# The Great Chicago Fire, 8 October 1871: how the O'Leary barn caught fire

**What happened.** On the warm, dry evening of Sunday 8 October 1871, a fire started in the cow barn behind the two small cottages of Patrick and Catherine O'Leary, an Irish couple who kept five cows and sold the milk, at 137 DeKoven Street on the West Side of Chicago; they lived in the rear cottage and rented the front one to the McLaughlin family. By Patrick O'Leary's account his wife went to bed about eight and he half an hour after her; on the neighbourhood's time that put him in bed about ten minutes, and her about forty, when the barn caught, though the O'Learys cited no clock. Daniel Sullivan, a neighbour from across the street with a wooden leg, saw it first, some time after half past eight by the neighbourhood's clocks, and ran to cut the cows loose. Within half an hour five buildings were burning; the fire crossed the river that night and burned through the centre of the city for two days.

**Why it is a mystery, and who says what.** Nobody saw the fire start. The newspapers blamed Mrs O'Leary and a cow that kicked over her lamp within a day, a story that Michael Ahern, a reporter in 1871, boasted forty years later he had made up. The Board of Police and Fire Commissioners, the city board that ran the police and the fire department, questioned fifty-one witnesses and found no proof that anyone had been in the barn after dark. H. A. Musham (1940) kept the cow story. Richard Bales, a lawyer with the Chicago Title insurance company, whose land records are the only ones to survive the fire, argued in 1997 and 2002 that Sullivan could not have seen the barn from where he said he sat, so must have been in it; on the strength of his research the Chicago City Council exonerated Mrs O'Leary and her cow in October 1997. Others have blamed a guest from the McLaughlins' party in the front cottage, boys gambling in the loft by lantern light (a claim made public in 1944 on behalf of Louis M. Cohn, a Chicago businessman), a chimney spark, the hay heating by itself, and Biela's comet. A. T. Andreas (1885) and Carl Smith (2011, 2020), the historians who have looked hardest, both concluded that the cause cannot be known.

**How the case was judged here.** The inquiry was read as the Chicago Tribune reported it day by day, with the sworn statements of 15 October 1871, Andreas's interviews of 1884, Bales's transcription of the witnesses (read in fragments), the city directories, an 1891 property atlas, a 1906 fire-insurance map, the present-day lot lines and one photograph. The block was rebuilt to the foot and a three-dimensional model run, with every uncertain dimension varied across its range, to ask what Sullivan could have seen from each of the four places the record puts him. Forty-six statements of the time were sorted by the clock each rested on. Each theory was set against the record and given a share, as a judgment, with the shares recomputed under the readings that change them most.

**Verdict: a flame carried into the barn by a person (about 85 percent), and the person cannot be identified.** The fire began inside the O'Leary barn on the side where wood and shavings were kept and was in the hay loft when the first neighbour reached it (about 95 percent). It began about 8.40 p.m. by the neighbourhood's clocks (range 8.30 to 8.50); the 9.20 to 9.30 of the Board of Police and the press is 35 to 45 minutes late on every clock, the fire department's included. Candidates, ranked as judgments with ranges: Daniel Sullivan alone or with Dennis Regan (another neighbour who called that evening) 22 percent (15 to 30); a member of the O'Leary household with a lamp 22 percent (15 to 30; Catherine O'Leary 15); boys, young men, vagrants or a passer-by from the alley 20 percent (10 to 25); a guest from the McLaughlin party 15 percent (10 to 20); chimney spark 8; arson 3; Regan alone 3; spontaneous heating of the hay 2; other accident 5; Biela's comet 0.

**The percentages are judgments made from the evidence.** They depend on the list of scenarios in report section 7 being complete and on the weight given to each piece of evidence in sections 3 to 7. In the central assessment no candidate reaches 30 percent. Each of three alternative readings in section 8 (accept Bales's blocked view as proof of a lie; accept a neighbour's memory of a woman's scream and the neighbours' affirmations of 1871; accept the Cohn story) lifts one candidate to between 38 and 45 percent, and none of the three rests on evidence that can be checked, which is why the central figures stand.

