diff --git a/book/The-Incompleteness-of-Observation-FULL.md b/book/The-Incompleteness-of-Observation-FULL.md
index 1e3e46f7c..dc40f22f2 100644
--- a/book/The-Incompleteness-of-Observation-FULL.md
+++ b/book/The-Incompleteness-of-Observation-FULL.md
@@ -816,7 +816,7 @@ The substratum construction provides a structural account of why the Standard Mo
 
 This is what makes the framework's scoped structural realism non-trivial. The invariants of the **reconstructed local lattice/gauge residue** are uniquely determined — under the reconstruction theorem's stated conditions — by observation plus structural assumptions. Individual representatives of that local residue related by the generators of $\mathcal{G}_{\text{sub}}$ are gauge expressions of the same reconstructed physical content, like coordinate choices on a manifold or specific representatives of a homotopy class. Bell-composite preparation data lie outside this uniqueness statement: H-Bell conditions compatibility/existence, and no result here selects a unique Bell-inclusive full representative.
 
-The reconstruction theorem identifies the **local lattice/gauge residue**; the gauge group identifies what within that scoped reconstruction has no observable content. With both in place, observed physics fixes that local residue under the theorem's stated conditions. H-Bell separately conditions Bell-inclusive existence, and Bell-inclusive uniqueness is not established. The next section turns to the synthesis: how quantum mechanics, general relativity, and the arrow of time relate to the substratum as three projections of the same construction.
+The reconstruction theorem identifies the **local lattice/gauge residue**; the gauge group identifies what within that scoped reconstruction has no observable content. With both in place, observed physics fixes that local residue under the theorem's stated conditions. H-Bell separately conditions Bell-inclusive existence, and Bell-inclusive uniqueness is not established. The next section turns to the synthesis: how quantum mechanics and general relativity relate to the substratum as projections of the same construction, and what a third projection establishes about the arrow of time.
 
 ***
 
@@ -824,7 +824,7 @@ The reconstruction theorem identifies the **local lattice/gauge residue**; the g
 
 The reconstruction theorem and the substratum gauge group have established what $(S, \varphi)$ is and what equivalences leave its observables unchanged. The question taken up in this section is what relationship $(S, \varphi)$ bears to the framework's three large emergent structures: quantum mechanics, general relativity, and the arrow of time.
 
-The framework's claim is that quantum mechanics, general relativity, and the arrow of time are not three independent theories. Conditional on H-Bell at the Bell/local-lattice interface, they are three projections of the same finite deterministic construction. Each projection applies the same trace-out machinery to the same $(S, \varphi)$ along a different axis, and each produces a distinct emergent structure as the observer sees the substratum from inside.
+The framework's claim is that quantum mechanics and general relativity are not two independent theories. Conditional on H-Bell at the Bell/local-lattice interface, they are projections of the same finite deterministic construction, and a third projection bears on the arrow of time. Each projection applies the same trace-out machinery to the same $(S, \varphi)$ along a different axis, and each produces a distinct emergent structure as the observer sees the substratum from inside.
 
 **The visible-sector projection: quantum mechanics.** Chapter 1 developed this projection in detail. At finite observable horizon the reduced stochastic law, a finite reversible realization, and a fixed-basis unitary/Born representation are exactly equivalent: $S\iff D\iff Q_{\mathrm{fb}}$. C4 readback then identifies the memory-bearing sector by forcing hidden predictive memory. In the spatial construction the coupling graph supplies exact causal cones for finite-range substrata. Bell is separate from the full-QM lift: a measurement-independent response-complete deterministic completion with parameter independence and a $\pm1$ outcome in every run would obey $|S_{\rm CHSH}|\le2$, so the framework's Bell-violating branch is ontically parameter-dependent with operational no-signaling required. Two independent qualifications remain beyond the finite-law equivalence: the simultaneous coherent instrument/composite lift ([Main §3.4]) and H-Bell, the compatibility of the local lattice sector with the ontically parameter-dependent Bell completion. Applied to the local lattice sector constructed in this chapter — a $d = 3$ cubic lattice with $K = 6$ internal components and the wave equation as its local propagating dynamics — the visible-sector projection produces the Standard Model: the gauge group $\mathrm{SU}(3) \times \mathrm{SU}(2) \times \mathrm{U}(1)$ from the commutant of the cubic-symmetric coupling matrix, three chiral fermion generations, one composite Higgs, and anomaly-free hypercharges. Chapters 5 and 6 develop the Standard Model derivation; Chapter 8 develops the specific JUNO neutrino measurement that the framework's parameter-free PMNS-angle value matches at $0.07\sigma$ (a retrodiction). The local nearest-neighbor sector cannot itself supply Bell parameter dependence for wings with one-step own-wing setting interventions outside each other's graph cone under measurement independence; identifying it with one Bell-violating total completion is the open H-Bell compatibility hypothesis.
 
@@ -832,13 +832,13 @@ The framework's claim is that quantum mechanics, general relativity, and the arr
 
 Applied to the cosmological horizon as a causal partition, the boundary-thermodynamic projection produces general relativity with quantitative predictions. The emergent action scale $\hbar = c^3 \epsilon^2 / (4G\sigma)$, with $\epsilon = 2\sqrt{\sigma}\,\ell_P$, follows from thermal self-consistency at the boundary layer under the infrared detailed-balance condition H-slope. The boundary-mode count reproduces the Bekenstein-Hawking entropy $S = A/(4 \ell_P^2)$, the $1/4$ coefficient being the ratio $2\pi/8\pi$ that the calibration matches and conditional on the horizon conditions of Chapter 7; GW250114 (September 2025) confirms the classical area theorem, which the framework preserves — the $1/4$ coefficient itself has no direct empirical test. The cosmological constant problem receives a conditional treatment: the quantum vacuum energy and the effective $\Lambda$ are properties of logically distinct levels of description rather than commensurable quantities whose $10^{122}$ ratio requires cancellation; the dissolution is conditional on the covariant coupling step (Chapter 7). The structured component of the boundary entropy produces dark matter phenomenology with MOND acceleration scale $a_0 = cH/6 \approx 1.1 \times 10^{-10}$ m/s² (the dimensional factor imported from Verlinde's emergent-gravity analysis). Chapter 7 develops each of these.
 
-**The horizon-clock projection: the arrow of time.** A third projection applies the trace-out machinery to the horizon-growth parameter along an observer worldline. The substratum is time-symmetric: $\varphi$ and $\varphi^{-1}$ are equally valid as forward-time maps, and there is no preferred direction at the substratum level. Observers, however, are not symmetric under this exchange. Condition C2 — persistence, in its slow-bath realization — picks out one direction along the observer's worldline as the direction in which the hidden sector's correlations with the visible sector's past are preserved. The structural observer-selection theorem developed in Chapter 1's appendix material — under the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), and in the coarse-grained description generated by iterating the one-step effective channel ([Main §4.6]) — restricts observers satisfying C1–C4, with C2 read in its timescale form (record persistence and readback there being suppressed, not excluded), to the non-equilibrium phase of $(S, \varphi)$; in the $\Lambda$-dominated cosmological phase, this phase is the one in which the horizon entropy $S_{\text{dS}}(t) = A(t)/(4\ell_P^2)$ is monotonically increasing.
+**The horizon-clock projection: the arrow of time.** A third projection applies the trace-out machinery to the horizon-growth parameter along an observer worldline. The substratum is time-symmetric: $\varphi$ and $\varphi^{-1}$ are equally valid as forward-time maps, and there is no preferred direction at the substratum level. Condition C2 — persistence, in its slow-bath realization — is a retention condition: it requires the hidden sector's correlations with the visible sector's past to persist to readback, and the coarse-grained mixing time it constrains is the same in the two directions; the direction in which records are written and read is fixed by a preparation and the observer's orientation convention. The structural observer-selection theorem developed in Chapter 1's appendix material — under the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), and in the coarse-grained description generated by iterating the one-step effective channel ([Main §4.6]) — restricts observers satisfying C1–C4, with C2 read in its timescale form (record persistence and readback there being suppressed, not excluded), to the non-equilibrium phase of $(S, \varphi)$; the restriction is time-symmetric and selects where observers can exist, not a direction. On an expanding cosmological background satisfying the null energy condition — inputs, not derived — the area $A(t)$ of the apparent horizon, and the horizon entropy $S_{\text{dS}}(t) = A(t)/(4\ell_P^2)$ assigned to it at the de Sitter calibration, is non-decreasing, strictly increasing while $\rho + p/c^2 > 0$ and constant in exact de Sitter. That the selected non-equilibrium phase is ordered by this growth is not established.
 
-The cosmological horizon entropy provides a structural clock. Every observer in the accelerating phase agrees that $S_{\text{dS}}$ increases — the monotonicity is a feature of the causal partition rather than of any local frame — and other arrows of time align with this structural clock through a cascade. The thermodynamic arrow descends from Jacobson's coupling between boundary entropy and matter entropy. The memory arrow descends from C2 — persistence, realized by the slow bath — which preserves correlations forward along the clock rather than backward. The measurement arrow — the observer records past outcomes and prepares future states — descends from the memory arrow. The chain grounds in one structural fact (horizon growth in the accelerating phase) and one structural constraint (C1–C4 observer selection, conditional on the effective-mixing hypothesis of [Main §4.6] and stated for the coarse-grained description generated by iterating the one-step effective channel). No independent past-hypothesis axiom is required to select where observers exist, and Boltzmann brains are excluded structurally rather than anthropically.
+The apparent-horizon area provides a structural clock on that background. Every comoving observer of an expanding homogeneous general-relativistic background satisfying the null energy condition agrees that it does not decrease — standard background kinematics, not a prediction derived from the substratum — and the criterion is expansion with the null energy condition, not acceleration: in exact de Sitter the area is constant. Whether the other arrows of time align with this clock is not established. Jacobson's coupling relates the boundary energy density to the expansion rate, not matter entropy to horizon area, so the direction of the thermodynamic arrow rests on a low-entropy or product boundary condition, and the generalized second law is open. C2 — persistence, realized by the slow bath — holds the same way in both directions; the direction of the memory arrow comes from a preparation and the observer's orientation convention. The measurement arrow — the observer records past outcomes and prepares future states — inherits the memory arrow's direction. The chain uses one input (horizon growth on the expanding branch) and one structural constraint (C1–C4 observer selection, conditional on the effective-mixing hypothesis of [Main §4.6] and stated for the coarse-grained description generated by iterating the one-step effective channel), which fixes where observers can exist, not a direction. No independent past-hypothesis axiom is required to select where observers exist, and Boltzmann brains are excluded structurally rather than anthropically.
 
-The framework's content here is precise: it derives a monotonic parameter along observer worldlines and four aligned arrows of time. What it does not derive is the convention of which direction along the horizon-clock parameter to call "future." The two directions are related by the layer swap at the substrate level ($\varphi \leftrightarrow \varphi^{-1}$), and the observer's choice of convention is residual freedom. The framework derives the existence of a monotonic parameter; the labeling is an observer-imposed convention. There is one further qualification worth stating: a monotonic parameter is monotonic with respect to an *ordering* of the worldline's moments, and the framework's foundational treatment axiomatizes that there is more than one moment while leaving the ordering of moments as a separate, open question. The derivation here is therefore conditional on that ordering — it establishes that ordered moments yield a directed parameter, not that directedness arises from the substratum with no assumption about ordering at all. CP violation in CKM and kaon/B systems is a flavor-basis feature of the specific bijection $\varphi$ rather than a consequence of the time-arrow cascade — the framework treats flavor-sector CP phases as solution-specific inputs that do not affect the cascade's structural content.
+The framework's content here is precise: on the expanding branch it establishes a non-decreasing horizon parameter along observer worldlines, and it does not establish that the four arrows of time align. Nor does it derive the convention of which direction along the horizon-clock parameter to call "future." The two directions are related by the layer swap at the substrate level ($\varphi \leftrightarrow \varphi^{-1}$), and the observer's choice of convention is residual freedom. The parameter is conditional on the expanding branch; the labeling is an observer-imposed convention. There is one further qualification worth stating: a monotonic parameter is monotonic with respect to an *ordering* of the worldline's moments, and the framework's foundational treatment axiomatizes that there is more than one moment while leaving the ordering of moments as a separate, open question. The result here is therefore conditional on that ordering as well — it establishes that ordered moments on the expanding branch yield a non-decreasing parameter, not that directedness arises from the substratum with no assumption about ordering at all. CP violation in CKM and kaon/B systems is a flavor-basis feature of the specific bijection $\varphi$ rather than a consequence of the time-arrow cascade — the framework treats flavor-sector CP phases as solution-specific inputs that do not affect the cascade's structural content.
 
-**The common structure.** The three projections share three features. Under H-Bell they apply to the same total substratum $(S, \varphi)$; without H-Bell, the local SM/GR lattice sector and the Bell-violating completion have not been shown to be one object. They use the same trace-out machinery — marginalization over the hidden sector under conditions C1–C4. They produce derived rather than fundamental structures: nothing in quantum mechanics, general relativity, or the arrow of time is a feature of $(S, \varphi)$ in itself; each is a feature of how the embedded observer sees the substratum from inside. The differences among the three projections lie entirely in which structure of the embedded observer's description they extract. The visible-sector projection works on the bulk dynamics and produces the reduced finite law — universally admitting the quantum representation — with the Standard Model as its specific content. The boundary-thermodynamic projection works on the classical thermal data at the partition boundary and produces general relativity with its gravitational and cosmological predictions. The horizon-clock projection works on the horizon-growth parameter along observer worldlines and produces the cascade of emergent arrows.
+**The common structure.** The three projections share three features. Under H-Bell they apply to the same total substratum $(S, \varphi)$; without H-Bell, the local SM/GR lattice sector and the Bell-violating completion have not been shown to be one object. They use the same trace-out machinery — marginalization over the hidden sector under conditions C1–C4. They produce derived rather than fundamental structures: nothing in quantum mechanics, general relativity, or the arrow of time is a feature of $(S, \varphi)$ in itself; each is a feature of how the embedded observer sees the substratum from inside. The differences among the three projections lie entirely in which structure of the embedded observer's description they extract. The visible-sector projection works on the bulk dynamics and produces the reduced finite law — universally admitting the quantum representation — with the Standard Model as its specific content. The boundary-thermodynamic projection works on the classical thermal data at the partition boundary and produces general relativity with its gravitational and cosmological predictions. The horizon-clock projection works on the horizon-growth parameter along observer worldlines together with the observer-selection rule; the alignment of the arrows with that parameter is open.
 
 The shared structure is not metaphor or analogy. The same local lattice sector that gives the Standard Model gauge group from the commutant of the coupling matrix gives the Bekenstein-Hawking entropy as the boundary mode count. H-Bell is required only for identifying that local sector with the Bell-violating total completion. The conditions C1–C4 that characterize the memory-bearing sector — whose conditional memory readback makes measurable — for the visible-sector projection produce the persistence-based observer selection for the horizon-clock projection. The substratum gauge group $\mathcal{G}_{\text{sub}}$ controls the gauge equivalences of the **locally reconstructed residue** across the three projections; it does not classify all preparation-indexed Bell-composite rules. The framework's claim is that the three projections are co-derived from a single construction, not three separate applications of similar machinery.
 
@@ -860,7 +860,7 @@ The framework's synthesis claim is structural and architectural, not predictive
 
 ## 2.8 What follows
 
-The reconstruction theorem and the substratum gauge group establish the substratum side of the framework. The three projections establish its relationship to the framework's three large emergent structures. With both in place, the technical core of Part I is complete; the chapters that follow develop its consequences in three directions.
+The reconstruction theorem and the substratum gauge group establish the substratum side of the framework. The three projections establish its relationship to quantum mechanics and general relativity, and what it establishes about the arrow of time. With both in place, the technical core of Part I is complete; the chapters that follow develop its consequences in three directions.
 
 Chapter 3 develops the framework's philosophical positioning. The substratum's ontological status — what it means to commit to the equivalence class $[(S, \varphi)] / \mathcal{G}_{\text{sub}}$ as the framework's metaphysical object — is articulated in its full hierarchical form, with realism operating at multiple levels of the observation hierarchy and the gauge hierarchy. The framework's relation to Bell, Kochen-Specker and PBR-type no-go theorems is developed through separate checks of their local-response, operational-algebra and preparation premises, with the stated unresolved obligations retained. The framework's intellectual position is located against the contemporary observer-dependence programs (QBism, relational quantum mechanics, the recent CLPW and Maldacena work on observer-incorporated entropy) and against the older philosophical lineage (Cusa's docta ignorantia, Spinoza's attribute theory, Worrall's epistemic structural realism). The relationship to Hofstadter's strange-loop framework is developed in detail.
 
@@ -929,7 +929,7 @@ The observer-admission reading at Level B sharpens this in two ways. First, it i
 
 A concrete example clarifies the structural distinction. Consider two finite cellular automata with the same configuration space size but different transition rules. The first has a Glauber dynamics on an Ising lattice at high temperature: fast mixing, short correlation times, no slow modes. The second has a topological phase-protected dynamics: slow modes from non-local order parameters, long correlation times in those modes, sufficient capacity in the topological sector. Both are well-defined deterministic systems on the same configuration space. In the coarse-grained description generated by its iterated one-step channel, with $\tau_S$ the observer's visible step, the first fails C2 in its timescale form, $\tau_S \ll \tau_B$, since it mixes fast relative to that step: writes into its hidden sector are scrambled before readback up to a suppressed residue, the absence of slow modes closing the thermal route to persistence; the second satisfies C1, C2, C3, C4 with appropriate partition choice. With C2 read in that timescale form, the first is observer-lacking; the second is observer-admitting. The distinction is structural — a property of the dynamics, not of the parameter values within a given dynamics — and applies prior to any specific universality-class commitment.
 
-This structural distinction has implications for several conceptual problems. The Boltzmann brain problem — the worry that a sufficiently large statistical-mechanical system at equilibrium would produce observers as random fluctuations more often than through evolutionary processes — is dissolved at Level B for C2 read in its timescale form. Under the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), and in the coarse-grained description generated by iterating the one-step effective channel ([Main §4.6]), equilibrium states are not observer-admitting in the sense of C1–C4 with C2 read in its timescale form, because that form, $\tau_S \ll \tau_B$, fails at equilibrium — its slow-bath route gone, no slow modes under that condition — and a bath that thermalizes between steps erodes the stored record before readback without establishing that none persists, suppressing the readback gap C4 requires exponentially in $\tau_S/\tau_B$ at every order $k \geq 2$ ([Main §2.3]). Boltzmann brains are not generic outcomes of long-time evolution that just happen to be rare; they are structurally excluded from the observer-admitting subset by C2 in that timescale form. The framework's account of the arrow of time, developed in Chapter 2 §2.6, follows from this exclusion: observers are restricted to the non-equilibrium phase of the substratum's dynamics, and the structural arrow of horizon growth provides the clock the observer's other arrows align with.
+This structural distinction has implications for several conceptual problems. The Boltzmann brain problem — the worry that a sufficiently large statistical-mechanical system at equilibrium would produce observers as random fluctuations more often than through evolutionary processes — is dissolved at Level B for C2 read in its timescale form. Under the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), and in the coarse-grained description generated by iterating the one-step effective channel ([Main §4.6]), equilibrium states are not observer-admitting in the sense of C1–C4 with C2 read in its timescale form, because that form, $\tau_S \ll \tau_B$, fails at equilibrium — its slow-bath route gone, no slow modes under that condition — and a bath that thermalizes between steps erodes the stored record before readback without establishing that none persists, suppressing the readback gap C4 requires exponentially in $\tau_S/\tau_B$ at every order $k \geq 2$ ([Main §2.3]). Boltzmann brains are not generic outcomes of long-time evolution that just happen to be rare; they are structurally excluded from the observer-admitting subset by C2 in that timescale form. The framework's treatment of the arrow of time, developed in Chapter 2 §2.6, starts from this exclusion: observers are restricted to the non-equilibrium phase of the substratum's dynamics. The exclusion is time-symmetric; on an expanding homogeneous general-relativistic background satisfying the null energy condition, the apparent-horizon area supplies a non-decreasing clock, and whether the observer's other arrows align with it is not established.
 
 The string-theory landscape problem looks different under the Level B reading as well. The conventional landscape picture posits roughly $10^{500}$ metastable vacua, each a different low-energy effective theory, with the vacuum-selection problem being to explain why our universe is in this particular vacuum. The Level B reading reframes this: most compactifications likely produce observer-lacking substrata at the relevant scales, because the slow-bath, sufficient-capacity, non-trivial-coupling conditions are non-generic. The SM-reproducing subset of the landscape is structurally smaller than the total landscape, because it is restricted to the observer-admitting subset. Whether this reframing fully resolves the landscape problem is an open technical question — it depends on whether observer-lacking compactifications can be characterized in string-theoretic terms, which requires substantive work in string-theoretic landscape physics — but the reframing is non-trivial.
 
@@ -1073,7 +1073,7 @@ Cusa is not the framework's source. The framework derives its claims from explic
 
 **Spinoza.** Spinoza's metaphysics is the deepest historical parallel to the framework's substratum-and-projections picture. The *Ethics* posits one substance — God or Nature — expressed through infinitely many attributes, each constituting the substance's essence under a different aspect. The two attributes humans can know are extension (the physical aspect) and thought (the mental aspect), but these are not two substances; they are the same substance expressed through different attributes, with no causal interaction between them because there is only one thing underneath. The doctrine that the order and connection of ideas is the same as the order and connection of things follows from this single-substance structure: the parallelism between mind and body is not an interaction but a consequence of their being the same underlying reality.
 
-The structural form of Spinoza's account — one substance, multiple incommensurable attributes, with the apparent multiplicity being a feature of how the substance is expressed rather than of the substance itself — maps directly onto the framework's three-projection structure. The substratum $(S, \varphi)$ is the single substance; quantum mechanics, general relativity, and the arrow of time are projections expressing the substratum's content under different aspects; the appearance of three distinct emergent theories is a feature of the projections rather than of the substratum. The framework's commitment to ontic structural realism on the **reconstructed local residue modulo** $\mathcal{G}_{\text{sub}}$ is Spinozistic in form: what is real is the gauge-invariant local structure, not any specific representative; the Bell-inclusive completion is not uniquely reconstructed.
+The structural form of Spinoza's account — one substance, multiple incommensurable attributes, with the apparent multiplicity being a feature of how the substance is expressed rather than of the substance itself — maps directly onto the framework's three-projection structure. The substratum $(S, \varphi)$ is the single substance; quantum mechanics and general relativity are projections expressing the substratum's content under different aspects, with a third bearing on the arrow of time; the appearance of three distinct emergent theories is a feature of the projections rather than of the substratum. The framework's commitment to ontic structural realism on the **reconstructed local residue modulo** $\mathcal{G}_{\text{sub}}$ is Spinozistic in form: what is real is the gauge-invariant local structure, not any specific representative; the Bell-inclusive completion is not uniquely reconstructed.
 
 The parallel is not complete. Spinoza's substance is unique (there is exactly one substance, by argument); the framework's substratum is a member of an equivalence class (there are many specific bijections, gauge-equivalent under $\mathcal{G}_{\text{sub}}$). Spinoza's attributes are infinite in number; the framework identifies three projections (quantum mechanics, general relativity, arrow of time) without claiming to enumerate all possible projections. The framework's claims are mathematical theorems rather than metaphysical doctrines. But the structural resemblance is closer than to any other position in the historical philosophical literature.
 
@@ -1146,7 +1146,7 @@ The benefit of the discipline is a clean audit trail. Every prediction the frame
 
 The cost of the discipline is that some natural extensions of the framework require derivational work that has not yet been done. The framework currently treats CP-violating phases in CKM and PMNS as solution-specific inputs rather than as derivations; the cosmological initial conditions are taken as given rather than derived; the specific values of fermion masses depend on one empirical input (the strange-quark mass) rather than being derived from the substratum alone. These are not failures of the framework but acknowledged scope limits. The methodological discipline distinguishes between "the framework predicts X" (X follows from the foundational premise plus the structural assumptions) and "the framework is consistent with X but does not predict it" (X is solution-specific or depends on inputs the framework has not yet derived).
 
-A useful diagnostic for the discipline is what the framework does *not* commit to. The framework does not commit to a specific interpretation of probability beyond standard Kolmogorov probability. The framework does not commit to a specific resolution of the measurement problem beyond the algorithmic-reading developed in Chapter 1. The framework does not commit to specific values of free parameters in the emergent quantum field theory beyond what the reconstruction theorem forces. The framework does not commit to a specific cosmological history beyond the structural arrow of horizon growth in the $\Lambda$-dominated phase. These non-commitments are part of the foundational economy: each additional commitment would expand the framework's claims, and the discipline is to make only those commitments that the foundational premise plus its derivations require.
+A useful diagnostic for the discipline is what the framework does *not* commit to. The framework does not commit to a specific interpretation of probability beyond standard Kolmogorov probability. The framework does not commit to a specific resolution of the measurement problem beyond the algorithmic-reading developed in Chapter 1. The framework does not commit to specific values of free parameters in the emergent quantum field theory beyond what the reconstruction theorem forces. The framework does not commit to a specific cosmological history beyond an expanding homogeneous general-relativistic background, taken as given, on which, with the null energy condition, the apparent-horizon area does not decrease. These non-commitments are part of the foundational economy: each additional commitment would expand the framework's claims, and the discipline is to make only those commitments that the foundational premise plus its derivations require.
 
 The single-foundational-commitment pattern distinguishes the framework from broader unification programs in a specific way. String theory commits to extended one-dimensional objects as the fundamental entities, plus supersymmetry, plus extra dimensions, plus a specific moduli structure. Loop quantum gravity commits to a graph-theoretic discrete geometry, plus a specific quantization procedure, plus a specific representation theory of the spatial diffeomorphism group. Causal sets commit to a Lorentz-invariant discrete causal structure, plus a specific dynamics for set growth. Each of these is a multi-commitment foundation, with each commitment adding to the framework's content. The framework here is a single-commitment foundation, with all content derived from the one premise plus its structural lemmas. Whether this is a methodological virtue or a methodological constraint depends on whether the derivations actually close — whether single-commitment foundations can produce the content that multi-commitment foundations produce by direct positing. The framework's claim is that they can; Parts II–IV develop this claim with the specific quantitative content.
 
@@ -5000,7 +5000,7 @@ The return itself is definite: the universe returns to its exact initial state a
 
 **Falsifiability.** The recurrence prediction is falsifiable only in the logical sense. For a substratum of given $|S|$ and $\varphi$, failure to recur within $g(|S|)$ steps would contradict a finite $S$ with bijection dynamics, and failure to recur within $|S| - 1$ steps would contradict only exact linearity (A5). But $|S|$ is not fixed by visible data: for any finite observation time $T$, a substratum with $|S| > T + 1$ is admissible, so no finite observation falsifies the framework as a whole, which would require an infinite observation time. An embedded observer's records are, moreover, part of $S$ and return with it. The recurrence is therefore a prediction at the structural level rather than at the empirical level: it follows necessarily from the framework's other commitments (finite substratum, bijection dynamics), and any framework consistent with those commitments must predict exact recurrence with a period at most $g(|S|)$ that divides $\mathrm{lcm}(1, \ldots, |S|)$, and below $|S|$ for the exactly linear rule.
 
-The conceptual content is significant. The framework predicts that the universe is *cyclic at the substratum level* — periodic with period $N$, the least common multiple of the bijection's cycle lengths. For a rule fixing the zero configuration on a nontrivial state space, no single cycle passes through every configuration. The exact linear rule of [SM §4.2] has this property and therefore at least two cycles. The cycle-ergodicity question that [SM §4.2] treats concerns sampling within an energy shell. Within a single cycle, the observer's experience is unidirectional (the arrow of time discussed in the next section), but at the substratum level the dynamics has no preferred direction. The framework's commitment to substratum cyclicity is one of the structural commitments that distinguishes it from frameworks treating the universe as either fundamentally non-cyclic (most cosmological models) or fundamentally infinite (most quantum gravity programs).
+The conceptual content is significant. The framework predicts that the universe is *cyclic at the substratum level* — periodic with period $N$, the least common multiple of the bijection's cycle lengths. For a rule fixing the zero configuration on a nontrivial state space, no single cycle passes through every configuration. The exact linear rule of [SM §4.2] has this property and therefore at least two cycles. The cycle-ergodicity question that [SM §4.2] treats concerns sampling within an energy shell. Within a single cycle no state function increases throughout; a direction of the observer's experience is fixed by boundary conditions on a stretch of the cycle (the arrow of time discussed in the next section), and at the substratum level the dynamics has no preferred direction. The framework's commitment to substratum cyclicity is one of the structural commitments that distinguishes it from frameworks treating the universe as either fundamentally non-cyclic (most cosmological models) or fundamentally infinite (most quantum gravity programs).
 
 **Implications for the framework's logical structure.** The recurrence theorem connects three structural features of the framework: substratum finiteness (the framework's commitment to $|S| < \infty$), bijection dynamics (the framework's commitment to $\varphi$ as a bijection), and P-indivisibility at the recurrence scale. The connection runs in three steps. Finiteness of $S$ guarantees exact recurrence ($\varphi^N = \mathrm{id}$). Recurrence together with C1 acting at the first step — which makes the one-step matrix a non-permutation under the uniform hidden prior, the cheapest non-permutation rooted map — gives returns of information from hidden to visible over the recurrence timescale (P-indivisibility across the recurrence cycle, not on accessible windows); the coupling is load-bearing rather than decorative, since for an uncoupled partition $\varphi = \varphi_V \times \varphi_H$ the visible sector evolves autonomously, its two-time family is divisible, and no distinguishability is ever restored — recurrence notwithstanding. With a full-support visible prior this indivisibility is equivalent to memory at some order below the recurrence time; the memory-bearing sector is the one in which C4 places that memory at an accessible order, and every finite law, memory-bearing or not, admits the quantum representation.
 
@@ -5012,21 +5012,21 @@ The framework's substratum dynamics is given by the bijection $\varphi: S \to S$
 
 This raises an obvious question. The universe we observe has an unambiguous arrow of time: entropy increases, biological systems age, memories form of the past but not the future, the universe expands rather than contracts. If the substratum dynamics is T-symmetric, where does the observed arrow come from?
 
-**The arrow as observer-emergent.** The framework's answer is that the arrow of time is *emergent from the observer's coarse-grained description*, not from the substratum dynamics. The substratum dynamics is T-symmetric; the observer's reduced description is not. The asymmetry is introduced by the trace-out — the marginalization over the hidden sector that produces the embedded observer's reduced description. The arrow of time is therefore a structural consequence of embedded observation, in the same way that quantum mechanics, the dark sector, and the Born rule are.
+**The arrow as observer-emergent.** The framework's answer is that the arrow of time is *emergent from the observer's coarse-grained description*, not from the substratum dynamics. The substratum dynamics is T-symmetric; the observer's reduced description of a non-equilibrium initial condition is not. The asymmetry enters through the trace-out of a non-equilibrium initial condition — the marginalization over the hidden sector that produces the embedded observer's reduced description; under the invariant measure the trace-out alone produces none. The arrow of time is therefore a feature of embedded observation of such a condition, in the same way that quantum mechanics, the dark sector, and the Born rule are.
 
 The mechanism is the standard Boltzmann argument applied to the framework's partition. The visible sector starts in a particular configuration that is, in some sense, "low entropy" relative to the typical configurations accessible at later times. The hidden sector starts in a configuration that contains all the correlations needed to specify the visible sector's future evolution. As the dynamics evolves, correlations flow from hidden sector to visible sector and back, with the time-reversibility of the underlying dynamics fully preserved at the substratum level. But the observer, who sees only the visible sector, sees an irreversible flow: information about the past is encoded in the visible sector's current state (entropy increasing), while information about the future is encoded in the hidden sector (inaccessible to the observer).
 
 The arrow of time is therefore *the observer's incompleteness* expressed temporally. The same partition that produces quantum randomness (by hiding the substratum information that determines outcomes) produces the arrow of time (by hiding the substratum information that would let the observer reverse the dynamics). Both are aspects of the same structural feature: the observer's reduced description is asymmetric in ways that the substratum dynamics is not.
 
-**Detailed balance and the thermodynamic arrow.** The framework's content on detailed balance from Chapter 6 §6.3 has a direct implication for the arrow of time. Detailed balance asserts that the transition probabilities $T_{ij}$ between any two states satisfy $T_{ij} = T_{ji}$ at all timescales — the rate of transition from $i$ to $j$ equals the rate of transition from $j$ to $i$. This is a *substratum-level* T-symmetry: the dynamics has no preferred direction.
+**Detailed balance and the thermodynamic arrow.** The framework's content on detailed balance from Chapter 6 §6.3 has a direct implication for the arrow of time. For a time reversal that preserves the partition, and the counting prior on hidden states, the visible transition probabilities satisfy reciprocity, $T_{ij} = T_{\theta j, \theta i}$ at all timescales, with $\theta$ the action of time reversal on visible states; when the visible states carry no T-odd data this is detailed balance, $T_{ij} = T_{ji}$ — the rate of transition from $i$ to $j$ equals the rate of transition from $j$ to $i$. This follows from *substratum-level* T-symmetry: the dynamics has no preferred direction.
 
 The thermodynamic arrow of time emerges not from any asymmetry in the transition probabilities but from the partition's *initial conditions*. The universe starts in a configuration that is far from equilibrium relative to the substratum's full state space, with most of the substratum's degrees of freedom on the hidden side of the partition and the visible sector occupying a small, structured region. As the dynamics evolves, the system moves toward equilibrium relative to the substratum's state space — entropy increasing in the observer's coarse-grained description — until the entire substratum is fully mixed. The arrow of time runs in the direction of this equilibration.
 
 The relation to detailed balance is that the H-theorem — the standard derivation of entropy increase from microscopic dynamics — is a theorem about the *coarse-grained* description, not about the underlying transition probabilities. The H-theorem assumes a particular kind of initial condition (low entropy at the coarse-grained level) and derives entropy increase from there. Detailed balance at the microscopic level is fully consistent with the H-theorem at the coarse-grained level, because the H-theorem is about how the observer's information about the system evolves rather than about the system's underlying dynamics.
 
-**The cosmological arrow.** The thermodynamic arrow and the cosmological arrow — the direction in which the universe expands — are aligned in the framework. The cosmological partition has a specific initial condition: the universe begins in a low-entropy configuration with most degrees of freedom on the hidden side of the cosmological horizon. Time evolution corresponds to the horizon entropy increasing toward equilibrium, with the boundary modes accumulating correlations and the hidden sector approaching maximum disorder.
+**The cosmological arrow.** Whether the thermodynamic arrow and the cosmological arrow — the direction in which the universe expands — are aligned is open in the framework: it is neither established nor refuted. The cosmological partition has a specific initial condition: the universe begins in a low-entropy configuration with most degrees of freedom on the hidden side of the cosmological horizon. On an expanding FRW background satisfying the null energy condition, the apparent-horizon area is non-decreasing. Its assigned entropy has that interpretation under the stated calibration assumptions; whether this behavior aligns with boundary correlations or hidden-sector disorder remains open.
 
-The alignment of the thermodynamic and cosmological arrows is structural rather than coincidental. Both arrows trace to the same initial conditions of the cosmological partition. The universe started with low entropy at the cosmological level; the framework's structural conditions C1–C4 hold at the cosmological boundary (Chapter 7 §7.2); the thermodynamic arrow within the visible sector tracks the cosmological arrow because the cosmological boundary is the primary partition for any embedded observer.
+Both arrows are traced to the same initial conditions of the cosmological partition. That is a common boundary condition, not a derived mechanism: that the thermodynamic arrow within the visible sector tracks the cosmological arrow is not established. The universe started with low entropy at the cosmological level, as an initial condition; at the cosmological boundary, coupling C1, persistence C2 and capacity C3 are verified, while history readback C4 is a named realization condition not presently discharged (Chapter 7 §7.2); the cosmological boundary is the primary partition for any embedded observer.
 
 **Why the framework reframes rather than explains.** The framework's content on the arrow of time is not an explanation of *why* the universe has a preferred direction; it is a *reframing* of the question. Standard accounts of the arrow of time treat it as an asymmetry that requires explanation — why is entropy increasing in the direction we call "future" rather than the direction we call "past"? The framework's answer is that there is no asymmetry in the substratum dynamics; the asymmetry is in the observer's reduced description, which inherits the initial conditions of the partition.
 
diff --git a/book/The-Incompleteness-of-Observation-FULL.pdf b/book/The-Incompleteness-of-Observation-FULL.pdf
index d3128f716..0785c2004 100644
Binary files a/book/The-Incompleteness-of-Observation-FULL.pdf and b/book/The-Incompleteness-of-Observation-FULL.pdf differ
diff --git a/book/The-Incompleteness-of-Observation-FULL.tex b/book/The-Incompleteness-of-Observation-FULL.tex
index b37c93bea..a22086694 100644
--- a/book/The-Incompleteness-of-Observation-FULL.tex
+++ b/book/The-Incompleteness-of-Observation-FULL.tex
@@ -3740,9 +3740,9 @@ \subsection{2.5 The substratum gauge
 physics fixes that local residue under the theorem's stated conditions.
 H-Bell separately conditions Bell-inclusive existence, and
 Bell-inclusive uniqueness is not established. The next section turns to
-the synthesis: how quantum mechanics, general relativity, and the arrow
-of time relate to the substratum as three projections of the same
-construction.
+the synthesis: how quantum mechanics and general relativity relate to
+the substratum as projections of the same construction, and what a third
+projection establishes about the arrow of time.
 
 \begin{center}\rule{0.5\linewidth}{0.5pt}\end{center}
 
@@ -3756,13 +3756,13 @@ \subsection{2.6 The three projections}\label{the-three-projections}}
 emergent structures: quantum mechanics, general relativity, and the
 arrow of time.
 
-The framework's claim is that quantum mechanics, general relativity, and
-the arrow of time are not three independent theories. Conditional on
-H-Bell at the Bell/local-lattice interface, they are three projections
-of the same finite deterministic construction. Each projection applies
-the same trace-out machinery to the same \((S, \varphi)\) along a
-different axis, and each produces a distinct emergent structure as the
-observer sees the substratum from inside.
+The framework's claim is that quantum mechanics and general relativity
+are not two independent theories. Conditional on H-Bell at the
+Bell/local-lattice interface, they are projections of the same finite
+deterministic construction, and a third projection bears on the arrow of
+time. Each projection applies the same trace-out machinery to the same
+\((S, \varphi)\) along a different axis, and each produces a distinct
+emergent structure as the observer sees the substratum from inside.
 
 \textbf{The visible-sector projection: quantum mechanics.} Chapter 1
 developed this projection in detail. At finite observable horizon the
@@ -3834,60 +3834,75 @@ \subsection{2.6 The three projections}\label{the-three-projections}}
 projection applies the trace-out machinery to the horizon-growth
 parameter along an observer worldline. The substratum is time-symmetric:
 \(\varphi\) and \(\varphi^{-1}\) are equally valid as forward-time maps,
-and there is no preferred direction at the substratum level. Observers,
-however, are not symmetric under this exchange. Condition C2 ---
-persistence, in its slow-bath realization --- picks out one direction
-along the observer's worldline as the direction in which the hidden
-sector's correlations with the visible sector's past are preserved. The
-structural observer-selection theorem developed in Chapter 1's appendix
-material --- under the effective-mixing hypothesis, for sufficiently
-fast-mixing equilibria (\(\tau_{\rm mix} \ll \tau_S\)), and in the
-coarse-grained description generated by iterating the one-step effective
-channel ({[}Main §4.6{]}) --- restricts observers satisfying C1--C4,
-with C2 read in its timescale form (record persistence and readback
-there being suppressed, not excluded), to the non-equilibrium phase of
-\((S, \varphi)\); in the \(\Lambda\)-dominated cosmological phase, this
-phase is the one in which the horizon entropy
-\(S_{\text{dS}}(t) = A(t)/(4\ell_P^2)\) is monotonically increasing.
-
-The cosmological horizon entropy provides a structural clock. Every
-observer in the accelerating phase agrees that \(S_{\text{dS}}\)
-increases --- the monotonicity is a feature of the causal partition
-rather than of any local frame --- and other arrows of time align with
-this structural clock through a cascade. The thermodynamic arrow
-descends from Jacobson's coupling between boundary entropy and matter
-entropy. The memory arrow descends from C2 --- persistence, realized by
-the slow bath --- which preserves correlations forward along the clock
-rather than backward. The measurement arrow --- the observer records
-past outcomes and prepares future states --- descends from the memory
-arrow. The chain grounds in one structural fact (horizon growth in the
-accelerating phase) and one structural constraint (C1--C4 observer
-selection, conditional on the effective-mixing hypothesis of {[}Main
-§4.6{]} and stated for the coarse-grained description generated by
-iterating the one-step effective channel). No independent
+and there is no preferred direction at the substratum level. Condition
+C2 --- persistence, in its slow-bath realization --- is a retention
+condition: it requires the hidden sector's correlations with the visible
+sector's past to persist to readback, and the coarse-grained mixing time
+it constrains is the same in the two directions; the direction in which
+records are written and read is fixed by a preparation and the
+observer's orientation convention. The structural observer-selection
+theorem developed in Chapter 1's appendix material --- under the
+effective-mixing hypothesis, for sufficiently fast-mixing equilibria
+(\(\tau_{\rm mix} \ll \tau_S\)), and in the coarse-grained description
+generated by iterating the one-step effective channel ({[}Main §4.6{]})
+--- restricts observers satisfying C1--C4, with C2 read in its timescale
+form (record persistence and readback there being suppressed, not
+excluded), to the non-equilibrium phase of \((S, \varphi)\); the
+restriction is time-symmetric and selects where observers can exist, not
+a direction. On an expanding cosmological background satisfying the null
+energy condition --- inputs, not derived --- the area \(A(t)\) of the
+apparent horizon, and the horizon entropy
+\(S_{\text{dS}}(t) = A(t)/(4\ell_P^2)\) assigned to it at the de Sitter
+calibration, is non-decreasing, strictly increasing while
+\(\rho + p/c^2 > 0\) and constant in exact de Sitter. That the selected
+non-equilibrium phase is ordered by this growth is not established.
+
+The apparent-horizon area provides a structural clock on that
+background. Every comoving observer of an expanding homogeneous
+general-relativistic background satisfying the null energy condition
+agrees that it does not decrease --- standard background kinematics, not
+a prediction derived from the substratum --- and the criterion is
+expansion with the null energy condition, not acceleration: in exact de
+Sitter the area is constant. Whether the other arrows of time align with
+this clock is not established. Jacobson's coupling relates the boundary
+energy density to the expansion rate, not matter entropy to horizon
+area, so the direction of the thermodynamic arrow rests on a low-entropy
+or product boundary condition, and the generalized second law is open.
+C2 --- persistence, realized by the slow bath --- holds the same way in
+both directions; the direction of the memory arrow comes from a
+preparation and the observer's orientation convention. The measurement
+arrow --- the observer records past outcomes and prepares future states
+--- inherits the memory arrow's direction. The chain uses one input
+(horizon growth on the expanding branch) and one structural constraint
+(C1--C4 observer selection, conditional on the effective-mixing
+hypothesis of {[}Main §4.6{]} and stated for the coarse-grained
+description generated by iterating the one-step effective channel),
+which fixes where observers can exist, not a direction. No independent
 past-hypothesis axiom is required to select where observers exist, and
 Boltzmann brains are excluded structurally rather than anthropically.
 
