As with many of the “classic” fire indicators, the presence of collapsed furniture springs has been misinterpreted to prove among other things that accelerants were or were not used to start a specific fire and that the fire dynamics were related to slow, smoldering fires such as those associated with cigarettes, as some texts indicate.1

As recently as 1994, an article on fire investigation stated that “springs at the point of origin will show localized annealing”2, which might also be interpreted to mean that the absence of annealed springs could be used to rule out that item as the point of origin. With only a marginal effort, an investigator can find some text (generally with unsupported references) to confirm what the investigator believes to be the cause of the fire. Considering the published research that concludes otherwise, it is puzzling that so much confusion surrounds the issue of collapsed springs as fire indicators.3 What is even more astounding is that many researchers encourage the use of the scientific method and highlight the need for basing decisions concerning fire origin and cause on and research and testing. In addition, as far as I know, there is no scientific testing to support the conclusions presented in texts that present collapsed springs as proof of a particular fire condition.


Prior to the core discussion regarding collapsed springs, some definitions or parameters covering terminology must be established. In this article, a collapsed spring will mean any spring that has been affected by “any permanent reduction in overall length (height) unless qualified by a percentage, in which case reference is made to the original height (length).”4

One of the investigative problems with judging postfire spring height is the lack of data concerning prefire spring condition. Even if we could determine the original spring height (at the time of manufacture), this datum may not correlate to the condition of the spring at the time of the fire.

In a number of instances, I have cut out the springs in used mattresses and found significant differences between the center springs and those on the sides or corners. In one mattress, there was an 11 percent difference between the lengths of the center springs and the heights of the corner free-standing springs. In this particular case, the original spring height was unknown, but based on observations of the mattresses, it could be assumed that it was close to the existing corner spring height. This example points out that before we could make scientific deductions concerning collapsed springs and the use or absence of accelerants, we would need to know the original spring height, the general condition of the furniture item, and the prefire spring height (plus additional information that will be presented)–conditions that rarely, if ever, are available to the investigator.


Various fire investigators have generated confusion over the definition of “annealing.” In relation to fire investigations, confusion even exists with regard to what annealing is (in some cases, it has been presented as an “accident” that happens during some fires). Technically, annealing is just one of a number of heat-treatment processes used to modify the properties of metal or alloys. Full annealing, isothermal annealing, normalizing, spheroidizing, quenching and tempering, martempering, austempering, and dual-phase processing are just some of the treatments used to modify metals. Metals are intentionally annealed to “relieve stresses induced during hot and cold working or to soften the metal to improve its machinability or formability.”5 The annealing processes are based on predetermined temperature/time parameters and controlled cooling rates.

What occurs to springs in a fire is probably a result of a number of these different processes, with the closest thing to true annealing occurring during a total structure burn where no subsequent suppression efforts take place (this might mimic the slow cooldown required for true annealing). For our discussion, annealing will be defined as the heating of a coiled compression spring (which is time and temperature dependent) that in turn produces changes in the properties of the metal or alloy.

Furniture springs are made by drawing high carbon steel through progressively smaller dies until the desired spring diameter is attained. This process is typically conducted at ambient temperatures (cold worked) and by its nature induces stresses and other physical changes to the properties of the metal. High-carbon steel is an alloy of ferrite (iron) and cementite (iron carbide), with final carbon contents of between 0.55 and 0.95 percent.6 Iron at room temperature contains very limited amounts of carbon interstitially. As the steel is cooled, the carbon changes into a new phase represented as cementite, which is hard and brittle. “In general, increasing the carbon concentration will increase the hardness, increase the yield strength, and lower the ductility (ability to deform without breaking) and toughness (the ability to absorb energy of impact)”7 (see Table 1).

In this table, yield strength is particularly important, since it relates to the “maximum stress that can be applied to the metal before it will undergo some permanent deformation….”8 This means that if we place a spring with 0.50 percent carbon and an identical (size, diameter, and so on…) spring with 0.70 percent carbon into a fire (and all other conditions are equal), the spring with the lower carbon content will collapse before the spring with the higher carbon content.

