The absolute clock: pinning the global zero-point from two measured anchors

The order-grammar completion converted the cascade's developmental order into a span of days using the measured segmentation clock, but left one piece explicitly open: a single global zero-point that would pin every gene not just to a cadence but to an absolute embryonic day. That open piece is the one most prone to being faked by tuning, because any zero-point can be slid to fit — so it is closed here in the one way that cannot be tuned. Two anchors are taken from two independent measurement modalities. The slope of the clock comes from the in-vitro segmentation oscillator: the cited five-hour period times the cited forty-two somite pairs is eight and three-quarter days of somitogenesis, spread across the cited stage bracket from the first somites to the bulk of the series. The zero-point comes from the in-vivo first heartbeat: the first cardiac contraction is a Carnegie-stage-ten event at about twenty-two days after fertilization, a functional landmark of the heart, measured separately from any clock. These two numbers fix a straight line from Carnegie stage to absolute day with no free parameters at all, because neither anchor uses any stage's tabulated day. Every cited Carnegie day and every gene's day is therefore a held-out prediction, and the line reproduces them inside the somite-clock window to within about half a day — five held-out stages to at most six tenths of a day, and thirty-one held-out genes to at most six tenths of a day. The clock is a measurement, so it is graded locked, never a verified invariant: the gate fails closed if anyone tries to promote it. And the honesty is kept where it belongs — the single somite-rate drifts after somitogenesis ends, by up to about six and a half days at the latest stages, exactly where the oscillator no longer sets the pace; that drift, and the deeper question of why the rates are what they are, are reported as bounds, not hidden. No body is built. Physical completion stays false, because two channels remain.

The completion appendix left a single named gap in absolute timing — the global zero-point — and this appendix closes it at the level the firewall allows, by a parameter-free calibration validated out of sample. The construction is a straight line from Carnegie stage number to post-fertilization day, with its slope and its intercept each supplied by a different real measurement so that nothing is fitted. The slope is the segmentation modality: the human presomitic-mesoderm oscillator's cited period of about five hours, multiplied by the cited count of about forty-two somite pairs, gives a somitogenesis duration of eight and three-quarter days, and dividing that by the cited stage span over which the somites form fixes the days advanced per Carnegie stage. The intercept is the cardiac modality: the first embryonic heart contraction, long described as a Carnegie-stage-ten event and dated to about twenty-two days after fertilization under the standard age convention, fixes one absolute point on the line. Crucially, neither anchor consumes any stage's tabulated day, so the cited Carnegie stage-day table and every gene's onset day become held-out targets. The line reproduces the held-out stage days inside the somite-clock window — stages eight through thirteen, excluding the stage-ten point the intercept passes through — to a maximum error near six tenths of a day, and it predicts the absolute days of thirty-one anchored genes in that window to the same tolerance, every one inside the acceptance band. The cardiac landmark independently agrees with the tabulated stage-ten day, a corroboration recorded but not used to fit. The result is graded locked real-data, never verified: there is no theorem in the substrate that forces a particular zero-point, only two cited measurements that happen to pin one, and the appendix says so. The honest residual is reported rather than buried — the single somite-derived rate drifts after somitogenesis, reaching several days of error by the late stages, because the oscillator stops setting the body's pace there; unifying the somite tempo with the later tempo into one derived rate law is left as a bound, as is the origin of the rates themselves, which is a magnitude question the firewall permanently excludes, exactly as it excludes the absolute strength of the order prediction. Inheritance is byte-exact: all sixty-three driver stiffnesses are checked digit-for-digit against the parent completion appendix read from disk, and the whole inherited apparatus — the acyclic cascade, the wavefront and quorum theorems, the sequence-derived drive, the strength floor, the segmentation sub-clock — is re-audited on the identical atlas. The fail-closed gate passes seventeen of seventeen deterministically under a double hash, holds both the strength claim and the clock claim at locked, and keeps physical completion false while two channels remain open.