-The framework's content here is precise: it derives a monotonic
-parameter along observer worldlines and four aligned arrows of time.
-What it does not derive is the convention of which direction along the
+The framework's content here is precise: on the expanding branch it
+establishes a non-decreasing horizon parameter along observer
+worldlines, and it does not establish that the four arrows of time
+align. Nor does it derive the convention of which direction along the
 horizon-clock parameter to call ``future.'' The two directions are
 related by the layer swap at the substrate level
 (\(\varphi \leftrightarrow \varphi^{-1}\)), and the observer's choice of
-convention is residual freedom. The framework derives the existence of a
-monotonic parameter; the labeling is an observer-imposed convention.
-There is one further qualification worth stating: a monotonic parameter
-is monotonic with respect to an \emph{ordering} of the worldline's
-moments, and the framework's foundational treatment axiomatizes that
-there is more than one moment while leaving the ordering of moments as a
-separate, open question. The derivation here is therefore conditional on
-that ordering --- it establishes that ordered moments yield a directed
-parameter, not that directedness arises from the substratum with no
-assumption about ordering at all. CP violation in CKM and kaon/B systems
-is a flavor-basis feature of the specific bijection \(\varphi\) rather
-than a consequence of the time-arrow cascade --- the framework treats
-flavor-sector CP phases as solution-specific inputs that do not affect
-the cascade's structural content.
+convention is residual freedom. The parameter is conditional on the
+expanding branch; the labeling is an observer-imposed convention. There
+is one further qualification worth stating: a monotonic parameter is
+monotonic with respect to an \emph{ordering} of the worldline's moments,
+and the framework's foundational treatment axiomatizes that there is
+more than one moment while leaving the ordering of moments as a
+separate, open question. The result here is therefore conditional on
+that ordering as well --- it establishes that ordered moments on the
+expanding branch yield a non-decreasing parameter, not that directedness
+arises from the substratum with no assumption about ordering at all. CP
+violation in CKM and kaon/B systems is a flavor-basis feature of the
+specific bijection \(\varphi\) rather than a consequence of the
+time-arrow cascade --- the framework treats flavor-sector CP phases as
+solution-specific inputs that do not affect the cascade's structural
+content.
 
 \textbf{The common structure.} The three projections share three
 features. Under H-Bell they apply to the same total substratum
@@ -3906,8 +3921,9 @@ \subsection{2.6 The three projections}\label{the-three-projections}}
 boundary-thermodynamic projection works on the classical thermal data at
 the partition boundary and produces general relativity with its
 gravitational and cosmological predictions. The horizon-clock projection
-works on the horizon-growth parameter along observer worldlines and
-produces the cascade of emergent arrows.
+works on the horizon-growth parameter along observer worldlines together
+with the observer-selection rule; the alignment of the arrows with that
+parameter is open.
 
 The shared structure is not metaphor or analogy. The same local lattice
 sector that gives the Standard Model gauge group from the commutant of
@@ -4034,9 +4050,10 @@ \subsection{2.8 What follows}\label{what-follows}}
 
 The reconstruction theorem and the substratum gauge group establish the
 substratum side of the framework. The three projections establish its
-relationship to the framework's three large emergent structures. With
-both in place, the technical core of Part I is complete; the chapters
-that follow develop its consequences in three directions.
+relationship to quantum mechanics and general relativity, and what it
+establishes about the arrow of time. With both in place, the technical
+core of Part I is complete; the chapters that follow develop its
+consequences in three directions.
 
 Chapter 3 develops the framework's philosophical positioning. The
 substratum's ontological status --- what it means to commit to the
@@ -4443,11 +4460,13 @@ \subsection{3.3 The observer-admission distinction in more
 ({[}Main §2.3{]}). Boltzmann brains are not generic outcomes of
 long-time evolution that just happen to be rare; they are structurally
 excluded from the observer-admitting subset by C2 in that timescale
-form. The framework's account of the arrow of time, developed in Chapter
-2 §2.6, follows from this exclusion: observers are restricted to the
-non-equilibrium phase of the substratum's dynamics, and the structural
-arrow of horizon growth provides the clock the observer's other arrows
-align with.
+form. The framework's treatment of the arrow of time, developed in
+Chapter 2 §2.6, starts from this exclusion: observers are restricted to
+the non-equilibrium phase of the substratum's dynamics. The exclusion is
+time-symmetric; on an expanding homogeneous general-relativistic
+background satisfying the null energy condition, the apparent-horizon
+area supplies a non-decreasing clock, and whether the observer's other
+arrows align with it is not established.
 
 The string-theory landscape problem looks different under the Level B
 reading as well. The conventional landscape picture posits roughly
@@ -5243,15 +5262,15 @@ \subsection{3.8 The philosophical
 feature of how the substance is expressed rather than of the substance
 itself --- maps directly onto the framework's three-projection
 structure. The substratum \((S, \varphi)\) is the single substance;
-quantum mechanics, general relativity, and the arrow of time are
-projections expressing the substratum's content under different aspects;
-the appearance of three distinct emergent theories is a feature of the
-projections rather than of the substratum. The framework's commitment to
-ontic structural realism on the \textbf{reconstructed local residue
-modulo} \(\mathcal{G}_{\text{sub}}\) is Spinozistic in form: what is
-real is the gauge-invariant local structure, not any specific
-representative; the Bell-inclusive completion is not uniquely
-reconstructed.
+quantum mechanics and general relativity are projections expressing the
+substratum's content under different aspects, with a third bearing on
+the arrow of time; the appearance of three distinct emergent theories is
+a feature of the projections rather than of the substratum. The
+framework's commitment to ontic structural realism on the
+\textbf{reconstructed local residue modulo} \(\mathcal{G}_{\text{sub}}\)
+is Spinozistic in form: what is real is the gauge-invariant local
+structure, not any specific representative; the Bell-inclusive
+completion is not uniquely reconstructed.
 
 The parallel is not complete. Spinoza's substance is unique (there is
 exactly one substance, by argument); the framework's substratum is a
@@ -5684,9 +5703,10 @@ \subsection{4.2 The single-foundational-commitment
 1. The framework does not commit to specific values of free parameters
 in the emergent quantum field theory beyond what the reconstruction
 theorem forces. The framework does not commit to a specific cosmological
-history beyond the structural arrow of horizon growth in the
-\(\Lambda\)-dominated phase. These non-commitments are part of the
-foundational economy: each additional commitment would expand the
+history beyond an expanding homogeneous general-relativistic background,
+taken as given, on which, with the null energy condition, the
+apparent-horizon area does not decrease. These non-commitments are part
+of the foundational economy: each additional commitment would expand the
 framework's claims, and the discipline is to make only those commitments
 that the foundational premise plus its derivations require.
 
@@ -22788,14 +22808,15 @@ \subsection{18.7 Finite information and exact finite
 space, no single cycle passes through every configuration. The exact
 linear rule of {[}SM §4.2{]} has this property and therefore at least
 two cycles. The cycle-ergodicity question that {[}SM §4.2{]} treats
-concerns sampling within an energy shell. Within a single cycle, the
-observer's experience is unidirectional (the arrow of time discussed in
-the next section), but at the substratum level the dynamics has no
-preferred direction. The framework's commitment to substratum cyclicity
-is one of the structural commitments that distinguishes it from
-frameworks treating the universe as either fundamentally non-cyclic
-(most cosmological models) or fundamentally infinite (most quantum
-gravity programs).
+concerns sampling within an energy shell. Within a single cycle no state
+function increases throughout; a direction of the observer's experience
+is fixed by boundary conditions on a stretch of the cycle (the arrow of
+time discussed in the next section), and at the substratum level the
+dynamics has no preferred direction. The framework's commitment to
+substratum cyclicity is one of the structural commitments that
+distinguishes it from frameworks treating the universe as either
+fundamentally non-cyclic (most cosmological models) or fundamentally
+infinite (most quantum gravity programs).
 
 \textbf{Implications for the framework's logical structure.} The
 recurrence theorem connects three structural features of the framework:
@@ -22852,12 +22873,14 @@ \subsection{18.8 The arrow of time}\label{the-arrow-of-time}}
 \textbf{The arrow as observer-emergent.} The framework's answer is that
 the arrow of time is \emph{emergent from the observer's coarse-grained
 description}, not from the substratum dynamics. The substratum dynamics
-is T-symmetric; the observer's reduced description is not. The asymmetry
-is introduced by the trace-out --- the marginalization over the hidden
-sector that produces the embedded observer's reduced description. The
-arrow of time is therefore a structural consequence of embedded
-observation, in the same way that quantum mechanics, the dark sector,
-and the Born rule are.
+is T-symmetric; the observer's reduced description of a non-equilibrium
+initial condition is not. The asymmetry enters through the trace-out of
+a non-equilibrium initial condition --- the marginalization over the
+hidden sector that produces the embedded observer's reduced description;
+under the invariant measure the trace-out alone produces none. The arrow
+of time is therefore a feature of embedded observation of such a
+condition, in the same way that quantum mechanics, the dark sector, and
+the Born rule are.
 
 The mechanism is the standard Boltzmann argument applied to the
 framework's partition. The visible sector starts in a particular
@@ -22883,12 +22906,15 @@ \subsection{18.8 The arrow of time}\label{the-arrow-of-time}}
 
 \textbf{Detailed balance and the thermodynamic arrow.} The framework's
 content on detailed balance from Chapter 6 §6.3 has a direct implication
-for the arrow of time. Detailed balance asserts that the transition
-probabilities \(T_{ij}\) between any two states satisfy
-\(T_{ij} = T_{ji}\) at all timescales --- the rate of transition from
+for the arrow of time. For a time reversal that preserves the partition,
+and the counting prior on hidden states, the visible transition
+probabilities satisfy reciprocity, \(T_{ij} = T_{\theta j, \theta i}\)
+at all timescales, with \(\theta\) the action of time reversal on
+visible states; when the visible states carry no T-odd data this is
+detailed balance, \(T_{ij} = T_{ji}\) --- the rate of transition from
 \(i\) to \(j\) equals the rate of transition from \(j\) to \(i\). This
-is a \emph{substratum-level} T-symmetry: the dynamics has no preferred
-direction.
+follows from \emph{substratum-level} T-symmetry: the dynamics has no
+preferred direction.
 
 The thermodynamic arrow of time emerges not from any asymmetry in the
 transition probabilities but from the partition's \emph{initial
@@ -22912,23 +22938,28 @@ \subsection{18.8 The arrow of time}\label{the-arrow-of-time}}
 about the system evolves rather than about the system's underlying
 dynamics.
 
-\textbf{The cosmological arrow.} The thermodynamic arrow and the
+\textbf{The cosmological arrow.} Whether the thermodynamic arrow and the
 cosmological arrow --- the direction in which the universe expands ---
-are aligned in the framework. The cosmological partition has a specific
-initial condition: the universe begins in a low-entropy configuration
-with most degrees of freedom on the hidden side of the cosmological
-horizon. Time evolution corresponds to the horizon entropy increasing
-toward equilibrium, with the boundary modes accumulating correlations
-and the hidden sector approaching maximum disorder.
-
-The alignment of the thermodynamic and cosmological arrows is structural
-rather than coincidental. Both arrows trace to the same initial
-conditions of the cosmological partition. The universe started with low
-entropy at the cosmological level; the framework's structural conditions
-C1--C4 hold at the cosmological boundary (Chapter 7 §7.2); the
-thermodynamic arrow within the visible sector tracks the cosmological
-arrow because the cosmological boundary is the primary partition for any
-embedded observer.
+are aligned is open in the framework: it is neither established nor
+refuted. The cosmological partition has a specific initial condition:
+the universe begins in a low-entropy configuration with most degrees of
+freedom on the hidden side of the cosmological horizon. On an expanding
+FRW background satisfying the null energy condition, the
+apparent-horizon area is non-decreasing. Its assigned entropy has that
+interpretation under the stated calibration assumptions; whether this
+behavior aligns with boundary correlations or hidden-sector disorder
+remains open.
+
+Both arrows are traced to the same initial conditions of the
+cosmological partition. That is a common boundary condition, not a
+derived mechanism: that the thermodynamic arrow within the visible
+sector tracks the cosmological arrow is not established. The universe
+started with low entropy at the cosmological level, as an initial
+condition; at the cosmological boundary, coupling C1, persistence C2 and
+capacity C3 are verified, while history readback C4 is a named
+realization condition not presently discharged (Chapter 7 §7.2); the
+cosmological boundary is the primary partition for any embedded
+observer.
 
 \textbf{Why the framework reframes rather than explains.} The
 framework's content on the arrow of time is not an explanation of
@@ -30964,4 +30995,4 @@ \subsection{Historical and philosophical
 {[}Origin of the molecular clock hypothesis.{]}
 
 \end{document}
-% source-sha256: 0a0d40464753a7d95cf0b71d3f52e6ac864ecff608db2df30622214a6021220f
+% source-sha256: 6a65fc11a8366ffcb5d40a59791177e010278de23242bafd5bfe4c038a2ce3f1
diff --git a/book/ch02-substratum.md b/book/ch02-substratum.md
index 750834410..e24dfc5e8 100644
--- a/book/ch02-substratum.md
+++ b/book/ch02-substratum.md
@@ -173,7 +173,7 @@ The substratum construction provides a structural account of why the Standard Mo
 
 This is what makes the framework's scoped structural realism non-trivial. The invariants of the **reconstructed local lattice/gauge residue** are uniquely determined — under the reconstruction theorem's stated conditions — by observation plus structural assumptions. Individual representatives of that local residue related by the generators of $\mathcal{G}_{\text{sub}}$ are gauge expressions of the same reconstructed physical content, like coordinate choices on a manifold or specific representatives of a homotopy class. Bell-composite preparation data lie outside this uniqueness statement: H-Bell conditions compatibility/existence, and no result here selects a unique Bell-inclusive full representative.
 
-The reconstruction theorem identifies the **local lattice/gauge residue**; the gauge group identifies what within that scoped reconstruction has no observable content. With both in place, observed physics fixes that local residue under the theorem's stated conditions. H-Bell separately conditions Bell-inclusive existence, and Bell-inclusive uniqueness is not established. The next section turns to the synthesis: how quantum mechanics, general relativity, and the arrow of time relate to the substratum as three projections of the same construction.
+The reconstruction theorem identifies the **local lattice/gauge residue**; the gauge group identifies what within that scoped reconstruction has no observable content. With both in place, observed physics fixes that local residue under the theorem's stated conditions. H-Bell separately conditions Bell-inclusive existence, and Bell-inclusive uniqueness is not established. The next section turns to the synthesis: how quantum mechanics and general relativity relate to the substratum as projections of the same construction, and what a third projection establishes about the arrow of time.
 
 ***
 
@@ -181,7 +181,7 @@ The reconstruction theorem identifies the **local lattice/gauge residue**; the g
 
 The reconstruction theorem and the substratum gauge group have established what $(S, \varphi)$ is and what equivalences leave its observables unchanged. The question taken up in this section is what relationship $(S, \varphi)$ bears to the framework's three large emergent structures: quantum mechanics, general relativity, and the arrow of time.
 
-The framework's claim is that quantum mechanics, general relativity, and the arrow of time are not three independent theories. Conditional on H-Bell at the Bell/local-lattice interface, they are three projections of the same finite deterministic construction. Each projection applies the same trace-out machinery to the same $(S, \varphi)$ along a different axis, and each produces a distinct emergent structure as the observer sees the substratum from inside.
+The framework's claim is that quantum mechanics and general relativity are not two independent theories. Conditional on H-Bell at the Bell/local-lattice interface, they are projections of the same finite deterministic construction, and a third projection bears on the arrow of time. Each projection applies the same trace-out machinery to the same $(S, \varphi)$ along a different axis, and each produces a distinct emergent structure as the observer sees the substratum from inside.
 
 **The visible-sector projection: quantum mechanics.** Chapter 1 developed this projection in detail. At finite observable horizon the reduced stochastic law, a finite reversible realization, and a fixed-basis unitary/Born representation are exactly equivalent: $S\iff D\iff Q_{\mathrm{fb}}$. C4 readback then identifies the memory-bearing sector by forcing hidden predictive memory. In the spatial construction the coupling graph supplies exact causal cones for finite-range substrata. Bell is separate from the full-QM lift: a measurement-independent response-complete deterministic completion with parameter independence and a $\pm1$ outcome in every run would obey $|S_{\rm CHSH}|\le2$, so the framework's Bell-violating branch is ontically parameter-dependent with operational no-signaling required. Two independent qualifications remain beyond the finite-law equivalence: the simultaneous coherent instrument/composite lift ([Main §3.4]) and H-Bell, the compatibility of the local lattice sector with the ontically parameter-dependent Bell completion. Applied to the local lattice sector constructed in this chapter — a $d = 3$ cubic lattice with $K = 6$ internal components and the wave equation as its local propagating dynamics — the visible-sector projection produces the Standard Model: the gauge group $\mathrm{SU}(3) \times \mathrm{SU}(2) \times \mathrm{U}(1)$ from the commutant of the cubic-symmetric coupling matrix, three chiral fermion generations, one composite Higgs, and anomaly-free hypercharges. Chapters 5 and 6 develop the Standard Model derivation; Chapter 8 develops the specific JUNO neutrino measurement that the framework's parameter-free PMNS-angle value matches at $0.07\sigma$ (a retrodiction). The local nearest-neighbor sector cannot itself supply Bell parameter dependence for wings with one-step own-wing setting interventions outside each other's graph cone under measurement independence; identifying it with one Bell-violating total completion is the open H-Bell compatibility hypothesis.
 
@@ -189,13 +189,13 @@ The framework's claim is that quantum mechanics, general relativity, and the arr
 
 Applied to the cosmological horizon as a causal partition, the boundary-thermodynamic projection produces general relativity with quantitative predictions. The emergent action scale $\hbar = c^3 \epsilon^2 / (4G\sigma)$, with $\epsilon = 2\sqrt{\sigma}\,\ell_P$, follows from thermal self-consistency at the boundary layer under the infrared detailed-balance condition H-slope. The boundary-mode count reproduces the Bekenstein-Hawking entropy $S = A/(4 \ell_P^2)$, the $1/4$ coefficient being the ratio $2\pi/8\pi$ that the calibration matches and conditional on the horizon conditions of Chapter 7; GW250114 (September 2025) confirms the classical area theorem, which the framework preserves — the $1/4$ coefficient itself has no direct empirical test. The cosmological constant problem receives a conditional treatment: the quantum vacuum energy and the effective $\Lambda$ are properties of logically distinct levels of description rather than commensurable quantities whose $10^{122}$ ratio requires cancellation; the dissolution is conditional on the covariant coupling step (Chapter 7). The structured component of the boundary entropy produces dark matter phenomenology with MOND acceleration scale $a_0 = cH/6 \approx 1.1 \times 10^{-10}$ m/s² (the dimensional factor imported from Verlinde's emergent-gravity analysis). Chapter 7 develops each of these.
 
-**The horizon-clock projection: the arrow of time.** A third projection applies the trace-out machinery to the horizon-growth parameter along an observer worldline. The substratum is time-symmetric: $\varphi$ and $\varphi^{-1}$ are equally valid as forward-time maps, and there is no preferred direction at the substratum level. Observers, however, are not symmetric under this exchange. Condition C2 — persistence, in its slow-bath realization — picks out one direction along the observer's worldline as the direction in which the hidden sector's correlations with the visible sector's past are preserved. The structural observer-selection theorem developed in Chapter 1's appendix material — under the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), and in the coarse-grained description generated by iterating the one-step effective channel ([Main §4.6]) — restricts observers satisfying C1–C4, with C2 read in its timescale form (record persistence and readback there being suppressed, not excluded), to the non-equilibrium phase of $(S, \varphi)$; in the $\Lambda$-dominated cosmological phase, this phase is the one in which the horizon entropy $S_{\text{dS}}(t) = A(t)/(4\ell_P^2)$ is monotonically increasing.
+**The horizon-clock projection: the arrow of time.** A third projection applies the trace-out machinery to the horizon-growth parameter along an observer worldline. The substratum is time-symmetric: $\varphi$ and $\varphi^{-1}$ are equally valid as forward-time maps, and there is no preferred direction at the substratum level. Condition C2 — persistence, in its slow-bath realization — is a retention condition: it requires the hidden sector's correlations with the visible sector's past to persist to readback, and the coarse-grained mixing time it constrains is the same in the two directions; the direction in which records are written and read is fixed by a preparation and the observer's orientation convention. The structural observer-selection theorem developed in Chapter 1's appendix material — under the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), and in the coarse-grained description generated by iterating the one-step effective channel ([Main §4.6]) — restricts observers satisfying C1–C4, with C2 read in its timescale form (record persistence and readback there being suppressed, not excluded), to the non-equilibrium phase of $(S, \varphi)$; the restriction is time-symmetric and selects where observers can exist, not a direction. On an expanding cosmological background satisfying the null energy condition — inputs, not derived — the area $A(t)$ of the apparent horizon, and the horizon entropy $S_{\text{dS}}(t) = A(t)/(4\ell_P^2)$ assigned to it at the de Sitter calibration, is non-decreasing, strictly increasing while $\rho + p/c^2 > 0$ and constant in exact de Sitter. That the selected non-equilibrium phase is ordered by this growth is not established.
 
-The cosmological horizon entropy provides a structural clock. Every observer in the accelerating phase agrees that $S_{\text{dS}}$ increases — the monotonicity is a feature of the causal partition rather than of any local frame — and other arrows of time align with this structural clock through a cascade. The thermodynamic arrow descends from Jacobson's coupling between boundary entropy and matter entropy. The memory arrow descends from C2 — persistence, realized by the slow bath — which preserves correlations forward along the clock rather than backward. The measurement arrow — the observer records past outcomes and prepares future states — descends from the memory arrow. The chain grounds in one structural fact (horizon growth in the accelerating phase) and one structural constraint (C1–C4 observer selection, conditional on the effective-mixing hypothesis of [Main §4.6] and stated for the coarse-grained description generated by iterating the one-step effective channel). No independent past-hypothesis axiom is required to select where observers exist, and Boltzmann brains are excluded structurally rather than anthropically.
+The apparent-horizon area provides a structural clock on that background. Every comoving observer of an expanding homogeneous general-relativistic background satisfying the null energy condition agrees that it does not decrease — standard background kinematics, not a prediction derived from the substratum — and the criterion is expansion with the null energy condition, not acceleration: in exact de Sitter the area is constant. Whether the other arrows of time align with this clock is not established. Jacobson's coupling relates the boundary energy density to the expansion rate, not matter entropy to horizon area, so the direction of the thermodynamic arrow rests on a low-entropy or product boundary condition, and the generalized second law is open. C2 — persistence, realized by the slow bath — holds the same way in both directions; the direction of the memory arrow comes from a preparation and the observer's orientation convention. The measurement arrow — the observer records past outcomes and prepares future states — inherits the memory arrow's direction. The chain uses one input (horizon growth on the expanding branch) and one structural constraint (C1–C4 observer selection, conditional on the effective-mixing hypothesis of [Main §4.6] and stated for the coarse-grained description generated by iterating the one-step effective channel), which fixes where observers can exist, not a direction. No independent past-hypothesis axiom is required to select where observers exist, and Boltzmann brains are excluded structurally rather than anthropically.
 
-The framework's content here is precise: it derives a monotonic parameter along observer worldlines and four aligned arrows of time. What it does not derive is the convention of which direction along the horizon-clock parameter to call "future." The two directions are related by the layer swap at the substrate level ($\varphi \leftrightarrow \varphi^{-1}$), and the observer's choice of convention is residual freedom. The framework derives the existence of a monotonic parameter; the labeling is an observer-imposed convention. There is one further qualification worth stating: a monotonic parameter is monotonic with respect to an *ordering* of the worldline's moments, and the framework's foundational treatment axiomatizes that there is more than one moment while leaving the ordering of moments as a separate, open question. The derivation here is therefore conditional on that ordering — it establishes that ordered moments yield a directed parameter, not that directedness arises from the substratum with no assumption about ordering at all. CP violation in CKM and kaon/B systems is a flavor-basis feature of the specific bijection $\varphi$ rather than a consequence of the time-arrow cascade — the framework treats flavor-sector CP phases as solution-specific inputs that do not affect the cascade's structural content.
+The framework's content here is precise: on the expanding branch it establishes a non-decreasing horizon parameter along observer worldlines, and it does not establish that the four arrows of time align. Nor does it derive the convention of which direction along the horizon-clock parameter to call "future." The two directions are related by the layer swap at the substrate level ($\varphi \leftrightarrow \varphi^{-1}$), and the observer's choice of convention is residual freedom. The parameter is conditional on the expanding branch; the labeling is an observer-imposed convention. There is one further qualification worth stating: a monotonic parameter is monotonic with respect to an *ordering* of the worldline's moments, and the framework's foundational treatment axiomatizes that there is more than one moment while leaving the ordering of moments as a separate, open question. The result here is therefore conditional on that ordering as well — it establishes that ordered moments on the expanding branch yield a non-decreasing parameter, not that directedness arises from the substratum with no assumption about ordering at all. CP violation in CKM and kaon/B systems is a flavor-basis feature of the specific bijection $\varphi$ rather than a consequence of the time-arrow cascade — the framework treats flavor-sector CP phases as solution-specific inputs that do not affect the cascade's structural content.
 
-**The common structure.** The three projections share three features. Under H-Bell they apply to the same total substratum $(S, \varphi)$; without H-Bell, the local SM/GR lattice sector and the Bell-violating completion have not been shown to be one object. They use the same trace-out machinery — marginalization over the hidden sector under conditions C1–C4. They produce derived rather than fundamental structures: nothing in quantum mechanics, general relativity, or the arrow of time is a feature of $(S, \varphi)$ in itself; each is a feature of how the embedded observer sees the substratum from inside. The differences among the three projections lie entirely in which structure of the embedded observer's description they extract. The visible-sector projection works on the bulk dynamics and produces the reduced finite law — universally admitting the quantum representation — with the Standard Model as its specific content. The boundary-thermodynamic projection works on the classical thermal data at the partition boundary and produces general relativity with its gravitational and cosmological predictions. The horizon-clock projection works on the horizon-growth parameter along observer worldlines and produces the cascade of emergent arrows.
+**The common structure.** The three projections share three features. Under H-Bell they apply to the same total substratum $(S, \varphi)$; without H-Bell, the local SM/GR lattice sector and the Bell-violating completion have not been shown to be one object. They use the same trace-out machinery — marginalization over the hidden sector under conditions C1–C4. They produce derived rather than fundamental structures: nothing in quantum mechanics, general relativity, or the arrow of time is a feature of $(S, \varphi)$ in itself; each is a feature of how the embedded observer sees the substratum from inside. The differences among the three projections lie entirely in which structure of the embedded observer's description they extract. The visible-sector projection works on the bulk dynamics and produces the reduced finite law — universally admitting the quantum representation — with the Standard Model as its specific content. The boundary-thermodynamic projection works on the classical thermal data at the partition boundary and produces general relativity with its gravitational and cosmological predictions. The horizon-clock projection works on the horizon-growth parameter along observer worldlines together with the observer-selection rule; the alignment of the arrows with that parameter is open.
 
 The shared structure is not metaphor or analogy. The same local lattice sector that gives the Standard Model gauge group from the commutant of the coupling matrix gives the Bekenstein-Hawking entropy as the boundary mode count. H-Bell is required only for identifying that local sector with the Bell-violating total completion. The conditions C1–C4 that characterize the memory-bearing sector — whose conditional memory readback makes measurable — for the visible-sector projection produce the persistence-based observer selection for the horizon-clock projection. The substratum gauge group $\mathcal{G}_{\text{sub}}$ controls the gauge equivalences of the **locally reconstructed residue** across the three projections; it does not classify all preparation-indexed Bell-composite rules. The framework's claim is that the three projections are co-derived from a single construction, not three separate applications of similar machinery.
 
@@ -217,7 +217,7 @@ The framework's synthesis claim is structural and architectural, not predictive
 
 ## 2.8 What follows
 
-The reconstruction theorem and the substratum gauge group establish the substratum side of the framework. The three projections establish its relationship to the framework's three large emergent structures. With both in place, the technical core of Part I is complete; the chapters that follow develop its consequences in three directions.
+The reconstruction theorem and the substratum gauge group establish the substratum side of the framework. The three projections establish its relationship to quantum mechanics and general relativity, and what it establishes about the arrow of time. With both in place, the technical core of Part I is complete; the chapters that follow develop its consequences in three directions.
 
 Chapter 3 develops the framework's philosophical positioning. The substratum's ontological status — what it means to commit to the equivalence class $[(S, \varphi)] / \mathcal{G}_{\text{sub}}$ as the framework's metaphysical object — is articulated in its full hierarchical form, with realism operating at multiple levels of the observation hierarchy and the gauge hierarchy. The framework's relation to Bell, Kochen-Specker and PBR-type no-go theorems is developed through separate checks of their local-response, operational-algebra and preparation premises, with the stated unresolved obligations retained. The framework's intellectual position is located against the contemporary observer-dependence programs (QBism, relational quantum mechanics, the recent CLPW and Maldacena work on observer-incorporated entropy) and against the older philosophical lineage (Cusa's docta ignorantia, Spinoza's attribute theory, Worrall's epistemic structural realism). The relationship to Hofstadter's strange-loop framework is developed in detail.
 
diff --git a/book/ch03-structural-realism.md b/book/ch03-structural-realism.md
index 7548977ef..d1d15b61d 100644
--- a/book/ch03-structural-realism.md
+++ b/book/ch03-structural-realism.md
@@ -55,7 +55,7 @@ The observer-admission reading at Level B sharpens this in two ways. First, it i
 
 A concrete example clarifies the structural distinction. Consider two finite cellular automata with the same configuration space size but different transition rules. The first has a Glauber dynamics on an Ising lattice at high temperature: fast mixing, short correlation times, no slow modes. The second has a topological phase-protected dynamics: slow modes from non-local order parameters, long correlation times in those modes, sufficient capacity in the topological sector. Both are well-defined deterministic systems on the same configuration space. In the coarse-grained description generated by its iterated one-step channel, with $\tau_S$ the observer's visible step, the first fails C2 in its timescale form, $\tau_S \ll \tau_B$, since it mixes fast relative to that step: writes into its hidden sector are scrambled before readback up to a suppressed residue, the absence of slow modes closing the thermal route to persistence; the second satisfies C1, C2, C3, C4 with appropriate partition choice. With C2 read in that timescale form, the first is observer-lacking; the second is observer-admitting. The distinction is structural — a property of the dynamics, not of the parameter values within a given dynamics — and applies prior to any specific universality-class commitment.
 
-This structural distinction has implications for several conceptual problems. The Boltzmann brain problem — the worry that a sufficiently large statistical-mechanical system at equilibrium would produce observers as random fluctuations more often than through evolutionary processes — is dissolved at Level B for C2 read in its timescale form. Under the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), and in the coarse-grained description generated by iterating the one-step effective channel ([Main §4.6]), equilibrium states are not observer-admitting in the sense of C1–C4 with C2 read in its timescale form, because that form, $\tau_S \ll \tau_B$, fails at equilibrium — its slow-bath route gone, no slow modes under that condition — and a bath that thermalizes between steps erodes the stored record before readback without establishing that none persists, suppressing the readback gap C4 requires exponentially in $\tau_S/\tau_B$ at every order $k \geq 2$ ([Main §2.3]). Boltzmann brains are not generic outcomes of long-time evolution that just happen to be rare; they are structurally excluded from the observer-admitting subset by C2 in that timescale form. The framework's account of the arrow of time, developed in Chapter 2 §2.6, follows from this exclusion: observers are restricted to the non-equilibrium phase of the substratum's dynamics, and the structural arrow of horizon growth provides the clock the observer's other arrows align with.
+This structural distinction has implications for several conceptual problems. The Boltzmann brain problem — the worry that a sufficiently large statistical-mechanical system at equilibrium would produce observers as random fluctuations more often than through evolutionary processes — is dissolved at Level B for C2 read in its timescale form. Under the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), and in the coarse-grained description generated by iterating the one-step effective channel ([Main §4.6]), equilibrium states are not observer-admitting in the sense of C1–C4 with C2 read in its timescale form, because that form, $\tau_S \ll \tau_B$, fails at equilibrium — its slow-bath route gone, no slow modes under that condition — and a bath that thermalizes between steps erodes the stored record before readback without establishing that none persists, suppressing the readback gap C4 requires exponentially in $\tau_S/\tau_B$ at every order $k \geq 2$ ([Main §2.3]). Boltzmann brains are not generic outcomes of long-time evolution that just happen to be rare; they are structurally excluded from the observer-admitting subset by C2 in that timescale form. The framework's treatment of the arrow of time, developed in Chapter 2 §2.6, starts from this exclusion: observers are restricted to the non-equilibrium phase of the substratum's dynamics. The exclusion is time-symmetric; on an expanding homogeneous general-relativistic background satisfying the null energy condition, the apparent-horizon area supplies a non-decreasing clock, and whether the observer's other arrows align with it is not established.
 
 The string-theory landscape problem looks different under the Level B reading as well. The conventional landscape picture posits roughly $10^{500}$ metastable vacua, each a different low-energy effective theory, with the vacuum-selection problem being to explain why our universe is in this particular vacuum. The Level B reading reframes this: most compactifications likely produce observer-lacking substrata at the relevant scales, because the slow-bath, sufficient-capacity, non-trivial-coupling conditions are non-generic. The SM-reproducing subset of the landscape is structurally smaller than the total landscape, because it is restricted to the observer-admitting subset. Whether this reframing fully resolves the landscape problem is an open technical question — it depends on whether observer-lacking compactifications can be characterized in string-theoretic terms, which requires substantive work in string-theoretic landscape physics — but the reframing is non-trivial.
 
@@ -199,7 +199,7 @@ Cusa is not the framework's source. The framework derives its claims from explic
 
 **Spinoza.** Spinoza's metaphysics is the deepest historical parallel to the framework's substratum-and-projections picture. The *Ethics* posits one substance — God or Nature — expressed through infinitely many attributes, each constituting the substance's essence under a different aspect. The two attributes humans can know are extension (the physical aspect) and thought (the mental aspect), but these are not two substances; they are the same substance expressed through different attributes, with no causal interaction between them because there is only one thing underneath. The doctrine that the order and connection of ideas is the same as the order and connection of things follows from this single-substance structure: the parallelism between mind and body is not an interaction but a consequence of their being the same underlying reality.
 
-The structural form of Spinoza's account — one substance, multiple incommensurable attributes, with the apparent multiplicity being a feature of how the substance is expressed rather than of the substance itself — maps directly onto the framework's three-projection structure. The substratum $(S, \varphi)$ is the single substance; quantum mechanics, general relativity, and the arrow of time are projections expressing the substratum's content under different aspects; the appearance of three distinct emergent theories is a feature of the projections rather than of the substratum. The framework's commitment to ontic structural realism on the **reconstructed local residue modulo** $\mathcal{G}_{\text{sub}}$ is Spinozistic in form: what is real is the gauge-invariant local structure, not any specific representative; the Bell-inclusive completion is not uniquely reconstructed.
+The structural form of Spinoza's account — one substance, multiple incommensurable attributes, with the apparent multiplicity being a feature of how the substance is expressed rather than of the substance itself — maps directly onto the framework's three-projection structure. The substratum $(S, \varphi)$ is the single substance; quantum mechanics and general relativity are projections expressing the substratum's content under different aspects, with a third bearing on the arrow of time; the appearance of three distinct emergent theories is a feature of the projections rather than of the substratum. The framework's commitment to ontic structural realism on the **reconstructed local residue modulo** $\mathcal{G}_{\text{sub}}$ is Spinozistic in form: what is real is the gauge-invariant local structure, not any specific representative; the Bell-inclusive completion is not uniquely reconstructed.
 
 The parallel is not complete. Spinoza's substance is unique (there is exactly one substance, by argument); the framework's substratum is a member of an equivalence class (there are many specific bijections, gauge-equivalent under $\mathcal{G}_{\text{sub}}$). Spinoza's attributes are infinite in number; the framework identifies three projections (quantum mechanics, general relativity, arrow of time) without claiming to enumerate all possible projections. The framework's claims are mathematical theorems rather than metaphysical doctrines. But the structural resemblance is closer than to any other position in the historical philosophical literature.
 
diff --git a/book/ch04-methodology.md b/book/ch04-methodology.md
index 78e83b7fa..6e201ed02 100644
--- a/book/ch04-methodology.md
+++ b/book/ch04-methodology.md
@@ -29,7 +29,7 @@ The benefit of the discipline is a clean audit trail. Every prediction the frame
 
 The cost of the discipline is that some natural extensions of the framework require derivational work that has not yet been done. The framework currently treats CP-violating phases in CKM and PMNS as solution-specific inputs rather than as derivations; the cosmological initial conditions are taken as given rather than derived; the specific values of fermion masses depend on one empirical input (the strange-quark mass) rather than being derived from the substratum alone. These are not failures of the framework but acknowledged scope limits. The methodological discipline distinguishes between "the framework predicts X" (X follows from the foundational premise plus the structural assumptions) and "the framework is consistent with X but does not predict it" (X is solution-specific or depends on inputs the framework has not yet derived).
 
-A useful diagnostic for the discipline is what the framework does *not* commit to. The framework does not commit to a specific interpretation of probability beyond standard Kolmogorov probability. The framework does not commit to a specific resolution of the measurement problem beyond the algorithmic-reading developed in Chapter 1. The framework does not commit to specific values of free parameters in the emergent quantum field theory beyond what the reconstruction theorem forces. The framework does not commit to a specific cosmological history beyond the structural arrow of horizon growth in the $\Lambda$-dominated phase. These non-commitments are part of the foundational economy: each additional commitment would expand the framework's claims, and the discipline is to make only those commitments that the foundational premise plus its derivations require.
+A useful diagnostic for the discipline is what the framework does *not* commit to. The framework does not commit to a specific interpretation of probability beyond standard Kolmogorov probability. The framework does not commit to a specific resolution of the measurement problem beyond the algorithmic-reading developed in Chapter 1. The framework does not commit to specific values of free parameters in the emergent quantum field theory beyond what the reconstruction theorem forces. The framework does not commit to a specific cosmological history beyond an expanding homogeneous general-relativistic background, taken as given, on which, with the null energy condition, the apparent-horizon area does not decrease. These non-commitments are part of the foundational economy: each additional commitment would expand the framework's claims, and the discipline is to make only those commitments that the foundational premise plus its derivations require.
 
 The single-foundational-commitment pattern distinguishes the framework from broader unification programs in a specific way. String theory commits to extended one-dimensional objects as the fundamental entities, plus supersymmetry, plus extra dimensions, plus a specific moduli structure. Loop quantum gravity commits to a graph-theoretic discrete geometry, plus a specific quantization procedure, plus a specific representation theory of the spatial diffeomorphism group. Causal sets commit to a Lorentz-invariant discrete causal structure, plus a specific dynamics for set growth. Each of these is a multi-commitment foundation, with each commitment adding to the framework's content. The framework here is a single-commitment foundation, with all content derived from the one premise plus its structural lemmas. Whether this is a methodological virtue or a methodological constraint depends on whether the derivations actually close — whether single-commitment foundations can produce the content that multi-commitment foundations produce by direct positing. The framework's claim is that they can; Parts II–IV develop this claim with the specific quantitative content.
 
diff --git a/book/ch18-beyond.md b/book/ch18-beyond.md
index 1e397116b..c06c0490a 100644
--- a/book/ch18-beyond.md
+++ b/book/ch18-beyond.md
@@ -187,7 +187,7 @@ The return itself is definite: the universe returns to its exact initial state a
 
 **Falsifiability.** The recurrence prediction is falsifiable only in the logical sense. For a substratum of given $|S|$ and $\varphi$, failure to recur within $g(|S|)$ steps would contradict a finite $S$ with bijection dynamics, and failure to recur within $|S| - 1$ steps would contradict only exact linearity (A5). But $|S|$ is not fixed by visible data: for any finite observation time $T$, a substratum with $|S| > T + 1$ is admissible, so no finite observation falsifies the framework as a whole, which would require an infinite observation time. An embedded observer's records are, moreover, part of $S$ and return with it. The recurrence is therefore a prediction at the structural level rather than at the empirical level: it follows necessarily from the framework's other commitments (finite substratum, bijection dynamics), and any framework consistent with those commitments must predict exact recurrence with a period at most $g(|S|)$ that divides $\mathrm{lcm}(1, \ldots, |S|)$, and below $|S|$ for the exactly linear rule.
 
-The conceptual content is significant. The framework predicts that the universe is *cyclic at the substratum level* — periodic with period $N$, the least common multiple of the bijection's cycle lengths. For a rule fixing the zero configuration on a nontrivial state space, no single cycle passes through every configuration. The exact linear rule of [SM §4.2] has this property and therefore at least two cycles. The cycle-ergodicity question that [SM §4.2] treats concerns sampling within an energy shell. Within a single cycle, the observer's experience is unidirectional (the arrow of time discussed in the next section), but at the substratum level the dynamics has no preferred direction. The framework's commitment to substratum cyclicity is one of the structural commitments that distinguishes it from frameworks treating the universe as either fundamentally non-cyclic (most cosmological models) or fundamentally infinite (most quantum gravity programs).
+The conceptual content is significant. The framework predicts that the universe is *cyclic at the substratum level* — periodic with period $N$, the least common multiple of the bijection's cycle lengths. For a rule fixing the zero configuration on a nontrivial state space, no single cycle passes through every configuration. The exact linear rule of [SM §4.2] has this property and therefore at least two cycles. The cycle-ergodicity question that [SM §4.2] treats concerns sampling within an energy shell. Within a single cycle no state function increases throughout; a direction of the observer's experience is fixed by boundary conditions on a stretch of the cycle (the arrow of time discussed in the next section), and at the substratum level the dynamics has no preferred direction. The framework's commitment to substratum cyclicity is one of the structural commitments that distinguishes it from frameworks treating the universe as either fundamentally non-cyclic (most cosmological models) or fundamentally infinite (most quantum gravity programs).
 
 **Implications for the framework's logical structure.** The recurrence theorem connects three structural features of the framework: substratum finiteness (the framework's commitment to $|S| < \infty$), bijection dynamics (the framework's commitment to $\varphi$ as a bijection), and P-indivisibility at the recurrence scale. The connection runs in three steps. Finiteness of $S$ guarantees exact recurrence ($\varphi^N = \mathrm{id}$). Recurrence together with C1 acting at the first step — which makes the one-step matrix a non-permutation under the uniform hidden prior, the cheapest non-permutation rooted map — gives returns of information from hidden to visible over the recurrence timescale (P-indivisibility across the recurrence cycle, not on accessible windows); the coupling is load-bearing rather than decorative, since for an uncoupled partition $\varphi = \varphi_V \times \varphi_H$ the visible sector evolves autonomously, its two-time family is divisible, and no distinguishability is ever restored — recurrence notwithstanding. With a full-support visible prior this indivisibility is equivalent to memory at some order below the recurrence time; the memory-bearing sector is the one in which C4 places that memory at an accessible order, and every finite law, memory-bearing or not, admits the quantum representation.
 
@@ -199,21 +199,21 @@ The framework's substratum dynamics is given by the bijection $\varphi: S \to S$
 
 This raises an obvious question. The universe we observe has an unambiguous arrow of time: entropy increases, biological systems age, memories form of the past but not the future, the universe expands rather than contracts. If the substratum dynamics is T-symmetric, where does the observed arrow come from?
 
-**The arrow as observer-emergent.** The framework's answer is that the arrow of time is *emergent from the observer's coarse-grained description*, not from the substratum dynamics. The substratum dynamics is T-symmetric; the observer's reduced description is not. The asymmetry is introduced by the trace-out — the marginalization over the hidden sector that produces the embedded observer's reduced description. The arrow of time is therefore a structural consequence of embedded observation, in the same way that quantum mechanics, the dark sector, and the Born rule are.
+**The arrow as observer-emergent.** The framework's answer is that the arrow of time is *emergent from the observer's coarse-grained description*, not from the substratum dynamics. The substratum dynamics is T-symmetric; the observer's reduced description of a non-equilibrium initial condition is not. The asymmetry enters through the trace-out of a non-equilibrium initial condition — the marginalization over the hidden sector that produces the embedded observer's reduced description; under the invariant measure the trace-out alone produces none. The arrow of time is therefore a feature of embedded observation of such a condition, in the same way that quantum mechanics, the dark sector, and the Born rule are.
 