The only way investigators could quantitatively compare all the collapsed springs they have observed would be to base their observations on the prefire carbon contents of the springs used to make the comparisons (clearly a daunting task). Even when comparing multiple furniture items in the same fire room (if all other conditions are equal), collapsed springs would have to be judged in relation to the carbon content of each spring.

One of the arguments you may use against this concept is “I may not know the prefire carbon content, but I can test the hardness of the springs after the fire and factor that into the evaluation of the collapsed springs.” The major problem with this theory is that, among other factors, the spring`s postfire hardness is dependent on its cooling rate during and after the fire (data obviously not available). Depending on the spring`s cooling rate after the fire, it would be possible for the spring to have a hardness rating that was the same, lower, or higher than its original prefire rating. Additionally, any attempt to extrapolate temperatures at the time of the fire by using postfire spring hardness ratings could result in temperature errors of more than 1,250 degrees F. As pointed out by Tobin & Monson: “Additionally, springs can become carburized or decarburized when exposed to reducing or oxidizing atmospheres at high temperatures, respectively, further altering surfaces` hardness values obtained from post-fire measurements.”9


From observations of damaged springs, investigators know that springs do collapse during fires. The obvious questions are Why? and Can this information be of practical use to the investigator? Elastic deformation and yield strength are key components relating to a compression spring`s response to a fire. A steel spring`s elasticity (ability of a spring to return to its original height [reversible deformation] after a load is removed) is independent of “the amount of cold work, annealing, or other heat treatments…” but it is strongly temperature dependent.10 Looked at another way, no matter what the spring`s fabrication history, on the basis of elasticity alone, the hotter the fire, the greater the probability that the spring will not return to its original height.

This was probably the starting point for the misconceptions regarding collapsed springs. Somebody saw that the elastic deformation spring response was primarily a function of temperature and deduced that the greater the spring`s collapse, the higher the temperature and that a drastic or total collapse must mean an accelerated fire. The basic problem is that all of the other properties and conditions that affect the performance of the spring during a fire and the conditions of the fire itself were not factored in.

Yield strength is also temperature dependent; but, unlike elasticity, it is very dependent on production factors such as cold working and heat treatments. Both yield strength and elasticity show significant reductions by the time the fire reaches 650°C (1,200°F), a temperature that can be easily attained without the use of accelerants.

When time and temperature are factored along with the cold work history of the metal, the possibility of spring failure or collapse can occur at a wide range of time/temperature conditions. An investigator might be able to roughly estimate temperature conditions within a room (based on known melting temperatures of materials) but will not be able to make a reliable estimate of time vs. temperature (that is, the fire growth rate) and will rarely (if ever) be able to attain the fabrication history of the spring. This means that the “classic collapsed” bed springs could be caused by a slow-growing smoldering fire, a fast-growing fire, or any combination in between. But the collapsed springs do not by themselves definitely indicate the rate of fire growth.

Cold working complicates the processes even more. Metallurgical testing has shown that the greater the amount of cold work in metal, the lower the temperature at which the metal will start to soften or collapse. Compression springs are cold worked to varying degrees during fabrication, depending on the application, which means that in a room subjected to fire conditions, collapsed springs in one location may be the result of the fabrication history of the metal only and do not reflect the fire`s point of origin or any other data pertinent to the fire.

An additional and key factor directly related to the collapse of a compression spring is the amount of weight on the spring at the time of the fire. The ability of steel to deform under load is referred to as “creep.” This is one of the effects fire protection engineers try to prevent when protecting structural steel from the effects of fire. For compression springs at room temperature, creep may be a consideration over a long period of time, but as the temperature increases during the fire, the effect of a load becomes greater and reduces the ability of the spring to resist creep. This is a key factor when trying to evaluate springs in the postfire environment.

