Building a Firearms Toolmark Database
Although they are mass produced, every firearm is unique, and when fired, they leave unique markings called toolmarks on the bullet and cartridge casing. Law enforcement agencies have used these “fingerprints” to match firearms with bullets as part of their criminal investigations for more than a century. While forensic evidence of this kind wouldn’t likely be enough to get a conviction on its own, it has played a crucial role in linking suspects to crimes, and the ability of firearms examiners to make those matches has never been a source of controversy … until recently.
In 2009, a report by the National Academy of Sciences questioned, among other things, the lack of objective methods for evaluating and identifying toolmarks. To address this, the Academies recommended development of objective methods rooted in mathematics and statistics. Development of objective methods, however, is hindered by the lack of access to diverse toolmark data sets. During a symposium held at NIST called “Measurement Science and Standards in Forensic Firearms Analysis 2012,” one of the main needs the attendees identified was for a database where bullets, cartridge cases, and toolmark surfaces could be shared among researchers. Having such a database, they said, would make it easier for them to develop and evaluate new systems, methods, and algorithms for matching bullets with firearms.
In 2013, I was awarded a grant by the National Institute of Justice (NIJ) to begin creating this database. As my colleagues Johannes Soons, Robert Thompson and I started to look for datasets of bullets and cartridges cases, we found that a lot of research had been conducted by firearms examiners, university researchers, and instrument manufacturers all over the U.S. since the 2009 report. Before we approached them, though, we decided to begin to design the open-access database using the 500 measurements of bullets and cartridge cases gathered as part of the Forensic Toolmark Analysis Project (FTAP).
Before we could even begin building the database, we had to figure out how to get all the data into the same format. We had generated all of our data at NIST using the same 3-D confocal microscope and saved all of it in the ASCII text format. This worked pretty well for us, but it wasn’t going to work once we expanded this database to include multiple instruments and other researchers. We needed to find a common standard file format for everything going into the database. This format needed to be efficient, standardized, and fit into the firearm toolmark measurement requirements.
Ryan Lilien from Cadre Research and I formed an informal group of researchers and practitioners and started discussing the need for an interoperable file format. The teleconferences were heated at times with everyone promoting their own file format as “the one,” but after a couple of meetings, we discovered an ISO (25178–72) format called XML 3-D Surface Profiles (X3P). After doing a bit a research, we agreed that this was the format that we should use to make all 3-D topography measurements of firearm toolmarks. Even better was the fact that the major instrument manufacturers either already supported X3P or were willing to adopt it.
Now that the file format issue was resolved, we focused on filling the database. The data we had gathered as part of FTAP came from the bullets and casings that we had fired under ideal laboratory conditions using only a couple types of guns. For the database to be successful, I desperately needed data from a diverse set of firearms and real world examples.
I first approached researchers I had met in the past, but for the most part they had built their datasets to answer their individual questions, so they were limited in size and scope. I knew I had to cast a wider net to find the diversity I needed. In 2014, I presented the NIST Ballistics Toolmarks Research Database (NBTRD) at an Association of Firearms and Toolmark Examiners (AFTE) training seminar. I gave the audience a sense of what I was trying to build, and, because I knew each major crime laboratory housed their own reference collection of firearms that could prove to be invaluable for the database, I asked them for their help. I expected to arouse some interest, but I was surprised at the number of examiners from local, state, and federal forensic laboratories who approached me once I stepped down from the podium. Many offered test fires from previous research they’ve conducted, and some offered to generate test fires from their entire reference collection.
Before I knew it, I was being inundated with data.
One of the laboratories who agreed to help was the Prince George’s County Police Department (PGPD) here in Maryland. I worked closely with Scott McVeigh from PGPD’s firearms examination unit and came up with a plan to use their reference collection. It was great having a local laboratory I could turn to for help. Before I knew it, I had collected over 1,200 test fires for the database, many more than I could’ve imagined at the beginning of the project. Now that I had all this data, I needed a good way to share it. Up to this point, I only had zipped files for each dataset with no real way to index or search the data.
In 2015, the NIJ awarded an extension for the NBTRD to develop a web database to index the collected data as well as allow users to upload their own data to further build the scale and diversity of the database. Luckily, NIST has the expertise for this task, and a year later, I announced the live launch of the NBTRD at the 2016 AFTE meeting in New Orleans.
A year later, I’m happy to report that the future of the database is a bright. I’ll personally be curating and adding more and more datasets. I’ve recently completed an agreement with the FBI’s Firearms and Toolmark Unit (FTU) to collaborate and share knowledge in the future. Most importantly, the FTU houses a reference collection of over 7,000 firearms and thousands of different ammunition types. They are currently going through their entire reference collection and shooting at least six different ammunition types from each firearm.
Toolmark testing has been an essential forensic tool for more than a century. It’s been great to have the opportunity to contribute toward the improvement of this discipline. My experience with building this database taught me that, while people may disagree about the specifics, any obstacle can be overcome so long as the group is committed to achieving the same goal. I hope the research generated from the NBTRD will one day enable examiners testifying in criminal trials to say, and back up, how confident they are that a bullet came from a particular gun.
Until that day, my colleagues and I will keep working to improve the science of firearms and toolmark identification.
*See NIST 3-D Ballistics Research Database Goes Live for more.
This post originally appeared on Taking Measure, the official blog of the National Institute of Standards and Technology (NIST), on July 22, 2016.
To make sure you never miss our blog posts or other news from NIST, sign up for our email alerts.
About the Author
Xiaoyu Alan Zheng is a mechanical engineer in the NIST Surface and Nanostructure Metrology Group. His research is focused on objective measurements and analysis of 2-D/3-D ballistics toolmarks. He is currently a member of the Subcommittee on Firearms & Toolmarks in the NIST OSACs as well as a technical adviser for AFTE.