 The mechanism is the standard Boltzmann argument applied to the framework's partition. The visible sector starts in a particular configuration that is, in some sense, "low entropy" relative to the typical configurations accessible at later times. The hidden sector starts in a configuration that contains all the correlations needed to specify the visible sector's future evolution. As the dynamics evolves, correlations flow from hidden sector to visible sector and back, with the time-reversibility of the underlying dynamics fully preserved at the substratum level. But the observer, who sees only the visible sector, sees an irreversible flow: information about the past is encoded in the visible sector's current state (entropy increasing), while information about the future is encoded in the hidden sector (inaccessible to the observer).
 
 The arrow of time is therefore *the observer's incompleteness* expressed temporally. The same partition that produces quantum randomness (by hiding the substratum information that determines outcomes) produces the arrow of time (by hiding the substratum information that would let the observer reverse the dynamics). Both are aspects of the same structural feature: the observer's reduced description is asymmetric in ways that the substratum dynamics is not.
 
-**Detailed balance and the thermodynamic arrow.** The framework's content on detailed balance from Chapter 6 §6.3 has a direct implication for the arrow of time. Detailed balance asserts that the transition probabilities $T_{ij}$ between any two states satisfy $T_{ij} = T_{ji}$ at all timescales — the rate of transition from $i$ to $j$ equals the rate of transition from $j$ to $i$. This is a *substratum-level* T-symmetry: the dynamics has no preferred direction.
+**Detailed balance and the thermodynamic arrow.** The framework's content on detailed balance from Chapter 6 §6.3 has a direct implication for the arrow of time. For a time reversal that preserves the partition, and the counting prior on hidden states, the visible transition probabilities satisfy reciprocity, $T_{ij} = T_{\theta j, \theta i}$ at all timescales, with $\theta$ the action of time reversal on visible states; when the visible states carry no T-odd data this is detailed balance, $T_{ij} = T_{ji}$ — the rate of transition from $i$ to $j$ equals the rate of transition from $j$ to $i$. This follows from *substratum-level* T-symmetry: the dynamics has no preferred direction.
 
 The thermodynamic arrow of time emerges not from any asymmetry in the transition probabilities but from the partition's *initial conditions*. The universe starts in a configuration that is far from equilibrium relative to the substratum's full state space, with most of the substratum's degrees of freedom on the hidden side of the partition and the visible sector occupying a small, structured region. As the dynamics evolves, the system moves toward equilibrium relative to the substratum's state space — entropy increasing in the observer's coarse-grained description — until the entire substratum is fully mixed. The arrow of time runs in the direction of this equilibration.
 
 The relation to detailed balance is that the H-theorem — the standard derivation of entropy increase from microscopic dynamics — is a theorem about the *coarse-grained* description, not about the underlying transition probabilities. The H-theorem assumes a particular kind of initial condition (low entropy at the coarse-grained level) and derives entropy increase from there. Detailed balance at the microscopic level is fully consistent with the H-theorem at the coarse-grained level, because the H-theorem is about how the observer's information about the system evolves rather than about the system's underlying dynamics.
 
-**The cosmological arrow.** The thermodynamic arrow and the cosmological arrow — the direction in which the universe expands — are aligned in the framework. The cosmological partition has a specific initial condition: the universe begins in a low-entropy configuration with most degrees of freedom on the hidden side of the cosmological horizon. Time evolution corresponds to the horizon entropy increasing toward equilibrium, with the boundary modes accumulating correlations and the hidden sector approaching maximum disorder.
+**The cosmological arrow.** Whether the thermodynamic arrow and the cosmological arrow — the direction in which the universe expands — are aligned is open in the framework: it is neither established nor refuted. The cosmological partition has a specific initial condition: the universe begins in a low-entropy configuration with most degrees of freedom on the hidden side of the cosmological horizon. On an expanding FRW background satisfying the null energy condition, the apparent-horizon area is non-decreasing. Its assigned entropy has that interpretation under the stated calibration assumptions; whether this behavior aligns with boundary correlations or hidden-sector disorder remains open.
 
-The alignment of the thermodynamic and cosmological arrows is structural rather than coincidental. Both arrows trace to the same initial conditions of the cosmological partition. The universe started with low entropy at the cosmological level; the framework's structural conditions C1–C4 hold at the cosmological boundary (Chapter 7 §7.2); the thermodynamic arrow within the visible sector tracks the cosmological arrow because the cosmological boundary is the primary partition for any embedded observer.
+Both arrows are traced to the same initial conditions of the cosmological partition. That is a common boundary condition, not a derived mechanism: that the thermodynamic arrow within the visible sector tracks the cosmological arrow is not established. The universe started with low entropy at the cosmological level, as an initial condition; at the cosmological boundary, coupling C1, persistence C2 and capacity C3 are verified, while history readback C4 is a named realization condition not presently discharged (Chapter 7 §7.2); the cosmological boundary is the primary partition for any embedded observer.
 
 **Why the framework reframes rather than explains.** The framework's content on the arrow of time is not an explanation of *why* the universe has a preferred direction; it is a *reframing* of the question. Standard accounts of the arrow of time treat it as an asymmetry that requires explanation — why is entropy increasing in the direction we call "future" rather than the direction we call "past"? The framework's answer is that there is no asymmetry in the substratum dynamics; the asymmetry is in the observer's reduced description, which inherits the initial conditions of the partition.
 
diff --git a/papers/Explainer.md b/papers/Explainer.md
index c185695b3..03b59f4f8 100644
--- a/papers/Explainer.md
+++ b/papers/Explainer.md
@@ -921,15 +921,15 @@ People ask: "Is the universe a computer?" The framework's answer: the universe i
 
 If the universe is a finite permutation, it must eventually cycle back: φ^N = id for some N. So where does the arrow of time come from? Why does entropy increase?
 
-The substratum itself has no arrow of time. φ and φ⁻¹ are equally valid. The total fine-grained entropy is exactly conserved. So the arrow, if there is one, has to come from somewhere else — specifically, from the observer.
+The substratum itself has no arrow of time. φ and φ⁻¹ are equally valid. The total fine-grained entropy is exactly conserved. So the arrow, if there is one, has to come from somewhere else — from the observer's description of the dynamics, given boundary conditions.
 
-A structural theorem (Main §4.6) establishes exactly that, under a stated mixing hypothesis for equilibrium regions and for the coarse-grained description generated by iterating the one-step effective channel. Observers satisfying C1–C4 — the conditions that make their description non-Markovian, quantum for the source's finite-configuration class — cannot exist in the equilibrium phase of (S, φ) when C2 is read in its timescale form, $\tau_S \ll \tau_B$. The reason is direct. C2 requires the written record to persist to readback — and in a thermal bath the record lives in correlations, so persistence essentially intact needs the hidden sector's correlation time $\tau_B$ to exceed the observer's measurement timescale $\tau_S$ (the conserve-and-route alternative being unavailable in an equilibrated bath). In equilibrium, local correlations decay on thermal timescales (a picosecond or so at ordinary temperatures), for a generic thermal bath (protected slow modes excepted), while any physical observer's internal processing time is many orders of magnitude longer. So in equilibrium, C2 fails in its timescale form ($\tau_S \ll \tau_B$) — and the readback C4 asks for is suppressed with it: a bath that thermalizes between steps erodes the stored record before it can be read, shrinking the readback gap exponentially in $\tau_S/\tau_B$ at every order $k \geq 2$ (the erasure lemma, [Main §2.3]). With C2 read in that timescale form, as one of the conditions C1–C4 that define the framework's observers, no observer satisfying that definition can exist there.
+A structural theorem (Main §4.6) constrains the observer side of this — where observers can exist, not which way time runs — under a stated mixing hypothesis for equilibrium regions and for the coarse-grained description generated by iterating the one-step effective channel. Observers satisfying C1–C4 — the conditions that make their description non-Markovian, quantum for the source's finite-configuration class — cannot exist in the equilibrium phase of (S, φ) when C2 is read in its timescale form, $\tau_S \ll \tau_B$. The reason is direct. C2 requires the written record to persist to readback — and in a thermal bath the record lives in correlations, so persistence essentially intact needs the hidden sector's correlation time $\tau_B$ to exceed the observer's measurement timescale $\tau_S$ (the conserve-and-route alternative being unavailable in an equilibrated bath). In equilibrium, local correlations decay on thermal timescales (a picosecond or so at ordinary temperatures), for a generic thermal bath (protected slow modes excepted), while any physical observer's internal processing time is many orders of magnitude longer. So in equilibrium, C2 fails in its timescale form ($\tau_S \ll \tau_B$) — and the readback C4 asks for is suppressed with it: a bath that thermalizes between steps erodes the stored record before it can be read, shrinking the readback gap exponentially in $\tau_S/\tau_B$ at every order $k \geq 2$ (the erasure lemma, [Main §2.3]). With C2 read in that timescale form, as one of the conditions C1–C4 that define the framework's observers, no observer satisfying that definition can exist there.
 
 This has two striking consequences. First, for C2 read in its timescale form, the "Boltzmann brain" problem dissolves at the level of definitions. A Boltzmann brain is by construction a local fluctuation out of a thermally equilibrated environment — that's what "fluctuation from equilibrium" means. Its hidden sector is therefore the thermally equilibrated bath around it, with τ_B bounded by thermal decorrelation — too short for C2's timescale form to hold, by six or more orders of magnitude for any brain-like observer when the bath's correlation time is set by the thermal scale (at most about 10⁻¹² s at temperatures that support structured matter, against an internal processing time of at least about 10⁻⁶ s), a generic thermal estimate that fails for protected slow modes. In the coarse-grained description generated by iterating the one-step effective channel, Boltzmann brains carry records eroded before readback and at most exponentially suppressed readback at orders beyond the first — though their statistics, like any finite law's, admit the representation; they aren't observers in the sense of C1–C4 with C2 read in its timescale form. The exclusion is structural, not anthropic — it doesn't depend on any argument about "typical observers" or the trustworthiness of memory.
 
-Second, the "past hypothesis" of cosmology gets replaced by a conditional theorem. Every standard cosmology has a mystery at t=0: why was the early universe so low-entropy? Boltzmann, Penrose, Albert, and Carroll all treat this as a postulate that has to be assumed separately. For the job of selecting where observers can exist, the framework replaces the postulate with a conditional derivation. You don't need to posit low initial entropy to explain why we see it; you need the definition of observation together with the mixing hypothesis for equilibrium regions, which select for the non-equilibrium phase. Observers exist where C1–C4 hold. Under that hypothesis, and in the coarse-grained description generated by iterating the one-step effective channel, C1–C4, with C2 read in its timescale form, don't hold in equilibrium regions that mix fast compared with the observer's own timescale; there, record persistence and readback are suppressed, not ruled out. Therefore, in that formulation, observers exist in non-equilibrium. The mixing hypothesis is the premise that takes the place of the postulate.
+Second, one role of the "past hypothesis" of cosmology — selecting where observers can exist — gets taken over by a conditional theorem. Every standard cosmology has a mystery at t=0: why was the early universe so low-entropy? Boltzmann, Penrose, Albert, and Carroll all treat this as a postulate that has to be assumed separately. For the job of selecting where observers can exist, the framework replaces the postulate with a conditional derivation. For that job you don't need to posit low initial entropy; you need the definition of observation together with the mixing hypothesis for equilibrium regions, which select for the non-equilibrium phase. The selection is time-symmetric: it does not explain why entropy is low at one end of the observed history, which remains a boundary condition. Observers exist where C1–C4 hold. Under that hypothesis, and in the coarse-grained description generated by iterating the one-step effective channel, C1–C4, with C2 read in its timescale form, don't hold in equilibrium regions that mix fast compared with the observer's own timescale; there, record persistence and readback are suppressed, not ruled out. Therefore, in that formulation, observers exist in non-equilibrium. The mixing hypothesis is the premise that takes the place of the postulate in that role.
 
-In our universe, the non-equilibrium phase is provided by the expanding cosmological horizon. The horizon's expansion sets $\tau_B \sim H^{-1}$ — about $10^{17}$ seconds — which is structurally (not thermally) generated and is vastly longer than any laboratory timescale. This is why we see the memory-bearing quantum phenomenology at all: we're in the one phase of the universe where the framework's observer-conditions, with C2 read in its timescale form, can be met — representability itself is universal; what the phase selects is the nontrivial sector. Entropy increase, the Second Law, and the felt direction of time emerge as coarse-grained descriptions of the dynamics accessible to observers in this non-equilibrium phase. Like quantum mechanics itself, the direction of time is emergent — a consequence of observing a lossless cycle from within, during the phase when such observation is structurally possible.
+In our universe, the non-equilibrium phase is associated with the expanding cosmological horizon — an expansion that is observed, not derived. The horizon's expansion sets $\tau_B \sim H^{-1}$ — about $10^{17}$ seconds — which is structurally (not thermally) generated and is vastly longer than any laboratory timescale. This is why we see the memory-bearing quantum phenomenology at all: we're in the one phase of the universe where the framework's observer-conditions, with C2 read in its timescale form, can be met — representability itself is universal; what the phase selects is the nontrivial sector. Entropy increase, the Second Law, and the felt direction of time emerge as coarse-grained descriptions of the dynamics accessible to observers in this non-equilibrium phase, given a low-entropy boundary condition that the selection rule does not supply; whether those arrows and the expansion share a common mechanism is open. Like quantum mechanics itself, the direction of time is emergent — a feature of observing a lossless cycle from within — but which way it points is fixed by boundary conditions, not derived.
 
 ### The incompleteness family
 
diff --git a/papers/Explainer.pdf b/papers/Explainer.pdf
index b40b4721a..446e088e0 100644
Binary files a/papers/Explainer.pdf and b/papers/Explainer.pdf differ
diff --git a/papers/Explainer.tex b/papers/Explainer.tex
index b02bba3a4..546d68e7a 100644
--- a/papers/Explainer.tex
+++ b/papers/Explainer.tex
@@ -3503,10 +3503,11 @@ \subsubsection{The arrow of time}\label{the-arrow-of-time}}
 
 The substratum itself has no arrow of time. φ and φ⁻¹ are equally valid.
 The total fine-grained entropy is exactly conserved. So the arrow, if
-there is one, has to come from somewhere else --- specifically, from the
-observer.
+there is one, has to come from somewhere else --- from the observer's
+description of the dynamics, given boundary conditions.
 
-A structural theorem (Main §4.6) establishes exactly that, under a
+A structural theorem (Main §4.6) constrains the observer side of this
+--- where observers can exist, not which way time runs --- under a
 stated mixing hypothesis for equilibrium regions and for the
 coarse-grained description generated by iterating the one-step effective
 channel. Observers satisfying C1--C4 --- the conditions that make their
@@ -3551,38 +3552,44 @@ \subsubsection{The arrow of time}\label{the-arrow-of-time}}
 exclusion is structural, not anthropic --- it doesn't depend on any
 argument about ``typical observers'' or the trustworthiness of memory.
 
-Second, the ``past hypothesis'' of cosmology gets replaced by a
-conditional theorem. Every standard cosmology has a mystery at t=0: why
-was the early universe so low-entropy? Boltzmann, Penrose, Albert, and
-Carroll all treat this as a postulate that has to be assumed separately.
-For the job of selecting where observers can exist, the framework
-replaces the postulate with a conditional derivation. You don't need to
-posit low initial entropy to explain why we see it; you need the
-definition of observation together with the mixing hypothesis for
-equilibrium regions, which select for the non-equilibrium phase.
-Observers exist where C1--C4 hold. Under that hypothesis, and in the
-coarse-grained description generated by iterating the one-step effective
-channel, C1--C4, with C2 read in its timescale form, don't hold in
-equilibrium regions that mix fast compared with the observer's own
-timescale; there, record persistence and readback are suppressed, not
-ruled out. Therefore, in that formulation, observers exist in
-non-equilibrium. The mixing hypothesis is the premise that takes the
-place of the postulate.
-
-In our universe, the non-equilibrium phase is provided by the expanding
-cosmological horizon. The horizon's expansion sets
-\(\tau_B \sim H^{-1}\) --- about \(10^{17}\) seconds --- which is
-structurally (not thermally) generated and is vastly longer than any
-laboratory timescale. This is why we see the memory-bearing quantum
-phenomenology at all: we're in the one phase of the universe where the
-framework's observer-conditions, with C2 read in its timescale form, can
-be met --- representability itself is universal; what the phase selects
-is the nontrivial sector. Entropy increase, the Second Law, and the felt
-direction of time emerge as coarse-grained descriptions of the dynamics
-accessible to observers in this non-equilibrium phase. Like quantum
-mechanics itself, the direction of time is emergent --- a consequence of
-observing a lossless cycle from within, during the phase when such
-observation is structurally possible.
+Second, one role of the ``past hypothesis'' of cosmology --- selecting
+where observers can exist --- gets taken over by a conditional theorem.
+Every standard cosmology has a mystery at t=0: why was the early
+universe so low-entropy? Boltzmann, Penrose, Albert, and Carroll all
+treat this as a postulate that has to be assumed separately. For the job
+of selecting where observers can exist, the framework replaces the
+postulate with a conditional derivation. For that job you don't need to
+posit low initial entropy; you need the definition of observation
+together with the mixing hypothesis for equilibrium regions, which
+select for the non-equilibrium phase. The selection is time-symmetric:
+it does not explain why entropy is low at one end of the observed
+history, which remains a boundary condition. Observers exist where
+C1--C4 hold. Under that hypothesis, and in the coarse-grained
+description generated by iterating the one-step effective channel,
+C1--C4, with C2 read in its timescale form, don't hold in equilibrium
+regions that mix fast compared with the observer's own timescale; there,
+record persistence and readback are suppressed, not ruled out.
+Therefore, in that formulation, observers exist in non-equilibrium. The
+mixing hypothesis is the premise that takes the place of the postulate
+in that role.
+
+In our universe, the non-equilibrium phase is associated with the
+expanding cosmological horizon --- an expansion that is observed, not
+derived. The horizon's expansion sets \(\tau_B \sim H^{-1}\) --- about
+\(10^{17}\) seconds --- which is structurally (not thermally) generated
+and is vastly longer than any laboratory timescale. This is why we see
+the memory-bearing quantum phenomenology at all: we're in the one phase
+of the universe where the framework's observer-conditions, with C2 read
+in its timescale form, can be met --- representability itself is
+universal; what the phase selects is the nontrivial sector. Entropy
+increase, the Second Law, and the felt direction of time emerge as
+coarse-grained descriptions of the dynamics accessible to observers in
+this non-equilibrium phase, given a low-entropy boundary condition that
+the selection rule does not supply; whether those arrows and the
+expansion share a common mechanism is open. Like quantum mechanics
+itself, the direction of time is emergent --- a feature of observing a
+lossless cycle from within --- but which way it points is fixed by
+boundary conditions, not derived.
 
 \hypertarget{the-incompleteness-family}{%
 \subsubsection{The incompleteness
@@ -4065,4 +4072,4 @@ \subsection{Appendix: Key Mathematical
 the author.
 
 \end{document}
-% source-sha256: 586174960112bcf83aaaf9889bfc15c4f265d9b9e3532e3dbb5e32a99eb9dd16
+% source-sha256: 5e79e7d1c0993c3d287e15135dac8dd04765eb263e4c33de4a211f1fcfa0411e
diff --git a/papers/GR.md b/papers/GR.md
index 1595cd6af..676a789c3 100644
--- a/papers/GR.md
+++ b/papers/GR.md
@@ -29,7 +29,7 @@ Three epistemic tiers are worth distinguishing before the technical development
 
 The tier system maps to the framework's broader hierarchical structure articulated in [Structure §2]. *Tier 1* — the energy-shell horizon temperature, the finite horizon counting, and the two-level distinction — operates at Level C × Level G4: these results hold for any partial-trace observational system on a horizon-like causal partition, not specifically for OI's cubic-lattice substratum. They are universality-class-level structural results, and the conditional results built on them ($\hbar$, $\epsilon = 2\sqrt{\sigma}\,l_p$, the $1/4$ coefficient) are universality-class-level in the same sense once their named conditions are granted, since those conditions concern the horizon and the emergent dynamics rather than the lattice (§8.5). *Tier 2* — the dark-sector phenomenology — sits at Level D × Levels G1–G2, deriving from the thermodynamic Clausius relation applied to the specific boundary reservoir of OI's universality-class representative. *Tier 3* — solution-specific quantities — is at the bijection-specific level below Level D, depending on the particular $\varphi$ within $[(S, \varphi)]/\mathcal{G}_{\text{sub}}$. This stratification clarifies what a conditional universality-class result claims: not "OI predicts $\hbar$ uniquely" but "any horizon-bounded embedded-observer system meeting the named conditions gets this $\hbar$, and OI provides one specific realization."
 
-The paper is organized as follows. §2 identifies the cosmological horizon as a causal partition and verifies the structural conditions. §3 derives the emergent action scale $\hbar$ in four steps: the classical horizon temperature, boundary-only dependence, dimensional determination, and thermal self-consistency. §4 fixes the discreteness scale $\epsilon = 2\sqrt{\sigma}\,l_p$. §5 derives the Bekenstein-Hawking entropy and discusses the GW250114 confirmation. §6 gives the conditional dissolution of the cosmological constant problem. §7 presents the quantitative predictions for dark energy, the dark sector, and dark matter, with a summary table in §7.4. §8 discusses scope, the status of the discreteness scale, vacuum energy and the Higgs potential, and the framework's falsifiability conditions. §9 concludes. The cubic-lattice realization of the substratum — which fixes the Standard Model gauge structure via the wave equation — is developed in a companion paper [SM] and is independent of the GR results presented here. The reconstruction theorem and the synthesis claim that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same $(S, \varphi)$ are developed in [Substratum].
+The paper is organized as follows. §2 identifies the cosmological horizon as a causal partition and verifies the structural conditions. §3 derives the emergent action scale $\hbar$ in four steps: the classical horizon temperature, boundary-only dependence, dimensional determination, and thermal self-consistency. §4 fixes the discreteness scale $\epsilon = 2\sqrt{\sigma}\,l_p$. §5 derives the Bekenstein-Hawking entropy and discusses the GW250114 confirmation. §6 gives the conditional dissolution of the cosmological constant problem. §7 presents the quantitative predictions for dark energy, the dark sector, and dark matter, with a summary table in §7.4. §8 discusses scope, the status of the discreteness scale, vacuum energy and the Higgs potential, and the framework's falsifiability conditions. §9 concludes. The cubic-lattice realization of the substratum — which fixes the Standard Model gauge structure via the wave equation — is developed in a companion paper [SM] and is independent of the GR results presented here. The reconstruction theorem and the synthesis claim that quantum mechanics and general relativity descend as projections of the same $(S, \varphi)$, with a third projection bearing on the arrow of time, are developed in [Substratum].
 
 ***
 
@@ -211,7 +211,7 @@ for the coherent completions that exist, subject to the stated carrier, locality
 
 $$\boxed{\;\mathrm{OI} \;+\; \text{coherent completion} \;+\; \text{positive thermodynamic orientation} \;\Longrightarrow\; \mathrm{QM}\;}$$
 
-up to ordinary unitary gauge. Two substantive questions accordingly remain: whether the actual OI substratum lies in the class admitting an OI-compatible coherent completion, and whether its derived classical thermodynamic arrow — $\beta_E = \partial S_H/\partial E > 0$, settled classically by exact counting (kernel: `counting_passive`) — enters that completion through an orientation-sensitive bridge. The classification of alternative coherent theories is otherwise closed: existing completions are ordinary complex quantum mechanics modulo one global antiunitary $\mathbb{Z}_2$, while some nominal carriers admit no coherent completion at all.
+up to ordinary unitary gauge. Two substantive questions accordingly remain: whether the actual OI substratum lies in the class admitting an OI-compatible coherent completion, and whether its classical thermodynamic orientation — the energy ordering under which exact finite-bath counting with $\beta_E = \partial S_H/\partial E > 0$ makes the classical profile passive (kernel: `counting_passive`, from a monotone bath count), an orientation of energy rather than a direction of time — enters that completion through an orientation-sensitive bridge. The classification of alternative coherent theories is otherwise closed: existing completions are ordinary complex quantum mechanics modulo one global antiunitary $\mathbb{Z}_2$, while some nominal carriers admit no coherent completion at all.
 
 **The operational-completion characterization.** The classification above is a uniqueness statement conditional on existence. The existence side — which operational theories generated on a finite carrier are exactly quantum mechanics at the level of instruments and composites — is settled by a characterization with no premise beyond its stated setting. Fix a nonempty finite visible system with configuration set $A$. The setting takes complex matrix kinematics as given: an operation on the system is a complex-linear map on the complex matrices $M_A(\mathbb{C})$ indexed by $A$, and an operation at ancilla level $n$ is a complex-linear map on $M_{A \times [n]}(\mathbb{C}) \cong M_A(\mathbb{C}) \otimes M_n(\mathbb{C})$, $[n] = \{1, \ldots, n\}$, the matrices indexed by a configuration paired with one of $n$ ancilla states. A derivation of this kinematics from operational premises that presuppose neither a matrix representation nor the complex field is open. An operational theory on the system specifies which finite outcome families of such maps are available on the system and at every finite ancilla level, and which preparation maps $M_A(\mathbb{C}) \to M_{A \times [n]}(\mathbb{C})$ are available at each level, subject to six rules (kernel: `FiniteOperationalTheory` in `OIBridge/OperationalAssembly.lean`): (a) the identity is available on the system; (b) coarse-graining the outcome labels of an available family gives an available family, on the system and at every level; (c) at every ancilla level, an available family followed by available families chosen by its outcome is available, with the pair of outcomes as the joint outcome — a composition rule stated at the ancilla levels and not on the system; (d) attaching a uniformly mixed ancilla, $\rho \mapsto \rho \otimes \mathbb{1}_n/n$, is an available preparation for every $n \geq 1$, and an available preparation followed by an available single-outcome operation at its level is an available preparation; (e) at every level $n$ an available $n$-outcome family reads the ancilla, its $k$-th branch sending every product $X \otimes Y$ to $X \otimes P_k Y P_k$, with $P_k$ the projector on the $k$-th ancilla state, a spectator-independence requirement from which the branch's form on every matrix follows (kernel: `readout_is_localLuders`); (f) an available preparation, followed by an available family at its level and then by discarding the ancilla through the partial trace, induces an available family on the system. No rule extends a system operation to a composite, and no rule imposes positivity or trace preservation; those enter through the conditions. The OI⁺, primitive-source, substratum-source, layer-flow and typed forms below are stated over the same kinematics, the typed form over the complex matrices of every finite carrier. Five conditions are in play. (i) *Valid probabilities*: at every ancilla level, every available family sends positive states to positive states and its outcomes' traces sum to the input trace. (ii) *Trivial-ancilla consistency*: an operation available on the system is available on the system with a one-state ancilla adjoined. (iii) *Inert spectators*: an untouched independent system can be adjoined without changing an intervention. (iv) *Full reversible control*: every finite unitary on every composite is available. (v) *Iterated composition*: a composite may itself serve as the working system of a larger experiment — an operation obtained by attaching a fresh uniformly prepared ancilla to a composite, acting on the enlarged carrier, and discarding the ancilla is available on the composite.
 
@@ -558,7 +558,7 @@ The gravitational sector's dark-energy content is geometry-dependent bookkeeping
 
 ## 8. Discussion
 
-The foundational emergent QM result is treated as established in [Main] and not re-derived here. The cubic-lattice realization of the substratum — which determines the Standard Model gauge structure via the wave equation — is independent of the present results and is developed in [SM]. The reconstruction theorem, the substratum gauge group, and the synthesis claim that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same $(S, \varphi)$ are developed in [Substratum].
+The foundational emergent QM result is treated as established in [Main] and not re-derived here. The cubic-lattice realization of the substratum — which determines the Standard Model gauge structure via the wave equation — is independent of the present results and is developed in [SM]. The reconstruction theorem, the substratum gauge group, and the synthesis claim that quantum mechanics and general relativity descend as projections of the same $(S, \varphi)$, with a third projection bearing on the arrow of time, are developed in [Substratum].
 
 ### 8.1 Scope
 
@@ -604,7 +604,7 @@ The falsifiability content is stratified by hierarchy level (per [Structure §2.
 
 *(i) Consistency of nested trace-outs (Appendix A, Theorems A.1–A.4).* Two procedures — tracing out the trans-cosmological region and the black hole interior simultaneously (direct) or sequentially (first cosmological, then black hole) — produce the same CPTP channel and the same $\hbar$. Hamiltonian reconstruction from the transition data is conditional on H-BG and the §3.3 hypotheses; for exactly unitary channel families, the generators differ only by an energy shift (Corollary A.3). The key result (Theorem A.2) is that the quantum partial trace within the emergent description agrees with the classical marginalization on the substratum: the emergent QM is self-consistent.
 
-*(ii) The generalized second law (Appendix A, §A.6), open.* Unitarity of $\varphi$ and subadditivity make the sum of the von Neumann entropies of $V$, $B$ and $D$ non-decreasing from a product initial state (Theorem A.6). That sum is $S_{\text{matter}} + S_{\text{BH}} + S_{\text{dS}}$ only if each horizon's Bekenstein-Hawking entropy equals the von Neumann entropy of its sector's reduced state, an identification the boundary-mode count does not supply.
+*(ii) The generalized second law (Appendix A, §A.6), open.* Unitarity of $\varphi$ and subadditivity bound the sum of the von Neumann entropies of $V$, $B$ and $D$ below by its value at a product initial state, at every earlier and later time (Theorem A.6); the bound is two-sided and does not make the sum monotone. That sum is $S_{\text{matter}} + S_{\text{BH}} + S_{\text{dS}}$ only if each horizon's Bekenstein-Hawking entropy equals the von Neumann entropy of its sector's reduced state, an identification the boundary-mode count does not supply.
 
 *(iii) The Page curve (Appendix A, Theorems A.7–A.9), conditional.* Under the scrambling condition H-scramble and the decoupling condition H-decouple of Appendix A, with the energy shell of $B \cup R$ the full product space $\mathcal{H}_B \otimes \mathcal{H}_R$, the entanglement entropy of the radiation of an evaporating black hole, averaged over full cycles of the dynamics at the bipartition of stage $t$, follows $\overline{S}_{\text{rad}}(t) = \min(n_R(t), n_B(t)) \cdot \ln q$ up to a correction of at most $\tfrac12$, reached at $n_R = n_B$ and exponentially small in $|n_R - n_B|$ away from it (Theorem A.9), where $n_B(t)$ is the shrinking boundary mode count and $n_R(t) = n_B(0) - n_B(t)$; whether the entropy along a single trajectory follows the same curve is open. Unitarity is preserved (bijectivity of $\varphi$); at $\beta = 1$, the radiation gaining one mode's state-space entropy per boundary mode lost (§A.7), the turnover occurs at the Page time $t_P \approx 0.646 \, t_{\text{evap}}$ (in general $t_P/t_{\text{evap}} = 1 - [\beta/(1+\beta)]^{3/2}$; independent of mass and of the emission coefficient when that coefficient is mass-independent). Theorem A.8 derives the Page value as a cycle average from H-scramble, which carries measure concentration [29, 61] to the orbit, the cycle decomposition of the finite configuration space supplying exact time averages. It counts initial configurations, with the bipartition aligned to the configuration coordinates, and there H-scramble fails for every bijection: every state on the orbit is a product state and the radiation's entropy is zero, so Theorems A.8–A.9 as stated have no instance. Under a bijection the radiation's entropy at an aligned bipartition never exceeds the entropy of the initial state's distribution over configurations (§A.7), so a Page value requires an initial superposition whose distribution over configurations has entropy close to the Page value or above it, or a bipartition not aligned with the configuration coordinates; neither is specified, and H-scramble is not established (uniformity over classical microstates is not uniformity over quantum pure states). H-decouple is not supplied by timescale ordering. On a proper energy shell the value differs in general, and the corresponding statement is open. The greybody factor is solution-specific.
 
@@ -706,7 +706,7 @@ The cosmological horizon realizes the embedded-observation framework of [Main],
 
 The empirical payoff is the set of predictions in §7. The dark-energy content is geometry-dependent bookkeeping whose smooth late-time expansion contains an $H^2 - H_0^2$ term shared by every smooth model, read in the Type II running-vacuum channel — a structural reading, not a derived law (§7.1): consistent with the Bertini et al. (2025) direct fit to DESI DR2 + DES-Y5 + Planck 2018 [15] (which reports $\nu$ consistent with zero at $2\sigma$), and decisively testable by DESI Year 5 and subsequent stage-IV surveys. The coefficient's magnitude is the self-consistent emergent-QFT value $|\nu_{\rm QFT}| \sim 10^{-32}$ (conditional on the §7.3 split, not yet established at theorem level), observationally indistinguishable from zero. Substratum contributions beyond this are constrained by the boundary mode count of §3.1 and §5 — spatial location alone, with no inherent frequency assignment — to be well below $|\nu_{\rm QFT}|$ under the natural 2D distribution. The framework's structural prediction is therefore the Type II channel with $|\nu| \approx |\nu_{\rm QFT}|$ and $w(z) \approx -1$. The dark-sector corollary explains ~95% of the gravitational budget of the universe without new fundamental particles, and the absence of direct dark-matter detection at the LZ and XENONnT experiments is interpreted within the framework as a structural consequence rather than a null result to be explained away. The MOND phenomenology emerges — at the G1 register, with the covariant completion (§7) the standing conditional — from entropy displacement at the cosmological boundary with $a_0 = cH/6$ (the factor $1/6$ imported from Verlinde), reproducing the baryonic Tully-Fisher relation, with rotation curves approaching flat at $r > r_M$. The Keplerian decline Jiao et al. (2023) observed over $19$–$26.5$ kpc, beyond the $9$–$16$ kpc Milky Way crossover, and the declining outer velocities Genzel et al. detect at $z = 0.9$–$2.4$, where at a fixed baryonic profile the $H(z)$ scaling of $a_0$ predicts flatter, less baryon-dominated curves, are in tension with the law (§7.3); JWST and ELT kinematics test the high-redshift prediction at higher precision.
 
-The foundational framework developed in [Main] and the cubic-lattice realization developed in [SM] are complementary to the present work: [Main] establishes a finite-horizon correspondence between the recordable laws of an embedded observer and fixed-basis quantum laws, the present paper applies it to the cosmological horizon, and the SM paper applies it to the microscopic lattice substratum. The reconstruction theorem and the synthesis claim — that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same deterministic $(S, \varphi)$ — are developed in [Substratum]. Together, this four-paper synthesis ([Main], [SM], [GR], [Substratum]) proposes that the Standard Model, general relativity, and the arrow of time descend from one finite deterministic construction, with the covariant field equations (G3) and the selection of full operational quantum mechanics open. The framework's hierarchical structural realism and its relationship to other unification programs are developed in the fifth core paper [Structure].
+The foundational framework developed in [Main] and the cubic-lattice realization developed in [SM] are complementary to the present work: [Main] establishes a finite-horizon correspondence between the recordable laws of an embedded observer and fixed-basis quantum laws, the present paper applies it to the cosmological horizon, and the SM paper applies it to the microscopic lattice substratum. The reconstruction theorem and the synthesis claim — that quantum mechanics and general relativity descend as projections of the same deterministic $(S, \varphi)$, with a third projection bearing on the arrow of time — are developed in [Substratum]. Together, this four-paper synthesis ([Main], [SM], [GR], [Substratum]) proposes that the Standard Model and general relativity descend from one finite deterministic construction, with the covariant field equations (G3) and the selection of full operational quantum mechanics open. The framework's hierarchical structural realism and its relationship to other unification programs are developed in the fifth core paper [Structure].
 
 ***
 
diff --git a/papers/GR.pdf b/papers/GR.pdf
index 2ab0ae597..b29f0caa1 100644
Binary files a/papers/GR.pdf and b/papers/GR.pdf differ
diff --git a/papers/GR.tex b/papers/GR.tex
index 763a1d783..3ea8345c1 100644
--- a/papers/GR.tex
+++ b/papers/GR.tex
@@ -453,9 +453,9 @@ \subsection{1. Introduction}\label{introduction}}
 Standard Model gauge structure via the wave equation --- is developed in
 a companion paper {[}SM{]} and is independent of the GR results
 presented here. The reconstruction theorem and the synthesis claim that
-quantum mechanics, general relativity, and the arrow of time descend as
-three projections of the same \((S, \varphi)\) are developed in
-{[}Substratum{]}.
+quantum mechanics and general relativity descend as projections of the
+same \((S, \varphi)\), with a third projection bearing on the arrow of
+time, are developed in {[}Substratum{]}.
 
 \begin{center}\rule{0.5\linewidth}{0.5pt}\end{center}
 
@@ -1465,10 +1465,12 @@ \subsubsection{3.3 Gauge-fixing and the dimensional
 
 up to ordinary unitary gauge. Two substantive questions accordingly
 remain: whether the actual OI substratum lies in the class admitting an
-OI-compatible coherent completion, and whether its derived classical
-thermodynamic arrow --- \(\beta_E = \partial S_H/\partial E > 0\),
-settled classically by exact counting (kernel:
-\texttt{counting\_passive}) --- enters that completion through an
+OI-compatible coherent completion, and whether its classical
+thermodynamic orientation --- the energy ordering under which exact
+finite-bath counting with \(\beta_E = \partial S_H/\partial E > 0\)
+makes the classical profile passive (kernel: \texttt{counting\_passive},
+from a monotone bath count), an orientation of energy rather than a
+direction of time --- enters that completion through an
 orientation-sensitive bridge. The classification of alternative coherent
 theories is otherwise closed: existing completions are ordinary complex
 quantum mechanics modulo one global antiunitary \(\mathbb{Z}_2\), while
@@ -3660,9 +3662,10 @@ \subsection{8. Discussion}\label{discussion}}
 substratum --- which determines the Standard Model gauge structure via
 the wave equation --- is independent of the present results and is
 developed in {[}SM{]}. The reconstruction theorem, the substratum gauge
-group, and the synthesis claim that quantum mechanics, general
-relativity, and the arrow of time descend as three projections of the
-same \((S, \varphi)\) are developed in {[}Substratum{]}.
+group, and the synthesis claim that quantum mechanics and general
+relativity descend as projections of the same \((S, \varphi)\), with a
+third projection bearing on the arrow of time, are developed in
+{[}Substratum{]}.
 
 \hypertarget{scope}{%
 \subsubsection{8.1 Scope}\label{scope}}
@@ -3939,9 +3942,10 @@ \subsubsection{8.4 Assumptions, limitations, and
 marginalization on the substratum: the emergent QM is self-consistent.
 
 \emph{(ii) The generalized second law (Appendix A, §A.6), open.}
-Unitarity of \(\varphi\) and subadditivity make the sum of the von
-Neumann entropies of \(V\), \(B\) and \(D\) non-decreasing from a
-product initial state (Theorem A.6). That sum is
+Unitarity of \(\varphi\) and subadditivity bound the sum of the von
+Neumann entropies of \(V\), \(B\) and \(D\) below by its value at a
+product initial state, at every earlier and later time (Theorem A.6);
+the bound is two-sided and does not make the sum monotone. That sum is
 \(S_{\text{matter}} + S_{\text{BH}} + S_{\text{dS}}\) only if each
 horizon's Bekenstein-Hawking entropy equals the von Neumann entropy of
 its sector's reduced state, an identification the boundary-mode count
@@ -4502,13 +4506,13 @@ \subsection{9. Conclusion}\label{conclusion}}
 the present paper applies it to the cosmological horizon, and the SM
 paper applies it to the microscopic lattice substratum. The
 reconstruction theorem and the synthesis claim --- that quantum
-mechanics, general relativity, and the arrow of time descend as three
-projections of the same deterministic \((S, \varphi)\) --- are developed
-in {[}Substratum{]}. Together, this four-paper synthesis ({[}Main{]},
-{[}SM{]}, {[}GR{]}, {[}Substratum{]}) proposes that the Standard Model,
-general relativity, and the arrow of time descend from one finite
-deterministic construction, with the covariant field equations (G3) and
-the selection of full operational quantum mechanics open. The
+mechanics and general relativity descend as projections of the same
+deterministic \((S, \varphi)\), with a third projection bearing on the
+arrow of time --- are developed in {[}Substratum{]}. Together, this
+four-paper synthesis ({[}Main{]}, {[}SM{]}, {[}GR{]}, {[}Substratum{]})
+proposes that the Standard Model and general relativity descend from one
+finite deterministic construction, with the covariant field equations
+(G3) and the selection of full operational quantum mechanics open. The
 framework's hierarchical structural realism and its relationship to
 other unification programs are developed in the fifth core paper
 {[}Structure{]}.
@@ -5472,4 +5476,4 @@ \subsection{References}\label{references}}
 arXiv:quant-ph/0407049.
 
 \end{document}
-% source-sha256: 48a41c9c8208192c233d02f9809eed1879d2ae7f4c359d9062d240f83a26cca7
+% source-sha256: 7304c21382c6f0bd3e77784b59fbddf7aff6a04645e99f9ca0929409873f3b7d
diff --git a/papers/Main.md b/papers/Main.md
index 14078969d..0f2849e77 100644
--- a/papers/Main.md
+++ b/papers/Main.md
@@ -741,7 +741,7 @@ Laboratory tests of the characterization theorem may be possible in tabletop sys
 
 ### 4.6 Structural Exclusion of Equilibrium-Phase Observers
 
-An immediate corollary of (C1)–(C4) is a structural selection rule on where in a finite bijective $(S, \varphi)$ observers can exist. The rule replaces a postulate often imported from elsewhere — the "past hypothesis" of low initial entropy, in its role of selecting observer-compatible configurations — with a theorem conditional on the effective-mixing hypothesis (EM) below and on the rate condition $\tau_{\rm mix}\ll\tau_S$ stated in the past-hypothesis corollary; under (EM) it is derived from the framework's own definitions, with (C2) read in its timescale form (C2-slow) — record persistence and readback being suppressed there, not excluded — for the coarse-grained description that the effective channel generates (Scope (ii)), and without (EM) it is a conjectural application.
+An immediate corollary of (C1)–(C4) is a structural selection rule on where in a finite bijective $(S, \varphi)$ observers can exist. The rule replaces a postulate often imported from elsewhere — the "past hypothesis" of low initial entropy, in its role of selecting observer-compatible configurations — with a theorem conditional on the effective-mixing hypothesis (EM) below and on the rate condition $\tau_{\rm mix}\ll\tau_S$ stated in the observer-selection corollary; under (EM) it is derived from the framework's own definitions, with (C2) read in its timescale form (C2-slow) — record persistence and readback being suppressed there, not excluded — for the coarse-grained description that the effective channel generates (Scope (ii)), and without (EM) it is a conjectural application.
 
 **Setup.** Let $V$ be a partition with hidden sector $B = S \setminus V$. Define the coupling-graph neighborhood $\Sigma_{\text{loc}}(V)$ as the subset of $S$ reached from $V$ by at most $\tau_S / \Delta t$ steps of $\varphi$, where $\Delta t$ is the update step. The exact Koopman operator of $\varphi$ — and of its restriction to any invariant subsystem — is a permutation operator: unitary, with modulus-one spectrum and exact recurrence. It has no dissipative gap; exponential mixing cannot hold for it. Mixing, where it holds, is a property of a *compression*: let $\mathcal{K}_\Sigma := P_\Sigma\, \mathcal{U}_\varphi\, P_\Sigma$ denote the coarse-grained effective channel on the observable algebra of a local region $\Sigma$. **Effective-mixing hypothesis (EM).** *The local effective channel $\mathcal{K}_{\Sigma_{\text{loc}}(V)}$ mixes, with mixing time $\tau_{\text{mix}}(\Sigma_{\text{loc}}(V))$ set by its (dissipative) spectral gap.* (EM) is an explicit physical hypothesis — motivated by local thermalization phenomenology, not derived from $\varphi$ — with the same status as the conditional-mixing curve of (C2).
 
