Round nine: fire in the Pine Barrens, 1818-1947 — research log (2026-09-13)

Lane 4 (“fire in the pines”). Working alone; no other entries, search.json, index pages, topics, mysteries.html, updates/changes.json, or the catalog script touched. No forks or WebSearch used.

Checked search.json (386 entries) first

Grepped for fire/arson/warden/burn-related titles and summaries. Confirmed existing coverage of: the 1936 Bass River/Chatsworth-to-Tuckerton fire (five deaths, conflicting descriptions — bass-river-fire-1936.html), Black Saturday 1963 (black-saturday-forest-fires-1963.html, which names “Coyle Field” incidentally), George Lemmon’s 1920 self-set fatal house fire near Retreat, the Brooksbrae fire (1917), the second Whiting station fire (1897), Medford Friends school/firehouse history, the BOMARC missile fire (1960), Silas Little’s prescribed-fire research, and several wildfire-science entries. Also searched for the four specific prior leads (chiselers, arends/arens, coyle, ridgeway, vannote, maple island, waterford, bowker) plus stokingham/centreton/1922/lightning/pointville/rahilly — no true duplicates found for anything published this round.

The four prior leads (Chiselers Club/Arends, Maple Island, Bowker, Roy Carson)

Read the prior round’s log (research/2026-09-12-round-eight-fire.md) first: that lane already ran all four of these same leads against the same Mirror index (and against panewsarchive, Chronicling America, archive.org) with keyword variants and limit=1000, and got no matching hits — Ocean/Camden county coverage gaps, as the assignment itself anticipated. Rather than repeat identical failed queries against the identical tool, I spent the session’s budget hunting fresh, datable fire events directly through the Mirror index’s productive search terms, which the assignment specifically invited (“fire,” “warden,” “sparks,” etc.). This paid off — see below. I did not re-run the four specific prior leads a third time; they remain unsourced from any tool available this session, for the same coverage-gap reason the prior lane documented.

Method and tool notes

netscape.bcls.lib.nj.us (Burlington County Library System, New Jersey Mirror index): confirmed working exactly as briefed — fields[] params required, limit can exceed 200, type=any is a simple LIKE match. Requests sometimes read as a curl timeout at 30-60s even though the server had already returned HTTP 200 with full content (confirmed by checking output size/status separately) — worth retrying with a longer --max-time or just reading the (complete) partial output rather than assuming failure.
Searched broad terms across the whole 1818-1947 run: “forest fire,” “fire warden,” “incendiary,” “sparks,” “damages,” “woods fire,” “cranberry bog fire,” “mill destroyed,” “schoolhouse fire,” “tavern fire,” “set fire to the woods,” “railroad set fire,” “railroad company for damages,” “suit against the railroad,” “bog fire,” “cedar swamp fire,” “Toms River and Forked River,” “Coyle,” “Stokingham.” Most surname/generic terms (“sparks,” “damages,” “tavern fire,” “schoolhouse fire”) returned mostly unrelated marriage/obituary/land-sale notices or already-covered material; several “railroad…damages” and “suit against the railroad” hits were grade-crossing collision suits, not fire-and-sparks damage suits — none of the located Mirror records this session describe a locomotive setting woods or a bog alight and a resulting damage suit, despite deliberately searching for exactly that. That specific sub-genre (mentioned in the beat description) was not found this round; it may exist in the Mirror’s run but did not surface under the terms tried.
No WebSearch, no forks, no other agents used, per instructions.

Published entries

1. Colonel Leonidas Coyle, State Forest Fire Warden (1937) — PUBLISHED

entry/leonidas-coyle-state-fire-warden-1937.html. Found via keyword=Coyle (limit 1000, 13 rows). Three index records used: his November 3, 1937 obituary (career since 1911, division chief from July 1923, death by apoplexy in his Trenton apartment); the May 27, 1936 fire-coverage index record, which names him repeatedly and describes his reconnaissance flights over the fatal Chatsworth-to-Tuckerton fire; and a July 6, 1938 item on a planned memorial and new landing field on Route 40 “about 12 miles below Four Mile Colony.” Flagged, rather than silently resolved, a discrepancy between the memorial item’s “over 25 years” heading the bureau and the obituary’s “past 14 years” as state fire warden — most likely the former counts his entire fire-service career from 1910-11. Did not claim to know the memorial’s current location or whether it survives; the map pin uses “Four Mile” near Browns Mills (the only “Four Mile” toponym in the NJ gazetteer) as an honestly-caveated reference point, with a wide “area” radius, explicitly not a verified site.

