Fire investigators are interested in knowing the best type of samples to obtain, best containers to use, and the appropriate number of samples that would be representative for a fire scene. Additional information can also be gained from the general review of the type of losses that most often occur, the type of accelerant most often found, and the number of positive results determined. Also the importance of new analytical tools such as the use of Gas Chromatography coupled with Mass Spectrometry with a second Mass Spectrometry operation conducted (GC/MS/MS). This paper will explore these questions by reviewing of our database of the results of samples and cases submitted to Activation Laboratories Ltd.

Activation Laboratories has studied, developed, and has advanced the technology used to determine the presence or absence of ignitable liquids in fire debris. Samples submitted for analysis are usually from the insurance industry in support of litigation in suspected arson cases. The samples are routinely analyzed in our laboratory by Gas Chromatography coupled with Mass Spectrometry (GC/MS), the gold standard for analysis in this forensic field. In early 1997, Activation Laboratories developed a second confirmatory analysis using technology that is currently the ultimate in sensitivity and specificity, GC/MS/MS. This is used when samples have residues in trace amounts, are very complex, or when the effects of the fire incident or subsequent environmental factors weather the samples.

During the time that GC/MS/MS has been offered at Activation Laboratories, a total of 381 samples were submitted. Of this total, 41% were determined to be positive for the presence of a ignitable liquid. The results are further broken down in the following table.

Note that the overall performance, in terms of the percent of samples submitted being positive, is often directly affected by the quality of the sample taken. Investigators from coast-to-coast have submitted these samples from a wide variety of situations. Three sample types stand out in this data. Concrete was the surprise in this review with 100% of the samples having positive results. We would add that these samples had chunks of concrete, say of 2 to 3 centimeters in length that had been chipped out. They were not small flakes from just the concrete surface. Carpet and wood have the best success following concrete.

The use of GC/MS/MS lead to a noteworthy and significant improvement in the number of positive results recorded for those sample types that it was used on. GC/MS/MS is only performed on select samples. When only these samples are considered, the performance of GC/MS/MS is even more impressive.

A number of submissions were identified where the investigator had used a trained K9 to aid in the locating of samples that may have contained traces of a ignitable liquid. The following table shows the performance of one individual K9.

This particular K9 located samples that were positive 46% of the time. This is an improvement of 5% more than the average of 41% in this single study. It may also be an important statistic that, when a K9 is used, almost 70% of the cases had a positive sample in that submission. In some cases it requires only one positive sample to potentially save thousands of dollars from fraudulent claims.

The 381 samples were submitted in 3 types of containers. The performance observed for each is outlined in the following table.

The large percentage of mason jars indicates the high number of Canadian based submissions. Mason jars are most often used in Canada and in this study significantly outperformed the results for samples stored in new lined metal paint cans by 15%. Some investigators worry about breakage in shipping and the effect of the red seal that is adhered to the tin lid. There were only two breakages noted from over 300 mason jars submitted and these were shipped with virtually no packing material. With the use of GC/MS and GC/MS/MS based methods, the red seal no longer presents a significant interference to the analysis of fire debris samples. The most surprising figure was the performance of the nylon evidence bags where over 80% of these samples were determined to be positive. Possibly the biggest reason for this is that good pre-tested nylon evidence bags have less chemical background than metal paint cans or even sterilized glass mason jars. This allows for the study of smaller amounts of ignitable liquids and simplifies the interpretation of the samples. Through our in-house research, Activation Laboratories is encouraging our customers to use pre-tested nylon evidence bags. The advantages include lightweight and reduced space requirements for carrying them from site-to-site as well as reduced shipping costs. Less chance of breakage in shipping, cost-competitive with other containers, and the better than expected ruggedness. However, of primary importance is the ultra-low level of their chemical background. This allows the best use of GC/MS and GC/MS/MS analysis especially for highly weathered or very small samples.

It should be remembered when reviewing these results, that the size of the structure can alter the reported percentages. As an example there might be more samples taken to adequately cover a case involving a warehouse than needed to cover a vehicle fire.

