Firearms and Toolmark Identification

Firearms examination rests on a physical claim that is easy to state and hard to bound: metal working against metal transfers microscopic marks, and the marks a particular firearm leaves are said to differ from those left by any other. Everything contested in the discipline concerns the second half of that sentence — how much agreement is enough, what can imitate agreement, and what an examiner is entitled to say once the agreement is found.

Three kinds of marks and where each comes from

Class characteristics are determined before a component is made. Caliber, the number of lands and grooves in a barrel, the width and direction of the rifling twist, the shape of a firing pin and the type of breech face finish are all design features shared by every item made to that specification. Class characteristics exclude efficiently. A bullet bearing six land impressions with a right twist did not come from a barrel cut the other way.

Individual characteristics are the microscopic irregularities left by imperfect manufacture, by wear, and by damage acquired in use. They are the features an identification rests on, on the premise that the accidental detail on any one working surface differs from that on every other. Subclass characteristics sit awkwardly between the two. They arise from the condition of the tool that shaped the surface, and they are shared by a subset of items — everything produced while the tool was in that condition.

Subclass carryover is the hardest case in the discipline, and it is hardest precisely where the components were produced one after another. A broach or a die wears progressively, so consecutively manufactured barrels or breech faces can carry common detail that did not originate in random imperfection at all. Distinguishing subclass from individual detail depends on knowing how the surface was made and on examining known non-matching test fires from the same production run.

The surfaces an examiner actually compares

A fired cartridge case records the chamber and the breech end of the action. The breech face impression is the flattened area struck when the case is driven rearward, and it carries the finish marks of that surface. The firing pin impression records the tip that struck the primer, including any drag as the case moved. Extractor and ejector marks record the parts that pulled the case from the chamber and threw it clear, and chamber marks record the walls the case expanded against.

A fired bullet records the barrel. Lands and grooves leave impressions whose number, width and twist are class features, and within those impressions run striations produced by the fine irregularities of the bore. Striations are the individual detail on a bullet, and they are also the detail most easily destroyed. A bullet that struck bone, masonry or a hard surface may retain enough class information to be useful and no comparable individual detail at all.

Sufficient agreement and who decides it

The comparison is made on a comparison microscope, which places two specimens under separate objectives and joins the fields so that a single image shows both surfaces meeting at a line. The examiner rotates and translates the specimens until striations or impressed detail align across that line, then judges the extent and quality of the correspondence. Test fires from the questioned firearm supply the known side, and several are usually produced to show how much the marks vary between shots.

The threshold applied is described in the discipline's own literature as sufficient agreement: correspondence exceeding the best agreement the examiner has observed between marks known to come from different sources, assessed against training and experience. That formulation is honest about what it is, and it is also the source of the principal criticism. It does not specify a count, a proportion or a statistic, so two examiners applying it are not applying an identical rule.

Correlation databases and the push toward measurement

Imaging systems capture digital images of fired cartridge cases and bullets and correlate them against stored images from other cases, returning a ranked candidate list. The correlation is an image comparison performed by software. It generates investigative leads and links between incidents, and it produces no conclusion about source. Only a microscopic examination of the physical components can do that, and a file describing a correlation result as a match has skipped the step that matters.

NIST has worked on replacing the visual judgment with a measurement. Three-dimensional topography captures the height of a surface rather than a photograph of it, which removes the dependence on lighting and orientation that makes two images of the same surface look different. From topographic data, similarity metrics can be computed and reference artifacts can be used to check that instruments agree. The aim is a statistical statement of evidentiary strength resembling the way a DNA match is expressed numerically.

That work is not finished, and objective measures are not yet what most casework relies on. NIST's scientific foundation reviews evaluate the empirical basis of forensic methods one discipline at a time, with firearm examination among them, and the Organization of Scientific Area Committees drafts the documentary standards a laboratory may adopt. Adoption is voluntary unless accreditation or a jurisdiction's rules make it otherwise.

Conclusion language and the file underneath it

The traditional wording was identification to the exclusion of all other firearms. The National Academies report on strengthening forensic science rejected that formulation, observing that the pattern disciplines had not been rigorously shown to support a claim of a single source and that they operated without enforceable standards across laboratories. The criticism was aimed at the claim, not at the practice of comparison, and the response has been to bound the wording rather than abandon the examination.

Bounded language states the conclusion as a source identification supported by agreement of individual characteristics, with an express acknowledgment that the method cannot exclude every other firearm in existence and that the conclusion reflects the examiner's judgment. Some laboratories go further and report the strength of the correspondence on a graded scale rather than collapsing it into a single word.

An NIJ-funded study using consecutively manufactured barrels reported examiners correctly associating bullets with the source firearm 98.8 percent of the time, an error rate below 1.2 percent, with experience level not significantly affecting accuracy. That result comes from a set-piece study on the hardest manufacturing condition. It does not establish an error rate for degraded casework specimens, for other firearm types, or for any individual examiner, and the same limitation applies to the black box work discussed in latent print comparison and the ACE-V method.

The material that makes a conclusion reviewable is the bench file rather than the report. It should contain the examiner's notes on class characteristics observed, a record of the test fires produced and how many, photographs of the aligned comparison at the magnification used, the reasoning for treating detail as individual, and the verification record naming the second examiner and describing what that examiner was given. What a proponent has to establish about the method itself is treated in the reliability showing for a forensic method, and the laboratory's own audit obligations in crime laboratory accreditation.