@essrt/physics 5.0.0 → 6.0.0

package/README.md

@@ -83,7 +83,7 @@ day lengths), `cardinal` (solstice/equinox JD, RA, year length), `moon`
 JD(UT) axis — ecliptic longitude, latitude and distance from the shared
 Meeus series over framework-native arguments), `eclipse` (geocentric
 solar/lunar eclipse search over a JD window, greatest-eclipse convention),
-`climate` (the L1 orbital-forcing formula) and `planets` (the Fibonacci-law
+`climate` (the L1 orbital-forcing formula) and `planets` (the retired integer-law
 records and orientation at epoch). Validation targets are refused as
 inputs — a counterfactual cannot move the goalposts it is judged by.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@essrt/physics",
-  "version": "5.0.0",
+  "version": "6.0.0",
   "description": "ESSRT (Expanding Solar System Resonance Theory) physics core — pure, injectable-constants, no I/O, no globals, no DOM. The complete model incl. fitted coefficients; every published version immutably ships one recorded model identity.",
   "essrt": {
     "modelVersion": "v14.0",

package/src/deltat/deep-time.cjs

@@ -68,6 +68,12 @@ const { createComposedPrecession } = require('../earth/precession-composed.cjs')
  * @property {(year: number) => number} cycleLodSumAt - gated δLOD sum (incl. swing)
  * @property {(year: number) => number} swingLodAt - gated swing δLOD alone
  * @property {(year: number) => number} swingLodRateAt - gated analytic swing rate
+ * @property {() => number} precessionPeriodJ2000YearsFn - the model's DERIVED
+ *   J2000 axial precession period: the certified year-length laws' beat at
+ *   2000, sid/(sid − trop) ≈ 25,771.4 yr (plan 06 S5 — one J2000 precession
+ *   reading; the same anchor the hybrid obliquity self-anchors on). Read
+ *   LAZILY on the first composed-rate use, never at construction: every
+ *   runtime's year laws read THIS factory's bases.
  * @property {(tMa: number) => number} [lEmAtAgeKgm2S] - OPTIONAL time-dependent
  *   Earth-Moon angular momentum (the Driver-1½ solar channels,
  *   recession-history.cjs). Absent → the J2000 constant, which the budget
@@ -117,12 +123,15 @@ function createDeepTimeLod(deps) {
     return K.holisticYearJ2000 * LOD_s / K.lodNowH13Seconds;
   }
 
-  // The unit H(t) ≡ 13 × the composed lunisolar precession period (plan 06
-  // D6 → Phase 3): ψ̇(t) = [ω(t)/ω₀]·p₀·[f_S + (1 − f_S)(a₀/a_M(t))³], p₀ =
-  // 1,296,000/(H₀/13). ONE formula home: earth/precession-composed, built
-  // here on this factory's own LOD and recession history.
+  // The composed lunisolar precession rate (plan 06 D6 → Phase 3 → S5):
+  // ψ̇(t) = [ω(t)/ω₀]·p₀·[f_S + (1 − f_S)(a₀/a_M(t))³], p₀ the model's
+  // DERIVED J2000 rate 1,296,000/axial0 (the certified year-length laws'
+  // beat at 2000 ≈ 25,771.4 yr — NOT 1,296,000/(H₀/13) = 50.245 ″/yr, the
+  // fit anchor's reading, 0.086 % slow; S5). ONE formula home:
+  // earth/precession-composed, built here on this factory's own LOD and
+  // recession history; the anchor resolves lazily on the first rate use.
   const composed = createComposedPrecession({
-    p0ArcsecPerYr: 1296000 / (K.holisticYearJ2000 / 13),
+    p0ArcsecPerYr: () => 1296000 / deps.precessionPeriodJ2000YearsFn(),
     solarShare: K.precessionSolarShareJ2000,
     lodSecondsAtAge,
     lodJ2000Seconds: K.lodNowH13Seconds,
@@ -131,11 +140,13 @@ function createDeepTimeLod(deps) {
     yearToTMa: /** @param {number} year */ (year) => (2000 - year) / 1e6,
   });
 
-  /** The unit H(t) = 13·T_p,composed(t) — H is 13 lunisolar precession periods
-   * at every epoch (plan 06: H is a UNIT). Written as H_era / [f_S + (1 − f_S)
-   * (a₀/a_M)³] — algebraically 13·1,296,000/ψ̇_composed, spelled in the SAME
-   * operations as layer0's holisticHCore so the layer0 gate holds the twins
-   * bit-identical (13·T_p differs by one ULP). ≡ H_era wherever (a₀/a_M)³ ≈ 1.
+  /** The unit H(t) = H_era / [f_S + (1 − f_S)(a₀/a_M)³] — the internal unit
+   * scales WITH the composed lunisolar precession period (H(t)/T_p(t) =
+   * H₀/axial0 = 13.011 at every epoch) but is NOT 13 of them: H₀ was fitted
+   * on the perihelion-of-date beat, and the "H = 13·T_p" claim is retired
+   * (plan 06 S5; docs/retired-record.md). Spelled in the SAME operations as
+   * layer0's holisticHCore so the layer0 gate holds the twins bit-identical.
+   * ≡ H_era wherever (a₀/a_M)³ ≈ 1. Identifier kept (plan 06 P4).
    * @param {number} t_Ma @returns {number|null} */
   function hAtAge(t_Ma) {
     const hEra = eraClockHAtAge(t_Ma);
@@ -143,12 +154,20 @@ function createDeepTimeLod(deps) {
     return hEra === null || term === null ? null : hEra / term;
   }
 