Even if we knew all of the other possible factors that can affect compression springs (carbon content, fabrication history, and so on), and we knew for certain there was no load on the springs at the time the fire started, we still could have problems evaluating the final condition of the spring. For example, as the fire progresses and the spring heats up, any material that falls down onto the spring will impose a load and may significantly affect the final height of the spring, even if the material subsequently burns away. Since investigators rarely can determine the load conditions prior to and during the fire, it is functionally impossible to determine if the collapse of a spring is a function of annealing or creep processes. As Tobin and Monson point out: Since creep is a function of time-dependent strain and the amount of the strain (weight) affects the ability of a spring to withstand collapse, then large amounts of weight may permanently compress a spring at elevated but noncritical temperatures. As an example, heated springs could permanently collapse when suppression crews step on them during fire operations.

No documented research studies could be found to support the contention that collapsed springs prove any particular fire cause. On the contrary, all of the available work indicates that collapsed compression springs should not be used to confirm a fire`s cause. The most technical evaluation of collapsed compression springs was conducted by Tobin and Monson at the Federal Bureau of Investigation`s Crime Laboratory and Forensic Science Research Unit. Their work included testing springs in electric furnaces at temperatures between 800}dgF and 1,800°F and controlled mattress test burns and tests during which impressed loads were placed on the springs. The FBI Metallurgy Unit evaluated the springs at the end of the tests. Tobin and Monson concluded that “…observation of the `collapsed` state of coiled furniture/bedding springs is not a reliable indicator of whether a fire was initiated by a smoldering cigarette or accelerated by the presence of a hydrocarbon.”

In 1992, Lentine, Smith, and Henderson published the results of a study conducted at the “Oakland” fire in 1991.11 They examined 50 nonincendiary structure fires that had burned to completion with the intent of analyzing the effect the fire had on glass, copper, steel, and concrete. The majority of the bed springs at the fires showed signs of physical deterioration/disintegration. Further, 84 percent of the structures showed signs of deteriorated springs, and 98 percent of the structures showed signs of deteriorated steel–many showing visual evidence that could have been misinterpreted to indicate that the steel springs had melted. They reported that steel samples from the fire gave the appearance of having melted at temperatures far below steel`s rated melting point even though they determined through metallurgical examination that no melting had taken place. In some cases, this observed deterioration occurred at temperatures as low as 593°C (1,100°F). In regard to coiled compression springs, they found that “No correlation was found in the Oakland Fire between the shape of the deteriorated area and any particular attribute of the fire environment.”

The inability to determine carbon content, applied loads, fabrication history (amount of cold working or other prefire conditions), and fire or postfire conditions such as temperature/time relationships, loads applied to the spring during the fire, and the rate of cooling make it impossible to determine anything about the fire`s growth rate, the heat energy transferred to the spring, or the relationship between the amount of damage to the spring and the cause of the fire. Even if all the necessary data could be gathered, there is no quantitative or qualitative research to judge the results against, bringing us back to the earlier question: Can the appearance of springs be of use to fire investigators?

As a general indicator of fire development and spread, collapsed springs may provide useful information with regard to the fire`s origin when evaluated with other fire patterns as part of the overall fire investigation. Walking into a room and seeing a collapsed set of bed springs and jumping to the conclusion that the fire started in the mattresses and was caused by smoking– “case closed”–would be scientifically unsound.

On the other hand, this same condition when considered with all of the other fire patterns within the room may provide the basis for determining the fire`s origin. Once we have to isolate the bed as the point of origin, numerous other factors have to be considered before the cause of the fire can be determined, including the following: What sources of ignition are available on or near the bed? What is the condition of the room? Is there any evidence to suggest that materials other than bedding were on the bed? What was the first material ignited? Do the fire patterns confirm that the fire spread outward from the suspected source of ignition?

When taken together with all the other indicators (and the rest of the investigation), the appearance of collapsed bedsprings as one of the fire patterns may be useful in reaching your conclusion. However, by no means can the presence of collapsed springs be used to “prove” the cause of the fire. The only thing they prove is that heating occurred in the area of the springs for the period of time it took to cause permanent spring collapse. n


l. Tobin, W.A., and K.L. Monson. Nov. 1989. “Collapsed Spring Observations in Arson Investigations: A Critical Metallurgical Evaluation.” Fire Technology–National Fire Protection Association, Nov. 1989.