2. The 1893 forest fire between Toms River and Barnegat — PUBLISHED

entry/toms-river-barnegat-forest-fire-1893.html. Found via the location-name search “Toms River and Forked River” (the same search that also surfaced the unrelated 1936 fire and an 1881 elopement story). A single May 17, 1893 local-briefs index record describes fires burning since the prior Tuesday, destroying the Quail Run tollgate/tollhouse (keeper Mrs. Abdel Imlay) on the Toms River and Forked River turnpike, then a house owned by Henry Bird, then a house owned by Mrs. Abbie Chamberlain and occupied by her son-in-law Nathan Moore’s family, who had to flee while Moore himself was fighting the fire elsewhere. No acreage, cost, or casualty figures given; no independent identification found for any of the four named people or for the tollgate’s exact site. This is a thin, single-source entry (much like the format model’s Dover wildman entry) and is presented as such — no filler claims added.

3. “Stokingham”: the Stokes mansion fire near Centerton (1905) — PUBLISHED

entry/stokingham-stokes-mansion-fire-1905.html. Found via keyword=sparks (a surname-noise search that happened to catch this headline) and confirmed via keyword=Stokingham (2 records). The December 27, 1905 index record describes the historic Stokes family homestead — birthplace of Governor Edward C. Stokes’s boyhood per the index’s reproduced Stokes-genealogy excerpt — destroyed by fire from overlooked lard-rendering sparks after a hog-killing, while occupied by Tylee B. Engle’s family. A second record, an April 17, 1907 wedding report, places the same family “at ‘Stokingham,’ the attractive home of Mr. and Mrs. Tylee B. Engle” roughly sixteen months later — the entry states this discrepancy plainly as unresolved rather than assuming rebuilding, partial damage, or any other explanation not in the sources. Location (Centerton, Mount Laurel Township, on the Rancocas) resolved via Nominatim after the NJ gazetteer file had no “Centreton”/”Centerton” entry matching a Rancocas-side hamlet in Burlington County by that exact spelling variant used in the index.

4. The July 1922 lightning-fire storm around Mount Holly — PUBLISHED

entry/july-1922-lightning-fires-mount-holly.html. Found via keyword=railroad set fire (an otherwise unproductive search whose last hit was this storm report, indexed under “Local News,” July 19, 1922). Describes a lightning strike that killed fourteen-year-old Myrtle Stewart, plus lightning-caused barn and tenant-house fires on three named farms (Asher B. Parker farm near Pointville, $3,000 tenant house + $10,000 barn; Richard Rahilly farm near Wrightstown, $10,000; John Nash farm, $5,000 grain barn), and non-fire lightning strikes at a parsonage, a farm weathervane, a cemetery tree and a boat landing. This is a natural-cause (lightning), not human-set, fire story but squarely within “what fires destroyed” — no arson or warden angle claimed for it. No follow-up coverage found.

Rejected/thin leads not published

“Fire warden William Cutts” preventing a small fire on the Sooy place at Vincentown, March 1918 — a single sentence, too thin even by the standard of the Dover-wildman model entry, and adds nothing beyond the one line. Not published.
Various “damages”/”suit against the railroad”/”railroad company for damages” hits were nineteenth/early-twentieth-century grade-crossing negligence suits (Doctor Heineken’s 1856 criminal trial after the backing-train collision that killed Mrs. Prescott and others; the 1906 Davis infant-death suit; the 1910 Allen/Worrell Mount Holly-Lumberton-Medford Railroad verdict) — all about collisions, not fire. None fit this lane’s beat and none were used.
The Howard Hulett Green plane-crash death in a cranberry bog (1938) was checked but explicitly involved no fire (“Harris said there had been no fire”) — out of scope for a fire-beat entry and not published.

Validation note

Every fact in the four published entries traces to an index record fetched and read via curl against netscape.bcls.lib.nj.us this session (screen output retained in this session’s scratchpad under scratchpad/fire/). No forum wording, no invented names, dates, or figures. Every discrepancy noticed between records (Coyle’s “25 years” vs. “14 years”; the Stokingham “destroyed” vs. the 1907 “attractive home”) is stated as an open discrepancy in the entry text itself rather than silently resolved or omitted.

Validation (root fact-check)

Adversarial re-check, September 13, 2026. Every index row re-retrieved from netscape.bcls.lib.nj.us and compared line by line against the published text. All four rows exist exactly as cited (date, page, column, headline, summary). Nothing in the four entries was invented. The corrections below are about overstatement, unattributed cause, unsourced geography, and one factual self-contradiction.

