# Shear Studies White Paper — Shape Against Magnitude Across Twenty-Seven Studies

**Program:** UUM-8D Solar Research / Shear Studies  
**Surface:** [uum8dSolarResearch wiki](https://github.com/gaiaftcl-sudo/uum8dSolarResearch/wiki)  
**Date:** 2026-08-26 (Version 3.1 — Study 27 added; v3.0 2026-08-26; v2.0 2026-08-13; first version 2026-07-23)  
**Status of this document:** Public synthesis of the twenty-seven-study board. Study 1: **DATA SEALED** — first look 2026-08-12, frozen integers + 82-row corpus; claim on apex not served yet; pinned-archive verdict ~2026-08-30. Studies 06/07: **LAW FROZEN**. Study 04: **PARTIAL_CORPUS_SEAL**. Study 02: **LIVE CLAIM**. Studies 11, 14, 16–21: **LAW FROZEN + LIVE CLAIM**. Study 15: **CITED** (court DARK). Studies 22–25: **LIVE ON GLAMA**. OPEN (data absent) is confined to 03, 05, 08, and Studies 26–27 (both chartered 2026-08-26). The live [index](Shear-Studies-Index) and the study pages are the single source of truth for every status; where this paper and a study page disagree, the study page wins.  
**Readers’ entry:** [Shear-Studies-Readers-Guide](Shear-Studies-Readers-Guide)

---

## Abstract

Science has spent decades measuring the world with rulers: percent drop, peak amplitude, χ² fit, image brightness. Rulers are cheap to fool. A geomagnetic superstorm digs a deeper ionospheric hole than an eclipse; a storm surge lifts a tide gauge as high as a tsunami; normalization moves a table’s absolute amplitudes before imaging ever begins. Magnitudes travel. **Appointments do not.**

This program replaces the ruler with a **shear**: a forcing that carries its own public clock and track; a raw public archive; a named adversary that matches size but not shape; and a frozen exact-integer law that grades confinement and traveling lag before any future event is scored. The eclipse of 2026-08-12 was the reference implementation — it happened, on schedule, on a day that measured quiet; the raw-channel first look measured a depletion at every sealed custody station, in its sealed window, in the sealed order; the data is sealed and the verdict awaits its pinned grading archive (§2.1). Twenty-six further studies now carry the same recipe on five wings: the **sky, sea, and ground studies** (rocket burns, solar flares, tsunamis, Forbush decreases, explosion-vs-earthquake discrimination, the EHT’s public visibilities of Sagittarius A\*, the orbit of Gaia BH1, a global convective bond, Fermi gamma-ray shear); the **exact-arithmetic courts** (Ehrhart counting, quantum parallel repetition, operator-algebra rigidity, four hard-problem courts live as public tools, and — chartered 2026-08-26 — exact nuclear scattering, replaying the field's own all-orders-versus-optical-limit divergence); the **industry-key courts** (protein, disease, chemistry, materials, a published dice tuple — identity by keys the public already owns); the **applied shears** (a historical-frequency test claim and a stellar-dynamo integer step law); and the **biology lane**, which carries the same discipline into medical territory as test claims on public data, never efficacy claims — newest, chartered 2026-08-26: [Study 26 — Master regulator bonds](Study-26-Master-Regulator-Bonds), grading cancer's load-bearing regulator networks in exact keys.

**The Study 07 ledger row, stated at its measured strength:** the public non-normalized Stokes-I CSV for Sgr A\* (product 2022-D02-01) passes a frozen integer law on short/long baseline ladder, closure-phase structure, and Earth-rotation arc order — **8/8 WIN** across days, bands, and both HOPS and CASA pipelines. The lightcurve-normalized tables used as static-imaging input fail the same absolute short-baseline floor — **8/8 MISS**, every failure `SHORT_FAIL`. Compact flux ~2.1–2.5 Jy on short baselines is in the sky’s table (VERIFIED against the sealed ledger). What the MISS measures is the definition of normalization doing exactly what the collaboration’s release notes document; the shape quantities imaging actually uses pass through the fork intact (§4.5). The program’s reading is that the absolute-jansky appointment and the imaging input are **different tables**, and that the fork must be named as provenance. That separation is a ledger row, not a slogan.

---

## 1. The thesis

### 1.1 Size is cheap; shape keeps appointments

| Domain | Magnitude mimic | What the mimic cannot do |
|---|---|---|
| Ionosphere | Gannon G5 storm (May 2024) | Thread the Moon’s umbral cone on the ephemeris clock |
| Rockets | Storm TEC holes, Tonga TIDs | Sit inside a launch corridor at T+5–15 min from a manifest second |
| Flares | Storms, radio bursts | Step the whole sunlit hemisphere on the GOES X-ray second |
| Ocean | Storm surge, meteotsunami | March gauge-to-gauge on a Huygens travel-time chart from a USGS origin |
| Cosmic rays | Pressure lows, diurnal wave | Strobe ~50 neutron monitors in one L1 shock minute, rigidity-ordered |
| Seismology | Near-site quakes, quarry blasts | Climb a Pn/Pg/Sn/Lg ladder from a fixed depth-0 point track |
| Galactic center VLBI | Ring PNG + regularizers + `csv_norm` | Invent short-baseline absolute janskys and closure topology absent from the sealed rows |
| Gaia BH1 astrometry | Published *N* M☉ dark companion narrative | Carry the horizon-scale label into the integer table — the mass is in the appointments (§9.4); the horizon narrative is a label the table does not adjudicate |
| Exact arithmetic | Float pipelines (triangulation volumes, tensor amplitudes, spectral norms, SDP eigenvalues, permanent sums) | Print the exact integer or rational the structure carries — float last bits fork across machines; the lattice answer is one string everywhere (§5) |
| Industry keys | Generated nicknames, predicted structures, minted identifiers | Bond a claim to the identifier the public already owns — PDB accession, ICD-10 + DOID, InChIKey + CID, COD number — verified against the hosted holdings (§5) |

### 1.1b Cosmology’s three black-hole pipelines

| # | Pipeline | Instrument writes | Model shows | Program study |
|---|---|---|---|---|
| 1 | High-precision astrometry | Star wobble (Gaia BH1) | Invisible black hole of *N* M☉ | [Study 08](Study-08-Gaia-BH1-Astrometric-Shear) |
| 2 | GW interferometry | Laser strain (LIGO/Virgo) | Merging compact objects | Holds no number on the live index; no charter yet |
| 3 | VLBI | Sparse visibilities (EHT) | Ring / shadow image | [Study 07](Study-07-SgrA-Milky-Way-Results) — **graded** |

**Numbering note (updated 2026-08-26):** v1 queued the GW-interferometry study as “Study 09”; v2 moved it to “Study 10.” Both numbers are now taken. *Study 09 — The Global Convective Bond* is **published on this surface at LIVE CLAIM** — frozen Int64 thresholds, standing rule REFUSED ≠ clear sky: [Study 09](Study-09-Global-Convective-Bond). *Study 10 — Fermi gamma shear / dark matter* stands at **CHARTER WITH ARCHIVE** — named HEASARC weekly photon FITS + paper, no keV ingest, not OPEN: [Study 10](Study-10-Fermi-Gamma-Shear-Dark-Matter). The GW study holds no number on the live index and is cited as pipeline 2 only.

**Origin of shear, scoped per pipeline:** the infinite-precision failure mode named in v1 belongs to **pipeline 3** — imaging regularizers filling sparse Fourier coverage on absolute grids. It does not describe pipeline 1: Gaia BH1’s published mass is a plain Keplerian two-body fit, no relativistic term, residuals consistent with noise (REPORTED — Nagarajan et al. 2024). For pipeline 1 the shear question is narrower and stated in §9.4: which channel of the public record carries the orbit at integer resolution, and what label rides on top of it. Ingest for Study 08 runs through a dedicated `corpus/study-08/` directory with integer quantization at ingest; ledger rows seal only after grading.

### 1.2 The four-piece shear recipe

Every study on this wiki must instantiate all four. Missing any one is a correlation hunt, not a shear study.