Why this appendix exists

The completion appendix closed four obstacles the order grammar had left open, and closed three of them fully, but it was careful to halve the fourth rather than overstate it. It attached the real measured period of the segmentation clock and the cited count of somite pairs and showed that their product, about eight and three-quarter days, matched the cited window over which the somites form — which converts the cadence of the clock into a span of days. But it stopped there and said so: a single absolute zero-point that would pin every gene in the atlas not merely to a cadence but to an absolute embryonic day would require a second independently measured anchor, and that the appendix did not yet have, so the global zero-point was left explicitly open. That open piece is the most delicate of all the timing claims, because it is the one most easily faked. A cadence is a difference and is hard to slide; a zero-point is an origin and can be slid anywhere to make a fit. So the discipline demands that the zero-point be closed only if it can be closed without tuning at all — with anchors that are cited measurements rather than free knobs, and with a prediction tested on data the anchors never touched. This appendix exists to do exactly that, and nothing more: to supply the one missing measured anchor, fix the clock with no free parameters, and prove the fix on held-out genes, while keeping the result a measurement and not pretending it is a law.

Two anchors from two modalities, and no free parameters

The clock is the simplest object that can carry the claim: a straight line from Carnegie stage number to post-fertilization day. A straight line has two degrees of freedom, a slope and an intercept, and the whole point is that each one is supplied by a separate real measurement, so that none is fitted. The slope comes from the segmentation modality, in vitro. The human presomitic-mesoderm oscillator runs with a cited period of about five hours, and a cited count of about forty-two somite pairs forms over the whole somite series, so somitogenesis lasts about eight and three-quarter days; spreading that duration across the cited stage bracket from the first somites to the bulk of the series fixes how many days the embryo advances per Carnegie stage. The intercept comes from the cardiac modality, in vivo. The first contraction of the embryonic heart has long been described as a Carnegie-stage-ten event and is dated, under the standard age convention, to about twenty-two days after fertilization — the heart being the first organ to function, its first beat is one of the most firmly fixed absolute landmarks in human embryology. That single absolute point fixes the intercept. What makes the construction tuning-proof is what it does not use: neither anchor consumes any stage's tabulated day. The slope is built from a period, a count, and a stage span; the intercept is built from one functional landmark. So the cited Carnegie stage-day table is never an input — it becomes a held-out target. Every gene's onset day, read through its cited stage, is held out the same way. The line is therefore not fitted to the days it is then asked to predict; it is constructed from two measurements in two modalities and tested against everything else.

The held-out test, and the honesty about where it ends

The test is what a parameter-free calibration earns the right to claim. With the slope at about two and two-tenths days per stage and the intercept near zero, the line predicts the absolute day of each Carnegie stage, and these predictions are compared with the cited stage-day table that was never used to build it. Across the somite-clock window — the stages over which the oscillator sets the pace, excluding only the stage-ten point that the intercept passes through by construction — the predictions land within about half a day of the cited days, with a maximum error near six tenths of a day. The same line predicts the absolute days of the thirty-one anchored genes whose cited stages fall in that window, and every one lands inside the acceptance band, the worst being the cardiac transcription factor at the late edge of the window, off by the same six tenths of a day. The cardiac anchor, fixed independently from the heart's first beat, agrees with the tabulated stage-ten day to within its stated uncertainty — a corroboration that the two modalities meet, recorded but deliberately not folded back into the fit. And then the appendix is honest about where the clock stops working. The same single rate, extended past the end of somitogenesis, drifts: by the latest stages it is off by several days, up to about six and a half, because once the somite series is complete the oscillator no longer sets the body's tempo and a single somite-derived rate cannot speak for the later embryo. That drift is reported as a bound, not hidden, and it is itself a kind of evidence — the clock works precisely where its mechanism applies and visibly fails where the mechanism stops, which is what a real mechanism does rather than a curve fitted to everything. Unifying the somite tempo with the later tempo into one derived rate law is left open as a bound; so is the deeper question of why the period is five hours and why there are forty-two somites, which is a magnitude-and-origin question the firewall permanently excludes, in the same way and for the same reason it permanently excludes the absolute strength of the order prediction. The rates are used as cited measurements to pin the zero-point; their origin is never claimed.