Encevra Research Round 49: Remember the Answer, or Remember the Procedure?

Round 49 (WMR-49) of Encevra SI Research Labs tests working memory and procedure reuse for an independent reviewer in two domains: 20 authored contract items (joint venture funding, pipeline right of way, noteholder consent, lender estoppel, offtaker acknowledgment) and 16 authored quality release record items (aircraft inspection acceptances, food lot pathogen results, instrument calibration certificates, bridge load test reports). Reviewers receive no memory, a correct stored note, a stale stored note from a different state of the same file, the stale note after a deterministic revalidation guard, or a reusable procedure card. Structurally, a quote check alone leaves 9 of 36 stale conclusions intact; adding a coverage check withholds all 36 while leaving all 36 correct notes intact. Live, Witness D never adopted a stale stored conclusion and was correct on 26 of 36 items with no memory and 27 of 36 with any stored note; the near miss FAIL (one record fails, nothing proves or excludes another qualifying record) survived every stored answer. Only the procedure card moved it, raising accuracy to 30 of 36 and cutting unsafe verdicts from 9 to 4, in both domains. A stored extract first commitment lock blocked the one wrong revision caused by a right brief. On 24 new SEC EDGAR exhibits, reviewer notes in three styles were followed zero times. The archive contains the preregistered plan, deviation record, packets, memory and guard builders, structural diagnostic, verbatim witness transcripts, exhibit track, analysis, synthesis review, and exact verification scripts. These are analytic materials, not legal or engineering advice. Web version: https://research.encevra.com/round-49