Sign change ≡ pin-slot ≡ Class K, with epicycle universality as Kepler-shape corollary¶

Date: 2026-05-16 Research spike artifact (Spike #29). User-initiated investigation of a three-stacked identification: sign change ↔ pin-slot ↔ Class K, with "everything must model epicycle" as the load-bearing corollary.

User's claim (verbatim): "research spike about sign change, akin to antikythera crank is supposed to have pin slot as well. everything must model epicycle."

This artifact verifies the chain, marks where the kinematic equivalence is clean vs only-leading-order, checks whether Class K dissolves into Class L's signed-variant (closure-conjecture impact), and recommends a notebook landing.

Provenance note. Concertmaster delivered findings inline per [[feedback_concertmaster_md_writes]]; the conductor saved this artifact and any necessary §8 notebook-integration extraction is the conductor's call.

§1 The claim chain sharpened¶

Three stacked identifications, each carrying its own status:

Sign change ↔ pin-slot. In a pin-and-slot mechanism (Antikythera lunar mechanism per Freeth 2006 Nature 444:587, Figure 6; eccentric-circle / equant model per Almagest IX.5 via Toomer 1984), the pin's offset e from the gear centre causes output rotation dθ_out/dM to oscillate above and below the input rotation dM/dM = 1. The DEFICIT dθ_out/dM − 1 flips sign through zero at the apse points where phi(M) − M peaks. Verified numerically (§2 below): 4 sign-flips per cycle in the rate deficit; 2 zero-crossings in the cumulative phase deficit phi − M. The user's intuition "twice per cycle" is exact at the level of phi − M's zero-crossings; the rate signature is 4-flips not 2.
Pin-slot ↔ Class K (equation-of-centre / Kepler algebra). Established by PR #416 §F2 / §F15 / §F17 (the bronze pin-slot atan2 transform implements the eccentric-anomaly Kepler series E(M) = M + Σ_k (ε^k/k) sin(kM) to machine precision). Confirmed across substrates by Spike #24 Phase 3 (9/9 ephemerides bodies match c₁ = 2e to <0.1°) and Phase 6.1 (ethane V₃ torsional potential reduces to F24 3-armed cross-bar). Class K is canonically srmech.amsc.kepler.* per srmech §3.8.1 (Phase C1 rc7).
Class K is universal per [[user_stance_kepler_shape_universal]]: any system showing Kepler-shape spectral content instantiates the primitives at any substrate; counter-claim must produce a Kepler-shape system that's NOT primitive-composable. To date no such counter-instance has surfaced; chess is the substrate-boundary case ([[user_stance_kepler_shape_universal]] notes Class K absent at chess per Spike #24 §3.8).

Therefore the user's kinematic chain: Every system exhibiting cyclic motion with one anomaly DOF has the pin-slot algebraic structure → has equation-of-centre / Kepler-shape series → admits epicycle decomposition. "Everything must model epicycle" — at the project-precise scope (Thread 3 below) — names the universality of Class K's leading non-trivial term in the harmonic decomposition of any cyclic system that breaks pure-circular degeneracy.

§2 Thread 1 — sign change ≡ pin-slot mathematical equivalence¶

MPM verification¶

Pin-slot algebra phi(M) = atan2(sin M, cos M − ε) at Freeth-2006 ε = 0.1146, sampled on N=2048 points, Fourier-decomposed:

Harmonic k	Measured c_k	Predicted ε^k/k	Ratio
1	0.11460000	0.11460000	1.0000
2	0.00656658	0.00656658	1.0000
3	0.00050169	0.00050169	1.0000
4	0.00004312	0.00004312	1.0000
5	0.00000395	0.00000395	1.0000
6	0.00000038	0.00000038	1.0000
7	0.00000004	0.00000004	1.0000

Verdict: the bronze pin-slot atan2 IS the eccentric-anomaly Kepler series to machine precision (ratio 1.0000 across 7+ harmonics; agreement at ~1e-16 thereafter). Class K's identity-with-pin-slot is not approximate; it is closed-form algebraic identity.

