The blueprint close-out: the jitter floor closed, the cis-drive shown data-blocked

The absolute-clock appendix left a binding blueprint with two open items, one permanent ceiling, and one declared scope, and this appendix takes that blueprint to its honest terminus — closing what can be closed and saying exactly what is missing where it cannot. The first open item was the jitter floor: the strength of the order prediction met its pre-registered bound on average, but at the lower edge of citation uncertainty it slipped below, because one muscle-lineage founder sat in the cascade as an artificial source with no depth, despite being a mid-stage gene. The fix is a single cited edge — the somite marker that precedes and gives rise to that founder — which removes the artifact, deepens the muscle genes to where the literature places them, and lifts the lower edge of the jitter band above the floor, with no inversion introduced and the cascade still acyclic. That closes the floor, and it is closed honestly: the strength is now robust to rank jitter but it stays a measured correlation, never promoted to a verified invariant, because no theorem in the substrate forces a number. The second open item was the cis-code to drive map: the wish to read each edge's drive off the sequence rather than set it by a uniform rule. Here the appendix does the real measurement and reports a negative. For every edge whose two genes have a cached real promoter, the parent regulator's cited binding motif is counted in the child's promoter and compared against a dinucleotide-shuffle background, and the result is that the signal is at or below background for most edges — and below background for the textbook direct edge of the skeletal program, where the repression is known to act through distal enhancers, not the proximal promoter. The drive is simply not in the window the model measures. That is not a flaw in the theory; it is a limit of the data, and the appendix names exactly what would lift it: distal-enhancer and chromatin-accessibility sequence the kit does not contain. So the cis-drive map stays a declared choice, the binding completion criterion is not met, and physical completion stays false — but the blueprint is now fully mapped, every remaining question resolved into closed, ceiling, data-blocked with the gap named, or declared scope.

The order-grammar program reached, with the absolute-clock appendix, a state where one channel (the global zero-point) was closed and a written blueprint enumerated what remained: B2 (derive the per-edge cis-drive from sequence), B4 (lift the rank-jitter lower edge above the strength floor), B3 (a permanent firewall ceiling on absolute strength), and B5 (a built organ, out of scope). This appendix resolves the two open items, add-only, on the identical sixty-three-gene atlas, with every inherited stiffness re-checked digit-for-digit against the parent database and no new γ introduced. B4 is closed by a single cited regulatory edge from the somite/dermomyotome marker MEOX1 to the myogenic founder PAX7, which in the prior appendix was an artificial cascade source — indegree zero, depth zero — despite a mid-range Carnegie rank, exactly the kind of coverage gap the program had already fixed for the limb and Hox founders. Adding the edge deepens PAX7 and the myogenic determination genes behind it; the global Spearman of depth against Carnegie rank rises from about 0.75 to about 0.86, and the lower fifth percentile of the ±1 rank-jitter distribution rises from about 0.62, below the pre-registered 0.70 floor, to about 0.73, above it, with the fraction of jittered draws meeting the floor going from under a half to over ninety-nine percent. The added edge is rank-tie-concordant — both genes carry the same cited Carnegie rank, so no inversion is introduced — and the cascade remains a directed acyclic graph. The strength claim is therefore now jitter-robust, but it stays locked, never verified; the firewall permanently forbids a verified absolute strength, so the gate fails closed on any such promotion. B2 is resolved as data-blocked by measurement. A deterministic, parameter-free probe scans each child promoter — the same ±2 kb window the stiffness operator reads — for the parent regulator's cited consensus motif on both strands, and compares the count against a dinucleotide-preserving shuffle null under the substrate seed. Only a minority of the both-cached edges exceed background, the mean standardized score is near zero, and decisively the canonical SOX9-represses-RUNX2 edge sits below background, consistent with that repression acting through distal enhancers rather than a proximal site array. The per-edge drive is not in the proximal window, so it cannot be read from it; closing B2 would require distal-enhancer and accessibility sequence the kit does not contain, which is named as the unblock condition. The cis-drive map therefore stays a declared modelling choice, not a measured one, and physical completion stays false because the binding criterion requires both B2 and B4 closed. The whole inherited apparatus — the acyclic cascade, the wavefront and quorum theorems, the sequence-derived order-invariance, the strength floor, the segmentation sub-clock, the two-anchor absolute clock — is re-audited byte-for-byte on the extended cascade. The fail-closed gate passes eighteen of eighteen deterministically under a double hash, holds B4 at locked and B2 at data-blocked, and keeps physical completion false while the named distal measurement is absent.