@@ -755,7 +755,7 @@ where $\varepsilon$ absorbs typical-set and finite-size corrections. In particul
 
 **Corollary (anti-Boltzmann-brain).** A Boltzmann brain is by construction a local fluctuation within a neighborhood $\Sigma_{\text{loc}}$ that is itself in local thermal equilibrium — that is what "fluctuation from equilibrium" means. The exclusion is the conditional chain, in the coarse-grained description generated by iterating $\mathcal{K}_{\Sigma_{\text{BB}}}$ (Scope (ii)): (EM) together with $\tau_{\text{mix}}(\Sigma_{\text{BB}}) \ll \tau_S$ implies the candidate partition fails (C2-slow), and hence — by the fast-bath erasure lemma of §2.3, under its uniform-erosion hypothesis — has its order-$k$ readback gap suppressed as $e^{-(k-1)\tau_S/\tau_B}$ for every $k \geq 2$; that gap is what (C4), the condition the characterization actually requires, asserts. For a numeric illustration under a further generic-thermal estimate — (GT): the region's effective-channel correlation time is set by the thermal scale, $\sim \hbar/(k_B T_{\text{local}}) \cdot O(1)$, an estimate that fails for protected slow modes (conserved-quantity bottlenecks, activation barriers, glassy or critical slowing) — the bound is at most $\sim 10^{-12}$ s for temperatures warm enough to support structured matter. The illustration rides (GT); the exclusion itself rides only the chain. Any finite-information observer has $\tau_S$ bounded below by its own internal processing time — for biological or synthetic BBs of any currently-envisioned type, $\tau_S \gtrsim 10^{-6}$ s. The slow form (C2-slow), $\tau_S \ll \tau_B$, therefore fails by at least six orders of magnitude for any BB whose local environment satisfies (GT) — and because (EM) posits genuine mixing rather than mere speed, the record the persistence form (C2-structural) requires is degraded with it: records are scrambled before readback up to a residue suppressed exponentially in $\tau_S/\tau_{\text{mix}}$. The chain bounds the readback gap: by the fast-bath erasure lemma, the failure of (C2-slow) with uniform erosion bounds the order-$k$ gap by $\delta_0\, e^{-(k-1)\tau_S/\tau_B}$ for every $k \geq 2$ — Boltzmann-brain observers so described carry at most exponentially suppressed readback at those orders, and since they fail (C2-slow), within that description they are not observers in the sense of (C1)–(C4) with (C2) read in its timescale form; the fixed-basis quantum representation they trivially possess ($S \iff D \iff Q_{\mathrm{fb}}$) carries no discriminating content. The exclusion is structural, not measure-theoretic — it follows from the incompatibility of (C2-slow) with fast-mixing local equilibrium, not from arguments about the ratio of BBs to "ordinary" observers in an anthropic sample.
 
-**Corollary (past-hypothesis replacement, with its rate condition).** The theorem above bounds $\tau_B(V)\le\tau_{\rm mix}(\Sigma_{\rm loc}(V))+O(\varepsilon)$, while (C2)-slow requires $\tau_S\ll\tau_B$; those do not compose. Contradicting (C2-slow) additionally requires $\tau_{\rm mix}\ll\tau_S$ — the condition the Boltzmann-brain corollary states explicitly. A slowly mixing equilibrium with $\tau_S\ll\tau_{\rm mix}$ is entirely compatible with $\tau_S\ll\tau_B\le\tau_{\rm mix}$, so the valid form is that a sufficiently fast-mixing equilibrium supports no observer satisfying (C2) in its timescale form (C2-slow); for (C2-structural) and (C4) it gives suppression, not exclusion (§2.3). This matters for the de Sitter route, where $\tau_{\rm thermal}\sim\hbar/(k_BT_{\rm GH})=2\pi/H$ is Hubble-order and hence enormously longer than laboratory $\tau_S$: a de Sitter thermal state does not erase observer memory on accessible timescales. Under $\tau_{\rm mix}\ll\tau_S$: For $(S, \varphi)$ in its equilibrium macrostate, canonical typicality [31] implies every local subsystem is also in its local equilibrium. By the theorem — under (EM) for each region's effective channel, in the coarse-grained description it generates — $\tau_B(V)$ is bounded by the local $\tau_{\text{mix}}$ for every bounded-diameter partition $V$. No bounded-diameter partition supports (C2-slow) in that description. That every relevant equilibrium region satisfies (EM) at the required rate is itself a physical hypothesis, to be established model by model; the theorem converts it into the selection rule and does not supply it. In that description, with (C2) read in its timescale form, the framework's observers therefore exist only in the non-equilibrium phase of $(S, \varphi)$. The low-entropy initial condition commonly imported as a "past hypothesis" to select observer-compatible configurations is replaced here, within that description, by a structural selection rule: observers exist where (C1)–(C4), with (C2) read as (C2-slow), are satisfiable, and (C1)–(C4), so read, are not satisfiable in the coarse-grained description of the equilibrium phase.
+**Corollary (observer-selection rule, with its rate condition).** The theorem above bounds $\tau_B(V)\le\tau_{\rm mix}(\Sigma_{\rm loc}(V))+O(\varepsilon)$, while (C2)-slow requires $\tau_S\ll\tau_B$; those do not compose. Contradicting (C2-slow) additionally requires $\tau_{\rm mix}\ll\tau_S$ — the condition the Boltzmann-brain corollary states explicitly. A slowly mixing equilibrium with $\tau_S\ll\tau_{\rm mix}$ is entirely compatible with $\tau_S\ll\tau_B\le\tau_{\rm mix}$, so the valid form is that a sufficiently fast-mixing equilibrium supports no observer satisfying (C2) in its timescale form (C2-slow); for (C2-structural) and (C4) it gives suppression, not exclusion (§2.3). This matters for the de Sitter route, where $\tau_{\rm thermal}\sim\hbar/(k_BT_{\rm GH})=2\pi/H$ is Hubble-order and hence enormously longer than laboratory $\tau_S$: a de Sitter thermal state does not erase observer memory on accessible timescales. Under $\tau_{\rm mix}\ll\tau_S$: For $(S, \varphi)$ in its equilibrium macrostate, canonical typicality [31] implies every local subsystem is also in its local equilibrium. By the theorem — under (EM) for each region's effective channel, in the coarse-grained description it generates — $\tau_B(V)$ is bounded by the local $\tau_{\text{mix}}$ for every bounded-diameter partition $V$. No bounded-diameter partition supports (C2-slow) in that description. That every relevant equilibrium region satisfies (EM) at the required rate is itself a physical hypothesis, to be established model by model; the theorem converts it into the selection rule and does not supply it. In that description, with (C2) read in its timescale form, the framework's observers therefore exist only in the non-equilibrium phase of $(S, \varphi)$. The low-entropy initial condition commonly imported as a "past hypothesis" to select observer-compatible configurations is replaced here, within that description, by a structural selection rule: observers exist where (C1)–(C4), with (C2) read as (C2-slow), are satisfiable, and (C1)–(C4), so read, are not satisfiable in the coarse-grained description of the equilibrium phase.
 
 **Scope.** The theorem uses bounded coupling, finiteness of $S$, the canonical-typicality result [31], and the effective-mixing hypothesis (EM) — the last an additional conditional physical premise, established model by model; it does not follow from isotropy or bounded-coupling regularity. Three scope notes. (i) *Locality of $V$.* The theorem assumes $V$ has bounded coupling-graph diameter. Adversarially non-local partitions — cherry-picking distant subsystems that happen to be correlated — are excluded by that assumption; bounded coupling degree does not by itself bound the diameter of $V$, and the finite response time of a physical observer motivates the assumption. (ii) *Ergodicity.* The gap the argument needs belongs to the coarse-grained effective channel $\mathcal{K}_\Sigma$, not to $\varphi$ itself: the exact Koopman operator of a finite permutation has modulus-one spectrum and cannot mix, so no spectral gap on typical orbits of $\varphi$ is available or assumed. A gap for $\mathcal{K}_\Sigma$ is exactly the content of (EM) and does not follow from statistical isotropy or bounded-coupling regularity. The cycle structure of $\varphi$ does not settle (EM), which concerns the one-step compression: a single cycle does not make the exact compressions relax (below), and orbits of length at most $2$ do not prevent a gap (on four states with cells $\{0,1\}$, $\{2,3\}$, the transposition $(0\,2)$ gives $\mathcal{K}_\Sigma$ with eigenvalues $1$ and $0$). The exact linear rule of [SM §4.2], whose largest energy shell for $d\ge2$ splits into exponentially many cycles, is therefore not excluded by its cycle structure; whether it or the rounded bijection satisfies (EM) is open. The exact $n$-step compressions $P_\Sigma\, \mathcal{U}_\varphi^n\, P_\Sigma$ are compressions of a permutation, recur, and need not equal $\mathcal{K}_\Sigma^n$: for the cyclic shift on five states with a coarse variable marking one of them, they equal $\mathcal{K}_\Sigma$ for $1 \leq n \leq 4$ and the identity at $n = 5$, while $\mathcal{K}_\Sigma^n$ relaxes. The theorem and its two corollaries are therefore statements about the coarse-grained description generated by iterating $\mathcal{K}_\Sigma$; that these iterates govern the exact compressions over the accessible window is not established, and the corollaries do not assert that the exact visible law carries no accessible memory. (iii) *The cosmological realization.* The de Sitter horizon is thermal in the static-patch sense at Hawking temperature, but the globally expanding system is not in equilibrium with respect to any static-observer partition. The horizon's expansion sets $\tau_B \sim H^{-1}$ structurally, not thermally, which is the OI-native mechanism making human-timescale laboratory observation compatible with (C2).
 
diff --git a/papers/Main.pdf b/papers/Main.pdf
index 3558ac2e3..c6d2d6d50 100644
Binary files a/papers/Main.pdf and b/papers/Main.pdf differ
diff --git a/papers/Main.tex b/papers/Main.tex
index 88c05ab59..24f423abe 100644
--- a/papers/Main.tex
+++ b/papers/Main.tex
@@ -5333,7 +5333,7 @@ \subsubsection{4.6 Structural Exclusion of Equilibrium-Phase
 hypothesis'' of low initial entropy, in its role of selecting
 observer-compatible configurations --- with a theorem conditional on the
 effective-mixing hypothesis (EM) below and on the rate condition
-\(\tau_{\rm mix}\ll\tau_S\) stated in the past-hypothesis corollary;
+\(\tau_{\rm mix}\ll\tau_S\) stated in the observer-selection corollary;
 under (EM) it is derived from the framework's own definitions, with (C2)
 read in its timescale form (C2-slow) --- record persistence and readback
 being suppressed there, not excluded --- for the coarse-grained
@@ -5431,8 +5431,8 @@ \subsubsection{4.6 Structural Exclusion of Equilibrium-Phase
 from arguments about the ratio of BBs to ``ordinary'' observers in an
 anthropic sample.
 
-\textbf{Corollary (past-hypothesis replacement, with its rate
-condition).} The theorem above bounds
+\textbf{Corollary (observer-selection rule, with its rate condition).}
+The theorem above bounds
 \(\tau_B(V)\le\tau_{\rm mix}(\Sigma_{\rm loc}(V))+O(\varepsilon)\),
 while (C2)-slow requires \(\tau_S\ll\tau_B\); those do not compose.
 Contradicting (C2-slow) additionally requires
@@ -5871,4 +5871,4 @@ \subsection{REFERENCES}\label{references}}
 \emph{Phys. Rev.~Lett.} \textbf{103}, 210401 (2009); arXiv:0908.0238.
 
 \end{document}
-% source-sha256: f041b9a6856791036bba5280a0187e49a3c5774e1bbfc45595392423bd4a5dde
+% source-sha256: 9a941c5feed1f7c74c8d440c5c93870a27f0d118312dc59db04b588e99328831
diff --git a/papers/Methodology.md b/papers/Methodology.md
index 5ec13ff63..65bb7909f 100644
--- a/papers/Methodology.md
+++ b/papers/Methodology.md
@@ -181,7 +181,7 @@ The distinction matters because the two questions are routinely conflated, and a
 
 Rung 4, as the framework needs it, is the domain-existence claim: that there is a temporal domain for the dynamics $\varphi$ to act across. The claim of this section is that this cannot be derived from rung 1. Tokened differentiation, as established in Section 2, is tenseless: it is satisfied by a single differentiated tableau — a configuration, one frame — with no before and no after. To reach even *two* moments from a single differentiated frame requires a premise the frame does not contain. The argument takes the form of a small no-go result.
 
-The two-question distinction also fixes the complete inventory of the framework's temporal commitments. The framework's downstream physics involves more temporal structure than the bare domain-existence claim — it involves an *ordering* of moments, a *causal topology*, and a derived *arrow* of time — and a reader is owed an account of how these relate to the two axioms. The inventory has four levels. *Multiplicity* — that there is more than one moment — is the domain-existence claim; it is the framework's second axiom. *Ordering* — that the moments form a sequence — is the ordering question above; the framework leaves it open. *Causal topology* — that the ordering carries influence-structure, that earlier moments can affect later ones — is what the dynamics requires to be stated at all. *The arrow* — that the ordering has an experienced direction — is derived in the framework's substratum treatment from horizon growth. The relations among these four are definite, and reduce the inventory to the two axioms-or-questions already named. The causal topology is not an independent commitment: an influence-structure on an ordered set of moments connected by $\varphi$ is nothing over and above the ordering together with $\varphi$ itself, since "earlier influences later" just means "related by forward iteration of $\varphi$"; the dynamics $\varphi$ is independently present, so the causal topology reduces to the ordering. The arrow is derived, not posited, and rests on the causal topology, hence ultimately on the ordering. So the framework's genuinely independent temporal commitments are exactly two — multiplicity and ordering — and they are precisely the two questions this section distinguishes: multiplicity is settled by the second axiom, ordering is left open. Causal topology and the arrow are downstream of these two and of structure (the dynamics, horizon growth) the framework has independently. No third temporal axiom is required, and none is concealed: the two-question framing of this section is the complete account of where the framework's temporal commitments lie.
+The two-question distinction also fixes the complete inventory of the framework's temporal commitments. The framework's downstream physics involves more temporal structure than the bare domain-existence claim — it involves an *ordering* of moments, a *causal topology*, and an *arrow* of time — and a reader is owed an account of how these relate to the two axioms. The inventory has four levels. *Multiplicity* — that there is more than one moment — is the domain-existence claim; it is the framework's second axiom. *Ordering* — that the moments form a sequence — is the ordering question above; the framework leaves it open. *Causal topology* — that the ordering carries influence-structure, that earlier moments can affect later ones — is what the dynamics requires to be stated at all. *The arrow* — that the ordering has an experienced direction — is addressed in the framework's substratum treatment: on an expanding homogeneous general-relativistic background satisfying the null energy condition, the apparent-horizon area is a non-decreasing parameter (background kinematics), and the direction rests on boundary conditions and the orientation convention. The relations among these four are definite, and reduce the inventory to the two axioms-or-questions already named. The causal topology is not an independent commitment: an influence-structure on an ordered set of moments connected by $\varphi$ is nothing over and above the ordering together with $\varphi$ itself, since "earlier influences later" just means "related by forward iteration of $\varphi$"; the dynamics $\varphi$ is independently present, so the causal topology reduces to the ordering. The arrow rests on the causal topology, hence on the ordering, and on those boundary conditions. So the framework's genuinely independent temporal commitments are exactly two — multiplicity and ordering — and they are precisely the two questions this section distinguishes: multiplicity is settled by the second axiom, ordering is left open. Causal topology and the arrow are downstream of these two and of inputs the framework takes as given (the dynamics, the expanding branch, the boundary conditions that fix the arrow's direction). No third temporal axiom is required: the two-question framing of this section accounts for the framework's temporal axioms, and the arrow's boundary conditions are inputs rather than axioms.
 
 ### 5.2 Six attempted derivations
 
@@ -224,7 +224,7 @@ That fact is a second axiom. It is tempting to call it a mere parameter — a se
 The two axioms are:
 
 - **Axiom 1.** *Tokened differentiation occurs.* Concrete differentiated content is instantiated; there is some this-not-that. This is the evidence of Section 2. It is indubitable — its denial is itself a tokened differentiation — and it is conceptually primitive: even the vocabulary needed to state the second axiom presupposes it.
-- **Axiom 2.** *Differentiation recurs.* The principle of Axiom 1 is instantiated more than once over — not only within a single configuration but across a further axis. The framework's temporal domain is the minimal such recurrence; the configurational and temporal instances are the two minimum witnesses, and the actual count of axes is settled downstream by derivation, not fixed here. What Axiom 2 asserts is bare recurrence — that the principle holds more than once over — and nothing more: it does not by itself supply an ordering of the recurrent instances, a causal topology, or the specific dynamics $\varphi$; these are either left open (ordering, §5.1) or derived downstream from structure the framework independently has (causal topology and arrow, §5.2 inventory). The floor is two posits at the granularity stated, and the dynamical content the downstream physics requires enters through later, separately-stated structural conditions rather than through Axiom 2's semantics.
+- **Axiom 2.** *Differentiation recurs.* The principle of Axiom 1 is instantiated more than once over — not only within a single configuration but across a further axis. The framework's temporal domain is the minimal such recurrence; the configurational and temporal instances are the two minimum witnesses, and the actual count of axes is settled downstream by derivation, not fixed here. What Axiom 2 asserts is bare recurrence — that the principle holds more than once over — and nothing more: it does not by itself supply an ordering of the recurrent instances, a causal topology, or the specific dynamics $\varphi$; these are either left open (ordering, §5.1), derived downstream from structure the framework independently has (causal topology, §5.2 inventory), or addressed downstream with boundary conditions taken as inputs (arrow, §5.2 inventory). The floor is two posits at the granularity stated, and the dynamical content the downstream physics requires enters through later, separately-stated structural conditions rather than through Axiom 2's semantics.
 
 The two axioms stand in a definite relationship, and stating it precisely is part of the result. Axiom 2 **presupposes** Axiom 1: there can be no recurrence of differentiation without differentiation, and the very notion of an "axis" or "domain" along which the recurrence runs is itself a differentiation-structure, unstatable without Axiom 1's content. But Axiom 2 is **not derivable from** Axiom 1: that a differentiation occurs does not yield that it recurs — this is the no-go of Section 5.5. The relationship is therefore one-directional presupposition without entailment. Axiom 1 is first — conceptually and epistemically. Axiom 2 is genuinely additional. They are two, ordered, and not reducible to one.
 
diff --git a/papers/Methodology.pdf b/papers/Methodology.pdf
index 7ab0166f8..f7c2f6d71 100644
Binary files a/papers/Methodology.pdf and b/papers/Methodology.pdf differ
diff --git a/papers/Methodology.tex b/papers/Methodology.tex
index 003cf6004..57f61e543 100644
--- a/papers/Methodology.tex
+++ b/papers/Methodology.tex
@@ -884,8 +884,8 @@ \subsubsection{5.1 Two questions about succession, and which one is the
 framework's temporal commitments. The framework's downstream physics
 involves more temporal structure than the bare domain-existence claim
 --- it involves an \emph{ordering} of moments, a \emph{causal topology},
-and a derived \emph{arrow} of time --- and a reader is owed an account
-of how these relate to the two axioms. The inventory has four levels.
+and an \emph{arrow} of time --- and a reader is owed an account of how
+these relate to the two axioms. The inventory has four levels.
 \emph{Multiplicity} --- that there is more than one moment --- is the
 domain-existence claim; it is the framework's second axiom.
 \emph{Ordering} --- that the moments form a sequence --- is the ordering
@@ -893,24 +893,29 @@ \subsubsection{5.1 Two questions about succession, and which one is the
 that the ordering carries influence-structure, that earlier moments can
 affect later ones --- is what the dynamics requires to be stated at all.
 \emph{The arrow} --- that the ordering has an experienced direction ---
-is derived in the framework's substratum treatment from horizon growth.
-The relations among these four are definite, and reduce the inventory to
-the two axioms-or-questions already named. The causal topology is not an
-independent commitment: an influence-structure on an ordered set of
-moments connected by \(\varphi\) is nothing over and above the ordering
-together with \(\varphi\) itself, since ``earlier influences later''
-just means ``related by forward iteration of \(\varphi\)''; the dynamics
+is addressed in the framework's substratum treatment: on an expanding
+homogeneous general-relativistic background satisfying the null energy
+condition, the apparent-horizon area is a non-decreasing parameter
+(background kinematics), and the direction rests on boundary conditions
+and the orientation convention. The relations among these four are
+definite, and reduce the inventory to the two axioms-or-questions
+already named. The causal topology is not an independent commitment: an
+influence-structure on an ordered set of moments connected by
+\(\varphi\) is nothing over and above the ordering together with
+\(\varphi\) itself, since ``earlier influences later'' just means
+``related by forward iteration of \(\varphi\)''; the dynamics
 \(\varphi\) is independently present, so the causal topology reduces to
-the ordering. The arrow is derived, not posited, and rests on the causal
-topology, hence ultimately on the ordering. So the framework's genuinely
+the ordering. The arrow rests on the causal topology, hence on the
+ordering, and on those boundary conditions. So the framework's genuinely
 independent temporal commitments are exactly two --- multiplicity and
 ordering --- and they are precisely the two questions this section
 distinguishes: multiplicity is settled by the second axiom, ordering is
 left open. Causal topology and the arrow are downstream of these two and
-of structure (the dynamics, horizon growth) the framework has
-independently. No third temporal axiom is required, and none is
-concealed: the two-question framing of this section is the complete
-account of where the framework's temporal commitments lie.
+of inputs the framework takes as given (the dynamics, the expanding
+branch, the boundary conditions that fix the arrow's direction). No
+third temporal axiom is required: the two-question framing of this
+section accounts for the framework's temporal axioms, and the arrow's
+boundary conditions are inputs rather than axioms.
 
 \hypertarget{six-attempted-derivations}{%
 \subsubsection{5.2 Six attempted
@@ -1111,9 +1116,10 @@ \subsubsection{5.4 The status of succession --- a second
   more than once over --- and nothing more: it does not by itself supply
   an ordering of the recurrent instances, a causal topology, or the
   specific dynamics \(\varphi\); these are either left open (ordering,
-  §5.1) or derived downstream from structure the framework independently
-  has (causal topology and arrow, §5.2 inventory). The floor is two
-  posits at the granularity stated, and the dynamical content the
+  §5.1), derived downstream from structure the framework independently
+  has (causal topology, §5.2 inventory), or addressed downstream with
+  boundary conditions taken as inputs (arrow, §5.2 inventory). The floor
+  is two posits at the granularity stated, and the dynamical content the
   downstream physics requires enters through later, separately-stated
   structural conditions rather than through Axiom 2's semantics.
 \end{itemize}
@@ -3102,4 +3108,4 @@ \subsection{REFERENCES}\label{references}}
 Time}, 2nd edn. Clarendon Press, Oxford.
 
 \end{document}
-% source-sha256: eb6974548ca3a290f33493aa63efca0d3ddfc7ce3be73ce1d6f0555df672ba87
+% source-sha256: ffb4c67b4d2b6e61466c47e09946e4b502c4b0d8f3e20e82274cb9648d9adf98
diff --git a/papers/SM.md b/papers/SM.md
index b55e55494..a3244e53c 100644
--- a/papers/SM.md
+++ b/papers/SM.md
@@ -1363,7 +1363,7 @@ The coincidence bound of §7.7 quantifies how concentrated the framework's *retr
 
 ## 8. Discussion
 
-The cosmological application — including the horizon calibration $\hbar = c^3\epsilon^2/(4G\sigma)$, the Bekenstein-Hawking area law, the conditional dissolution of the cosmological constant problem ([GR §6.2]), and the dynamical dark energy prediction — is developed in [GR]. The substratum-level reconstruction theorem and the synthesis claim that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same $(S, \varphi)$ are developed in [Substratum].
+The cosmological application — including the horizon calibration $\hbar = c^3\epsilon^2/(4G\sigma)$, the Bekenstein-Hawking area law, the conditional dissolution of the cosmological constant problem ([GR §6.2]), and the dynamical dark energy prediction — is developed in [GR]. The substratum-level reconstruction theorem and the synthesis claim that quantum mechanics and general relativity descend as projections of the same $(S, \varphi)$, with a third projection bearing on the arrow of time, are developed in [Substratum].
 
 ### 8.1 Structural realism
 
@@ -1589,7 +1589,7 @@ This paper has shown that the Standard Model's gauge group, matter content, and
 
 The framework is sharply falsifiable by experiments now in progress or under construction. JUNO's first results (November 2025) are matched at $0.14\sigma$, and the subsequent post-JUNO global fit at $0.07\sigma$, by the parameter-free value $\sin^2\theta_{12} = 1/3 - 1/(4\pi^2)$ — a retrodiction; sub-percent precision over the 30-year design lifetime provides the pre-registered forward test at the level required to distinguish it from competing tribimaximal variants. The construction supplies no quantitative neutron EDM prediction; H-top and H-det remain open (§5.3). Neutrinoless double beta decay searches will test the Majorana-neutrino prediction. The continuing absence of any new gauge bosons, fundamental scalars, or supersymmetric partners at the LHC and its successors corroborates structural predictions of the framework, which forbids all such states.
 
-The cosmological application of the same framework — including the horizon calibration $\hbar = c^3\epsilon^2/(4G\sigma)$ and the Bekenstein-Hawking area law with its $1/4$ coefficient (consistency relations, both conditional on the horizon conditions recorded there), the dissolution of the cosmological constant problem at the descriptive level, and dark energy in running-vacuum form (the functional form the local expansion shared by every smooth model; the content the Type II channel, with magnitude not derived at theorem level but consistent with the Bertini et al. 2025 direct DESI-DR2-era fit at $2\sigma$) — is developed in [GR]. The substratum-level reconstruction theorem, the substratum gauge group, and the synthesis claim that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same $(S, \varphi)$ are developed in [Substratum]. Together, this four-paper synthesis ([Main], [SM], [GR], [Substratum]) makes a single claim: the Standard Model, general relativity, and the arrow of time are three projections of one finite deterministic construction — the visible-sector projection developed here, the boundary-level thermodynamic limit developed in [GR], and the horizon-clock-plus-observer-selection projection developed in [Substratum §5.4]. The fifth core paper [Structure] articulates the framework's two-dimensional hierarchical structural realism and develops its relationship to other unification programs through universality classes of embedded observers.
+The cosmological application of the same framework — including the horizon calibration $\hbar = c^3\epsilon^2/(4G\sigma)$ and the Bekenstein-Hawking area law with its $1/4$ coefficient (consistency relations, both conditional on the horizon conditions recorded there), the dissolution of the cosmological constant problem at the descriptive level, and dark energy in running-vacuum form (the functional form the local expansion shared by every smooth model; the content the Type II channel, with magnitude not derived at theorem level but consistent with the Bertini et al. 2025 direct DESI-DR2-era fit at $2\sigma$) — is developed in [GR]. The substratum-level reconstruction theorem, the substratum gauge group, and the synthesis claim that quantum mechanics and general relativity descend as projections of the same $(S, \varphi)$, with a third projection bearing on the arrow of time, are developed in [Substratum]. Together, this four-paper synthesis ([Main], [SM], [GR], [Substratum]) makes a single claim: the Standard Model and general relativity are projections of one finite deterministic construction — the visible-sector projection developed here and the boundary-level thermodynamic limit developed in [GR] ; the horizon-clock-plus-observer-selection projection is developed in [Substratum §5.4], where the alignment of the arrows of time is open. The fifth core paper [Structure] articulates the framework's two-dimensional hierarchical structural realism and develops its relationship to other unification programs through universality classes of embedded observers.
 
 ***
 
diff --git a/papers/SM.pdf b/papers/SM.pdf
index 1483fb4ab..8aef259e6 100644
Binary files a/papers/SM.pdf and b/papers/SM.pdf differ
diff --git a/papers/SM.tex b/papers/SM.tex
index 6f6c29de2..eac4fd82a 100644
--- a/papers/SM.tex
+++ b/papers/SM.tex
@@ -7524,9 +7524,9 @@ \subsection{8. Discussion}\label{discussion}}
 the conditional dissolution of the cosmological constant problem ({[}GR
 §6.2{]}), and the dynamical dark energy prediction --- is developed in
 {[}GR{]}. The substratum-level reconstruction theorem and the synthesis
-claim that quantum mechanics, general relativity, and the arrow of time
-descend as three projections of the same \((S, \varphi)\) are developed
-in {[}Substratum{]}.
+claim that quantum mechanics and general relativity descend as
+projections of the same \((S, \varphi)\), with a third projection
+bearing on the arrow of time, are developed in {[}Substratum{]}.
 
 \hypertarget{structural-realism}{%
 \subsubsection{8.1 Structural realism}\label{structural-realism}}
@@ -8750,19 +8750,20 @@ \subsection{9. Conclusion}\label{conclusion}}
 channel, with magnitude not derived at theorem level but consistent with
 the Bertini et al.~2025 direct DESI-DR2-era fit at \(2\sigma\)) --- is
 developed in {[}GR{]}. The substratum-level reconstruction theorem, the
-substratum gauge group, and the synthesis claim that quantum mechanics,
-general relativity, and the arrow of time descend as three projections
-of the same \((S, \varphi)\) are developed in {[}Substratum{]}.
-Together, this four-paper synthesis ({[}Main{]}, {[}SM{]}, {[}GR{]},
-{[}Substratum{]}) makes a single claim: the Standard Model, general
-relativity, and the arrow of time are three projections of one finite
+substratum gauge group, and the synthesis claim that quantum mechanics
+and general relativity descend as projections of the same
+\((S, \varphi)\), with a third projection bearing on the arrow of time,
+are developed in {[}Substratum{]}. Together, this four-paper synthesis
+({[}Main{]}, {[}SM{]}, {[}GR{]}, {[}Substratum{]}) makes a single claim:
+the Standard Model and general relativity are projections of one finite
 deterministic construction --- the visible-sector projection developed
-here, the boundary-level thermodynamic limit developed in {[}GR{]}, and
-the horizon-clock-plus-observer-selection projection developed in
-{[}Substratum §5.4{]}. The fifth core paper {[}Structure{]} articulates
-the framework's two-dimensional hierarchical structural realism and
-develops its relationship to other unification programs through
-universality classes of embedded observers.
+here and the boundary-level thermodynamic limit developed in {[}GR{]} ;
+the horizon-clock-plus-observer-selection projection is developed in
+{[}Substratum §5.4{]}, where the alignment of the arrows of time is
+open. The fifth core paper {[}Structure{]} articulates the framework's
+two-dimensional hierarchical structural realism and develops its
+relationship to other unification programs through universality classes
+of embedded observers.
 
 \begin{center}\rule{0.5\linewidth}{0.5pt}\end{center}
 
@@ -9155,4 +9156,4 @@ \subsection{References}\label{references}}
 \textbf{07}, 104 (2017); arXiv:1705.01853.
 
 \end{document}
-% source-sha256: d438d343fd92d92f446ba2f62048d1819868d0d55a482c61e6d29655c0f58fa5
+% source-sha256: 1a908ba9fa329c5d83dd96349dc81c4c0c75182163b1e0a7e804ad71c045eebd
diff --git a/papers/Substratum.md b/papers/Substratum.md
index 815258fcd..cb0b25c98 100644
--- a/papers/Substratum.md
+++ b/papers/Substratum.md
@@ -11,7 +11,7 @@
 
 ## Abstract
 
-Quantum mechanics and general relativity are conventionally treated as two fundamental theories awaiting unification. This paper develops the alternative that they are two projections of a single finite deterministic construction, joined by a third projection that produces the arrow of time. The underlying object is the bijection $(S,\varphi)$ on a finite configuration space. Its visible-sector reduction has the exact finite stochastic/quantum-representation equivalence established in [Main]: finite observable laws, finite reversible realizations, and fixed-basis unitary/Born representations coincide, while genuine history readback forces hidden predictive memory. Its spatial coupling graph supplies finite causal cones; equality with the complete standard quantum instrument theory holds, on a finite carrier with complex matrix kinematics taken as given, exactly under the five completion conditions characterized in [GR §3.3], and which of them the substratum supplies is open. The boundary-thermodynamic projection yields the gravitational description developed in [GR], and the lattice dynamics yields the Standard-Model structure developed in [SM].
+Quantum mechanics and general relativity are conventionally treated as two fundamental theories awaiting unification. This paper develops the alternative that they are two projections of a single finite deterministic construction, joined by a third projection that supplies a time-symmetric observer-selection rule, read against the area of the apparent horizon, which on an expanding homogeneous general-relativistic background satisfying the null energy condition is non-decreasing as a matter of background kinematics. The underlying object is the bijection $(S,\varphi)$ on a finite configuration space. Its visible-sector reduction has the exact finite stochastic/quantum-representation equivalence established in [Main]: finite observable laws, finite reversible realizations, and fixed-basis unitary/Born representations coincide, while genuine history readback forces hidden predictive memory. Its spatial coupling graph supplies finite causal cones; equality with the complete standard quantum instrument theory holds, on a finite carrier with complex matrix kinematics taken as given, exactly under the five completion conditions characterized in [GR §3.3], and which of them the substratum supplies is open. The boundary-thermodynamic projection yields the gravitational description developed in [GR], and the lattice dynamics yields the Standard-Model structure developed in [SM].
 
 The two substratum-level results that make these emergences one construction rather than independent applications are the reconstruction theorem (Theorem 23) and the substratum gauge group (Theorem 24). Theorem 23 reconstructs the **local propagating lattice/gauge residue** from its stated empirical inputs and structural assumptions, conditional on the named modeling and nearest-neighbor premises; Bell-inclusive existence additionally requires H-Bell and Bell-inclusive uniqueness is not established. Theorem 24 identifies generators of the observables-preserving kernel of the dilation-datum/local reconstruction; their completeness is open. The synthesis claim is therefore structural rather than rhetorical: the same finite deterministic object supports the visible stochastic/quantum projection, the boundary-thermodynamic gravitational projection, and the lattice gauge projection. The traditional QM–GR “unification problem” is correspondingly recast as a category error between two projections of one object, while the theorem statuses and remaining conditional steps are kept explicit.
 
@@ -23,7 +23,7 @@ This paper argues that the QM-GR incompatibility is not a technical problem to b
 
 The present paper develops the substratum-level results that make these three emergences — emergent QM in [Main], the Standard Model derivation in [SM], and the gravitational sector in [GR] — a single construction rather than three separate applications of the same formalism. The two technical results that do this work are the *reconstruction theorem* (§3) and the *substratum gauge group* (§4). The reconstruction theorem shows — under its stated conditions (M1-T; C2-mixing; the Stage-2 structural inputs, including the nearest-neighbor premise) — that observed physics fixes the **local propagating lattice/gauge residue**. Bell violation enters separately through M1-B: a common measurement-independent deterministic completion exists only conditional on H-Bell, and no theorem here makes the Bell-inclusive completion unique. The substratum gauge group is the kernel of this inverse map; §4 identifies three families of transformations of $(S, \varphi)$ in it, one of them (alphabet change) under the alphabet-size hypothesis of [SM §2.7], whose exhaustion of all observables-preserving operations of the dilation-datum substratum is open. The combination of these two results turns the framework's three derivations into a single object: the SM gauge group is not chosen from a landscape, the gravitational thermodynamics are not appended to a quantum description, and the emergent QM is not postulated as a starting point. All three emerge from the same $(S, \varphi)$, and the substratum gauge group is the symmetry that makes the relationship between them exact rather than approximate.
 
-The synthesis claim that follows from these results — developed in §5 — is that quantum mechanics and general relativity are not two theories awaiting unification but two projections of the same construction, viewed at different levels of description. Quantum mechanics lives at the level of the visible-sector trace-out: the embedded observer sees the bijection through the hole of finite causal access, and the resulting compressed description admits the universal unitary quantum representation. General relativity lives at the level of the boundary thermodynamics: the same causal partition that produces, at the visible-sector level, the reduced description admitting the quantum representation produces classical horizon thermodynamics at the boundary level, and Jacobson's thermodynamic argument promotes this to Einstein's equations. The two descriptions are not in tension because they refer to different objects in the construction — not different *aspects of the same regime*, which would invite unification, but different *projections of the same substratum*, which makes unification a category error. The substratum gauge group makes this precise: at the level of $(S, \varphi)$, there is no quantum and no classical, no metric and no wave function; there is only the bijection and its symmetries. The trace-out at the visible-sector level produces one set of derived structures (the wave function, the SM gauge group, particle content), and the thermodynamic limit at the boundary level produces another (the metric, $\hbar$, the BH entropy, the dynamical dark energy), and the framework provides the construction that makes both of these descriptions exact projections of the same object. The same construction produces a third projection (§5.4) that resolves the arrow-of-time question by descending the observer's emergent arrows — cosmological, thermodynamic, memory, measurement — from the horizon-growth clock and the C1–C4 observer-selection theorem of [Main, §4.6], answering the Ellis objection to bijective substrates without adding a structural commitment.
+The synthesis claim that follows from these results — developed in §5 — is that quantum mechanics and general relativity are not two theories awaiting unification but two projections of the same construction, viewed at different levels of description. Quantum mechanics lives at the level of the visible-sector trace-out: the embedded observer sees the bijection through the hole of finite causal access, and the resulting compressed description admits the universal unitary quantum representation. General relativity lives at the level of the boundary thermodynamics: the same causal partition that produces, at the visible-sector level, the reduced description admitting the quantum representation produces classical horizon thermodynamics at the boundary level, and Jacobson's thermodynamic argument promotes this to Einstein's equations. The two descriptions are not in tension because they refer to different objects in the construction — not different *aspects of the same regime*, which would invite unification, but different *projections of the same substratum*, which makes unification a category error. The substratum gauge group makes this precise: at the level of $(S, \varphi)$, there is no quantum and no classical, no metric and no wave function; there is only the bijection and its symmetries. The trace-out at the visible-sector level produces one set of derived structures (the wave function, the SM gauge group, particle content), and the thermodynamic limit at the boundary level produces another (the metric, $\hbar$, the BH entropy, the dynamical dark energy), and the framework provides the construction that makes both of these descriptions exact projections of the same object. The same construction gives a third projection (§5.4): the C1–C4 observer-selection theorem of [Main, §4.6], which is time-symmetric and selects where observers can exist, read against the apparent-horizon area, which is non-decreasing on an expanding homogeneous general-relativistic background satisfying the null energy condition (background kinematics, not a substratum prediction); a common mechanism aligning the cosmological, thermodynamic, memory and measurement arrows is not established.
 
 The paper is organized as follows. §1.1 states the framework's primary metaphysical commitment and its relation to explanatory and empirical claims. §2 develops the interpretation of $(S, \varphi)$ as a finite lossless memory with partial read-write access and distinguishes the physical implementation question from the fixed-interface invariances of §4. §3 states the reconstruction theorem (Theorem 23) with its conditions. §4 states the substratum invariance results (Theorem 24), their open completeness question and the three-level gauge comparison. §5 develops the synthesis claim, citing [Main], [SM] and [GR] for the empirical content. §6 discusses structural realism, the ontological hierarchy and the measurement problem. §7 concludes.
 
@@ -53,7 +53,7 @@ The $10^{122}$ CC discrepancy is the compression ratio between total storage and
 
 **Remark (effective finiteness and the computation comparison).** The finite visible sector and boundary layer can determine the leading-order accessible-time law under [GR §3.2]'s hypotheses. This is a comparison of observed laws, not a proof that total-carrier finiteness is physically unnecessary or that every infinite tape realizes the same law. The exact normalized-product-ancilla identity of Theorem 24(iii) concerns finite added factors. The representation result of [Main §3.4] covers finite observable laws, while the history-readback sector retains its separate memory hypotheses.
 
-**The arrow of time.** The substratum has no arrow of time — φ and φ⁻¹ are equally valid. Observers satisfying C1–C4, with C2 read in its timescale form, are nonetheless structurally restricted to the non-equilibrium phase of $(S, \varphi)$: the theorem in [Main, §4.6] shows, under its effective-mixing hypothesis and for the coarse-grained description generated by iterating the one-step effective channel, that $\tau_B(V)$ is bounded by the local mixing time on equilibrium microstates, so C2 fails there in its timescale form, record persistence being degraded rather than shown to fail, wherever that mixing time is short compared with $\tau_S$ — and the readback gap of C4, the condition the characterization registers, is suppressed exponentially in $\tau_S/\tau_B$ at every order $k \geq 2$ by the erasure lemma of [Main §2.3]. In that description, this replaces the low-entropy "past hypothesis", in its role of selecting observer-compatible configurations, with a structural observer-selection rule and excludes Boltzmann brains structurally rather than anthropically. Entropy increase is then emergent as a coarse-grained description of the non-equilibrium dynamics observers necessarily see.
+**The arrow of time.** The substratum has no arrow of time — φ and φ⁻¹ are equally valid. Observers satisfying C1–C4, with C2 read in its timescale form, are nonetheless structurally restricted to the non-equilibrium phase of $(S, \varphi)$: the theorem in [Main, §4.6] shows, under its effective-mixing hypothesis and for the coarse-grained description generated by iterating the one-step effective channel, that $\tau_B(V)$ is bounded by the local mixing time on equilibrium microstates, so C2 fails there in its timescale form, record persistence being degraded rather than shown to fail, wherever that mixing time is short compared with $\tau_S$ — and the readback gap of C4, the condition the characterization registers, is suppressed exponentially in $\tau_S/\tau_B$ at every order $k \geq 2$ by the erasure lemma of [Main §2.3]. In that description, this replaces the low-entropy "past hypothesis", in its role of selecting observer-compatible configurations, with a structural observer-selection rule and excludes Boltzmann brains structurally rather than anthropically. The rule is time-symmetric: the direction in which coarse-grained entropy increases rests on a low-entropy or product boundary condition, not on the rule (§5.4).
 
 **The incompleteness family.** The framework's central result belongs to a family of impossibility-with-structure theorems. Gödel: a formal system cannot prove all truths about itself — the unprovable truths have rigid structure. Turing: a computer cannot decide all questions about its own behavior — the undecidable problems have rigid structure. OI: an observer embedded in a deterministic system cannot access the complete state — the emergent description has rigid structure (quantum mechanics). The common structure is *self-reference under finite resources*.
 
@@ -364,19 +364,19 @@ For (ii), partition $\Omega$ by its first $j$ readings, starting with the indisc
 
 ## 5. Synthesis: Three Projections of (S, φ)
 
-The reconstruction theorem (§3) establishes that the stated empirical and structural inputs, together with M1-T, C2-mixing and the nearest-neighbor premise of Stage 2(b), fix the **local propagating lattice/gauge residue**. M1-B and H-Bell separately condition the existence of a measurement-independent Bell-violating completion; Bell-inclusive uniqueness is open. The substratum gauge group (§4) makes the equivalence relation precise. This section makes the synthesis claim explicit: quantum mechanics, general relativity, and the arrow of time are not three independent theories but three projections of the same finite deterministic construction.
+The reconstruction theorem (§3) establishes that the stated empirical and structural inputs, together with M1-T, C2-mixing and the nearest-neighbor premise of Stage 2(b), fix the **local propagating lattice/gauge residue**. M1-B and H-Bell separately condition the existence of a measurement-independent Bell-violating completion; Bell-inclusive uniqueness is open. The substratum gauge group (§4) makes the equivalence relation precise. This section makes the synthesis claim explicit: quantum mechanics and general relativity are not two independent theories but projections of the same finite deterministic construction, and a third projection bears on the arrow of time (§5.4).
 
 ### 5.1 The three projections
 
-The framework's derivations apply the same trace-out machinery to the same $(S, \varphi)$ along three projections: a visible-sector projection that produces the reduced description admitting the quantum representation ([Main]) and the Standard Model ([SM]); a boundary-thermodynamic projection that produces general relativity ([GR]); and an arrow-of-time projection (§5.4) that produces the emergent cosmological, thermodynamic, memory, and measurement arrows from a single structural clock and the C1–C4 observer-selection theorem of [Main, §4.6].
+The framework's derivations apply the same trace-out machinery to the same $(S, \varphi)$ along three projections: a visible-sector projection that produces the reduced description admitting the quantum representation ([Main]) and the Standard Model ([SM]); a boundary-thermodynamic projection that produces general relativity ([GR]); and an arrow-of-time projection (§5.4) that supplies the time-symmetric C1–C4 observer-selection theorem of [Main, §4.6], read against the apparent-horizon area, which is non-decreasing, as background kinematics, on an expanding homogeneous general-relativistic background satisfying the null energy condition.
 