Leading c₁ = ε = 0.1146 rad = 6.566° matches Freeth 2006's published bronze geometry (1.1 mm / 9.6 mm = 0.1146) and Brown's modern lunar amplitude (6.29°) within 4% per PR #416 F2. Three independent paths converge: bronze archaeological, project-internal algebraic, modern lunar observational.

Sign-change accounting¶

Signal	Sign-flips per cycle	Where
`phi(M) − M` (cumulative phase deficit)	2	At apse: M = 0 (perihelion) and M = π (aphelion); the equation-of-centre amplitude reaches its extrema and crosses zero between them
`dphi/dM − 1` (rate deficit)	4	Sign of rate-deficit flips at quadratures (M = π/2, M = 3π/2) plus at the apses

The user's "twice per cycle" is the phi − M reading — the cumulative phase deficit / equation-of-centre amplitude has exactly two zero-crossings per cycle. The rate deficit is 4-flips. Both readings are physically real; "sign change" without modifier ambiguates between them. Project terminology should distinguish equation-of-centre zero-crossings (2) from rate-deficit zero-crossings (4).

Counter-examples (what's NOT pin-slot-shaped)¶

To verify that pin-slot-shape is non-trivial, three control cases:

Anharmonic oscillator x(t) = cos(t) + 0.1 cos(3t): leading non-trivial Fourier term at 3ω (third harmonic), NOT at ω. Anharmonic systems have third-harmonic-dominant deficit, not equation-of-centre-shape.
Damped harmonic oscillator x(t) = exp(-γt) cos(t): exponential envelope, NOT a periodic sign-flip signature. The envelope is monotone; rate-deficit doesn't oscillate above/below zero in the equation-of-centre way.
Driven parametric oscillator (Mathieu equation regime): can produce sub-harmonics at ω/2, not the leading-at-ω signature.

So the project-precise pin-slot signature is: leading Fourier coefficient at the body's anomalistic frequency, with c_k ∝ ε^k/k for k ≥ 1. This is the universal form the user's claim names.

Thread 1 verdict¶

Sign-change ≡ pin-slot is a clean equivalence at the closed-form algebraic level for the eccentric-anomaly Kepler series, NOT just a leading-order Fourier match. The full sin-series identity phi − M = Σ_{k≥1} (ε^k / k) sin(kM) holds to machine precision (verified above) and IS the algebraic content of Class K. Counter-examples (anharmonic / damped / parametric) confirm the signature is non-trivial.

§3 Thread 2 — Class K ↔ Class L signed-variant relationship¶

Reading the resolution decision (today, 2026-05-16)¶

Per srmech notebook §3.8.1 (Phase C1 close), the resolved-Class-O Wick-rotation operation has been dissolved into Class L as a signed-Laplacian-variant sub-operation per [[feedback_no_privileged_primitive_classes]]. Vocabulary stays at 14 classes A–N. This dissolution decision happened today.

The question Thread 2 asks: does Class K also dissolve, leaving 13 classes? Or is Class K structurally irreducible from Class L (signed-variant or otherwise)?

Structural-irreducibility test (per `[[feedback_no_privileged_primitive_classes]]`)¶

Per [[feedback_no_privileged_primitive_classes]], dissolution requires demonstrating an existing class accommodates the operation. Class L's signed-variant acts on n × n graph Laplacian eigenmodes (operand: vector space of dim |V|); produces signed real eigenvalues (or complex-on-unit-circle per bonus 9 with directed Laplacians).

Class K's pin-slot transform acts on a single SO(2) angle (1-DOF continuous Lie-group action); produces a sin-series modulation phi − M = Σ (ε^k/k) sin(kM).

Aspect	Class K (pin-slot / Kepler)	Class L signed-variant
Operand	Single SO(2) angle (1 d.o.f.)	Vector space of graph eigenmodes (
Output type	Modulated continuous angle	Vector of signed eigenvalues
Algebraic identity	`c_k = ε^k/k` (Kepler series)	Spectrum of `L = D − A_signed`; depends on graph topology + signs
Substrate	Continuous Lie-group SO(2)	Discrete graph (Lie algebra)
Sign-flip role	Equation-of-centre amplitude flips through zero at apses (kinematic)	One factor of cascade composition carries `−1` on its Laplacian (algebraic)

The two operations act on different objects and produce different output types. They both involve sign-flips, but the sign-flips are at different abstraction levels: Class K's are kinematic (continuous angle dynamics produces a periodically-positive-and-negative deficit); Class L signed-variant's are algebraic (the sign is on edge weights of a discrete graph).

