# Main acceptance: bounded date/era addendum

2026-09-08 UTC. **Accepted with the existing interpretive qualifications.**
The original565-scenario run is unchanged. This acceptance closes the finite
two-event numerical follow-through, not release activation or deployment.

## What was changed and what was reused

The saved Batch18 decisions replace La Gomera's nonhistorical1449 proxy with
the1488–1489 sequence and replace the West Bank/East Jerusalem2024 evidence-
cutoff proxy with the1967 onset. The latter moves from1990-present to1945–1989.
The detailed evidence cutoff remains2024; no terminal year is asserted.
Candidate nominal-date eligibility rises627→629; baseline stays628. There
are still no separately measured continuous movement durations.

The canonical score matrix is exactly
`2cb65a7ef78805045ec44061e6561d50b133fb217f9f56d109631ee9bdfc1b18`.
Its accepted fit was reused, not recomputed. Scores, U/NA, families, membership,
the six fixed program/influence weight vectors and the frozen-v6 baseline are
unchanged. Two new unique matrices were fitted: equal-era weighting and
era-specific median imputation. Their14 scenario aliases share those two
results rather than repeating fits. All565 original scenarios are accounted
for;511 original date results reproduced before their amended calculations.
Those511 date records and seven weighted aliases comprise518 explicit patches.

## Numerical interpretation

Both affected branches retain their previous factor count: eight for equal-
era weighting and seven for era-median imputation. Consequently the complete
scenario accounting remains524 nine-factor,29 eight-factor and12 seven-factor
records. These are reused/related diagnostics, not565 independent replications.
The two canonical unweighted fits still retain nine factors.

Compared with their own earlier branch, the amended equal-era fit has maximum
eight-/nine-dimensional subspace angles of0.38750°/0.29929°. Era-median
imputation changes by3.47764°/2.42355°. Both produce finite full-rank8- and9-axis
loadings without fit warnings. Small incremental changes from this particular
correction do not negate the much larger previously reported sensitivities to
weighting, missingness or indicator selection.

The canonical nominal-date correlations change by at most0.002213 in ordinary
Spearman rho across the eight named axes; the largest absolute change after
controlling both unknown and structural-missing counts is0.007795. Their raw
signs do not reverse. This comparison describes a two-event metadata amendment,
not an inferential fairness threshold, a causal effect of historical time or
permission to use calendar envelopes as process durations. Full coefficients,
partial correlations, nominal p-values and all amended scenario records remain
available rather than selecting only favorable results.

## Verification and preservation

- Six small dependency/refusal tests pass normally and with optimization.
- The separate serialized checker passes normally and with optimization,
  checks559 addendum/reused-fit bindings and all63 original scenario-input
  bindings, and rejects missing-scenario, invented-third-fit and altered-date-
  subset controls. It performs no new estimator fit. `CHECKS_FINAL.json` is
  the complete receipt; `CHECKS.json` preserves the earlier narrower check.
- Completed raw result SHA-256:
  `56a3a5db90f84c1e8691160ed6eb565b30df60f5f8062eb4bbb8a1d44c745287`.
- Addendum run manifest:
  `3fbdfb07642d9aac23cbddd4d98e220a590d2e8b0fcca6191128b80c8d304cf7`.
- `readout/RESULTS.json.gz` restores the exact32,385,697 result bytes; gzip
  SHA-256:`09c32316f1f6f749561503e8017cd1cd818195dc6f20bcfbbdd911b96ddfd3e0`.
- Full raw output and two typed checkpoints are privately backed up on Latrian:
  seven files/33,452,447 bytes, exact local/remote match. Receipt:
  `maintenance/date_metadata_addendum_20260908_backup.json` at the repo root.

This is main's bounded implementation and result check, not a new independent
human/source review. The previous prescribed numerical review stays closed;
it must not be described as having verified this later addendum. No more
historical research, coding or full565-cycle rerun is required for this item.

## Remaining publication work

Consume the amended publication metadata in the derived name/date/era tables;
keep original diagnostic inputs and original565 results as dated immutable
artifacts. Add this explicit overlay and amended metadata to the release with
unambiguous links. The portable addendum path is now built in
`portable/date_metadata_reproduction_20260908a.zip` (109 members/8,139,495 bytes;
all99 original numerical members unchanged). It reconstructs original fits
through the existing public numerical runner, not a private-cache fallback.
The scientific function ASTs are identical; only root/storage/dependency/output
plumbing changes. Five normal/optimized tests and isolated normal/optimized
input checks pass, with private-project reads, writes and network denied,
no research cache, both amended matrix fingerprints reproduced and altered
metadata rejected. `PORTABLE_CHECKS.json` records the bounded scope: zero new
fits, not a new full565 numerical reproduction or all-platform bit guarantee.
Select this explicit ZIP/member inventory in the final release. Finish current manuscript/case
status wording and the coordinated qualified activation/deployment. No public
GitHub switch, Zenodo publication or new authoring is included.