Arson is one of the hardest crimes to solve because the evidence is actively being destroyed as the crime is committed. Dr. Mark Sandercock, a scientist in Trace Evidence at the RCMP's National Forensic Laboratory Services in Edmonton, Alta., and Professor James Harynuk of the University of Alberta (U of A) developed a computer program that can analyze arson debris in seconds. Deidre Seiden spoke to Sandercock about the benefits of the tool.
How do you currently analyze arson debris?
When a fire investigator thinks a fire is suspicious, he will collect samples taken from the possible area where the fire originated. The samples are usually very burnt, often unrecognizable bits of material from the scene such as flooring or furniture. At Trace Evidence, we extract the volatiles from the arson debris and analyze them for an ignitable liquid, such as gasoline. But in doing so, we're also extracting all of the other volatile compounds produced by the burning of the materials at the fire scene. We're looking for something in a very complex background, much like looking for a needle in a haystack.
Two scientists are assigned to examine the data we obtain from the arson debris analysis in a process we call peer review. One scientist reviews the data and writes up her findings. Then that result will go to a second scientist who reviews all of the first person's work. If their findings are in agreement, a report will be written and it will go back to the investigator.
Does it take a long time?
It can take about an hour to process the data from a straightforward sample. But some samples are quite complex, and can take several hours, or even a full day just to interpret the data from one sample. If you have two people examining the data for each sample, our scientists can spend a significant amount of time reviewing data.
Is this why you collaborated with the U of A?
Yes, we worked with the U of A to develop a computer program that can examine the data from arson debris in seconds and correctly identify if gasoline is present. If we can implement this model in our laboratory, it could eliminate the need for the second scientist to review the data because the computer can look through the data set from one sample in about 10 seconds, saving us a lot of time.
How did you get involved?
As the program technical leader, I'm always looking at what's out there and what we can use to help us do our jobs better. So when a new professor in the chemistry department at the U of A called me looking for research ideas, we started discussing various possibilities. This particular one caught James' interest so he ran with it. The university did the work and I had the great job of being their consultant and helping them stay on track.
How would this impact an investigator?
It'll have a two-fold impact for our investigators. The obvious one is the arson cases will leave our building sooner and the results will go back to the investigator that much faster. The second benefit is, because we're spending less time doing these reviews, the scientists are freed up to do other casework.
Has it been used yet?
Not yet. But we did a pilot study where we took several hundred samples that we had processed and gave the data to the U of A. They ran the data through their computer program and they got the same results that we did. If there wasn't gasoline in the sample, the computer program didn't find it, and where there was, the computer program did find it. That was basically proof of concept that it worked.
Two things need to happen before the program is ready to be used in our lab. First, the U of A is building a user-friendly interface so it's easier for us to run the program. They also want to build their database so it covers a broader range of ignitable liquids. Optimistically, we are about two or three years away from trying it out in our laboratory.