# Appendix BS: mathematical and evidence audit Date: 21 September 2026. Scope: the supplied `Theory_of_Reality_v1.70_draft.md`, especially Appendix BS. This is a separate audit, not a revised edition of the paper. The supplied draft and other engines were left unchanged. ## Result The continuous rigid-motion obstruction and the explicit extra-coordinate rotation survive. Several proposed **inferences from observations do not**: visible slice dimension need not count hidden dimensions; a nonrigid inversion need not pass through a globally achiral configuration; no observed flips do not determine a compactification radius; and a temperature-independent rate floor does not certify rigidity. There are two explicit geometric counterexamples below, with written all-time arguments, 16 regression tests, and seven separately implemented exact polynomial checks. These are mathematical checks, **not physical observations or evidence for an additional spatial dimension**. The separate implementation is a computational cross-check, not independent peer review. ## Provenance and limits Source SHA-256: `8932ff1185c821eba3fe63458d7cc7fef42724cfc80e3b110aab1566da55f489` The source's 17-row ledger does contain the stated counts: six deformation routes, five law/not-object entries, three developmental origins, and three null bounds. This is a text tally only. The actual `dimensional_verifier.py` was not supplied or located, so its execution, rule precedence, inputs, and claimed empirical verification remain unverified. This audit does not independently establish the history of every phenomenon in that ledger. The mathematical catalog in BS.B is not certified by this audit. Numerical entries and broad historical claims there require a separate source-by-source review. ## 1. What survives, with the correct scope **BS.T1.** If a labeled chiral body follows a continuous Euclidean isometry path starting at the identity, the determinant cannot change from +1 to −1. It cannot end at its mirror under those assumptions. This is a valid written proof, not a new physical law. **BS.T2.** One additional spatial coordinate suffices for the construction \[ R_\theta(x,y,z,0)=(x,y,z\cos\theta,z\sin\theta),\qquad 0\leq\theta\leq\pi. \] All ambient pairwise distances are preserved, and the endpoint is \((x,y,-z,0)\). This establishes geometric sufficiency, not physical accessibility, a transition rate, or an electromagnetic observation rule. **BS.T3.** A genuinely continuous, identity-preserving, fully rigid mirror conversion is inconsistent with this frozen Euclidean 3D model. Identifying its physical cause still requires checking the observation map, continuity, identity, geometry, and dynamics. Four supplied `PASS` flags do not themselves establish those conditions experimentally. **Correction to the earlier audit:** BS.A.3 already restricts its two-orbit claim to the full isometry orbit of a fixed chiral body, modulo proper motions. That restricted claim is valid. The standard cover fact in BS.A.5 is also correct for a connected base manifold: its orientation double cover is disconnected exactly when the base is orientable. An orientation cover records local orientation choices; its existence does not establish a second material universe. The surrounding mirror-world interpretation goes beyond the theorem. ## 2. Counterexample: a point-sized slice with one hidden coordinate Give the vertices distinct permanent labels and take the entire solid tetrahedron \[ K=\operatorname{conv}\{(0,0,0),(1,0,1),(0,1,1),(0,0,1)\}. \] Its signed edge determinant is 1, so its 3D volume is \(1/6\); it is not a lower-dimensional input. The fixed labels make it chiral. Apply exactly BS.T2's rotation in the \(z,w\) plane, with **one** extra coordinate and no translation. Every point of \(K\) is \[ (a,b,a+b+d),\qquad a,b,d\geq0,\quad a+b+d\leq1. \] For every intermediate angle, \(\sin\theta>0\). Thus \[ w=(a+b+d)\sin\theta=0 \quad\Longleftrightarrow\quad a=b=d=0. \] The entire visible literal slice is the origin throughout \(0<\theta<\pi\). The body reappears mirrored at the end, with every ambient distance unchanged. **BS.9's rule “point implies at least three extra dimensions” is therefore false as stated.** A line-slice counterexample is \[ \operatorname{conv}\{0,(1,0,0),(0,1,1),(0,0,2)\}. \] The same one-axis rotation leaves the segment from \(0\) to \((1,0,0)\) as the slice. Its initial signed edge determinant is 2, again nonzero. The failure is boundary tangency. The rank formula the verifier appears to intend can be recovered **conditionally**: write the hidden-coordinate map on a full affine 3-plane as \(Bu+b\). A nonempty solution set has affine dimension \(3-\operatorname{rank}B\). If the intersection meets the relative interior of the solid, its local slice has that dimension, implying \(m\geq3-d_s\) for \(m\) hidden coordinates. Without that interior-intersection condition, a boundary slice may be much smaller. Sparse atoms, finite visibility, occlusion, and projected images require their own models. **Slice is not projection.