 **Visible-sector projection.** At the level of the embedded observer's epistemic access, the trace-out over the hidden sector produces the exact finite observable-law correspondence established in [Main]: $S\iff D\iff Q_{\mathrm{fb}}$. The fixed-basis unitary/Born representation is therefore exact and bidirectional but universal. The nontrivial OI content is history readback: C4 forces conditional memory and hidden predictive structure in every faithful completion; C1 and C3 diagnose the mediation and its capacity, while C2-structural, the persistence of the record to readback, holds by reversibility in every bijective realization and constrains coarse-grained descriptions of the hidden sector. In the spatial theory the dependency graph additionally supplies exact causal cones for finite-range couplings. Full equality with the standard coherent instrument/composite theory holds, on a finite carrier with complex matrix kinematics taken as given, exactly under the five completion conditions of [Main §3.4] and [GR §3.3]; which of them the substratum supplies is open. Bell is separate: the measurement-independent deterministic completion reaches quantum Bell violation only by ontic parameter dependence, with operational no-signaling [Main §3.3]. Applied to a cubic lattice with the wave equation as substratum dynamics — itself uniquely selected, up to the coefficient $\alpha$, among second-order reversible nearest-neighbor dynamics by center independence, isotropy, and linearity ([SM, §4.1]) — this projection produces the Standard Model gauge group $\mathrm{SU}(3) \times \mathrm{SU}(2) \times \mathrm{U}(1)$ (from the commutant of the coupling matrix, [SM, §§4.4–4.6]), three generations (physical reading conditional on H-spin', [SM §4.7]; chirality on H-χ', [SM §4.8]) and the Higgs as a composite scalar in the singlet staggered taste ([SM, §4.7], Theorems 8–11), and anomaly-free hypercharges ([SM, §4.9], Theorems 14–15). Twenty-two quantitative predictions follow ([SM, §7]), including the Cabibbo angle from a single Brillouin-zone distance, the Koide angle from a cubic-group quadratic Casimir, all three PMNS angles within $1.1\sigma$ (vs NuFIT 6.0), six fermion masses from one empirical input to better than $1\%$, and all three SM gauge couplings at $M_Z$ ([SM, §6.3]).
 
 **Boundary-thermodynamic projection.** At the level of the partition boundary, classical horizon thermodynamics — the temperature, entropy, and dynamical evolution of the causal horizon as a gravitating object — produces general relativity through Jacobson's thermodynamic argument, conditionally on the open covariant step ([GR, §8.7]): the Clausius relation $dE = T \, dS$ applied at local causal horizons yields Einstein's equations with the same structural inputs that give the visible-sector projection its form. Applied to the cosmological horizon as a causal partition ([GR, §2]), this projection determines the emergent action scale $\hbar = c^3 \epsilon^2 / (4G\sigma)$ from thermal self-consistency ([GR, §3]), fixes the discreteness scale $\epsilon = 2\sqrt{\sigma}\, l_p$, for an entropy $\sigma$ per boundary mode ($\epsilon = 2 l_p$ at one nat per mode), as the unique simultaneous solution ([GR, §4]), recovers the Bekenstein-Hawking entropy $S = A/(4 l_p^2)$, the mode count reproducing it once the calibration fixes $\epsilon/\sqrt{\sigma} = 2 l_p$ and the $1/4$ being the ratio $2\pi/8\pi$ of the KMS and first-law normalizations ([GR, §5]) — the action scale, the discreteness scale and the coefficient conditional on H-slope with the horizon and frame conditions of [GR, §2, §8.5]; GW250114 confirms the classical area theorem, the coefficient itself untested directly — and addresses the cosmological constant problem by identifying the quantum vacuum energy and the effective $\Lambda$ as properties of logically distinct levels of description rather than commensurable quantities whose $10^{122}$ ratio requires cancellation ([GR, §6]), the descriptive distinction established and the dissolution conditional on the open coupling step ([GR, §6.2]). The framework carries dark energy in running-vacuum form — the functional form the local expansion shared by every smooth model, the content the Type II channel with magnitude $|\nu| \sim 10^{-32}$ (not derived at theorem level) — consistent with the Bertini et al. (2025) direct DESI-DR2-era fit at $2\sigma$ (which is consistent with the ΛCDM limit); a ~$95\%$ dark sector corollary as a structural feature rather than a new particle species; and a MOND acceleration scale $a_0 = cH/6$ from entropy displacement at the boundary ([GR, §7]).
 
-The three projections share the same source ($(S, \varphi)$), the same trace-out machinery (marginalization over the hidden sector under conditions C1–C4), and the same structural inputs (the partition geometry, the boundary entropy, the substratum gauge group). They differ only in which structure of the embedded observer's description they derive: the visible-sector projection (§5.1, §5.3) works on the bulk dynamics and produces the quantum-representable reduced description and the Standard Model; the boundary-thermodynamic projection (§5.1) works on the classical thermal data at the partition boundary and produces general relativity; the arrow-of-time projection (§5.4) works on the horizon-clock parameter along observer worldlines and produces the cascade of emergent arrows.
+The three projections share the same source ($(S, \varphi)$), the same trace-out machinery (marginalization over the hidden sector under conditions C1–C4), and the same structural inputs (the partition geometry, the boundary entropy, the substratum gauge group). They differ only in which structure of the embedded observer's description they derive: the visible-sector projection (§5.1, §5.3) works on the bulk dynamics and produces the quantum-representable reduced description and the Standard Model; the boundary-thermodynamic projection (§5.1) works on the classical thermal data at the partition boundary and produces general relativity; the arrow-of-time projection (§5.4) works on the horizon-area parameter along observer worldlines together with the observer-selection rule.
 
-**The projections and the Poincaré structure of the partition.** The three projections are organized by how the observer-worldline-centered partition treats the Poincaré group, which makes the boost/rotation result of [Main §3.5] one facet of a larger pattern. The partition selects a preferred event — a spatial origin and a rest frame, jointly the observer's worldline — so the compact rotations about that worldline are not excluded by the partition's rest-frame data, while the non-compact generators are broken, given that the coarse-graining map determines the horizon foliation ([Main §3.5]); spatial isotropy enters as an empirical input. Boosts, tied to the selected rest frame, surface as the cosmic rest frame (operationally the CMB frame) and the boost-sector Lorentz residual ([SM §3.1]); spatial translations, which merely relabel the observer's origin, are broken only observer-relatively and are recovered as spatial homogeneity — the equivalence of all worldline-centered descriptions; and time translation is broken intrinsically by the growth of the horizon, which is the boundary-thermodynamic and arrow projections (§5.4, [GR]). The partition's symmetry-breaking signature — isotropy not broken by the partition, a preferred rest frame, spatial homogeneity, a distinguished cosmic time — is the symmetry structure of a Friedmann-Robertson-Walker cosmology. The framework does not derive that structure; it exhibits the partition's symmetry-breaking pattern as compatible with it, with ordinary momentum conservation holding in the visible sector up to corrections of order (length / horizon radius).
+**The projections and the Poincaré structure of the partition.** The three projections are organized by how the observer-worldline-centered partition treats the Poincaré group, which makes the boost/rotation result of [Main §3.5] one facet of a larger pattern. The partition selects a preferred event — a spatial origin and a rest frame, jointly the observer's worldline — so the compact rotations about that worldline are not excluded by the partition's rest-frame data, while the non-compact generators are broken, given that the coarse-graining map determines the horizon foliation ([Main §3.5]); spatial isotropy enters as an empirical input. Boosts, tied to the selected rest frame, surface as the cosmic rest frame (operationally the CMB frame) and the boost-sector Lorentz residual ([SM §3.1]); spatial translations, which merely relabel the observer's origin, are broken only observer-relatively and are recovered as spatial homogeneity — the equivalence of all worldline-centered descriptions; and time translation is broken by the growth of the apparent horizon on an expanding background with $\rho + p/c^2 > 0$, an input to the boundary-thermodynamic and arrow projections; in exact de Sitter the horizon area is constant (§5.4, [GR]). The partition's symmetry-breaking signature — isotropy not broken by the partition, a preferred rest frame, spatial homogeneity, a distinguished cosmic time — is the symmetry structure of a Friedmann-Robertson-Walker cosmology. The framework does not derive that structure; it exhibits the partition's symmetry-breaking pattern as compatible with it, with ordinary momentum conservation holding in the visible sector up to corrections of order (length / horizon radius).
 
 ### 5.2 The QM-GR incompatibility as a category error
 
@@ -398,21 +398,21 @@ The boundary-thermodynamic projection that produces GR is a parallel structure a
 
 ### 5.4 The arrow of time as a third projection
 
-After Main §4.6 (structural observer-selection theorem), the arrow of time joins quantum mechanics and general relativity as a third emergent structure that the framework derives from the same underlying $(S, \varphi)$ and the same trace-out machinery. Before §4.6, the arrow of time was treated as external to the derivation chain — a past hypothesis or an anthropic selection, imposed on the framework rather than produced by it. With §4.6 in place, the observer's confinement to the non-equilibrium phase of $(S, \varphi)$, with C2 read in its timescale form (record persistence and readback there being suppressed, not excluded), is a structural consequence of the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), in the coarse-grained description generated by iterating the one-step effective channel, and every arrow the observer sees descends from this confinement together with the horizon growth that the accelerating phase produces. The third projection has the same form as the first two: a feature of how the embedded observer sees $(S, \varphi)$ from inside, rather than a feature of $(S, \varphi)$ itself.
+Main §4.6 (structural observer-selection theorem) supplies the selection content of the third projection, applied to the same underlying $(S, \varphi)$ with the same trace-out machinery. The observer's confinement to the non-equilibrium phase of $(S, \varphi)$, with C2 read in its timescale form (record persistence and readback there being suppressed, not excluded), is a structural consequence of the effective-mixing hypothesis, for sufficiently fast-mixing equilibria ($\tau_{\rm mix} \ll \tau_S$), in the coarse-grained description generated by iterating the one-step effective channel. The confinement is time-symmetric and fixes where observers can exist, not a direction of time; that the arrows the observer sees descend from it together with horizon growth is not established. The third projection has the same form as the first two: a feature of how the embedded observer sees $(S, \varphi)$ from inside, rather than a feature of $(S, \varphi)$ itself.
 
-A distinction should be kept in view here. The *existence* of the temporal domain — that differentiation recurs, that there is more than one moment at all — is not a projection of $(S, \varphi)$ but the framework's second foundational axiom (the recurrence axiom of the two-axiom observation base; see [Main §1.2] and Part I of the companion paper *Physics Modulo Gauge*). What is a projection, and what this section derives, is the *arrow* — the directedness of time within that domain. The framework derives the arrow; it posits the existence of the domain. Conflating the two would overstate what the projection delivers: §5.4 produces directedness, not the bare existence of succession.
+A distinction should be kept in view here. The *existence* of the temporal domain — that differentiation recurs, that there is more than one moment at all — is not a projection of $(S, \varphi)$ but the framework's second foundational axiom (the recurrence axiom of the two-axiom observation base; see [Main §1.2] and Part I of the companion paper *Physics Modulo Gauge*). What this section addresses is the *arrow* — the directedness of time within that domain. The framework posits the existence of the domain; for the arrow it establishes the conditional and time-symmetric results stated below, and the direction rests on boundary conditions. Conflating the two would overstate what the projection delivers.
 
-**Horizon growth as the structural clock.** The framework's primary arrow is the cosmological horizon entropy $S_{\rm dS}(t) = A(t)/(4 l_p^2)$, monotonically increasing along any observer worldline in the $\Lambda$-dominated phase. The monotonicity is a feature of the causal partition rather than of any local frame: every observer in the accelerating phase agrees that $S_{\rm dS}$ increases, just as every observer in a Schwarzschild geometry agrees on the exterior direction. The horizon clock is observer-independent, continuous, and monotonic for as long as $\ddot a > 0$, which the framework identifies as the regime of interest for the derivations of [GR]. At the substrate level, $(S, \varphi)$ has no preferred direction: $\varphi$ and $\varphi^{-1}$ are equally valid as forward-time maps. At the observer level, the horizon clock picks out a direction, and all other arrows align with it.
+**Horizon growth as the structural clock.** The framework's primary clock is the area $A(t) = 4\pi c^2/H^2$ of the apparent (Hubble) horizon, with the horizon entropy $S_{\rm dS}(t) = A(t)/(4 l_p^2)$ assigned to it at the de Sitter calibration ([GR, §2.3]). On an expanding flat Friedmann-Robertson-Walker background ($H > 0$) satisfying the null energy condition, $\dot H = -4\pi G(\rho + p/c^2) \leq 0$ and $dA/dt = -8\pi c^2\dot H/H^3 \geq 0$: the area is non-decreasing along any comoving observer worldline, strictly increasing while $\rho + p/c^2 > 0$, and constant in exact de Sitter. This is standard general-relativistic background kinematics, computed from the Friedmann equations of the background taken as given ([GR, §8.7]), not a prediction derived from the substratum. The criterion is expansion with the null energy condition, not $\ddot a > 0$: the area also grows in the decelerating era, and is constant in exact de Sitter, where $\ddot a > 0$. Both are inputs: the expanding branch is observed, not derived, and the null energy condition is assumed of the matter content; reading $A/(4 l_p^2)$ as the entropy of a dynamical horizon goes beyond the de Sitter calibration ([GR, §2.3]). The monotonicity holds for every comoving observer of the homogeneous background. At the substrate level, $(S, \varphi)$ has no preferred direction: $\varphi$ and $\varphi^{-1}$ are equally valid as forward-time maps. At the observer level, the horizon parameter orders the expanding branch non-strictly; that the other arrows align with it is not established.
 
-**The layered cascade.** Four derived arrows align with the horizon clock through a structural cascade. *Cosmological:* $S_{\rm dS}(t)$ monotonic in the $\Lambda$-dominated phase, as above. *Thermodynamic:* via Jacobson's coupling $\rho_s = \rho_{\rm crit}$ ([GR, §3]), the entropy of any horizon-bounded system increases with the observer's clock — the thermodynamic arrow descends from the structural arrow without an independent postulate. *Memory:* C2 (memory persistence) implies the observer's hidden sector preserves correlations forward along the clock rather than backward, because the coarse-grained mixing time of the hidden-sector dynamics is uniform in the two directions but the observer's measurement protocol commits to one. *Measurement:* the observer records "past" outcomes and prepares "future" states — an asymmetry inherited from the memory layer, not added on top of it. The chain grounds in one structural fact (horizon growth) and one structural constraint (C1–C4 observer selection via Main §4.6, conditional on its effective-mixing hypothesis and stated for the coarse-grained description generated by iterating the one-step effective channel). The effective-mixing hypothesis is the premise that takes the place of an initial-condition axiom in selecting where observers exist.
+**The layered cascade.** Four arrows are stated against the horizon parameter, with different status. *Cosmological:* $A(t)$ non-decreasing on the expanding branch, as above; the branch is an input, and no result here selects it. *Thermodynamic:* Jacobson's coupling $\rho_s = \rho_{\rm crit}$ ([GR, §7.2]) is an instantaneous identity, even in $H$, between the boundary energy density and $H^2$; it does not relate matter entropy to horizon area. The direction of coarse-grained entropy increase rests on a low-entropy or product boundary condition: from a product state at $t = 0$, the sum of the sector entropies is bounded below by its value at $t = 0$ at every earlier and later time ([GR, §A.6]), a two-sided bound rather than a monotonicity statement, and the generalized second law is open. *Memory:* C2 (memory persistence) is a retention condition; the coarse-grained mixing time of the hidden-sector dynamics is the same in the two directions, and the direction in which records are written and read is fixed by a non-stationary preparation together with the observer's orientation convention. *Measurement:* the observer records "past" outcomes and prepares "future" states — an asymmetry inherited from the memory layer, not added on top of it. That the thermodynamic, memory and measurement arrows align with the horizon parameter through a common mechanism is not established. The chain uses one input (horizon growth on the expanding branch) and one structural constraint (C1–C4 observer selection via Main §4.6, conditional on its effective-mixing hypothesis and stated for the coarse-grained description generated by iterating the one-step effective channel), which fixes where observers exist, not a direction. The effective-mixing hypothesis is the premise that takes the place of an initial-condition axiom in selecting where observers exist.
 
-**What the framework derives, and the directional-language subtlety.** The framework's quantitative content at the level of Main §4.6 is *symmetric* in time direction — the spectral-gap bound $\tau_B(V) \leq \tau_{\rm mix}(\Sigma_{\rm loc}(V)) + O(\epsilon)$, stated for the coarse-grained description generated by iterating the one-step effective channel, applies equally in both directions along any observer worldline. What is *imposed* by the observer, not derived, is the convention of which direction along the horizon-clock parameter to call "future." The two directions are gauge-equivalent at the substrate level ($\varphi \leftrightarrow \varphi^{-1}$), and the observer's choice of convention is the residual freedom after all physical content has been extracted. The framework therefore derives the existence of a monotonic parameter along observer worldlines, not a substrate-level arrow of time. This distinction answers Ellis's objection (a bijective substratum cannot carry an arrow of time) by agreeing with the objection's premise while deriving what the observer actually sees from the structure the substratum does have.
+**What the framework establishes, and the directional-language subtlety.** The framework's quantitative content at the level of Main §4.6 is *symmetric* in time direction — the spectral-gap bound $\tau_B(V) \leq \tau_{\rm mix}(\Sigma_{\rm loc}(V)) + O(\epsilon)$, stated for the coarse-grained description generated by iterating the one-step effective channel, applies equally in both directions along any observer worldline. What is *imposed* by the observer, not derived, is the convention of which direction along the horizon-clock parameter to call "future." The two directions are gauge-equivalent at the substrate level ($\varphi \leftrightarrow \varphi^{-1}$), and the observer's choice of convention is the residual freedom after all physical content has been extracted. What the framework establishes is therefore a horizon parameter that is non-decreasing along observer worldlines on the expanding branch, given the inputs above, not a substrate-level arrow of time. This agrees with the premise of Ellis's objection (a bijective substratum cannot carry an arrow of time); the observed direction does not follow from the results above without the boundary conditions they name.
 
-**Scope note: causal topology as load-bearing.** The memory-arrow step uses "the observer's worldline has a causal topology" as the carrier of the asymmetry. This is not a new axiom — [Main, §1.3] already assumes a causal topology on the partition — but it is load-bearing for the cascade and is noted explicitly here rather than treated as obvious. A framework in which the observer's worldline had no causal topology could not run the memory-arrow step, and the cascade would halt at the thermodynamic layer.
+**Scope note: causal topology as load-bearing.** The memory-arrow step uses "the observer's worldline has a causal topology" as the carrier of its orientation convention. This is not a new axiom — it reduces to the ordering of moments together with $\varphi$ (scope note below) — but it is load-bearing for the cascade and is noted explicitly here rather than treated as obvious. A framework in which the observer's worldline had no causal topology could not state the memory-arrow step.
 
-**Scope note: the arrow derivation is conditional on the ordering of moments.** The cascade derives a *direction* along the observer's worldline. It presupposes that the worldline's moments are *ordered* — that they form a sequence — since a monotonic parameter is monotonic with respect to an ordering. This ordering is distinct from the bare existence of more than one moment: the framework's foundational treatment (Part I of the companion paper *Physics Modulo Gauge*) axiomatizes the existence of a temporal domain but treats the *ordering* of its moments as a separate question, which it leaves open. The arrow derivation of this section is therefore conditional: it derives the direction *given* an ordering, and inherits the open status of the ordering question. What this section establishes is the conditional — ordered moments yield a derived monotonic parameter and an observer-imposed direction-convention — not an unconditional derivation of temporal directedness from the substratum alone. The causal topology of the worldline reduces to this ordering together with the dynamics $\varphi$ (it introduces no commitment beyond the ordering), so the arrow's single open dependency is the ordering question and nothing further.
+**Scope note: the results of this section are conditional on the ordering of moments and on further inputs.** The cascade concerns a *direction* along the observer's worldline. It presupposes that the worldline's moments are *ordered* — that they form a sequence — since a monotonic parameter is monotonic with respect to an ordering. This ordering is distinct from the bare existence of more than one moment: the framework's foundational treatment (Part I of the companion paper *Physics Modulo Gauge*) axiomatizes the existence of a temporal domain but treats the *ordering* of its moments as a separate question, which it leaves open. The results of this section therefore inherit the open status of the ordering question. What this section establishes is conditional — given ordered moments and the expanding branch, a non-decreasing horizon parameter and an observer-imposed direction-convention — not a derivation of temporal directedness from the substratum alone. The causal topology of the worldline reduces to this ordering together with the dynamics $\varphi$ (it introduces no commitment beyond the ordering). The ordering is not the only dependency: the horizon parameter rests on the expanding branch and the null energy condition; the observer selection on the effective-mixing hypothesis and the iterated-channel description; the thermodynamic, memory and measurement directions on a non-stationary or low-entropy preparation and the orientation convention; and the alignment of the arrows is open.
 
-**CP violation as the residual Ellis-class objection.** The layered cascade closes the Ellis-axis objection for the arrow of time but does not automatically close the related question of CP violation. The substratum is T-symmetric (load-bearing for the reciprocity of visible transition counts, [SM, §5]; it does not by itself fix $\bar\theta$), and observed CP violation in CKM and kaon/B systems is a flavor-basis feature of the specific bijection rather than a consequence of the cascade. The framework treats flavor-sector CP phases as solution-specific inputs — see [SM, §5.4] for the scope statement and §6.4 "what does not falsify" for the commitment that the framework makes no prediction of $\delta_{\text{CKM}}$, $\delta_{\text{PMNS}}$, or the Wolfenstein $(\rho, \eta)$.
+**CP violation and the Ellis-class objection.** The related question of CP violation is separate from the cascade. The substratum is T-symmetric (load-bearing for the reciprocity of visible transition counts, [SM, §5]; it does not by itself fix $\bar\theta$), and observed CP violation in CKM and kaon/B systems is a flavor-basis feature of the specific bijection rather than a consequence of the cascade. The framework treats flavor-sector CP phases as solution-specific inputs — see [SM, §5.4] for the scope statement and §6.4 "what does not falsify" for the commitment that the framework makes no prediction of $\delta_{\text{CKM}}$, $\delta_{\text{PMNS}}$, or the Wolfenstein $(\rho, \eta)$.
 
 ### 5.5 What the synthesis does and does not claim
 
diff --git a/papers/Substratum.pdf b/papers/Substratum.pdf
index 713b47ad1..4b0fa3b95 100644
Binary files a/papers/Substratum.pdf and b/papers/Substratum.pdf differ
diff --git a/papers/Substratum.tex b/papers/Substratum.tex
index aa9ff7207..b0a4ce26e 100644
--- a/papers/Substratum.tex
+++ b/papers/Substratum.tex
@@ -200,8 +200,11 @@ \subsection{Abstract}\label{abstract}}
 Quantum mechanics and general relativity are conventionally treated as
 two fundamental theories awaiting unification. This paper develops the
 alternative that they are two projections of a single finite
-deterministic construction, joined by a third projection that produces
-the arrow of time. The underlying object is the bijection
+deterministic construction, joined by a third projection that supplies a
+time-symmetric observer-selection rule, read against the area of the
+apparent horizon, which on an expanding homogeneous general-relativistic
+background satisfying the null energy condition is non-decreasing as a
+matter of background kinematics. The underlying object is the bijection
 \((S,\varphi)\) on a finite configuration space. Its visible-sector
 reduction has the exact finite stochastic/quantum-representation
 equivalence established in {[}Main{]}: finite observable laws, finite
@@ -354,12 +357,14 @@ \subsection{1. Introduction}\label{introduction}}
 another (the metric, \(\hbar\), the BH entropy, the dynamical dark
 energy), and the framework provides the construction that makes both of
 these descriptions exact projections of the same object. The same
-construction produces a third projection (§5.4) that resolves the
-arrow-of-time question by descending the observer's emergent arrows ---
-cosmological, thermodynamic, memory, measurement --- from the
-horizon-growth clock and the C1--C4 observer-selection theorem of
-{[}Main, §4.6{]}, answering the Ellis objection to bijective substrates
-without adding a structural commitment.
+construction gives a third projection (§5.4): the C1--C4
+observer-selection theorem of {[}Main, §4.6{]}, which is time-symmetric
+and selects where observers can exist, read against the apparent-horizon
+area, which is non-decreasing on an expanding homogeneous
+general-relativistic background satisfying the null energy condition
+(background kinematics, not a substratum prediction); a common mechanism
+aligning the cosmological, thermodynamic, memory and measurement arrows
+is not established.
 
 The paper is organized as follows. §1.1 states the framework's primary
 metaphysical commitment and its relation to explanatory and empirical
@@ -554,8 +559,9 @@ \subsection{2. The Physical Interpretation of (S,
 low-entropy ``past hypothesis'', in its role of selecting
 observer-compatible configurations, with a structural observer-selection
 rule and excludes Boltzmann brains structurally rather than
-anthropically. Entropy increase is then emergent as a coarse-grained
-description of the non-equilibrium dynamics observers necessarily see.
+anthropically. The rule is time-symmetric: the direction in which
+coarse-grained entropy increases rests on a low-entropy or product
+boundary condition, not on the rule (§5.4).
 
 \textbf{The incompleteness family.} The framework's central result
 belongs to a family of impossibility-with-structure theorems. Gödel: a
@@ -2100,9 +2106,9 @@ \subsection{5. Synthesis: Three Projections of (S,
 the existence of a measurement-independent Bell-violating completion;
 Bell-inclusive uniqueness is open. The substratum gauge group (§4) makes
 the equivalence relation precise. This section makes the synthesis claim
-explicit: quantum mechanics, general relativity, and the arrow of time
-are not three independent theories but three projections of the same
-finite deterministic construction.
+explicit: quantum mechanics and general relativity are not two
+independent theories but projections of the same finite deterministic
+construction, and a third projection bears on the arrow of time (§5.4).
 
 \hypertarget{the-three-projections}{%
 \subsubsection{5.1 The three projections}\label{the-three-projections}}
@@ -2112,10 +2118,11 @@ \subsubsection{5.1 The three projections}\label{the-three-projections}}
 projection that produces the reduced description admitting the quantum
 representation ({[}Main{]}) and the Standard Model ({[}SM{]}); a
 boundary-thermodynamic projection that produces general relativity
-({[}GR{]}); and an arrow-of-time projection (§5.4) that produces the
-emergent cosmological, thermodynamic, memory, and measurement arrows
-from a single structural clock and the C1--C4 observer-selection theorem
-of {[}Main, §4.6{]}.
+({[}GR{]}); and an arrow-of-time projection (§5.4) that supplies the
+time-symmetric C1--C4 observer-selection theorem of {[}Main, §4.6{]},
+read against the apparent-horizon area, which is non-decreasing, as
+background kinematics, on an expanding homogeneous general-relativistic
+background satisfying the null energy condition.
 
 \textbf{Visible-sector projection.} At the level of the embedded
 observer's epistemic access, the trace-out over the hidden sector
@@ -2201,8 +2208,8 @@ \subsubsection{5.1 The three projections}\label{the-three-projections}}
 the Standard Model; the boundary-thermodynamic projection (§5.1) works
 on the classical thermal data at the partition boundary and produces
 general relativity; the arrow-of-time projection (§5.4) works on the
-horizon-clock parameter along observer worldlines and produces the
-cascade of emergent arrows.
+horizon-area parameter along observer worldlines together with the
+observer-selection rule.
 
 \textbf{The projections and the Poincaré structure of the partition.}
 The three projections are organized by how the
@@ -2219,15 +2226,17 @@ \subsubsection{5.1 The three projections}\label{the-three-projections}}
 spatial translations, which merely relabel the observer's origin, are
 broken only observer-relatively and are recovered as spatial homogeneity
 --- the equivalence of all worldline-centered descriptions; and time
-translation is broken intrinsically by the growth of the horizon, which
-is the boundary-thermodynamic and arrow projections (§5.4, {[}GR{]}).
-The partition's symmetry-breaking signature --- isotropy not broken by
-the partition, a preferred rest frame, spatial homogeneity, a
-distinguished cosmic time --- is the symmetry structure of a
-Friedmann-Robertson-Walker cosmology. The framework does not derive that
-structure; it exhibits the partition's symmetry-breaking pattern as
-compatible with it, with ordinary momentum conservation holding in the
-visible sector up to corrections of order (length / horizon radius).
+translation is broken by the growth of the apparent horizon on an
+expanding background with \(\rho + p/c^2 > 0\), an input to the
+boundary-thermodynamic and arrow projections; in exact de Sitter the
+horizon area is constant (§5.4, {[}GR{]}). The partition's
+symmetry-breaking signature --- isotropy not broken by the partition, a
+preferred rest frame, spatial homogeneity, a distinguished cosmic time
+--- is the symmetry structure of a Friedmann-Robertson-Walker cosmology.
+The framework does not derive that structure; it exhibits the
+partition's symmetry-breaking pattern as compatible with it, with
+ordinary momentum conservation holding in the visible sector up to
+corrections of order (length / horizon radius).
 
 \hypertarget{the-qm-gr-incompatibility-as-a-category-error}{%
 \subsubsection{5.2 The QM-GR incompatibility as a category
@@ -2341,74 +2350,89 @@ \subsubsection{5.3 The three-level hierarchy made
 \subsubsection{5.4 The arrow of time as a third
 projection}\label{the-arrow-of-time-as-a-third-projection}}
 
-After Main §4.6 (structural observer-selection theorem), the arrow of
-time joins quantum mechanics and general relativity as a third emergent
-structure that the framework derives from the same underlying
-\((S, \varphi)\) and the same trace-out machinery. Before §4.6, the
-arrow of time was treated as external to the derivation chain --- a past
-hypothesis or an anthropic selection, imposed on the framework rather
-than produced by it. With §4.6 in place, the observer's confinement to
-the non-equilibrium phase of \((S, \varphi)\), with C2 read in its
-timescale form (record persistence and readback there being suppressed,
-not excluded), is a structural consequence of the effective-mixing
-hypothesis, for sufficiently fast-mixing equilibria
+Main §4.6 (structural observer-selection theorem) supplies the selection
+content of the third projection, applied to the same underlying
+\((S, \varphi)\) with the same trace-out machinery. The observer's
+confinement to the non-equilibrium phase of \((S, \varphi)\), with C2
+read in its timescale form (record persistence and readback there being
+suppressed, not excluded), is a structural consequence of the
+effective-mixing hypothesis, for sufficiently fast-mixing equilibria
 (\(\tau_{\rm mix} \ll \tau_S\)), in the coarse-grained description
-generated by iterating the one-step effective channel, and every arrow
-the observer sees descends from this confinement together with the
-horizon growth that the accelerating phase produces. The third
-projection has the same form as the first two: a feature of how the
-embedded observer sees \((S, \varphi)\) from inside, rather than a
-feature of \((S, \varphi)\) itself.
+generated by iterating the one-step effective channel. The confinement
+is time-symmetric and fixes where observers can exist, not a direction
+of time; that the arrows the observer sees descend from it together with
+horizon growth is not established. The third projection has the same
+form as the first two: a feature of how the embedded observer sees
+\((S, \varphi)\) from inside, rather than a feature of \((S, \varphi)\)
+itself.
 
 A distinction should be kept in view here. The \emph{existence} of the
 temporal domain --- that differentiation recurs, that there is more than
 one moment at all --- is not a projection of \((S, \varphi)\) but the
 framework's second foundational axiom (the recurrence axiom of the
 two-axiom observation base; see {[}Main §1.2{]} and Part I of the
-companion paper \emph{Physics Modulo Gauge}). What is a projection, and
-what this section derives, is the \emph{arrow} --- the directedness of
-time within that domain. The framework derives the arrow; it posits the
-existence of the domain. Conflating the two would overstate what the
-projection delivers: §5.4 produces directedness, not the bare existence
-of succession.
+companion paper \emph{Physics Modulo Gauge}). What this section
+addresses is the \emph{arrow} --- the directedness of time within that
+domain. The framework posits the existence of the domain; for the arrow
+it establishes the conditional and time-symmetric results stated below,
+and the direction rests on boundary conditions. Conflating the two would
+overstate what the projection delivers.
 
 \textbf{Horizon growth as the structural clock.} The framework's primary
-arrow is the cosmological horizon entropy
-\(S_{\rm dS}(t) = A(t)/(4 l_p^2)\), monotonically increasing along any
-observer worldline in the \(\Lambda\)-dominated phase. The monotonicity
-is a feature of the causal partition rather than of any local frame:
-every observer in the accelerating phase agrees that \(S_{\rm dS}\)
-increases, just as every observer in a Schwarzschild geometry agrees on
-the exterior direction. The horizon clock is observer-independent,
-continuous, and monotonic for as long as \(\ddot a > 0\), which the
-framework identifies as the regime of interest for the derivations of
-{[}GR{]}. At the substrate level, \((S, \varphi)\) has no preferred
-direction: \(\varphi\) and \(\varphi^{-1}\) are equally valid as
-forward-time maps. At the observer level, the horizon clock picks out a
-direction, and all other arrows align with it.
-
-\textbf{The layered cascade.} Four derived arrows align with the horizon
-clock through a structural cascade. \emph{Cosmological:}
-\(S_{\rm dS}(t)\) monotonic in the \(\Lambda\)-dominated phase, as
-above. \emph{Thermodynamic:} via Jacobson's coupling
-\(\rho_s = \rho_{\rm crit}\) ({[}GR, §3{]}), the entropy of any
-horizon-bounded system increases with the observer's clock --- the
-thermodynamic arrow descends from the structural arrow without an
-independent postulate. \emph{Memory:} C2 (memory persistence) implies
-the observer's hidden sector preserves correlations forward along the
-clock rather than backward, because the coarse-grained mixing time of
-the hidden-sector dynamics is uniform in the two directions but the
-observer's measurement protocol commits to one. \emph{Measurement:} the
-observer records ``past'' outcomes and prepares ``future'' states --- an
-asymmetry inherited from the memory layer, not added on top of it. The
-chain grounds in one structural fact (horizon growth) and one structural
+clock is the area \(A(t) = 4\pi c^2/H^2\) of the apparent (Hubble)
+horizon, with the horizon entropy \(S_{\rm dS}(t) = A(t)/(4 l_p^2)\)
+assigned to it at the de Sitter calibration ({[}GR, §2.3{]}). On an
+expanding flat Friedmann-Robertson-Walker background (\(H > 0\))
+satisfying the null energy condition,
+\(\dot H = -4\pi G(\rho + p/c^2) \leq 0\) and
+\(dA/dt = -8\pi c^2\dot H/H^3 \geq 0\): the area is non-decreasing along
+any comoving observer worldline, strictly increasing while
+\(\rho + p/c^2 > 0\), and constant in exact de Sitter. This is standard
+general-relativistic background kinematics, computed from the Friedmann
+equations of the background taken as given ({[}GR, §8.7{]}), not a
+prediction derived from the substratum. The criterion is expansion with
+the null energy condition, not \(\ddot a > 0\): the area also grows in
+the decelerating era, and is constant in exact de Sitter, where
+\(\ddot a > 0\). Both are inputs: the expanding branch is observed, not
+derived, and the null energy condition is assumed of the matter content;
+reading \(A/(4 l_p^2)\) as the entropy of a dynamical horizon goes
+beyond the de Sitter calibration ({[}GR, §2.3{]}). The monotonicity
+holds for every comoving observer of the homogeneous background. At the
+substrate level, \((S, \varphi)\) has no preferred direction:
+\(\varphi\) and \(\varphi^{-1}\) are equally valid as forward-time maps.
+At the observer level, the horizon parameter orders the expanding branch
+non-strictly; that the other arrows align with it is not established.
+
+\textbf{The layered cascade.} Four arrows are stated against the horizon
+parameter, with different status. \emph{Cosmological:} \(A(t)\)
+non-decreasing on the expanding branch, as above; the branch is an
+input, and no result here selects it. \emph{Thermodynamic:} Jacobson's
+coupling \(\rho_s = \rho_{\rm crit}\) ({[}GR, §7.2{]}) is an
+instantaneous identity, even in \(H\), between the boundary energy
+density and \(H^2\); it does not relate matter entropy to horizon area.
+The direction of coarse-grained entropy increase rests on a low-entropy
+or product boundary condition: from a product state at \(t = 0\), the
+sum of the sector entropies is bounded below by its value at \(t = 0\)
+at every earlier and later time ({[}GR, §A.6{]}), a two-sided bound
+rather than a monotonicity statement, and the generalized second law is
+open. \emph{Memory:} C2 (memory persistence) is a retention condition;
+the coarse-grained mixing time of the hidden-sector dynamics is the same
+in the two directions, and the direction in which records are written
+and read is fixed by a non-stationary preparation together with the
+observer's orientation convention. \emph{Measurement:} the observer
+records ``past'' outcomes and prepares ``future'' states --- an
+asymmetry inherited from the memory layer, not added on top of it. That
+the thermodynamic, memory and measurement arrows align with the horizon
+parameter through a common mechanism is not established. The chain uses
+one input (horizon growth on the expanding branch) and one structural
 constraint (C1--C4 observer selection via Main §4.6, conditional on its
 effective-mixing hypothesis and stated for the coarse-grained
-description generated by iterating the one-step effective channel). The
-effective-mixing hypothesis is the premise that takes the place of an
-initial-condition axiom in selecting where observers exist.
+description generated by iterating the one-step effective channel),
+which fixes where observers exist, not a direction. The effective-mixing
+hypothesis is the premise that takes the place of an initial-condition
+axiom in selecting where observers exist.
 
-\textbf{What the framework derives, and the directional-language
+\textbf{What the framework establishes, and the directional-language
 subtlety.} The framework's quantitative content at the level of Main
 §4.6 is \emph{symmetric} in time direction --- the spectral-gap bound
 \(\tau_B(V) \leq \tau_{\rm mix}(\Sigma_{\rm loc}(V)) + O(\epsilon)\),
@@ -2419,53 +2443,57 @@ \subsubsection{5.4 The arrow of time as a third
 to call ``future.'' The two directions are gauge-equivalent at the
 substrate level (\(\varphi \leftrightarrow \varphi^{-1}\)), and the
 observer's choice of convention is the residual freedom after all
-physical content has been extracted. The framework therefore derives the
-existence of a monotonic parameter along observer worldlines, not a
-substrate-level arrow of time. This distinction answers Ellis's
-objection (a bijective substratum cannot carry an arrow of time) by
-agreeing with the objection's premise while deriving what the observer
-actually sees from the structure the substratum does have.
+physical content has been extracted. What the framework establishes is
+therefore a horizon parameter that is non-decreasing along observer
+worldlines on the expanding branch, given the inputs above, not a
+substrate-level arrow of time. This agrees with the premise of Ellis's
+objection (a bijective substratum cannot carry an arrow of time); the
+observed direction does not follow from the results above without the
+boundary conditions they name.
 
 \textbf{Scope note: causal topology as load-bearing.} The memory-arrow
 step uses ``the observer's worldline has a causal topology'' as the
-carrier of the asymmetry. This is not a new axiom --- {[}Main, §1.3{]}
-already assumes a causal topology on the partition --- but it is
-load-bearing for the cascade and is noted explicitly here rather than
-treated as obvious. A framework in which the observer's worldline had no
-causal topology could not run the memory-arrow step, and the cascade
-would halt at the thermodynamic layer.
-
-\textbf{Scope note: the arrow derivation is conditional on the ordering
-of moments.} The cascade derives a \emph{direction} along the observer's
-worldline. It presupposes that the worldline's moments are
-\emph{ordered} --- that they form a sequence --- since a monotonic
-parameter is monotonic with respect to an ordering. This ordering is
-distinct from the bare existence of more than one moment: the
-framework's foundational treatment (Part I of the companion paper
-\emph{Physics Modulo Gauge}) axiomatizes the existence of a temporal
-domain but treats the \emph{ordering} of its moments as a separate
-question, which it leaves open. The arrow derivation of this section is
-therefore conditional: it derives the direction \emph{given} an
-ordering, and inherits the open status of the ordering question. What
-this section establishes is the conditional --- ordered moments yield a
-derived monotonic parameter and an observer-imposed direction-convention
---- not an unconditional derivation of temporal directedness from the
-substratum alone. The causal topology of the worldline reduces to this
-ordering together with the dynamics \(\varphi\) (it introduces no
-commitment beyond the ordering), so the arrow's single open dependency
-is the ordering question and nothing further.
-
-\textbf{CP violation as the residual Ellis-class objection.} The layered
-cascade closes the Ellis-axis objection for the arrow of time but does
-not automatically close the related question of CP violation. The
-substratum is T-symmetric (load-bearing for the reciprocity of visible
-transition counts, {[}SM, §5{]}; it does not by itself fix
-\(\bar\theta\)), and observed CP violation in CKM and kaon/B systems is
-a flavor-basis feature of the specific bijection rather than a
-consequence of the cascade. The framework treats flavor-sector CP phases
-as solution-specific inputs --- see {[}SM, §5.4{]} for the scope
-statement and §6.4 ``what does not falsify'' for the commitment that the
-framework makes no prediction of \(\delta_{\text{CKM}}\),
+carrier of its orientation convention. This is not a new axiom --- it
+reduces to the ordering of moments together with \(\varphi\) (scope note
+below) --- but it is load-bearing for the cascade and is noted
+explicitly here rather than treated as obvious. A framework in which the
+observer's worldline had no causal topology could not state the
+memory-arrow step.
+
+\textbf{Scope note: the results of this section are conditional on the
+ordering of moments and on further inputs.} The cascade concerns a
+\emph{direction} along the observer's worldline. It presupposes that the
+worldline's moments are \emph{ordered} --- that they form a sequence ---
+since a monotonic parameter is monotonic with respect to an ordering.
+This ordering is distinct from the bare existence of more than one
+moment: the framework's foundational treatment (Part I of the companion
+paper \emph{Physics Modulo Gauge}) axiomatizes the existence of a
+temporal domain but treats the \emph{ordering} of its moments as a
+separate question, which it leaves open. The results of this section
+therefore inherit the open status of the ordering question. What this
+section establishes is conditional --- given ordered moments and the
+expanding branch, a non-decreasing horizon parameter and an
+observer-imposed direction-convention --- not a derivation of temporal
+directedness from the substratum alone. The causal topology of the
+worldline reduces to this ordering together with the dynamics
+\(\varphi\) (it introduces no commitment beyond the ordering). The
+ordering is not the only dependency: the horizon parameter rests on the
+expanding branch and the null energy condition; the observer selection
+on the effective-mixing hypothesis and the iterated-channel description;
+the thermodynamic, memory and measurement directions on a non-stationary
+or low-entropy preparation and the orientation convention; and the
+alignment of the arrows is open.
+
+\textbf{CP violation and the Ellis-class objection.} The related
+question of CP violation is separate from the cascade. The substratum is
+T-symmetric (load-bearing for the reciprocity of visible transition
+counts, {[}SM, §5{]}; it does not by itself fix \(\bar\theta\)), and
+observed CP violation in CKM and kaon/B systems is a flavor-basis
+feature of the specific bijection rather than a consequence of the
+cascade. The framework treats flavor-sector CP phases as
+solution-specific inputs --- see {[}SM, §5.4{]} for the scope statement
+and §6.4 ``what does not falsify'' for the commitment that the framework
+makes no prediction of \(\delta_{\text{CKM}}\),
 \(\delta_{\text{PMNS}}\), or the Wolfenstein \((\rho, \eta)\).
 
 \hypertarget{what-the-synthesis-does-and-does-not-claim}{%
@@ -3303,4 +3331,4 @@ \subsection{References}\label{references}}
 (2026).
 