1. **Forcing with computable clock and track** — announced in public numbers before the response is examined (in the courts: a presented claim or configuration, against the frozen law it must clear).
2. **Public raw response archive** — exact URLs, formats, cadence; anyone pulls the same bytes (in the courts: the public key registry or the sealed fixture).
3. **Sealed adversary** — matches magnitude, fails shape; graded raw; firings published, never renamed (in the courts: the float pipeline or the generated identity).
4. **Standing future events** — law frozen before scoring events that had not yet happened (or products not yet examined).

**Verdict rules (program-wide):** exact integers only — no floating point crosses a seal. Every sky study has a **confinement** analog and a **traveling-lag** analog; every court has an **identity** analog and an **exactness** analog (§5). Thresholds derive from historical corpus, then freeze. A recorded failure is a finding. **VOID means the data is absent** — never a failure verdict.

### 1.3 The road, and the words on the board

Charter → Corpus → Frozen law → Pre-registration → Public resolution. No stage revisited; status lines only advance. Since v2 the live index has defined the controlled vocabulary the §7 board uses: **OPEN** (the program does not have the data — only 03, 05, 08), **CHARTER WITH ARCHIVE**, **DATA SEALED**, **LAW FROZEN**, **LIVE CLAIM**, **LIVE ON GLAMA**, **CITED** — with the rule that a named archive in the charter is not OPEN if a stronger term already applies.

---

## 2. Study 1 — Eclipse 2026: the reference implementation

**Status:** **DATA SEALED** — first look 2026-08-12; frozen integers + the 82-row corpus. The claim on apex is not served yet. The sealed verdict grades on the pinned archive (~2026-08-30).

On 2026-08-12 the Moon’s umbral cone crossed western Iceland and northern Spain, on schedule — the sealed stations spanned 16:43:40–19:22:06 UT, totality peak off Iceland 17:45:54 UT. Six decades of eclipse-ionosphere literature reported scalars (“TEC dropped 40%” — REPORTED). On 2024-05-11 the Gannon superstorm produced deeper depletions under an unobscured Sun; graded precisely, **the program’s own frozen magnitude-gate law false-positived at all seven 2024 stations that day** — sealed raw in the historical Form, nine false positives in total (VERIFIED, ledger-frozen). The inference that any detector keyed to magnitude alone would have fired that day is the program’s reading, not a measurement.

Study 1 asks a different question: does the ionospheric dent **thread** the shadow’s timetable — confined to the geometry’s minutes, deepest near each station’s maximum, marching at the cone’s ground speed? Storms dig holes; they have no cone, no track, no clock.

**Already sealed (public Forms):**

| Stage | Claim | Status |
|---|---|---|
| 2024 proof | Shape-read verifies April 2024 eclipse from raw archives | SEALED — 7/7, 21 controls null, 0 false positives |
| Historical universality | One frozen law across 2017 / 2023 annular / 2024 + storm shear | SEALED RAW **as a failure** — 15/16, the Eugene OR 2023 miss named in the seal, 9 false positives published; current registry stamp 2026-08-18: **16/20 · CURE · `cdf43e80f06f10a3`** |
| 2026 prediction | Named totality stations must thread; named nulls must refuse | PRE-REGISTERED 2026-07-16 |
| Extensions | Station order and storm-decidable confinement | PRE-REGISTERED 2026-07-16 |

All four Forms — fingerprints `89afd825011f060e`, `6fe49bee70289fee`, `c3e6841a206b8058`, `b106d2926b1bf2cf` — re-verified byte-for-byte on 2026-08-13, 28 days after sealing (VERIFIED — a dated statement from the 27-day review). Since then the live surface has re-stamped the historical-universality Form: every render-time Form-registry block now shows **16/20 · CURE · `cdf43e80f06f10a3` · 2026-08-18T16:42:21Z** — the current registry state, named here alongside the original 2026-07-16 seal `6fe49bee70289fee` at 15/16.

**The frozen numbers a reader needs to falsify Study 1** (all ledger-frozen; re-derived exactly at T+1):

- **Detection law** (`89afd825011f060e`): 2024 benchmark 7/7 detections, 21 control pairs null, 0 false positives — re-run green 2026-08-13.
- **Two-clause discriminant** (`b106d2926b1bf2cf`): confinement excess ≥ **50,189 ppm**; lag gate **[−300, +1200] s** around each station’s own maximum; traveling order latrabjarg +192 s → reykjavik +2,372 s → coruna +53 s → leon +32 s → zaragoza.
- **Storm gate:** est-Kp ≥ 5 voids the clean-test framing. It never fired (§2.1).
- **Quantified nulls:** norwich must refuse at 40,790–73,923 ppm; millstone at 49,036–88,868 ppm.
- **Separation:** minimum true-eclipse excess **57,543 ppm** vs maximum storm impostor **42,836 ppm** (Gannon May-11, Dallas) — a 14,707 ppm gap around the frozen threshold. All 16 true eclipse rows pass both clauses, including Eugene, which the base magnitude law missed.
- **Empirical coupling band:** 309–560 permille of obscuration.
- **Historical crucible** (`6fe49bee70289fee`, sealed as a failure, raw): 2017 totality 5/5 · 2023 annular 3/4 (the Eugene, OR miss named in the seal, never renamed) · 2024 7/7; Gannon May-10 7/7 null; May-11’s naive-gate false positives rejected by confinement + lag. 15/16 overall, 9 false positives, across **82 station-day pairs** (a birth-day commit said 96 — an overcount, corrected in the [27-day review](Eclipse-2026-First-27-Days-Review)). The live registry has since re-stamped this Form: every render-time Form-registry block now shows **16/20 · CURE · `cdf43e80f06f10a3` · 2026-08-18T16:42:21Z** — the current surface state, standing beside the original 2026-07-16 seal stated above.
- **2026 prediction** (`c3e6841a206b8058`): the named totality stations must thread; the named nulls must refuse.

A 2026 verdict that fails any of these terms is a miss and will be published as one.

**Why it is the reference:** it proved that a magnitude-mimicking adversary (Gannon) fails a shape test that true eclipse days pass — and that the law can be sealed weeks ahead: 2026-07-16, 27 days before the eclipse, untouched since. Every later study reuses that logic.

### 2.1 T+1 — the day answered quiet; the archive has not yet spoken (added 2026-08-13; status restated 2026-08-26)

The eclipse happened. Whether the prediction won is not yet decidable — the reason is a measured archive calendar, not a hedge. The index’s words: **DATA SEALED** — the raw first look shows the TEC bite riding the shadow’s ground track at all four custody stations, geomagnetic storms the sealed adversary; the sealed verdict grades on the pinned archive (~2026-08-30). The claim on apex is not served yet — an index-recorded serving gap, stated so paper and surface agree.

**The fair test is verified.** The confounds are quantified from captured public feeds (all VERIFIED):

- **Geomagnetic quiet (GFZ definitive Kp):** eclipse-day maximum Kp **2.667**; the sealed storm gate (est-Kp ≥ 5) never fired.
- **Flares absent (captured GOES archive):** one B5.4 flare ending 43 minutes before the Reykjavik totality window; zero C/M/X flares in 15:30–20:00 UT.
- **Solar wind quantified (captured L1 feed):** speed 382–478 km/s, density 1.48–9.39 cm⁻³, Bz −5.0 to +3.4 nT with 56 minutes below −3 nT — a mild residual tail, a named caveat for the auroral-adjacent Iceland corridor only; Spain is clean.
- **Auroral flank quiet (captured Narsarsuaq magnetometer):** X-range 107 nT in the eclipse window vs 453 nT pre-dawn; no substorm. Leirvogur cross-check pending.

A quiet day is a fair test: a match will be a clean win with no confound to explain it away; a miss a clean miss with no storm to blame.

**Raw custody is held.** Twenty-seven RINEX 30-second files (93 MB) captured at T+1 from anonymous public mirrors (BKG IGS/EUREF, NOAA CORS) for nine stations, covering eclipse day (DOY 224) and control days 217/197, alongside the GOES, solar-wind, and magnetometer captures, under a SHA-256 manifest (digest, retrieval timestamp, source URL per file). Custody doctrine, stated plainly: **the sealed verdict grades on the pinned Madrigal gridded-TEC surface; the raw RINEX is custody and cross-check, not a substitute grading surface.**

**The pinned-archive verdict date is measured, not promised.** Madrigal instrument-8000 gridded TEC for 2026-08-12 was NOT POSTED at T+1; the posting boundary was measured at a two-to-three-week lag (between 2026-07-25 and 2026-08-01), so eclipse day is expected to post around **2026-08-30**, with control day 2026-08-19 also up by then. Grading runs then, on the pinned surface, against the untouched fingerprints. Version 1 of this paper promised a public verdict on 2026-08-13; that promise is graded **MISSED** in the published [operational review](Eclipse-2026-First-27-Days-Review) — the archive’s own calendar made it impossible, and the date lived only in prose.

