The input ledger

What we assume

A framework that claims zero free parameters owes you a list of everything it takes from outside. This is that list — every external input the theory permits, what happened to it under audit, and what is still owed.

9inputs on the old list

5irreducible after audit

3were unit definitions

0fitted parameters

What this page is not

This is not a set of derivations. What the audit achieved is narrower and more defensible: it reduced the external input set from nine members to five, mostly by identifying which members were never physics in the first place but unit definitions. Two items are left standing with no line drawn through them, and they are named below.

The nine, and what became of each

Input	Standing	Reason
c	unit bridge	Since 1983 the metre is defined from c, so no c-free wavelength measurement exists and there is no independent extraction to be had. The frequency/wavenumber ratio returns c to 7 parts in 10⁹ only because CODATA tabulates cR_∞ as c × R_∞ — an identity that cannot fail. The Mercury-precession closure is withdrawn as a derivation: the ephemeris elements it consumes are produced by fits that model light propagation with c. Physical content retained: one relay speed closes every line.
k_B	unit bridge	SI-exact since 2019 — it defines the kelvin. Temperature itself is counted natively from relay throughput, and the blackbody suite follows from counted lattice modes. A defined conversion has no physics left to derive.
e	unit bridge	The permitted set contains no current dimension, so no combination yields coulombs. There is no charge substance — charge is handed redirection — so "derive e" is a category error. The physical magnitude behind it, the redirection pressure, is derived.
ℏ	collapsed	The circulation action of the fundamental W=1 vortex. Its value has no ℏ-free route — a dimensional result, not a failure of effort — so it merges into the single mass/action seat below rather than standing as its own input.
m_e	collapsed	Mass is the medium's resistance to a change in a form's vector. The mechanism is native and recovers Newton I and II, E = mc², and the equivalence principle as a theorem. The value is the same one seat.
m_p	collapsed	The W=3 trefoil's occlusion cost. Pinned to the measured boundary radius through W+1 = 4 at 0.02%, and to m_e by the pure ratio (4/α)(r_e/R_p). R_p and m_p are one fact, not two — the 0.02% residual is shared between them.
α	extracted	Read off the hydrogen spectrum two independent ways — a pure fine/gross frequency ratio needing no mass or charge, and the movement-budget seat from the ionisation limit. It is the koppa rung of hydrogen's ground state, the same kind of object as the Sun's. Not dialled in from nowhere; not yet derived either.
ℓ_P	stands	No line drawn, and that is a theorem. Every clean route to subdividing the boundary distance reaches ~10³ where ~10⁶¹ is needed. Exactly one length, mass or action anchor is mathematically unavoidable, and this is it.
T_CMB	stands	Measured, and reclassified rather than derived: it is the present convergence temperature — a boundary state, a clock reading, not a constant of nature. Deriving its absolute value would be deriving today's date.

The five that remain

ℓ_P              the one dimensional seed — unavoidable, by theorem
one mass seat    {ℏ ≡ m_e ≡ m_p} — a single anchor, a count of matter
α                extracted from the hydrogen spectrum; derivation open
a₀               measured — the length the seat form rides on
T_CMB            the present boundary state

[R_p ≡ m_p        one fact, not two]
[c, k_B, e        unit bridges — SI definitions, no physics to derive]

The correction that matters

An earlier count said four. It was wrong by one, and the missing member is a length: the route that draws the line through the electron mass consumes a₀, the Bohr radius, for which no derived route exists — so it is a measured input and must be listed. This names the open problem precisely rather than hiding it: the collapse moves the mystery from a mass to a length.

What would actually shorten this list

Close a₀. A derived route to the Bohr radius removes the fifth input outright.
Land the 6π⁵ wake integral. The proton-to-electron mass ratio matches 6π⁵ to 19 parts per million with zero adjustable parameters, and it is the only simple form of its kind that does. Nobody has computed the wake phase-space integral that would produce it, so it stands as an unexplained match — privileged, not earned.
Derive α geometrically. Being dimensionless, α is not blocked by the seed theorem the way a length is. The route is open; the closed-form search for a clean π-expression is ruled out.

Provenance: the ledger is produced by derivelist_rederive.cpp, run against the canonical engine header. It was independently rebuilt on a second toolchain by a checker the same day — output byte-identical — and four findings were raised and applied, two of which corrected labels, one of which corrected this count upward from four to five, and one of which improved an SDT result that had been scored against a stale reference. Circular routes are printed as circular and are not laundered into derivations.

J. C. Harvey, Melbourne.