 \end{document}
-% source-sha256: 94a21b775531457c91af8e79b4f9dc561de3ba64feeedd02a9b53b1d1c746162
+% source-sha256: 986b259db5439232a985de28e38f52b85569f7bd5d12aadc648b233a20f7a0f5
diff --git a/verification/audit-census.json b/verification/audit-census.json
index dfac8b943..2ae8ed425 100644
--- a/verification/audit-census.json
+++ b/verification/audit-census.json
@@ -156,9 +156,10 @@
             "papers/GR.md": 2,
             "book/ch01-observation.md": 1,
             "book/ch07-gravity.md": 2,
-            "book/The-Incompleteness-of-Observation-FULL.md": 3
+            "book/ch18-beyond.md": 1,
+            "book/The-Incompleteness-of-Observation-FULL.md": 4
           },
-          "why": "the status sentence in running text: GR's prerequisites and 8.4, the book's 7.1 and 7.2, and since B5 the paragraph after Chapter 1's C4 definition, in both parallel sources. GR 2.2's own entry closes with the capitalized verdict form, counted separately. A count below these means a mirror lost it."
+          "why": "the status sentence in running text: GR's prerequisites and 8.4, the book's 7.1 and 7.2, and since B5 the paragraph after Chapter 1's C4 definition, in both parallel sources; since ARROW-1 (#1112) also the cosmological-arrow paragraph of Chapter 18 section 18.8, in both parallel sources. GR 2.2's own entry closes with the capitalized verdict form, counted separately. A count below these means a mirror lost it."
         },
         {
           "id": "B1-hbar-not-consuming",
diff --git a/verification/audits/manuscript/arrow-1-20261105/CHANGES.md b/verification/audits/manuscript/arrow-1-20261105/CHANGES.md
new file mode 100644
index 000000000..9918bb778
--- /dev/null
+++ b/verification/audits/manuscript/arrow-1-20261105/CHANGES.md
@@ -0,0 +1,407 @@
+# ARROW-1 Phase 2 change record
+
+**Status: WIP / UNVERIFIED.** This record awaits independent scientific and minimum-change
+review. Nothing here is merged.
+
+- **Base:** `f5c281ae10a913ec68fa41b3b839ad0b164979cd`.
+- **Authorization:**
+  - #1112 c.6649271094: Option 3, the plan in c.6643840329, the side corrections.
+  - #1112 c.6649643759 (amaybaum, 2026-11-05): the Main:744 label phrase, the
+    `LEDGER.json` `id`/`label`/`fingerprint`, ch18:190, the ch02:192 and ch04:32
+    reclassification, and Main:760(iii) NO CHANGE.
+- **Basis:** `RECOVERY.md`, as reviewed in #1113 c.6649600806 and c.6649698455.
+- **Commits:**
+  - `e133ffbc`: record;
+  - `d3e13c99`: Main and LEDGER;
+  - `eb74b3e7`: papers;
+  - `4081cba0`: book and FULL;
+  - `e4ab279a`: regenerated `.tex`/`.pdf`;
+  - `24452b01`: this record;
+  - `6010cf44`: `audit-census.json` `B1-required` counts, accepted by ruling A4;
+  - `bd926c76`: Substratum:405 FLRW → Friedmann-Robertson-Walker;
+  - `f6f7dd13`: follow-up edits under ruling c.6650301410, with rebuilds;
+  - `a762d418`: SC-42 result regenerated (ruling A3);
+  - `98d04808`: record update;
+  - `ca21185e`: ch18:214 second sentence under ruling c.6651202921, with the book rebuild;
+  - then this record update.
+
+**How to read the table.**
+- Each row is one fragment replacement. Its old text is unique in the file at the base, and
+  each replacement stays within one paragraph line. Line numbers are therefore the same
+  before and after.
+- Each book row is applied identically to the FULL mirror. After the edits, the chapter
+  paragraph and the FULL paragraph are byte-identical.
+- Class labels:
+  - *retain*: the derived claim stays, with its hypotheses.
+  - *conditional*: the conditions are named.
+  - *unresolved*: the assertion is removed and the open status stated.
+  - *side-correction*: owner-listed, c.6649271094.
+  - *owner-ruled*: c.6649643759.
+- RECOVERY references are to sections of `RECOVERY.md`.
+
+## Site-by-site changes
+
+| # | Site (before → after) | Mirror | Class | RECOVERY | Old text | New text |
+|---|---|---|---|---|---|---|
+| 1 | `papers/Substratum.md`:14 → :14 | — | unresolved | B.8, C.3, E.2 | joined by a third projection that produces the arrow of time. | joined by a third projection that supplies a time-symmetric observer-selection rule and, on the expanding cosmological branch, a non-decreasing horizon-area parameter. |
+| 2 | `papers/Substratum.md`:26 → :26 | — | unresolved | B.7, B.8 | The same construction produces a third projection (§5.4) that resolves the arrow-of-time question by descending the observer's emergent arrows — cosmological, thermodynamic, memory, measurement — from the horizon-growth clock and the C1–C4 observer-selection theorem of [Main, §4.6], answering the Ellis objection to bijective substrates without adding a structural commitment. | The same construction gives a third projection (§5.4): the C1–C4 observer-selection theorem of [Main, §4.6], which is time-symmetric and selects where observers can exist, together with a horizon-area parameter that is non-decreasing on the expanding cosmological branch; a common mechanism aligning the cosmological, thermodynamic, memory and measurement arrows is not established. |
+| 3 | `papers/Substratum.md`:56 → :56 | — | conditional | B.3, B.9, C.3 | Entropy increase is then emergent as a coarse-grained description of the non-equilibrium dynamics observers necessarily see. | The rule is time-symmetric: the direction in which coarse-grained entropy increases is fixed by a low-entropy or product boundary condition, not by the rule (§5.4). |
+| 4 | `papers/Substratum.md`:367 → :367 | — | conditional | B.8 | This section makes the synthesis claim explicit: quantum mechanics, general relativity, and the arrow of time are not three independent theories but three projections of the same finite deterministic construction. | This section makes the synthesis claim explicit: quantum mechanics and general relativity are not two independent theories but projections of the same finite deterministic construction, and a third projection bears on the arrow of time (§5.4). |
+| 5 | `papers/Substratum.md`:371 → :371 | — | conditional | B.1, B.8, B.9 | and an arrow-of-time projection (§5.4) that produces the emergent cosmological, thermodynamic, memory, and measurement arrows from a single structural clock and the C1–C4 observer-selection theorem of [Main, §4.6]. | and an arrow-of-time projection (§5.4) that supplies the time-symmetric C1–C4 observer-selection theorem of [Main, §4.6] and, on the expanding cosmological branch, a non-decreasing horizon-area parameter. |
+| 6 | `papers/Substratum.md`:377 → :377 | — | conditional | B.7, B.8 | the arrow-of-time projection (§5.4) works on the horizon-clock parameter along observer worldlines and produces the cascade of emergent arrows. | the arrow-of-time projection (§5.4) works on the horizon-area parameter along observer worldlines together with the observer-selection rule. |
+| 7 | `papers/Substratum.md`:379 → :379 | — | conditional | B.1, B.2, E.2, E.4 | and time translation is broken intrinsically by the growth of the horizon, which is the boundary-thermodynamic and arrow projections (§5.4, [GR]). | and time translation is broken by the growth of the apparent horizon on an expanding background with $\rho + p/c^2 > 0$, an input to the boundary-thermodynamic and arrow projections; in exact de Sitter the horizon area is constant (§5.4, [GR]). |
+| 8 | `papers/Substratum.md`:401 → :401 | — | conditional | B.8, B.9; §A.33 history voice | After Main §4.6 (structural observer-selection theorem), the arrow of time joins quantum mechanics and general relativity as a third emergent structure that the framework derives from the same underlying $(S, \varphi)$ and the same trace-out machinery. Before §4.6, the arrow of time was treated as external to the derivation chain — a past hypothesis or an anthropic selection, imposed on the framework rather than produced by it. With §4.6 in place, the observer's confinement | Main §4.6 (structural observer-selection theorem) supplies the selection content of the third projection, applied to the same underlying $(S, \varphi)$ with the same trace-out machinery. The observer's confinement |
+| 9 | `papers/Substratum.md`:401 → :401 | — | unresolved | B.7, C.3 | in the coarse-grained description generated by iterating the one-step effective channel, and every arrow the observer sees descends from this confinement together with the horizon growth that the accelerating phase produces. The third projection | in the coarse-grained description generated by iterating the one-step effective channel. The confinement is time-symmetric and fixes where observers can exist, not a direction of time; that the arrows the observer sees descend from it together with horizon growth is not established. The third projection |
+| 10 | `papers/Substratum.md`:403 → :403 | — | unresolved | B.8 | What is a projection, and what this section derives, is the *arrow* — the directedness of time within that domain. The framework derives the arrow; it posits the existence of the domain. Conflating the two would overstate what the projection delivers: §5.4 produces directedness, not the bare existence of succession. | What this section addresses is the *arrow* — the directedness of time within that domain. The framework posits the existence of the domain; for the arrow it establishes the conditional and time-symmetric results stated below, and the direction rests on boundary conditions. Conflating the two would overstate what the projection delivers. |
+| 11 | `papers/Substratum.md`:405 → :405 | — | conditional | B.1, B.2, B.7, E.1–E.4 | **Horizon growth as the structural clock.** The framework's primary arrow is the cosmological horizon entropy $S_{\rm dS}(t) = A(t)/(4 l_p^2)$, monotonically increasing along any observer worldline in the $\Lambda$-dominated phase. The monotonicity is a feature of the causal partition rather than of any local frame: every observer in the accelerating phase agrees that $S_{\rm dS}$ increases, just as every observer in a Schwarzschild geometry agrees on the exterior direction. The horizon clock is observer-independent, continuous, and monotonic for as long as $\ddot a > 0$, which the framework identifies as the regime of interest for the derivations of [GR]. At the substrate level, $(S, \varphi)$ has no preferred direction: $\varphi$ and $\varphi^{-1}$ are equally valid as forward-time maps. At the observer level, the horizon clock picks out a direction, and all other arrows align with it. | **Horizon growth as the structural clock.** The framework's primary clock is the area $A(t) = 4\pi c^2/H^2$ of the apparent (Hubble) horizon, with the horizon entropy $S_{\rm dS}(t) = A(t)/(4 l_p^2)$ assigned to it at the de Sitter calibration ([GR, §2.3]). On an expanding flat Friedmann-Robertson-Walker background ($H > 0$) satisfying the null energy condition, $dA/dt = -8\pi c^2\dot H/H^3 \geq 0$: the area is non-decreasing along any comoving observer worldline, strictly increasing while $\rho + p/c^2 > 0$, and constant in exact de Sitter. The criterion is expansion with the null energy condition, not $\ddot a > 0$: the area also grows in the decelerating era, and is constant in exact de Sitter, where $\ddot a > 0$. Both conditions are inputs: the expanding branch is observed, not derived, and reading $A/(4 l_p^2)$ as the entropy of a dynamical horizon goes beyond the de Sitter calibration ([GR, §2.3]). The monotonicity is a feature of the causal partition rather than of any local frame, just as every observer in a Schwarzschild geometry agrees on the exterior direction. At the substrate level, $(S, \varphi)$ has no preferred direction: $\varphi$ and $\varphi^{-1}$ are equally valid as forward-time maps. At the observer level, the horizon parameter orders the expanding branch; that the other arrows align with it is not established. |
+| 12 | `papers/Substratum.md`:407 → :407 | — | conditional | B.2–B.7, C.4; side correction (citation) | **The layered cascade.** Four derived arrows align with the horizon clock through a structural cascade. *Cosmological:* $S_{\rm dS}(t)$ monotonic in the $\Lambda$-dominated phase, as above. *Thermodynamic:* via Jacobson's coupling $\rho_s = \rho_{\rm crit}$ ([GR, §3]), the entropy of any horizon-bounded system increases with the observer's clock — the thermodynamic arrow descends from the structural arrow without an independent postulate. *Memory:* C2 (memory persistence) implies the observer's hidden sector preserves correlations forward along the clock rather than backward, because the coarse-grained mixing time of the hidden-sector dynamics is uniform in the two directions but the observer's measurement protocol commits to one. *Measurement:* the observer records "past" outcomes and prepares "future" states — an asymmetry inherited from the memory layer, not added on top of it. The chain grounds in one structural fact (horizon growth) and one structural constraint (C1–C4 observer selection via Main §4.6, conditional on its effective-mixing hypothesis and stated for the coarse-grained description generated by iterating the one-step effective channel). | **The layered cascade.** Four arrows are stated against the horizon parameter, with different status. *Cosmological:* $A(t)$ non-decreasing on the expanding branch, as above; the branch is an input, and no result here selects it. *Thermodynamic:* Jacobson's coupling $\rho_s = \rho_{\rm crit}$ ([GR, §7.2]) is an instantaneous identity, even in $H$, between the boundary energy density and $H^2$; it does not relate matter entropy to horizon area. The direction of coarse-grained entropy increase rests on a low-entropy or product boundary condition: from a product state at $t = 0$, the sum of the sector entropies is bounded below by its value at $t = 0$ at every earlier and later time ([GR, §A.6]), a two-sided bound rather than a monotonicity statement, and the generalized second law is open. *Memory:* C2 (memory persistence) is a retention condition; the coarse-grained mixing time of the hidden-sector dynamics is the same in the two directions, and the direction in which records are written and read is fixed by a non-stationary preparation together with the observer's orientation convention. *Measurement:* the observer records "past" outcomes and prepares "future" states — an asymmetry inherited from the memory layer, not added on top of it. That the thermodynamic, memory and measurement arrows align with the horizon parameter through a common mechanism is not established. The chain uses one input (horizon growth on the expanding branch) and one structural constraint (C1–C4 observer selection via Main §4.6, conditional on its effective-mixing hypothesis and stated for the coarse-grained description generated by iterating the one-step effective channel), which fixes where observers exist, not a direction. |
+| 13 | `papers/Substratum.md`:409 → :409 | — | conditional / unresolved | B.1, B.8, C.1 | The framework therefore derives the existence of a monotonic parameter along observer worldlines, not a substrate-level arrow of time. This distinction answers Ellis's objection (a bijective substratum cannot carry an arrow of time) by agreeing with the objection's premise while deriving what the observer actually sees from the structure the substratum does have. | What the framework establishes is therefore a horizon parameter that is non-decreasing along observer worldlines on the expanding branch, given the inputs above, not a substrate-level arrow of time. This agrees with the premise of Ellis's objection (a bijective substratum cannot carry an arrow of time); the observed direction does not follow from the results above without the boundary conditions they name. |
+| 14 | `papers/Substratum.md`:411 → :411 | — | side-correction | README side corrections; B.10 | This is not a new axiom — [Main, §1.3] already assumes a causal topology on the partition — but it is load-bearing | This is not a new axiom — it reduces to the ordering of moments together with $\varphi$ (scope note below) — but it is load-bearing |
+| 15 | `papers/Substratum.md`:413 → :413 | — | conditional | B.10 | **Scope note: the arrow derivation is conditional on the ordering of moments.** The cascade derives a *direction* along the observer's worldline. It presupposes that the worldline's moments are *ordered* — that they form a sequence — since a monotonic parameter is monotonic with respect to an ordering. | **Scope note: the results of this section are conditional on the ordering of moments and on further inputs.** The cascade concerns a *direction* along the observer's worldline. It presupposes that the worldline's moments are *ordered* — that they form a sequence — since a monotonic parameter is monotonic with respect to an ordering. |
+| 16 | `papers/Substratum.md`:413 → :413 | — | conditional | B.10 | The arrow derivation of this section is therefore conditional: it derives the direction *given* an ordering, and inherits the open status of the ordering question. What this section establishes is the conditional — ordered moments yield a derived monotonic parameter and an observer-imposed direction-convention — not an unconditional derivation of temporal directedness from the substratum alone. The causal topology of the worldline reduces to this ordering together with the dynamics $\varphi$ (it introduces no commitment beyond the ordering), so the arrow's single open dependency is the ordering question and nothing further. | The results of this section therefore inherit the open status of the ordering question. What this section establishes is conditional — given ordered moments and the expanding branch, a non-decreasing horizon parameter and an observer-imposed direction-convention — not a derivation of temporal directedness from the substratum alone. The causal topology of the worldline reduces to this ordering together with the dynamics $\varphi$ (it introduces no commitment beyond the ordering). The ordering is not the only dependency: the horizon parameter rests on the expanding branch and the null energy condition; the observer selection on the effective-mixing hypothesis and the iterated-channel description; the thermodynamic, memory and measurement directions on a non-stationary or low-entropy preparation and the orientation convention; and the alignment of the arrows is open. |
+| 17 | `papers/Substratum.md`:415 → :415 | — | unresolved | B.8 | **CP violation as the residual Ellis-class objection.** The layered cascade closes the Ellis-axis objection for the arrow of time but does not automatically close the related question of CP violation. | **CP violation and the Ellis-class objection.** The related question of CP violation is separate from the cascade. |
+| 18 | `papers/GR.md`:32 → :32 | — | conditional | B.8 | The reconstruction theorem and the synthesis claim that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same $(S, \varphi)$ are developed in [Substratum]. | The reconstruction theorem and the synthesis claim that quantum mechanics and general relativity descend as projections of the same $(S, \varphi)$, with a third projection bearing on the arrow of time, are developed in [Substratum]. |
+| 19 | `papers/GR.md`:214 → :214 | — | side-correction | README side corrections | and whether its derived classical thermodynamic arrow — $\beta_E = \partial S_H/\partial E > 0$, settled classically by exact counting (kernel: `counting_passive`) — enters that completion | and whether its classical thermodynamic orientation — the energy ordering under which exact finite-bath counting with $\beta_E = \partial S_H/\partial E > 0$ makes the classical profile passive (kernel: `counting_passive`, from a monotone bath count), an orientation of energy rather than a direction of time — enters that completion |
+| 20 | `papers/GR.md`:561 → :561 | — | conditional | B.8 | and the synthesis claim that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same $(S, \varphi)$ are developed in [Substratum]. | and the synthesis claim that quantum mechanics and general relativity descend as projections of the same $(S, \varphi)$, with a third projection bearing on the arrow of time, are developed in [Substratum]. |
+| 21 | `papers/GR.md`:607 → :607 | — | side-correction | B.4, C.4 | Unitarity of $\varphi$ and subadditivity make the sum of the von Neumann entropies of $V$, $B$ and $D$ non-decreasing from a product initial state (Theorem A.6). | Unitarity of $\varphi$ and subadditivity bound the sum of the von Neumann entropies of $V$, $B$ and $D$ below by its value at a product initial state, at every earlier and later time (Theorem A.6); the bound is two-sided and does not make the sum monotone. |
+| 22 | `papers/GR.md`:709 → :709 | — | conditional | B.8 | The reconstruction theorem and the synthesis claim — that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same deterministic $(S, \varphi)$ — are developed in [Substratum]. Together, this four-paper synthesis ([Main], [SM], [GR], [Substratum]) proposes that the Standard Model, general relativity, and the arrow of time descend from one finite deterministic construction, | The reconstruction theorem and the synthesis claim — that quantum mechanics and general relativity descend as projections of the same deterministic $(S, \varphi)$, with a third projection bearing on the arrow of time — are developed in [Substratum]. Together, this four-paper synthesis ([Main], [SM], [GR], [Substratum]) proposes that the Standard Model and general relativity descend from one finite deterministic construction, |
+| 23 | `papers/SM.md`:1366 → :1366 | — | conditional | B.8 | The substratum-level reconstruction theorem and the synthesis claim that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same $(S, \varphi)$ are developed in [Substratum]. | The substratum-level reconstruction theorem and the synthesis claim that quantum mechanics and general relativity descend as projections of the same $(S, \varphi)$, with a third projection bearing on the arrow of time, are developed in [Substratum]. |
+| 24 | `papers/SM.md`:1592 → :1592 | — | conditional | B.7, B.8 | and the synthesis claim that quantum mechanics, general relativity, and the arrow of time descend as three projections of the same $(S, \varphi)$ are developed in [Substratum]. Together, this four-paper synthesis ([Main], [SM], [GR], [Substratum]) makes a single claim: the Standard Model, general relativity, and the arrow of time are three projections of one finite deterministic construction — the visible-sector projection developed here, the boundary-level thermodynamic limit developed in [GR], and the horizon-clock-plus-observer-selection projection developed in [Substratum §5.4]. | and the synthesis claim that quantum mechanics and general relativity descend as projections of the same $(S, \varphi)$, with a third projection bearing on the arrow of time, are developed in [Substratum]. Together, this four-paper synthesis ([Main], [SM], [GR], [Substratum]) makes a single claim: the Standard Model and general relativity are projections of one finite deterministic construction — the visible-sector projection developed here and the boundary-level thermodynamic limit developed in [GR] — with the horizon-clock-plus-observer-selection projection developed in [Substratum §5.4], whose alignment of the arrows of time is open. |
+| 25 | `papers/Methodology.md`:184 → :184 | — | conditional | B.1, B.2, B.8 | *The arrow* — that the ordering has an experienced direction — is derived in the framework's substratum treatment from horizon growth. | *The arrow* — that the ordering has an experienced direction — is addressed in the framework's substratum treatment: on the expanding cosmological branch, horizon growth supplies a non-decreasing parameter, and the direction rests on boundary conditions and the orientation convention. |
+| 26 | `papers/Methodology.md`:184 → :184 | — | unresolved | B.8, B.10 | The arrow is derived, not posited, and rests on the causal topology, hence ultimately on the ordering. | The arrow rests on the causal topology, hence on the ordering, and on those boundary conditions. |
+| 27 | `papers/Methodology.md`:184 → :184 | — | unresolved | B.8, B.10 | Causal topology and the arrow are downstream of these two and of structure (the dynamics, horizon growth) the framework has independently. No third temporal axiom is required, and none is concealed: the two-question framing of this section is the complete account of where the framework's temporal commitments lie. | Causal topology and the arrow are downstream of these two and of inputs the framework takes as given (the dynamics, the expanding branch, the boundary conditions that fix the arrow's direction). No third temporal axiom is required: the two-question framing of this section accounts for the framework's temporal axioms, and the arrow's boundary conditions are inputs rather than axioms. |
+| 28 | `papers/Methodology.md`:227 → :227 | — | conditional | B.2, B.8 | these are either left open (ordering, §5.1) or derived downstream from structure the framework independently has (causal topology and arrow, §5.2 inventory). | these are either left open (ordering, §5.1), derived downstream from structure the framework independently has (causal topology, §5.2 inventory), or addressed downstream with boundary conditions taken as inputs (arrow, §5.2 inventory). |
+| 29 | `papers/Explainer.md`:930 → :930 | — | unresolved | B.9 | Second, the "past hypothesis" of cosmology gets replaced by a conditional theorem. | Second, one role of the "past hypothesis" of cosmology — selecting where observers can exist — gets taken over by a conditional theorem. |
+| 30 | `papers/Explainer.md`:930 → :930 | — | unresolved | B.9, C.3 | You don't need to posit low initial entropy to explain why we see it; you need the definition of observation together with the mixing hypothesis for equilibrium regions, which select for the non-equilibrium phase. | For that job you don't need to posit low initial entropy; you need the definition of observation together with the mixing hypothesis for equilibrium regions, which select for the non-equilibrium phase. The selection is time-symmetric: it does not explain why entropy is low at one end of the observed history, which remains a boundary condition. |
+| 31 | `papers/Explainer.md`:930 → :930 | — | unresolved | B.9 | The mixing hypothesis is the premise that takes the place of the postulate. | The mixing hypothesis is the premise that takes the place of the postulate in that role. |
+| 32 | `papers/Explainer.md`:932 → :932 | — | conditional | B.2 | In our universe, the non-equilibrium phase is provided by the expanding cosmological horizon. | In our universe, the non-equilibrium phase is associated with the expanding cosmological horizon — an expansion that is observed, not derived. |
+| 33 | `papers/Explainer.md`:932 → :932 | — | conditional / unresolved | B.3, B.7, C.1 | Entropy increase, the Second Law, and the felt direction of time emerge as coarse-grained descriptions of the dynamics accessible to observers in this non-equilibrium phase. Like quantum mechanics itself, the direction of time is emergent — a consequence of observing a lossless cycle from within, during the phase when such observation is structurally possible. | Entropy increase, the Second Law, and the felt direction of time emerge as coarse-grained descriptions of the dynamics accessible to observers in this non-equilibrium phase, given a low-entropy boundary condition that the selection rule does not supply; whether those arrows and the expansion share a common mechanism is open. Like quantum mechanics itself, the direction of time is emergent — a feature of observing a lossless cycle from within — but which way it points is fixed by boundary conditions, not derived. |
+| 34 | `papers/Main.md`:744 → :744 | — | owner-ruled consequence | F.4 (c.6649643759 item 1) | the rate condition $\tau_{\rm mix}\ll\tau_S$ stated in the past-hypothesis corollary | the rate condition $\tau_{\rm mix}\ll\tau_S$ stated in the observer-selection corollary |
+| 35 | `papers/Main.md`:758 → :758 | — | unresolved (title) | B.9, F.4 | **Corollary (past-hypothesis replacement, with its rate condition).** | **Corollary (observer-selection rule, with its rate condition).** |
+| 36 | `book/ch02-substratum.md`:176 → :176 | FULL:819 → :819 | conditional | B.8 | The next section turns to the synthesis: how quantum mechanics, general relativity, and the arrow of time relate to the substratum as three projections of the same construction. | The next section turns to the synthesis: how quantum mechanics and general relativity relate to the substratum as projections of the same construction, and what a third projection establishes about the arrow of time. |
+| 37 | `book/ch02-substratum.md`:184 → :184 | FULL:827 → :827 | conditional | B.8 | The framework's claim is that quantum mechanics, general relativity, and the arrow of time are not three independent theories. Conditional on H-Bell at the Bell/local-lattice interface, they are three projections of the same finite deterministic construction. | The framework's claim is that quantum mechanics and general relativity are not two independent theories. Conditional on H-Bell at the Bell/local-lattice interface, they are projections of the same finite deterministic construction, and a third projection bears on the arrow of time. |
+| 38 | `book/ch02-substratum.md`:192 → :192 | FULL:835 → :835 | conditional | B.5, C.3 (ruling item 3: C2 retention) | Observers, however, are not symmetric under this exchange. Condition C2 — persistence, in its slow-bath realization — picks out one direction along the observer's worldline as the direction in which the hidden sector's correlations with the visible sector's past are preserved. | Condition C2 — persistence, in its slow-bath realization — is a retention condition: it requires the hidden sector's correlations with the visible sector's past to persist to readback, and the coarse-grained mixing time it constrains is the same in the two directions; the direction in which records are written and read is fixed by a preparation and the observer's orientation convention. |
+| 39 | `book/ch02-substratum.md`:192 → :192 | FULL:835 → :835 | conditional / unresolved | B.1, B.7, B.9, E.2 (ruling item 3) | to the non-equilibrium phase of $(S, \varphi)$; in the $\Lambda$-dominated cosmological phase, this phase is the one in which the horizon entropy $S_{\text{dS}}(t) = A(t)/(4\ell_P^2)$ is monotonically increasing. | to the non-equilibrium phase of $(S, \varphi)$; the restriction is time-symmetric and selects where observers can exist, not a direction. On an expanding cosmological background satisfying the null energy condition — inputs, not derived — the area $A(t)$ of the apparent horizon, and the horizon entropy $S_{\text{dS}}(t) = A(t)/(4\ell_P^2)$ assigned to it, is non-decreasing, strictly increasing while $\rho + p/c^2 > 0$ and constant in exact de Sitter. That the selected non-equilibrium phase is ordered by this growth is not established. |
+| 40 | `book/ch02-substratum.md`:194 → :194 | FULL:837 → :837 | conditional / unresolved | B.1–B.7, E.4 | The cosmological horizon entropy provides a structural clock. Every observer in the accelerating phase agrees that $S_{\text{dS}}$ increases — the monotonicity is a feature of the causal partition rather than of any local frame — and other arrows of time align with this structural clock through a cascade. The thermodynamic arrow descends from Jacobson's coupling between boundary entropy and matter entropy. The memory arrow descends from C2 — persistence, realized by the slow bath — which preserves correlations forward along the clock rather than backward. The measurement arrow — the observer records past outcomes and prepares future states — descends from the memory arrow. The chain grounds in one structural fact (horizon growth in the accelerating phase) and one structural constraint (C1–C4 observer selection, conditional on the effective-mixing hypothesis of [Main §4.6] and stated for the coarse-grained description generated by iterating the one-step effective channel). | The horizon area provides a structural clock on that background. Every comoving observer on an expanding background satisfying the null energy condition agrees that it does not decrease — the monotonicity is a feature of the causal partition rather than of any local frame — and the criterion is expansion with the null energy condition, not acceleration: in exact de Sitter the area is constant. Whether the other arrows of time align with this clock is not established. Jacobson's coupling relates the boundary energy density to the expansion rate, not matter entropy to horizon area, so the direction of the thermodynamic arrow rests on a low-entropy or product boundary condition, and the generalized second law is open. C2 — persistence, realized by the slow bath — holds the same way in both directions; the direction of the memory arrow comes from a preparation and the observer's orientation convention. The measurement arrow — the observer records past outcomes and prepares future states — inherits the memory arrow's direction. The chain uses one input (horizon growth on the expanding branch) and one structural constraint (C1–C4 observer selection, conditional on the effective-mixing hypothesis of [Main §4.6] and stated for the coarse-grained description generated by iterating the one-step effective channel), which fixes where observers can exist, not a direction. |
+| 41 | `book/ch02-substratum.md`:196 → :196 | FULL:839 → :839 | unresolved | B.7, B.8 | The framework's content here is precise: it derives a monotonic parameter along observer worldlines and four aligned arrows of time. What it does not derive is the convention | The framework's content here is precise: on the expanding branch it establishes a non-decreasing horizon parameter along observer worldlines, and it does not establish that the four arrows of time align. Nor does it derive the convention |
+| 42 | `book/ch02-substratum.md`:196 → :196 | FULL:839 → :839 | conditional | B.1, B.2 | The framework derives the existence of a monotonic parameter; the labeling is an observer-imposed convention. | The parameter is conditional on the expanding branch; the labeling is an observer-imposed convention. |
+| 43 | `book/ch02-substratum.md`:196 → :196 | FULL:839 → :839 | conditional | B.10 | The derivation here is therefore conditional on that ordering — it establishes that ordered moments yield a directed parameter, not that directedness arises from the substratum with no assumption about ordering at all. | The result here is therefore conditional on that ordering as well — it establishes that ordered moments on the expanding branch yield a non-decreasing parameter, not that directedness arises from the substratum with no assumption about ordering at all. |
+| 44 | `book/ch02-substratum.md`:198 → :198 | FULL:841 → :841 | conditional / unresolved | B.7 | The horizon-clock projection works on the horizon-growth parameter along observer worldlines and produces the cascade of emergent arrows. | The horizon-clock projection works on the horizon-growth parameter along observer worldlines together with the observer-selection rule; the alignment of the arrows with that parameter is open. |
+| 45 | `book/ch03-structural-realism.md`:58 → :58 | FULL:932 → :932 | unresolved | B.7, B.9 | The framework's account of the arrow of time, developed in Chapter 2 §2.6, follows from this exclusion: observers are restricted to the non-equilibrium phase of the substratum's dynamics, and the structural arrow of horizon growth provides the clock the observer's other arrows align with. | The framework's treatment of the arrow of time, developed in Chapter 2 §2.6, starts from this exclusion: observers are restricted to the non-equilibrium phase of the substratum's dynamics. The exclusion is time-symmetric; on the expanding branch, horizon growth supplies a non-decreasing clock, and whether the observer's other arrows align with it is not established. |
+| 46 | `book/ch03-structural-realism.md`:202 → :202 | FULL:1076 → :1076 | conditional | B.8 | quantum mechanics, general relativity, and the arrow of time are projections expressing the substratum's content under different aspects; | quantum mechanics and general relativity are projections expressing the substratum's content under different aspects, with a third bearing on the arrow of time; |
+| 47 | `book/ch04-methodology.md`:32 → :32 | FULL:1149 → :1149 | conditional | B.1, B.2 (ruling item 3) | The framework does not commit to a specific cosmological history beyond the structural arrow of horizon growth in the $\Lambda$-dominated phase. | The framework does not commit to a specific cosmological history beyond the expanding branch, taken as given, on which the horizon area does not decrease. |
+| 48 | `book/ch18-beyond.md`:190 → :190 | FULL:5003 → :5003 | owner-ruled site | C.1 (c.6649643759 item 2) | Within a single cycle, the observer's experience is unidirectional (the arrow of time discussed in the next section), but at the substratum level the dynamics has no preferred direction. | Within a single cycle no state function increases throughout; a direction of the observer's experience is fixed by boundary conditions on a stretch of the cycle (the arrow of time discussed in the next section), and at the substratum level the dynamics has no preferred direction. |
+| 49 | `book/ch18-beyond.md`:202 → :202 | FULL:5015 → :5015 | conditional | C.3, C.5 (A11) | The asymmetry is introduced by the trace-out — the marginalization over the hidden sector that produces the embedded observer's reduced description. The arrow of time is therefore a structural consequence of embedded observation, | The asymmetry enters through the trace-out of a non-equilibrium initial condition — the marginalization over the hidden sector that produces the embedded observer's reduced description; under the invariant measure the trace-out alone produces none. The arrow of time is therefore a feature of embedded observation of such a condition, |
+| 50 | `book/ch18-beyond.md`:208 → :208 | FULL:5021 → :5021 | side-correction | README side corrections | Detailed balance asserts that the transition probabilities $T_{ij}$ between any two states satisfy $T_{ij} = T_{ji}$ at all timescales — the rate of transition from $i$ to $j$ equals the rate of transition from $j$ to $i$. This is a *substratum-level* T-symmetry: the dynamics has no preferred direction. | For a time reversal that preserves the partition, and the counting prior on hidden states, the visible transition probabilities satisfy reciprocity, $T_{ij} = T_{\theta j, \theta i}$ at all timescales, with $\theta$ the action of time reversal on visible states; when the visible states carry no T-odd data this is detailed balance, $T_{ij} = T_{ji}$ — the rate of transition from $i$ to $j$ equals the rate of transition from $j$ to $i$. This follows from *substratum-level* T-symmetry: the dynamics has no preferred direction. |
+| 51 | `book/ch18-beyond.md`:216 → :216 | FULL:5029 → :5029 | unresolved / side-correction | B.7, C.6(β); README side corrections | The alignment of the thermodynamic and cosmological arrows is structural rather than coincidental. Both arrows trace to the same initial conditions of the cosmological partition. The universe started with low entropy at the cosmological level; the framework's structural conditions C1–C4 hold at the cosmological boundary (Chapter 7 §7.2); the thermodynamic arrow within the visible sector tracks the cosmological arrow because the cosmological boundary is the primary partition for any embedded observer. | Both arrows are traced to the same initial conditions of the cosmological partition. That is a common boundary condition, not a derived mechanism: that the thermodynamic arrow within the visible sector tracks the cosmological arrow is not established. The universe started with low entropy at the cosmological level, as an initial condition; at the cosmological boundary, coupling C1, persistence C2 and capacity C3 are verified, while history readback C4 is a named realization condition not presently discharged (Chapter 7 §7.2); the cosmological boundary is the primary partition for any embedded observer. |
+
+The LEDGER row is not in the table above. In `verification/coverage/LEDGER.json`, the
+entry `MAIN:C-past-hypothesis-replacement-rate-condition` is renamed to
+`MAIN:C-observer-selection-rule-rate-condition`:
+- `label`: `(past-hypothesis replacement, with its rate condition)` becomes
+  `(observer-selection rule, with its rate condition)`;
+- `fingerprint`: `a2f5c0c423aa9ca2` becomes `df04fba6b2d81e20`, as computed by
+  `tools/proof_census.py --json` on the edited Main.
+- Unchanged: `kernel: GAP`, `assumptions`, `area`, `line` and every other field.
+
+Authorized by owner ruling c.6649643759, item 1; justified in `RECOVERY.md` §F.4.
+
+## Follow-up edits (owner rulings #1112 c.6650301410)
+
+These answer the Phase 2 reviews at `bd926c76`: the scientific review (#1113 c.6650236439) and
+the minimum-change review (#1113 c.6650241392). The owner ruled on both in
+[#1112 c.6650301410](https://github.com/amaybaum/incompleteness/issues/1112#issuecomment-6650301410)
+(amaybaum, 2026-11-05).
+
+Each row is a fragment replacement on the text at `bd926c76`. The old text is unique in its
+file, and line numbers are unchanged. Book rows are applied identically to FULL.
+
+**Ruling items.**
+- **A1** adds ch18:214.
+- **A2** adds Explainer:924–926.
+- **A5:**
+  - the Substratum:409 heading changes. The section still separates the symmetric §4.6
+    content from the parameter that is conditional on its inputs.
+  - ch02:220 changes. "Establish its relationship to the framework's three large emergent
+    structures" overstated the arrow item, so it is narrowed to match ch02:184.
+- **B1:** the apparent horizon replaces the "causal partition" clause, at Substratum:405 and
+  ch02:194.
+- **B2:** the conditions are added at the summary sites:
+  - expanding homogeneous general-relativistic background;
+  - apparent horizon;
+  - null energy condition;
+  - conclusion stated as non-decreasing.
+- **B3:** Methodology:184.
+- **B5:** Substratum:405 now says the result is background kinematics, and :14, :26 and :371
+  say "read against", not that the projection supplies the parameter.
+- **N1–N8** as reviewed:
+  - N1, N2, N7 at Substratum:405;
+  - N3 at ch02:192;
+  - N4 at ch18:202;
+  - N5 at Substratum:411;
+  - N6 at Substratum:56;
+  - N8 at SM:1592.
+
+**Left as it stands.** ch18:214's second sentence, "Time evolution corresponds to the
+horizon entropy increasing toward equilibrium", is unchanged. Ruling A1 covers only the
+alignment sentence. The scientific review's B4 suggestion to make this clause conditional is
+recorded here for the next review.
+
+| # | Site (before → after) | Mirror | Class | Ruling | RECOVERY | Old text | New text |
+|---|---|---|---|---|---|---|---|
+| 52 | `papers/Substratum.md`:14 → :14 | — | conditional | B2, B5 | B.1, E.2 | joined by a third projection that supplies a time-symmetric observer-selection rule and, on the expanding cosmological branch, a non-decreasing horizon-area parameter. | joined by a third projection that supplies a time-symmetric observer-selection rule, read against the area of the apparent horizon, which on an expanding homogeneous general-relativistic background satisfying the null energy condition is non-decreasing as a matter of background kinematics. |
+| 53 | `papers/Substratum.md`:26 → :26 | — | conditional | B2, B5 | B.1, E.2 | together with a horizon-area parameter that is non-decreasing on the expanding cosmological branch; | read against the apparent-horizon area, which is non-decreasing on an expanding homogeneous general-relativistic background satisfying the null energy condition (background kinematics, not a substratum prediction); |
+| 54 | `papers/Substratum.md`:371 → :371 | — | conditional | B2, B5 | B.1, E.2 | that supplies the time-symmetric C1–C4 observer-selection theorem of [Main, §4.6] and, on the expanding cosmological branch, a non-decreasing horizon-area parameter. | that supplies the time-symmetric C1–C4 observer-selection theorem of [Main, §4.6], read against the apparent-horizon area, which is non-decreasing, as background kinematics, on an expanding homogeneous general-relativistic background satisfying the null energy condition. |
+| 55 | `papers/Substratum.md`:56 → :56 | — | conditional | N6 | B.3 | the direction in which coarse-grained entropy increases is fixed by a low-entropy or product boundary condition, not by the rule (§5.4). | the direction in which coarse-grained entropy increases rests on a low-entropy or product boundary condition, not on the rule (§5.4). |