**The operational record is on the page.** The [27-day review](Eclipse-2026-First-27-Days-Review) also records, as failures: the live capture loop dark from 2026-07-21 through eclipse day; the watch daemon staged but never installed; the program’s pages split across three unreconciled surfaces; the fossil-pair count corrected from 96 to 82. The sealed core held while the live surface failed. Both facts belong in the record.

### 2.1b Predictions vs first measured data — the raw-channel first look (added 2026-08-13, T+1)

The requirement on resolution day was plain: predictions and measured data, side by side, now. The pinned archive had not posted eclipse day — but the raw receiver files SHA-sealed at T+1 contain the ionosphere's answer, and the program computed it directly from those bytes: GPS dual-frequency geometry-free phase (L1C/L2W), sign-calibrated, arcs leveled, 300-second bins, differenced against sealed control day 2026-08-05. This is **relative slant TEC from single stations — a different observable from the pinned gridded surface the frozen law grades on**: the first look, not the sealed verdict, which still runs ~2026-08-30 on the pinned archive, untouched.

**What the raw bytes say, station by sealed station:**

| Sealed station | Sealed maximum (27 days prior) | Deepest dip (raw first look) | Lag vs sealed max | Depth vs control | Lag gate [−300, +1200] s |
|---|---|---|---|---|---|
| **Reykjavik** | 17:48:46 UT | 17:40–17:45 UT bin | **−376 s** (next-deepest bin −76 s) | **−4.5 TECU** | outside early edge by 76 s at this binning; bracketing bins inside |
| **A Coruña** | 18:28:18 UT | 18:45 UT bin | **+1,002 s** | **−6.2 TECU** | **inside** |
| **León** | 18:29:11 UT | 18:45 UT bin | **+952 s** | **−5.8 TECU** | **inside** |
| **Zaragoza** | 18:29:43 UT | 18:35 UT bin | **+467 s** | **−3.1 TECU** (measured curve) | **inside** |

Reading it plainly:

- **Every sealed totality station with a receiver in our custody measurably dipped, in its sealed window, on the day.** Reykjavik shows the textbook signature — onset ~17:30, bottom straddling the sealed maximum, asymmetric recovery. The control day shows nothing at those minutes.
- **The traveling structure is present at first-look resolution:** Iceland first, the Spanish trio ~fifty minutes later — the sealed order. The gaps *within* the trio (+53 s, +32 s) sit below 300-second binning; they are for the pinned surface to grade.
- **Depths are consistent with the sealed band** — 3–6 TECU against quiet ~8–15 TECU sits inside the pre-registered 31–56%-of-quiet expectation — stated as consistency, not a graded fraction, since arc-leveled relative TEC has no absolute denominator.
- **Recorded raw, not smoothed:** a pre-maximum positive excursion at A Coruña and León (one bin +13 TECU, flagged as a possible arc-leveling artifact); Reykjavik's deepest bin just outside the early edge of the lag gate at this resolution, neighboring bin inside; a late León feature near local sunset. Printed before the sealed grading, so nobody can accuse the record of meeting the data halfway.
- **The null channel** (Westford) is captured, not yet processed — a different format generation. The sealed null claim is *silence*, and it grades on the pinned surface with everything else.

The derivation is in custody beside the raw files it read — scripts, per-station outputs, method README, SHA-256-sealed in the same manifest; anyone holding the RINEX files can re-run the arithmetic.

**What this is and is not:** four sealed stations, four measured dips, in the sealed windows, in the sealed order, at depths inside the sealed band, read from bytes no one can alter — and not the verdict, which belongs to the frozen law on the pinned archive, ~2026-08-30. If the pinned surface disagrees with this first look, both go on the page.

### 2.2 The arithmetic underneath — no supercomputer, no floats, and why that changes what a verdict is

There is a fact about this experiment that the tables do not show: **none of it was computed the way modern geophysics is normally computed.** No supercomputer. No floating point anywhere a claim lives. Every sealed number — obscuration in parts-per-million, time in whole UT seconds, depletion as integer ppm of baseline, thresholds like 50,189 — exists on a whole-integer lattice, and every verdict is an integer comparison. The law never asks whether 0.5754 exceeds 0.5019 in some machine's rounding; it asks whether 57,543 ≥ 50,189, and that question has exactly one answer on every computer ever built.

**A floating-point verdict is a property of a machine. An integer verdict is a property of the numbers.** Two computers running the same float pipeline can disagree in the last bits, and at a decision boundary the last bit *is* the verdict; a tolerance is a place where an argument lives. Integer arithmetic has no last bit to argue about: the grading can be re-run by a stranger, on any machine, in any decade, and must produce **byte-identical** output. That is why four seals re-verified **byte-for-byte, 28 days after stamping, 4/4** — a sentence that cannot be written about a float pipeline. The honesty architecture rests on arithmetic that cannot drift.

**No supercomputer was involved, and that is the point.** The full experiment runs on one desk machine reading public archives. The scarce resource was never float throughput; it was **exactness and discipline**, and those are free. The lattice touches the world at exactly one guarded crossing — a single quantization step (ppm, milli-units, whole seconds) — and from there no approximate number touches the record. (Stated honestly: the T+1 first look was an exploratory calculation on the custody bytes; the sealed grading that supersedes it runs entirely on the integer law.)

**What actually changes:** on conventional arithmetic a verdict is a computation you are asked to trust; on the lattice it is a **fact you can check** — the same fact on every machine, for every person, at every time. That is what let a fully public operation seal a claim twenty-seven days ahead and hand the grading to anyone. The doctrine’s titled page: [Zero Float · Zero Shear](Zero-Float-Zero-Shear-Paradigm).

**What resolution will mean.** When the archive posts, the prediction grades at full strength on a verified-quiet day. One resolved prediction is one entry in the track record — no more, no less.

**Entry points:** [Overview](Eclipse-2026-Overview) · [Model shear](Eclipse-2026-Model-Shear) · [Prediction registry](Eclipse-2026-Prediction-Registry) · [Blind spots](Eclipse-2026-Blind-Spots) · [First 27 days — full review](Eclipse-2026-First-27-Days-Review)

---

## 3. Studies 02–06 — the same recipe, statuses now diverged

These five share Study 1’s recipe. In v2 all five stood at charter; their statuses have since diverged: **02 LIVE CLAIM** with a dated data row, **04 PARTIAL_CORPUS_SEAL**, **06 LAW FROZEN** and graded, **03 and 05 OPEN** — and OPEN means the program does not have the data, applying program-wide only to 03, 05, 08. Each charter page names archives, adversaries, integer metrics, and success criteria, and carries its own honest-limits section.

### 3.1 Study 02 — Launch ionospheric holes · **LIVE CLAIM**

| Piece | Instantiation |
|---|---|
| Clock / track | Launch Library 2 / GCAT liftoff second; second-stage corridor ±3° |
| Archive | Madrigal world-gridded TEC (kindat 3500) + captured CORS RINEX |
| Adversary | Gannon, Hunga Tonga TIDs, Falcon deorbit burns |
| Shear | Corridor confinement + clock-locked onset/recovery in percent-of-baseline integers |

**The live data row:** Starlink Group 10-49 · **2026-08-25T09:33:38Z** · SLC-40. What is VERIFIED is the integer table: 21 RINEX files from 7 CORS stations (fetched 2026-08-25T20:33:34Z), first-epoch L1/L2 as integer milli-cycles, `REFUSED_FLOAT = 0`, with a precautionary lockout window 09:35–11:30 UTC sealed so a terrestrial plasma hole is not read as cosmological shear. Cited-versus-measured discipline: the **~900 km hole diameter and 10–70% depletion range are REPORTED** (FORMOSAT-5 founding scale, Chou et al. 2018) and stay **cited, not measured**, until Madrigal posts launch-day TEC — Madrigal returned HTTP 200 and 0 bytes for 2026-08-25, graded as a data absence (GAP), not a failure.

