Working notes from the research, kept verbatim — what was read, what it proved, what stays open. Many open with a line grading the parentage; what the grades mean.

Research note · 1 Aug 2026

Audited 1 Aug 2026 — AND DELIBERATELY WITHOUT A CITATION. ⛔ SHE IS ABSENT FROM HER HUSBAND'S HOUSEHOLD. The 1850 census of District 2, Clark Co., Kentucky, dwelling 97, holds [1086] Ambrose Bush 72, 'Julia Bush' 23 and Robert Berry, AND NOBODY ELSE — confirmed on the image, so the absence is real and not an indexing artifact. The tree has her b.1782 d.1859; she should be present, aged 68. THREE LIVE POSSIBILITIES, NONE PREFERRED: she was enumerated elsewhere; her recorded 1859 death is wrong; or the household is misidentified. ⛔ NO DEATH IS INFERRED FROM AN ABSENCE. She remains UNSOURCED and that is the honest state. ⭐⭐ THE GENERATION'S FINDING, from the four households read together: A MIGRATION COHORT IN SILVER CREEK, RANDOLPH CO., MISSOURI. Two unrelated families arrive in the same township within a few years of each other in the 1830s, from two different states — MANSFIELD from Virginia (flip at Elizabeth, 17) and CAVENS from Kentucky (flip at Margaret E, 12). That is REGION x ERA x ANCHOR as `stopping-rules.md` sec.3 asks for it: Randolph Co., Missouri, the 1830s, the push west, with four named ancestors standing in it. THIS IS CARPENTER'S ERA HOOK AND IT DID NOT EXIST BEFORE 1 Aug 2026. ⛔ THE LIMIT, STATED: in both households the entries above the flip are DITTO MARKS whose referent is inferred from the parents' state. Each flip is legible; the extent above it is not. Re-read at higher zoom before either bracket is written as a fact. Worksheet: ancestry/carpenter-gen7-census-worksheet-2026-08-01.md

Elsewhere

This person on Ancestry (sign-in required)