+| 56 | `papers/Substratum.md`:405 → :405 | — | conditional | B5, N2 | B.1, E.2 | satisfying the null energy condition, $dA/dt = -8\pi c^2\dot H/H^3 \geq 0$: the area is non-decreasing along any comoving observer worldline, strictly increasing while $\rho + p/c^2 > 0$, and constant in exact de Sitter. | satisfying the null energy condition, $\dot H = -4\pi G(\rho + p/c^2) \leq 0$ and $dA/dt = -8\pi c^2\dot H/H^3 \geq 0$: the area is non-decreasing along any comoving observer worldline, strictly increasing while $\rho + p/c^2 > 0$, and constant in exact de Sitter. This is standard general-relativistic background kinematics, computed from the Friedmann equations of the background taken as given ([GR, §8.7]), not a prediction derived from the substratum. |
+| 57 | `papers/Substratum.md`:405 → :405 | — | conditional | N1 | B.1, E.2 | Both conditions are inputs: the expanding branch is observed, not derived, and reading $A/(4 l_p^2)$ | Both are inputs: the expanding branch is observed, not derived, and the null energy condition is assumed of the matter content; reading $A/(4 l_p^2)$ |
+| 58 | `papers/Substratum.md`:405 → :405 | — | conditional | B1 | B.1, E.1 | The monotonicity is a feature of the causal partition rather than of any local frame, just as every observer in a Schwarzschild geometry agrees on the exterior direction. | The monotonicity holds for every comoving observer of the homogeneous background. |
+| 59 | `papers/Substratum.md`:405 → :405 | — | conditional | N7 | E.4 | At the observer level, the horizon parameter orders the expanding branch; that the other arrows | At the observer level, the horizon parameter orders the expanding branch non-strictly; that the other arrows |
+| 60 | `papers/Substratum.md`:409 → :409 | — | unresolved (heading) | A5 | B.8 | **What the framework derives, and the directional-language subtlety.** | **What the framework establishes, and the directional-language subtlety.** |
+| 61 | `papers/Substratum.md`:411 → :411 | — | conditional | N5 | B.5 | The memory-arrow step uses "the observer's worldline has a causal topology" as the carrier of the asymmetry. | The memory-arrow step uses "the observer's worldline has a causal topology" as the carrier of its orientation convention. |
+| 62 | `papers/Substratum.md`:411 → :411 | — | conditional | N5 | B.5, B.7 | A framework in which the observer's worldline had no causal topology could not run the memory-arrow step, and the cascade would halt at the thermodynamic layer. | A framework in which the observer's worldline had no causal topology could not state the memory-arrow step. |
+| 63 | `papers/SM.md`:1592 → :1592 | — | conditional | N8 | B.7 | — with the horizon-clock-plus-observer-selection projection developed in [Substratum §5.4], whose alignment of the arrows of time is open. | ; the horizon-clock-plus-observer-selection projection is developed in [Substratum §5.4], where the alignment of the arrows of time is open. |
+| 64 | `papers/Methodology.md`:184 → :184 | — | unresolved | B3 | B.8 | a *causal topology*, and a derived *arrow* of time | a *causal topology*, and an *arrow* of time |
+| 65 | `papers/Methodology.md`:184 → :184 | — | conditional | B2 | B.1, E.2 | on the expanding cosmological branch, horizon growth supplies a non-decreasing parameter, | on an expanding homogeneous general-relativistic background satisfying the null energy condition, the apparent-horizon area is a non-decreasing parameter (background kinematics), |
+| 66 | `papers/Explainer.md`:924 → :924 | — | unresolved | A2 | B.9, C.3 | So the arrow, if there is one, has to come from somewhere else — specifically, from the observer. | So the arrow, if there is one, has to come from somewhere else — from the observer's description of the dynamics, given boundary conditions. |
+| 67 | `papers/Explainer.md`:926 → :926 | — | unresolved | A2 | B.9, C.3 | A structural theorem (Main §4.6) establishes exactly that, under a stated mixing hypothesis | A structural theorem (Main §4.6) constrains the observer side of this — where observers can exist, not which way time runs — under a stated mixing hypothesis |
+| 68 | `book/ch02-substratum.md`:192 → :192 | FULL:835 → :835 | conditional | N3 | B.1 | and the horizon entropy $S_{\text{dS}}(t) = A(t)/(4\ell_P^2)$ assigned to it, is non-decreasing, | and the horizon entropy $S_{\text{dS}}(t) = A(t)/(4\ell_P^2)$ assigned to it at the de Sitter calibration, is non-decreasing, |
+| 69 | `book/ch02-substratum.md`:194 → :194 | FULL:837 → :837 | conditional | B1 | B.1, E.1, E.2 | The horizon area provides a structural clock on that background. Every comoving observer on an expanding background satisfying the null energy condition agrees that it does not decrease — the monotonicity is a feature of the causal partition rather than of any local frame — and the criterion | The apparent-horizon area provides a structural clock on that background. Every comoving observer of an expanding homogeneous general-relativistic background satisfying the null energy condition agrees that it does not decrease — standard background kinematics, not a prediction derived from the substratum — and the criterion |
+| 70 | `book/ch02-substratum.md`:220 → :220 | FULL:863 → :863 | conditional | A5 | B.8 | The three projections establish its relationship to the framework's three large emergent structures. | The three projections establish its relationship to quantum mechanics and general relativity, and what it establishes about the arrow of time. |
+| 71 | `book/ch03-structural-realism.md`:58 → :58 | FULL:932 → :932 | conditional | B2 | B.1, E.2 | on the expanding branch, horizon growth supplies a non-decreasing clock, | on an expanding homogeneous general-relativistic background satisfying the null energy condition, the apparent-horizon area supplies a non-decreasing clock, |
+| 72 | `book/ch04-methodology.md`:32 → :32 | FULL:1149 → :1149 | conditional | B2 | B.1, B.2, E.2 | beyond the expanding branch, taken as given, on which the horizon area does not decrease. | beyond an expanding homogeneous general-relativistic background, taken as given, on which, with the null energy condition, the apparent-horizon area does not decrease. |
+| 73 | `book/ch18-beyond.md`:202 → :202 | FULL:5015 → :5015 | conditional | N4 | C.3, C.5 | The substratum dynamics is T-symmetric; the observer's reduced description is not. | The substratum dynamics is T-symmetric; the observer's reduced description of a non-equilibrium initial condition is not. |
+| 74 | `book/ch18-beyond.md`:214 → :214 | FULL:5027 → :5027 | unresolved | A1 | B.7, C.6 | The thermodynamic arrow and the cosmological arrow — the direction in which the universe expands — are aligned in the framework. | Whether the thermodynamic arrow and the cosmological arrow — the direction in which the universe expands — are aligned is open in the framework: it is neither established nor refuted. |
+
+## Follow-up edit (owner ruling #1112 c.6651202921)
+
+The Phase 2 scientific review at `98d04808` (#1113 c.6650406439) blocked on item C1. The
+sentence of ch18:214 that follows the initial-condition sentence stated horizon-entropy
+growth without the expanding background, the null energy condition or the calibration
+caveat, and tied it to boundary correlations and hidden-sector disorder. That tie is the
+relation GR:805 leaves open (`RECOVERY.md` B.3, B.7), and the paragraph's first sentence
+calls the alignment open. The owner ruled the replacement text in
+[#1112 c.6651202921](https://github.com/amaybaum/incompleteness/issues/1112#issuecomment-6651202921)
+(amaybaum, 2026-11-05). The ruled text does not assert that the horizon approaches a de
+Sitter value, or that the other arrows align.
+
+The row is a fragment replacement on the text at `98d04808`. The old text is unique in each
+file, line numbers are unchanged, and after the edit ch18:214 and FULL:5027 are
+byte-identical. Nothing else in the paragraph changes.
+
+| # | Site | FULL mirror | Class | Ruling | RECOVERY | Old text | New text |
+|---|---|---|---|---|---|---|---|
+| 75 | `book/ch18-beyond.md`:214 → :214 | FULL:5027 → :5027 | conditional / unresolved | c.6651202921 (C1) | B.1, B.3, B.7 | Time evolution corresponds to the horizon entropy increasing toward equilibrium, with the boundary modes accumulating correlations and the hidden sector approaching maximum disorder. | On an expanding FRW background satisfying the null energy condition, the apparent-horizon area is non-decreasing. Its assigned entropy has that interpretation under the stated calibration assumptions; whether this behavior aligns with boundary correlations or hidden-sector disorder remains open. |
+
+**Corpus grep (§A.25 step 1).** The old sentence and the paraphrases "horizon entropy
+increas", "toward equilibrium", "maximum disorder", "accumulat… correlations", "hidden
+sector approach", "horizon entropy grows/rises" and "de Sitter equilibrium" were searched
+over `papers/*.md` and `book/*.md`. After the edit the old sentence has no manuscript
+occurrence. The one remaining hit is ch18:210 (FULL:5023), "the system moves toward
+equilibrium relative to the substratum's state space". It describes the coarse-grained
+thermodynamic arrow under a non-equilibrium initial condition, not the horizon, so it is
+not a mirror of this sentence and is not edited.
+
+**Consequences.**
+- **Build:** `sh ./build.sh --book` reported `build: OK`, 561 pages (unchanged), mean 2979
+  characters per page, and no dropped glyphs.
+- **SC-42:** the integration check needs no regeneration. ch18:214 already has its "chapter
+  line mirrored in FULL" record, which is keyed by the line's 60-character prefix, and the
+  mirror stays byte-identical. A fresh `--verbose` run equals the committed
+  `sc42-integration-result.json` (sha256 `fc79539f…`, unchanged), with verdict PASS, and
+  `--self-test` equals the committed controls file.
+- **`audit-census.json`:** no count changes; `audit_census_probe.py` passes.
+
+## Gate-manifest consequence: `verification/audit-census.json`
+
+The ordinary PR `verify` run 41969128746 on `24452b01` failed its job "Numerical probes /
+core B". The cause was `audit_census_probe.py`, AUDIT-B, vocabulary `B1-required`
+(pattern "not presently discharged").
+
+**The drift.** Row 51, the authorized ch18:216 side correction, adds a correct instance of
+the required C4 status phrase at ch18:216 and FULL:5029. The counts move as follows:
+- `book/ch18-beyond.md`: 0 → 1;
+- FULL: 3 → 4.
+
+**Disposition.** "repaired" still holds at that location. The phrase is the C4 status
+sentence that the vocabulary requires, now in one more paragraph.
+
+**Change.** Following the manifest's own drift rule, only `expect` and `why` of
+`B1-required` change. The pattern is not widened. `python3 verification/lean/audit_census_probe.py`
+passes locally after the change.
+
+**Scope.** The manifest is a mutable gate file, not a frozen record; it is not listed in
+`verification/infrastructure/legacy-records.json`. It was not in the authorized file list,
+so this consequence is recorded here for the owner and reviewers to confirm, as the
+LEDGER rename was.
+
+**Owner ruling.** Accepted by owner ruling c.6650301410, item A4: ch18 0 → 1 and FULL 3 → 4.
+The rule also requires the update to share a commit with the text change; this update came
+in `6010cf44`, after the text change in `4081cba0`. The follow-up edits change no
+audit-census count, and `audit_census_probe.py` passes.
+
+## CI finding: the SC-42 integration step
+
+The ordinary PR `verify` run 41970039595 on `6010cf44` failed one job, "Numerical probes /
+core B", at the step "SC-42 integration source scope and must-fail controls". That step
+runs `verification/audits/manuscript/bell-operational-reconciliation-20261029/check_sc42_integration.py`,
+diffing against its pinned base `ca3a5bb2`.
+
+1. **Wording hit: fixed.** The checker's added-text scan found one hit,
+   `('papers/Substratum.md', 'caps', 'FLRW')`. The acronym in row 11 broke the
+   caps-emphasis rule (§A.33). Row 11 now reads "flat Friedmann-Robertson-Walker
+   background", the form Substratum:379 already uses, and Substratum `.tex`/`.pdf` are
+   rebuilt (still 55 pages). Run locally after the fix, the checker gives:
+   - `verdict: PASS`, with no failed predicates;
+   - a `--self-test` byte-identical to the committed `sc42-integration-controls.json`.
+2. **Byte comparison with the committed result: resolved by regeneration (ruling A3).**
+   The step also compares the checker's output byte for byte (`cmp`) against the
+   committed `sc42-integration-result.json`. The checker adds one "chapter line mirrored in
+   FULL" predicate for each book-chapter line changed since `ca3a5bb2`.
+   - **At `bd926c76`** the count was 9 added and 1 replaced, checks 678 → 687. (An
+     earlier draft of this record said "10 and replace 1".)
+   - **After the follow-up edits.** Ruling A1 (ch18:214) and ruling A5 (ch02:220) add two
+     more lines. The final count is 12 new passing records, one of which replaces the old
+     ch02:194 record: 11 added and 1 replaced. Checks go from 678 to 689.
+   - **The new records, by chapter line:**
+     - ch02:176, 184, 194 (replacing "The cosmological horizon entropy provides a
+       structural clock…"), 196, 198, 220;
+     - ch03:202;
+     - ch04:32;
+     - ch18:202, 208, 214, 216.
+   - **Edited lines that add no new record:** ch02:192, ch03:58 and ch18:190. Their
+     60-character prefixes were already recorded for edits made since `ca3a5bb2`.
+   - **Regeneration.** Commit `a762d418` regenerates the file with the existing checker:
+     the stdout of `python3 -I -B verification/audits/manuscript/bell-operational-reconciliation-20261029/check_sc42_integration.py --verbose`,
+     the same procedure as `d83f8723`. The checker, `sc42-integration-controls.json` and
+     the tests are unchanged.
+   - **Verification:**
+     - Only the `checks` array differs from the previous file; `base`, `checker_sha256`,
+       `site_table_sha256`, the counts, `mode`, `schema` and `verdict` are equal.
+     - The verdict is PASS with `failed_predicates: []`, and every record has `ok: true`.
+     - The `--verbose` run and the `-O` run give identical output.
+     - `--self-test` gives the same output in both modes, equal to the committed controls
+       file.
+     - A fresh run equals the committed file.
+     - The git diff is +45/−1 lines: 11 new four-line records, plus the one predicate line
+       of the replaced record.
+   - **sha256** of the committed result: `fc79539fb4d1dc4eef184320ee5f39b3f5a85ba02d273126abf4d6a78ed96565`.
+
+## Planned sites left unchanged
+
+Each was reviewed against `RECOVERY.md`. No change was needed, under minimum change
+(§A.25).
+
+- **Main:744 body.** The replacement is scoped "in its role of selecting
+  observer-compatible configurations". Only the label phrase changed (row 34).
+- **Main:758 body** (B.9, F.4): unchanged; the title is row 35.
+- **Main:760 (iii):** NO CHANGE by ruling c.6649643759, item 4.
+- **Selection content kept as written:**
+  - Substratum:56's selection sentence (row 3 changes only the last sentence);
+  - Substratum:409's symmetry and convention content (row 13 changes only the last two
+    sentences).
+- **GR:797–805:** Theorem A.6 and "the generalized second law is open" are correct as
+  written. GR:797 defines the generalized second law (delta review c.6649698455, item D).
+- **Other retained sites:** Explainer:928 (Explainer:924–926 were later changed by ruling A2); Complexity:206; Structure:717;
+  ch04:30; ch07:155–157.
+- **Asking or listing, not asserting:**
+  - ch02:182 (FULL:825) asks what relationship (S, φ) bears to the three structures;
+  - ch03:62 (FULL:936) lists the arrow among consequences of the observer-admission
+    boundary, which does bear on it through selection;
+  - ch03:204 (FULL:1078) counts three projections without stating what the third derives.
+- **Already stated as unresolved:** ch18:20, 204–212, 218–222, 328, 334; ch00:101;
+  `book/README.md`:90; GR:805.
+
+## Unlisted sites found by the corpus-wide grep, not edited
+
+Grep run per §A.25 step 1, for `\barrows?\b`, "horizon clock", "structural clock",
+"accelerating phase", "direction of time", "past hypothesis", "Ellis", "second law",
+"horizon growth", "monotonically increasing", "unidirectional", "three projections" and
+"third projection" over `papers/*.md` and `book/*.md`. These sites carry related wording but
+are not on the authorized list. They are listed for an owner ruling.
+
+1. **ch18:214 (FULL:5027).** The paragraph says "The thermodynamic arrow and the
+   cosmological arrow — the direction in which the universe expands — are aligned in the
+   framework" and "Time evolution corresponds to the horizon entropy increasing toward
+   equilibrium".
+   - It asserts alignment (`RECOVERY.md` B.7). It sits beside the edited ch18:216.
+   - The plan listed ch18:204–214 as "already correct as unresolved". This record finds
+     :214 an exception.
+   - Recommended: the same treatment as ch18:216. **Resolved:** added by ruling A1, row 74;
+     the horizon sentence by ruling c.6651202921, row 75.
+2. **Substratum:399**, the heading "5.4 The arrow of time as a third projection", and
+   **Substratum:393**, "(the arrow-of-time projection is developed in §5.4 …)".
+   - These are labels for the narrowed framing.
+   - Recommended: no change.
+3. **Substratum:409's heading** "What the framework derives, and the directional-language
+   subtlety". Its paragraph now states what the framework *establishes*. **Resolved:**
+   changed to "establishes" by ruling A5, row 60.
+4. **ch02:200 (FULL:843).** It reads "produce the persistence-based observer selection for
+   the horizon-clock projection … the three projections are co-derived from a single
+   construction". This is framing. Recommended: no change.
+5. **ch02:220 (FULL:863).** It reads "The three projections establish its relationship to
+   the framework's three large emergent structures." **Resolved:** narrowed to match
+   ch02:184 by ruling A5, row 70, because it overstated the arrow item.
+6. **GR:190.** It reads "Its classical layer is settled: exact finite-bath counting with
+   $\beta_E>0$ makes the counting-selected profile passive (kernel: `counting_passive`)".
+   It states $\beta_E>0$ as the condition, consistent with the GR:214 correction.
+   Recommended: no change.
+7. **Methodology:212, 214.** These describe other authors' derivations of ordering and
+   arrow. Not OI claims; no change.
+8. **Explainer:924–926.** Found by the Phase 2 scientific review, c.6650236439. "A
+   structural theorem (Main §4.6) establishes exactly that" attributed the arrow's origin to
+   §4.6. **Resolved:** added by ruling A2, rows 66–67.
+
+## Build
+
+- **Toolchain:** pandoc 3.1.3; XeTeX 3.141592653-2.6-0.999995 (TeX Live 2023/Debian).
+- **Commands:** `sh ./build.sh Substratum GR SM Methodology Explainer Main`, then
+  `sh ./build.sh --book`. Both reported `build: OK`.
+- **Missing-character warnings:** 0. `build.sh` lists dropped glyphs per file and listed
+  none.
+- **Each `.tex`/`.pdf`** is built from the committed `.md`. `tools/staleness_check.py`
+  reports 13 matched, 0 unstamped.
+
+| Artifact | Pages at base | Pages after | Mean characters per page after |
+|---|---|---|---|
+| Substratum | 55 | 55 | 3185 |
+| GR | 91 | 91 | 3166 |
+| SM | 150 | 150 | 3156 |
+| Methodology | 53 | 53 | 3021 |
+| Explainer | 71 | 72 | 2786 |
+| Main | 100 | 100 | 3277 |
+| Book (FULL) | 560 | 561 | 2978 |
+
+Commit `f6f7dd13` rebuilt the sources it edited, with `sh ./build.sh Substratum SM
+Methodology Explainer` and `sh ./build.sh --book`. Both reported `build: OK`, with 0
+missing-character warnings, and `staleness_check` reports 13 matched.
+
+| Artifact | Pages | Mean characters per page |
+|---|---|---|
+| Substratum | 55 | 3194 |
+| SM | 150 | 3156 |
+| Methodology | 53 | 3022 |
+| Explainer | 72 | 2788 |
+| Book (FULL) | 561 | 2978 |
+
+GR and Main were not edited in the follow-up and were not rebuilt.
+
+## Local checks
+
+These were run in the worktree after the edits. They are offline and cheap, and none is
+a scientific computation. All exited 0.
+
+| Command | Result |
+|---|---|
+| `python3 tools/coverage_check.py` | `coverage_check: OK (134 canonical statements, 19 unattached checkers)` |
+| `python3 tools/voice_check.py` | `voice_check: OK (no manuscript-voice history narration; 40 manuscript file(s) scanned)` |
+| `python3 tools/claims_check.py` | `claims_check: OK (no withdrawn result asserted unconditionally)` |
+| `python3 tools/duplicate_check.py` | `duplicate_check: OK (no paragraph repeated within a file)` |
+| `python3 tools/staleness_check.py` | `staleness_check: OK (13 matched, 0 unstamped)` |
+| `python3 tools/lean_manuscript_census.py` | `lean_manuscript_census: OK (… 25 current families; complete relative to the maintained registry)` |
+| `python3 tools/artifact_placement_check.py` | `artifact_placement_check: OK (… self-test passed)` |
+| `python3 verification/lean/audit_census_probe.py` | `audit_census_probe: ALL CHECKS PASS`, after the `B1-required` update |
+| `python3 -I -B verification/audits/manuscript/bell-operational-reconciliation-20261029/check_sc42_integration.py` | verdict PASS, no failed predicates; after `a762d418` the output equals the committed result (sha256 `fc79539f…`) |
+
+The same checks passed at the base before the edits. No workflow was dispatched, and no
+Lean module was built.
diff --git a/verification/audits/manuscript/arrow-1-20261105/README.md b/verification/audits/manuscript/arrow-1-20261105/README.md
new file mode 100644
index 000000000..7a5716f35
--- /dev/null
+++ b/verification/audits/manuscript/arrow-1-20261105/README.md
@@ -0,0 +1,179 @@
+# ARROW-1 manuscript correction round (#1112)
+
+**Status: WIP / UNVERIFIED.** Phase 1 (the theorem-recovery note, `RECOVERY.md`) was
+reviewed independently twice, both reviews model-authored and both ACCEPT WITH CHANGES:
+- at `bd398f05af3d778c149db9d1a627f80efffd998e`:
+  [#1113 c.6649600806](https://github.com/amaybaum/incompleteness/pull/1113#issuecomment-6649600806);
+- at `37b39d1c5bbea2efd4a10d9a8c69d127dd593bf5` (delta review):
+  [#1113 c.6649698455](https://github.com/amaybaum/incompleteness/pull/1113#issuecomment-6649698455).
+
+The delta review judged the note a sound basis for Phase 2. The owner ruled on the open
+items in
+[#1112 c.6649643759](https://github.com/amaybaum/incompleteness/issues/1112#issuecomment-6649643759)
+(amaybaum, Claude Code session chat, 2026-11-05). Phase 2's manuscript edits follow this
+record commit, and `CHANGES.md` records them site by site.
+
+## Scope
+
+Issue #1112 asks whether any argument in the corpus derives a time asymmetry without
+time-asymmetric initial or boundary conditions. The read-only analysis
+([#1112 c.6643691865](https://github.com/amaybaum/incompleteness/issues/1112#issuecomment-6643691865))
+and its independent verification
+([#1112 c.6643840329](https://github.com/amaybaum/incompleteness/issues/1112#issuecomment-6643840329))
+found that none does. This round applies the owner's decision (Option 3, split): it
+separates what is proved under stated conditions from inherited boundary inputs and from
+unresolved problems, and propagates the result corpus-wide under AGENTS.md §A.25.
+
+- **Base:** `main` at `f5c281ae10a913ec68fa41b3b839ad0b164979cd`. All line locators in this
+  directory are at that commit.
+- **Branch:** `fix/arrow-1-20261105`. Designated writer: `writer-arrow`.
+- **Underlying research question:** open. This round changes manuscript statements to
+  match their support. It does not resolve whether OI explains the observed direction of
+  time, and #1112 stays open.
+
+## Authorization
+
+Owner decision, recorded in
+[#1112 c.6649271094](https://github.com/amaybaum/incompleteness/issues/1112#issuecomment-6649271094)
+(amaybaum, Claude Code session chat, 2026-11-05). It approves the plan in c.6643840329 and
+quotes the owner:
+
+> Alignment: Before narrowing the claims, prepare the short theorem-recovery note
+> identifying what a common alignment mechanism would need to establish. If no existing
+> derivation survives review, mark alignment unresolved rather than refuted.
+
+> Authorize one designated writer on a fresh branch for this manuscript correction round,
+> under §A.25. First record the recovery analysis, then implement the supported edits with
+> complete cross-document propagation, generated artifacts and mirrors. Obtain independent
+> scientific and minimum-change reviews.
+>
+> No new physical premise, theorem promotion, computation, workflow dispatch or merge is
+> authorized.
+
+**Not authorized:** computation, Lean builds, workflow dispatch, new physical premises,
+theorem promotion, merge, and changes to frozen records.
+
+## Phases
+
+1. **Phase 1 — recovery note (this checkpoint).** `RECOVERY.md`: the unproved claims quoted
+   at the base, the strongest existing result for each, the target statement a proof would
+   need, the remaining gap, the obstructions that bound any recovery, a verdict per claim,
+   the horizon kinematics written out, and wording principles for Phase 2. Reviewed
+   independently before any manuscript edit.
+2. **Phase 2 — manuscript edits**, only after that review: the sites below, their FULL
+   mirrors, the regenerated `.tex`/`.pdf`, and a change note mapping each edited passage to
+   its finding. Independent scientific and minimum-change reviews run on the exact commit.
+
+## Sites Phase 2 will touch
+
+Taken from the plan in c.6643840329, as approved in c.6649271094. Book sites carry their
+FULL mirrors (§A.14). Locators are at the base.
+
+**(i) Retain a derived claim (expected unchanged, or conditions only restated where they
+already are):** Main:744–756 (selection theorem); Main:758 (selection-rule content);
+Substratum:56 (selection sentence); Substratum:409 ("§4.6 is symmetric; the direction is a
+convention"); GR:797–805 (Theorem A.6 and "the generalized second law is open");
+Explainer:924, 926–928; Complexity:206; Structure:717; ch02:194 (only the clause on no
+independent past-hypothesis axiom being required to select where observers exist, with
+(EM)); ch03:58 (the Boltzmann-brain exclusion); ch04:30; ch07:155–157.
+
+**(ii) Conditional language, conditions named:**
+- monotone clock (expanding branch, null energy condition; weak, constant in exact de
+  Sitter; replace the ä > 0 criterion): Substratum:379, 405, and 409's "derives the
+  existence of a monotonic parameter"; ch02:192's last clause ("this phase is the one in
+  which the horizon entropy … is monotonically increasing"; also alignment content, see
+  (iii)); ch04:32's "the structural arrow of horizon growth in the $\Lambda$-dominated phase"
+  (clock-as-arrow wording); Methodology:184's "from horizon growth";
+- cosmological arrow (the expanding branch is an input, selected by no theorem):
+  Substratum:407 (cosmological step); Explainer:932; Methodology:227;
+- thermodynamic arrow (product or low-entropy preparation, plus the open identification of
+  horizon area with entropy): Substratum:407 (thermodynamic step); Substratum:56's last
+  sentence; GR:607;
+- memory and measurement arrows (orientation convention and a non-stationary preparation):
+  Substratum:407 (memory and measurement steps); ch02:192's "Condition C2 … picks out one
+  direction along the observer's worldline";
+- "three projections" framing: Substratum:367, 371, 377; GR:32, 561, 709; SM:1366, 1592;
+  ch02:176, 182, 184, 198; ch03:62, 202, 204;
+- other: Substratum:401 (the confinement; remove the §A.33 history voice), 411 (citation),
+  413 (list every dependency); GR:214 (energy orientation, not time direction); ch18:202
+  (the trace-out alone gives no asymmetry).
+
+**(iii) Describe an unresolved problem:**
+- alignment of arrows: Substratum:401's "every arrow … descends" and 405's "all other
+  arrows align"; ch02:192's identification of the selected phase with the phase of growing
+  horizon entropy; ch02:196 ("four aligned arrows"); ch18:216;
+- claims that the arrow is derived or the objection closed: Substratum:14, 26, 403, 409's
+  "answers Ellis's objection", 415; Methodology:184's "none is concealed"; Explainer:930's
+  sentence "You don't need to posit low initial entropy to explain why we see it"; ch03:58's "account of the arrow … follows from this exclusion"; ch02:194's
+  thermodynamic-arrow and alignment content;
+- Main:758's corollary title, retitled "Corollary (observer-selection rule, with its rate
+  condition)" with the theorem and hypotheses preserved (see `RECOVERY.md` §F).
+- Already correct as unresolved, expected unchanged: ch18:20, 204–214, 218–222, 328, 334;
+  ch00:101; `book/README.md`:90; GR:805.
+- **Main:760 (iii): NO CHANGE** (owner ruling c.6649643759 item 4). `RECOVERY.md` §F.4
+  finds no ARROW-1 defect there: it makes no monotonicity, direction or low-entropy claim.
+
+## Side corrections (owner, c.6649271094)
+
+- ch18:208 states detailed balance as $T_{ij}=T_{ji}$ "at all timescales", as a bare
+  consequence of substratum T-symmetry. The corpus result (SM Theorem 19; ch06:62) is
+  $T_{ij}=T_{\theta j,\theta i}$ at every time scale, under its hypotheses:
+  - an involutive time reversal that preserves the partition;
+  - the counting (uniform) hidden prior.
+
+  It reduces to $T_{ij}=T_{ji}$ only when the visible states carry no T-odd data
+  ($\theta=\mathrm{id}$). The error is the dropped θ and the dropped hypotheses. Under the
+  counting prior, $T_{ij}$ is both the normalized count and the transition probability.
+- ch18:216: "the framework's structural conditions C1–C4 hold at the cosmological boundary";
+  (C4) is a named realization condition there, not presently discharged (GR:52; ch07:12).
+- GR:214: the kernel theorem `counting_passive` proves passivity *given* a monotone bath
+  count `Ω`; the "arrow" there is an energy-sign orientation, not a time direction.
+- GR:607: "non-decreasing from a product initial state" overstates Theorem A.6, which is a
+  bound relative to $t=0$ holding at every $t$ in both directions.
+- Substratum:407: the citation "[GR, §3]" for $\rho_s=\rho_{\rm crit}$; the identity is
+  derived in GR §7.2 (GR:428–430), with its temperature input from GR §3.1.
+- Substratum:411: "[Main, §1.3] already assumes a causal topology on the partition"; Main
+  contains no such assumption.
+
+## Files
+
+- Allowed (c.6649271094): the passages above in `papers/{Substratum,GR,SM,Methodology,
+  Explainer,Main}.md` (Main: the §4.6 corollary title and :760 only); `book/ch02`, `ch03`,
+  `ch04`, `ch18` and their FULL mirrors; the regenerated `.tex`/`.pdf`; this directory.
+- Not allowed: GR:260 and Main:590 (the separate #1029 unit); Lean modules; frozen records;
+  every other passage.
+- **Added by owner ruling** [#1112 c.6649643759](https://github.com/amaybaum/incompleteness/issues/1112#issuecomment-6649643759)
+  (amaybaum, 2026-11-05). Quoting item 1: "Approve the retitle, including the necessary
+  Main:744 reference correction and atomic updates to `LEDGER.json` (`id`, `label`,
+  `fingerprint`). Preserve `kernel: GAP`, existing assumptions and traceability." Quoting
+  item 2: "Correct the unconditional unidirectionality claim without introducing a new
+  assertion that observer experience must reverse during recurrence." The added sites:
+  1. **Main:744's label phrase.** It reads "the rate condition … stated in the
+     past-hypothesis corollary", the only use of that label, and would dangle after the
+     Main:758 retitle. Recommended scope: that phrase only.
+  2. **`verification/coverage/LEDGER.json`.** The entry
+     `MAIN:C-past-hypothesis-replacement-rate-condition` would no longer match, because
+     `tools/proof_census.py` builds the identifier from the parenthetical label.
+     `tools/coverage_check.py` would then fail. Recommended scope: rename `id` to
+     `MAIN:C-observer-selection-rule-rate-condition` and update `label` and `fingerprint`,
+     with `kernel: GAP` and the assumptions unchanged. The ledger is mutable, not frozen.
+  3. **ch18:190.** It reads "Within a single cycle, the observer's experience is
+     unidirectional (the arrow of time discussed in the next section)". That conflicts with
+     `RECOVERY.md` §C.1: within a cycle no state function is globally monotone, and any
+     monotonicity is windowed and anchored by boundary data. It is added with its FULL
+     mirror. Its ground is `RECOVERY.md` §C.1.
+
+  See `RECOVERY.md` §F.4 for items 1–2.
+- **Main:760 (iii): NO CHANGE**, by ruling c.6649643759 item 4.
+- **Ruling item 3:** ch02:192 and ch04:32 are reclassified within the round. The ruling
+  asks that observational conditions, C2 retention, conditional horizon growth and any
+  genuinely established temporal direction be kept separate.
+
+## Contents
+
+- `README.md` — this file.
+- `RECOVERY.md` — the theorem-recovery note (Phase 1).
+- `CHANGES.md` — the Phase 2 site-by-site change record.
+
+No scientific computation was run and no Lean module was built for this record. Every argument in
+`RECOVERY.md` is a paper derivation or a reading of the cited source, and is marked as such.
diff --git a/verification/audits/manuscript/arrow-1-20261105/RECOVERY.md b/verification/audits/manuscript/arrow-1-20261105/RECOVERY.md
new file mode 100644
index 000000000..431e82ecd
--- /dev/null
+++ b/verification/audits/manuscript/arrow-1-20261105/RECOVERY.md
@@ -0,0 +1,846 @@
+# ARROW-1 theorem-recovery note
+
+**Status: WIP / UNVERIFIED.** Phase 1 of the #1112 correction round. Reviewed
+independently twice, both reviews model-authored and both ACCEPT WITH CHANGES:
+- at `bd398f05`: #1113 c.6649600806;
+- at `37b39d1c` (delta review): #1113 c.6649698455.
+
+This revision addresses both. Nothing in this note changes a manuscript statement or
+promotes a result.
+
+- **Base:** `f5c281ae10a913ec68fa41b3b839ad0b164979cd`. Every locator `File:line` is at
+  this commit.
+- **Authorization:** #1112 c.6649271094. No computation was run and no Lean module was
+  built.
+- **Support labels.** Each result below carries one of these labels:
+  - **[K]** kernel-checked theorem, cited by identifier;
+  - **[PP]** paper proof in the corpus;
+  - **[EA]** exact algebra written out in this note;
+  - **[HA]** hand argument in this note, not in the corpus;
+  - **[AMB]** ambient FLRW/GR background calculation, not a native OI derivation;
+  - **[LIT]** literature read first-hand (exact arXiv version given);
+  - **[LIT2]** literature known only second-hand.
+
+  Literature never establishes that a result follows from OI.
+- **Purpose.** The rule is "theorem recovery before concession" (AGENTS.md, Code Review
+  Rules). For each unproved claim, this note:
+  - identifies the strongest existing result;
+  - states what a proof would have to establish;
+  - attempts recovery inside the corpus;
+  - records the exact gap.
+
+  Failure to recover a proof is not disproof. "Refuted" is used only where a counterexample
+  is exhibited.
+
+## Summary
+
+| # | Claim (short) | Verdict |
+|---|---|---|
+| A1 | Horizon clock is monotone along observer worldlines | **Area: proved [EA, AMB]** as ambient FLRW kinematics, given an expanding background with the null energy condition (NEC).<br>• **Apparent horizon** (conditions on the interval considered): non-decreasing; strict iff ρ + p/c² > 0; constant in exact de Sitter.<br>• **Event horizon** (flat; conditions on all of [t, ∞)): non-decreasing; Ṙ_e(t) > 0 iff H is non-constant on [t, ∞); constant iff exact de Sitter on [t, ∞).<br>Reading the area as *entropy* along a dynamical horizon is **open** (GR:56). "Monotonic for as long as ä > 0" is **refuted as a general kinematic criterion** (phantom, NEC-violating), and **false as strict increase in exact de Sitter**. In the corpus's NEC-respecting background, ä > 0 is not the operative condition (§E.4). |
+| A2 | Cosmological arrow | **Conditional.** The direction of A1 is the sign of H, which is an input. Why the expanding branch is the one realized: **open.** |
+| A3 | Thermodynamic arrow descends from ρ_s = ρ_crit | **Open.** The cited identity is instantaneous and concerns the boundary entropy only; it does not reach the step. The physical direction of entropy increase is **conditional on asymmetric boundary data**. |
+| A4 | Generalized second law; "non-decreasing from a product initial state" | GSL **open** (as GR:805 states). Theorem A.6 is **proved [PP]** as a two-sided bound relative to t = 0. Its monotone reading at GR:607 is **refuted** (two-qubit counterexample, §C.4). |
+| A5 | Memory arrow aligns with the clock | Direction: **conditional** on a non-stationary preparation and the orientation convention. Alignment with the clock: **open.** |
+| A6 | Measurement arrow | Inherits A5: **conditional** / **open**. |
+| A7 | Alignment ("all other arrows align", "every arrow … descends", "four aligned arrows") | **Open — unresolved, not refuted.** Alignment is T-even, and none of the obstructions forbids it (§C.6). No common mechanism is proved; the gap is stated in §B.7. |
+| A8 | "The framework derives the arrow" / "resolves the arrow-of-time question" / closes the Ellis objection | **Not supported as stated.** What is derived reduces to A1 (conditional, ambient) plus A9 (symmetric). The rest is open. Only the reading "a globally monotone state function of (S, φ)" is excluded (§C.1). |
+| A9 | Observer-selection rule (Main §4.6) | **Proved [PP]**, conditional on (EM), the rate condition, bounded diameter and canonical typicality, in the iterated-channel description. **Time-symmetric** (§C.3). It supplies no direction and no low-entropy boundary. |
+| A10 | "The arrow's single open dependency is the ordering question and nothing further" | **Refuted as stated.** The section's own chain uses further inputs (§B.10). |
+| A11 | "The asymmetry is introduced by the trace-out" (ch18:202) | **Refuted in its unconditional reading** (§C.3, §C.5). Under the invariant measure the traced-out process carries no entropy or memory asymmetry. |
+
+The owner's expectation was "unresolved rather than refuted" for alignment. That holds:
+no counterexample to alignment exists, and §C.6 explains why none of the obstructions
+yields one. The refutations in the table concern narrower statements, each for the class actually
+ruled out:
+- the ä > 0 criterion, as a general kinematic criterion, and as strict increase in exact
+  de Sitter;
+- the monotone reading of Theorem A.6;
+- the two unconditional inventory and trace-out sentences.
+
+***
+
+## A. The claims, quoted at `f5c281ae`
+
+**A1 — monotone horizon clock.** Substratum:405:
+> The framework's primary arrow is the cosmological horizon entropy
+> $S_{\rm dS}(t) = A(t)/(4 l_p^2)$, monotonically increasing along any observer worldline in
+> the $\Lambda$-dominated phase. … every observer in the accelerating phase agrees that
+> $S_{\rm dS}$ increases … The horizon clock is observer-independent, continuous, and
+> monotonic for as long as $\ddot a > 0$ …
+
+Also at:
+- Substratum:409: "The framework therefore derives the existence of a monotonic parameter along observer worldlines";
+- ch02:192–194;
+- Methodology:184.
+
+**A2 — cosmological arrow.** Substratum:407:
+> *Cosmological:* $S_{\rm dS}(t)$ monotonic in the $\Lambda$-dominated phase, as above.
+
+Also Substratum:379: "time translation is broken intrinsically by the growth of the horizon".
+
+**A3 — thermodynamic step.** Substratum:407:
+> *Thermodynamic:* via Jacobson's coupling $\rho_s = \rho_{\rm crit}$ ([GR, §3]), the
+> entropy of any horizon-bounded system increases with the observer's clock — the
+> thermodynamic arrow descends from the structural arrow without an independent postulate.
+
+ch02:194: "The thermodynamic arrow descends from Jacobson's coupling between boundary
+entropy and matter entropy."
+
+**A4 — generalized second law.** GR:607:
+> Unitarity of $\varphi$ and subadditivity make the sum of the von Neumann entropies of $V$,
+> $B$ and $D$ non-decreasing from a product initial state (Theorem A.6).
+
+GR:805: "**The generalized second law is open.**"
+
+**A5 — memory step.** Substratum:407:
+> *Memory:* C2 (memory persistence) implies the observer's hidden sector preserves
+> correlations forward along the clock rather than backward, because the coarse-grained
+> mixing time of the hidden-sector dynamics is uniform in the two directions but the
+> observer's measurement protocol commits to one.
+
+**A6 — measurement step.** Substratum:407:
+> *Measurement:* the observer records "past" outcomes and prepares "future" states — an
+> asymmetry inherited from the memory layer, not added on top of it.
+
+**A7 — alignment.**
+- Substratum:401: "every arrow the observer sees descends from this confinement together with the horizon growth that the accelerating phase produces".
+- Substratum:405: "At the observer level, the horizon clock picks out a direction, and all other arrows align with it."
+- Substratum:407: "Four derived arrows align with the horizon clock through a structural cascade."
+- ch02:196: "it derives a monotonic parameter along observer worldlines and four aligned arrows of time".
+- ch18:216: "The alignment of the thermodynamic and cosmological arrows is structural rather than coincidental. Both arrows trace to the same initial conditions of the cosmological partition."
+
+**A8 — "derived" headlines.**
+- Substratum:403: "The framework derives the arrow; it posits the existence of the domain."
+- Substratum:26: "a third projection (§5.4) that resolves the arrow-of-time question …, answering the Ellis objection".
+- Substratum:14: "joined by a third projection that produces the arrow of time".
+- Substratum:415: "The layered cascade closes the Ellis-axis objection for the arrow of time".
+- Methodology:184: "The arrow is derived, not posited … No third temporal axiom is required, and none is concealed."
+- Explainer:930: "You don't need to posit low initial entropy to explain why we see it".
+- ch03:58: "The framework's account of the arrow of time, developed in Chapter 2 §2.6, follows from this exclusion".
+
+**A9 — selection.**
+- Main:744: "The rule replaces a postulate often imported from elsewhere — the 'past hypothesis' of low initial entropy, in its role of selecting observer-compatible configurations — with a theorem conditional on the effective-mixing hypothesis (EM)".
+- Main:758, titled "**Corollary (past-hypothesis replacement, with its rate condition).**", ends: "The low-entropy initial condition commonly imported as a 'past hypothesis' to select observer-compatible configurations is replaced here, within that description, by a structural selection rule".
+
+**A10 — inventory.** Substratum:413: "so the arrow's single open dependency is the ordering
+question and nothing further."
+
+**A11 — trace-out.** ch18:202: "The asymmetry is introduced by the trace-out — the
+marginalization over the hidden sector that produces the embedded observer's reduced
+description."
+
+***
+
+## B. Per claim: strongest result, target, closest results, gap, route
+
+### B.1 Monotone horizon clock (A1)
+
+- **Strongest existing result.** The corpus does not prove it. GR:56 names the two horizons:
+  - the event horizon, $R_{\rm event} = a(t)\int_t^\infty c\,dt'/a(t')$;
+  - the apparent/Hubble horizon, $R_A = c/H$, $A = 4\pi c^2/H^2$.
+
+  GR:56 also states that they coincide only in exact de Sitter. No corpus document computes
+  $dA/dt$. The Lean grep in c.6643840329, re-checked here, finds no horizon-monotonicity
+  theorem.
+- **Target statement (T1).** Let the background be an FLRW metric obeying the Friedmann
+  equations with total density ρ and pressure p (Λ included as a component).
+  - **(T1a) Apparent horizon, any curvature.** If $H > 0$ and $\rho + p/c^2 \ge 0$ on an
+    interval $I$, then on $I$ the apparent-horizon area is non-decreasing. It is strictly
+    increasing where $\rho + p/c^2 > 0$, and constant where $\rho + p/c^2 \equiv 0$ (exact
+    de Sitter).
+  - **(T1b) Event horizon, flat.** The event horizon at $t$ depends on the whole future
+    $[t,\infty)$. A later phantom phase toward a big rip gives $R_e \to 0$, and so
+    $\dot R_e < 0$ on $I$ can occur whatever holds on a bounded interval $I$. The
+    hypotheses are therefore imposed on all of $[t,\infty)$; this is sufficient, not
+    necessary. If the event horizon exists, $H > 0$ and $\dot H \le 0$ on $[t,\infty)$
+    (NEC, flat), then $\dot R_e(t) \ge 0$. Equality holds iff H is constant on
+    $[t,\infty)$. So $\dot R_e(t) > 0$ iff H is non-constant on $[t,\infty)$. That is not
+    the apparent horizon's "strict iff $\rho+p/c^2>0$" condition: an exact de Sitter
+    stretch followed by an NEC-respecting decrease of H gives $\dot R_e > 0$ during the
+    stretch.
+  - Phase 2 names the apparent horizon (F.2), so T1a is the statement Phase 2 relies on.
