campaign experiment and a measured relativity between 0 and
   N*, programme repository, 2026. (The first paper: the instrument, the seven
   campaigns, the prediction, and the named mechanism this paper reports.)
2. The design record: *oxexp-relativity-ingress — the Relativity Ingress Experiment*,
   programme repository, 2026. (The sealed pre-registrations, verdicts, and pinned
   values for all thirty-five attempts; the sole source of truth.)
3. The sealed digest: *oxexp-relativity-ingress — THE HARVEST (A1..A35)*, programme
   repository, 2026. (The record's consumable hand-off; restatement by citation only.)
4. S. Wolfram, *A New Kind of Science*, Wolfram Media, 2002. (The 256 elementary rules
   as the Boolean functions of three inputs.)
5. F. Rosenblatt, The perceptron: a probabilistic model for information storage and
   organization in the brain, *Psychological Review* **65** (1958), 386–408.
6. K. Popper, *The Logic of Scientific Discovery*, Hutchinson, 1959. (The refutability
   standard the seal protocol operationalises.)

*(Items 1–3 resolve to public URLs at release.)*