** At \((\cos\theta,\sin\theta)=(3/5,4/5)\), the first tetrahedron has a point slice but a full-dimensional 3D projection, with signed edge determinant \(3/5\). The actual projected volume is \(1/10\). A camera rendering a projection cannot be analyzed as a literal hyperplane intersection without justification. ## 3. Counterexample: deformation into a mirror while remaining chiral Consider five distinctly labeled points moving entirely in ordinary 3D: \[ \begin{aligned} p_0&=(0,0,0),&p_1&=(1,0,0),&p_2&=(0,1,0),\\ p_3&=(2,0,\cos\theta),&p_4&=(0,2,\sin\theta), \end{aligned} \qquad 0\leq\theta\leq\pi. \] They never collide. Their two oriented tetrahedral determinants are \(\cos\theta\) and \(\sin\theta\), so their squared sum is 1. Consequently the whole configuration spans 3D at every time. An isometry fixing all distinct labels must then be the identity; no orientation-reversing symmetry fixes the configuration. It remains chiral throughout. The final configuration is exactly the initial configuration reflected in \(z=0\). Intermediate distances change: this is a **nonrigid** path, not a counterexample to BS.T1. It is a counterexample to the universal claim that every in-space inversion must pass through a globally achiral state. A particular tetrahedral determinant can cross zero without the whole multi-point object becoming achiral. The safe replacement is: a continuous 3D chirality conversion cannot stay within the original body's proper rigid-motion orbit. Do not require an achiral waypoint for every deformable object. ## 4. What rigidity measurements must establish Unchanged endpoints do not imply unchanged intermediate distances. Even constant central bond lengths are insufficient. Hold \(p_0=0,p_1=e_1,p_2=e_2\) fixed and move \[ p_3=(3\sin\theta/5,4\sin\theta/5,\cos\theta). \] All three distances from the central point stay 1. Handedness reverses, with changing bond angles and other pairwise distances, without collisions. Therefore bond lengths alone cannot certify the required rigidity. Finite sampling needs explicit temporal-resolution and dynamical assumptions to exclude intersample rearrangements. This retains the earlier correction that integer-tick capture does not enforce continuous non-crossing. A sealed laboratory also does not prove contractibility as a purely topological fact; locality is a justified model assumption requiring an appropriate physical argument. BS.T1′ correctly points to orientation-reversing transport as a loophole for local frames. But orientation holonomy alone does not prove preservation of every finite-body distance on an arbitrary curved manifold. A metric and a finite-body transport construction must be specified. ## 5. Null observations do not automatically constrain hidden radius H-Inv allows \(\Gamma_4\geq0\) with no positive lower bound as a function of radius. For **every** radius, \(\Gamma_4=0\) gives probability 1 of no flips. Therefore no number of null observations can upper-bound the radius within that hypothesis alone. This remains true even if a proposed geometric size threshold is granted. BS.T7 analyzes a particular rotation and wrap-avoidance criterion. Its heuristic \(h\lesssim\pi R_c\) is not, as written, a necessary theorem for all shapes, placements, product metrics, and possible paths. A defensible radius inference would first specify geometry and a quantitative rate/accessibility law. The quartz estimate lacks an observed event history, detection efficiency, justified independent-unit exposure, and a statistical model. Ancient bulk enantiopurity is not a record of zero detected unit-cell flips. For illustration only, if zero independent Poisson events really had been monitored perfectly over \(N=10^{22}\) units for \(10^8\) Julian years, then \[ \Gamma_{95}=\frac{-\log(0.05)}{NT} =9.492902735\times10^{-38}\ \mathrm{s}^{-1}. \] The manuscript's approximately \(3\times10^{-38}\) is an inverse-exposure scale, not that 95% limit. **Neither number is a measured quartz bound in this audit.** Two primary-source checks also narrow the claimed external support: - Planck's search allowed both orientable and nonorientable phase matching, but its circle search was restricted to antipodal/back-to-back pairs and had masking and detectability qualifications. The no-circle result constrains specified topology signatures; it does not show that every orientation-reversing loop exceeds 27 Gpc or universally close the laboratory loophole. See [Planck 2015 XVIII, revised June 2016, §§4.2 and 5.1.1](https://arxiv.org/pdf/1502.01593). - The 38.6 μm result is a 95% bound on a **gravity-strength Yukawa interaction range**, not a universal compactification-radius or molecular-flip bound. That conversion requires a physical model relating the parameters. See [Lee et al., published 13 March 2020](https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.124.101101). ## 6. The proposed experiment is not yet a dimensional detector An isotope-insensitive low-temperature floor does not demonstrate unchanged internal geometry throughout a transition. Background processes, readout artifacts, residual environmental effects, and other 3D mechanisms must be modeled and tested. Conversely, higher-dimensional