SF-0004Strange America
A fire has burned under Centralia, Pennsylvania since 1962 — and the state does not claim to know how it started
The fire is real and the town is gone. The two details everyone repeats — the trash fire, and the 250 years — are the two the record supports least.
Route 61 south of Centralia, buckled and venting where the fire burns beneath it. The road was closed to traffic and has never reopened.
Cracked highway from subsurface coal fire (Route 61, near Centralia, Pennsylvania, USA) 2" — the photographer's own title. His accompanying description states that the fire "was accidentally set by firemen hired by town officials", which is the confident version of the origin that this file exists to qualify. The description is not used as evidence for anything; it is recorded here because a caption repeating the popular account, attached to an otherwise excellent photograph, is a neat demonstration of how durable that account is.
- Image
- James St. John
- Taken
- not established — see reproductionNote
- Location
- Route 61, near Centralia, Columbia County, Pennsylvania, USA
- Licence
- CC BY 2.0 (licence record)
- The image itself
- Authentic, unaltered
Photograph by James St. John, licensed CC BY 2.0. Rescaled and cropped at the bottom edge for the card.
Verdict
"In 1962 the town of Centralia burned its trash in an old strip mine, accidentally set the coal seam underneath on fire, and it's still burning today — it has enough coal to burn for another 250 years."
The core of the claim holds up, but a specific detail as it circulates is wrong or overstated.
The core of the claim is straightforwardly true and meets the Verified bar on a single Tier 1 source: Pennsylvania's Department of Environmental Protection records that the Centralia Mine Fire was first reported on 27 May 1962 and is still monitored today. Two specific circulating details fail. First, the origin: DEP states there "have been many plausible theories on how the Centralia Mine Fire got started", calls the deliberate-burning account the "most widely accepted ignition scenario" rather than the established one, and lists three alternatives — a 1930s mine explosion, spontaneous combustion of trash, and vandalism. Second, the duration: DEP's own estimate is that the fire "could burn for over 100 years" if left uncontrolled. The 250-year figure in circulation is not the state's. Mostly True is the correct verdict because the claim's spine holds and two named details do not.
Mostly True requires 1 primary or authoritative source. This file cites 2.
In short
A coal seam fire has burned beneath Centralia, Pennsylvania since May 1962. That much is documented and uncontested. The two details that travel with the story are both softer than they sound. Pennsylvania's own environmental agency lists four competing origin theories and calls the trash-fire account "most widely accepted" rather than established — and the state's own estimate of how long the fire could burn is over 100 years, not the 250 that circulates.
Research packet — complete
What the file turned out to be
The usual Centralia story is fire real, town gone, still burning. All true, all well covered, nothing for this publication to add.
The file the research actually found is better: the two details people state most confidently are the two the record supports least.
Pennsylvania’s own Department of Environmental Protection — the agency that monitors the fire, drilled the boreholes, and paid for the relocations — does not claim to know how it started. It says there “have been many plausible theories”, names the deliberate-burning account as “most widely accepted”, and lists a 1930s mine explosion, spontaneous combustion, and vandalism alongside it.
And the “250 more years” that travels with every retelling is not the state’s number. DEP’s estimate is over 100 years.
Why it is MOSTLY TRUE rather than MISLEADING
The spine of the claim is correct. There is a fire, it started in May 1962, it is still burning, and it destroyed the town. A reader who believed only that would not be misled about anything that matters.
Two named details fail, and both are specific enough to correct precisely. That is exactly the shape Mostly True exists for — and the verdict standard requires stating plainly which part is true and which is not, which this file does.
What remains before READY
- The cooling-off period. The verdict was assigned 2026-07-26 and no file publishes on the day its verdict is first assigned. This clears on 2026-07-27.
- The image attribution. The Bureau of Mines photograph is almost certainly public domain, but the federal authorship is asserted on a Commons page rather than confirmed against a government catalogue. Either confirm it, or switch to the verified CC BY 2.0 fallback. Our own standard says not to reason around that gap.
- Optional, not blocking: confirm the Christian Science Monitor attribution at the original, and re-check the current population.
The blocker that dissolved
The previous pass recorded this file as blocked on Bureau of Mines RI 8799, which
stacks.cdc.gov refuses to serve to automated requests.
That was correct at the time and turned out to be the wrong thing to wait for. The PA DEP FAQ — already cited, already downloaded, and previously read only through a failed automated extraction — answers both load-bearing questions directly and at Tier 1. Reading a source we already had, properly, was worth more than obtaining a source we did not.
