# C149 — The exact rows pair Noah with the next Flood Gear

23 September 2026 · Bounded step 14 of 20 (C136–C155).

**The four exact rows have a specific selector structure: the MT Flood Gear is one above the SP Noah Gear, and the companion/close bits numerically match.** This is stronger than a bare equality of interval lengths, while preserving the distinction between those two kinds of event state.

Let gN be the SP Noah Gear, cN its companion bit; let gF be the MT Flood Gear, cF its close bit. Source tables give

\[
X=3704+2g_N-c_N,\qquad M=2644+2g_F-c_F,
\]
\[
\delta=2(g_N-g_F+1)+c_F-c_N.
\]

Because cN and cF are 0 or 1, δ=0 requires **gF=gN+1 and cF=cN**.

| Exact row | SP Noah | Alternate MT Flood |
|---|---|---|
| N01 | G1 companion | G2 close |
| N02 | G1 primary | G2 start |
| N03 | G2 companion | G3 close |
| N04 | G2 primary | G3 start |

At P6, Noah is already G3 companion. An exact selector continuation would exceed the allowed three-Gear Flood domain; it is not licensed. The nearest admitted Flood is G3 start, giving δ=1. No fourth Gear is introduced.

Equal bits are a correspondence between two distinct source selectors. Flood closes remain event-year boundaries; they are not Noah/Shem companions. Creation and Conquest receive no Gear. This typed result explains the finite upper-boundary failure without recasting it as a calendar error.

## Source and scope control

All numerical bases in this packet stand for paired coordinates: BC base B is (B+1)t/Bn BC; reflected AD base A is AD At/(A+1)n. Thus a cross-Mirror same-season span is B+A, after the civil no-year-zero subtraction. AD stations are reflected source events, not historical AD occurrences. K=529 throughout.

Regular Gears remain restricted to Noah/Shem/Flood. Creation and Conquest are fixed source-field selections. C133–C135 corrections remain controlling. The C127 graph and canonical controls are frozen; source-field admission is distinguished from an existing graph node and from a generated comparison coordinate. No wider N01/N03 expansion or missing N03 Shem join is repaired. Original Reports 01–10 remain previously verified references; interrupted original Reports 11–25 remain unestablished. Check counts are computational assertions, not independent historical witnesses.

Evidence: `CERTIFICATE.json`; **78 passing assertions** in `RESULTS.json`. The assertion list identifies the actual tests.