**The regulatory alarm:** the 7-station integer table plus one-way IONEX/FITS bridges — TECU explicitly marked UNKNOWN — is packaged for FAA docket FAA-2026-8614, open through 2026-08-31, in formats FAA AST, NOAA SWPC, and MAST/ESA already load. The founder files; no agency receipt id exists. This is the only class of F-region hole that arrives on a schedule published in advance — and that conventional TID filters delete by construction.  
**Pages:** [Study 02](Study-02-Launch-Ionospheric-Holes) · [Regulatory alarm](Study-02-Regulatory-Alarm)

### 3.2 Study 03 — Solar-flare sudden ionospheric disturbances · **OPEN**

Clock: GOES X-ray peak; track: the sunlit hemisphere (eclipse geometry in reverse); adversary: storms + radio bursts; shear: hemisphere step in lockstep with the X-ray clock. OPEN, stated plainly: no SID corpus; detection threshold not sealed. Since v1 the study has grown a second layer — SHARP integer helicity, a signed crossing-count exact in integers — and a [decidable checklist page](Study-03-Predictions-and-Validations). Polar aviation HF and crew dose ride the same daylit physics.  
**Page:** [Study 03](Study-03-Solar-Flare-SIDs)

### 3.3 Study 04 — Tsunami vs storm surge · **PARTIAL_CORPUS_SEAL**

| Piece | Instantiation |
|---|---|
| Clock / track | USGS origin second + Huygens bathymetric travel-time chart |
| Archive | NOAA CO-OPS / DART public water levels |
| Adversary | Storm surge, meteotsunami, Lamb waves |
| Shear | Gauge-to-gauge arrival on the sealed chart — the oceanic traveling signature |

**Graded so far (index wording):** 2/2 tsunami gauge WIN; Sandy adversary WIN; Tonga **VOID_NEED_PRESSURE** — VOID meaning data absent, never a failure: the full multi-gauge Form and pressure branch are still unsealed. The data is sealed; the claim on apex is not served yet. False alarms and missed arrivals are measured in coastlines.  
**Pages:** [Study 04](Study-04-Tsunami-vs-Storm-Surge) · [Results](Study-04-Tsunami-vs-Storm-Surge-Results)

### 3.4 Study 05 — Forbush decreases · **OPEN**

Clock: the L1 shock minute; archive: NMDB / OMNI public counts; adversary: pressure lows + the traveling diurnal wave (the “corrections” conventional pipelines apply); shear: a rigidity-ordered, longitude-flat global step on raw uncorrected counts. OPEN, stated plainly: no NMDB corpus ingested; thresholds not frozen. One clock, many stations, magnitude mimics from weather — Study 1’s exact discipline on particle-physics instruments.  
**Page:** [Study 05](Study-05-Forbush-Decreases)

### 3.5 Study 06 — Explosion vs earthquake · **LAW FROZEN**

| Piece | Instantiation |
|---|---|
| Clock / track | Catalog origin + fixed depth-0 point (Punggye-ri); Pn/Pg/Sn/Lg ladder |
| Archive | EarthScope / IRIS miniSEED integer counts; USGS ComCat labels |
| Adversary | Near-site quakes, ripple-fired quarry blasts |
| Shear | Energy confined inside sealed P windows while Lg stays starved — shape, not mb/Ms scalars |

**Graded (index wording):** IC.MDJ **6/6 explosion WIN; 3/3 adversary reject**, on frozen integers — catalog clock + 53±2 s travel + SNR ≥ 20,000 ppt on raw integer counts. IU.INCN is VOID this seal (data absent, not a failure). The data is sealed; the claim on apex is not served yet. This is nuclear-treaty verification’s working problem, stated in public raw integers.  
**Pages:** [Study 06](Study-06-Explosion-vs-Earthquake) · Corpus, Results, and Registry pages on this surface

---

## 4. Study 07 — Sagittarius A\*: what the Milky Way wrote in the table

**Status:** LAW FROZEN 2026-07-23T20:00:59Z · primary `csv/` **8/8 WIN** · `csv_norm/` adversary **8/8 MISS** · all rows re-verified against the local ledger 2026-08-13

### 4.1 The provenance line this study draws

The public retelling often treats the EHT ring image as if it were the measurement. The source collaboration itself does not claim that: it publishes the calibrated visibility tables (the very release this study graded), documents the lightcurve normalization in its release notes, and describes imaging as reconstruction with quantified uncertainty (REPORTED — EHT Paper III). The measurement is a **visibility table** — time, station pair, `U`, `V`, Stokes-I amplitude, phase, error — with Earth’s rotation dragging baselines into arcs across a mostly empty Fourier plane. Image pipelines minimize a floating-point objective with entropy, total-variation, and compactness priors — and, for Sgr A\* static imaging, start from **lightcurve-normalized** visibilities (`csv_norm/`), not the absolute-jansky `csv/` (VERIFIED from the release’s own README).

Study 07 grades the table first and formalizes the provenance line as a ledger row. The PNG is the adversary class not because images are fabricated bytes — they are not — but because an image is a reconstruction from a named product, and which fork it came from matters. The image-for-measurement conflation lives chiefly in the press layer; this study makes the distinction gradeable rather than rhetorical.

### 4.2 Forcing, archive, adversary

| Piece | Instantiation |
|---|---|
| Clock / track | Observation UT; Earth-rotation `(u,v)` from station geodesy + Sgr A\* direction |
| Archive | EHT 2022-D02-01 — tarball SHA-256 `709f322cd825669e1b30dacb3c9093f016ae3c931a19366b7b29f536ed0baebb` (recomputed byte-exact 2026-08-13; 7,083,259-byte size VERIFIED live) |
| Adversary | Ring images / pipelines; **processing adversary** = `csv_norm/` (Paper III static-imaging input) |
| Integer seals | µJy amplitudes, millidegree phases, kλ baselines, 0.01 h bins |

### 4.3 Frozen law (exact integers)

| Symbol | Sealed threshold |
|---|---|
| Short-baseline median (`q` < 1000 Mλ) | ≥ **1,500,000 µJy** (1.5 Jy) |
| Long-baseline median (`q` ≥ 4000 Mλ) | ≤ **400,000 µJy** (0.4 Jy) |
| Short/long ratio | ≥ **5,000 ppt** (×5) |
| Closure-phase sector concentration | ≥ **200 ppt** |
| Concentration − phase-scramble null | ≥ **20 ppt** (seed **7**) |
| Earth-rotation arc-order agreement | ≥ **400 ppt** |

WIN iff all hold. No float crosses a seal.

### 4.4 Primary grades — the sky’s table (`csv/`)

**8 / 8 WIN** (days 096 & 097 × lo/hi × casa/hops):

| Day | Band | Pipeline | Short (Jy) | Long (Jy) | Ratio | Clos. conc | vs null | Verdict |
|---|---|---|---|---|---|---|---|---|
| 096 | hi | casa | 2.503 | 0.134 | ×18.6 | 452 ppt | +149 | WIN |
| 096 | hi | hops | 2.517 | 0.135 | ×18.6 | 463 ppt | +160 | WIN |
| 096 | lo | casa | 2.443 | 0.119 | ×20.5 | 445 ppt | +137 | WIN |
| 096 | lo | hops | 2.444 | 0.118 | ×20.7 | 473 ppt | +182 | WIN |
| 097 | hi | casa | 2.183 | 0.109 | ×20.1 | 509 ppt | +161 | WIN |
| 097 | hi | hops | 2.186 | 0.103 | ×21.2 | 492 ppt | +136 | WIN |
| 097 | lo | casa | 2.100 | 0.112 | ×18.7 | 499 ppt | +155 | WIN |
| 097 | lo | hops | 2.106 | 0.108 | ×19.5 | 501 ppt | +122 | WIN |

The magnitudes agree with the collaboration’s own literature — ~2.4 Jy short, ~0.1 Jy long at 230 GHz (REPORTED) — corroboration, not independent reproduction.

