Independent review of C932–C951, with C952 correction

Disposition: the arithmetic and bounded source interpretation pass. The material continuation-premise omission in C949–C950 is resolved by the appended C952 correction. Read C949–C950 together with C952; their unchanged historical records preserve the issue and its resolution.

Reviewed all twenty C932–C951 records, their scripts and generated models, the research/replay helpers, File 52c §§3.3–3.4, and the C952 corrective record. File 52c is the frozen latest draft, SHA-256 a5ea84562101158b60d0cf296765d6eff38e7a2abda4e74ad1b353dfd13b9530. No journal, numbered script, or canonical source was edited by this review.

• Source roles and correction. The regular 1650 + 1050 and primary cumulative 9170 + 3430 partitions independently give inverse duration 10620 and, with the Conquest anchor, 12026 BC. The continuation requires the additional original tail 1400 = 1406 − 6. C950 had substituted the regular total 2700 for the continuation gain. C952 now derives that gain from I(1400) − 1400 = 4100 − 1400 = 2700, retaining Nativity 6 and the source transition rule. Thus 14726 is dependent on the first endpoint and the retained tail premise; equality of the gain and the regular total is not an identity of their roles. This matches File 52c §3.4 and leaves all endpoint numbers unchanged.
• Independent arithmetic. Exhaustive integer checking in the stated aggregate box confirms the unique regular aggregate (2,6,10), the two cumulative branches (11,15,10) and (12,5,10), and their inverse sums 11610 and 10620. The kernel (10,−101,10), the finite-box uniqueness argument, all 63 ordered regular allocations, the diagnostic counterexample, the selected-shift weight 4/5, and the corrected tail calculation check exactly.
• Interpretive scope. The fixed decimal register, source-appointed partitions and cumulative carry branch remain premises. Aggregate identifiability does not recover individual allocations, an ordered historical breakpoint, or historical causation. Diagnostic allocations are arithmetic controls, not admitted chronology families or independent statistical hits. C932's four categories are descriptive; C949 appropriately retracts any implication that their independence was proved. No second decimal inversion was executed in the reviewed actions.
• Wording to carry forward. C949's “two three-digit cores, each with one trailing zero” should be expressed as “two durations equal to ten times three-digit cores ending in a nonzero digit.” Earlier register calculations use the correct meaning; this wording does not change their arithmetic. Preserve the correction in subsequent synthesis rather than rewriting history.
• Helper and evidence checks. The writer enforces sequence and links the first record to the pinned C931 predecessor. The independent verifier substitutes its own begin/finish/artifact operations, checks every recorded field and hash, blocks replay writes, uses a separate decimal-reversal implementation, and guards against nested or provenance-tracked second inversion. The guard is correctly described as protection for this trusted packet, not a proof against arbitrary deliberately evasive code. The existing C951 replay reports 20 actions, 49 checks, 44 frozen snapshots and 44 available current copies; all 20 reviewed script hashes still match that report.

The checkpoint is frozen through C952: 21 completed actions, with last record SHA-256 5f366b8672367fb2c64d67e6234ef0b1230fdfbee17424f6de6834fd546c5eae. C951's earlier 20-action checkpoint remains historical. Later current-cycle actions are outside this review and will be excluded from the numbered checkpoint record.