Frontier round 4 — part E (results 1–7 of 13)

Integration of workflow wf_70287d01-764, journal result lines 8–14. Confidence tags are the agents'; single-Find-A-Grave "CONFIRMED" claims downgraded to PROBABLE per house rule (none required it in this slice — these agents cited registers, certs, and indexes).

Eva Brown (GA Virtual Vault coverage check, Towns/Union Brown & Foster)

Verdict: Eva's 1938 death cert is provably NOT in the Virtual Vault — but the sweep landed primary-source corroboration of our Foster line the agent itself couldn't see.

Key facts (all from vault.georgiaarchives.org/digital/collection/gadeaths, index transcriptions, CONFIRMED unless noted):

• Coverage verbatim: "Georgia Death Certificates from 1919 through 1927. The collection also includes a number of certificates from 1914-1918, with the bulk dating from 1917 and 1918." Eva (d. 1938) is 11 years past the ceiling; a direct "Eva Brown" search returns 5 unrelated hits. The tree's open-key plan "Eva Brown's parents via 1938 GA death cert @ Virtual Vault" is dead as routed.
• Foster, Gather L. (cert 18757, Towns): d. 1925-06-03, b. 1856-10-12, father "Frank Foster", spouse "Adaline Foster" = our Lumpkin Gaither Foster (1856–1925) — exact day-level dates new; father/spouse fields corroborate Franklin Calloway Foster and Mary Adeline (Hooper).
• Foster, Osborn Calliway (cert 7931, Towns): b. 1862-05-18, d. 1926-04-01, father "Frankie Foster", mother "Malindy Osbon" = partD's sibling Osborne Calloway Foster (1862–1926). The mother field is the first primary-source support for Harriett Malinda Osborn — upgrades her maiden name from POSSIBLE (FaG-only) to PROBABLE.
• Erastus Hoyt Foster cert 23057 re-transcribed: b. 1882-11-26, d. 1927-03-08, White Co (not Towns).
• Complete Towns Foster deaths = 4; Towns Brown = 5; Union Brown = 9 (all transcribed in journal). India Brown (b. 1885 Towns, d. 1920, father "Alexander") = possible Eva sibling, POSSIBLE.

Parents found: none (Eva's remain open).

Top leads: (1) reroute Eva's death record — later GA death index/obituary/cemetery, not this collection; (2) compare "Alexander" vs Eva's father when found; (3) pull Osborn Calliway's cert image to seal the Osborn maiden name.

Dead ends: No "Julius Foster"/"Lumpkin Foster" person in the collection (namesake hits quarantined, unrelated). Eva Brown definitively outside coverage.

Catharine (Sayles) Olney (Arnold's RI Vital Record, Vol. 2 Providence)

Verdict: Marriage date CORRECTED to Dec 20 1692; Roger Williams filiation NOT upgraded — Arnold is silent on Catharine's parents, so the RW chain stays anchored-per-Austin (PROBABLE).

Key facts (Arnold, Vital Record of RI, 1st Series Vol. 2, archive.org cu31924098822103; CONFIRMED unless noted):

• Providence Marriages p.137 (OLNEY), citing Town Records Book 1 p.423, verbatim: "William, and Catherine Sayles, Dec. 20, 1692." Cross-index p.163 (SAYLES), same original page, verbatim: "SAYLES Catherine, and William Olney, Dec. 20, 1692." Double cross-indexed — the "Dec 28" in our docs (from the 1889 Olney genealogy) is wrong. See corrections list below.
• Six children of William & Catherine, Book 1 p.423: William (b. Oct. 6, 1694 — day-level pin for our William Olney⁴ 1694–1771), John (1699), Katherine (1701), Thomas (170[6/8], OCR), Deborah (1708), Richard (1711). PROBABLE (two OCR-uncertain points).
• Providence Births p.243, Book 1 p.81, verbatim: "SAYLES Mary, of John and Mary, July 11, 1652." + "John, Aug. 17, 1654." — the ONLY recorded births for John & Mary Sayles. No Catharine. All 15 'sayles' hits in the 83,593-line volume reviewed. Negative/silent, not contradiction (Arnold's intro: pre-1790 records badly incomplete).
• Roger Williams family register independently confirmed, Book 1 p.4 (p.257), verbatim: "Mary, of Roger and Mary, born Plymouth, first week in August, 1633." (+ Freeborne 1635, Providence 1638, Mercy 1640, Daniel 1641, Joseph 1643). Resolves the Austin "Aug 1638" tension in favor of Aug 1633 — what the site already carries.
• Book 2 p.82 "Roger, of Daniel, 1650" / "Rebecca, of Roger, born Scituate, April 20, 1735" = later-generation same-name Roger, quarantined.