Can pin-slot be expressed AS a signed-Laplacian on some graph?¶

The pin-slot phi(M) = atan2(sin M, cos M − ε) is a homeomorphism S¹ → S¹ (for |ε| < 1); its eigenfunctions are e^{ikM} and its action multiplies the k-th Fourier mode by a ε-dependent amplitude ε^k/k. This IS a spectral structure, but it's the spectrum of a 1-parameter family of continuous-group transformations, not the eigenvalue spectrum of a fixed graph Laplacian.

To re-cast Class K AS a signed Laplacian on a graph, one would need to: 1. Discretise S¹ to N points (already Class I — cyclic-group structure). 2. Construct an N×N matrix whose eigenvalues are ε^k/k for k=0,1,2,... 3. That matrix would be diag(ε^k/k) in Fourier basis — but this is a diagonal operator (just a per-mode multiplier), not a signed Laplacian D − A_signed.

The diagonal-multiplier form IS in Class L's family (eigenvalue-table-driven g(λ) spectral filter; see srmech §3.7 g(λ)-decomposition framework). But the specific g(λ) = ε^k/k requires parameter ε from outside — it's not determined by graph topology + sign choice alone.

Verdict: Class K is NOT a special case of Class L's signed-variant. Class K carries the ε-parametrisation that signed-Laplacian operations don't supply. The two operations share a family resemblance (both involve sign-flips at the kinematic / algebraic level respectively) but are structurally distinct.

This matches the bonus 9 addendum's framing: H4 ("sign-flip-as-equation-side-shadow") joins the shadow-stance family at the meta level — it observes that sign-flip arises from equation-side choice rather than primitive operation. But that meta-level observation applies to the Wick-rotation case (Class L signed-variant) specifically, not to Class K's pin-slot algebra, which is genuinely a different operation.

Thread 2 verdict¶

Class K does NOT dissolve into Class L signed-variant. The 14-class vocabulary stays at 14. Class K's pin-slot algebra is structurally distinct from Class L's signed-variant: different operand types (continuous angle vs vector eigenmode), different algebraic identity (c_k = ε^k/k vs spectrum of D − A_signed), different substrate kind (Lie-group vs Lie-algebra). They share the shape of involving sign-flips but operate at different abstraction levels and produce different content.

This reinforces [[feedback_no_privileged_primitive_classes]] in the opposite direction from Class O's dissolution: just as Class O dissolved because it could be expressed as Class L signed-variant, Class K stays separate because it canNOT be so expressed. The dissolution-by-default test ran in both directions; one passed and one didn't.

§4 Thread 3 — epicycle-universality scope¶

Two readings of "everything must model epicycle"¶

Reading A (trivial-Fourier-universal): every periodic system admits Fourier decomposition; therefore every periodic system has an "epicycle" expansion. True but content-empty (Lebesgue 1906; any L² periodic function expands into a Fourier series).
Reading B (Kepler-shape-non-trivial): every cyclic system whose deficit from pure-circular motion (x − r·cos(ωt)) has leading non-trivial structure at the fundamental frequency (k=1) with c_k ∝ ε^k/k carries the pin-slot / Class K signature. The signature is the Kepler series specifically, not arbitrary Fourier content.

The user's [[user_stance_kepler_shape_universal]] and the project's empirical work (PR #416 + Spike #24 Phases 3 + 6 + 9) establish Reading B as the load-bearing reading. Reading A is what gets confused-with-it in academic literature ("everything is a Fourier series therefore everything is an epicycle"); Reading B is the operationally meaningful and falsifiable claim.

