# C319 — Paired-selection sensitivity: source/method review

24 September 2026. Separate review of C319 only; no later calculation.

## What is permitted before testing

C318 authenticated two different choices: an annual member selection in the cumulative LXX_ON Shem envelope, and an ordinary regular MT Flood Gear selection after declared restoration and Terah context changes. Neither is obligatory in the reviewed sources. Holding the original member and Gear is a legitimate comparison state. File_18's modal-processing rules, regular variant fork, and Flood tables admit these base states; File_09 keeps members of the seven-year object separately labelled and prohibits treating regular Gears as cumulative operators.

Accordingly both the00 and22 cells of the proposed a,b∈{0,2} sensitivity grid must be allowed wherever their endpoint domains permit them. The mixed02 and20 cells also deserve explicit source/domain checks, even if they fail a desired529 relation. They are controls for the coupling claim, not discarded chronologies.

## Interpretation of a conditional winner

The formulas S=p−460+a and R=f−460+b use cumulative signed coordinates for S and regular positive-BC/reflected coordinates for R. Keep that convention visible. In all cells the regular endpoint retains Flood start/close with rail=null, CainanON/Terah60/full430, and a named Gear; the cumulative endpoint retains LXX_ON Shem plus a named annual index. No MT companion operation occurs.

A finding that only paired22 gives the full2K/15K/17K/19K configuration would establish **uniqueness conditional on this specified four-case comparison and its held contexts**. It would not show that the sources prescribe22, that no other source configuration elsewhere works, or that the all-K criterion is independently justified by historical evidence. The successful scalar equations can identify the selection necessary to realize this already chosen pattern; they cannot retrospectively convert the selection into an independent prediction.

If00 also preserves equal17K translations, it remains a valid source-qualified rectangle. Thus failure of its shorter edges to be integralK is not a source failure. Conversely, formal all-K equalities at indices beyond the admitted source field remain failed state admissions. Source legality and scalar-pattern membership must be recorded separately.

The four-case equations share p,q,f,n and fixed600,460,8993. Their differences and repeated phase results are consequences of those inputs, not a multiplication of independent evidence. A direct phase check is still necessary to prevent a center-only comparison from silently changing ancestry or seasonal ordering.

## Useful reporting distinctions

Report, for each cell: source-admitted rows; equal17K lower translation; both same-side short spans; full all-K status; and failed cumulative versus regular boundaries. If restricting to five i3…7 cases, state that restriction explicitly. C287 originally contains an additional supported i2 row; a five-case result must not be worded as a new exhaustive count of all retained C287 rows.

The exact25/23 seasonal-width failure is already C290's result. It should not be offered again as a new discovery from this sensitivity check.

## Best subsequent bounded question

After C319, compare the five-case restriction with the **six already admitted C271 SP Noah/Flood records**, examining the omitted G1-companion boundary case only. Ask which of the source600 rectangle, C287 all-K rectangle, and reversed500 cross-bridge relations it can satisfy within their original domains. Do not extend the cumulative window or create the missing annual selector. This would clarify whether the present five-case restriction reflects source absence or the earlier comparison's boundary, and prepare an accurate final synthesis. No such boundary test is executed here.