kinematics alone predicts neither a nonzero rate nor isotope independence; inertia, confinement and coupling parameters can still matter. Three numerical corrections to BS.6: | Quantity | Checked result | Scope | |---|---:|---| | Mass of \(10^{18}\) molecules at 100 g/mol | 166.054 μg | Not “about a microgram”; mass requires molar mass | | Net conversions giving \(10^{-4}\) loss of initially unit enantiomeric excess | \(5\times10^{13}\) | Each net L→R conversion changes excess by \(2/N\) | | Probability of at least 10 Poisson events when the mean is 10 | 54.207% | Expected count is not a 95%-power calculation | For symmetric first-order L↔R conversion at per-direction rate \(\Gamma\), \(ee(t)=ee(0)e^{-2\Gamma t}\). A stipulated \(10^{-4}\) relative loss over \(3\times10^7\) s corresponds to \(\Gamma=1.666750006\times10^{-12}\ \mathrm{s}^{-1}\). This is a sensitivity conversion, not a confidence limit or validated instrument specification. Ordinary ammonia has indistinguishable hydrogens and mirror symmetry; its umbrella inversion is not an enantiomer conversion of a chiral labeled body unless distinguishing labels are actually supplied. Its near-24-GHz inversion spectrum is a quantum energy splitting, with rotational-state dependence—not a universal classical count of flips per second. A stationary energy eigenstate need not oscillate between localized configurations. See [Flambaum and Kozlov, 2007](https://arxiv.org/pdf/0704.2301), especially the inversion-doublet discussion and equation (1). BS.T5 also needs a dynamical and measurement model. A geometric path and additivity do not establish universal charge/energy leakage, or torque-free changes in observed angular momentum. A stipulated split \(Q_{\rm total}=Q_{\rm visible}+Q_{\rm hidden}\) is bookkeeping; whether a detector measures that split is an additional assumption. ## 7. Definitions and classifier repairs In Definition A.1, an observation-expressible invariant differing from its value at \(x_0\) already implies unreachability: if \(b=gx_0\), invariance forces \(I(b)=I(x_0)\). The two listed conditions are therefore not independent. The following 2×2 table quietly changes “visible” to ordinary observation distinguishability. Keep that table if desired, but distinguish it from the stronger invariant-separation certificate. BS.A.4's displayed semicircle has endpoints \((z,0)\) and \((-z,0)\): its diameter is along the **visible** axis, not the hidden axis. Fixed points with \(z=0\) remain visible throughout. Recommended output contract for a future verifier: | Input evidence | Permissible conclusion | |---|---| | Calibrated evidence against a frozen continuous, rigid, identity-preserving local 3D model | Conditional rejection of that model, with stated error control | | Known deformation/reconnection | The rigid/topological obstruction does not apply; no universal achiral waypoint inferred | | Metadata flags without supporting measurements | Claimed certificate conditions, not verified detection | | Literal slice plus known affine support and certified interior intersection | Conditional rank/codimension bound | | Arbitrary visible slice or rendered projection | No dimension-count inference | | Null events with justified exposure, efficiency, background and rate model | Model-conditional rate limit | | Null events without a rate-versus-radius model | No radius bound | The seven BS.9 classes are not mutually exclusive from the displayed rules alone: for example, `invariant FAIL` also occurs in the null-bound rows. An implementation needs explicit precedence and separation between event type, certificate evidence, and inferential output. `UNKNOWN` must never be silently promoted to `PASS`. ## 8. Reproduction and next test Run with Python 3; no third-party dependencies: ```bash python3 audit_bs.py python3 independent_symbolic_check.py ``` Optional source-specific check: ```bash python3 audit_bs.py --source /path/to/Theory_of_Reality_v1.70_draft.md --output new_results.json ``` Both scripts refuse to overwrite an existing requested output file. `BS_CHECKED_RESULTS.json` records the 16 passing tests, source hash and calculations; `BS_SYMBOLIC_RESULTS.json` records seven passing polynomial identities. The polynomial checker shares no functions with the regression script and uses a different determinant algorithm. Written arguments above establish the continuous geometric results; finite rational sampling alone does not. **Next bounded test:** obtain the original sighting-verifier source and feed it the point-slice tetrahedron, the nonrigid always-chiral five-point path, an unobserved intersample deformation, and a zero-rate model at arbitrary radius. It must not convert these records into extra dimension counts or radius bounds. Then freeze a sensor-level likelihood model and test detection power and false positives on fresh synthetic records before any physical experiment. No physical measurement, extra-dimensional detection, universal no-detection survey, or theological inference is supplied by this work. No previous closure, recurrence-class, angular-impulse, or continuous-noncrossing correction is changed.