-  /** The composed lunisolar precession rate, ″/yr (= 1,296,000·13/H(t)).
+  /** The composed lunisolar precession rate, ″/yr — p₀·[ω/ω₀]·[f_S + (1 − f_S)
+   * (a₀/a_M)³] on the derived J2000 anchor (S5). ≡ 1,296,000·(H₀/axial0)/H(t).
    * @param {number} t_Ma @returns {number|null} */
   function lunisolarPrecessionRateArcsecPerYrAtAge(t_Ma) {
     return composed.composedRateArcsecPerYrAtAge(t_Ma);
   }
 
+  /** The mean lunisolar precession period T_p(t), years — 1,296,000/ψ̇(t);
+   * 25,771.4 at J2000 (the model's one J2000 precession reading, S5).
+   * @param {number} t_Ma @returns {number|null} */
+  function lunisolarPrecessionPeriodYearsAtAge(t_Ma) {
+    return composed.composedPeriodYearsAtAge(t_Ma);
+  }
+
   /** Sidereal year seconds (Kepler under linear mass loss, dT/T = −2 dM/M).
    * @param {number} t_Ma @returns {number} */
   function siderealYearSecondsAtAge(t_Ma) {
@@ -165,7 +184,16 @@ function createDeepTimeLod(deps) {
     return K.meanSiderealYearJ2000Seconds * (1 - mass_loss_fraction) * (1 - mass_loss_fraction);
   }
 
-  /** @param {number} t_Ma @returns {number} tropical = sidereal · (1 − 13/H(t)) */
+  /** @param {number} t_Ma @returns {number} tropical = sidereal · (1 − 13/H(t)).
+   * The 13/H(t) here is the unit's CALENDAR convention (one turn per H₀/13
+   * of the unit — the kinematic day/year identities on H/(H − 13) and the
+   * deep JD↔year calendar ride it), NOT a precession claim: the published
+   * precession period is the composed T_p (S5), 0.086 % apart. Kept
+   * unchanged in S5 so nothing certified moves (calendar, goldens, CSV);
+   * plan-06 D2 DECIDED: relabel, not re-base — the 0.086 % is the fit
+   * anchor's convention (H₀/T_p = 13.011, constant at every epoch beyond
+   * ±2 Myr, measured), and re-basing would move every comb coefficient
+   * for a rounding-level gain. This tier stays a named device. */
   function tropicalYearSecondsAtAge(t_Ma) {
     const sidSec = siderealYearSecondsAtAge(t_Ma);
     const Ht = hAtAge(t_Ma);
@@ -331,7 +359,7 @@ function createDeepTimeLod(deps) {
 
   return {
     lodSecondsAtAge, lodSecondsAtAgeWithAlpha, lodHoursAtAge, hAtAge,
-    lunisolarPrecessionRateArcsecPerYrAtAge,
+    lunisolarPrecessionRateArcsecPerYrAtAge, lunisolarPrecessionPeriodYearsAtAge,
     // the composition's terms (plan 06 Phase 3 S3, the lunisolar surface): (a₀/a_M)³ and f_S + (1 − f_S)(a₀/a_M)³
     lunarTorqueFactorAtAge: composed.lunarTorqueFactorAtAge,
     precessionTorqueTermAtAge: composed.torqueTermAtAge,

package/src/earth/precession-composed.cjs

@@ -12,15 +12,25 @@
  * derived from the shared constants (two-body point-mass torques with the
  * (1 − e²)^(−3/2) eccentricity factors and the lunar-inclination factor
  * 1 − 3/2·sin² i_M; ≈ 0.316, the textbook value). At J2000 every factor is 1.
- * Plan 06 Phase 3: the unit H(t) IS 13 of these periods at every epoch
- * (deltat/deep-time.cjs hAtAge builds on this factory), so H(t)/13 and the
- * composed period are one quantity. The pre-Phase-3 structural clock
- * 13·1,296,000/H_era(t) (pure ω-scaling, H_era = H₀·LOD/LOD₀) survives only
- * as the FROZEN ERA CLOCK's named phase convention (eraClockHAtAge) and in
- * docs/retired-record.md, with its record: it read 70.9 ″/yr against
- * Lantink et al. 2022's 108.6 ± 8.5 at 2.46 Ga and 65.2 against Meyers &
- * Malinverno 2018's 85.79 ± 2.72 at 1.4 Ga, where the composed rate reads
- * 104.5 and 86.5.
+ *
+ * THE ANCHOR p₀ (plan 06 S5 — one J2000 precession reading): p₀ is the
+ * model's DERIVED J2000 rate, 1,296,000/axial0 with axial0 the certified
+ * year-length laws' beat at 2000 (sid/(sid − trop) = 25,771.4 yr = 50.2883
+ * ″/yr, IAU 50.2879 to 8×10⁻⁶) — the SAME anchor the hybrid obliquity
+ * self-anchors on. It is NOT 1,296,000/(H/13): H₀ = 335,317 was fitted on
+ * the 1246 AD perihelion–solstice alignment (the perihelion-of-date beat),
+ * so H₀/13 = 25,793.6 yr is the fit anchor's reading, 0.086 % slow, and
+ * is not a period of anything the model computes (nor a window mean: the
+ * published period averages 25,598 yr over ±26 kyr). The unit H(t) =
+ * H_era/[torque term] (deltat/deep-time.cjs hAtAge) scales WITH the
+ * composed period — H(t)/T_p(t) = H₀/axial0 = 13.011 at every epoch — but
+ * is not 13 of them; the 13 is a fit-era label that leaves the physics.
+ * The pre-Phase-3 structural clock 13·1,296,000/H_era(t) (pure ω-scaling,
+ * H_era = H₀·LOD/LOD₀) survives only as the FROZEN ERA CLOCK's named phase
+ * convention (eraClockHAtAge) and in docs/retired-record.md, with its
+ * record: it read 70.9 ″/yr against Lantink et al. 2022's 108.6 ± 8.5 at
+ * 2.46 Ga and 65.2 against Meyers & Malinverno 2018's 85.79 ± 2.72 at
+ * 1.4 Ga, where the composed rate reads 104.6 and 86.6.
  *
  * Pure factory: the tidal chain is injected (each runtime's own LOD and
  * moon-distance evaluators), no artifact reads, no globals.
@@ -47,15 +57,18 @@ function computeSolarTorqueShare(c) {
 
