Fire patterns: the measurements behind a claim of arson
Fire patterns can record damage without identifying what started the fire. In a 2015 Scientific American article, Michael Shermer reported investigator John J. Lentini's criticism of using the size and sheen of char blisters to classify a fire as fast or slow. He also rejected the leap from those classifications to a confident statement about cause. His example involved the intense burning of furniture such as a couch or bed. This was an experienced investigator's criticism of an interpretive shortcut. The stronger foundation for investigating that shortcut lies in controlled burns, measurements and procedures that preserve the surrounding evidence.
The National Institute of Justice's fire investigation guide distinguishes origin from cause. Origin concerns where a fire began; cause concerns the ignition source, the first fuel and the circumstances bringing them together. Those are different findings. Locating the most conspicuous damage does not by itself identify an ignition source. The guide directs investigators to document a scene systematically and consult specialists when origin or cause is uncertain. Fire patterns belong within that larger process. A couch's appearance has to be considered alongside the room, the available fuels, witness observations and what happened during firefighting, rather than functioning as an independent verdict about arson.
Anthony Putorti's 1997 NIST study used full-scale rooms furnished as residential bedrooms to investigate burn patterns. Researchers measured temperatures and heat flux at different positions. In some experiments they also measured oxygen, carbon monoxide and carbon dioxide. Post-fire inspection could then be compared with recorded conditions during the burn. This reverses the uncertainty inherent in finding a damaged room after the event: the experiment supplies measurements of what occurred before investigators examine the resulting marks. The published research description explicitly identifies the limited scientific foundation for many traditional fire patterns and indicators as a reason for the work.
A 2010 NIST laboratory study by Daniel Madrzykowski and Charles Fleischmann examined repeatability on painted gypsum wallboard. Researchers used a natural-gas burner, a gasoline pan and polyurethane foam as distinct fuel sources. Each source had a top surface 1 foot square, approximately 3 inches above the floor, positioned against the middle of the rear wall. The fire patterns were studied before flashover. The three-sided test structure had a full floor and partial ceiling, with painted wallboard lining its wood framing. Fresh painted wallboard was used for each experiment. An oxygen-depletion calorimeter measured heat release, and photographs and video documented flame behavior. The investigators compared the height, width and area of the resulting fire patterns. Repetition and controlled placement made variability something to measure, rather than something to explain away after a single fire.
That wallboard experiment was not a direct test of every blister on upholstered furniture. Its scope matters precisely because the material, fuel and geometry were specified. The NIJ guide addresses a related problem at actual scenes: conditions can change before evidence is collected. It calls for observations about doors and windows and changes made during suppression, as well as photographs before objects are moved. It warns that fuel-powered equipment can contaminate possible ignitable-liquid evidence. For laboratory identification of liquids, it specifies immediate sealing in clean, unused, vaportight containers. Labels identify the collection and recovery location, while custody records track transfers to named recipients. Recorded interviews with the first firefighters and preservation of the original fire report can retain observations made before the scene changed. Comparison samples help place laboratory findings in context. Fire patterns and laboratory findings become more useful when their physical context and handling history remain available.
The consequences of lost opportunities for testing appear in the Third Circuit's 2012 decision in Han Tak Lee v. Glunt. Lee challenged the scientific basis of fire testimony used at his murder and arson trial. Lentini supplied a cautious assessment: he could not determine the cause without further access to the evidence, although he considered the prosecution's observations consistent with an accidental fire. The appellate court held that denying discovery was an abuse of discretion and required further proceedings. That ruling did not itself establish an accidental cause. It recognized the importance of letting an independent expert examine the physical material using subsequent scientific developments.
The court also noted a trial expert's acknowledgment that indicators accepted when the investigation occurred might be viewed differently after later research. That distinction separates a historically common practice from a scientifically supported inference. Lentini's criticism of blister categories likewise does not mean that all damage is meaningless. NIST studied fire patterns through measured experiments, and NIJ specifies procedures for documenting observations and preserving evidence. A small, flat blister and a large, shiny one cannot independently establish why a couch caught fire. Identifying the ignition source still requires evidence connecting the damage to a particular sequence of events.
- Anthony D. Putorti Jr., NIST, Full Scale Room Burn Pattern Study, NIJ Report 601-97, 1997 — www.nist.gov
- Daniel Madrzykowski and Charles Fleischmann, NIST, Fire Pattern Repeatability: A Laboratory Study on Gypsum Wallboard, 2010 — www.nist.gov
- National Institute of Justice, A Guide for Investigating Fire and Arson — nij.ojp.gov
- U.S. Court of Appeals for the Third Circuit, Han Tak Lee v. Glunt, No. 10-4133, 27 January 2012 — www.govinfo.gov
- Michael Shermer, Scientific American, Can We Trust Crime Forensics?, 1 September 2015, reporting John J. Lentini's remarks — www.scientificamerican.com