Suspicious Fire Patterns and Burn Damage Interpreted as the Result of Ignitable Liquids

NFPA 921 | Law Review Articles | Research ArticlesCases | Other

Until the first edition of NFPA 921: Guide for Fire and Explosion Investigation was published in 1992, there were a number of widely accepted myths and misconceptions regarding how a fire investigator could examine fire patterns and burn damage at a fire scene and deduce the presence of an ignitable liquid.

NFPA 921


NFPA 921 (2014) at 70
Irregular patterns are common in situations of post-flashover conditions, long extinguishing times, or building collapse. These patterns may result from the effects of hot gasses, flaming and smoldering debris, melted plastics, or ignitable liquids. If the presence of ignitable liquids is suspected, supporting evidence in the form of a laboratory analysis should be sought.


NFPA 921 (2014) at 63
Holes in floors may be caused by glowing combustion, radiation, or an ignitable liquid. The surface below a liquid remains cool (or at least below the boiling point of the liquid) until the liquid is consume. Hoses in the floor from burning ignitable liquids may result when the ignitable liquid has soaked into the floor or accumulated below the floor level. Evidence other than the hole or its shape is necessary to confirm the cause of a given pattern.

NFPA 921 (2014) at 64
Penetrations in a downward direction are often considered unusual because the more natural direction of heat movement is upward because of the buoyancy of heated gases. In fully involved compartments, however, hot gases may be forced through small, pre-existing openings in a floor, resulting in a penetration. Penetrations may also arise as the result of intense burning under furniture items such as polyurethane mattresses, couches, or chairs. Flaming or smoldering under collapsed floors or roofs can also cause floor penetrations.


NFPA 921 (2014) at 65
The angle of the borders of the V pattern does not indicate the speed of fire growth or rate of heat release of the fuel alone; that is, a wide V does not indicate a slowly growing (“slow”) fire and a narrow V does not indicate a rapidly growing (“fast”) fire.


NFPA 921 (2014) at 56
Crazing is a term used to describe a complicated pattern of short cracks in glass. These cracks may be straight or crescent-shaped and may or may not extend through the thickness of the glass. Crazing has been claimed to be the result of very rapid heating of one side of the glass while the other side remains cool. Despite widespread publication of this claim, there is no scientific basis for it. In fact, published research has shown that crazing cannot be caused by rapid heating, but can only be caused by rapid cooling. Regardless of how rapidly it was heated, hot glass will reproducibly craze when sprayed with water.


NFPA 921 (2014) at 49
In the past, the appearance of the char and cracks had been given meaning by the fire investigation community beyond what has been substantiated by controlled testing. The presence of large shiny blisters (alligator char) is not evidence that a liquid accelerant was present during the fire, or that a fire spread rapidly or burned with different greater intensity. These types of blisters can be found in many different types of fire. There is no justification for the inference that the appearance of large, curved blisters is an indicator of an accelerated fire. Figure (at 49), showing boards exposed to the same fire, illustrates the variability of char blister.


NFPA 921 (2014) 6.2.5
The presence of absence of spalling at a fire scene should not, in and of itself, be construed as an indicator of the presence or absence of liquid fuel accelerant. The presence of ignitable liquids will not normally cause spalling beneath the surface of the liquid. Rapid and intense heat development from an ignitable liquid fire may cause spalling on adjacent surfaces, or a resultant fire may cause spalling on the surface after the ignitable liquid burns away.


NFPA (2014)
Wood and gasoline burn at essentially the same flame temperature. The turbulent diffusion flame temperatures of all hydrocarbon fuels (plastics and ignitable liquids) and cellulosic fuels are approximately the same, although the fuels release heat at different rates.

Law Review Articles

Giannelli, Paul, Junk Science and the Execution of an Innocent Man (2013)

Shifted Science and Post-Conviction Relief, C. Plummer and I. Syed, 8 Stanford Journal of Civil Rights and Civil Liberties, Vol. 8 (2012).

Research Articles

Lentini, Scientific Protocols for Fire Investigation, Ch. 8 (2006) (discussing myths of arson investigation).

Grann, D., Trial by Fire: Did Texas Execute an Innocent Man?, The New Yorker, Sept. 7, 2009.

National Research Council, National Academy of Sciences, Strengthening Forensic Science in the United States: A Path Forward (2009).

Lentini, The Mythology of Arson Investigation 

Lentini, Evolution of Investigation, at 4-6 

Beyler, C.L., Analysis of the Fire Investigation Methods and Procedures Used in the Criminal Arson Cases Against Ernest Ray Willis and Cameron Todd Willingham, Report to the Texas Forensic Science Commission, August 17, 2009.


R. Ogle, et al., The Scientific Investigation of Arson Fires, Exponent Failure Analysis Associates, Fire and Casualty (Nov. 2003).

“[i]n free-burning, pre-flashover fires, floor burn-through patterns are most likely caused by solid fuel fires.  Liquid fuel fires do not penetrate floors in free-burning fires.  Therefore, floor burn-through patterns are not probative evidence of an incendiary fire started with an ignitable liquid.


State of Texas v. Cameron Todd Willingham, 897 S.W.2d 351 (Tex.Cr.App. 1995)

Han Tak Lee v. Houtzdale SCI, 798 F.3d 159, 166-69 (3d Cir. 2015)

Souliotes v. Hedgpeth, No. 1:06-cv-00667, 2012 WL 1458087 at *18-22 (E.D. Cal. Apr. 26, 2012), adopted at 2012 WL 2684972 (E.D. Cal. Jul. 6, 2012)

United States v. Hebshie, 754 F. Supp. 2d 89, 92 (D. Mass. 2010)


NAS Report (2009) at 5, 34, 35

[M]any of the rules of thumb that are typically assumed to indicate that an accelerant was used . . . have been shown not to be true.