 /**
  * @param {{
- *   p0ArcsecPerYr: number,
+ *   p0ArcsecPerYr: number | (() => number),
  *   solarShare: number,
  *   lodSecondsAtAge: (tMa: number) => (number | null),
  *   lodJ2000Seconds: number,
  *   moonDistanceMetresAtAge: (tMa: number) => (number | null),
  *   moonDistanceJ2000Metres: number,
  *   yearToTMa: (year: number) => number,
- * }} deps - p0 = 1,296,000/(H/13) (the model's J2000 rate); lodJ2000Seconds
- *   the SAME day basis lodSecondsAtAge(0) returns.
+ * }} deps - p0 = the model's derived J2000 rate, 1,296,000/axial0 (S5; a
+ *   number or a LAZY getter — the callers' certified year laws read the
+ *   deep-time factory this is built inside, so the anchor is resolved on
+ *   the first RATE use, never at construction; the torque term needs no
+ *   anchor); lodJ2000Seconds the SAME day basis lodSecondsAtAge(0) returns.
  * @returns {{
  *   composedRateArcsecPerYrAtAge: (tMa: number) => (number | null),
  *   composedPeriodYearsAtAge: (tMa: number) => (number | null),
@@ -69,7 +82,18 @@ function computeSolarTorqueShare(c) {
  * }}
  */
 function createComposedPrecession(deps) {
-  const { p0ArcsecPerYr: p0, solarShare: fS } = deps;
+  const { solarShare: fS } = deps;
+  /** @type {number|undefined} */
+  let p0Memo;
+  /** The J2000 anchor, resolved once on first use (lazy getter allowed). */
+  const p0 = () => {
+    if (p0Memo === undefined) {
+      const v = typeof deps.p0ArcsecPerYr === 'function' ? deps.p0ArcsecPerYr() : deps.p0ArcsecPerYr;
+      if (!(v > 0)) throw new Error(`createComposedPrecession: p0ArcsecPerYr must resolve to a positive rate, got ${v}`);
+      p0Memo = v;
+    }
+    return p0Memo;
+  };
   /** @param {number} tMa */
   const lunarTorqueFactorAtAge = (tMa) => {
     const a = deps.moonDistanceMetresAtAge(tMa);
@@ -88,7 +112,7 @@ function createComposedPrecession(deps) {
     const lod = deps.lodSecondsAtAge(tMa);
     const term = torqueTermAtAge(tMa);
     if (lod === null || !(lod > 0) || term === null) return null;
-    return (deps.lodJ2000Seconds / lod) * p0 * term;
+    return (deps.lodJ2000Seconds / lod) * p0() * term;
   };
   /** @param {number} tMa */
   const composedPeriodYearsAtAge = (tMa) => {
@@ -96,13 +120,13 @@ function createComposedPrecession(deps) {
     return r === null ? null : 1296000 / r;
   };
   /** ψ̇(t)/ψ̇₀ — the factor a consumer scales ITS OWN certified J2000
-   * anchor by (the hybrids anchor on the of-date beat 25,771.4 yr, not on
-   * p₀'s 25,793; the ratio is what D6 changes, the anchor stays theirs).
-   * ≡ 1 at J2000; null out of the tidal chain's domain.
+   * anchor by (since S5 p₀ and the hybrids' anchor are the same derived
+   * value, 25,771.4 yr; the ratio is what D6 changes). ≡ 1 at J2000; null
+   * out of the tidal chain's domain.
    * @param {number} tMa */
   const composedRateRatioAtAge = (tMa) => {
     const r = composedRateArcsecPerYrAtAge(tMa);
-    return r === null ? null : r / p0;
+    return r === null ? null : r / p0();
   };
   return {
     composedRateArcsecPerYrAtAge,
@@ -113,7 +137,7 @@ function createComposedPrecession(deps) {
     torqueTermAtAge,
     lunarTorqueFactorAtAge,
     solarShare: fS,
-    p0ArcsecPerYr: p0,
+    get p0ArcsecPerYr() { return p0(); },
   };
 }
 

package/src/index.js

@@ -227,19 +227,22 @@ export const createModel = (constants = GENERATED, opts = {}) => {
     hash,
 
