Firearms matches between the marks left on casings and the grooves in a firing mechanism continue to be a minute forensic trail leading the way to killers, and resulting in convictions. An alleged serial killer in Phoenix was identified just last week based on the weapons he had used in otherwise seemingly-random crimes over a two-month period.

But some forensic scientists hold that firearms analysis needs to be more quantifiable, similar to DNA match probabilities.

A new study undertaken by scientists at the National Institute of Standards and Technology proposes a new method of assessing 3-D topographical images of the minute tool marks left behind after a round is fired—and then mathematically understanding how likely one gun could be mistaken for another.

Their conclusion: a database big enough would provide reproducible identification evidence—but it will never be up to the “gold standard” of the best DNA matches. Variations such as firing conditions, wear in the gun, and contaminants cause potential differences in the tool marks left behind, they write in the journal Forensic Science International.

“Because of the inherent variability of the firing process, we do not expect evidence from firearms to exhibit the extremely low error rates that are characteristic of DNA evidence,” they write. “However … the extremely small false positive errors rates calculated from the models suggest the feasibility of applying (this method) to a large number of firearms manufactured under similar conditions and producing correlation results to support identification, if not exclusion, decisions.”

The engineers’ method is called “congruent matching cells,” or CMC. It divides full topographical images of the breech face impressions into dozens of separate pieces, and each are scrutinized down to the minutest level for differences. (The images of the breech face impressions were highly detailed, made under a white light source by disk scanning confocal microscopy, as they describe at some length.)

For the initial test, they looked at 40 cartridge cases ejected from 10 Ruger 9 mm pistol slides—the 10 guns of which were manufactured right consecutively at the factory. They then tested the methodology on an additional 95 cartridge cases from different slides.

The CMC method, with the breakdowns of each image, then provided a statistical foundation for calculating error rates. They found the numbers bore out the minute differences among the small group of guns in the study. However, the identification power will increase as the database expands, as well.

“For realistic databases with many entries of firearms and ammunition, even when classified according to model and manufacturer, the overlap of (known match) and (known non-match) distributions can become significant and the error rates will likely increase significantly,” they write.

Criticism of firearms and toolmark matching first really gained ground with the watershed 2009 report by the National Research Council entitled “Strengthening Forensic Science in the United States: A Path Forward.” Another critical document issued in September 2016 by the President’s Council of Advisors on Science and Technology, or PCAST, similarly criticized the process of firearms matching for not being quantitative, or fully explained to juries.

Defense attorneys have picked up on these arguments—like when the counsel for convicted “Grim Sleeper” serial killer Lonnie Franklin Jr. argued that the 2-D images matching the bullets fired to Franklin’s .25-caliber handgun were not sufficient to link the weapon to the victims. Similarly, just weeks after the PCAST report was issued, attorneys for six alleged members of the Chicago-based Hobos gang argued that the ballistics evidence against the six accused racketeers should be thrown out. Neither argument worked. Franklin was eventually sentenced to death, and the half-dozen Hobos received life sentences.  

NIST has several projects in the works with 3-D databases and other advanced analysis of firearms and toolmarks comparison. Another study the agency published in Forensic Science International this month assesses using computer matching algorithms to compare 3-D topography measurements of breech face impressions.

The National Integrated Ballistic Information Network (NIBIN) that is run by the Buearu of Alcohol Tobacco and Firearms (ATF) is currently the only national network for comparison of firearms. Esbalished in 1999, it currently touts more than 74,000 hits.