**A caveat carried on the row itself:** the six thresholds were derived from this same eight-file corpus and set outside its observed envelope, so 8/8 WIN was guaranteed by construction. The WIN is **descriptive**; discriminating power exists only against future products the law has not seen — the Registry’s S4 rows (§4.7). The seal timestamp is ledger-internal, with no third-party time anchor (recorded program debt, §9).

### 4.5 Processing adversary — `csv_norm/`

**8 / 8 MISS**, every file `SHORT_FAIL`. Short-baseline medians fall to ~**0.997–0.999 Jy** — below the 1.5 Jy floor — while ratios remain steep.

This outcome is predictable from the product’s definition: an amplitude divided by the concurrent total-flux light curve has short-baseline amplitude near 1.0 Jy-equivalent **by construction**, and 1.0 < 1.5. The 8/8 MISS therefore re-derives the definition of the normalized product — a provenance fact, not a discovery about the sky; the normalization is a documented step whose stated purpose is to mitigate Sgr A\*’s intraday variability so static full-track imaging is possible (REPORTED — Paper III and the release README). The same graded table shows the ratio ladder (18,785–21,716 ppt) survives normalization intact, and closure phases are exactly invariant under a per-time amplitude scale: **the shape information imaging actually uses passes through the fork unchanged.** What normalization destroys is the absolute jansky appointment on short baselines; naming that is the row’s whole content.

### 4.6 What this means — and what it does not

**Means:**

1. Compact emission at the galactic center is a **table fact** (~2.1–2.5 Jy short baselines) — a number the collaboration itself publishes in this open release.
2. The Fourier profile is a **steep radial ladder** (~×18–×21), not a flat point source.
3. Closure phases are **structured** against a deterministic scramble null.
4. `(u,v)` samples **march on Earth-rotation arcs**.
5. HOPS and CASA **agree** under identical integers.
6. The static-imaging input is a **different table**: the absolute short-baseline appointment lives only in `csv/`. Treating the ring PNG as “the raw data” confuses a reconstruction with a measurement — a confusion the collaboration’s publications avoid and press simplification routinely commits.

**Does not mean:**

- That there is no compact source at Sgr A\* (the opposite is measured).
- That published images are fabricated bytes (they are reconstructions from named products).
- That this study has imaged an event horizon in integer pixels (WIN is network shape, not a ring map).
- That the ring was painted by priors. The quantities that pin morphology pass through normalization unchanged, and the collaboration reports non-ring morphologies in ≤5% of its descattered top-set reconstructions, plus synthetic-data tests against non-ring models (REPORTED — Paper III). This ledger neither confirms nor refutes the ring morphology; the frozen law grades absolute amplitudes and network shape, not images.

**The program’s reading, stated as the program’s reading:** the Milky Way’s center wrote arcs, janskys, and closures into the public table; the ring image is a regularized reconstruction built on the normalized fork, filling sparse Fourier coverage. Naming the fork is provenance discipline — a distinction the collaboration also draws in its own release documentation. Whether the reconstruction’s morphology is right is a question this ledger does not grade, and one the collaboration addresses with its own published tests.

### 4.7 Standing future test — where the law can actually discriminate

The Registry pre-registers named future portal products (2025-D01-01, 2025-D02-01, 2026-D01-01) under the same frozen law — the one place the law earns non-circular discriminating power: a future non-normalized Sgr A\* release passing or failing the 1.5 Jy floor would be a genuine out-of-sample result. The next portal product is a **registry slot, not an open study** — the index reserves OPEN for studies where the program does not have the data (03, 05, 08). The named product codes have not been checked against the portal, and nothing prospective has been graded.

**Entry points:** [Charter](Study-07-SgrA-Milky-Way-Raw-Visibilities) · [Corpus](Study-07-SgrA-Milky-Way-Corpus) · [Results](Study-07-SgrA-Milky-Way-Results) · [Registry](Study-07-SgrA-Milky-Way-Registry)

---

## 5. One method — the isomorphism tables

The sky, sea, and ground studies grade **confinement** and **traveling lag** against a forcing that carries its own clock and track:

| Study | Confinement analog | Traveling-lag analog | Adversary |
|---|---|---|---|
| 01 Eclipse | Umbra / obscuration gate | Station maxima march with cone ground speed | Gannon storm (graded); Halloween 2003 archive reference only, never graded as shear pairs |
| 02 Launch holes | ±3° ascent corridor | Onset T+5–15 min from liftoff; recovery hours | Storms, Tonga TIDs, deorbit burns |
| 03 Flare SIDs | Sunlit hemisphere bands | Lockstep with GOES X-ray peak | Storms, radio bursts |
| 04 Tsunami | Gauge on travel-time chart | Origin → arrival by Huygens time | Surge, meteotsunami, Lamb |
| 05 Forbush | Global NM network, rigidity order | Single L1 shock minute | Pressure, diurnal wave |
| 06 Explosion | P-window vs Lg-window energy | Ladder from catalog clock + point track | Quakes, quarry blasts |
| 07 Sgr A\* | Short/long `q` ladder + closure sectors | Earth-rotation arc order in UT | Ring PNG + `csv_norm` |
| 09 Convective bond | Frozen Int64 thresholds bond traffic + atmosphere to solar forcing | Graded on the forcing’s clock; the page’s standing rule: REFUSED ≠ clear sky | A refusal read as a clear sky |

(Study 08 holds a sealed charter and Study 10 a named archive, with no graded row yet; their pieces are on their pages.)

In the **courts**, the study pages state the analogs differently: confinement becomes **identity** — the claim must bond to a key or invariant the public already owns — and traveling lag becomes **exactness** — an exact integer or rational verdict that replays byte-identically where the adversary’s float forks:

| Study | Identity / exactness analog | Adversary |
|---|---|---|
| 11 Ehrhart | Exact rational volume Δᵈh(0)/d! by lattice counting — **169** points at t = 12 | round(float_vol × t^d) predicts **216** — 5/5 MISS; dilation 1.5 → REFUSED_FLOAT |
| 12 Parallel repetition | Exact rational **27/64** sealed at rounds = 3; receipt STUDY12_QPR_PROVEN | Float-amplitude tensor’s 751/1000; float payload → HTTP 400 |
| 13 Connes rigidity | Integer linking L(γ) = (q,r) ∈ ℤ² — (0,1) RIGID_CYCLE for Z2_a2b | Float spectral-radius estimates of group-algebra norms |
| 14 Protein lattice | 4-char PDB accession bonded to hosted holdings, N = 258,616 (2026-08-24); unknown ids refused | Predicted structures (AlphaFold, float MD, DFT+DL) treated as the molecule |
| 15 Skala DFT | (court DARK — no integer thermochemistry court exists today; numbers CITED) | The continuously improving float functional — 2.72 / 2.8 kcal/mol, each version superseding the last |
| 16 Disease | ICD-10 + DOID bonded pair (E11 + DOID:9352; C67 + DOID:11054), OLS4-fetched | Nickname → REFUSED_NO_STD; unattributed → REFUSED_UNATTRIBUTED; float score → REFUSED_FLOAT |
| 17 Chemistry | InChIKey + integer CID (aspirin CID 2244), PubChem-fetched | SMILES refused as cookbook; float MW / logP → REFUSED_FLOAT |
| 18 Materials | COD / mp-id / space-group 1–230 + Pearson (diamond COD 9008564, 227, cF8) | Minted ids and nicknames do not win; float density → REFUSED_FLOAT |
| 19 Dice | All 120 orderings at exactly **1/120** by rational cross-multiplication — verified as presented, not searched for | Float simplex solvers — epsilon pruning, uncertifiable, forks across machines |
| 20 Frequency claims | Attributed integer Hz / nm only (440 Hz ISO 16; 380/750 nm); none invented | Floats (432.0 → REFUSED_FLOAT) and unattributed claims |
| 21 Stellar dynamo | Integer Jordan step (q,r) + exact rational spin n/d; (0,1), 1/2 → WIN | Float decay curve + bolted Rossby 2.16 ± 0.09 (REPORTED); k = 0.15 → REFUSED_FLOAT |
| 22–25 QMA / #P courts | Exact printable answers: energy **−1/1**, energy **−1**, **WIN**, permanent **6** | VQE landscapes, annealing samples, float-SDP drift, float cancellation |

Same grammar. Different skies — and now different courts. That is the program.

