Round eight — lane A (“deaths in the woods, 1831-1958”) — September 12, 2026

Outcome

Zero new entries published this round. Every one of the five assigned leads traces back, in the njpb-forum.md digest, to a Newspapers.com scan paraphrased by a forum poster — never a source this session could open directly. I could not independently corroborate any of the five in a source I actually opened, so none were written up. Detail on each attempt is below.

search.json check

Grepped search.json (374 entries) for “lungin”, “hyland”, “tomlin”, “atwood”, “david thomas”, “edgar old”, “roosevelt deer club”, “oswego gun club” — no matches. None of the five leads are already covered.

Lead 1 — George Lungin, frozen to death, Dec. 21, 1831, Martha Furnace–Tuckerton road

  • Confirmed the Burlington County Library System’s New Jersey Mirror index (https://netscape.bcls.lib.nj.us/newspapers/newjerseymirror/) is a live, working, keyword-searchable database (84,133 records, full run Sept. 16, 1818 – March 5, 1947), so 1831/1832 is in scope.
  • Ran the POST search (njmirror.phtml) for: Lungin, Tuckerton (all 25-and full result sets, scanned for a Jan. 1832 hit), Martha Furnace, Martha, Furnace (date-range filter, see below), frozen, frozen to death (exact phrase), found frozen, perished, Woodland, Washington Township, wolves, died in the woods, man found dead. None returned a Dec. 1831/Jan. 1832 item about a frozen traveler. The one plausible “Tuckerton” obituary in that window (Dec. 29, 1831, “Died… Aaron Spragg, aged 49”) is a different, named decedent.
  • The site’s daterange form field appears broken server-side: submitting daterange=date with start/end month-day-year fields always returns a literal SQL-fragment string (= "1831-12-1" and date) and zero rows, regardless of the keyword paired with it. Confirmed this is a site bug, not absence of data, by testing the same broken pattern on unrelated 1903 date ranges (same failure signature).
  • Tried Chronicling America (chroniclingamerica.loc.gov / loc.gov/chroniclingamerica) by direct curl and by WebFetch: both paths return HTTP 403 (Cloudflare “Just a moment…” challenge or a flat 403) even for a single, well-spaced request — not a rate-limit situation where waiting helps, a hard block for this session’s fetchers.
  • WebSearch: session’s search budget was already exhausted (200/200) when I reached this lead, so no general web search was possible for “George Lungin” 1831/1832.
  • HathiTrust babel.hathitrust.org full-text search: 403 Forbidden via WebFetch.
  • Google Books search page: WebFetch returned only an error/help shell, no results.
  • archive.org advancedsearch.php (metadata-only, not full-text-in-scan): 0 hits for related terms; this endpoint does not do OCR full-text search across arbitrary newspaper scans, so a null result here doesn’t clear the source, it just means this tool can’t reach it.
  • Conclusion: could not source. Not written. The forum’s own note that “paper not named” makes this doubly hard — even if it is in fact an NJ Mirror item, I could not locate it in the index by any keyword I could construct from the paraphrase.

Lead 2 — Thomas Hyland, 65, Atsion carpet weaver, hanged himself April 19, 1903

  • Same NJ Mirror index, well within its 1818–1947 coverage window, and the index does contain material from April 1903 generally (confirmed via a Mount Holly keyword search restricted to April 1903 — 20+ obituary/marriage notices returned for that month, so the reel for that period is indexed).
  • Searched: Hyland (16 hits total, none in 1903 — closest are 1892, 1907, 1910, 1916, 1933+, all unrelated Hylands), hanged himself, Stone Bridge, carpet weaver, East Fruitland, Atsion (102 hits spanning 1819–1939 but with a hole at exactly 1902→1905 that skips 1903 outright), suicide (274 hits spanning 1819–1948, also skipping 1903 outright, and the one Atsion-area suicide the index does carry is an 1853 Batsto Iron Works item — a different case, correctly excluded), Batsto (37 hits, none in 1903).
  • The fact that both an “Atsion” search and a “suicide” search each show a clean gap at exactly 1903 (with hits on both sides, 1902 and 1904/1905) suggests either a true indexing gap for that specific stretch of reel, or the item ran in a paper other than the Mirror. I could not resolve which.
  • Chronicling America, Google Books, HathiTrust: same blocks as above (403s). WebSearch budget exhausted.
  • Conclusion: could not source. Not written.