RI 8799 would still strengthen the file and remains on the list. It is no longer what stands between this file and publication.
The evidence
Pennsylvania DEP: "The Centralia Mine Fire was first reported on May 27, 1962, just prior to Memorial Day." Note the wording — first reported, not first started.
Cited sources:S1
DEP on the origin: "Because of the initial rapid and unexpected spread of the fire, there have been many plausible theories on how the Centralia Mine Fire got started… The most widely accepted ignition scenario is that the fire was deliberately set by Borough personnel." Most widely accepted is not the same as established.
Cited sources:S1
DEP lists three competing origins alongside the landfill account: "a mine explosion in the 1930s, the spontaneous combustion of trash and even vandalism as the cause of the initial flare-up."
Cited sources:S1
DEP's duration estimate: "If left uncontrolled, it is estimated that the Centralia Mine Fire could burn for over 100 years." The state's figure is over 100 years, not 250.
Cited sources:S1
The widely repeated "over 250 years" figure is attributed to a February 2010 Christian Science Monitor article rather than to any government or geological estimate.
Cited sources:S2
DEP gives the 1962 population as "approximately 1,100 to 1,200 people" — lower than the 1,500 that commonly circulates — and records 5 remaining property owners as of January 2013.
Cited sources:S1
The fire is measured, not merely observed. Borehole records from the 1970s and 1980s show many readings above 1,000°F, with a maximum of 1,350°F; ground-surface temperatures reached 900°F where the fire came close to the surface. Over 2,000 boreholes have been drilled since 1966, and the state still monitors the surface monthly.
Cited sources:S1
By January 2012 the fire involved approximately 400 acres, having expanded beyond the borough boundary, advancing on average 50–75 feet per year over five decades in a "highly variable or uneven fashion".
Cited sources:S1
Cost: over $7.0 million was spent on fire control and emergency relocation by 1983, and approximately $41.6 million on relocation between 1983 and January 2012.
Cited sources:S1
The fire is not, by DEP's account, unstoppable in principle — only unaffordable. "The Centralia Mine Fire could feasibly be excavated, but would require a very large and expensive project. The cost… is currently beyond the capacity of Pennsylvania's AML Program to address."
Cited sources:S1
DEP's own chronology of the fire records only "May 1962 — Fire Commences" and assigns no cause. It is a document that lists the interception trench, the flushing projects, the buyout and the relocations across five decades — and it does not say how the fire started.
Cited sources:S3
Evidence that cuts the other way
Findings that complicate or argue against the verdict above. They are listed because leaving them out would make the file look stronger than it is.
If "many plausible theories" was DEP's position, then the claim's most confident element — a specific date and a specific mechanism — is the least established part of it. That inverts the usual shape of these files, and it is the reason this one is worth publishing.
Cited sources:S1
DEP does not endorse the vaguer end of the legend either. The fire does not burn beneath the cemeteries — most of that ground is barren or was isolated by excavation — and it is judged "highly unlikely" ever to reach Mount Carmel. Several of the more lurid framings of the Centralia story fail against the same document that undercuts the claim.
Cited sources:S1
What we still don’t know
- The withdrawn image. A 1969 Bureau of Mines photograph was researched and rejected rather than published: its public-domain claim depended on federal authorship asserted on a Wikimedia Commons file page, from a scan sourced to a 1986 trade book, with no Bureau of Mines catalogue record located. It may well be public domain. That is not the same as being documented, and this file does not publish the difference.
- When the replacement photograph was taken. The file carries no EXIF and Flickr's "17 December 2012" is an upload timestamp; the 4:3 frame and autumn foliage are inconsistent with it. No claim here depends on the date, so it is left unstated.
- How the fire actually started is not established, and this file does not resolve it. DEP names a most-widely-accepted scenario and three alternatives; nothing read here adjudicates between them.
- Whether US Bureau of Mines Report of Investigations 8799 (Chaiken et al., 1983) treats the origin as settled. It is the most authoritative technical study of this specific fire and could not be retrieved — stacks.cdc.gov returns 403 to automated requests. It is no longer load-bearing now that DEP answers the question directly, but it would strengthen the file.
- Current population and the status of the 2013 agreement with the remaining residents. DEP's figures stop at January 2013 and must be re-checked immediately before publication.
- Whether any source supports a 250-year estimate. This file establishes that DEP says over 100 years, and that the 250 figure is attributed to journalism — not that no such estimate exists anywhere.