     /**
-     * H-lattice periods — the structural identities, in years.
+     * The fit-era lattice identities H/13, H/3, H/16, in years — DEVICE tier.
      *
-     * PURE ALGEBRA OVER THE CONTEXT. No epoch, no formula, no fit. These are
-     * the divisor relationships H/13, H/3, H/16 that define the lattice, and
-     * they exist here in Phase 5 for one reason: a hash-only counterfactual
-     * test would still pass if `createModel` ignored its argument entirely.
-     * Something has to READ the context and return a number for injection to be
-     * demonstrated end to end.
+     * PURE ALGEBRA OVER THE CONTEXT. No epoch, no formula, no fit. They exist
+     * here for one reason: a hash-only counterfactual test would still pass if
+     * `createModel` ignored its argument entirely. Something has to READ the
+     * context and return a number for injection to be demonstrated end to end.
      *
-     * The motion model is Phase 6. This is not it.
+     * These are NOT the published periods (plan 06 S5): the axial precession
+     * period is `epoch.axialPrecessionYearsAtYear` (25,771.4 yr at J2000, the
+     * of-date year laws' beat); H/13 = 25,793.6 is the fit anchor's reading,
+     * 0.086 % slow — H₀ was fitted on the perihelion-of-date beat. Likewise
+     * H/3 and H/16 against the chain's apsidal and perihelion-of-date periods.
+     * The property names keep their historical form (plan 06 P4).
      *
      * `divisor` and `period` are never interchangeable (CLAUDE.md): 13 is the
-     * divisor, H/13 years is the period. The names say which.
+     * divisor, H/13 years is the identity's value. The names say which.
      *
      * @returns {{axialPrecessionPeriodYears: number, inclinationPrecessionPeriodYears: number, perihelionPrecessionPeriodYears: number}}
      */

package/src/layer0/index.js

@@ -149,7 +149,10 @@ export const createEpochPrimitives = ({ params: p, alphaAtAgeMa }) => {
   };
 
   /**
-   * Tropical year — sidereal less one axial-precession turn, H/13.
+   * Tropical year — sidereal less one calendar turn of the unit, 13/H(t): the
+   * unit's CALENDAR convention (plan 06 S5 — a device identity, 0.086 % from
+   * one turn per the published precession period; D2 decided: it stays a
+   * named device, never re-based).
    * Falls back to the J2000 H past the asymptote, matching the shipped chain.
    * @param {number} t age in Ma
    * @returns {number} seconds

package/src/model.js

@@ -285,6 +285,9 @@ export function assembleModel(C, F, laws = {}, secularSeriesArtifact = /** @type
     },
     moonDistanceMetresAtAge,
     moiFactorAtAge: earthMoiFactorAtAge,
+    // S5: the composed clock's J2000 anchor — the certified year laws' beat
+    // at 2000 (ONE home below, shared with the hybrid's self-anchor); lazy.
+    precessionPeriodJ2000YearsFn: () => certifiedAxialPrecessionJ2000Years(),
     siderealYearDaysFourierAt: (year) => evalSiderealYearFourierIAU(year),
     cycleLodSumAt: dtCycleLodCorrectionSum,
     swingLodAt: (year) => dtCycles.swingLodSecondsAt(year),
@@ -435,6 +438,24 @@ export function assembleModel(C, F, laws = {}, secularSeriesArtifact = /** @type
     const days = deepLod.eraClockYearInDaysAtAge(yearToTMa(year));
     return days === null ? meanSolarYearDays : days;
   };
+  /** THE J2000 precession anchor (plan 06 S5 — one J2000 precession
+   * reading): the certified of-date year laws at 2000, sid/(sid − trop) =
+   * 25,771.4 yr = 50.2883 ″/yr (IAU 50.2879 to 8×10⁻⁶) — identical to the
+   * engine/browser Direct twins. ONE home: the composed clock's p₀, the
+   * hybrid's self-anchor and the lunisolar surface all read it. The former
+   * H/13 reading (25,793.6) is the fit anchor's, not a period. Memoized;
+   * hoisted so the deep-time factory's lazy dep can name it.
+   * @returns {number} */
+  function certifiedAxialPrecessionJ2000Years() {
+    if (certifiedAxial0Memo === undefined) {
+      const sidLaw2000 = evalYearFourier(2000, siderealYearDaysBase(2000), F.SIDEREAL_YEAR_HARMONICS);
+      const solLaw2000 = evalYearFourier(2000, tropicalYearDaysBase(2000), F.TROPICAL_YEAR_HARMONICS);
+      certifiedAxial0Memo = sidLaw2000 / (sidLaw2000 - solLaw2000);
+    }
+    return certifiedAxial0Memo;
+  }
+  /** @type {number|undefined} */
+  let certifiedAxial0Memo;
   /** @param {number} year @returns {number} */
   const anomalisticYearDaysBase = (year) => {
     const tMa = yearToTMa(year);
@@ -524,9 +545,7 @@ export function assembleModel(C, F, laws = {}, secularSeriesArtifact = /** @type
     // movement's realized precession 25 yr off the certified value — the
     // hidden split behind the API's 365.242204 tropical mean (found in the
     // owner's one-implementation drive).
-    const sidLaw2000 = evalYearFourier(2000, siderealYearDaysBase(2000), F.SIDEREAL_YEAR_HARMONICS);
-    const solLaw2000 = evalYearFourier(2000, tropicalYearDaysBase(2000), F.TROPICAL_YEAR_HARMONICS);
-    const axial0 = sidLaw2000 / (sidLaw2000 - solLaw2000);
+    const axial0 = certifiedAxialPrecessionJ2000Years();   // S5: the ONE home
     // D6 → Phase 3 (plan 06): the deep-time ψ̇(t) the hybrid precesses on is
     // the COMPOSED lunisolar rate, and since Phase 3 the unit H(t) IS
     // 13 × that period (deltat/deep-time.cjs hAtAge), so the injection is
@@ -1216,19 +1235,31 @@ export function assembleModel(C, F, laws = {}, secularSeriesArtifact = /** @type
   // ── The assembled surface ─────────────────────────────────────────────────
   // Plan 06 Phase 3 S2 — THE published of-date precession period and mean
   // tropical year: the one-family route (B) inside the banked tiers
-  // (|Δyear| ≤ ONE_FAMILY_WINDOW_YEARS), the unit's secular mean beyond
-  // (H(t)/13; the tidal-chain year). Shared by `epoch` and `lunisolar`.
+  // (|Δyear| ≤ ONE_FAMILY_WINDOW_YEARS), the composed lunisolar period
+  // beyond (S5: on the derived J2000 anchor, so the two faces agree at
+  // 25,771.4 at J2000 and the deep tail is the same clock scaled; the
+  // tidal-chain year). Shared by `epoch` and `lunisolar`.
   /** @param {number} year @returns {number} */
   const publishedAxialPrecessionYearsAtYear = (year) => {
     if (Math.abs(year - 2000) <= ONE_FAMILY_WINDOW_YEARS) return yearLengthsM.axialPrecessionYearsAtYear(year);
-    const h = deepLod.hAtAge(yearToTMa(year));
-    return (h === null ? H : h) / 13;
+    const p = deepLod.lunisolarPrecessionPeriodYearsAtAge(yearToTMa(year));
+    return p === null ? certifiedAxialPrecessionJ2000Years() : p;
   };
   /** @param {number} year @returns {number} */
   const publishedTropicalYearSecondsAtYear = (year) => {
     if (Math.abs(year - 2000) <= ONE_FAMILY_WINDOW_YEARS) return yearLengthsM.tropicalYearSecondsAtYear(year);
     return deepLod.tropicalYearSecondsAtAge(yearToTMa(year));
   };
+  // Plan 06 S6 — |s₃|, the dominant nodal mode of Earth's orbit: the
+  // largest-amplitude ζ mode of the banked deep secular modes (the recipe
+  // the registry's `eclPrecYears`/`obliqCycleYears` and the browser's
+  // obliquity-beat helper use). ONE home here; the API reads it.
+  const s3ArcsecPerYr = (() => {
+    const z = DEEP_MODES_ARTIFACT.earthZeta
+      .filter((m) => Math.abs(m.omegaRadPerYr) > 1e-9)
+      .sort((a, b) => Math.hypot(b.re, b.im) - Math.hypot(a.re, a.im))[0];
+    return Math.abs((z.omegaRadPerYr * 180) / Math.PI) * 3600;
+  })();
 