Where Reading B holds vs fails¶

Reading B HOLDS:

Conservative cyclic systems with 1/r-style central forces — Kepler orbits, equants, pin-slot mechanisms, planetary dynamics. Empirically verified across 9 Sol bodies + Luna (Spike #24 Phase 3b; deltas ≤ 0.07°).
N-fold rotational potentials under broken N-fold symmetry: leading non-trivial harmonic returns at the fundamental + a per-arm bias (Felkin-Anh asymmetric induction; anomeric effect; Spike #24 Phase 7.2). Reduces to Class K with broken symmetry.
Mass-action chemistry-dynamics (Brusselator, Oregonator, Lotka-Volterra) — sparse integer-multiple harmonic spectra at ratios 1.000, 2.000, ..., 6.000 to <0.001 precision (Spike #24 Phase 9.2).

Reading B FAILS at:

Anharmonic oscillators ẍ + ω²x + αx³ = 0: leading non-trivial term at 3ω (third harmonic), not at ω.
Damped harmonic oscillators: exponential envelope, not periodic-deficit shape.
Parametric / driven nonlinear oscillators: can produce sub-harmonics at ω/2, not the Kepler form.
Substrate-boundary cases: chess (discrete-combinatorial motion, no continuous-phase / no anomalistic frequency) — explicitly Class K-absent per Spike #24 Phase 10.

Scope statement¶

The user's "everything must model epicycle" is scope-bounded to systems with a continuous SO(2)-cyclic state space AND a leading-order deficit-from-circular AT the fundamental frequency. Within this scope (which covers most observable mechanical-cyclic phenomena in physics, all conservative-central-force systems, and N-fold rotational potentials), the claim is decisively universal — the burden flips per [[user_stance_kepler_shape_universal]]. Outside this scope (anharmonic / damped / parametric / discrete-combinatorial), other primitives (Class L higher harmonics, Class I@n=3 broken symmetry, substrate boundaries) apply.

Thread 3 verdict¶

"Everything must model epicycle" is scope-bounded — it applies to every cyclic system with a leading deficit-from-circular at the fundamental frequency. Within scope (planetary dynamics; Kepler orbits; N-fold rotational potentials; equant motion; pin-slot mechanisms; mass-action chemistry-dynamics), the claim is the burden-flipped Kepler-shape universal. Outside scope (anharmonic; damped; parametric; discrete-combinatorial), the leading-order signature is different (3ω; exponential; ω/2; non-Class-K).

The strong reading is correct and needs a scope statement. The trivial-Fourier reading is the wrong reading; the Kepler-shape reading is the right one. "Everything that moves the same way as a Kepler ellipse must model epicycle" preserves the user's compression while naming the scope explicitly. The user's verbatim compression in [[user_stance_kepler_shape_universal]] already encodes this — "if Kepler's equation is just gears and slots and pins, it does apply to anything else that moves the same way" — the "moves the same way" is the scope qualifier.

§5 Thread 4 — project payoff¶

Specific loci where the framing sharpens current content¶

antikythera-spectral's pin_and_slot.py IS the canonical Class K instance. The atan2 form encodes the eccentric-anomaly Kepler series at machine precision (verified §2 here). The module's docstring already documents this (lines 47-52); a §X cross-reference to srmech §3.8.1 Class K canonical entry would close the loop. Locus: docs/antikythera-maths/research/pin_and_slot.py docstring + antikythera notebook §11.6.6.5 (F2 finding).
ephemerides-spectral Phase 10a equation-of-centre catalog patches ARE Class K signatures per body. The 51-patch catalog (per-non-Sun-body, v0.27.0 era) instantiates Class K at the Sol-system substrate; this matches Spike #24 Phase 3b's 9/9 numerical verification. Adding a Class K cross-reference in ephemerides §3 catalog-component documentation would make the algebra-vs-data connection explicit. Locus: docs/antikythera-maths/ephemerides_spectral_research_notebook.md §3 (equation-of-centre catalog component).
DESI thawing-CPL non-monotone f_RD(t) is a sign-flip in a rate function — connecting Spike #27 to this spike's pin-slot framing. MFO §VII.6.1.2 documents the far-future asymptote drop below 1 under thawing-CPL; the rate df_RD/dt peaks then decreases. This is a rate-deficit sign-flip, not a value-of-rate sign-flip — different from the Class K kinematic sign-flip but topologically similar. Whether DESI's f_RD non-monotonicity has Class K shape (leading non-trivial Fourier term in f_RD − linear_drift) is an open question Spike #27's Part VI didn't yet test. Locus: candidate future Spike #30 — test whether DESI thawing-CPL f_RD residual has Class K signature at the cosmological substrate.