2. Mammone, M. “Determining the Point of Origin,” Fire House, Sept. 1994.

3. Tobin and Monson, 1989. Tobin, W.A. Mar./Apr. 1990. “What Collapsed Springs Really Tell Arson Investigators,” Fire Journal, National Fire Protection Association (NFPA). Lentini, J.J., D.M. Smith, and Dr. R.W. Henderson. Aug. 1992. “Baseline Characteristics of Residential Structures Which Have Burned to Completion: The Oakland Experience.” Fire Technology, NFPA.

4. Tobin and Monson, 1989.

5. Ginzburg, Vladmir B. 1989. Steel Rolling Technology–Heat Treatment of Steel. Unk. Publisher.

6. Harmathy, T.Z. 1993. Fire Safety Design & Concrete. New York, N.Y.: John Wiley and Sons.

7. Tobin and Monson, 1989.

8. Ibid.

9. Ibid.

10. Ibid.

11. Lentini et al, 1992.

Additional References

Carrol, John R. 1979. Physical and Technical Aspects of Fire and Arson Investigation. Thomas Books: Springfield, Ill.

Cole, Lee S. 1992. Investigation of Motor Vehicle Fires, Third ed. Lee Books: Novato, Calif.

Cote, Arthur E. 1991. Fire Protection Handbook, 17th ed. National Fire Protection Association (NFPA): Quincy, Mass.

Damant, Gordon H. and Said Nurakhsh, “Heat Release Tests of Mattresses and Bedding Systems,” Journal of Fire Sciences Vol. 10: Sept./Oct. 1992.

DeHaan, J.D. 1991. Kirk`s Fire Investigation. Brady: Englewood Cliffs, N.J.

Drysdale, Dougal. 1985. An Introduction to Fire Dynamics. John Wiley and Sons: New York, N.Y.

Friedman, Raymond. 1989. Principles of Fire Protection Chemistry, 2nd ed. NFPA: Quincy, Mass.

Haessler, Walter M. 1974. The extinguishment of Fire. NFPA.

921 Standard for Fire and Explosion Investigations, NFPA, 1992 and 1993 eds.

Mcgill, Lamont “Monty.” 1994. Unpublished teaching document.

Paul, K.T., and S.D. Christan. “Standard Flaming Ignition Sources for Upholstered Composites, Furniture and Bed Assembly Tests,” Journal of FIRE SCIENCES, Vol. 5:May/June 1987.

Perry, Donald G. 1990. Wildland Firefighting, 2nd ed. Fire Publications Inc.

Pocket guide to Arson and Fire Investigation, 2nd and 3rd eds. Factory Mutual Engineering and Research: Norwood, Mass. 1979, 1990.

Posey, James E. and Eleanor P. Posey, “Using Calcination of Gypsum Wallboad to Reveal Burn Patterns,” The Fire and Arson Investigator, International Association of Arson Investigators, Inc., Vol 33:3, March 1983.

Roberts, Wm. L. 1983. “Hot Rolling of Steel.” Marcel Dekker, Inc.: New York, N.Y.

Savitsky, E.M. 1961. The Influence of Temperature on the Mechanical Properties of Metals and Alloys. Stanford University Press.

n RONALD K. MARLEY is fire chief of the Shasta College Fire Department in Redding, California, and an instructor in the college`s fire technology program. He has 13 years of explosive ordnance disposal experience within the military and local law enforcement community. Involved with fire investigations since 1984, he previously served as the chief fire investigator for the City of Dixon (CA)Fire Department. Marley is a certified fire officer, fire instructor, and fire investigator and an accredited master instructor in the California State Fire Marshal`s Office. He has various degrees including a bachelor`s degree in fire and explosion origin and cause investigation and is currently conducting a variety of long-term research projects related to fire and explosion analysis.

No posts to display