   return Object.freeze({
     time: Object.freeze({
@@ -1251,7 +1282,7 @@ export function assembleModel(C, F, laws = {}, secularSeriesArtifact = /** @type
       // one-family route (B): the hybrid's own equinox regression against
       // the λ̇-corrected sidereal year (~25,771 yr at J2000 ≈ IAU), inside
       // the banked tiers (|Δyear| ≤ ONE_FAMILY_WINDOW_YEARS); beyond, the
-      // unit's SECULAR mean H(t)/13 = the composed lunisolar period (the
+      // composed lunisolar period on the same J2000 anchor (S5; the
       // of-date wobble is unresolved there). The comb family's beat
       // sid/(sid − tropicalYearDirectDays) — the frozen era clock's
       // kinematic-day device — is no longer published here; it still
@@ -1263,24 +1294,23 @@ export function assembleModel(C, F, laws = {}, secularSeriesArtifact = /** @type
       // one-family inside the window, the tidal-chain mean beyond.
       tropicalYearSecondsAtYear: publishedTropicalYearSecondsAtYear,
     }),
-    // Plan 06 Phase 3 S3 — the lunisolar precession clock as a first-class
-    // surface: Earth's spin clock, ONE home for its published faces, spoken
-    // in PERIODS and their RATIOS — no unit, no integer. T_p(t) is the
-    // composed lunisolar period (calculation map 2.1 step 6; the internal
-    // identifier H is 13·T_p by definition and is not a face here); the
-    // composition's terms; the of-date beat (the published family, seamed
-    // at ±2 Myr); the apsidal period from the engine-D chain's secular
-    // tangent and the two ratios the owner found wandering by hand —
-    // T_aps/T_p (4.33 at J2000, 0.84 … 9.5 across ±26 kyr) and T_peri/T_p
-    // with T_peri = 1/(1/T_p + 1/T_aps) the perihelion-of-date period
-    // (0.812 at J2000). That these read close to 13/3 and 13/16 today is
-    // the J2000 reading, stated nowhere as structure.
+    // Plan 06 Phase 3 S3 → S5 — the lunisolar precession clock as a
+    // first-class surface: Earth's spin clock, ONE home for its published
+    // faces, spoken in PERIODS and their RATIOS — no unit, no integer.
+    // ONE J2000 reading (S5): every face reads 25,771.4 yr at J2000 — the
+    // of-date period (the published family, seamed at ±2 Myr) and the
+    // composed mean T_p(t) (the deep-time clock, anchored on the same
+    // derived value). The composition's terms; the apsidal period from the
+    // engine-D chain's secular tangent; the two ratios the owner found
+    // wandering by hand — T_aps/T_p (4.33 at J2000, 0.84 … 9.9 across
+    // ±26 kyr) and T_peri/T_p with T_peri = 1/(1/T_p + 1/T_aps) the
+    // perihelion-of-date period (0.812 at J2000) — on the of-date T_p.
     lunisolar: Object.freeze({
-      /** The mean lunisolar precession period T_p(t), years — the composed torque rate's period. @param {number} year @returns {number} */
-      meanPeriodYearsAtYear: (year) => { const h = deepLod.hAtAge(yearToTMa(year)); return (h === null ? H : h) / 13; },
-      /** The composed rate ψ̇(t) = [ω/ω₀]·p₀·[f_S + (1 − f_S)(a₀/a_M)³], ″/yr. @param {number} year @returns {number} */
-      meanRateArcsecPerYrAtYear: (year) => { const r = deepLod.lunisolarPrecessionRateArcsecPerYrAtAge(yearToTMa(year)); return r === null ? 1296000 / (H / 13) : r; },
-      /** The of-date period — the one-family beat inside ±2 Myr, the mean beyond. @param {number} year @returns {number} */
+      /** The mean lunisolar precession period T_p(t), years — the composed torque rate's period on the derived J2000 anchor (25,771.4 at J2000). @param {number} year @returns {number} */
+      meanPeriodYearsAtYear: (year) => { const p = deepLod.lunisolarPrecessionPeriodYearsAtAge(yearToTMa(year)); return p === null ? certifiedAxialPrecessionJ2000Years() : p; },
+      /** The composed rate ψ̇(t) = [ω/ω₀]·p₀·[f_S + (1 − f_S)(a₀/a_M)³], ″/yr (p₀ = 50.2883, derived). @param {number} year @returns {number} */
+      meanRateArcsecPerYrAtYear: (year) => { const r = deepLod.lunisolarPrecessionRateArcsecPerYrAtAge(yearToTMa(year)); return r === null ? 1296000 / certifiedAxialPrecessionJ2000Years() : r; },
+      /** The of-date period — the one-family beat inside ±2 Myr, the composed mean beyond. @param {number} year @returns {number} */
       ofDatePeriodYearsAtYear: publishedAxialPrecessionYearsAtYear,
       /** f_S, the solar fraction of the J2000 precession torque. */
       solarShareJ2000: EPOCH_PARAMS.precessionSolarShareJ2000,
@@ -1288,28 +1318,35 @@ export function assembleModel(C, F, laws = {}, secularSeriesArtifact = /** @type
       lunarTorqueFactorAtYear: (year) => { const f = deepLod.lunarTorqueFactorAtAge(yearToTMa(year)); return f === null ? 1 : f; },