Rows confirmed:

Coyle: Nov 3, 1937, p. 10 col. 3, category “Obituaries”, summary opening “Leonidas S. Coyle”; Jul 6, 1938, p. 1 col. 2, “Memorial to Colonel Coyle”; May 27, 1936, p. 1 cont. p. 5 col. 1, “Five Killed, Many Injured in Greatest Forest Fires…”. 13 rows for keyword “Coyle”.
1893: May 17, 1893, p. 3 col. 1, “LOCAL AFFAIRS. / LOCAL BRIEFS.” Keyword “Quail Run” returns this one row only — a far better citation link than the broad “Toms River and Forked River” search, and now used.
Stokingham: Dec 27, 1905, p. 3 col. 7, “‘STOKINGHAM’ DESTROYED BY FIRE”; Apr 17, 1907, p. 3 col. 4, “WETHERILL–ENGLE”. Both filed under category “Marriages”. 2 rows total.
1922: Jul 19, 1922, p. 10 cols. 5-6, “Local News”, “One Death and Much Damage Caused by Storm”. Keyword “Rahilly” returns it as hit 5 of 14; now the citation link, replacing the unhelpful “railroad set fire” search.

Weekday arithmetic checked and correct: Dec 27, 1905 / Apr 17, 1907 / May 17, 1893 / Jul 19, 1922 / Nov 3, 1937 were all Wednesdays, so the “Thursday” inferences (Dec 21, 1905; Apr 11, 1907; Jul 13, 1922) and “found dead yesterday” = Nov 2, 1937 all hold. But they are inferences, and the entries now say so at the point of use instead of stating the dates flat in the lede.

Corrections made:

Coyle. The nickname was the main problem: “Flying Colonel” stood in the title, H1 and breadcrumb while the entry’s own limits paragraph said “nothing in the record explains why he was nicknamed the ‘Flying Colonel’” — leaving the archive’s most prominent claim about the man looking invented. It is not invented: the July 6, 1938 row reads “for over 25 years head of the bureau, and known as the ‘Flying Colonel.’” That attestation was in the retrieved row but never quoted. Now quoted, in the lede and in the memorial section, and the false “nothing explains the nickname” sentence replaced with an accurate one. Also: the middle-initial conflict (obituary “Leonidas S.”, 1936 coverage “Leonidas J.”, 1938 notice no initial) was silently smoothed over — the lede and items.json said “J.”, the body said “S.”, neither flagged it; now flagged explicitly and the entry uses no initial by default. Removed “in his own plane” (source says “his plane”); corrected the sequence of the final flight (he returned to Trenton after a plane trip and announced the fires “officially out” — the entry had him flying after the fire was declared under control). Added the caveat that a 1938 “Route 40” is not necessarily today’s U.S. 40. Softened “probably reflects” to “may be”. Added columns to the source citation.

1893 fire. Deleted a paragraph of pure speculation about the item’s composition (“appears to have been written after the immediate danger had passed, as a summary of a already-known event rather than breaking news”) — nothing supports it, and the source’s “have raged” arguably cuts against it; it also carried a typo. Deleted the unsourced assertion that the turnpike “ran roughly along today’s Route 166 corridor” — no source consulted establishes the alignment. Added that the fire as described ran south to Barnegat, beyond what the pin covers. Added column number and the narrower search link.

Stokingham. The entry asserted a supposed cause as fact in three places: title (“lost to a lard-rendering fire”), meta description (“after sparks from a lard-rendering furnace were overlooked”) and lede (“A fire that started from an overlooked hog-rendering furnace”). The source says “it is believed that some of the sparks were overlooked” and the flames “are supposed to have started” from them. All three rewritten; a paragraph now sets out the cause as the household’s own supposition, with the source’s hedging quoted. The section heading said “eighteen months later” while its own body said sixteen — fixed to sixteen (Dec 1905 to Apr 1907). Lede stated “the night of December 21, 1905” as fact when the paper says only “Thursday night” — now hedged, with the arithmetic shown. The factbox asserted “Mount Laurel Township”, which no source gives and which sits awkwardly against the genealogy quoted in the very same report placing Stokingham in Willingboro Township on the northerly side of the creek; factbox corrected and the tension noted.

On the contradiction the lane flagged: it is real and was honestly presented, and it survives scrutiny. Same family, same house name, same “along the Rancocas creek, near Centreton”, sixteen months apart — these are not two different houses in two different places. The 1907 item even puts the ceremony in a “spacious parlor”. The likeliest reading is a rebuild on the site under the old name, but nothing found establishes it, so the entry now lays out the candidate readings and picks none. Added a supporting detail the lane had in hand but did not use: the genealogy in the 1905 item itself says Annie Stokes lived there “up to the time of the fire” — the paper treated the residence as ended.