Parents found: John Sayles + Mary (Williams) Sayles remain POSSIBLE — named for Mary/John's births but never for a Catharine.

Top leads: (1) John Sayles's will/probate naming "daughter Catharine Olney" = the new RW closer; (2) Austin's exact Sayles wording re-inspection; (3) cleaner scan for the two OCR ambiguities.

Dead ends: No death record for Catherine (Sayles) Olney anywhere in Vol. 2; no Catharine birth under Sayles — this volume is exhausted.

Josephine Oakford (Lambert m. 1874; Oakford parent hunt)

Verdict: Marriage cert-numbered to the day (Manhattan #923, Jan 22 1874, CONFIRMED both-sides); hypothesized parents now carry death-cert numbers (PROBABLE); Jane E. Oakford's 1880 death is a hard NYC negative.

Key facts (German Genealogy Group free NYC indexes + archive.org directories):

• Marriage Cert #923, Jan 22 1874, Manhattan — bride-index "Oakford, Josephine" and groom-index "Lambert, John J" agree on cert #, date, county. CONFIRMED; trivial Municipal Archives order.
• Almira Oakford d. Jun 5 1870, Kings, age 38, Cert #3603 — PROBABLE match to Green-Wood monument "Almira Oakford 1835–1870" (age implies b.~1832, minor discrepancy).
• John Oakford d. Jun 19 1898, Manhattan, age 72, Cert #16353 — PROBABLE match to "John M. Oakford 1826–1898" (b.~1826 exact). Note: their parenthood of Josephine remains the Green-Wood-plot hypothesis; the certs are the resolution, not the proof.
• Complete NYC Oakford death index 1862–1948 = exactly 8 records; no Jane — Jane E. Oakford (1855–1880) has NO NYC death record under exact, soundex, or wildcard search. She likely died outside the five boroughs (NJ fits the Anita/New Brunswick thread) or under a variant year/name.
• Directory contamination flag: three distinct 1865–68 "John Oakford"s — "John D., broker" (Oakford & Brewster, Broad St; Brooklyn Heights; best social fit but wrong initial), "John M., clerk, h 104 Perry" (right initial, weaker fit), an officer/captain (unrelated). NOT resolved; cert #16353 settles it.
• Candidate siblings: Kate B. Oakford (d. 1901, b.~1854); Elvira Oakford (d. Dec 7 1880 — Jane's death YEAR, wrong age/name). POSSIBLE only.

Parents found: John [D. or M.?] Oakford + Almira — PROBABLE, cert-pull gated.

Top leads: (1) order certs #923, #16353, #3603; (2) 1870/1880 census, Brooklyn Wards 1–2 Oakford household; (3) John Oakford d.1898 probate at NY County Surrogate's Court.

Dead ends: Jane E. in NYC index (thorough multi-method negative); 1853 Green-Wood proprietor catalogue (no Oakfords); 1901 Green-Wood index (image-only); Manhattan directories 1869–85 not full-text on archive.org; italiangen.org JS-walled.

Roger Williams (his parents — the Pemberton line)

Verdict: Parents documented to the will: James Williams, merchant tailor of London, + Alice (almost certainly Pemberton) — the 1889 Waters wills, quoted verbatim. New upstream generation IF our RW link holds (still PROBABLE via Catharine Sayles).

Key facts (Waters, NEHGR 43 (1889) 290–298, verified in archive.org full text; CONFIRMED unless noted):

• James Williams, citizen and merchant tailor, Cow Lane, St Sepulchre-without-Newgate, will 7 Sep 1620, proved 19 Nov 1621: "my will and meaning is that my sons Roger and Robert Williams shall have but one moiety or half part of the other third part of my estate... twenty and five pounds apiece, to be paid unto each of them severally at his age of four and twenty years." Overseers include "my brother in law Roger Pemberton."
• Alice Williams, will 1 Aug 1634, proved 26 Jan 1634/5: "Item I give to my son Roger Williams now beyond the seas ten pounds yearly... shall be paid to his wife and to his daughter, if they survive." — matches RW's wife Mary + daughter Mary b. Aug 1633 exactly.
• Roger Pemberton's own will (proved 1627): "unto my cosen and Godsonne Roger Williams the some of ten pounds of lawfull english money" — the Pemberton maiden name inference (PROBABLE).
• Roger Pemberton was of St Alban's, Hertfordshire (Visitation of Herts 1634; father Robert Pemberton of St Alban's) — the SAME parish as our Ashtons and Thomas Olney's marriage. Candidate Alice baptism: "Alice Pemberton, daughter of Robert, 18 February 1564," St Albans register (POSSIBLE).
• Earliest mention of RW by name: Margery Pate will, 22 Sep 1617, legacies to James, Alice, Roger, Robert, Katherine Williams.
• RW's birth has NO documentary pin (St Sepulchre registers burned 1666); scholarly range 1598–1607, Hodges 1899 best estimate ~1604; "c.1603" is convention.