Sources
Every source below was read in full before it was cited. Archive snapshots are recorded at research time, because a file whose sources have evaporated is indistinguishable from one that never had any.
S1Tier 1Read in full
The Centralia Mine Fire — Frequently Asked Questions
Pennsylvania Department of Environmental Protection, Bureau of Abandoned Mine Reclamation · 2013
The state agency responsible for the fire, answering 41 questions across six pages. Source of every quantitative and qualitative finding in this file: the 27 May 1962 report date, the four competing origin theories and DEP's own hedge on them, the over-100-year duration estimate, 1962 population, borehole temperatures, acreage, advance rate, spending, and the judgement that excavation is feasible but unaffordable. Read in full.
S3Tier 1Read in full
Centralia Mine Fire — Chronology
Pennsylvania Department of Environmental Protection
The state's own timeline of the fire, read in full. Corroborates the 1962 start independently of the FAQ, and establishes something the FAQ cannot: that DEP's chronology records only "May 1962 — Fire Commences" and states no cause at all. A chronology is the document in which an agency would record an origin if it considered one settled.
S2Tier 3
Wikipedia
Finding aid only, as the source hierarchy permits. Used solely to establish that the 250-year figure is attributed to a February 2010 Christian Science Monitor article, and to locate Bureau of Mines RI 8799. Not evidence for any claim; the CSM attribution is carried as suggestive and should be confirmed at the original before publication.
Checked for circularity: S1 and S2; S3 and S2 were confirmed not to derive from one another. Two sources that trace back to the same origin are one source, and counting them twice is the most common way a false claim looks well-evidenced.
How we checked this
What we went looking for, and what we found
The disconfirming search produced this file's actual subject. Searching for the case against the standard landfill account surfaced DEP's own hedge — "many plausible theories" — plus three named alternatives, which is a stronger and more interesting finding than confirmation would have been. Searching against the 250-year figure produced the state's own materially different estimate. Both were found by looking for the weakness rather than the support.
- We searched for evidence against our own conclusion
- We could not trace where the claim started
- The claim is stated in its strongest form, not a weak version
- The photograph was verified independently of the story
- We checked whether this was settled after the popular version froze
- Image rights were traced to a stated licence
- We checked what we were leaving out
- The verdict sat before publication, and was re-read cold
One check was not completed on this file. It is listed above rather than omitted, because a checklist that shows only its ticks is decoration.
- Confidence in the verdict
- high
- Verdict assigned
- Most recent authoritative source
- Researched by
- The Strange File
Cleared for READY on 2026-07-30. Both remaining blockers are resolved, and one of them by replacing the image rather than by arguing the old one through.
The image blocker was real and the previous note called it correctly: the withdrawn photograph's public-domain basis rested on a Commons uploader's assertion of federal authorship, from a scan that came via DeKok's 1986 book rather than from a Bureau of Mines catalogue record. That is an inference, not a licence. It was replaced with James St. John's CC BY 2.0 photograph, where the licence is stated by the creator on his own page, has a 2019 Wayback snapshot showing the same terms, and was independently confirmed by Commons' Flickr review in January 2026. Better photograph, and no inference required.
The cooling-off period is satisfied: the verdict was assigned 2026-07-26 and this is a separate pass four days later that re-read the FAQ's origin wording, added the chronology, and changed the image.
US Bureau of Mines RI 8799 remains unretrieved — stacks.cdc.gov returns 403 to automated requests, and that did not change on re-attempt. It is recorded as an open question rather than a blocker because DEP answers the origin and duration questions directly at Tier 1, and RI 8799 would corroborate rather than establish. Anyone with browser access to CDC Stacks can close it.
Archiving is incomplete and deliberately not faked. S1 carries a verified Wayback snapshot from 2026-07-26. S3 does not: a save was requested and returned 302, but the availability and CDX endpoints both returned errors, so no snapshot URL could be confirmed. An unverified archiveUrl is worse than an absent one, because it looks like a safety net that is not there. The field is left empty, the advisory gate is allowed to say so, and the next pass should retry.
Part of a collection
The Qualifier Is The StoryThe interesting content is in the clause everyone skims.
What The Record Declines To SayAn authority’s silence, treated as evidence rather than as a gap.
Places America LeftTowns emptied by industry, and the sequence the retelling flattens.
Also related
The USGS calls 1905 the Salton Sea's most recent formation, not its first
The Colorado broke through a half-finished irrigation headgate in 1905 and took until 1907 to put back. But the river had filled that basin on its own before, and each time the basin dried out again.