1922 storm. The lede claimed barn fires “near Pointville, Wrightstown and Florence”. Florence is wrong: the only Florence item in the source is a lightning strike on the Rev. B. E. Eldridge’s house that started no fire. The third barn (John Nash’s) is not placed by the report at all — which the entry’s own limits paragraph admitted while the lede placed it anyway. Lede and meta description corrected; the Florence point is now stated explicitly as not a fire. “Tenant farmer Charles Stewart” was an inference — the source says “a farmer, who occupies the Dennis farm”; corrected. Noted that the Rahilly loss is not actually described as fire in the source. “County fair grounds” → “fair grounds” (source says only “the Fair Grounds”). Storm date hedged. Cut an editorial inference about the absence of insurance or negligence detail being “consistent with the incident being accidental”.

On tone for the child’s death: the account was already restrained and remains so. Reworked into plain sequence (she went up to lower the windows, the bolt struck, her parents heard the crash and found her), added the coroner’s name from the source, and continued to omit the report’s description of the state of her hair and clothing, which adds nothing. Added an explicit statement that her identification rests on this single newspaper report and has not been checked against a death record, census return or burial — the entry previously named and aged her without saying how thin that footing is.

Map pins (research/locations/round-nine-fire.json) — all four rewritten, all four now “area”:

Coyle: the pin was titled “Colonel Leonidas Coyle memorial (approximate area, unconfirmed)”, which pins as a memorial a memorial the entry admits it cannot locate. Renamed to “Four Mile, Burlington County — reference point named in the 1938 Coyle memorial notice”, with the basis opening “This pin is NOT the memorial.” Coordinates were fine (USGS GNIS Four Mile, 39.891505/-74.580988, confirmed in the gazetteer).
1893: correctly an area, not a site. Kept the midpoint and 9 km radius, added the Barnegat caveat, cited the GNIS coordinates used.
Stokingham: was “locality” at Centerton. Downgraded to area, radius 3000, because the reports say near Centreton and the genealogy points across the creek toward Willingboro/Rancocas. The log claimed the NJ gazetteer had no Burlington County Centerton and used Nominatim; the gazetteer does have it, at 39.9956677/-74.8729433 — the same coordinates, now cited to GNIS.
1922: geometry re-checked. Mount Holly centre with 20 km covers Florence (~14 km), Wrightstown (~15 km), Pointville (~16 km) and Smithville (~3 km). Kept, with the note that the Nash farm cannot be placed.

All four gazetteer lookups re-verified against DomesticNames_NJ.txt; every coordinate in the file matches GNIS to the digits given. All four source URLs re-requested and returning HTTP 200; limit normalised to 200 across the entries. research/round-nine-fire-items.json details rewritten so they match the corrected entries.

Duplication: re-checked all 386 entries in search.json — no overlap with any of the four. black-saturday-forest-fires-1963.html mentions “Coyle Field” in passing only, and the Coyle entry does not assert any connection between the man and that field.

Scope, referred up, not acted on. Entries 3 and 4 are both outside the Pines, and unevenly so.

Stokingham (Centreton) is the weaker of the two by a distance. Western Burlington County farm country on the Rancocas, a domestic house fire with no pines dimension at all; the only hooks are the county and a governor’s boyhood home. Recommendation: keep only if the archive’s remit genuinely extends to Burlington County’s general fire record. Otherwise this is the one to drop. I have added a “Why this is in a Pine Barrens index” section stating the connection plainly and admitting it is weak, so the entry is honest either way — but that framing is a holding position, not an argument for inclusion.
July 1922 storm is defensible. Mount Holly, Pointville and Wrightstown lie along the agricultural edge of the region, and lightning-ignited farm fires reported in the same columns as the forest fires are reasonable regional context. It is still not a Pine Barrens wildfire and should not be indexed as one; a “Why this is in a Pine Barrens index” section has been added saying so.

Neither entry deleted, and no index, topic, search.json or catalog file touched.

Editorial removal — September 13, 2026 (root)

Dropped stokingham-stokes-mansion-fire-1905 and deleted the file, on the fact-checker’s scope recommendation. The Stokes homestead at Centreton sits in inner farm country with no Pine Barrens dimension; the sourcing was sound, but the subject is outside what this archive covers. The entry never appeared in a published index or changelog, so no removal record is required.