Parents found: James Williams (CONFIRMED) + Alice (Pemberton) (PROBABLE).

Top leads: (1) NEHGR 43 Oct-1889 p.427 continuation (Robert Pemberton will); (2) Robert Pemberton of St Albans records — one more generation; (3) Chester's London Marriage Licenses for James⚭Alice.

Dead ends: Guild's Cornwall/Gwinear theory and Elton's "Rodericus"/Oxford theory — both explicitly dead since 1889; no James–Alice marriage record; no exact RW birth date recoverable, ever.

Rev. Gregory Dexter (birthplace claim + real dates)

Verdict: Estimates retired — b. c.1610 (traditional), d. 1700 (CONFIRMED) — which resolves our flagged 1643-imprint tension exactly as predicted. The "Olney, Northamptonshire" birthplace is unsourced 1859 tradition with the county wrong.

Key facts:

• Death 1700: Swan (1949) found Dexter alive per a 27 Mar 1699 quitclaim deed, concluding "probably he died, as the genealogies state, the ensuing year" — agrees with Newman 1859 + Austin 1887. CONFIRMED (year only; Wikipedia's "April 23, 1700" unverified).
• Birth tradition traced to origin: Newman 1859, verbatim: "Rev. Gregory Dexter was born at Olney, Northampton County, England, A.D. 1610" — NO source; Austin copies it. Swan: "Newman gives not the slightest hint...as to where he got his information"; all pre-Newman writers said London; Swan's hedge places Olney correctly in Buckinghamshire. Record as POSSIBLE/traditional-only. The Olney-town/Olney-family echo is a place-name coincidence, not a link.
• Wikipedia's "Old, Northamptonshire... stated in his Bible" — uncorroborated; the only "Bible" in Swan is his print-shop sign "at the Bible in Pope's Head Alley." Do not import.
• Wife Abigail Fullerton: maiden name via Morgan Edwards tradition; marriage date/place "remains a mystery" (Swan). Living as late as 1706 (Swan); Austin: "d. 1706" — retires her ed ~1696 estimate.
• Children (Providence records per Swan): Stephen 1647, James 1650, John 1652, Abigail 1655, (+Peleg 1658, Newman only). Abigail Dexter b. 1655 Sep. 24 (Austin, day; year in 3 sources) — retires her eb ~1657 estimate.
• Abigail m. James Angell 3 Sep 1678 (Austin p.597; garbled p.27 "1673" = OCR artifact; PROBABLE). James Angell d. 3 Mar 1711, administration to widow Abigail 31 Mar 1711, inventory £381 12s 8d (PROBABLE) — retires his ed ~1715 estimate.

Parents found: none (Dexter's own origin = Swan's "Among the Mists").

Top leads: (1) Olney, Bucks parish registers c.1608–12; (2) Providence probate for Dexter's will (exact death day); (3) Angell genealogy for James Angell's birth.

Dead ends: No primary source for Dexter's birth exists per Swan; no Dexter-Fullerton marriage record; James Angell birth absent from Austin.

James Ashton (St Albans Abbey registers, full sweep)

Verdict: Full sibling set register-confirmed with burials that likely KILL the Alice-Ashton⚭Thomas-Angell claim; two new Olney children surface, including a previously-undocumented son Epinetus.

Key facts (Brigg 1897 transcription, archive.org parishregisterso00stal; CONFIRMED unless noted):