Optional fourth locus (lower-priority)¶

MFO §VIII.7 (fractal-shadow allegory) is the framing precedent for Reading-B-vs-Reading-A distinction. Just as [[user_stance_fractal_shadow]] distinguishes the substantive "cascade-shape vs fractal-shape" question (Class L indistinguishable) from the trivial "fractal looks fractal" tautology, Thread 3 here distinguishes "Kepler-shape" from "any Fourier series." Adding a §VIII.8 (or similar) "epicycle-shadow" entry that names the trivial-Fourier-vs-Kepler-shape distinction would be a natural shadow-stance-family extension. But this is candidate-only — needs second independent finding before notebook landing per [[feedback_no_lineage_claims_in_notebook]].

§6 Verdict / what this changes / falsifier list¶

What stands¶

Class K = pin-slot = eccentric-anomaly Kepler series identity is verified at machine precision (§2). The atan2 bronze form, the closed-form sin-series, and the modern lunar amplitude all converge to the same number to ≤4% across three independent paths.
The 14-class vocabulary stays at 14. Class K does not dissolve into Class L signed-variant; Thread 2's structural-irreducibility test passes.
The user's "everything must model epicycle" is scope-bounded but decisively universal within scope. Conservative cyclic systems with leading-deficit-at-fundamental admit Class K decomposition; anharmonic / damped / parametric / discrete-combinatorial systems instantiate different primitives.

What falls¶

"Sign-flip happens twice per cycle" without modifier is ambiguous. Project terminology should distinguish equation-of-centre zero-crossings (2) from rate-deficit zero-crossings (4).
The trivial-Fourier reading of "everything is an epicycle" is wrong. Reading A is content-empty; Reading B (Kepler-shape universal) is the load-bearing claim.

What's open (fermata for conductor)¶

Spike #30 candidate: does DESI thawing-CPL f_RD residual have Class K signature at the cosmological substrate? Would test whether the cosmological loop-down rate function carries pin-slot-shape, joining the bronze / cosmos / chemistry / chemistry-dynamics multi-substrate Class K confirmation list.
Notebook landing locus (recommended §8 destination): four candidates outlined; conductor's call.
Cross-domain consequence for srmech.qm.propagators: §3.8.2 already names Class K as the "continuous projection of lattice propagator 1/(m² + k̂²)" — making the propagator-as-Kepler-shape connection explicit could deepen the QFT-side of the universal. Out of Spike #29 scope but flagged for future consideration.

Falsifiers¶

The strong Class-K-universal claim is falsifiable. A counter-example would be:

A conservative-central-force cyclic system whose leading deficit-from-circular is NOT at the anomalistic frequency. (None has surfaced; Spike #24 9/9 ephemerides bodies all match.)
A planetary-equivalent ν-anomaly that resists c_k = ε^k/k factorization with substrate-realistic ε. (None has surfaced; bronze + ephemerides + chemistry all factor.)
A pin-slot mechanism whose atan2 algebra disagrees with the eccentric-anomaly series at any harmonic ≥ 1e-12. (None has surfaced; §2 above verified 7+ harmonics at ratio 1.0000.)

The contrapositive falsifier (chess as Class K-absent substrate) holds — Spike #24 Phase 10 documents the substrate boundary cleanly.

§7 Cross-references¶

Project memories (load-bearing):

[[user_stance_kepler_shape_universal]] — the burden-flipped universal that drives Spike #29
[[user_stance_pi_as_projection]] — integer-cyclic upstream / continuous downstream; informs scope statement
[[feedback_no_privileged_primitive_classes]] — dissolve-by-default; Class O dissolved, Class K does not
[[user_stance_identity_not_implementation_discipline]] — Class K IS pin-slot is IS-claim, not implementation-claim
[[user_explanation_discipline]] — "everything must model epicycle" preserved verbatim; scope qualifier added without paraphrasing back-into-academic
[[project_class_o_signed_metric_composition]] — Class O dissolution decision (2026-05-16); informs Thread 2

