> **These notes are now carried in the paper itself.** They stood beside AP28 for part of 7 September 2026, while no source of record for the paper had been established. One was established the same day, and the paper was republished with the three notes appended after §12 and the U+2728 ornament removed as dress. This file is kept, never deleted, as the record of the interval; the digest it cites, `aecc98d3…`, is the pre-republish PDF, preserved at `proofs/instruments/AP28/AP28_The_Constant_FINAL_v1_0_published_to_2026-09-06.pdf`. Note 2 as built quotes the registry's words for KS-CCC.4a and KS-CCC.4b, which this file paraphrases.

---

# Dated notes beside AP28 The Constant — issued 7 September 2026

**The 420 Code · Studio G · a supersession note; the frozen file is never edited**

AP28 (*The Constant*, Forces and Constants, Notebook IV) is published at
[the420code.org/AP28_The_Constant.pdf](/AP28_The_Constant.pdf), 34 pages,
sha256 `aecc98d3e38a24c9c8bfa3ec3608918f13c6cb823c5098ca9823798ff6253bc3`.
Three matters of record have arisen since its lock. **None alters the body of the
paper, and the paper is not edited.**

---

## 1. Gravity, realised — AP44 supplies the factor this paper provisioned

AP44 (*The Snap*), locked 11 August 2026, supplies the one factor this paper's
architecture provisioned and never booked — the arena's single constitution, one
unpaired ε held once at the first record — and realises the gravitational
constant as

> **G = α_em²¹ × (1 + 1/π) / (1 + α_em) × ℏc/m_e² = 6.6719 × 10⁻¹¹ N·m²/kg²**

a residual of **−0.036 %** against the recommended value, and about 1 ppm from
the atom-interferometry determination (LENS 2014, 6.67191 ± 0.00099 × 10⁻¹¹).

§7's structural value, **6.7206 × 10⁻¹¹ at +0.69 %**, stands beside it as the
record of the omission, never in its place. The channel count, the compounding
law and the puncture factor are untouched. **KS-R.7 as registered in this paper
did not fire, its bar being one per cent.**

## 2. The scatter reading — what has and has not been tested

§7 and §0.7 offer the inter-laboratory scatter in measured G as the leakage
signature of AP06 — a structural explanation, testable in principle and not then
confirmed. AP44's Appendix C took it up under two switches. Both **stand LIVE —
EMPIRICAL** in the Master Kill Switch Registry v5.28; neither is discharged.
Quoted from the registry as it prints them:

> **KS-CCC.4a** — Gap-scale exclusion. No physical characterisation of the
> published apparatuses supports a common leakage term of order 6,900 ppm (the
> Appendix C meta-analysis; **Tier 1 executed twice independently, outcome N in
> both**).

> **KS-CCC.4b** — Scatter structure. Residual boundary-dependence exists in the
> inter-laboratory scatter at AP06's honest size, order 10² ppm (**Tier 2's blind
> boundary coding**; **Tier 1 found the structure at best marginal, p ≈ 0.06–0.2**).

So: gap-scale corrections of the size the reading would need are excluded, and
Tier 1 found the scatter structure at best marginal. **Tier 2's blind boundary
coding is registered and has not yet been run.** AP06's reading of the scatter
weakens on Tier 1's evidence; it is not closed. The realised value's central
claim is unaffected either way, as KS-CCC.4b provides.

## 3. One decorative character, to be removed as dress

The Artist's Note carries one decorative character, **U+2728**, at the end of its
final paragraph — "…its formal verification against the full puncture topology is
owed. [U+2728]" (page 5 of the published PDF — the character is named here by
codepoint rather than reproduced, so this note carries no instance of it). Under
the glyph ruling of 7 September
2026 no decorative glyph appears anywhere in the corpus. Its removal is **dress,
not body**: it happens at this paper's next republish and is recorded there by
codepoint. It is logged here so the record shows it was seen before it was
removed.

---

## Why these notes are beside the paper and not inside it

AP30 was republished on 7 September with its notes appended inside the paper,
because a source existed whose body was byte-identical to the published text.
**For AP28 no such source exists.** The only candidate,
`420code_final/Standalone_Artists_Proofs/AP28_The_Constant_FINAL.docx`, is an
earlier generation: it names the axioms differently in §0.3 (*"Axiom C
(propagation)"* where the published paper reads *"Axiom C (Constraint —
locality)"*, and likewise for B, R and S), cites *"Master Kill Switch Registry
v5.1 (May 2026)"* where the published paper cites the registry without a version,
carries a sentence in §11 that the published paper does not, and has lost every
heading style — all 275 of its paragraphs are Normal.

Rebuilding the body from the published PDF would mean re-typesetting a locked
paper, and an error introduced that way would be invisible in a few years' time.
The corpus's own rule answers it: **frozen files are never edited; corrections
supersede, dated, in the open.** These notes are that supersession. The published
PDF is untouched, so its digest holds and every record that cites it still
resolves.

To append the notes inside AP28 as AP30 has them, the desk would need to produce
for AP28 what it produced for AP30 — the published body, verified paragraph by
paragraph, with the notes appended and the glyph removed. That is a desk task,
and it is recorded as owed.

---

*Frozen files are never edited; corrections supersede, dated, in the open.*