Leads 3 & 4 — Atwood Tomlin (Oct. 1947, Tabernacle) and David Thomas (Dec. 13, 1947, Chatsworth–Tabernacle Rd.)

  • Checked the NJ Mirror index’s own end date first: the site’s history note states the Mirror ran “October 14, 1818 to March 5, 1947” — i.e., it stopped publishing roughly seven months before either of these October/December 1947 deaths. Confirmed empirically: keyword searches for Tomlin (164 hits), Grover, Prickett, New Lisbon (county burial ground) all return zero 1947 hits, and the full Tomlin result set’s latest year is 1940. These two leads fall entirely outside the New Jersey Mirror’s coverage — the index cannot help here regardless of search terms.
  • Tried Chronicling America (403, both direct curl and WebFetch), archive.org full-text-style search (metadata search only, 0 hits, not a real full-text tool for uncatalogued newspaper scans), findagrave.com search for “Atwood Tomlin” (reachable, 200 OK, but no clear matching memorial in the returned name list — closest are unrelated “Tomlin”/”Tomlinson” entries with no obvious Atwood), fultonhistory.com (unreachable by curl — times out/refuses non-browser client; this is a JS-driven ASP search site that this session’s tools could not drive).
  • WebSearch budget was exhausted before I could run general queries for either name.
  • Conclusion: could not source either. Not written.

Lead 5 — Edgar Old Sr., 72, Feb. 17, 1958, Oswego Gun Club cabin, Bass River Township

  • 1958 is eleven years past the NJ Mirror’s own end date — not a candidate for that index at all.
  • Chronicling America: same 403 block. Chronicling America’s core scanned run in any case thins out after the 1920s–40s for most titles, so even without the block this was a long shot.
  • WebSearch budget exhausted before I could query for “Edgar Old” / “Oswego Gun Club” / “Micks Place” generally.
  • findagrave.com: not attempted for this one given the pattern above (a name match alone would not meet the sourcing bar without an actual newspaper account of the rescue/death).
  • Conclusion: could not source. Not written.

Access-failure summary (for other lanes / future rounds)

  • chroniclingamerica.loc.gov and its loc.gov/chroniclingamerica redirect target: hard-blocked (HTTP 403, Cloudflare challenge page) for both direct curl and the WebFetch tool this session, not merely rate-limited. Waiting between requests did not appear to be the issue — the very first request of the session to this host failed the same way.
  • babel.hathitrust.org search: HTTP 403 via WebFetch.
  • fultonhistory.com: unreachable via curl (connection/timeout failure); it is a legacy JavaScript/ASP search UI that likely needs an interactive browser.
  • Google web/book search pages fetched directly by URL return only the no-JS error shell, not results.
  • archive.org/advancedsearch.php only searches item metadata (title/identifier/description fields), not the OCR text of arbitrary uploaded newspaper scans — a 0-result response from it does not mean a topic isn’t in some archive.org-hosted paper, only that no item itself is titled or catalogued with that phrase.
  • The BCLS New Jersey Mirror search form (njmirror.phtml) works for keyword search but its date-range filter (daterange=date plus month/day/year fields) is broken server-side: it returns a literal SQL fragment as page text and zero rows for every combination tried, independent of the keyword. Keyword-only queries (type=any/exact, sort=dateasc, limit=100000) work correctly and were used throughout instead.
  • This session’s WebSearch tool budget (200 calls) was already fully consumed before I reached leads 3–5, appearing to be a shared session-wide cap rather than per-lane; no general web search was available for the back half of this round.

Files touched

  • research/round-eight-woods-items.json — written as [] (no new entries this round).
  • research/locations/round-eight-woods.json — written as [] (no map pins this round).
  • This log.
  • No existing entry, search.json, index page, topics page, mysteries.html, or updates/changes.json was read for editing purposes or modified, per the hard rules for this lane.