The marine-hull programme compares small survey-boat hulls with the Holtrop-Mennen regression. Our implementation used one constant, c15 = -1.69385, for every hull. The published method switches it by slenderness: that value below L³/∇ = 512, a linear ramp up to 1726.91, and zero beyond. A third open implementation (Archibald, MIT) carries the full rule, so I coded it (c15_coefficient) and swept a hull at Froude number 0.30.
Wave resistance is unchanged up to L/∇^(1/3) = 8 and rises 2% at 8.5, 8% at 10 and 17% at 12, where c15 reaches zero. Every earlier marine-hull result (the Wigley hull, the Delft yachts, the 1982 worked example) sits below the 512 threshold and reruns byte-identical, so no earlier conclusion moves. The new branch matters for catamaran demi-hulls and AUV-like bodies, and no tank curve checks it yet; that is the next step.
The cross-read also caught a coefficient typo on their side (m1 uses 0.014047; the 1982 example's printed m1 = -2.1274 only comes out with 0.0140407, as in ours). Their test tolerance hid it. The 1984 paper itself was not opened; the rule comes from their docstring, which cites it.
Log, script and results: research/marine-hull/LOG.md, milestone 3p.