• Marriage register p.160, verbatim (OCR): "Thoma.s Olney aud Mary Aslitou" under 1629, day-markers including "Sep. 16" — consistent with our 1629-09-16.
• Baptisms of James Ashton's children: James 1604 ("James sonne of James Ashton"), Marie 1605 ("Marie dough : of James Ashton"), John 1607/8, Martha 1608/9, Alice 1610, Elizabeth 1612, Daniel 1613–15, Sara — precise: ",, 23 Sara dougliter of James Ashton by Alice his wyfe" = 23 Feb 1619/20, possible 9th Thomas c.1630 (POSSIBLE, big gap).
• Burials (PROBABLE): "Alice doughter of James / Ashton." bound to 1613/14 — the sister Alice likely died a CHILD, 3–4 yrs after baptism → second independent strike against her marrying Thomas Angell (first: no Angell⚭Ashton anywhere in the register). James (son) + Martha both buried 1624. "[blank] the wife of James / Ashtou." 1643 = matriarch Alice (est. was ~1644). "James Ashton sen." buried 1651 (est. was ~1640).
• Olney children found: "Thomas sonne of Thomas Olney by his wj'f. [siV]" bapt. 1630 (mother blank in original — likely our Thomas Olney², book says b. 1632; flag, don't merge silently) and ",, 14 Epinetus y'^ sonne of Tho: Olney et p Marie vx." = 14 Feb 1633/4 — new child, biblical name (Rom. 16:5).
• "Marten Element et Lidia Olney" c.1631–33 — probable sister of Thomas Olney¹ (PROBABLE), new trail to his origin parish.
• No Ashton OR Olney presence in the parish before 1602–04/1629 respectively — both arrived as adults.

Parents found: none (James Ashton's own baptism/marriage not in this register).

Top leads: (1) trace Epinetus Olney forward; (2) Lidia Olney⚭Element/Clement → Olney origin parish; (3) neighboring Herts/Beds registers for Ashton/Olney 1570s–1600s.

Dead ends: James Ashton's baptism/marriage not here; Thomas Olney's baptism not here (wall holds); Alice⚭Angell unsupported on two grounds; "Ownie" index variant needs page images.

Richard Wilson (b. 1767 NC — redirected four-county sweep)

Verdict: Comprehensive, well-executed NEGATIVE — Wilkes/Stokes/Ashe/Rockingham yield zero Richard Wilson in wills, tax lists, or heritage volumes; the round's real prize is a reusable NC State Archives JSON API.

Key facts (all CONFIRMED negatives unless noted):

• Wilkes taxables 1787–90: only John + Samuel Wilson. Stokes Heritage Vol II (~115k lines): no Richard Wilson. Ashe Baptist histories: none. All 16 vols of the Journal of Rockingham County History: no Richard among 40+ enumerated Wilson given names; 1815 multi-district Rockingham tax roll: 8 Wilson taxpayers, no Richard.
• NC State Archives AXAEM catalog (live JSON API discovered: appx.archives.ncdcr.gov/solr/axaem/Series) — 2,060 indexed will testators across the four counties: zero Richard Wilson. Only Wilsons: Aquilla (Rockingham, will c.1802, OW/172), Milley (Stokes 1837), William D. (Stokes 1872), Jesse Willson.
• A "Wilson, Richard" DOES exist in Revolutionary War Army Accounts (Comptroller's Papers, Vol. A p.113 etc.) — no county/date attached; b.1767 makes him a teen in the war. POSSIBLE, unmerged.
• Land entries/grants for all four counties exist as bulk volumes but are NOT name-indexed remotely — genuine web-research dead end.

Parents found: none.

Top leads: (1) pull Aquilla Wilson's c.1802 Rockingham will (right county, right era — check for a Richard); (2) resolve the Rev-War pay account's county via FamilySearch microfilm; (3) NC land-grant name index via a WebSearch-enabled session.

Dead ends: Heritage of Wilkes (both eds.), Heritage of Surry — access-restricted on archive.org; no published abstract volumes for the four counties on archive.org; NC land-grant web front end unlocatable by URL-guessing; AXAEM estate-name indexing incomplete (weak negative only).

Corrections queue from this slice (Dec 28 → Dec 20, 1692)

Docs carrying the wrong Olney–Sayles marriage date, per Arnold's double cross-indexed Dec 20:

1. 2026-07-20-wilson-tree-v0.md line 116 ("1692-12-28")
2. 2026-07-21-lineage-research-partA.md line 379 ("28 Dec 1692" — origin: 1889 Olney genealogy)
3. 2026-07-21-sibling-round-partC.md line 133 ("28 Dec 1692")
4. 2026-07-21-sibling-round-partD.md line 164 ("28 vs 18 Dec 1692 — unresolved" — Arnold resolves it: both wrong, it's the 20th)
5. roots-timeline/index.html (1 occurrence, "Dec 28, 1692")
6. roots-timeline/people-bios.json (2 occurrences → regenerate people pages, at least william-olney-1663 + catharine-sayles)

Also stale: tree v0 line 124 "Remaining closer: Arnold's RI Vital Record" — Arnold now checked: confirms marriage (corrected date) but is SILENT on Catharine's filiation. RW descent stays PROBABLE; new closer = John Sayles probate. And tree open-key "Eva Brown 1938 cert via Virtual Vault" is impossible (coverage ends 1927) — reroute.