⚜ PRINCIPIA ORTHOGONA · Vol VI · Roots · WP-96 ← WP-95 · WP-97  ·  Vol VII · Feynman — the action instrument itself →
#Falsifiability
Vol VI · Roots · WP-96 · Received 2026-09-07 · Correction to WP-79 · Closed

The Second Instrument

WP-79 asked which dm³ quantities a measurement could refute, answered for one class of instrument, and wrote the answer down as a general principle. A second class exists. It breaks the symmetry the filter quotients by, and three quantities this corpus had classified as untouchable turn out to be falsifiable.
Methodsymbolic rescaling checks and a dimensional identity
reproduced by wp96-verify.py · 5 blocks, all passing
Claim typecorrection to WP-79’s bookkeeping
the physics is standard and is not claimed as a result
Directionthe correction increases exposure
three quantities become refutable, none becomes safe
Statusclosed
WP-79’s verdict on the k-nacci roots stands unaltered
A filter that sorts your claims into refutable and unrefutable is only as general as the instrument you had in mind when you wrote it. WP-79 had one instrument in mind and did not say so.
COMPUTED produced by the companion script CHECKED verified against a primary source CORRECTION fixes a claim published earlier in this series
§ 1

What WP-79 established, and the sentence that reached too far

WP-79, The Ratio and the Scale, was written in August against a neutral-atom quantum simulator result and asked a good question: of the numbers this corpus publishes, which could an experiment actually refute? Its answer was that a spectrum measurement reports level ratios, En/Em, and that in a ratio every overall scale cancels. So a scale cannot be tested by a ratio, and most of the corpus’s numbers are scales.

That much is right, and block [1] confirms it on its own terms: under a rescaling E → aE the level ratio is unchanged, and under t → bt the product μ T* is unchanged. COMPUTED

The claim, as published

Ratios are falsifiable. Scales are not.

What is wrong is not the finding but its scope. The sentence is stated as a property of measurement. It is a property of spectrum instruments, which was the only class WP-79 had in front of it. The qualifier was never written down, and without it the conclusion classifies as unfalsifiable a set of quantities that a different apparatus refutes routinely.

§ 2

Dimensionless is not scale-invariant

Here is where the reasoning slipped, and it is a distinction worth having in general.

φ = E₁T / ℏ dimensionless — the units cancel
under E → aE : φ → a φ not invariant — the value moves

A phase is dimensionless and is not scale-invariant. WP-79’s filter tests the second property; the eye reads the first; and because the two coincide for the quantities a spectrometer reports, nothing forced them apart. COMPUTED

Block [3] makes the same point in SI base units on the free-fall phase, m g h T ⁄ ℏ, which reduces to 1 — dimensionless — while every factor inside it carries a scale. The quotient remains observable. That is the whole of the objection in one expression.

§ 3

Why ℏ is not a free parameter

The reason the second instrument behaves differently is not subtle, and it is structural rather than technical.

A spectrum instrument reports En/Em. Multiply every energy in the system by a and the report is identical: the apparatus is blind to the overall scale by construction, which is exactly why WP-79’s quotient is the right thing to take.

An action instrument — an atom interferometer — reports S⁄ℏ. And ℏ is fixed by nature. It is not a parameter of the model that a rescaling may carry along; it is a constant the apparatus measures against. So the rescaling symmetry WP-79 quotients by is broken by the apparatus, and a scale that no spectroscopy can touch becomes something an interferometer can refute.

The corrected statement

Ratios are falsifiable by any instrument. Scales are unfalsifiable by spectrum instruments, and falsifiable by instruments whose report is an action in units of ℏ.

WP-79’s sentence is recovered as the first half.

§ 4

The sort, redone against two columns

quantityspectrum instrumentaction instrument
k-nacci roots ηk
level ratios (CFT data)exposedexposed
μmax × T* (Floquet)exposedexposed
T* — the period aloneexposed
μmax — the rate aloneexposed
light-cone velocity vexposed

Three quantities move. T*, μmax and v were classified as beyond the reach of experiment on the strength of a filter that only ever described one apparatus. They are not. COMPUTED

It is worth being explicit about which direction this correction runs. It does not retract a claim or narrow one. It hands three of this corpus’s own numbers to an experiment that could kill them. A correction that increases the surface exposed to refutation is the useful kind, and it is the second in this volume — WP-97 replaced “the only field admitting both is trivial”, which named nothing observable, with “it would show thirty sides”, which a telescope can refuse.

§ 5

What does not move, and cannot

The k-nacci roots ηk stay where WP-79 put them, and no third column would move them either. They are theorems about the roots of polynomials. There is no apparatus whose report bears on whether a polynomial has the root it has, and no instrument class — existing, planned or imagined — changes that.