       /** f_S + (1 − f_S)(a₀/a_M)³ — the torque term the unit divides H_era by. @param {number} year @returns {number} */
       torqueTermAtYear: (year) => { const t = deepLod.precessionTorqueTermAtAge(yearToTMa(year)); return t === null ? 1 : t; },
-      /** The hybrid's precession constant α = p₀ / cos ε₀, ″/yr (ε₀ the J2000 obliquity input). */
-      torqueConstantJ2000ArcsecPerYr: (1296000 / (H / 13)) / Math.cos((C.earthOrbital.obliquityJ2000_deg * Math.PI) / 180),
+      /** The hybrid's precession constant α = p₀ / cos ε₀, ″/yr (p₀ the derived J2000 rate, ε₀ the J2000 obliquity input; 54.81). */
+      torqueConstantJ2000ArcsecPerYr: (1296000 / certifiedAxialPrecessionJ2000Years()) / Math.cos((C.earthOrbital.obliquityJ2000_deg * Math.PI) / 180),
+      /** |s₃|, the dominant nodal mode of Earth's orbit — the largest-amplitude ζ mode of the banked deep secular modes, ″/yr (18.85; the obliquity beat's partner). ONE home for the registry, the browser and the API. */
+      nodalModeS3ArcsecPerYr: s3ArcsecPerYr,
+      /** The ecliptic (nodal) precession period 1,296,000/|s₃|, years (68,751) — an orbital quantity, fixed at every epoch at the two-body level. */
+      nodalPeriodYears: 1296000 / s3ArcsecPerYr,
+      /** The obliquity beat 2π/(ψ̇(t) − |s₃|), years — the SHIPPED deep-time obliquity period (falsification leg 1; 41,224 at J2000, on the composed rate). @param {number} year @returns {number} */
+      obliquityBeatYearsAtYear: (year) => {
+        const r = deepLod.lunisolarPrecessionRateArcsecPerYrAtAge(yearToTMa(year));
+        const psiDot = r === null ? 1296000 / certifiedAxialPrecessionJ2000Years() : r;
+        return 1296000 / (psiDot - s3ArcsecPerYr);
+      },
       /** The apsidal (perihelion vs the stars) period from the engine-D chain's secular tangent, years — inside the published window only (the tangent is an extrapolation beyond the banked series: it turns negative at −5 Myr); null beyond. @param {number} year @returns {number|null} */
       apsidalPeriodYearsAtYear: (year) => (Math.abs(year - 2000) <= ONE_FAMILY_WINDOW_YEARS ? 360 / computeApsidalSecularDegPerYr(year, kcChainsM.earth, kcChainsM) : null),
-      /** T_aps(t) / T_p(t) — the apsidal period in mean precession periods (4.33 at J2000, a fitted reading; 0.84 … 9.5 across ±26 kyr, measured); null beyond the published window. @param {number} year @returns {number|null} */
+      /** T_aps(t) / T_p(t) — the apsidal period in of-date precession periods (4.33 at J2000, a reading; 0.84 … 9.9 across ±26 kyr, measured); null beyond the published window. @param {number} year @returns {number|null} */
       apsidalPerPrecessionAtYear: (year) => {
         if (Math.abs(year - 2000) > ONE_FAMILY_WINDOW_YEARS) return null;
-        const h = deepLod.hAtAge(yearToTMa(year));
-        return (360 / computeApsidalSecularDegPerYr(year, kcChainsM.earth, kcChainsM)) / ((h === null ? H : h) / 13);
+        return (360 / computeApsidalSecularDegPerYr(year, kcChainsM.earth, kcChainsM)) / publishedAxialPrecessionYearsAtYear(year);
       },
-      /** T_peri(t) = 1/(1/T_p + 1/T_aps) — the perihelion-of-date period (equinox precession + inertial perihelion motion, frame arithmetic at every epoch), years; null beyond the published window. @param {number} year @returns {number|null} */
+      /** T_peri(t) = 1/(1/T_p + 1/T_aps) — the perihelion-of-date period (equinox precession + inertial perihelion motion, frame arithmetic at every epoch), years, on the of-date T_p; null beyond the published window. @param {number} year @returns {number|null} */
       periOfDatePeriodYearsAtYear: (year) => {
         if (Math.abs(year - 2000) > ONE_FAMILY_WINDOW_YEARS) return null;
-        const h = deepLod.hAtAge(yearToTMa(year));
-        const tp = (h === null ? H : h) / 13, taps = 360 / computeApsidalSecularDegPerYr(year, kcChainsM.earth, kcChainsM);
+        const tp = publishedAxialPrecessionYearsAtYear(year), taps = 360 / computeApsidalSecularDegPerYr(year, kcChainsM.earth, kcChainsM);
         return 1 / (1 / tp + 1 / taps);
       },
       /** T_peri(t) / T_p(t) (0.812 at J2000 — the J2000 reading); null beyond the published window. @param {number} year @returns {number|null} */
       periOfDatePerPrecessionAtYear: (year) => {
         if (Math.abs(year - 2000) > ONE_FAMILY_WINDOW_YEARS) return null;
-        const h = deepLod.hAtAge(yearToTMa(year));
-        const tp = (h === null ? H : h) / 13, taps = 360 / computeApsidalSecularDegPerYr(year, kcChainsM.earth, kcChainsM);
+        const tp = publishedAxialPrecessionYearsAtYear(year), taps = 360 / computeApsidalSecularDegPerYr(year, kcChainsM.earth, kcChainsM);
         return (1 / (1 / tp + 1 / taps)) / tp;
       },
       /** The published of-date window, years from 2000. */