---

## 6. Why this changes how archives are read

### 6.1 From pictures and percent to appointments

Operational and public science default to **percent / amplitude** (ionosphere, cosmic rays), **images** (VLBI, medical, remote sensing), **scalar discriminants** (mb/Ms, P/Lg), and **baseline-corrected residuals** (tide gauges, neutron monitors). Each default invites a magnitude mimic. The shear default asks: **did the response keep the forcing’s appointment?** That question is decidable in integers on public bytes.

### 6.2 Processing forks become first-class adversaries

Study 07 makes explicit what every field quietly knows: the “science ready” product may not carry the absolute appointment. Naming `csv` vs `csv_norm` as separate graded surfaces — and publishing MISS on the imaging fork — is the discipline. It applies wherever a community “corrects,” “detrends,” or “normalizes” before the claim is made — and, in the courts, wherever a float pipeline stands between the structure and the printed answer.

### 6.3 Pre-registration is not optional decoration

Study 1 sealed its 2026 claims 27 days before the eclipse, which occurred with the fingerprints untouched, re-verified byte-exact the day after. Study 07 froze thresholds before the next portal release is scored; Study 06 froze and then graded 6/6 with 3/3 adversary rejects; Study 04 sealed 2/2 gauge wins plus the Sandy adversary win; Study 02’s live claim carries a dated data row with its lockout window sealed in advance. Studies 03 and 05 will freeze before the next flare or L1 shock. Without pre-registration, shape tests collapse into post-hoc storytelling — the failure mode the program was built to exit.

### 6.4 Falsifiers stay on the page

Study 1’s historical Form carries its falsifiers in the seal itself: 15/16 at the original 2026-07-16 seal, the Eugene 2023 miss named, nine false positives published, never renamed — and the registry’s current stamp on that Form (**16/20 · CURE · `cdf43e80f06f10a3`**, 2026-08-18) is named beside that original seal, both on the record. Study 07’s `csv_norm` MISS rows stay published; Study 23’s excited-class MISS (+3) stays beside its ground-state row. Operational failures get the same treatment: the [27-day review](Eclipse-2026-First-27-Days-Review) grades every public promise KEPT or MISSED by name. **A recorded failure is how the experiment learns to see.** Renaming MISS as success is forbidden — and VOID (data absent) is never renamed as either.

---

## 7. Program status board (as of 2026-08-26 — statuses copied from the live index, which is the single source of truth)

| # | Study | Status (index wording) | Headline fact |
|---|---|---|---|
| 1 | Eclipse 2026 | **DATA SEALED** | First look 2026-08-12; frozen integers + 82-row corpus; apex claim not served; verdict ~2026-08-30 |
| 2 | Launch holes | **LIVE CLAIM** | Starlink 10-49 · 2026-08-25T09:33:38Z; alarm package ready, no agency receipt id |
| 3 | Flare SIDs | **OPEN** | No SID corpus; detection threshold not sealed |
| 4 | Tsunami vs surge | **PARTIAL_CORPUS_SEAL** | 2/2 tsunami gauge WIN; Sandy adversary WIN; Tonga VOID_NEED_PRESSURE |
| 5 | Forbush | **OPEN** | No NMDB corpus ingested; thresholds not frozen |
| 6 | Explosion vs quake | **LAW FROZEN** | IC.MDJ 6/6 explosion WIN; 3/3 adversary reject; IU.INCN VOID this seal |
| 7 | Sgr A\* Milky Way | **LAW FROZEN** | csv 8/8 WIN; csv_norm 8/8 MISS |
| 8 | Gaia BH1 | **OPEN** | No µas ingest; thresholds not frozen |
| 9 | Global convective bond | **LIVE CLAIM** | Traffic + atmosphere + solar; frozen Int64 thresholds; REFUSED ≠ clear sky |
| 10 | Fermi gamma shear / dark matter | **CHARTER WITH ARCHIVE** | Named HEASARC weekly photon FITS + paper; no keV ingest; not OPEN |
| 11 | Ehrhart volume | **LAW FROZEN + LIVE CLAIM** | 5/5 WIN · 5/5 float adversary MISS; 169 vs 216 at t = 12 |
| 12 | Quantum parallel repetition | **LIVE CLAIM** | Exact 27/64 sealed at rounds = 3; float 751/1000 is the miss |
| 13 | Connes rigidity | **LIVE CLAIM** | Linking (0,1) RIGID_CYCLE for Z2_a2b |
| 14 | Protein lattice manifold | **LAW FROZEN + LIVE CLAIM** | PDB bond, N = 258,616; occupancy/docking DARK |
| 15 | Skala DFT shear | **CITED** | GMTKN55 2.72 / 2.8 kcal/mol fetched 2026-08-24; court DARK; not OPEN |
| 16 | Disease type | **LAW FROZEN + LIVE CLAIM** | ICD-10 + DOID bonded pairs; nicknames REFUSED_NO_STD |
| 17 | Chemistry InChIKey | **LAW FROZEN + LIVE CLAIM** | Aspirin CID 2244 heads fixture; SMILES refused as cookbook |
| 18 | Material STD | **LAW FROZEN + LIVE CLAIM** | Diamond COD 9008564 heads fixture; minted ids do not win |
| 19 | Go-first dice | **LAW FROZEN + LIVE CLAIM** | 120/120 orderings at exactly 1/120 |
| 20 | Frequency claims (Rife record) | **LAW FROZEN + LIVE CLAIM** | Test claim, not efficacy; no integer table exists, none invented |
| 21 | Stellar dynamo | **LAW FROZEN + LIVE CLAIM** | (0,1) + spin 1/2 → WIN; k = 0.15 → REFUSED_FLOAT |
| 22 | QMA 2-local Hamiltonian | **LIVE ON GLAMA** | Energy −1/1, tool `execute_2local_hamiltonian` |
| 23 | QMA spin glass | **LIVE ON GLAMA** | Ground −1; all-up +3 published MISS, tool `route_spin_glass_manifold` |
| 24 | QMA N-representability | **LIVE ON GLAMA** | WIN; adversary MISS on negative eigenvalue, tool `verify_n_representability` |
| 25 | Exact permanent | **LIVE ON GLAMA** | 3×3 all-ones → 6; above 3×3 refused, tool `execute_exact_permanent` |
| 26 | Master regulator bonds | **OPEN** — charter only; gate not run | Load-bearing regulator sets as bonds — network shape vs mutation magnitude on public TCGA/LINCS keys; test claim, not efficacy — [Study 26](Study-26-Master-Regulator-Bonds) |
| 27 | Exact nuclear scattering | **OPEN** — charter only; gate not run | Glauber all-orders vs optical-limit smear; the field's own 30–50 mb divergence replayed on EXFOR/AME2020 keys — [Study 27](Study-27-Exact-Nuclear-Scattering) |

The index states the data-absent set plainly: **OPEN stays 03 / 05 / 08.** The v2 build-order sentence is retired — 04 and 06, last in that queue, reached PARTIAL_CORPUS_SEAL and LAW FROZEN ahead of it, and 02 carries a dated live claim.

---

## 8. What each sealed win protects

| Study / lane | Downstream |
|---|---|
| 01 / 02 / 03 | GNSS, HF aviation, scheduled ionospheric integrity. The sealed record supports **retrospective grading on definitive archives**; live shape warnings with clocks remain the goal, not a delivered capability ([27-day review](Eclipse-2026-First-27-Days-Review)) |
| 02 — the alarm | The launch-hole record ships in the formats FAA AST, NOAA SWPC, and MAST/ESA already load — one-way IONEX/FITS bridges with every unmeasured quantity labeled **cited**, not painted — so an agency inherits the constraint without inheriting a heatmap ([regulatory-alarm page](Study-02-Regulatory-Alarm)) |
| 04 | Tsunami warning vs surge false alarm — travel-time appointments at gauges |
| 05 | Honest cosmic-ray steps vs weather-shaped phantoms |
| 06 | Treaty-relevant explosion/quake shape in public integer counts |
| 07 | Public clarity about what the galactic-center table contains, and which product fork any image was built on, before any ring is shown |
| 11–13, 22–25 — the Math Court | Hard-problem verification as a **public utility**: any laptop can POST a decimal-string configuration and replay the exact sealed verdict, so quantum-advantage claims can be checked against printable exact answers rather than sampled distributions (§10b) |
| 14–19 — industry keys | Identity by keys the public already owns — PDB accession, ICD/DOID pair, InChIKey, COD number, a published tuple — so a generated nickname or minted identifier never wins a claim |
| 20 — the biology lane | **Test claims, never efficacy.** Study 20 is the standing precedent for medical scoping: no integer table exists, none is invented, and no statement about treating any disease is made or graded |

---

## 9. Limitations and falsification — what is not yet won, and what would count as losing

Every charter page on this surface carries its own honest-limits section; this is the program-level one, covering every open study.