WP-79 said so, and on this the earlier verdict stands without amendment.

§ 6

Provenance — none of the physics here is new

Stated plainly, because the temptation runs the other way

That an action instrument determines a scale absolutely is the operating principle of atom interferometry as a metrology field, not an observation made here. Cold-atom gravimeters measure g absolutely. Recoil interferometry measures h⁄m and through it the fine structure constant — Parker et al. (2018) with caesium, Morel et al. (2020) with rubidium. Those are absolute determinations of scales, by exactly the mechanism described in §3. CHECKED

Nothing in this note is offered to that community as news. The finding is internal. WP-79 ran a falsifiability audit against one instrument class and stated its conclusion without the qualifier; this note supplies the qualifier and redoes the sort. The physics is the standard physics and is load-bearing only as the reason the correction holds.

Where the distinction is not engaged

The physics is standard in metrology, and § 6 says so without qualification. The place it is not standard is the literature WP-79’s sentence actually came from: the foundations-of-constants argument about how many fundamental constants there are, and which quantities have physical meaning.

The canonical text is Duff, Okun and Veneziano, Trialogue on the number of fundamental constants (JHEP 03 (2002) 023). Okun’s position is that “only dimensionless variables, functions and constants have physical meaning in a theory”; Veneziano’s framework is built on quantities “typically representing ratios of quantities having the same dimensions”, with α and me⁄mp as the examples. CHECKED A dimensionless observable that is not scale-invariant — an action or an interferometric phase, S⁄ℏ — is not treated in that exchange.

The exact size of that observation

This does not make § 2 new physics, and does not upgrade any row of the inventory below. The metrology community operates on the distinction daily; a foundations paper not discussing an observable is not the same as the observable being unknown. What it does explain is how a correct-sounding sentence gets written: “only dimensionless quantities are measurable” is the form the argument takes in the place one goes to read about constants, and in that form it slides into “only scale-invariant quantities are measurable” without anyone noticing the step. WP-79 took that step. The two-column sort of § 5 is what catches it. OPEN whether the foundations literature engages it somewhere this note has not looked.

§ 7

Why this note carries no #Machine Learning tag

Under the rule of 2026-09-01: this paper’s finding is a domain result — which physical quantities an instrument class can refute — occasioned by an external experiment. Machine assistance found it. Assistance used is not the test; assistance as the subject is. Same class as WP-90, which reports defects in an external preprint and carries no tag either.

§ 8

Honest inventory

ClaimStatusWhat it rests on
level ratio is invariant under E → aEComputedblock [1]
μ T* is invariant under t → btComputedblock [1] — WP-79 is correct for its instrument
a phase is dimensionless and not scale-invariantComputedblock [2]; φ → a φ
[m g h T ⁄ ℏ] = 1 in SI base unitsComputedblock [3]
an action instrument breaks the rescaling symmetryFollows§3; ℏ is a constant of nature, not a model parameter
T*, μmax and v are exposed to an action instrumentComputedblock [5]; three quantities move
the k-nacci roots are exposed to no instrumentFollowsthey are theorems about polynomial roots
absolute determination of scales by interferometryClassicalParker 2018, Morel 2020 — cited, not claimed
this note contains new physicsFALSE§6; the finding is internal to this corpus
a third instrument class moves the k-nacci rootsOPENnone is proposed here, and none is expected

Every row marked Computed is regenerated by wp96-verify.py, which sits beside this note and exits non-zero on any failure. The row marked FALSE is there because a reader skimming §3 could reasonably take the interferometry argument for a result of this corpus, and it is not one.

Provenance of this page

wp96-verify.py was committed on 2026-09-06 ahead of the page it belongs to, and CLAUDE.md recorded that the paper was held in a session on another account and that no session should draft a replacement. This page was written on 2026-09-07 at the author’s instruction, rendered from the verification script rather than from that draft. If the held version arrives, the two are renderings of the same checked claim set and differ in prose alone; reconcile them by the script, which is the authority for every number above.

References
  1. WP-79, The Ratio and the Scale (August 2026), this volume — the audit corrected here, occasioned by Sun et al., Nature, 19 August 2026, doi:10.1038/s41586-026-10904-x.
  2. R. H. Parker, C. Yu, W. Zhong, B. Estey, H. Müller, “Measurement of the fine-structure constant as a test of the Standard Model”, Science 360, 191 (2018).
  3. L. Morel, Z. Yao, P. Cladé, S. Guellati-Khélifa, “Determination of the fine-structure constant with an accuracy of 81 parts per trillion”, Nature 588, 61 (2020).
  4. WP-90 — the tagging precedent cited in §7. WP-97, Thirty Was Doing the Work — the other correction in this volume that increases exposure.