**Where this stands against earlier accounts.** It agrees with the Board of Police (December 1871), Andreas (1885) and Carl Smith (2011, 2020) that the cause cannot be determined from the evidence. It agrees with Bales (1997, 2002) that the case against Mrs O'Leary was never more than hearsay, and with Fred Fedler (1985), a journalism scholar, that the cow story was in print before any reporter could have invented it as later claimed. The Chicago City Council adopted the same view by resolution on 28 October 1997: introduced by Alderman Edward M. Burke in September 1997 and approved by the Council's Committee on Police and Fire on 6 October 1997 after a hearing at the fire academy that stands on the site of the barn, it resolved to "forever exonerate Mrs. O'Leary and her cow from all blame in regard to the Great Chicago Fire of 1871" (Journal of the Proceedings of the City Council of the City of Chicago, 28 October 1997; text as reproduced by the Glessner House Museum, "Chicago Fire Stories Part III: Catherine O'Leary", 19 October 2021; the hearing in the Chicago Tribune, 6 and 8 October 1997, and the Knight-Ridder news service, 12 October 1997). The act was the City Council's; Cook County took no part in it. It is a legislative judgment on Bales's research and adds nothing to the evidence about the night. It differs from Bales on Daniel Sullivan: his sightline argument holds for the one seat he chose (the midpoint of William White's west lot, where the model shows nothing under about 16 ft above the barn's ridge: median 15.6 ft with the eye at the height of a man seated on the raised sidewalk, 14 to 16 ft across the three eye heights tried) and fails for the three other placements the record gives (his affidavit's "a few lots east", the Board's "nearly opposite", Andreas's "in front of his own house", which by Bales's own deeds is the next lot west, where a flame within 5 ft of the ridge is visible in 61 percent of the model's draws and the barn itself in 17 percent). The difference is one 25-ft lot and about three feet of house height. His time argument turns on a late time that the Board and the press adopted as well as Sullivan; the error belongs to the whole official record.

**New evidence** (documents not previously used in the published discussion, as far as could be checked against a book read only in fragments): the 1871 Edwards directory entries showing Sullivan and White as teamsters for the same firm (Chase, Hanford & Co.); the Tribune's reports of the inquiry sittings of 29 November and 4 December 1871, on pages missing from the Library of Congress set (30 November p. 4 and 5 December p. 2) and recovered from another scan; the Illinois Staats-Zeitung's silence on the inquiry and the Board's report; the dating of the Evening Journal's O'Leary interview to 21 October 1871; a second Chicago extra of 9 October 1871 carrying the cow story.

**New analysis of old evidence:** the sightline test (a 3-D model built from the 1891 atlas, the 1906 Sanborn map, the present-day parcel, one photograph and Bales's deeds, run with uncertainty ranges over all four placements of Sullivan already in the record, 7,200 draws; it adds no document to the record); the three-cluster reading of 46 time statements and the 10 to 15 minute offset between neighbourhood and department clocks (a re-reading of the Tribune's and Andreas's reports); the inference from computed sunrise that the 5 a.m. milking was in darkness, so a lamp lived in the barn and its remains prove nothing; the re-working of Bales's Appendix B arithmetic showing that his 193 feet depends on the seat he chose. Anthony DeBartolo (Chicago Tribune, 1997 and 1998) found that Cohn's recorded will contains no confession; that finding is his, and this report only repeats it.