### 9.1 Study 1 — Eclipse 2026

- **Nothing about the 2026 prediction is yet won or lost.** The pinned-archive verdict runs ~2026-08-30; the falsification terms are printed in full in §2 — a verdict failing any of them is a miss and will be published as one.
- **The claim on apex is not served yet** — the index records the serving gap; closing it is program work, not grading work.
- **Named residual caveat:** the Iceland corridor carries a mild solar-wind tail (56 minutes of Bz below −3 nT); Spain is clean. The Leirvogur cross-check is pending.
- **The live operational surface failed while the sealed core held** — itemized with dates in §2.1 and the [27-day review](Eclipse-2026-First-27-Days-Review).
- **The fossil corpus file sits outside version control** — custody rests on the ledger rendering, not git (recorded debt).

### 9.2 Studies 02–06

The v2 sentence “charters only — no result exists” no longer describes this block. What stands now, honestly scoped:

- **Study 02 (LIVE CLAIM):** the ~900 km diameter and 10–70% depletion range are REPORTED and stay **cited, not measured**, until Madrigal posts launch-day TEC (the 2026-08-25 pull: HTTP 200, 0 bytes — a GAP, not a failure). No agency receipt id; TECU in the bridge files is marked UNKNOWN.
- **Study 04 (PARTIAL_CORPUS_SEAL):** Tonga is **VOID_NEED_PRESSURE** — data absent, never a failure; the full multi-gauge Form and pressure branch are still unsealed; the claim on apex is not served yet.
- **Study 06 (LAW FROZEN):** IU.INCN is VOID this seal; the claim on apex is not served yet. Falsifier: the frozen integers failing on the next scored event.
- **Studies 03 and 05 (OPEN):** no thresholds frozen, no corpus ingested, no result exists — favorable or otherwise. If either fails at corpus stage, the failure publishes under the same rules as Study 1’s historical Form.

### 9.3 Study 07 — Sgr A\*

- **The 8/8 WIN is in-sample by construction** (§4.4); discriminating power is prospective only (§4.7), and nothing prospective has been graded — the pre-registered future product codes have not been checked against the portal.
- **The 8/8 MISS is true by construction:** it re-derives the definition of lightcurve normalization — a provenance result, not a sky result.
- **Reproducibility is partial:** the medians, ratio, and both amplitude floors — the fork carrying the `SHORT_FAIL` headline — are algorithm-complete on the public pages. Three of six law components are not (closure-sector triangle-enumeration/binning; the shuffle algorithm behind seed 7; arc-sign agreement); a third party cannot yet reproduce the full six-inequality verdict from published material alone. Publishing those three specifications is open debt.
- **Seal timestamps are ledger-internal** with no third-party time anchor (program-wide debt, §9.7).
- **Falsifier:** a future non-normalized Sgr A\* release failing the 1.5 Jy short-baseline floor — or a normalized one passing it — breaks the law’s reading of the archive.

### 9.4 Study 08 — Gaia BH1

- **The title channel is undecidable today.** Gaia DR3 publishes no epoch astrometry for this source; the astrometric-residual track cannot be constructed until Gaia DR4, announced for **2026-12-02** (VERIFIED). The charter’s VOID-by-inventory clause covers this honestly. The decidable raw channel at charter stage is the radial-velocity series — 115 public epochs across four instruments (VERIFIED).
- **The arithmetic already fixes the honest expectation.** The star’s barycentric orbit subtends ~2,600 integer µas and the RV semi-amplitude is ~66,000 integer m/s — 10³–10⁴ times the sealed integer quanta. The charter-stage expectation, to be pre-registered at S4, is an adversary **WIN on the mass object**: the ~9.3 M☉ dark companion is fully supported by integer appointments. Only the horizon-scale **label** stays at the model layer — the published papers themselves claim no horizon-scale measurement; that narrative is the press layer. The charter grades solution and narrative separately, and this paper adopts that split.
- **The sealed adversary must pin which published solution is graded:** the 2023 discovery mass (9.62 ± 0.18 M☉) and the 2024 refit (9.27 ± 0.10 M☉) differ by ~0.35 M☉ (both REPORTED).
- **The public sampling clock is a forecast:** only ~80% of predicted transits yield accepted data, and as-flown per-transit times are unpublished until DR4. The frozen law must grade windows, not guaranteed transits.

### 9.5 Studies 09 and 10

Study 09 is **published on this surface at LIVE CLAIM** — frozen Int64 thresholds, standing rule REFUSED ≠ clear sky; the v2 “not yet published here” sentence is retired. Study 10 is now **Fermi gamma shear / dark matter** at **CHARTER WITH ARCHIVE** — no keV ingest yet; nothing is graded until ingest runs. The GW-interferometry study holds no number on the live index and has no charter.

### 9.6 The courts and the new wings — honest limits from their own pages

- **The Glama courts (22–25) grade wired fixtures** — the surface’s own words, verbatim: “These are wired fixtures. A presented configuration is verified. An unknown ground state is not searched.” The bounded claim is the point and the limit: nothing about unpresented instances is claimed, and the permanent court refuses instances above its sealed 3×3 bound.
- **In-sample by construction, like Study 07’s WIN:** the fixtures were locked (2026-08-26) by the program; a win on a locked fixture is descriptive of the fixture. Discriminating power lives in the open door — any visitor’s presented configuration (§10b).
- **DARK rows are stated as DARK.** Study 14: lattice occupancy, pocket geometry, and any docking claim are DARK — data absent — until an integer occupancy key exists; the win is the identity bond only. Study 15: no integer thermochemistry court exists today — the court is DARK, the Skala numbers (2.72 / 2.8 kcal/mol, both spellings kept distinct) are CITED as fetched 2026-08-24, and the program does not call the functional useless.
- **Verified as presented, not searched:** Study 19’s tuple is Meyer’s published set, checked, not discovered. Studies 12 and 13’s live rows are single presented instances; their pre-registered criteria are the standing tests.
- **The biology lane claims tests, never treatment.** Study 20 makes and grades no statement about treating any disease — the lane’s standing precedent.
- **Study 21 grades a law against literature, not a telescope observation:** the critical Rossby number is REPORTED; the program’s integer step is the claim under test on future spin-down data.

### 9.7 Program-wide

Seal timestamps across the program are ledger-internal; third-party time anchoring is open debt. Sealed studies 01, 04, and 06 carry an index-recorded gap between sealed data and the served claim on apex. The program was split across three surfaces during the eclipse cycle and has not finished reconciling them. External literature is tagged REPORTED; bytes fetched, hashed, or re-derived by the program are tagged VERIFIED. And program-wide: this paper is a synthesis — **the live index and the study pages are the single source of truth for every status**; where they and this paper disagree, the study page wins.

