So I am fond of long form journalism, one genre of which is “true crime” (hopefully no relationship to the 80s era tabloid mag!)
How can identifying a specific gun from a bullet even be feasible? I mean, sure caliber can be easy to determine.
My objections:
All (modern) guns are rifled. That, I assume, could help identify the barrel of the gun, because there may well be marks on the bullet,
But - once a bullet, especially a lead bullet, a soft metal, hits, it is surely going to eradicate all reliable evidence of the original rifling marks as the bullet deforms.
Also, every, say 1980 era AK-47 is identical in the rifling. How could simple wear and tear to the rifling influence a full metal jacket bullet in an identifiable way?
AIUI, any specific gun, even if maintained in good order, has some sort of tiny irregularity - grooves, bumps, whatnot - inside the barrel which will engrave identifiable traces in the bullet as it travels through this barrel.
Source: Nothing in writing, and I’m not a gun expert myself, but my mother used to be a police detective and would also investigate homicides, and this is how she explained it to me.
The claim is that individual guns leave unique tool marks on bullets and casings, and that these tool marks can be used to link the bullet and casing to the gun that fired it, and determine if multiple bullets and casings were fired from the same gun.
Going back about 25 years this has been called into question. There have been a variety of studies revealing various levels of reliability for firearm forensics. Methodological and analytic flaws are the general criticism.
It is important to remember that scientists often don’t have a “side” in a debate when they are pointing out flaws in an experiment. The flaw does not mean the conclusions of the researchers are wrong, it means that the conclusions are not supported by the experiment.
Anyway, below is a short editorial which discusses why this type of forensics should not be trusted. I do not have enough expertise to defend the author’s opinion or refute it.
Even finger prints aren’t unique, not when you take into account how much a print from the same finger can vary. I would assume that analyzing the bullet would let you say that a gun wasn’t used, or might have been used, or really looks like it was likely used. And sometimes the bullet is squashed enough that you can’t even do that.
Another couple of points, which may or may not be covered in the linked article, but not everyone is going to read it.
One, bullets deform differently depending on what they impact and/or pass through. Often, at the scene of a shooting, you’ll recover multiple spent rounds, some of which may have very limited deformation. And especially with FMJ (Full Metal Jacket with the copper covered lead rounds most commonly seen) they tend to deform a lot less than specially designed HP (hollow points). So, even if one round is totally deformed, you may be testing others that make for a better comparison.
Two, IF the weapon is recovered, they normally fire multiple rounds from the gun at different, ideally matching final materials (gelatin, water, etc.) so that they can make it more of an apples to apples comparison. Multiple firings also help because barrels heat up over firing, and the rifling can show minor differences between the first (cold) and last (warm to hot depending on how many rounds, how “hot” the rounds are, and everything else).
See the classic SD article for just how hot some weapons can get, and how relatively quickly (or not, again, depending on the weapons):
All that said, @puzzlegal and the rest are correct, it’s as much art as (faux) science. Generally I think it’s in the realm of “one more strike against” a possible criminal, if a recovered firearm is tied to both the possible perp and victim.
Speaking only as a semi-fan of ballistics with a tiny bit of experience in cleaning and maintaining their own firearms and having watched much more involved individuals talk shop who had actual gunsmithing experience.
True, the bullet may be squashed*, but often it is marks on the ejected cartridge that can be matched to other cartridges, and eventually to a gun barrel.
Locally, police have been able to recover cartridges from a shots-fired location, and match them to cartridges from other crime scenes. When they eventually recover the gun that matches those marks, the possessor of that gun can be charged with all the crimes where the bullet cartridges were fired from that gun. This marks that person as a serial criminal, and will increase the punishment if they are convicted.
But bullets do not squash much when they hit something soft, like a human body.
I once read a murder mystery where the culprit was a cop. He’d been on the police range firing box after box of ammunition through his service pistol. It was to cause enough wear on the rifling that it wouldn’t match what was found in the victim’s body and was a red herring early on – “It couldn’t be him; the ballistics don’t match.” – but was caught through other evidence.
It would take a very large number (thousands) to make any difference.
Examination can demonstrate that a particular gun could not have been used in the crime.
Past damage, or corrosion, can mark projectiles, or cartridges, in unique ways. Both often, the most that can be said, is that the evidence is compatible with a particular brand of gun being used. If such a gun is available to the defendant, it can support the prosecution case.
There are several problems with that opinion piece, starting with the authors trumpeting themselves as “anti-expert experts”, while only one of them has a bio suggesting some expertise in firearms forensic analysis, apparently having testified in cases as a defense witness.
They want to label inconclusive findings as errors to inflate the error rate, rather than as honest appraisals that insufficient evidence exists for a definitive conclusion. For comparison, if an autopsy of a decomposed body reveals no certain cause of death, has the medical examiner made an “error”?
It’s true that we should be wary of conclusions made by experts in borderline cases. To again draw a parallel with medicine: breast pathology experts commonly disagree about diagnoses of atypical ductal hyperplasia, ductal carcinoma in situ and minimally invasive ductal carcinoma. Tightening of standards to minimize subjectivity is one answer. I doubt AI will eliminate all doubts.
Yeah, I imagine it’s mostly a statistical thing. Maybe you can narrow it down to “This bullet likely came from this group of about 10% of this model ever sold. The defendant had one of those 10% in their possession when arrested.” By itself, it’s not enough for a conviction, but when added to other evidence, it can make a difference.