# PRE-REGISTRATION — is there a veil at the bottom? Written 29 September 2026, before the run. Covers two worklist pairs, **The Ghost + The Veil** and **The Veil + The Climb Out**, both rank 11, both sharing C, k, p. ## Where the question comes from **The Veil** prints `u* ≈ (C/k)^(1/(1−p)) = 0.00224` — a leading-order gap between heaven and the wall, with the exact gap **0.00276** and heaven at **0.997239**. It describes the balance near **x = +1**: the light's pull and the sideways push meet, and the balance holds. **The Ghost** and **The Climb Out** both describe the **bottom**, and both declare C, k, p — the same three constants. The Veil has never been asked about the bottom. Near **x = −1** the same two forces meet, but differently. As x → −1 the sideways push `F = −2Cx/(1−x²)` **diverges** (away from the wall), while the light's pull `T = k[(1−x)^(−p) − 1]` stays **finite**, tending to `k(2^(−p) − 1)`. A divergence against a finite quantity is a balance, and it should have its own leading-order law. ## The derivation being tested Writing x = −1 + g for small g: `1 − x² = (1−x)(1+x) ≈ 2g`, and `−2Cx ≈ 2C`, so `F ≈ C/g`. Setting `F + T = 0` at leading order: g ≈ C / (k(1 − 2^(−p))) **This is not in the house.** The Veil gives the top only. ## Criterion, fixed before the run **LEG 1 — the run reproduces The Veil as published.** `(C/k)^(1/(1−p))` equals **0.00224 ± 1%**, the true gap at the top equals **0.00276 ± 1%**, and the top attractor sits at **0.997239 ± 10⁻⁵**. If this fails the run is not the house's flow and nothing else counts. **LEG 2 — the bottom's law holds at the working value.** `C/(k(1 − 2^(−p)))` at p = 0.35 matches the true distance of the bottom attractor from x = −1 to within **2%**. **LEG 3 — it is a law, not a coincidence at one p.** Across p ∈ [0.05, 0.95] at 41 points, the relative error of the bottom formula stays below **5%** at every point. **WHAT KILLS IT:** leg 3 failing means the agreement at p = 0.35 is a coincidence, the bottom has no such law, and that is what gets written — not a softened version. Leg 2 failing kills it outright. **REPORTED, NOT SCORED:** which veil is thicker, and which leading-order law is the more accurate. The Veil's own leading order is 0.00224 against an exact 0.00276 — **19% out** — so the published room already carries a loose approximation, and I do not know in advance how the bottom compares. **Prediction, recorded now:** legs 1 and 2 hold. On leg 3 I expect it to hold but I am less sure, because the expansion drops a term of order g relative to 1 and p enters twice. On the comparison I expect the bottom's law to be **more** accurate than the top's, because at the bottom the neglected term is small against a finite pull, whereas at the top both forces are diverging. ## Discipline Verdict assembled from measured quantities. Leg 1 is placed first as a guard, as in entries 29 and 30.