---

## 10. Reproducibility

| Object | Location |
|---|---|
| Program index | [Shear-Studies-Index](Shear-Studies-Index) |
| Integer-lattice doctrine | [Zero Float · Zero Shear](Zero-Float-Zero-Shear-Paradigm) |
| Study 07 ledger JSON | `corpus/study-07/study07_ledger.json` in [uum8dSolarResearch](https://github.com/gaiaftcl-sudo/uum8dSolarResearch) |
| Study 07 tarball digest | SHA-256 `709f322cd825669e1b30dacb3c9093f016ae3c931a19366b7b29f536ed0baebb` (recomputed byte-exact 2026-08-13) |
| EHT source | https://github.com/eventhorizontelescope/2022-D02-01 |
| Eclipse Forms / live tables | Ledger-rendered status blocks on the wiki — the v2-era measurement: five pages rendered one identical ledger state at 2026-08-13 10:41 UT; the pages now render the 2026-08-18 registry state (historical universality **16/20 · CURE · `cdf43e80f06f10a3` · 2026-08-18T16:42:21Z**); trial tables match the 82-row corpus row-for-row. Recorded debt: the corpus file sits outside version control |
| Eclipse T+1 raw custody | 27 RINEX 30-s files (93 MB, 9 stations, DOY 224 + controls 217/197) plus GOES / solar-wind / magnetometer captures, under a SHA-256 manifest — custody and cross-check; the grading surface remains the pinned Madrigal archive |
| Study 02 custody | 21 RINEX files / 7 CORS stations for Starlink 10-49 (fetched 2026-08-25T20:33:34Z), first-epoch integers on the study page |
| The Math Court | Open MCP endpoint + Glama listing + example client — §10b; [user guide](Math-Court-User-Guide) · [example app](Math-Court-Example-App) · [court page](Affine-Math-Court-Glama) |

Anyone with the charter pages and a terminal can pull the named archives — and since 2026-08-26 the courts upgrade that sentence: anyone with a **JSON-RPC POST** can replay a verdict (§10b). Study 07’s medians, ratio, and amplitude floors are reproducible from the public CSV by the sealed quantization rules on the Results page; the closure-sector, scramble-null, and arc-order components are not yet algorithm-complete in public — recorded debt (§9.3).

## 10b. The Math Court — the program’s public door

The program’s public verification surface is the Affine.Earth Math Court: an open MCP endpoint at `https://affine.earth/language-invariant/mcp`, listed on Glama at `https://glama.ai/mcp/connectors/earth.affine/affine-earth-math-court-remote` (registry `earth.affine/math-court`). No laboratory seat is required — **a JSON-RPC POST is the whole entry fee.**

On 2026-08-26 the program’s own example client walked the entire court and measured (VERIFIED): `tools/list` advertised 21 tools, and the full walk — the `math_court` catalog of 24 domains, one ingest per domain, the four QMA tools by name, the closing bond check — graded **29 LIVE · 0 LEARN of 29**. Floats are refused on the wire: every number travels as a decimal string, so `1/2` is graded and `0.5` is refused — no IEEE-754 value can reach a seal. The QMA fixtures returned exact values (2-local energy −1/1, spin-glass energy −1, N-representability WIN, 3×3 permanent 6), and `verify_jordan_bond` returned `AFFINE_JZ_SHEAR_ZERO` on the uniform lock.

The claim class is stated on the surface itself, verbatim: “These are wired fixtures. A presented configuration is verified. An unknown ground state is not searched.” That bounded claim is the point: the answer is a printable exact string any visitor can replay, not a distribution to be averaged. Reproducibility upgrades from *anyone with a terminal* to *anyone with a JSON-RPC POST*.

Bind and run: [user guide](Math-Court-User-Guide) · [example app](Math-Court-Example-App) · [Affine Math Court on Glama](Affine-Math-Court-Glama).

---

## 11. Conclusion

The shear program is one claim, now restated twenty-seven times:

> **An adversary can copy the size of a response. It cannot keep the forcing’s appointment. Grade the appointment in exact integers on public raw bytes. Freeze the law before the next event. Publish WIN and MISS alike.**

Study 1 proved it on the Moon’s shadow against a superstorm: sixteen true eclipse rows pass confinement and lag, every Gannon row is rejected, and the threshold was frozen 27 days before the 2026 eclipse, which arrived on a verified-quiet day with the fingerprints untouched. The data is sealed; the pinned-archive verdict runs ~2026-08-30; when it resolves, it will be one entry in the track record — no more, no less.

Study 07 proved the same grammar on the Milky Way’s center against the imaging stack: the table carries compact flux, a steep radial ladder, structured closures, and arc order; the normalized imaging input fails the absolute short-baseline seal exactly as its own definition requires; and the ring remains a reconstruction from a named product — which fork it was built on is a ledger row.

The program now flies on five wings. The **sky, sea, and ground studies** (01–10) grade public raw archives against forcings carrying their own clock and track — laws frozen on 06 and 07; sealed data on 01 and 04; a dated live claim on 02; OPEN confined to 03, 05, 08. The **exact-arithmetic courts** (11–13, 22–25) freeze integer and rational law where the standard method computes in floats — four courts live on Glama since 2026-08-26 as public tools that refuse a float at the wire. The **industry-key courts** (14–19) bond every claim to an identifier the public already owns, so a generated nickname never wins. The **applied shears** (20–21) carry the method into contested territory as test claims, never efficacy. And the **biology lane** (14, 16, 20) carries the same discipline into medical territory as test claims, never efficacy. The live board is [Shear-Studies-Index](Shear-Studies-Index) — with each study page, the single source of truth for every status.

And beneath all twenty-seven studies runs the fact that makes the grading portable: none of these verdicts is a floating-point computation on somebody's machine. Every sealed threshold, comparison, and fingerprint lives on a whole-integer lattice — arithmetic that produces the identical byte on every computer ever built, incapable of taking a side. That is what lets a sealed claim be handed to a stranger for grading — since 2026-08-26, through an open door any JSON-RPC POST can walk (§10b) — and why the seals re-verify byte-for-byte weeks after stamping. The doctrine has its own titled page: [Zero Float · Zero Shear](Zero-Float-Zero-Shear-Paradigm). The lattice is not an implementation detail. It is what makes the appointment checkable.

The sky writes geometry. The ledger is how we read it — wins, misses, voids, and the program’s own failures alike.

---

## Document control

| Field | Value |
|---|---|
| Title | Shear Studies White Paper — Shape Against Magnitude Across Twenty-Seven Studies |
| Version | 3.0 |
| Sealed Study 07 law timestamp | 2026-07-23T20:00:59Z |
| Eclipse event date | 2026-08-12 — DATA SEALED; day verified quiet |
| Eclipse pinned-archive verdict | ~2026-08-30, on Madrigal instrument-8000 posting (lag measured at 2–3 weeks between 2026-07-25 and 2026-08-01) |
| Contact surface | This wiki — append-only study pages; white paper revised by versioned section |

**Version 1.0 (2026-07-23)** — initial public synthesis coinciding with Study 07 law freeze and full program index of Studies 01–07. Carried a title/doc-control contradiction (Eight vs Seven Skies) and a prose-only 2026-08-13 resolution date, both corrected in v2.

**Version 2.0 (2026-08-13, T+1)** — post-eclipse revision. Past-tense sweep; new §2.1; Study 1’s frozen falsification numbers printed in full; interpretive claims re-scoped to their evidence class throughout; renumbering note added; new §9 (limitations and falsification); status board refreshed to T+1; title/doc-control contradiction fixed.

**Version 3.1 (2026-08-26)** — Study 27 — Exact nuclear scattering added (charter, board row, abstract wing): the Horiuchi–Suzuki–Wiringa all-orders Glauber result (PRL/PRC 2026) becomes a replay court on EXFOR/AME2020 keys.

**Version 3.0 (2026-08-26)** — program-scale revision synchronized to the live index. The board grows from eight skies to twenty-six studies across five wings, every status line copied from the index (the single source of truth, with the study pages). Study 1 restated as **DATA SEALED** with the apex-serving gap named; Studies 02/04/06 advanced past their v2 charters (LIVE CLAIM; PARTIAL_CORPUS_SEAL with Tonga VOID stated as data absence; LAW FROZEN 6/6 + 3/3); §1.1b/§9.5 corrected for the published Study 09 and the Fermi Study 10; §1.3 carries the index’s controlled vocabulary and §4.7 aligns to it; §5 gains the courts’ identity/exactness table; §7 carries all 25 rows; §8 gains the alarm, Math Court, industry-key, and biology lanes; §9 gains the new wings’ honest limits; new §10b (the Math Court) with §10 pointing at it; Appendix A carries the index’s WIN grade; title, abstract, and conclusion counts recast honestly. Prose compacted with no sealed number changed.

---

## Appendix A — UUM-8D vs IUT (2026-07-24)

A separate program page records the comparison of the program’s runtime formalization against Inter-universal Teichmüller theory’s manual symbolic verification, with public seal and graphics — graded on the live index as a **WIN**: [UUM8D-vs-IUT — WIN](UUM8D-vs-IUT-Topological-Projection).