Of the samples with positive results during the time period of this study, the following accelerants were identified with the following observed frequency.

Gasoline was identified in 65% of the samples that were positive for a ignitable liquid. There are similar statistics for most places in the world. Petrol/gasoline was found to be the accelerant of choice in 69% of the positive results in a study in Australia in 1996 (Jackowski). With these results in mind, Activation Laboratories has focused their research on improving the technology to detect all ignitable liquids with a special focus on gasoline. Our confirmatory analysis by GC/MS/MS is extremely sensitive to the detection of gasoline.

A total of 74% of the cases had only 1 or 2 samples submitted. Of these small submissions an average of 71% of the cases had a positive result. This indicates that many investigators will only take samples if they are almost certain that the samples will lead to a positive result. Thus for cases that had 1 or 2 samples, 71% of these cases had a positive result. Investigators most likely believe that by taking only 1 or 2 samples they are essentially certain to be positive, that they will be seen as saving their client money and enhance their image. It may also show a lack of confidence in past laboratory analysis where samples that they firmly believed should have been positive were reported as negative. Thus investigators are taking fewer samples less often as they do not perceive the value in taking more than 2 samples.

In this study it was observed that those cases that had 6 samples submitted, resulted in one or more positive samples in 100% of the submissions. GC/MS/MS was used in 20% of these submissions adding to this very high success rate. Thus the success rate in finding evidence of ignitable liquids from the cases chosen to be investigated could possibly increase by almost 30% by taking 6 samples instead of 1 or 2. An increase of 30% in successfully identifying potential fraudulent claims could mean significantly more in savings than the few hundred dollars to test 4 or 5 additional samples.

Other publications support these findings (Byron). From a larger study in the U.S. it was found that “representative sampling occurs at 5 samples. Further sampling in unnecessary except to help determine multiple points of origin or with very extensive fires”.

Thus the less than adequate sampling, by taking 1 or 2 samples and these only under certain conditions, must result in a dramatic under reporting of positive identifications for ignitable liquids in these cases. By under sampling cases, the insurance companies may be losing up to 30% of the investigations that they believe to have suspicious circumstances.

Today the fire investigator can significantly improve the probability of finding positive samples should ignitable liquids be present at the site of investigation. Choosing good samples, with the use of a K9 if appropriate, taking 5 or 6 samples for better site representation, storing the samples in pre-tested nylon evidence bags, and using GC/MS analysis with the availability of GC/MS/MS confirmation, yields results with the highest confidence attainable anywhere today.

The more sensitive and specific analysis by GC/MS/MS helps the investigator, as it is more forgiving to the sample location. The samples can still be proven positive a further distance from the fire origin since it can detect more minute residues of ignitable liquids. It is also important in the analysis of samples from hot or arid climates for the same reason. Adding to this extra sensitivity is the superior selectivity that GC/MS/MS brings to the analysis of these complex samples. It is significantly better than GC/MS at filtering out pyrolysates and making the identification of ignitable liquids more definitive. Use of this second confirmatory analysis is a very significant advancement in this field as outlined in an article published recently in the United States, entitled “Canadians Use New Technique to Find Arson” (Hays).

If insurance companies were 30% more successful in defending against fraudulent claims by having a more representative number of samples taken in an investigation (5 or 6), and by using the latest technology in the laboratory (GC/MS and GC/MS/MS), they could then recoup significant amounts of monies paid out to false claims. Ultimately this would reduce the cost of related insurance to their policyholders. Publicizing this success may also reduce the incidence of arson as the would be Arsonist would know that the odds of their success has been reduced, which in turn may also save the lives of innocent victims.

(1) Jackowski, Incidence of Ignitable Liquid Residues in Fire Debris as Determined by a Sensitive and Comprehensive Analytical Scheme. J. Forensic Science 1997; 42(5): 828-832.
(2) Byron, Adequate Sampling of Fire Scenes. Fire and Arson Investigator, December 1990, Volume 41, #2.
(3) Hays, Canadians Use New Technique To Find Arson. National Underwriter; Property & Casualty/Risk & Benefits Management Edition, November 9, 1998.