+    T1b is recorded for completeness.
+- **Proof.** §E [EA, AMB].
+- **Gap.** None for T1 as stated. T1 is an ambient GR calculation: OI's covariant field
+  equations are the open Level-G3 step (GR §8.7). T1 is therefore a statement about the
+  background OI is applied to, not a derivation from (S, φ). Two further points:
+  - Reading $A/(4l_p^2)$ along a *dynamical* FLRW horizon as the horizon's entropy goes
+    beyond GR's calibration surface. GR:56 states that the dynamical (Kodama–Hayward)
+    extension "is not established here".
+  - So "the horizon entropy increases" is the area statement T1 plus an identification
+    that is open.
+- **Route.** Complete for the area (T1). The entropy reading is open.
+
+### B.2 Cosmological arrow (A2)
+
+- **Strongest existing result.** T1 orients A by the sign of H. Under $t \to -t$, H changes
+  sign while $\dot H$, A and ρ + p do not, so $dA/dt$ is T-odd (§E.3). The horizon clock's
+  direction is therefore *the direction of expansion*, re-expressed.
+- **Inputs.** The expanding branch is taken as given:
+  - Main:760 (iii): "The horizon's expansion sets $\tau_B \sim H^{-1}$";
+  - ch04:30: "the cosmological initial conditions are taken as given";
+  - Substratum:379: "The framework does not derive that structure". This is weak support:
+    "that structure" there is the FRW symmetry pattern, not the expanding branch as such.
+- **Target (T2).** A theorem from (S, φ), the partition and stated preparation data
+  selecting $H > 0$ over $H < 0$ for the observer's accessible window.
+- **Closest result.** None. Main §4.6 is symmetric (§C.3) and does not mention expansion.
+- **Gap.** All of T2.
+- **Route.** None inside the corpus. Under T-invariance the reversed history is the
+  contracting branch, admissible on equal terms (§C.5). Any T2 needs boundary data or a
+  measure that is not T-invariant.
+
+### B.3 Thermodynamic step (A3)
+
+- **Strongest existing results.**
+  1. **$\rho_s = \rho_{\rm crit}$** [PP], GR §7.2 (GR:428–430), with its temperature input
+     from GR §3.1. The identity is $\rho_s = 3c^2H^2/(8\pi G)$: an instantaneous equality,
+     even in H. It concerns the *boundary* entropy's energy density, not the entropy of
+     matter inside the horizon.
+  2. **Theorem A.6** [PP]. See B.4 and §C.4.
+  3. **The unital-channel remark at GR:805** [PP]. If the hidden state is maximally mixed,
+     the marginal channel $\Phi_V$ is unital, so $S_{\rm matter}$ does not decrease *over
+     that one channel application*. Composing steps needs the hidden state re-prepared
+     maximally mixed at each step, which is a Markov assumption the exact dynamics does not
+     supply (Main:760 (ii)).
+  4. **The fast-bath erasure lemma** [PP], Main:165–167. In the iterated-channel
+     description under uniform erosion, the readback gap decays as
+     $\delta_0 e^{-(k-1)\tau_S/\tau_B}$. This is a statement about steps *away from a
+     rooted record*, and applies in either direction (§C.3).
+- **Target (T3, relational).** For a stated class of histories and a coarse-grained
+  visible (matter) entropy $S_{\rm cg}$, on the observer's window,
+  $\operatorname{sign}\dot S_{\rm cg} = \operatorname{sign}\dot A$. Explicit premises would
+  include:
+  - (P1) a non-stationary (low-entropy or product) preparation;
+  - (P2) a *monotonicity* theorem for $S_{\rm cg}$ in the exact dynamics, not a bound, over
+    the window;
+  - (P3) a relation tying the entropy gradient to $\dot A$: either a common boundary for
+    (P1) and the minimum of A, or an identity linking $S_{\rm cg}$ (or the hidden sectors'
+    coarse-grained entropy) to A.
+- **Closest results.**
+  - (P1) is the hypothesis of Theorem A.6.
+  - (P2) is supplied only in the iterated-channel description under (EM). Main:760 (ii)
+    states that its governance of the exact compressions "is not established".
+  - (P3) is the identification GR:805 names as missing: "a relation between horizon area
+    and the hidden sectors' coarse-grained entropy along the evaporation".
+- **Gap.** (P2) for the exact dynamics, and all of (P3). The cited identity (1) cannot
+  supply (P3): it relates $\rho_s$ to H², not the matter entropy to A. The step as written
+  ("via Jacobson's coupling … the entropy of any horizon-bounded system increases") has
+  no derivation behind it.
+- **Route inside the corpus.** Partial. The most direct candidate is capacity growth:
+  - GR §5's boundary-mode count makes the hidden capacity scale with $A/\epsilon^2$;
+  - Theorem A.6 already allows the sector grouping $V_t, B_t, D_t$ to change with t;
+  - so a growing A enlarges the maximum coarse-grained entropy.
+
+  This supplies room for entropy growth, not growth itself: it still needs (P1), (P2) and
+  the GR:805 identification. The literature analogue (Layzer; Frautschi, "Entropy in an
+  expanding universe") is **[LIT2]**, not read first-hand for this note.
+
+### B.4 Generalized second law and Theorem A.6 (A4)
+
+- **Strongest existing result. Theorem A.6** [PP], GR:799–803. For a product state at
+  $t = 0$ evolving under $U_\varphi$, and for every $t$ and every regrouping:
+  $\sum_X S(\rho_{X_t}(t)) \ge \sum_X S(\rho_X(0))$.
+  - The proof (subadditivity, then unitarity) uses no sign of t.
+  - $U_\varphi$ is invertible, so the bound holds for $t < 0$ as well.
+  - It is a two-sided bound with its minimum at the preparation time.
+- **Target (T4, GSL).** $S_{\rm matter}+S_{\rm BH}+S_{\rm dS}$ is non-decreasing along the
+  observer's clock. Premises:
+  - (P1) as above;
+  - the GR:805 identification, $S_{\rm BH}(t) = S(\rho_{B_t}(t))$ and
+    $S_{\rm dS}(t) = S(\rho_{D_t}(t))$, or a coarse-grained replacement;
+  - a monotonicity theorem: Theorem A.6 bounds against $t=0$ only, and §C.4 shows it does
+    not give monotonicity.
+- **Closest.** Theorem A.6 and the unital-channel remark, both stated at GR:805 with their
+  limits.
+- **Gap.** Exactly as GR:805 states, plus monotonicity.
+- **Route.** None inside the corpus. **GSL: open**, already stated correctly at GR:805 and
+  ch07:155–157.
+- **GR:607.** It reads "non-decreasing from a product initial state". That is refuted as a
+  monotonicity statement by §C.4 and should state the bound.
+
+### B.5 Memory step (A5)
+
+- **Strongest existing results.** Main's realization and readback theorems are *rooted*:
+  - the process dilation starts from $P_0 \otimes \mu_H$ at $t = 0$ (Main:518);
+  - the recurrence/indivisibility theorem uses rooted maps $\Gamma_t$ with a shared hidden
+    prior (Main:137);
+  - the conditions (C2) and (C4) are phrased forward from a write to a readback (Main:76,
+    :80).
+
+  Each holds for every bijection φ. Since $\varphi^{-1}$ is also a bijection, each has a
+  mirror statement for $t < 0$ from the same root (§C.3, *closure under inversion*).
+- **Target (T5).** In the realization class used for the cosmological cut, records of the
+  visible past are written and read back in the direction of increasing A. Premises:
+  - a preparation (root) at $t_0$, the low-A end of the window;
+  - persistence and readback forward from $t_0$;
+  - a reason the root sits at the low-A end and not the high-A end.
+- **Closest.** The rooted theorems give "away from the root", in both directions. The
+  orientation of the protocol is a convention (Substratum:409; Methodology:184, "related by
+  forward iteration of φ").
+- **Gap.** The link between the root time and the horizon clock (the third premise of T5).
+  The text's own mechanism ("the observer's measurement protocol commits to one") is the
+  orientation convention.
+- **Route.** None inside the corpus beyond the common-boundary form of §C.6 (β).
+
+### B.6 Measurement step (A6)
+
+The step inherits A5 by its own text. The record/prepare asymmetry is the memory
+asymmetry under the protocol convention. Target, gap and route are those of B.5.
+
+### B.7 Alignment: a common mechanism (A7)
+
+- **Strongest existing result.** None proves alignment.
+  - Substratum §5.4 asserts it.
+  - ch18:216 attributes it to "the same initial conditions", that is, to boundary data
+    posited at the same time. That supplies a common cause by stipulation, not a mechanism.
+  - ch18:216 also asserts "C1–C4 hold at the cosmological boundary", against GR:52 ("(C4)
+    … **Not presently discharged.**").
+- **Target (T7, common alignment mechanism).** For a class $\mathcal C$ of tuples
+  (finite bijection φ, partition family $V_t$, preparation), stated in advance, and for
+  every history in $\mathcal C$ on the observer's accessible window $W$:
+  $$\operatorname{sign}\dot A(t) = \operatorname{sign}\dot S_{\rm cg}(t) = \text{(direction of record formation)} = \text{(direction of preparation→registration)},\quad t\in W,$$
+  derived from one property of the substratum and partition, not from four separate
+  inputs. Explicit premises any proof would need:
+  - (Q1) **A as a substratum quantity.** The horizon area, or a proxy, is a function of the
+    substratum configuration and partition, so that "$\dot A$" is defined inside OI. At
+    present A enters only through the ambient FLRW background (B.1).
+  - (Q2) **Entropy–area relation.** An identity or inequality linking $S_{\rm cg}$, or the
+    hidden sectors' coarse-grained entropy, to A: the GR:805 gap.
+  - (Q3) **Exact monotonicity on W.** For $S_{\rm cg}$ under the exact compressions, not
+    only the iterated channel: Main:760 (ii).
+  - (Q4) **Record–entropy link.** Records form in the direction in which $S_{\rm cg}$
+    increases. The corpus has the ingredients (erasure lemma, readback lemma), but only
+    rooted and two-sided.
+- **What the obstructions allow (§C.6).** T7 is a *relational*, T-even statement, and none
+  of §C.1–§C.5 excludes it. They exclude only:
+  - a global version (beyond a recurrence window);
+  - a version that fixes the *direction* from (S, φ) alone.
+- **Closest results.** Theorem A.6 (for Q3, but two-sided and a bound); `counting_passive`
+  (orients energy given monotone Ω, not time); the boundary-mode count (for Q1/Q2, at the
+  exact de Sitter calibration only).
+- **Gap.** Q1–Q4, each unproved.
+- **Route.** None found. The weak conditional statement available now is §C.6 (β): with a
+  product preparation and the minimum of A at the same time, both quantities are bounded
+  below by their values there. Its content is the posited coincidence. That is the ch18
+  position, not a mechanism.
+- **Verdict: open (unresolved).** No counterexample.
+
+### B.8 "The framework derives the arrow" (A8)
+
+- **What the text itself concedes.**
+  - Substratum:409: the §4.6 content is symmetric; which direction is "future" is
+    convention; "The framework therefore derives the existence of a monotonic parameter
+    along observer worldlines, not a substrate-level arrow of time."
+  - Substratum:413: the derivation is conditional on the ordering of moments.
+- **What survives.**
+  - A1 (conditional, ambient, constant in exact de Sitter);
+  - A9 (symmetric selection);
+  - the convention.
+- **Not supported as stated.**
+  - The arrow as a physical direction: A2, A3, A5, A6.
+  - The arrow as a set of aligned arrows: A7.
+  - "Resolves the arrow-of-time question" and "closes the Ellis-axis objection" rest on
+    the cascade, which is open.
+- **Excluded by §C.1.** Only one reading is excluded: a globally monotone function of the
+  substratum state. Substratum:409 already agrees with that ("agreeing with the objection's
+  premise").
+- **Not checked.** The source of the "Ellis objection" is uncited in the corpus and was
+  not checked.
+
+### B.9 Observer-selection rule (A9)
+
+- **Strongest existing result.** Main §4.6 theorem and corollaries [PP]. Conditional on:
+  - (EM);
+  - the rate condition $\tau_{\rm mix}\ll\tau_S$;
+  - bounded coupling-graph diameter;
+  - canonical typicality [31];
+  - stated for the description generated by iterating $\mathcal K_\Sigma$ (Main:760 (ii)).
+
+  No kernel counterpart: the coverage ledger grades it `GAP`.
+- **Time symmetry.** By §C.3 the theorem holds verbatim for $\varphi^{-1}$, with the same
+  gap.
+- **What it does and does not do.**
+  - It selects *where* (non-equilibrium regions).
+  - It does not select *which way*.
+  - It does not supply a low-entropy boundary. A selected non-equilibrium region of a
+    reversible system is, absent past-boundary data, as likely to sit at a local entropy
+    minimum as on a monotone slope: the Boltzmann/Loschmidt point, in c.6643691865.
+- **Verdict.**
+  - Retain as proved under its hypotheses.
+  - The title "past-hypothesis replacement" attributes to it the replacement of a boundary
+    condition. Its body scopes that replacement to the role "to select observer-compatible
+    configurations". The title does not.
+
+### B.10 Inventory (A10)
+
+Substratum:413 states that "the arrow's single open dependency is the ordering question and
+nothing further". The same section conditions its chain on further inputs:
+- (EM) (Substratum:401, :407);
+- the iterated-channel description (Substratum:401, :407, :409);
+- the expanding branch ($H > 0$), implicit in "the $\Lambda$-dominated phase" and "the
+  accelerating phase" (Substratum:405). The condition $\ddot a > 0$ named there does not
+  itself imply $H > 0$.
+- a measurement-protocol orientation (Substratum:407).
+
+The thermodynamic step also needs (P1)–(P3) of B.3. Each of these is a dependency other
+than the ordering, so the sentence is false as stated: **refuted** by exhibition. The fix
+is a complete list, not a proof.
+
+***
+
+## C. Obstructions that bound any recovery
+
+Each is stated as a constraint on what a recovery can deliver, not as a refutation of
+alignment.
+
+### C.1 Reversibility with finite recurrence: no globally monotone state function [EA]
+
+Let φ be a bijection of the finite set S, so $\varphi^N = \mathrm{id}$ for some N
+(Main:123, Step 1). Suppose $F: S \to \mathbb R$ satisfies $F(\varphi s) \ge F(s)$ for all
+s. Summing $F(\varphi s) - F(s) \ge 0$ around each cycle gives 0, so every term vanishes,
+and $F\circ\varphi = F$.
+
+The same holds for any functional of the evolving distribution:
+- $p_t = p_0\circ\varphi^{-t}$ is N-periodic;
+- so is every coarse-grained entropy $S_{\rm cg}(p_t)$;
+- so are the sector entropies of Theorem A.6.
+
+**Consequence.** Any monotonicity is confined to a window shorter than the recurrence time
+and anchored by data on that window. Main §2.3's own theorem exhibits this: the one-step
+distinguishability contracts, $d(\delta_iT,\delta_jT) < 1$, and returns to 1 at $t = N$
+(Main:125–129).
+
+### C.2 Rooted results are closed under inversion [HA]
+
+Main's realization, readback and indivisibility results, and Theorem A.6:
+- quantify over all bijections φ of a finite set;
+- start from a preparation at a root time, $P_0\otimes\mu_H$ or a product state.
+
+$\varphi^{-1}$ is a bijection of the same set, so each result applies to the process run
+backward from the same root. For the existence theorem at Main:518, this means the class of
+laws realizable from a root is the same for backward runs. For the bound results
+(Main:137, Theorem A.6), the conclusion holds for $t < 0$. Their conclusions describe
+evolution *away from the root* in both directions.
+
+**The abstract step is kernel-checked [K].** `no_universal_oriented_property`
+(`verification/lean-mathlib/OIBridge/OrientationClosure.lean`:249) states:
+- let a class C be closed under a map Θ and inhabited;
+- let a property P satisfy $P(R) \Rightarrow \neg P(\Theta R)$ on C;
+- then P cannot hold throughout C.
+
+The instantiation by hand [HA]:
+- Θ is $\varphi\mapsto\varphi^{-1}$ (with the same root);
+- C is a hypothesis class closed under inversion;
+- P is "the conclusion points toward $t > 0$ and not toward $t < 0$".
+
+A one-sided arrow therefore cannot come from these results without one of:
+- a hypothesis not closed under $\varphi\mapsto\varphi^{-1}$;
+- a second boundary.
+
+### C.3 Stationarity: block entropies and $M_k$ are reversal-invariant; §4.6 is symmetric [HA]
+
+Take the canonical measure μ of Main Lemma 3 (uniform, φ-invariant) on
+$S = \mathcal C_V\times\mathcal C_H$, and $X_t = \pi_V\varphi^t(s)$ with $s\sim\mu$,
+$t\in\mathbb Z$.
+
+1. **Stationarity.** $\mu\circ\varphi^{-1} = \mu$, so $(X_{t+m})_t$ has the law of
+   $(X_t)_t$, and the block entropy $H_n := H(X_t,\dots,X_{t+n-1})$ does not depend on t.
+2. **Reversal invariance of $M_k$.** Main defines
+   $M_k := I(X_{<k};X_{k+1}\mid X_k)$ (Main:471). By the chain rule,
+   $$M_k = H(X_0..X_k) + H(X_k,X_{k+1}) - H(X_k) - H(X_0..X_{k+1}) = H_{k+1}+H_2-H_1-H_{k+2}.$$
+   For the reversed process $Y_t = X_{-t}$, the same four blocks appear, shifted, so
+   $M_k^{\rm rev} = M_k$. Every block-entropy functional is reversal-invariant, including
+   excess entropy and entropy rate.
+3. **Consequence for (C4) and (C2).** Under the invariant measure, (C4) holds forward at an
+   order exactly when it holds backward, as measured by $M_k>0$. The condition then
+   distinguishes no direction.
+4. **Main §4.6 is symmetric, under two stated assumptions.**
+   - **Assumption (P_Σ).** Main:746 does not define $P_\Sigma$. Read it as the
+     μ-orthogonal conditional expectation onto the coarse observables of Σ, self-adjoint on
+     $L^2(\mu)$. Main's own examples (Main:760 (ii)) fit this reading.
+     - With it, the compression for $\varphi^{-1}$ is
+       $P_\Sigma\mathcal U_\varphi^{\dagger}P_\Sigma = \mathcal K_\Sigma^{\dagger}$.
+     - Its spectrum is the complex conjugate of $\mathcal K_\Sigma$'s, with the same moduli.
+     - With Π the projection onto the invariant part,
+       $\|(\mathcal K_\Sigma^\dagger)^n-\Pi\| = \|(\mathcal K_\Sigma^n-\Pi)^\dagger\| = \|\mathcal K_\Sigma^n-\Pi\|$,
+       so the gap and $\tau_{\rm mix}$ are the same, in the spectral/$L^2$ reading of
+       $\tau_{\rm mix}$.
+   - **Assumption (bounded range of $\varphi^{-1}$).** The proof uses φ's causal cone
+     (Main:754, :364), so the φ⁻¹ version needs $\varphi^{-1}$ to have bounded per-step
+     range. A local bijection's inverse need not be local: $x_i \mapsto x_i + x_{i+1}$
+     mod 3 on an odd ring has a dense inverse. The assumption holds under the corpus's
+     T-invariance posit with a site-local T, where $\varphi^{-1} = \mathcal T\varphi\mathcal T$.
+     That is the case of the second-order wave update of SM Theorem 17 (kernel:
+     `wave_time_reversal_invariant`, `verification/lean/OI_Time_Reversal.lean`), whose time
+     reversal swaps the two time layers site by site.
+   - Under both assumptions, (EM) holds for φ exactly when it holds for $\varphi^{-1}$, and
+     the theorem and both corollaries hold verbatim for $\varphi^{-1}$. Substratum:409
+     already says the bound "applies equally in both directions"; this is the reason.
+   - Items 1–3 need neither assumption.
+
+A stationary process need not be reversible in law. A 3-cycle on visible states has
+$P(0\to1)\ne P(1\to0)$. That asymmetry is a chirality of φ, not an entropy or memory
+gradient: items 1–3 still hold. Under the corpus's T-invariance posit it disappears up to
+relabeling (§C.5).
+
+### C.4 Theorem A.6 is two-sided, and its monotone reading fails [EA, counterexample]
+
+**Two-sidedness.** By C.2, the bound $\Sigma(t)\ge\Sigma(0)$ holds for $t<0$ too. The
+preparation time is a global minimum of the bound, not the start of a monotone increase.
+
+**Counterexample to "non-decreasing" (GR:607).** Take:
+- $V$ and $B$ one bit each, $D$ trivial;
+- $\varphi(v,b) = (v\oplus b, b)$, a permutation of $\{0,1\}^2$ (CNOT, an involution);
+- the product initial state $\delta_{v=0}\otimes\mathrm{unif}_b$, diagonal, so all
+  entropies are Shannon entropies.
+
+Then:
+
+| t | joint state | $S_V$ | $S_B$ | $\Sigma(t)$ |
+|---|---|---|---|---|
+| 0 | uniform on $\{(0,0),(0,1)\}$ | 0 | ln 2 | ln 2 |
+| 1 | uniform on $\{(0,0),(1,1)\}$ | ln 2 | ln 2 | 2 ln 2 |
+| 2 | $\varphi^2 = \mathrm{id}$: as at t = 0 | 0 | ln 2 | ln 2 |
+
+- Theorem A.6's hypotheses hold and its conclusion $\Sigma(t)\ge\ln 2$ holds.
+- $\Sigma(2) < \Sigma(1)$, so the sum is not non-decreasing for $t>0$.
+- The same pattern recurs for every φ that correlates the sectors at some t (C.1).
+
+This is a counterexample to the monotone reading only. Theorem A.6 itself is correct.
+
+### C.5 T-invariance: the reversed visible process is the relabeled forward process [HA]
+
+Assume the corpus's posits:
+- an involution $\mathcal T$ with $\varphi = \mathcal T\varphi^{-1}\mathcal T$ (ch06:62;
+  SM Theorem 19);
+- $\mathcal T$ preserves the partition, acting as θ on visible states;
+- the counting (uniform) measure.
+
+Then for every n and every history $(i_0,\dots,i_n)$:
+$$P(X_0=i_0,\dots,X_n=i_n) = P(X_0=\theta i_n,\dots,X_n=\theta i_0).$$
+
+*Proof.*
+1. $s\mapsto s' := \mathcal T\varphi^n s$ is a μ-preserving bijection.
+2. $\varphi\mathcal T = \mathcal T\varphi^{-1}$ gives $\varphi^k s' = \mathcal T\varphi^{n-k}s$.
+3. Its visible part is $\theta\,\pi_V\varphi^{n-k}s$.
+
+This extends SM Theorem 19 from two-time counts to histories. It is a hand argument in
+this note, not a corpus result; a reviewer should check it.
+
+**Consequence.** Under the invariant measure the trace-out produces a process whose time
+reversal is its θ-image. The trace-out alone introduces no asymmetry. That refutes the
+unconditional reading of ch18:202. ch18:210's conditional reading (initial conditions)
+stands.
+
+**Witness: memory without an arrow [EA].** Take $\mathcal C_V = \mathcal C_H = \{0,1\}$,
+$\varphi(v,h) = (h,v)$ and the uniform prior.
+- The visible process is $(a,b,a,b,\dots)$ with $a, b$ independent and uniform.
+- Its law is reversal-symmetric.
+- $M_1 = I(X_0;X_2\mid X_1) = 1$ bit, since $X_2 = X_0$.
+
+So history readback ((C4) at order 1) holds at full strength with no time asymmetry.
+
+**Alignment is unaffected.** This assumes, as standard, that the horizon area A and the
+coarse-grained entropy $S_{\rm cg}$ are T-even: T leaves the value at each instant fixed
+and reverses the order of instants. Then $\mathcal T$ maps an expanding history in which
+entropy increases to a contracting history in which entropy decreases. Both are *aligned*.
+The T-image of an aligned history is aligned.
+
+### C.6 What alignment can still mean
+
+The obstructions constrain direction, globality and stationarity. Alignment survives them
+in four senses.
+
+- **(α) Relational alignment (T-even).**
+  - The statement $\operatorname{sign}\dot S_{\rm cg} = \operatorname{sign}\dot A$ is
+    invariant under $t\to-t$, because both sides flip.
+  - Neither C.1 nor C.3 nor C.5 excludes it. For example, an identity
+    $S_{\rm cg} = f(A)$ with f increasing would give it under any measure.
+  - Recovering alignment means proving T7 (B.7): premises Q1–Q4. The direction would then
+    still come from boundary data or convention.
+- **(β) Conditional alignment from a common boundary [EA, weak].**
+  - Take a product state at $t_0$ (Theorem A.6), with $A(t_0)$ the minimum of A on the
+    window. Then $\Sigma(t)\ge\Sigma(t_0)$ and $A(t)\ge A(t_0)$ for all t on the window.
+  - The statement is true and trivial. Its content is the posited coincidence of the two
+    boundary data, which is ch18:216's "same initial conditions".
+  - A non-trivial version needs Q3 (monotonicity) and a reason for the coincidence (Q2).
+- **(γ) Windowed alignment.** On finite S any alignment holds at most on windows shorter
+  than the recurrence time (C.1).
+- **(δ) Direction.** The direction never comes from (S, φ) alone (C.1, C.3). It comes from
+  boundary data (a preparation, the expanding branch) or from convention.
+
+***
+
+## D. Verdicts
+
+| Claim | Verdict | Precise content |
+|---|---|---|
+| A1 clock (area) | **proved** (ambient) | **T1a** (apparent horizon): FLRW + Friedmann, with $H>0$ and $\rho+p/c^2\ge0$ on the interval; non-decreasing, strict iff $\rho+p/c^2>0$, constant in exact de Sitter.<br>**T1b** (event horizon, flat): with the conditions on all of $[t,\infty)$, non-decreasing; $\dot R_e(t)>0$ iff H is non-constant on $[t,\infty)$; constant iff exact de Sitter on $[t,\infty)$.<br>Entropy reading: **open** (GR:56).<br>The ä > 0 criterion is **refuted as a general kinematic criterion** (phantom, NEC-violating) and **false as strict increase in exact de Sitter**. In the corpus's NEC-respecting background, ä > 0 is not the operative condition (§E.4). |
+| A2 cosmological | **conditional** | On the expanding branch, an input. Why that branch: **open**. |
+| A3 thermodynamic | **open** | Gap: (P2) exact monotonicity and (P3) the entropy–area relation. Direction: conditional on asymmetric boundary data. The cited ρ_s = ρ_crit does not reach the step. |
+| A4 GSL | **open** | Gap as at GR:805, plus monotonicity. |
+| A4 Theorem A.6 | **proved** [PP] | Two-sided bound relative to t = 0. Monotone reading (GR:607) **refuted** by §C.4. |
+| A5 memory | **conditional** (direction) / **open** (alignment) | Direction from a non-stationary preparation and the orientation convention. Alignment gap: the link between root time and the horizon clock. |
+| A6 measurement | as A5 | — |
+| A7 alignment | **open (unresolved)** | Not refuted. Target T7, premises Q1–Q4 open. Weak conditional form (β) only. |
+| A8 "derives the arrow" | **not supported as stated** | Reduces to A1 + A9 + convention. The global reading is excluded (§C.1). |
+| A9 selection | **proved** [PP], conditional | Hypotheses as at Main:744–760; time-symmetric; no direction; no low-entropy boundary. |
+| A10 inventory | **refuted as stated** | Further dependencies exhibited (B.10). |
+| A11 trace-out | **refuted** (unconditional reading) | §C.3, §C.5. The conditional reading (ch18:210) stands. |
+
+***
+
+## E. Horizon kinematics
+
+These are ambient GR computations [AMB], written out as exact algebra [EA]:
+- c is kept explicit in the flat case;
+- c = 1 is used for general curvature;
+- ρ is mass density and p pressure, with Λ included as a component ($p_\Lambda = -\rho_\Lambda c^2$).
+
+### E.1 Which horizon
+
+The corpus defines the horizon causally as the **event horizon** (GR:56), but computes with
+the **apparent/Hubble horizon**, area $A = 4\pi c^2/H^2$ (GR:56; Substratum:405's
+$S_{\rm dS}=A/4l_p^2$). They coincide only in exact de Sitter. Phase 2 text must name the
+one it uses. Below, both are treated.
+
+Equations used:
+- From Cai & Kim, arXiv:hep-th/0501055v1, read first-hand **[LIT]**, with n = 3 and c = 1:
+  - $\dot H - k/a^2 = -4\pi G(\rho+p)$ (eq. 2.14);
+  - $\tilde r_A = (H^2+k/a^2)^{-1/2}$ (eq. 2.3);
+  - the event horizon (eq. 2.4);
+  - $\dot{\tilde r}_A = -H\tilde r_A^3(\dot H - k/a^2)$ (eq. 2.15).
+- Standard, not from Cai & Kim: the acceleration equation
+  $\ddot a/a = -\tfrac{4\pi G}{3}(\rho+3p)$.
+
+### E.2 Apparent horizon
+
+- **General curvature (c = 1).** From the two displayed relations,
+  $$\dot{\tilde r}_A = 4\pi G\,H\,\tilde r_A^{\,3}\,(\rho+p),\qquad \frac{dA}{dt} = 8\pi\tilde r_A\dot{\tilde r}_A = 32\pi^2G\,H\,\tilde r_A^{\,4}(\rho+p).$$
+- **Flat (k = 0), c explicit.** Here $R_A = c/H$ and $A = 4\pi c^2/H^2$, so
+  $$\frac{dA}{dt} = -\frac{8\pi c^2\dot H}{H^3},\qquad \dot H = -4\pi G\Big(\rho+\frac{p}{c^2}\Big)\ \Rightarrow\ \frac{dA}{dt} = \frac{32\pi^2Gc^2}{H^3}\Big(\rho+\frac{p}{c^2}\Big).$$
+
+**Role of each condition.**
+- **The sign is $\operatorname{sign}(H)\cdot\operatorname{sign}(\rho+p/c^2)$.**
+- **Given H, the NEC ($\rho+p/c^2\ge0$) fixes the sign.**
+  - With $H>0$ the area is non-decreasing.
+  - On the contracting branch ($H<0$) it is non-increasing.
+  - In the equality case $\rho+p/c^2\equiv0$ (exact de Sitter) it is constant on either
+    branch.
+- **Strict growth** on the expanding branch holds iff $\rho+p/c^2>0$.
+- **In ΛCDM** today, $\rho+p/c^2 = \rho_m + \tfrac43\rho_r > 0$, so A grows strictly. It
+  approaches the de Sitter value asymptotically, as $\rho_m\to0$.
+
+### E.3 Event horizon (flat)
+
+$R_e(t) = a(t)\int_t^\infty c\,dt'/a(t')$, which exists iff the integral converges. Then:
+- $\dot R_e = HR_e - c$.
+- If $H>0$ and $\dot H\le0$ on $[t,\infty)$ (flat with NEC), then $H(t')\le H(t)$ for
+  $t'\ge t$. So $a(t)/a(t')\ge e^{-H(t)(t'-t)}$, hence $R_e\ge c/H$ and $\dot R_e\ge0$.
+- Equality holds iff H is constant on $[t,\infty)$: exact de Sitter from t on.
+
+**The event horizon is defined by the future boundary $t\to\infty$.** Its time reverse is
+the particle horizon $R_p(t) = a(t)\int_{t_i}^t c\,dt'/a(t')$, defined by the past boundary.
+Choosing the event horizon (GR:56) therefore already fixes which end of time is the
+reference. Under $t\to-t$ they exchange: H flips sign, while $\dot H$, A and $\rho+p$ do
+not. So $dA/dt$ is T-odd, and the clock's direction is the direction of expansion.
+
+### E.4 Exact de Sitter, and why ä > 0 is the wrong criterion
+
+**Exact de Sitter.** $\rho+p/c^2 = 0$, so $\dot H = 0$ and $R_A = R_e = c/H$ are constant
+(also Cai & Kim, below their eq. (2.4)). The horizon clock **stops** on GR's own
+calibration surface (GR:56).
+
+**The criteria compared.** $\ddot a>0 \iff \rho+3p/c^2<0$, whereas area growth (with
+$H>0$) $\iff \rho+p/c^2>0$. Single-fluid flat solutions with
+$w = p/(\rho c^2)$ constant give:
+
+| Regime | ä | $dA/dt$ | Shows |
+|---|---|---|---|
+| matter/radiation ($w = 0, \tfrac13$) | $<0$ | $>0$ | ä > 0 is **not necessary** |
+| quintessence ($-1<w<-\tfrac13$) | $>0$ | $>0$ | — |
+| exact Λ ($w=-1$) | $>0$ | $=0$ | ä > 0 does **not give increase** |
+| phantom ($w<-1$; $a\propto(-t)^{2/(3(1+w))}$, $t<0$) | $>0$ | $<0$ | ä > 0 is **not sufficient** (NEC violated) |
+
+In the phantom row, $H = 2/(3(1+w)t) > 0$ and $\dot H = -2/(3(1+w)t^2) > 0$, so A shrinks
+while $\ddot a/a = \dot H + H^2 > 0$.
+
+**The correct criterion is "$H>0$ and the NEC".** It gives non-decreasing A, strict where
+$\rho+p/c^2>0$, and constant in exact de Sitter.
+
+**Scope of the verdict on ä > 0.** The verdict covers only the class each row rules out.
+Substratum:405 says "monotonic for as long as $\ddot a>0$" and "every observer in the
+accelerating phase agrees that $S_{\rm dS}$ increases".
+- **As a general kinematic criterion it is refuted** by the phantom row. That row violates
+  the NEC.
+- **As strict increase it is false in exact de Sitter**: ä > 0 there, and A is constant.
+  Exact de Sitter is GR's own calibration surface.
+- **Inside the corpus's background it is weakly true but not operative.** That background
+  respects the NEC, and GR excludes phantom crossing, conditionally (GR:12, :358, :553).
+  There, ä > 0 with H > 0 gives $dA/dt\ge0$, so the statement holds weakly. ä > 0 is still
+  not the operative condition: growth is governed by $H>0$ and $\rho+p/c^2>0$, and the
+  criterion omits the decelerating era, in which A also grows.
+- The "not necessary" row shows the criterion is incomplete. It does not refute a
+  sufficiency claim.
+
+***
+
+## F. Wording principles for Phase 2
+
+General rules:
+- Minimum change (§A.25).
+- Remove an assertion rather than tag it (§A.30).
+- No self-narration of revisions (§A.27, §A.30, §A.33). Substratum:401's "Before §4.6 … With
+  §4.6 in place" goes.
+- Neutral register in `papers/` (§A.32). The book keeps its essay register but not a
+  stronger status.
+- Every book edit is mirrored in FULL.
+
+### F.1 Retain
+
+- Keep a derived or proved claim only with its hypotheses stated where it is proved.
+- The selection theorem is described as selecting *where* observers can exist, in the
+  iterated-channel description under (EM) and the rate condition. It is never described as
+  selecting a time direction or supplying a low-entropy boundary.
+- Theorem A.6 and "the generalized second law is open" stay as written at GR:797–805.
+
+### F.2 Conditional
+
+Name every input in the sentence that uses it:
+- **The background.** "An expanding FLRW background ($H>0$) satisfying the null energy
+  condition", as an input, not derived.
+- **The horizon.** Name it: "the apparent (Hubble) horizon, $A = 4\pi c^2/H^2$".
+- **The clock.** Replace "monotonic / monotonically increasing" with "non-decreasing,
+  strictly increasing while $\rho+p/c^2>0$, constant in exact de Sitter". Drop the
+  "$\ddot a>0$" and "accelerating phase" criteria (§E.4).
+- **The thermodynamic direction.** Conditional on a low-entropy or product preparation;
+  the horizon-area/entropy identification is open.
+- **Memory and measurement.** Conditional on a non-stationary preparation and the
+  orientation convention.
+- **GR:607.** State the bound: "bounded below by its value at a product initial state, at
+  every t". GR:797's "non-decreasing" defines the generalized second law and is followed by
+  "supplies the following inequality toward it". It is a correct definition, not an
+  overstatement (delta review c.6649698455, item D), and stays unchanged.
+- **Substratum:407's thermodynamic step.** Drop the universal "the entropy of any
+  horizon-bounded system increases with the observer's clock". No result supports it (B.3).
+- **GR:214.** "monotone bath count ⇒ passive (energy orientation)", not a time direction.
+
+### F.3 Unresolved
+
+- State declaratively, where the cascade is stated, that the alignment of the
+  thermodynamic, memory and measurement arrows with the horizon parameter is not
+  established. Do not call it refuted or impossible.
+- Remove the arrow-as-derived claims from the headline sentences (Substratum:14, 26, 403,
+  415; Methodology:184 "none is concealed"; Explainer:930's sentence; ch03:58), rather than
+  qualifying them in place.
+- Where the "three projections" framing lists the arrow, narrow the item to what is
+  established: the symmetric selection rule and, on the expanding branch, a non-decreasing
+  horizon parameter. Do not delete the framing wholesale.
+- Substratum:413: list the dependencies (B.10) in place of "nothing further".
+
+### F.4 Main:758 retitle, and Main's upstream independence
+
+- **The retitle (owner-approved).** "**Corollary (past-hypothesis replacement, with its rate
+  condition).**" becomes "**Corollary (observer-selection rule, with its rate
+  condition).**".
+  - The statement and hypotheses are unchanged.
+  - The retitle removes an attribution and adds no claim, citation or premise, so it
+    creates no dependence of Main on downstream content.
+  - The body's last sentence is already scoped ("to select observer-compatible
+    configurations … within that description"). The authorization covers the title and
+    :760 only, so the body stays.
+- **Necessary consequences of the retitle**, authorized by the owner ruling #1112
+  c.6649643759 (amaybaum, 2026-11-05):
+  1. **Main:744.** It refers to "the rate condition $\tau_{\rm mix}\ll\tau_S$ stated in the
+     past-hypothesis corollary". After the retitle that label dangles. The minimal fix
+     changes the label phrase only.
+  2. **`verification/coverage/LEDGER.json`.** `tools/proof_census.py` builds a statement's
+     identifier from its parenthetical label. The entry
+     `MAIN:C-past-hypothesis-replacement-rate-condition` would no longer match, and
+     `tools/coverage_check.py` (release gate) would fail. The entry has to be renamed to
+     the new identifier, with its fingerprint updated, in the same commit. The ledger is
+     mutable, not a frozen record.
+- **Main:760 (iii): NO CHANGE**, by owner ruling c.6649643759 item 4 ("Record NO CHANGE
+  unless a concrete defect is identified and reviewed").
+  - **Earlier recommendation, now ruled.** The sentence reads "The horizon's expansion
+  sets $\tau_B \sim H^{-1}$ structurally, not thermally". It makes:
+  - no monotonicity claim;
+  - no time-direction claim;
+  - no low-entropy claim.
+
+  It describes the cosmological realization, which Main states is developed in companion
+  work and consumed by no proof in Main (Main:72). This note finds no ARROW-1 defect in it.
+  If the owner or a reviewer wants the expanding branch named there, the only admissible
+  form names it as an observed input, without importing §E or citing GR or Substratum, so
+  that Main stays independent of downstream results (Code Review Rules).
+- **Authorized scope** (c.6649643759 item 1):
+  - the Main:744 label phrase only;
+  - in LEDGER.json, the `id`, `label` and `fingerprint` fields only, with `kernel: GAP`
+    and the assumptions unchanged. The new identifier is
+    `MAIN:C-observer-selection-rule-rate-condition`.
+- **ch18:190 (c.6649643759 item 2).** Correct the unconditional unidirectionality claim
+  (§C.1), without asserting that observer experience must reverse during recurrence.
+
+***
+
+## Literature
+
+**Read first-hand [LIT]:**
+- R. M. Wald, "The Arrow of Time and the Initial Conditions of the Universe",
+  **arXiv:gr-qc/0507094v1** (21 Jul 2005). Its argument: a present thermodynamic arrow
+  requires a very special early state, and Wald argues that a dynamical origin of that
+  state is not plausible. Its footnote 1 describes Carroll–Chen as giving a *locally*
+  observed arrow in a universe time-symmetric on large scales, with inflation and
+  "deflation" episodes on equal footing. This parallels C.2/C.6 (δ). It supports nothing
+  about OI.
+- R.-G. Cai and S. P. Kim, "First Law of Thermodynamics and Friedmann Equations of
+  Friedmann-Robertson-Walker Universe", **arXiv:hep-th/0501055v1** (8 Jan 2005), JHEP 0502
+  (2005) 050. Used for:
+  - eqs. (2.3) apparent-horizon radius, (2.4) event-horizon definition, (2.14) Friedmann
+    $\dot H$ equation, (2.15) $\dot{\tilde r}_A$;
+  - the remark that in pure de Sitter the three horizons coincide and are constant.
+
+**Known only second-hand [LIT2], not used as support:** Carroll–Chen, hep-th/0410270 and
+gr-qc/0505037 (via Wald's footnote only); Layzer and Frautschi on entropy in an expanding
+universe; Davies on cosmological event-horizon area and the generalized second law.
+
+## Not checked
+
+- The source of the "Ellis objection".
+- The companion *Physics Modulo Gauge* (cited at Substratum:403, :413).
+- Main reference [31].
+- Any Lean counterpart: none was found by grep, and no Lean build was run.
+- Byte identity of the FULL mirrors across whole chapters.
+- §C.2, §C.3 and §C.5 are hand arguments written for this note, and §C.4 and §E are exact
+  algebra written for this note. None was machine-checked. The review packet below lists
+  them.
+
+## Review packet: what an independent reviewer must verify
+
+1. **§E.2.** The apparent-horizon derivative in both forms, the sign role of H, the NEC
+   role, and the equality case.
+2. **§E.3 and T1b.** The event-horizon bound $R_e\ge c/H$ when $\dot H\le0$ on
+   $[t,\infty)$; the de Sitter equality case; the event/particle exchange under $t\to-t$;
+   and that T1b's hypotheses are stated on $[t,\infty)$.
+3. **§E.4.** The four regimes, especially the phantom solution's signs, and the scope of
+   the ä > 0 verdict: general criterion refuted; strict increase false in exact de Sitter;
+   weakly true but not operative in the NEC background.
+4. **§C.4.** The CNOT table, that it satisfies Theorem A.6's hypotheses, and that it
+   refutes only the monotone reading.
+5. **§C.3.** The block-entropy identity for $M_k$ and its reversal invariance under the
+   invariant measure. The adjoint-compression argument under the stated $P_\Sigma$
+   reading. The bounded-range assumption for $\varphi^{-1}$, and that it holds for a
+   site-local T.
+6. **§C.5.** The multi-time extension of SM Theorem 19; the swap witness; and, assuming A
+   and $S_{\rm cg}$ are T-even, that T maps aligned histories to aligned histories.
+7. **§C.2.** That each rooted result cited (Main:137, :518, Theorem A.6) has its backward
+   counterpart. That `no_universal_oriented_property` (OrientationClosure.lean:249) states
+   the abstract step as quoted, and that the hand instantiation is valid.
+8. **§B.3 (P3).** That $\rho_s=\rho_{\rm crit}$ concerns the boundary entropy only, and
+   cannot by itself supply the entropy–area relation.
+9. **§B.7.** That T7's premises Q1–Q4 are each unproved in the corpus, and that no
+   counterexample to alignment exists, so the verdict is open, not refuted.
+10. **§B.10.** That each listed dependency is used by Substratum §5.4's own text.
+11. **§F.4.** Whether the Main:744 label phrase and the LEDGER.json rename are in scope as
+    necessary consequences of the retitle.
+12. **Locators.** That every quoted locator matches `f5c281ae`.
diff --git a/verification/audits/manuscript/bell-operational-reconciliation-20261029/sc42-integration-result.json b/verification/audits/manuscript/bell-operational-reconciliation-20261029/sc42-integration-result.json
index 9343f7f97..f3366b70c 100644
--- a/verification/audits/manuscript/bell-operational-reconciliation-20261029/sc42-integration-result.json
+++ b/verification/audits/manuscript/bell-operational-reconciliation-20261029/sc42-integration-result.json
@@ -1562,6 +1562,14 @@
       "ok": true,
       "predicate": "chapter line mirrored in FULL: **Substrate freedom and the realism question.** A uniform hi..."
     },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: The reconstruction theorem identifies the **local lattice/ga..."
+    },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: The framework's claim is that quantum mechanics and general ..."
+    },
     {
       "ok": true,
       "predicate": "chapter line mirrored in FULL: **The visible-sector projection: quantum mechanics.** Chapte..."
@@ -1572,7 +1580,19 @@
     },
     {
       "ok": true,
-      "predicate": "chapter line mirrored in FULL: The cosmological horizon entropy provides a structural clock..."
+      "predicate": "chapter line mirrored in FULL: The apparent-horizon area provides a structural clock on tha..."
+    },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: The framework's content here is precise: on the expanding br..."
+    },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: **The common structure.** The three projections share three ..."
+    },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: The reconstruction theorem and the substratum gauge group es..."
     },
     {
       "ok": true,
@@ -1702,6 +1722,10 @@
       "ok": true,
       "predicate": "chapter line mirrored in FULL: **The ledger.** Conceded: Bell's mathematics; EPR's logic an..."
     },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: The structural form of Spinoza's account — one substance, mu..."
+    },
     {
       "ok": true,
       "predicate": "chapter line mirrored in FULL: The framework's programme relative to Worrall is to specify ..."
@@ -1738,6 +1762,10 @@
       "ok": true,
       "predicate": "chapter line mirrored in FULL: The chapter has three pieces of work. §4.2 develops the sing..."
     },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: A useful diagnostic for the discipline is what the framework..."
+    },
     {
       "ok": true,
       "predicate": "chapter line mirrored in FULL: The framework's general rule is *group structural, couplings..."
@@ -2410,6 +2438,22 @@
       "ok": true,
       "predicate": "chapter line mirrored in FULL: The framework's commitment to finite information is therefor..."
     },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: **The arrow as observer-emergent.** The framework's answer i..."
+    },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: **Detailed balance and the thermodynamic arrow.** The framew..."
+    },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: **The cosmological arrow.** Whether the thermodynamic arrow ..."
+    },
+    {
+      "ok": true,
+      "predicate": "chapter line mirrored in FULL: Both arrows are traced to the same initial conditions of the..."
+    },
     {
       "ok": true,
       "predicate": "chapter line mirrored in FULL: Second, the framework's finiteness *does not require* compac..."
diff --git a/verification/coverage/LEDGER.json b/verification/coverage/LEDGER.json
index 8d5977e0c..617d8f61c 100644
--- a/verification/coverage/LEDGER.json
+++ b/verification/coverage/LEDGER.json
@@ -1420,12 +1420,12 @@
       "note": ""
     },
     {
-      "id": "MAIN:C-past-hypothesis-replacement-rate-condition",
+      "id": "MAIN:C-observer-selection-rule-rate-condition",
       "paper": "papers/Main.md",
       "line": 752,
       "kind": "Corollary",
-      "label": "(past-hypothesis replacement, with its rate condition)",
-      "fingerprint": "a2f5c0c423aa9ca2",
+      "label": "(observer-selection rule, with its rate condition)",
+      "fingerprint": "df04fba6b2d81e20",
       "area": "oi-foundations",
       "assumptions": "the observer-selection theorem plus its rate condition — the manuscript is explicit that the two bounds do NOT compose without it; stated for the coarse-grained description generated by iterating the effective channel",
       "checks": [],