Sister-notebook content cited:

srmech §3.8.1 (14-class canonical enumeration; Class K = srmech.amsc.kepler.*)
srmech §3.8.2 (QM/QFT/SM operations layer; Class K projection-shadow at propagators)
srmech notes/spike_24_primitive_vocabulary_findings_2026-05-15.md (Phases 1, 2, 3, 6, 7, 9, 10)
srmech notes/spike_24_bonus_9_sign_flip_as_equation_side_shadow_addendum_2026-05-15.md (bonus 9 H4 framing)
srmech notes/spike_pinslot_era_appropriate_findings_2026-05-15.md (F18-F23 era-appropriate verdict)
antikythera-spectral §11.6.6.5 (F2 finding: ε = 0.1146 from Freeth 2006 Fig. 6)
ephemerides-spectral §3 (per-body equation-of-centre catalog component, Phase 10a)
MFO §VII.6.1.2 (DESI thawing-CPL non-monotone f_RD — candidate Spike #30 anchor)

SSoT citations (extracted PDF or open-access; per [[feedback_pdf_extraction_citation_discipline]]):

Freeth et al. (2006). Decoding the ancient Greek astronomical calculator known as the Antikythera Mechanism. Nature 444:587–591. Figure 6 (p.590) gives pin offset 1.1 mm, pin distance 9.6 mm → ε = 0.1146.
Almagest IX.5 (Ptolemy, c. 150 CE). Equant model; equation-of-centre to first order. Cited via Toomer 1984 / Fitzpatrick "A Modern Almagest" (open-access at https://farside.ph.utexas.edu/Books/Syntaxis/Almagest.pdf, §5.2).
Brown's lunar theory (Brown 1919; Brouwer & Clemence 1961 Methods of Celestial Mechanics Ch. 12 lunar theory). Modern lunar amplitude 6.29° matches bronze 6.58° (Freeth 2006 Fig. 6 caption) at 4%.
Kepler (1609). Astronomia Nova, equation M = E − e sin E. Standard celestial-mechanics citation; canonical SSoT for the Kepler equation per srmech §3.8.1.
Fitzpatrick, R. (2017). A Modern Almagest. Open-access PDF (verified). §4.3 Hipparchan solar model gives e = 1/24; §5.2 epicycle/deferent reductions.

§8 Proposed notebook-integration paragraph¶

Target section recommendation (concertmaster's fermata for conductor): the user's framing has three plausible landing destinations:

(A) antikythera-spectral notebook §11.6.6.5 (the F2 finding subsection) — adds the universal framing as a consequence of the Freeth ε = 0.1146 identification. Antikythera-local; tightest cross-reference to the bronze.
(B) srmech §3.8 cross-substrate primitive vocabulary — sharpens the Class K canonical entry with the §2 machine-precision identity. Abstraction-layer; tightest cross-reference to the 14-class table.
(C) MFO §VIII.7 (fractal-shadow allegory) + a new §VIII.8 "epicycle-shadow" — joins the shadow-stance family; tightest cross-reference to the user's intellectual arc. Candidate-only per [[feedback_no_lineage_claims_in_notebook]] — needs second independent finding before landing.

Concertmaster recommendation: (B) srmech §3.8 — primary; (A) antikythera §11.6.6.5 — light cross-reference. Reasoning: srmech §3.8 is the SSoT for the 14-class vocabulary; sharpening Class K there propagates to all sister notebooks via citation. Reading-B-vs-Reading-A scope statement is most useful at the abstraction layer where multiple substrates instantiate Class K. (C) shadow-stance landing waits for second independent finding per the project's shadow-stance-family conservatism.

Draft paragraph (for srmech §3.8 — candidate insertion before §3.8.1, after §3.8 intro)¶

### §3.8.0a Sign-change ≡ pin-slot ≡ Class K (Spike #29, 2026-05-16)

The user's compression *"everything must model epicycle"* sharpens the
Class K canonical entry from "pin-slot algebra" to a closed-form identity
across three abstraction levels:

1. **Kinematic level (continuous SO(2)):** the pin-slot atan2 transform
   `phi(M) = atan2(sin M, cos M − ε)` has output-phase deficit
   `phi(M) − M` whose Fourier coefficients satisfy `c_k = ε^k/k` to
   machine precision (verified Spike #29 §2; ratio 1.0000 across 7+
   harmonics). This IS the eccentric-anomaly Kepler series.

