# 490d C374 — Paired repair in the full Mirror span matrix

24 September 2026. Step 1 of 10 in the C374–C383 authorization.

**The two local repairs preserve the outer Mirror rectangle and its endpoint-exchange relation, while changing five of its nine cross-field spans.** This is checked on both sourced primary rows, i=3 and i=5.

Write the cumulative chain as S,u,q and the reflected regular chain in ascending order as R,m,N. The repairs are u→u−2 and m→m+2; S,q,R,N remain fixed.

| Cross-span row | To R, before → after | To m, before → after | To N, before → after |
|---|---|---|---|
| From S |17K → 17K|18K+69 → 18K+71|19K → 19K|
| From u |16K+69 → 16K+71|17K+138 → 17K+142|18K+69 → 18K+71|
| From q |15K → 15K|16K+69 → 16K+71|17K → 17K|

The old column/row labels identify the source roles; the after values use their repaired source records. The four outer cross-spans and both 2K sides are unchanged. The same-offset ±1/4 comparisons retain this table exactly.

The central span receives both two-year displacements, so it grows by four. Applying P to the remainder138 would instead give140; the actual result is142. Thus the successful local138→140 comparison is not an expansion of the complete17K+138 bridge.

A useful symmetry does survive: `S+N=q+R=u+m=(u−2)+(m+2)`. The existing source corners therefore determine a finite order-reversing correspondence that pairs both original and repaired middle points. Its diagnostic sum is −2639 at i=3 and −2635 at i=5; no common global axis is adopted. Each side changes from460+598 to458+600 when read from its outer Noah/Shem end. The segment ratios229/230 and300/299 describe that redistribution, not a new source protocol.

No single affine map can fix two distinct endpoints of one side and move its middle point: the two fixed endpoints force the identity. What survives is a source-labelled correspondence between the two repaired configurations, not a nontrivial affine motion fixing all four original corners.

## Verification and limits

23 main checks and 16 frozen-input/predecessor authentication checks passed. A second main replay produced byte-identical DATA/RESULTS. These checks and the source interpretation were performed by the root agent; no new independent-agent review is claimed. All frozen originals and standing controls remain unchanged. Verification counts are not independent-witness counts.

The diagnostic endpoint-exchange sums and229/230 segment ratio are not new adopted axes or chronology operators. The local P/J pairing remains conditional.

Mirror/529 focus: K=529, E=25/23, P=70/69. Global paired Mirror January AD1; local junction January AD2; no civil year zero. Regular Gears are Noah/Shem/Flood only. SP minus-one person companions remain local to SP Noah/Shem; Flood close is a separate boundary. No upstream/cumulative Gear, new endpoint or phase, fitted anchor, global axis, primer reopening, canonical or graph edit. Capture now; canonize later. Source identity, context, modal binding, generated coordinates and phase ancestry stay distinct.

## Decision after this result

**C375, proposed and unexecuted:** Test the exact conjugacy of the cumulative and reflected regular P/J comparisons under their existing source-corner endpoint exchange. Include the four complete C287 rows, source-role membership and inherited phase ancestry.

The paired endpoint sum survives the repair. Checking whether it also pairs the actual two Flood pivots will distinguish an exact cross-field correspondence from a superficial similarity of partitions.