Forum/GNIS place-mapping pass — working log
Forum/GNIS place-mapping pass — working log
Session date: 2026-09-11. Target: research/locations/forum-places.json, 40+ mappable Pine Barrens
places with real coordinates. Method actually used, in order of discovery (not the order the brief
suggested, because the second source found — the mapping site’s own data file — turned out to be far
richer than crawling forum threads one at a time):
What was opened
-
research/agents/njpb-forum.md(already in the repo, a prior crawl of forums.njpinebarrens.com). Read in full. Supplied the explicit Hampton Gate AME church cemetery coordinate (39.8169, -74.6998) and the Warren Grove/Zeldin search-point coordinate (39.7522, -74.3952), both of which the forum post itself says came from maps.njpinebarrens.com. Also supplied place names, dates and case narratives used in blurbs (Rider/Wilkinson bogs at Hampton Furnace, the Brooksbrae/Pasadena fire, Woodmansie in the Carslake search, Sim Place’s disputed position relative to Martha Furnace and Calico, etc.). -
search.json(site’s own entry index). Read the first ~1020 lines (through “L”); did not need to read the rest because the places that matter for this task (Atsion, Batsto, Calico, Carranza, Cedar Bridge, Double Trouble, Friendship, Hampton Gate, Harrisville, Mary Ann Forge, Brooksbrae, Whitesbog, Bamber, Warren Grove, Lopez/Bear Swamp Hill, Buddtown, Brotherton/Indian Mills) all sort before “L” alphabetically or were already known from the forum file. Slugs were copied verbatim from this file’surlfield, trimmed to the basename. -
https://maps.njpinebarrens.com/— page source, thenjs/app.js, thenjs/places.js, which revealed the site fetchesdata/locations.json(105 named Pines localities) anddata/poi.json(10 points of interest: hills, Pakim Pond, Carranza Memorial). Fetched both directly. This became the primary coordinate source for most entries — it is exactly the “companion site… inspect its page source and any data file it loads” the brief asked for, and it turned out to expose a clean, already-geocoded gazetteer rather than requiring per-thread scraping. -
USGS GNIS, per a correction mid-task: the
edits.nationalmap.gov/apps/gaz-domestic/public/search/namesURL in the original brief returns only the React app shell (confirmed withcurl -sv, got HTML not JSON — the endpoint is client-side, not a public REST route). The coordinator supplied a local flat-file extract instead (DomesticNames_NJ.txt, 7,553 rows, pipe-delimited). Grepped it for every maps.njpinebarrens.com name. It independently confirmed 33 of them to within ~0.001–0.015° (i.e. GNIS and the hobbyist map agree on where these vanished villages sit), so those 33 entries were switched to cite GNISfeature_idas the primary basis, with the map site noted as corroboration. Two useful corrections came out of this cross-check: Hanover Furnace and Pleasant Mills are officially in Burlington and Atlantic County respectively — not the counties I’d have guessed from geography alone — so thetownshipfield was corrected to match GNIS’s county attribution. -
OpenStreetMap Nominatim (one request per second,
User-Agent: PineBarrensFieldNotesResearch/1.0 (andrew@leahey.org)), used only for three names the mapping site’slocations.jsondid not carry: Brooksbrae (returned an actualhistoric=ruinsway — “Brooksbrae/Pasadena Terracotta Factory” — better than a plain place name, so it gotprecision: site), Munion Field, and Bulltown (both returned OSM hamlet nodes; GNIS subsequently confirmed both almost exactly, so those two also ended up citing GNIS as primary).
Rejected / not locatable this session
- Bulltown was briefly at risk of being dropped (first two Nominatim queries came back empty for “New Jersey Pine Barrens” and “Woodland Township” variants) — a plain “Bulltown New Jersey” query succeeded, then GNIS confirmed it.
- GNIS had no entry for Calico, Hermann City, Pasadena (as a populated place — only reachable via the Brooksbrae ruins way), Ong’s Hat, Quaker Bridge, Aetna Furnace, Martha Furnace, Mary Ann Forge, or Cedar Bridge (the historic tavern in Barnegat Township). These entries therefore cite only the maps.njpinebarrens.com pin (with the forum’s own corroborating narrative where available) and are marked accordingly — I did not invent a GNIS citation for them.
- GNIS does list a “Cedar Bridge” populated place, but at 40.056, -74.128 in Ocean County — roughly 30 km northeast of the historic Cedar Bridge Tavern site the map pins at 39.7836, -74.3656. These are evidently two different places sharing a name (there is also a separate “Cedar Bridge Manor” nearby in GNIS). I did not merge them; the tavern entry keeps its original map-only basis and no GNIS citation.
- Cemetery searches on Nominatim for “Tabernacle Cemetery,” “Green Bank Cemetery,” and “Chairville Cemetery” (all named in the forum notes for specific burials — Kell, Maxwell, Bakely) returned no results, so none of those cemeteries are in the output file. Only the one cemetery with an explicit forum-given coordinate (Hampton Gate AME) made it in.
- Robert L. Burns’s 1963 skeleton site, the Marlton-man 1978 stabbing site, and several other forum “new-entry seed” cases were left out entirely: the forum gives only relative directions (“about 14 miles south of Vincentown,” “a secluded section of Washington Township”), not a coordinate, and no independent source narrowed them further in the time available.
Result
49 places written to forum-places.json: 21 site-precision, 28 locality-precision, none at area
precision (no source in hand gave only a radius/zone rather than a point). Kinds: 21 ghost town, 12
other (still-inhabited villages / natural landmarks), 5 furnace, 4 ruin, 4 structure, 2 incident site,
1 cemetery.
October 6, 2026 correction: removed the duplicate Warren Grove/Zeldin forum-area marker after contemporary CBS and Inquirer reports supported a single broader case marker sourced to Warren Grove Recreation Area and Beaver Dam Road. The current JSON has 49 place records; the counts in the original result paragraph describe that earlier research pass, not the present precision mix.