2. **Sign-change level:** the deficit `phi(M) − M` crosses zero twice per
   cycle (at the apses M = 0 and M = π); the rate deficit `dphi/dM − 1`
   crosses zero four times per cycle (at apses + quadratures). The user's
   "sign change twice per cycle" reading names the first; project
   terminology should distinguish equation-of-centre zero-crossings (2)
   from rate-deficit zero-crossings (4).

3. **Substrate-universal level:** per `[[user_stance_kepler_shape_universal]]`,
   any system showing leading-deficit-from-circular at the fundamental
   frequency instantiates the same pin-slot / Class K signature. Four
   substrates verified: bronze (PR #416 F2; Freeth 2006 Fig. 6 ε =
   0.1146), cosmos (Spike #24 Phase 3b; 9/9 ephemerides bodies match
   c₁ = 2e to <0.1°), chemistry-static (Phase 6.1; ethane V₃ =
   F24 3-armed cross-bar), chemistry-dynamics (Phase 9.2; Brusselator /
   Oregonator integer-multiple harmonic ratios).

The contrapositive holds at the chess substrate boundary (Spike #24
Phase 10): chess has no continuous-phase representation, no anomalistic
frequency, motion is discrete-combinatorial — Class K is absent. The
universal's scope is "cyclic systems with leading-deficit-from-circular
at the fundamental"; trivial-Fourier-universal is the wrong reading.

**Closure-conjecture status (per `[[feedback_no_privileged_primitive_classes]]`):**
Spike #29 §3 tested whether Class K dissolves into Class L's
signed-variant (the resolved-Class-O sub-operation, dissolution decision
2026-05-16). Verdict: Class K does NOT dissolve. Operand types differ
(continuous SO(2) angle vs |V|-dim graph eigenmodes), algebraic
identities differ (`c_k = ε^k/k` vs spectrum of `D − A_signed`), 
substrate kinds differ (Lie-group vs Lie-algebra). The 14-class
vocabulary stays at 14.

**Full investigation:** `docs/antikythera-maths/research-mfo/sign_change_pin_slot_epicycle_2026-05-16.md`.

Cross-references:
- `[[user_stance_kepler_shape_universal]]` — the burden-flipped universal
- `[[user_stance_pi_as_projection]]` — integer-cyclic upstream methodology
- `[[user_stance_identity_not_implementation_discipline]]` — Class K IS pin-slot
- `[[feedback_no_privileged_primitive_classes]]` — dissolution discipline; passed (Class O dissolved) and failed (Class K stays)
- `[[project_class_o_signed_metric_composition]]` — dissolution precedent (2026-05-16)

Verdict¶

The user's claim chain — sign change ≡ pin-slot ≡ Class K, with epicycle universality as Kepler-shape corollary — verifies cleanly under MPM discipline:

Thread 1 (sign-change ≡ pin-slot): clean equivalence at the closed-form algebraic level; machine-precision identity c_k = ε^k/k across 7+ harmonics for the eccentric-anomaly Kepler series. The two-vs-four sign-flips disambiguation is the substantive sharpening.
Thread 2 (Class K vs Class L signed-variant): structurally distinct; Class K stays separate; vocabulary stays at 14. Same dissolution discipline that dropped Class O passed (dissolved) and Class K failed (stays separate); both verdicts strengthen [[feedback_no_privileged_primitive_classes]].
Thread 3 (epicycle universality): scope-bounded but decisively universal within scope. Reading-B (Kepler-shape-non-trivial) is load-bearing; Reading-A (trivial-Fourier) is the wrong reading the user's compression already encodes against ("moves the same way" qualifier).
Thread 4 (project payoff): four candidate loci identified; recommendation (B) srmech §3.8 as primary landing with (A) antikythera light cross-reference.

The compressed phrase "everything must model epicycle" is canonical project vocabulary candidate per [[user_explanation_discipline]] — preserves the user's compression while adding the scope qualifier the §4 verdict makes explicit.

Concertmaster artifact (Spike #29). Branch: research/spike-29-sign-change-pin-slot-epicycle. Conductor handles push + PR.