Why this appendix exists

The absolute-clock appendix did two things deliberately. It closed the one timing channel that could be closed without tuning — the global zero-point — and it wrote down, as a binding part of the kit, exactly what remained before the program could call itself complete. That blueprint named four things. Two were genuinely open: a way to read each regulatory edge's drive off the sequence rather than setting it by a uniform rule, and a way to lift the lower edge of the strength test above its pre-registered floor, since the strength met the floor on average but slipped below it at the margin of citation uncertainty. One was a permanent ceiling the firewall will never let past — a verified, invariant absolute strength — and one was declared out of scope, a built or simulated organ. The honest thing, having written that blueprint, is to act on it: to close the two open items if they can be closed under the same discipline, and, where one cannot, to prove that it cannot and say precisely what is missing. That is what this appendix is for. It is not a new theory; it is the close-out of an existing one, and its value is in being exact about the difference between a gap that a single cited fact fills and a gap that only new measurement can fill.

B4 — the jitter floor, closed by one cited source-fix

The strength of the order prediction is the Spearman correlation between a gene's depth in the regulatory cascade and its Carnegie onset rank, and the pre-registered discipline was that this correlation must clear a floor of seven tenths and stay there under a small perturbation of the cited ranks. In the absolute-clock appendix it cleared the floor on average but its lower fifth percentile under a one-rank jitter sat just below, and the appendix reported that honestly and kept the strength a locked, jitter-marginal claim rather than promoting it. The diagnosis here is specific. The genes dragging the lower edge down were the muscle lineage — the myogenic determination factors and their founder, PAX7 — and the reason was a coverage gap, not a failure of the idea: PAX7 carried a mid-range Carnegie rank but sat in the cascade as an artificial source, with no regulator above it and therefore zero depth, because the somite marker that precedes it had no edge drawn to it. That is the very same artifact the program had already fixed twice before, when the limb founders and the terminal Hox genes were given their true upstream regulators. The fix is one cited edge from the somite and dermomyotome marker to the myogenic founder — the somite gives rise to the Pax7-positive progenitors, a textbook lineage relation — and its effect is exactly what the diagnosis predicts. The founder gains real depth, the determination genes behind it deepen with it, the global correlation rises from about three-quarters to about six-sevenths, and the lower fifth percentile of the jittered distribution rises from below the floor to above it, with almost every jittered draw now meeting the floor. The edge introduces no inversion, because the two genes carry the same cited Carnegie rank — a concordant tie — and the cascade stays acyclic. The closure is real and minimal: one edge, cited, mechanistic, with the statistic now robust to the jitter that previously broke it. And it stays locked, not verified: a robust correlation on cited data is still a measurement, and the firewall permanently forbids calling an absolute strength a verified invariant.

B2 — the cis-drive, shown to be data-blocked, with the gap named

The second open item asked for more: not a robustness fix but a derivation, reading each edge's drive off the regulatory sequence instead of setting it by the uniform amplitude rule the prior appendices declared. The honest test of that wish is to measure whether the parent regulator's binding motif is actually present in the child gene's promoter — the same window the stiffness operator reads — above what chance composition would give. The probe is parameter-free: for every edge whose two genes have a cached real promoter, it counts the parent's cited consensus motif on both strands of the child promoter and compares that count against a dinucleotide-preserving shuffle of the same sequence, under the substrate's own seed, calling an edge above background only when its count exceeds the shuffle's upper tail. The result is a clean negative. Only a fifth of the both-cached edges exceed background, the mean standardized score across edges is near zero, and the single most diagnostic edge — the skeletal program's textbook direct repression, where the master chondrogenic factor represses the osteogenic one — sits below background, exactly as the biology predicts, since that repression is known to act through distal enhancers and protein contacts rather than a proximal array of binding sites. The conclusion is not that the regulation is absent but that it is not in the window the model measures: the per-edge drive lives in distal enhancers and accessible chromatin outside the proximal promoter, and a proximal scan therefore cannot recover it. This is a limit of the data, not a defect of the theory, and it has a precedent in the program's own history, where an order-score was shown to be non-identified because the measurement could not separate two collinear quantities. So the appendix does the disciplined thing: it grades the cis-drive map a declared choice rather than a measured one, it names the exact data that would lift the block — distal-enhancer and chromatin-accessibility sequence for the cascade genes — and it holds physical completion false, because the binding criterion requires both open items closed and this one is blocked by measurement.