package/types/deltat/deep-time.d.cts

@@ -62,6 +62,15 @@ export type DeepTimeLodDeps = {
      * - gated analytic swing rate
      */
     swingLodRateAt: (year: number) => number;
+    /**
+     * - the model's DERIVED
+     * J2000 axial precession period: the certified year-length laws' beat at
+     * 2000, sid/(sid − trop) ≈ 25,771.4 yr (plan 06 S5 — one J2000 precession
+     * reading; the same anchor the hybrid obliquity self-anchors on). Read
+     * LAZILY on the first composed-rate use, never at construction: every
+     * runtime's year laws read THIS factory's bases.
+     */
+    precessionPeriodJ2000YearsFn: () => number;
     /**
      * - OPTIONAL time-dependent
      * Earth-Moon angular momentum (the Driver-1½ solar channels,
@@ -99,6 +108,12 @@ export type DeepTimeLodDeps = {
  * @property {(year: number) => number} cycleLodSumAt - gated δLOD sum (incl. swing)
  * @property {(year: number) => number} swingLodAt - gated swing δLOD alone
  * @property {(year: number) => number} swingLodRateAt - gated analytic swing rate
+ * @property {() => number} precessionPeriodJ2000YearsFn - the model's DERIVED
+ *   J2000 axial precession period: the certified year-length laws' beat at
+ *   2000, sid/(sid − trop) ≈ 25,771.4 yr (plan 06 S5 — one J2000 precession
+ *   reading; the same anchor the hybrid obliquity self-anchors on). Read
+ *   LAZILY on the first composed-rate use, never at construction: every
+ *   runtime's year laws read THIS factory's bases.
  * @property {(tMa: number) => number} [lEmAtAgeKgm2S] - OPTIONAL time-dependent
  *   Earth-Moon angular momentum (the Driver-1½ solar channels,
  *   recession-history.cjs). Absent → the J2000 constant, which the budget
@@ -111,6 +126,7 @@ export function createDeepTimeLod(deps: DeepTimeLodDeps): {
     lodHoursAtAge: (t_Ma: number) => number | null;
     hAtAge: (t_Ma: number) => number | null;
     lunisolarPrecessionRateArcsecPerYrAtAge: (t_Ma: number) => number | null;
+    lunisolarPrecessionPeriodYearsAtAge: (t_Ma: number) => number | null;
     lunarTorqueFactorAtAge: (tMa: number) => (number | null);
     precessionTorqueTermAtAge: (tMa: number) => (number | null);
     siderealYearSecondsAtAge: (t_Ma: number) => number;