## Files
- `report/chicago-fire.html` — the report (published at https://historical-mysteries.netlify.app/chicago-fire/report), 12 sections, two figures, six tables, sensitivity table.
- `analysis/` — `plan_inputs.json` (every dimension with its source line), `sightline_mc.py` (the model; `python3 sightline_mc.py 400`), `summary.csv` and `results/` (raw draws), `draw_figures.py`, `fig_plan.png`, `fig_profile.png`, `timeline.csv` (46 time statements with source and clock basis), `mc_run.log`.
- `maps/` — notes on the block plan and their source log, `plan/plan_coordinates.csv` and `plan_coords.py` (the coordinate frame from the 1891 atlas and the modern parcel), the digitised Musham plan, directory extracts and the three Edwards directories (OCR), page images of Andreas's plat (p. 717), his "West line of fire" plat (p. 714), Musham's plan (p. 98) and the Abbott stereograph.
- `press/` — notes on the 1871 press (about 120 issues, 9 October to 16 December 1871, with a day-by-day chronology of the legend) and their source log; `pages/` holds the OCR text of every Tribune page 11 October to 16 December and of the out-of-town pages used, including the two recovered pages.
- `literature/` — notes on the later literature (Musham, Bales, Cromie and Kogan, Sawislak, Smith, Miller, Pauly, Fedler, Cohn, Ahern, the comet, the 1997 resolution and its press, 2021 coverage), their source log, Bales's Google Books snippets by query (128 queries, 534 snippets), and full texts of Fedler 1985, DeBartolo 1997 and 1998, Smith's American Heritage excerpt and Musham's origin sections.
- `bales/` — the recovery of Bales's text: `REPORT.md`, `bales_argument.md` (his Sullivan case as recovered, with every fragment and query id), `chains_dump.md` (the 76 passages), `appendixD/` (fragments of his transcript of Sullivan, Regan, the O'Learys, the Sullivans, McLaughlin, Brown), `query_log.tsv` (435 queries), tooling.
- `transcript/` — text extracts: the Tribune's report of each witness, Andreas 1885 pp. 706 to 719, the Board's 1872 report, Sheahan and Upton, Colbert and Chamberlin, Goodspeed, Luzerne; index in `transcript/text/INDEX.md`.
- `ACCESS_NOTES.md` — what was blocked (Chicago History Museum transcript site down; HathiTrust, Chicago Collections and the museum's image server behind bot walls; archive.org lending copies) and what was not attempted.

## What this report could not do (report section 11)
The inquiry transcript was not read in the original. Bales's book was read through two fragment services and never in full: 128 Google Books queries returned 534 page-numbered snippets from 187 pages, and 435 Internet Archive search-inside queries returned 1,439 fragments of about 100 characters, mostly repeats, joined into 76 passages; his Figures 54, 64 and 78 and his 1997 article were not seen. Carl Smith 2020 was read through its American Heritage excerpt; Sawislak, Pauly, Miller, Cromie and Kogan and Cromie were not obtained; Ahern's and Cohn's originals are known through later quotation. The 1997 resolution is cited from its own text as reproduced by the Glessner House Museum and from the press of October 1997; the Journal of the Proceedings for 28 October 1997 was not reached online. The barn's orientation is unsettled between Andreas's plan and Musham's, and the heights of Dalton's house, of the O'Learys' two cottages and of the cottage at No. 133 rest on one photograph and typical construction; the model carries these as ranges and says so. No paywall or bot wall was bypassed.

## Summary for the case index
- Title: The Great Chicago Fire
- Verdict line: Leading candidate. A flame carried into the barn by a person, about 85 percent; the person cannot be identified from the surviving evidence. Among the possible carriers, Daniel Sullivan and a member of the O'Leary household stand first, close to even.
- Confidence and sensitivity: about 85 percent for a carried flame. Leading candidates 22 / 22 / 20 / 15, none above 30 in the central assessment; three alternative readings lift one candidate each to 38 to 45, none on evidence that can be checked.
- Evidence label: leading candidate.
- What is new: five documents (the 1871 directory entries, the two Tribune pages missing from the Library of Congress set, the Staats-Zeitung's silence, the date of the Journal's O'Leary interview, a second extra of 9 October) and new analysis of old evidence (the sightline model over all four placements of Sullivan, the three-cluster timeline, the milking-in-darkness point, the re-working of Bales's arithmetic).
- Where it stands: agrees with the Board of Police (1871), Andreas (1885) and Carl Smith (2020) that the cause cannot be determined; differs from Bales (2002) on Sullivan.
- Images: Figures 1 and 2 (the block plan and the sightline sections) and Tables 1 to 6 were made for this report from the sources named in their captions; the report reproduces no historical images. Figures 1 and 2 are `analysis/fig_plan.png` and `analysis/fig_profile.png`, drawn by `analysis/draw_figures.py`; Table 3 is built from `analysis/timeline.csv` and Table 4 from `analysis/summary.csv`.
- Opening: the report begins with what happened, why it is a mystery and who says what, and how the case was judged, then the verdict panel.