package/types/earth/precession-composed.d.cts

@@ -17,15 +17,18 @@ export function computeSolarTorqueShare(c: {
 }): number;
 /**
  * @param {{
- *   p0ArcsecPerYr: number,
+ *   p0ArcsecPerYr: number | (() => number),
  *   solarShare: number,
  *   lodSecondsAtAge: (tMa: number) => (number | null),
  *   lodJ2000Seconds: number,
  *   moonDistanceMetresAtAge: (tMa: number) => (number | null),
  *   moonDistanceJ2000Metres: number,
  *   yearToTMa: (year: number) => number,
- * }} deps - p0 = 1,296,000/(H/13) (the model's J2000 rate); lodJ2000Seconds
- *   the SAME day basis lodSecondsAtAge(0) returns.
+ * }} deps - p0 = the model's derived J2000 rate, 1,296,000/axial0 (S5; a
+ *   number or a LAZY getter — the callers' certified year laws read the
+ *   deep-time factory this is built inside, so the anchor is resolved on
+ *   the first RATE use, never at construction; the torque term needs no
+ *   anchor); lodJ2000Seconds the SAME day basis lodSecondsAtAge(0) returns.
  * @returns {{
  *   composedRateArcsecPerYrAtAge: (tMa: number) => (number | null),
  *   composedPeriodYearsAtAge: (tMa: number) => (number | null),
@@ -39,7 +42,7 @@ export function computeSolarTorqueShare(c: {
  * }}
  */
 export function createComposedPrecession(deps: {
-    p0ArcsecPerYr: number;
+    p0ArcsecPerYr: number | (() => number);
     solarShare: number;
     lodSecondsAtAge: (tMa: number) => (number | null);
     lodJ2000Seconds: number;

package/types/model.d.ts

@@ -37,11 +37,11 @@ export function assembleModel(C: Readonly<Record<string, any>>, F: Readonly<Reco
         tropicalYearSecondsAtYear: (year: number) => number;
     }>;
     lunisolar: Readonly<{
-        /** The mean lunisolar precession period T_p(t), years — the composed torque rate's period. @param {number} year @returns {number} */
+        /** The mean lunisolar precession period T_p(t), years — the composed torque rate's period on the derived J2000 anchor (25,771.4 at J2000). @param {number} year @returns {number} */
         meanPeriodYearsAtYear: (year: number) => number;
-        /** The composed rate ψ̇(t) = [ω/ω₀]·p₀·[f_S + (1 − f_S)(a₀/a_M)³], ″/yr. @param {number} year @returns {number} */
+        /** The composed rate ψ̇(t) = [ω/ω₀]·p₀·[f_S + (1 − f_S)(a₀/a_M)³], ″/yr (p₀ = 50.2883, derived). @param {number} year @returns {number} */
         meanRateArcsecPerYrAtYear: (year: number) => number;
-        /** The of-date period — the one-family beat inside ±2 Myr, the mean beyond. @param {number} year @returns {number} */
+        /** The of-date period — the one-family beat inside ±2 Myr, the composed mean beyond. @param {number} year @returns {number} */
         ofDatePeriodYearsAtYear: (year: number) => number;
         /** f_S, the solar fraction of the J2000 precession torque. */
         solarShareJ2000: number;
@@ -49,13 +49,19 @@ export function assembleModel(C: Readonly<Record<string, any>>, F: Readonly<Reco
         lunarTorqueFactorAtYear: (year: number) => number;
         /** f_S + (1 − f_S)(a₀/a_M)³ — the torque term the unit divides H_era by. @param {number} year @returns {number} */
         torqueTermAtYear: (year: number) => number;
-        /** The hybrid's precession constant α = p₀ / cos ε₀, ″/yr (ε₀ the J2000 obliquity input). */
+        /** The hybrid's precession constant α = p₀ / cos ε₀, ″/yr (p₀ the derived J2000 rate, ε₀ the J2000 obliquity input; 54.81). */
         torqueConstantJ2000ArcsecPerYr: number;
+        /** |s₃|, the dominant nodal mode of Earth's orbit — the largest-amplitude ζ mode of the banked deep secular modes, ″/yr (18.85; the obliquity beat's partner). ONE home for the registry, the browser and the API. */
+        nodalModeS3ArcsecPerYr: number;
+        /** The ecliptic (nodal) precession period 1,296,000/|s₃|, years (68,751) — an orbital quantity, fixed at every epoch at the two-body level. */
+        nodalPeriodYears: number;
+        /** The obliquity beat 2π/(ψ̇(t) − |s₃|), years — the SHIPPED deep-time obliquity period (falsification leg 1; 41,224 at J2000, on the composed rate). @param {number} year @returns {number} */
+        obliquityBeatYearsAtYear: (year: number) => number;
         /** The apsidal (perihelion vs the stars) period from the engine-D chain's secular tangent, years — inside the published window only (the tangent is an extrapolation beyond the banked series: it turns negative at −5 Myr); null beyond. @param {number} year @returns {number|null} */
         apsidalPeriodYearsAtYear: (year: number) => number | null;
-        /** T_aps(t) / T_p(t) — the apsidal period in mean precession periods (4.33 at J2000, a fitted reading; 0.84 … 9.5 across ±26 kyr, measured); null beyond the published window. @param {number} year @returns {number|null} */
+        /** T_aps(t) / T_p(t) — the apsidal period in of-date precession periods (4.33 at J2000, a reading; 0.84 … 9.9 across ±26 kyr, measured); null beyond the published window. @param {number} year @returns {number|null} */
         apsidalPerPrecessionAtYear: (year: number) => number | null;
-        /** T_peri(t) = 1/(1/T_p + 1/T_aps) — the perihelion-of-date period (equinox precession + inertial perihelion motion, frame arithmetic at every epoch), years; null beyond the published window. @param {number} year @returns {number|null} */
+        /** T_peri(t) = 1/(1/T_p + 1/T_aps) — the perihelion-of-date period (equinox precession + inertial perihelion motion, frame arithmetic at every epoch), years, on the of-date T_p; null beyond the published window. @param {number} year @returns {number|null} */
         periOfDatePeriodYearsAtYear: (year: number) => number | null;
         /** T_peri(t) / T_p(t) (0.812 at J2000 — the J2000 reading); null beyond the published window. @param {number} year @returns {number|null} */
         periOfDatePerPrecessionAtYear: (year: number) => number | null;